blob: 2de70f5d9405dcbc477ea76883f36c45ddd9afea [file] [log] [blame]
Eric Snow2ebc5ce2017-09-07 23:51:28 -06001
2/* Core extension modules are built-in on some platforms (e.g. Windows). */
3#ifdef Py_BUILD_CORE
Eric Snowfc1bf872017-09-11 18:30:43 -07004#define Py_BUILD_CORE_BUILTIN
Eric Snow2ebc5ce2017-09-07 23:51:28 -06005#undef Py_BUILD_CORE
6#endif
7
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00008#include "Python.h"
9#include "structmember.h"
10
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -080011PyDoc_STRVAR(pickle_module_doc,
12"Optimized C implementation for the Python pickle module.");
13
Larry Hastings61272b72014-01-07 12:41:53 -080014/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -080015module _pickle
Larry Hastingsc2047262014-01-25 20:43:29 -080016class _pickle.Pickler "PicklerObject *" "&Pickler_Type"
17class _pickle.PicklerMemoProxy "PicklerMemoProxyObject *" "&PicklerMemoProxyType"
18class _pickle.Unpickler "UnpicklerObject *" "&Unpickler_Type"
19class _pickle.UnpicklerMemoProxy "UnpicklerMemoProxyObject *" "&UnpicklerMemoProxyType"
Larry Hastings61272b72014-01-07 12:41:53 -080020[clinic start generated code]*/
Serhiy Storchaka1009bf12015-04-03 23:53:51 +030021/*[clinic end generated code: output=da39a3ee5e6b4b0d input=4b3e113468a58e6c]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -080022
Łukasz Langac51d8c92018-04-03 23:06:53 -070023/* Bump HIGHEST_PROTOCOL when new opcodes are added to the pickle protocol.
24 Bump DEFAULT_PROTOCOL only when the oldest still supported version of Python
25 already includes it. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +000026enum {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +010027 HIGHEST_PROTOCOL = 4,
Łukasz Langac51d8c92018-04-03 23:06:53 -070028 DEFAULT_PROTOCOL = 4
Alexandre Vassalottica2d6102008-06-12 18:26:05 +000029};
30
Alexandre Vassalottica2d6102008-06-12 18:26:05 +000031/* Pickle opcodes. These must be kept updated with pickle.py.
32 Extensive docs are in pickletools.py. */
33enum opcode {
34 MARK = '(',
35 STOP = '.',
36 POP = '0',
37 POP_MARK = '1',
38 DUP = '2',
39 FLOAT = 'F',
40 INT = 'I',
41 BININT = 'J',
42 BININT1 = 'K',
43 LONG = 'L',
44 BININT2 = 'M',
45 NONE = 'N',
46 PERSID = 'P',
47 BINPERSID = 'Q',
48 REDUCE = 'R',
49 STRING = 'S',
50 BINSTRING = 'T',
51 SHORT_BINSTRING = 'U',
52 UNICODE = 'V',
53 BINUNICODE = 'X',
54 APPEND = 'a',
55 BUILD = 'b',
56 GLOBAL = 'c',
57 DICT = 'd',
58 EMPTY_DICT = '}',
59 APPENDS = 'e',
60 GET = 'g',
61 BINGET = 'h',
62 INST = 'i',
63 LONG_BINGET = 'j',
64 LIST = 'l',
65 EMPTY_LIST = ']',
66 OBJ = 'o',
67 PUT = 'p',
68 BINPUT = 'q',
69 LONG_BINPUT = 'r',
70 SETITEM = 's',
71 TUPLE = 't',
72 EMPTY_TUPLE = ')',
73 SETITEMS = 'u',
74 BINFLOAT = 'G',
75
76 /* Protocol 2. */
77 PROTO = '\x80',
78 NEWOBJ = '\x81',
79 EXT1 = '\x82',
80 EXT2 = '\x83',
81 EXT4 = '\x84',
82 TUPLE1 = '\x85',
83 TUPLE2 = '\x86',
84 TUPLE3 = '\x87',
85 NEWTRUE = '\x88',
86 NEWFALSE = '\x89',
87 LONG1 = '\x8a',
88 LONG4 = '\x8b',
89
90 /* Protocol 3 (Python 3.x) */
91 BINBYTES = 'B',
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +010092 SHORT_BINBYTES = 'C',
93
94 /* Protocol 4 */
95 SHORT_BINUNICODE = '\x8c',
96 BINUNICODE8 = '\x8d',
97 BINBYTES8 = '\x8e',
98 EMPTY_SET = '\x8f',
99 ADDITEMS = '\x90',
100 FROZENSET = '\x91',
101 NEWOBJ_EX = '\x92',
102 STACK_GLOBAL = '\x93',
103 MEMOIZE = '\x94',
104 FRAME = '\x95'
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000105};
106
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000107enum {
108 /* Keep in synch with pickle.Pickler._BATCHSIZE. This is how many elements
109 batch_list/dict() pumps out before doing APPENDS/SETITEMS. Nothing will
110 break if this gets out of synch with pickle.py, but it's unclear that would
111 help anything either. */
112 BATCHSIZE = 1000,
113
114 /* Nesting limit until Pickler, when running in "fast mode", starts
115 checking for self-referential data-structures. */
116 FAST_NESTING_LIMIT = 50,
117
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000118 /* Initial size of the write buffer of Pickler. */
119 WRITE_BUF_SIZE = 4096,
120
Antoine Pitrou04248a82010-10-12 20:51:21 +0000121 /* Prefetch size when unpickling (disabled on unpeekable streams) */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100122 PREFETCH = 8192 * 16,
123
Serhiy Storchaka1211c9a2018-01-20 16:42:44 +0200124 FRAME_SIZE_MIN = 4,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100125 FRAME_SIZE_TARGET = 64 * 1024,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100126 FRAME_HEADER_SIZE = 9
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000127};
128
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800129/*************************************************************************/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000130
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800131/* State of the pickle module, per PEP 3121. */
132typedef struct {
133 /* Exception classes for pickle. */
134 PyObject *PickleError;
135 PyObject *PicklingError;
136 PyObject *UnpicklingError;
Larry Hastings61272b72014-01-07 12:41:53 -0800137
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800138 /* copyreg.dispatch_table, {type_object: pickling_function} */
139 PyObject *dispatch_table;
Antoine Pitroud9dfaa92009-06-04 20:32:06 +0000140
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800141 /* For the extension opcodes EXT1, EXT2 and EXT4. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000142
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800143 /* copyreg._extension_registry, {(module_name, function_name): code} */
144 PyObject *extension_registry;
145 /* copyreg._extension_cache, {code: object} */
146 PyObject *extension_cache;
147 /* copyreg._inverted_registry, {code: (module_name, function_name)} */
148 PyObject *inverted_registry;
149
150 /* Import mappings for compatibility with Python 2.x */
151
152 /* _compat_pickle.NAME_MAPPING,
153 {(oldmodule, oldname): (newmodule, newname)} */
154 PyObject *name_mapping_2to3;
155 /* _compat_pickle.IMPORT_MAPPING, {oldmodule: newmodule} */
156 PyObject *import_mapping_2to3;
157 /* Same, but with REVERSE_NAME_MAPPING / REVERSE_IMPORT_MAPPING */
158 PyObject *name_mapping_3to2;
159 PyObject *import_mapping_3to2;
160
161 /* codecs.encode, used for saving bytes in older protocols */
162 PyObject *codecs_encode;
Serhiy Storchaka58e41342015-03-31 14:07:24 +0300163 /* builtins.getattr, used for saving nested names with protocol < 4 */
164 PyObject *getattr;
Serhiy Storchaka0d554d72015-10-10 22:42:18 +0300165 /* functools.partial, used for implementing __newobj_ex__ with protocols
166 2 and 3 */
167 PyObject *partial;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800168} PickleState;
169
170/* Forward declaration of the _pickle module definition. */
171static struct PyModuleDef _picklemodule;
172
173/* Given a module object, get its per-module state. */
174static PickleState *
175_Pickle_GetState(PyObject *module)
176{
177 return (PickleState *)PyModule_GetState(module);
178}
179
180/* Find the module instance imported in the currently running sub-interpreter
181 and get its state. */
182static PickleState *
183_Pickle_GetGlobalState(void)
184{
185 return _Pickle_GetState(PyState_FindModule(&_picklemodule));
186}
187
188/* Clear the given pickle module state. */
189static void
190_Pickle_ClearState(PickleState *st)
191{
192 Py_CLEAR(st->PickleError);
193 Py_CLEAR(st->PicklingError);
194 Py_CLEAR(st->UnpicklingError);
195 Py_CLEAR(st->dispatch_table);
196 Py_CLEAR(st->extension_registry);
197 Py_CLEAR(st->extension_cache);
198 Py_CLEAR(st->inverted_registry);
199 Py_CLEAR(st->name_mapping_2to3);
200 Py_CLEAR(st->import_mapping_2to3);
201 Py_CLEAR(st->name_mapping_3to2);
202 Py_CLEAR(st->import_mapping_3to2);
203 Py_CLEAR(st->codecs_encode);
Serhiy Storchaka58e41342015-03-31 14:07:24 +0300204 Py_CLEAR(st->getattr);
Victor Stinner9ba97df2015-11-17 12:15:07 +0100205 Py_CLEAR(st->partial);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800206}
207
208/* Initialize the given pickle module state. */
209static int
210_Pickle_InitState(PickleState *st)
211{
Serhiy Storchaka58e41342015-03-31 14:07:24 +0300212 PyObject *builtins;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800213 PyObject *copyreg = NULL;
214 PyObject *compat_pickle = NULL;
215 PyObject *codecs = NULL;
Serhiy Storchaka0d554d72015-10-10 22:42:18 +0300216 PyObject *functools = NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800217
Serhiy Storchaka58e41342015-03-31 14:07:24 +0300218 builtins = PyEval_GetBuiltins();
219 if (builtins == NULL)
220 goto error;
221 st->getattr = PyDict_GetItemString(builtins, "getattr");
222 if (st->getattr == NULL)
223 goto error;
224 Py_INCREF(st->getattr);
225
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800226 copyreg = PyImport_ImportModule("copyreg");
227 if (!copyreg)
228 goto error;
229 st->dispatch_table = PyObject_GetAttrString(copyreg, "dispatch_table");
230 if (!st->dispatch_table)
231 goto error;
232 if (!PyDict_CheckExact(st->dispatch_table)) {
233 PyErr_Format(PyExc_RuntimeError,
234 "copyreg.dispatch_table should be a dict, not %.200s",
235 Py_TYPE(st->dispatch_table)->tp_name);
236 goto error;
237 }
238 st->extension_registry = \
239 PyObject_GetAttrString(copyreg, "_extension_registry");
240 if (!st->extension_registry)
241 goto error;
242 if (!PyDict_CheckExact(st->extension_registry)) {
243 PyErr_Format(PyExc_RuntimeError,
244 "copyreg._extension_registry should be a dict, "
245 "not %.200s", Py_TYPE(st->extension_registry)->tp_name);
246 goto error;
247 }
248 st->inverted_registry = \
249 PyObject_GetAttrString(copyreg, "_inverted_registry");
250 if (!st->inverted_registry)
251 goto error;
252 if (!PyDict_CheckExact(st->inverted_registry)) {
253 PyErr_Format(PyExc_RuntimeError,
254 "copyreg._inverted_registry should be a dict, "
255 "not %.200s", Py_TYPE(st->inverted_registry)->tp_name);
256 goto error;
257 }
258 st->extension_cache = PyObject_GetAttrString(copyreg, "_extension_cache");
259 if (!st->extension_cache)
260 goto error;
261 if (!PyDict_CheckExact(st->extension_cache)) {
262 PyErr_Format(PyExc_RuntimeError,
263 "copyreg._extension_cache should be a dict, "
264 "not %.200s", Py_TYPE(st->extension_cache)->tp_name);
265 goto error;
266 }
267 Py_CLEAR(copyreg);
268
269 /* Load the 2.x -> 3.x stdlib module mapping tables */
270 compat_pickle = PyImport_ImportModule("_compat_pickle");
271 if (!compat_pickle)
272 goto error;
273 st->name_mapping_2to3 = \
274 PyObject_GetAttrString(compat_pickle, "NAME_MAPPING");
275 if (!st->name_mapping_2to3)
276 goto error;
277 if (!PyDict_CheckExact(st->name_mapping_2to3)) {
278 PyErr_Format(PyExc_RuntimeError,
279 "_compat_pickle.NAME_MAPPING should be a dict, not %.200s",
280 Py_TYPE(st->name_mapping_2to3)->tp_name);
281 goto error;
282 }
283 st->import_mapping_2to3 = \
284 PyObject_GetAttrString(compat_pickle, "IMPORT_MAPPING");
285 if (!st->import_mapping_2to3)
286 goto error;
287 if (!PyDict_CheckExact(st->import_mapping_2to3)) {
288 PyErr_Format(PyExc_RuntimeError,
289 "_compat_pickle.IMPORT_MAPPING should be a dict, "
290 "not %.200s", Py_TYPE(st->import_mapping_2to3)->tp_name);
291 goto error;
292 }
293 /* ... and the 3.x -> 2.x mapping tables */
294 st->name_mapping_3to2 = \
295 PyObject_GetAttrString(compat_pickle, "REVERSE_NAME_MAPPING");
296 if (!st->name_mapping_3to2)
297 goto error;
298 if (!PyDict_CheckExact(st->name_mapping_3to2)) {
299 PyErr_Format(PyExc_RuntimeError,
300 "_compat_pickle.REVERSE_NAME_MAPPING should be a dict, "
301 "not %.200s", Py_TYPE(st->name_mapping_3to2)->tp_name);
302 goto error;
303 }
304 st->import_mapping_3to2 = \
305 PyObject_GetAttrString(compat_pickle, "REVERSE_IMPORT_MAPPING");
306 if (!st->import_mapping_3to2)
307 goto error;
308 if (!PyDict_CheckExact(st->import_mapping_3to2)) {
309 PyErr_Format(PyExc_RuntimeError,
310 "_compat_pickle.REVERSE_IMPORT_MAPPING should be a dict, "
311 "not %.200s", Py_TYPE(st->import_mapping_3to2)->tp_name);
312 goto error;
313 }
314 Py_CLEAR(compat_pickle);
315
316 codecs = PyImport_ImportModule("codecs");
317 if (codecs == NULL)
318 goto error;
319 st->codecs_encode = PyObject_GetAttrString(codecs, "encode");
320 if (st->codecs_encode == NULL) {
321 goto error;
322 }
323 if (!PyCallable_Check(st->codecs_encode)) {
324 PyErr_Format(PyExc_RuntimeError,
325 "codecs.encode should be a callable, not %.200s",
326 Py_TYPE(st->codecs_encode)->tp_name);
327 goto error;
328 }
329 Py_CLEAR(codecs);
330
Serhiy Storchaka0d554d72015-10-10 22:42:18 +0300331 functools = PyImport_ImportModule("functools");
332 if (!functools)
333 goto error;
334 st->partial = PyObject_GetAttrString(functools, "partial");
335 if (!st->partial)
336 goto error;
337 Py_CLEAR(functools);
338
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800339 return 0;
340
341 error:
342 Py_CLEAR(copyreg);
343 Py_CLEAR(compat_pickle);
344 Py_CLEAR(codecs);
Serhiy Storchaka0d554d72015-10-10 22:42:18 +0300345 Py_CLEAR(functools);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800346 _Pickle_ClearState(st);
347 return -1;
348}
349
350/* Helper for calling a function with a single argument quickly.
351
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800352 This function steals the reference of the given argument. */
353static PyObject *
354_Pickle_FastCall(PyObject *func, PyObject *obj)
355{
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800356 PyObject *result;
357
Victor Stinner7bfb42d2016-12-05 17:04:32 +0100358 result = PyObject_CallFunctionObjArgs(func, obj, NULL);
Victor Stinner75210692016-08-19 18:59:15 +0200359 Py_DECREF(obj);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800360 return result;
361}
362
363/*************************************************************************/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000364
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200365/* Retrieve and deconstruct a method for avoiding a reference cycle
366 (pickler -> bound method of pickler -> pickler) */
367static int
368init_method_ref(PyObject *self, _Py_Identifier *name,
369 PyObject **method_func, PyObject **method_self)
370{
371 PyObject *func, *func2;
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200372 int ret;
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200373
374 /* *method_func and *method_self should be consistent. All refcount decrements
375 should be occurred after setting *method_self and *method_func. */
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200376 ret = _PyObject_LookupAttrId(self, name, &func);
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200377 if (func == NULL) {
378 *method_self = NULL;
379 Py_CLEAR(*method_func);
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200380 return ret;
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200381 }
382
383 if (PyMethod_Check(func) && PyMethod_GET_SELF(func) == self) {
384 /* Deconstruct a bound Python method */
385 func2 = PyMethod_GET_FUNCTION(func);
386 Py_INCREF(func2);
387 *method_self = self; /* borrowed */
388 Py_XSETREF(*method_func, func2);
389 Py_DECREF(func);
390 return 0;
391 }
392 else {
393 *method_self = NULL;
394 Py_XSETREF(*method_func, func);
395 return 0;
396 }
397}
398
399/* Bind a method if it was deconstructed */
400static PyObject *
401reconstruct_method(PyObject *func, PyObject *self)
402{
403 if (self) {
404 return PyMethod_New(func, self);
405 }
406 else {
407 Py_INCREF(func);
408 return func;
409 }
410}
411
412static PyObject *
413call_method(PyObject *func, PyObject *self, PyObject *obj)
414{
415 if (self) {
416 return PyObject_CallFunctionObjArgs(func, self, obj, NULL);
417 }
418 else {
419 return PyObject_CallFunctionObjArgs(func, obj, NULL);
420 }
421}
422
423/*************************************************************************/
424
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000425/* Internal data type used as the unpickling stack. */
426typedef struct {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000427 PyObject_VAR_HEAD
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000428 PyObject **data;
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200429 int mark_set; /* is MARK set? */
430 Py_ssize_t fence; /* position of top MARK or 0 */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000431 Py_ssize_t allocated; /* number of slots in data allocated */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000432} Pdata;
433
434static void
435Pdata_dealloc(Pdata *self)
436{
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200437 Py_ssize_t i = Py_SIZE(self);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000438 while (--i >= 0) {
439 Py_DECREF(self->data[i]);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000440 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000441 PyMem_FREE(self->data);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000442 PyObject_Del(self);
443}
444
445static PyTypeObject Pdata_Type = {
446 PyVarObject_HEAD_INIT(NULL, 0)
447 "_pickle.Pdata", /*tp_name*/
448 sizeof(Pdata), /*tp_basicsize*/
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +0200449 sizeof(PyObject *), /*tp_itemsize*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000450 (destructor)Pdata_dealloc, /*tp_dealloc*/
451};
452
453static PyObject *
454Pdata_New(void)
455{
456 Pdata *self;
457
458 if (!(self = PyObject_New(Pdata, &Pdata_Type)))
459 return NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000460 Py_SIZE(self) = 0;
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200461 self->mark_set = 0;
462 self->fence = 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000463 self->allocated = 8;
464 self->data = PyMem_MALLOC(self->allocated * sizeof(PyObject *));
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000465 if (self->data)
466 return (PyObject *)self;
467 Py_DECREF(self);
468 return PyErr_NoMemory();
469}
470
471
472/* Retain only the initial clearto items. If clearto >= the current
473 * number of items, this is a (non-erroneous) NOP.
474 */
475static int
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200476Pdata_clear(Pdata *self, Py_ssize_t clearto)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000477{
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200478 Py_ssize_t i = Py_SIZE(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000479
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200480 assert(clearto >= self->fence);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000481 if (clearto >= i)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000482 return 0;
483
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000484 while (--i >= clearto) {
485 Py_CLEAR(self->data[i]);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000486 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000487 Py_SIZE(self) = clearto;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000488 return 0;
489}
490
491static int
492Pdata_grow(Pdata *self)
493{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000494 PyObject **data = self->data;
Victor Stinnerf13c46c2014-08-17 21:05:55 +0200495 size_t allocated = (size_t)self->allocated;
496 size_t new_allocated;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000497
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000498 new_allocated = (allocated >> 3) + 6;
499 /* check for integer overflow */
Victor Stinnerf13c46c2014-08-17 21:05:55 +0200500 if (new_allocated > (size_t)PY_SSIZE_T_MAX - allocated)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000501 goto nomemory;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000502 new_allocated += allocated;
Benjamin Peterson59b08c12015-06-27 13:41:33 -0500503 PyMem_RESIZE(data, PyObject *, new_allocated);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000504 if (data == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000505 goto nomemory;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000506
507 self->data = data;
Victor Stinnerf13c46c2014-08-17 21:05:55 +0200508 self->allocated = (Py_ssize_t)new_allocated;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000509 return 0;
510
511 nomemory:
512 PyErr_NoMemory();
513 return -1;
514}
515
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200516static int
517Pdata_stack_underflow(Pdata *self)
518{
519 PickleState *st = _Pickle_GetGlobalState();
520 PyErr_SetString(st->UnpicklingError,
521 self->mark_set ?
522 "unexpected MARK found" :
523 "unpickling stack underflow");
524 return -1;
525}
526
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000527/* D is a Pdata*. Pop the topmost element and store it into V, which
528 * must be an lvalue holding PyObject*. On stack underflow, UnpicklingError
529 * is raised and V is set to NULL.
530 */
531static PyObject *
532Pdata_pop(Pdata *self)
533{
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200534 if (Py_SIZE(self) <= self->fence) {
535 Pdata_stack_underflow(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000536 return NULL;
537 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000538 return self->data[--Py_SIZE(self)];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000539}
540#define PDATA_POP(D, V) do { (V) = Pdata_pop((D)); } while (0)
541
542static int
543Pdata_push(Pdata *self, PyObject *obj)
544{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000545 if (Py_SIZE(self) == self->allocated && Pdata_grow(self) < 0) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000546 return -1;
547 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000548 self->data[Py_SIZE(self)++] = obj;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000549 return 0;
550}
551
552/* Push an object on stack, transferring its ownership to the stack. */
553#define PDATA_PUSH(D, O, ER) do { \
554 if (Pdata_push((D), (O)) < 0) return (ER); } while(0)
555
556/* Push an object on stack, adding a new reference to the object. */
557#define PDATA_APPEND(D, O, ER) do { \
558 Py_INCREF((O)); \
559 if (Pdata_push((D), (O)) < 0) return (ER); } while(0)
560
561static PyObject *
562Pdata_poptuple(Pdata *self, Py_ssize_t start)
563{
564 PyObject *tuple;
565 Py_ssize_t len, i, j;
566
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200567 if (start < self->fence) {
568 Pdata_stack_underflow(self);
569 return NULL;
570 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000571 len = Py_SIZE(self) - start;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000572 tuple = PyTuple_New(len);
573 if (tuple == NULL)
574 return NULL;
575 for (i = start, j = 0; j < len; i++, j++)
576 PyTuple_SET_ITEM(tuple, j, self->data[i]);
577
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000578 Py_SIZE(self) = start;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000579 return tuple;
580}
581
582static PyObject *
583Pdata_poplist(Pdata *self, Py_ssize_t start)
584{
585 PyObject *list;
586 Py_ssize_t len, i, j;
587
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000588 len = Py_SIZE(self) - start;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000589 list = PyList_New(len);
590 if (list == NULL)
591 return NULL;
592 for (i = start, j = 0; j < len; i++, j++)
593 PyList_SET_ITEM(list, j, self->data[i]);
594
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000595 Py_SIZE(self) = start;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000596 return list;
597}
598
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000599typedef struct {
600 PyObject *me_key;
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200601 Py_ssize_t me_value;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000602} PyMemoEntry;
603
604typedef struct {
605 Py_ssize_t mt_mask;
606 Py_ssize_t mt_used;
607 Py_ssize_t mt_allocated;
608 PyMemoEntry *mt_table;
609} PyMemoTable;
610
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000611typedef struct PicklerObject {
612 PyObject_HEAD
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000613 PyMemoTable *memo; /* Memo table, keep track of the seen
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000614 objects to support self-referential objects
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000615 pickling. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000616 PyObject *pers_func; /* persistent_id() method, can be NULL */
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200617 PyObject *pers_func_self; /* borrowed reference to self if pers_func
618 is an unbound method, NULL otherwise */
Antoine Pitrou8d3c2902012-03-04 18:31:48 +0100619 PyObject *dispatch_table; /* private dispatch_table, can be NULL */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000620
621 PyObject *write; /* write() method of the output stream. */
622 PyObject *output_buffer; /* Write into a local bytearray buffer before
623 flushing to the stream. */
624 Py_ssize_t output_len; /* Length of output_buffer. */
625 Py_ssize_t max_output_len; /* Allocation size of output_buffer. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000626 int proto; /* Pickle protocol number, >= 0 */
627 int bin; /* Boolean, true if proto > 0 */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100628 int framing; /* True when framing is enabled, proto >= 4 */
629 Py_ssize_t frame_start; /* Position in output_buffer where the
Martin Pantera90a4a92016-05-30 04:04:50 +0000630 current frame begins. -1 if there
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100631 is no frame currently open. */
632
633 Py_ssize_t buf_size; /* Size of the current buffered pickle data */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000634 int fast; /* Enable fast mode if set to a true value.
635 The fast mode disable the usage of memo,
636 therefore speeding the pickling process by
637 not generating superfluous PUT opcodes. It
638 should not be used if with self-referential
639 objects. */
640 int fast_nesting;
Antoine Pitroud9dfaa92009-06-04 20:32:06 +0000641 int fix_imports; /* Indicate whether Pickler should fix
642 the name of globals for Python 2.x. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000643 PyObject *fast_memo;
644} PicklerObject;
645
646typedef struct UnpicklerObject {
647 PyObject_HEAD
648 Pdata *stack; /* Pickle data stack, store unpickled objects. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000649
650 /* The unpickler memo is just an array of PyObject *s. Using a dict
651 is unnecessary, since the keys are contiguous ints. */
652 PyObject **memo;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100653 Py_ssize_t memo_size; /* Capacity of the memo array */
654 Py_ssize_t memo_len; /* Number of objects in the memo */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000655
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000656 PyObject *pers_func; /* persistent_load() method, can be NULL. */
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200657 PyObject *pers_func_self; /* borrowed reference to self if pers_func
658 is an unbound method, NULL otherwise */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000659
660 Py_buffer buffer;
661 char *input_buffer;
662 char *input_line;
663 Py_ssize_t input_len;
664 Py_ssize_t next_read_idx;
Antoine Pitrou04248a82010-10-12 20:51:21 +0000665 Py_ssize_t prefetched_idx; /* index of first prefetched byte */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100666
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000667 PyObject *read; /* read() method of the input stream. */
668 PyObject *readline; /* readline() method of the input stream. */
Antoine Pitrou04248a82010-10-12 20:51:21 +0000669 PyObject *peek; /* peek() method of the input stream, or NULL */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000670
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000671 char *encoding; /* Name of the encoding to be used for
672 decoding strings pickled using Python
673 2.x. The default value is "ASCII" */
674 char *errors; /* Name of errors handling scheme to used when
675 decoding strings. The default value is
676 "strict". */
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -0500677 Py_ssize_t *marks; /* Mark stack, used for unpickling container
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000678 objects. */
679 Py_ssize_t num_marks; /* Number of marks in the mark stack. */
680 Py_ssize_t marks_size; /* Current allocated size of the mark stack. */
Antoine Pitroud9dfaa92009-06-04 20:32:06 +0000681 int proto; /* Protocol of the pickle loaded. */
682 int fix_imports; /* Indicate whether Unpickler should fix
683 the name of globals pickled by Python 2.x. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000684} UnpicklerObject;
685
Serhiy Storchaka3c1f0f12014-01-27 10:34:22 +0200686typedef struct {
687 PyObject_HEAD
688 PicklerObject *pickler; /* Pickler whose memo table we're proxying. */
689} PicklerMemoProxyObject;
690
691typedef struct {
692 PyObject_HEAD
693 UnpicklerObject *unpickler;
694} UnpicklerMemoProxyObject;
695
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000696/* Forward declarations */
697static int save(PicklerObject *, PyObject *, int);
698static int save_reduce(PicklerObject *, PyObject *, PyObject *);
699static PyTypeObject Pickler_Type;
700static PyTypeObject Unpickler_Type;
701
Serhiy Storchaka3c1f0f12014-01-27 10:34:22 +0200702#include "clinic/_pickle.c.h"
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000703
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000704/*************************************************************************
Serhiy Storchaka95949422013-08-27 19:40:23 +0300705 A custom hashtable mapping void* to Python ints. This is used by the pickler
706 for memoization. Using a custom hashtable rather than PyDict allows us to skip
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000707 a bunch of unnecessary object creation. This makes a huge performance
708 difference. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000709
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000710#define MT_MINSIZE 8
711#define PERTURB_SHIFT 5
712
713
714static PyMemoTable *
715PyMemoTable_New(void)
716{
717 PyMemoTable *memo = PyMem_MALLOC(sizeof(PyMemoTable));
718 if (memo == NULL) {
719 PyErr_NoMemory();
720 return NULL;
721 }
722
723 memo->mt_used = 0;
724 memo->mt_allocated = MT_MINSIZE;
725 memo->mt_mask = MT_MINSIZE - 1;
726 memo->mt_table = PyMem_MALLOC(MT_MINSIZE * sizeof(PyMemoEntry));
727 if (memo->mt_table == NULL) {
728 PyMem_FREE(memo);
729 PyErr_NoMemory();
730 return NULL;
731 }
732 memset(memo->mt_table, 0, MT_MINSIZE * sizeof(PyMemoEntry));
733
734 return memo;
735}
736
737static PyMemoTable *
738PyMemoTable_Copy(PyMemoTable *self)
739{
740 Py_ssize_t i;
741 PyMemoTable *new = PyMemoTable_New();
742 if (new == NULL)
743 return NULL;
744
745 new->mt_used = self->mt_used;
746 new->mt_allocated = self->mt_allocated;
747 new->mt_mask = self->mt_mask;
748 /* The table we get from _New() is probably smaller than we wanted.
749 Free it and allocate one that's the right size. */
750 PyMem_FREE(new->mt_table);
Benjamin Peterson59b08c12015-06-27 13:41:33 -0500751 new->mt_table = PyMem_NEW(PyMemoEntry, self->mt_allocated);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000752 if (new->mt_table == NULL) {
753 PyMem_FREE(new);
Victor Stinner42024562013-07-12 00:53:57 +0200754 PyErr_NoMemory();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000755 return NULL;
756 }
757 for (i = 0; i < self->mt_allocated; i++) {
758 Py_XINCREF(self->mt_table[i].me_key);
759 }
760 memcpy(new->mt_table, self->mt_table,
761 sizeof(PyMemoEntry) * self->mt_allocated);
762
763 return new;
764}
765
766static Py_ssize_t
767PyMemoTable_Size(PyMemoTable *self)
768{
769 return self->mt_used;
770}
771
772static int
773PyMemoTable_Clear(PyMemoTable *self)
774{
775 Py_ssize_t i = self->mt_allocated;
776
777 while (--i >= 0) {
778 Py_XDECREF(self->mt_table[i].me_key);
779 }
780 self->mt_used = 0;
781 memset(self->mt_table, 0, self->mt_allocated * sizeof(PyMemoEntry));
782 return 0;
783}
784
785static void
786PyMemoTable_Del(PyMemoTable *self)
787{
788 if (self == NULL)
789 return;
790 PyMemoTable_Clear(self);
791
792 PyMem_FREE(self->mt_table);
793 PyMem_FREE(self);
794}
795
796/* Since entries cannot be deleted from this hashtable, _PyMemoTable_Lookup()
797 can be considerably simpler than dictobject.c's lookdict(). */
798static PyMemoEntry *
799_PyMemoTable_Lookup(PyMemoTable *self, PyObject *key)
800{
801 size_t i;
802 size_t perturb;
803 size_t mask = (size_t)self->mt_mask;
804 PyMemoEntry *table = self->mt_table;
805 PyMemoEntry *entry;
Benjamin Peterson8f67d082010-10-17 20:54:53 +0000806 Py_hash_t hash = (Py_hash_t)key >> 3;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000807
808 i = hash & mask;
809 entry = &table[i];
810 if (entry->me_key == NULL || entry->me_key == key)
811 return entry;
812
813 for (perturb = hash; ; perturb >>= PERTURB_SHIFT) {
814 i = (i << 2) + i + perturb + 1;
815 entry = &table[i & mask];
816 if (entry->me_key == NULL || entry->me_key == key)
817 return entry;
818 }
Barry Warsawb2e57942017-09-14 18:13:16 -0700819 Py_UNREACHABLE();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000820}
821
822/* Returns -1 on failure, 0 on success. */
823static int
824_PyMemoTable_ResizeTable(PyMemoTable *self, Py_ssize_t min_size)
825{
826 PyMemoEntry *oldtable = NULL;
827 PyMemoEntry *oldentry, *newentry;
828 Py_ssize_t new_size = MT_MINSIZE;
829 Py_ssize_t to_process;
830
831 assert(min_size > 0);
832
833 /* Find the smallest valid table size >= min_size. */
834 while (new_size < min_size && new_size > 0)
835 new_size <<= 1;
836 if (new_size <= 0) {
837 PyErr_NoMemory();
838 return -1;
839 }
840 /* new_size needs to be a power of two. */
841 assert((new_size & (new_size - 1)) == 0);
842
843 /* Allocate new table. */
844 oldtable = self->mt_table;
Benjamin Peterson59b08c12015-06-27 13:41:33 -0500845 self->mt_table = PyMem_NEW(PyMemoEntry, new_size);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000846 if (self->mt_table == NULL) {
Victor Stinner8ca72e22013-07-12 00:53:26 +0200847 self->mt_table = oldtable;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000848 PyErr_NoMemory();
849 return -1;
850 }
851 self->mt_allocated = new_size;
852 self->mt_mask = new_size - 1;
853 memset(self->mt_table, 0, sizeof(PyMemoEntry) * new_size);
854
855 /* Copy entries from the old table. */
856 to_process = self->mt_used;
857 for (oldentry = oldtable; to_process > 0; oldentry++) {
858 if (oldentry->me_key != NULL) {
859 to_process--;
860 /* newentry is a pointer to a chunk of the new
861 mt_table, so we're setting the key:value pair
862 in-place. */
863 newentry = _PyMemoTable_Lookup(self, oldentry->me_key);
864 newentry->me_key = oldentry->me_key;
865 newentry->me_value = oldentry->me_value;
866 }
867 }
868
869 /* Deallocate the old table. */
870 PyMem_FREE(oldtable);
871 return 0;
872}
873
874/* Returns NULL on failure, a pointer to the value otherwise. */
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200875static Py_ssize_t *
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000876PyMemoTable_Get(PyMemoTable *self, PyObject *key)
877{
878 PyMemoEntry *entry = _PyMemoTable_Lookup(self, key);
879 if (entry->me_key == NULL)
880 return NULL;
881 return &entry->me_value;
882}
883
884/* Returns -1 on failure, 0 on success. */
885static int
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200886PyMemoTable_Set(PyMemoTable *self, PyObject *key, Py_ssize_t value)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000887{
888 PyMemoEntry *entry;
889
890 assert(key != NULL);
891
892 entry = _PyMemoTable_Lookup(self, key);
893 if (entry->me_key != NULL) {
894 entry->me_value = value;
895 return 0;
896 }
897 Py_INCREF(key);
898 entry->me_key = key;
899 entry->me_value = value;
900 self->mt_used++;
901
902 /* If we added a key, we can safely resize. Otherwise just return!
903 * If used >= 2/3 size, adjust size. Normally, this quaduples the size.
904 *
905 * Quadrupling the size improves average table sparseness
906 * (reducing collisions) at the cost of some memory. It also halves
907 * the number of expensive resize operations in a growing memo table.
908 *
909 * Very large memo tables (over 50K items) use doubling instead.
910 * This may help applications with severe memory constraints.
911 */
912 if (!(self->mt_used * 3 >= (self->mt_mask + 1) * 2))
913 return 0;
914 return _PyMemoTable_ResizeTable(self,
915 (self->mt_used > 50000 ? 2 : 4) * self->mt_used);
916}
917
918#undef MT_MINSIZE
919#undef PERTURB_SHIFT
920
921/*************************************************************************/
922
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000923
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000924static int
925_Pickler_ClearBuffer(PicklerObject *self)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000926{
Serhiy Storchaka48842712016-04-06 09:45:48 +0300927 Py_XSETREF(self->output_buffer,
Serhiy Storchaka4a1e70f2015-12-27 12:36:18 +0200928 PyBytes_FromStringAndSize(NULL, self->max_output_len));
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000929 if (self->output_buffer == NULL)
Amaury Forgeot d'Arc87eee632008-10-17 20:15:53 +0000930 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000931 self->output_len = 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100932 self->frame_start = -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000933 return 0;
934}
Amaury Forgeot d'Arc87eee632008-10-17 20:15:53 +0000935
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100936static void
Antoine Pitrou8f2ee6e2013-11-23 21:05:08 +0100937_write_size64(char *out, size_t value)
938{
Victor Stinnerf13c46c2014-08-17 21:05:55 +0200939 size_t i;
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -0800940
Serhiy Storchakafad85aa2015-11-07 15:42:38 +0200941 Py_BUILD_ASSERT(sizeof(size_t) <= 8);
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -0800942
943 for (i = 0; i < sizeof(size_t); i++) {
944 out[i] = (unsigned char)((value >> (8 * i)) & 0xff);
945 }
946 for (i = sizeof(size_t); i < 8; i++) {
947 out[i] = 0;
Alexandre Vassalottided929b2013-11-24 22:41:13 -0800948 }
Antoine Pitrou8f2ee6e2013-11-23 21:05:08 +0100949}
950
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100951static int
952_Pickler_CommitFrame(PicklerObject *self)
953{
954 size_t frame_len;
955 char *qdata;
956
957 if (!self->framing || self->frame_start == -1)
958 return 0;
959 frame_len = self->output_len - self->frame_start - FRAME_HEADER_SIZE;
960 qdata = PyBytes_AS_STRING(self->output_buffer) + self->frame_start;
Serhiy Storchaka1211c9a2018-01-20 16:42:44 +0200961 if (frame_len >= FRAME_SIZE_MIN) {
962 qdata[0] = FRAME;
963 _write_size64(qdata + 1, frame_len);
964 }
965 else {
966 memmove(qdata, qdata + FRAME_HEADER_SIZE, frame_len);
967 self->output_len -= FRAME_HEADER_SIZE;
968 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100969 self->frame_start = -1;
970 return 0;
971}
972
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000973static PyObject *
974_Pickler_GetString(PicklerObject *self)
975{
976 PyObject *output_buffer = self->output_buffer;
977
978 assert(self->output_buffer != NULL);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100979
980 if (_Pickler_CommitFrame(self))
981 return NULL;
982
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000983 self->output_buffer = NULL;
984 /* Resize down to exact size */
985 if (_PyBytes_Resize(&output_buffer, self->output_len) < 0)
986 return NULL;
987 return output_buffer;
988}
989
990static int
991_Pickler_FlushToFile(PicklerObject *self)
992{
993 PyObject *output, *result;
994
995 assert(self->write != NULL);
996
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100997 /* This will commit the frame first */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000998 output = _Pickler_GetString(self);
999 if (output == NULL)
1000 return -1;
1001
Alexandre Vassalotti20c28c12013-11-27 02:26:54 -08001002 result = _Pickle_FastCall(self->write, output);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001003 Py_XDECREF(result);
1004 return (result == NULL) ? -1 : 0;
1005}
1006
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01001007static int
1008_Pickler_OpcodeBoundary(PicklerObject *self)
1009{
1010 Py_ssize_t frame_len;
1011
1012 if (!self->framing || self->frame_start == -1) {
1013 return 0;
1014 }
1015 frame_len = self->output_len - self->frame_start - FRAME_HEADER_SIZE;
1016 if (frame_len >= FRAME_SIZE_TARGET) {
1017 if(_Pickler_CommitFrame(self)) {
1018 return -1;
1019 }
Leo Ariasc3d95082018-02-03 18:36:10 -06001020 /* Flush the content of the committed frame to the underlying
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01001021 * file and reuse the pickler buffer for the next frame so as
1022 * to limit memory usage when dumping large complex objects to
1023 * a file.
1024 *
1025 * self->write is NULL when called via dumps.
1026 */
1027 if (self->write != NULL) {
1028 if (_Pickler_FlushToFile(self) < 0) {
1029 return -1;
1030 }
1031 if (_Pickler_ClearBuffer(self) < 0) {
1032 return -1;
1033 }
1034 }
1035 }
1036 return 0;
1037}
1038
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001039static Py_ssize_t
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001040_Pickler_Write(PicklerObject *self, const char *s, Py_ssize_t data_len)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001041{
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001042 Py_ssize_t i, n, required;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001043 char *buffer;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001044 int need_new_frame;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001045
1046 assert(s != NULL);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001047 need_new_frame = (self->framing && self->frame_start == -1);
1048
1049 if (need_new_frame)
1050 n = data_len + FRAME_HEADER_SIZE;
1051 else
1052 n = data_len;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001053
1054 required = self->output_len + n;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001055 if (required > self->max_output_len) {
1056 /* Make place in buffer for the pickle chunk */
1057 if (self->output_len >= PY_SSIZE_T_MAX / 2 - n) {
1058 PyErr_NoMemory();
1059 return -1;
1060 }
1061 self->max_output_len = (self->output_len + n) / 2 * 3;
1062 if (_PyBytes_Resize(&self->output_buffer, self->max_output_len) < 0)
1063 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001064 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001065 buffer = PyBytes_AS_STRING(self->output_buffer);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001066 if (need_new_frame) {
1067 /* Setup new frame */
1068 Py_ssize_t frame_start = self->output_len;
1069 self->frame_start = frame_start;
1070 for (i = 0; i < FRAME_HEADER_SIZE; i++) {
1071 /* Write an invalid value, for debugging */
1072 buffer[frame_start + i] = 0xFE;
1073 }
1074 self->output_len += FRAME_HEADER_SIZE;
1075 }
1076 if (data_len < 8) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001077 /* This is faster than memcpy when the string is short. */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001078 for (i = 0; i < data_len; i++) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001079 buffer[self->output_len + i] = s[i];
1080 }
1081 }
1082 else {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001083 memcpy(buffer + self->output_len, s, data_len);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001084 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001085 self->output_len += data_len;
1086 return data_len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001087}
1088
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001089static PicklerObject *
1090_Pickler_New(void)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001091{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001092 PicklerObject *self;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001093
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001094 self = PyObject_GC_New(PicklerObject, &Pickler_Type);
1095 if (self == NULL)
1096 return NULL;
1097
1098 self->pers_func = NULL;
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01001099 self->dispatch_table = NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001100 self->write = NULL;
1101 self->proto = 0;
1102 self->bin = 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001103 self->framing = 0;
1104 self->frame_start = -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001105 self->fast = 0;
1106 self->fast_nesting = 0;
1107 self->fix_imports = 0;
1108 self->fast_memo = NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001109 self->max_output_len = WRITE_BUF_SIZE;
1110 self->output_len = 0;
Victor Stinner68c8ea22013-07-11 22:56:25 +02001111
1112 self->memo = PyMemoTable_New();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001113 self->output_buffer = PyBytes_FromStringAndSize(NULL,
1114 self->max_output_len);
Victor Stinner68c8ea22013-07-11 22:56:25 +02001115
1116 if (self->memo == NULL || self->output_buffer == NULL) {
Victor Stinnerc31df042013-07-12 00:08:59 +02001117 Py_DECREF(self);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001118 return NULL;
1119 }
1120 return self;
1121}
1122
1123static int
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001124_Pickler_SetProtocol(PicklerObject *self, PyObject *protocol, int fix_imports)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001125{
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001126 long proto;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001127
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001128 if (protocol == NULL || protocol == Py_None) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001129 proto = DEFAULT_PROTOCOL;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001130 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001131 else {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001132 proto = PyLong_AsLong(protocol);
1133 if (proto < 0) {
1134 if (proto == -1 && PyErr_Occurred())
1135 return -1;
1136 proto = HIGHEST_PROTOCOL;
1137 }
1138 else if (proto > HIGHEST_PROTOCOL) {
1139 PyErr_Format(PyExc_ValueError, "pickle protocol must be <= %d",
1140 HIGHEST_PROTOCOL);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001141 return -1;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001142 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001143 }
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001144 self->proto = (int)proto;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001145 self->bin = proto > 0;
1146 self->fix_imports = fix_imports && proto < 3;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001147 return 0;
1148}
1149
1150/* Returns -1 (with an exception set) on failure, 0 on success. This may
1151 be called once on a freshly created Pickler. */
1152static int
1153_Pickler_SetOutputStream(PicklerObject *self, PyObject *file)
1154{
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001155 _Py_IDENTIFIER(write);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001156 assert(file != NULL);
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001157 if (_PyObject_LookupAttrId(file, &PyId_write, &self->write) < 0) {
1158 return -1;
1159 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001160 if (self->write == NULL) {
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001161 PyErr_SetString(PyExc_TypeError,
1162 "file must have a 'write' attribute");
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001163 return -1;
1164 }
1165
1166 return 0;
1167}
1168
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001169/* Returns the size of the input on success, -1 on failure. This takes its
1170 own reference to `input`. */
1171static Py_ssize_t
1172_Unpickler_SetStringInput(UnpicklerObject *self, PyObject *input)
1173{
1174 if (self->buffer.buf != NULL)
1175 PyBuffer_Release(&self->buffer);
1176 if (PyObject_GetBuffer(input, &self->buffer, PyBUF_CONTIG_RO) < 0)
1177 return -1;
1178 self->input_buffer = self->buffer.buf;
1179 self->input_len = self->buffer.len;
1180 self->next_read_idx = 0;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001181 self->prefetched_idx = self->input_len;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001182 return self->input_len;
1183}
1184
Antoine Pitrou04248a82010-10-12 20:51:21 +00001185static int
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001186bad_readline(void)
1187{
1188 PickleState *st = _Pickle_GetGlobalState();
1189 PyErr_SetString(st->UnpicklingError, "pickle data was truncated");
1190 return -1;
1191}
1192
1193static int
Antoine Pitrou04248a82010-10-12 20:51:21 +00001194_Unpickler_SkipConsumed(UnpicklerObject *self)
1195{
Victor Stinnerb43ad1d2013-10-31 13:38:42 +01001196 Py_ssize_t consumed;
1197 PyObject *r;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001198
Victor Stinnerb43ad1d2013-10-31 13:38:42 +01001199 consumed = self->next_read_idx - self->prefetched_idx;
1200 if (consumed <= 0)
1201 return 0;
1202
1203 assert(self->peek); /* otherwise we did something wrong */
Martin Panter6245cb32016-04-15 02:14:19 +00001204 /* This makes a useless copy... */
Victor Stinnerb43ad1d2013-10-31 13:38:42 +01001205 r = PyObject_CallFunction(self->read, "n", consumed);
1206 if (r == NULL)
1207 return -1;
1208 Py_DECREF(r);
1209
1210 self->prefetched_idx = self->next_read_idx;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001211 return 0;
1212}
1213
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001214static const Py_ssize_t READ_WHOLE_LINE = -1;
1215
1216/* If reading from a file, we need to only pull the bytes we need, since there
1217 may be multiple pickle objects arranged contiguously in the same input
1218 buffer.
1219
1220 If `n` is READ_WHOLE_LINE, read a whole line. Otherwise, read up to `n`
1221 bytes from the input stream/buffer.
1222
1223 Update the unpickler's input buffer with the newly-read data. Returns -1 on
1224 failure; on success, returns the number of bytes read from the file.
1225
1226 On success, self->input_len will be 0; this is intentional so that when
1227 unpickling from a file, the "we've run out of data" code paths will trigger,
1228 causing the Unpickler to go back to the file for more data. Use the returned
1229 size to tell you how much data you can process. */
1230static Py_ssize_t
1231_Unpickler_ReadFromFile(UnpicklerObject *self, Py_ssize_t n)
1232{
1233 PyObject *data;
Serhiy Storchaka6fe39b72013-11-30 23:15:38 +02001234 Py_ssize_t read_size;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001235
1236 assert(self->read != NULL);
Victor Stinner121aab42011-09-29 23:40:53 +02001237
Antoine Pitrou04248a82010-10-12 20:51:21 +00001238 if (_Unpickler_SkipConsumed(self) < 0)
1239 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001240
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08001241 if (n == READ_WHOLE_LINE) {
Victor Stinner559bb6a2016-08-22 22:48:54 +02001242 data = _PyObject_CallNoArg(self->readline);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08001243 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001244 else {
Serhiy Storchaka6fe39b72013-11-30 23:15:38 +02001245 PyObject *len;
1246 /* Prefetch some data without advancing the file pointer, if possible */
1247 if (self->peek && n < PREFETCH) {
1248 len = PyLong_FromSsize_t(PREFETCH);
1249 if (len == NULL)
1250 return -1;
1251 data = _Pickle_FastCall(self->peek, len);
1252 if (data == NULL) {
1253 if (!PyErr_ExceptionMatches(PyExc_NotImplementedError))
1254 return -1;
1255 /* peek() is probably not supported by the given file object */
1256 PyErr_Clear();
1257 Py_CLEAR(self->peek);
1258 }
1259 else {
1260 read_size = _Unpickler_SetStringInput(self, data);
1261 Py_DECREF(data);
1262 self->prefetched_idx = 0;
1263 if (n <= read_size)
1264 return n;
1265 }
1266 }
1267 len = PyLong_FromSsize_t(n);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001268 if (len == NULL)
1269 return -1;
Alexandre Vassalotti20c28c12013-11-27 02:26:54 -08001270 data = _Pickle_FastCall(self->read, len);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001271 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001272 if (data == NULL)
1273 return -1;
1274
Serhiy Storchaka6fe39b72013-11-30 23:15:38 +02001275 read_size = _Unpickler_SetStringInput(self, data);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001276 Py_DECREF(data);
1277 return read_size;
1278}
1279
Victor Stinner19ed27e2016-05-20 11:42:37 +02001280/* Don't call it directly: use _Unpickler_Read() */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001281static Py_ssize_t
Victor Stinner19ed27e2016-05-20 11:42:37 +02001282_Unpickler_ReadImpl(UnpicklerObject *self, char **s, Py_ssize_t n)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001283{
Antoine Pitrou04248a82010-10-12 20:51:21 +00001284 Py_ssize_t num_read;
1285
Benjamin Peterson6aa15642015-09-27 01:16:03 -07001286 *s = NULL;
Benjamin Petersone48cf7e2015-09-26 00:08:34 -07001287 if (self->next_read_idx > PY_SSIZE_T_MAX - n) {
1288 PickleState *st = _Pickle_GetGlobalState();
1289 PyErr_SetString(st->UnpicklingError,
1290 "read would overflow (invalid bytecode)");
1291 return -1;
1292 }
Victor Stinner19ed27e2016-05-20 11:42:37 +02001293
1294 /* This case is handled by the _Unpickler_Read() macro for efficiency */
1295 assert(self->next_read_idx + n > self->input_len);
1296
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001297 if (!self->read)
1298 return bad_readline();
1299
Antoine Pitrou04248a82010-10-12 20:51:21 +00001300 num_read = _Unpickler_ReadFromFile(self, n);
1301 if (num_read < 0)
1302 return -1;
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001303 if (num_read < n)
1304 return bad_readline();
Antoine Pitrou04248a82010-10-12 20:51:21 +00001305 *s = self->input_buffer;
1306 self->next_read_idx = n;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001307 return n;
1308}
1309
Victor Stinner19ed27e2016-05-20 11:42:37 +02001310/* Read `n` bytes from the unpickler's data source, storing the result in `*s`.
1311
1312 This should be used for all data reads, rather than accessing the unpickler's
1313 input buffer directly. This method deals correctly with reading from input
1314 streams, which the input buffer doesn't deal with.
1315
1316 Note that when reading from a file-like object, self->next_read_idx won't
1317 be updated (it should remain at 0 for the entire unpickling process). You
1318 should use this function's return value to know how many bytes you can
1319 consume.
1320
1321 Returns -1 (with an exception set) on failure. On success, return the
1322 number of chars read. */
1323#define _Unpickler_Read(self, s, n) \
Victor Stinnerda230562016-05-20 21:16:59 +02001324 (((n) <= (self)->input_len - (self)->next_read_idx) \
Victor Stinner19ed27e2016-05-20 11:42:37 +02001325 ? (*(s) = (self)->input_buffer + (self)->next_read_idx, \
1326 (self)->next_read_idx += (n), \
1327 (n)) \
1328 : _Unpickler_ReadImpl(self, (s), (n)))
1329
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001330static Py_ssize_t
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001331_Unpickler_CopyLine(UnpicklerObject *self, char *line, Py_ssize_t len,
1332 char **result)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001333{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001334 char *input_line = PyMem_Realloc(self->input_line, len + 1);
Victor Stinner42024562013-07-12 00:53:57 +02001335 if (input_line == NULL) {
1336 PyErr_NoMemory();
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001337 return -1;
Victor Stinner42024562013-07-12 00:53:57 +02001338 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001339
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001340 memcpy(input_line, line, len);
1341 input_line[len] = '\0';
1342 self->input_line = input_line;
1343 *result = self->input_line;
1344 return len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001345}
1346
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001347/* Read a line from the input stream/buffer. If we run off the end of the input
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001348 before hitting \n, raise an error.
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001349
1350 Returns the number of chars read, or -1 on failure. */
1351static Py_ssize_t
1352_Unpickler_Readline(UnpicklerObject *self, char **result)
1353{
1354 Py_ssize_t i, num_read;
1355
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001356 for (i = self->next_read_idx; i < self->input_len; i++) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001357 if (self->input_buffer[i] == '\n') {
1358 char *line_start = self->input_buffer + self->next_read_idx;
1359 num_read = i - self->next_read_idx + 1;
1360 self->next_read_idx = i + 1;
1361 return _Unpickler_CopyLine(self, line_start, num_read, result);
1362 }
1363 }
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001364 if (!self->read)
1365 return bad_readline();
Victor Stinner121aab42011-09-29 23:40:53 +02001366
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001367 num_read = _Unpickler_ReadFromFile(self, READ_WHOLE_LINE);
1368 if (num_read < 0)
1369 return -1;
1370 if (num_read == 0 || self->input_buffer[num_read - 1] != '\n')
1371 return bad_readline();
1372 self->next_read_idx = num_read;
1373 return _Unpickler_CopyLine(self, self->input_buffer, num_read, result);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001374}
1375
1376/* Returns -1 (with an exception set) on failure, 0 on success. The memo array
1377 will be modified in place. */
1378static int
1379_Unpickler_ResizeMemoList(UnpicklerObject *self, Py_ssize_t new_size)
1380{
1381 Py_ssize_t i;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001382
1383 assert(new_size > self->memo_size);
1384
Sergey Fedoseev67b9cc82018-08-16 09:27:50 +05001385 PyObject **memo_new = self->memo;
1386 PyMem_RESIZE(memo_new, PyObject *, new_size);
1387 if (memo_new == NULL) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001388 PyErr_NoMemory();
1389 return -1;
1390 }
Sergey Fedoseev67b9cc82018-08-16 09:27:50 +05001391 self->memo = memo_new;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001392 for (i = self->memo_size; i < new_size; i++)
1393 self->memo[i] = NULL;
1394 self->memo_size = new_size;
1395 return 0;
1396}
1397
1398/* Returns NULL if idx is out of bounds. */
1399static PyObject *
1400_Unpickler_MemoGet(UnpicklerObject *self, Py_ssize_t idx)
1401{
1402 if (idx < 0 || idx >= self->memo_size)
1403 return NULL;
1404
1405 return self->memo[idx];
1406}
1407
1408/* Returns -1 (with an exception set) on failure, 0 on success.
1409 This takes its own reference to `value`. */
1410static int
1411_Unpickler_MemoPut(UnpicklerObject *self, Py_ssize_t idx, PyObject *value)
1412{
1413 PyObject *old_item;
1414
1415 if (idx >= self->memo_size) {
1416 if (_Unpickler_ResizeMemoList(self, idx * 2) < 0)
1417 return -1;
1418 assert(idx < self->memo_size);
1419 }
1420 Py_INCREF(value);
1421 old_item = self->memo[idx];
1422 self->memo[idx] = value;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001423 if (old_item != NULL) {
1424 Py_DECREF(old_item);
1425 }
1426 else {
1427 self->memo_len++;
1428 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001429 return 0;
1430}
1431
1432static PyObject **
1433_Unpickler_NewMemo(Py_ssize_t new_size)
1434{
Benjamin Peterson59b08c12015-06-27 13:41:33 -05001435 PyObject **memo = PyMem_NEW(PyObject *, new_size);
Victor Stinner42024562013-07-12 00:53:57 +02001436 if (memo == NULL) {
1437 PyErr_NoMemory();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001438 return NULL;
Victor Stinner42024562013-07-12 00:53:57 +02001439 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001440 memset(memo, 0, new_size * sizeof(PyObject *));
1441 return memo;
1442}
1443
1444/* Free the unpickler's memo, taking care to decref any items left in it. */
1445static void
1446_Unpickler_MemoCleanup(UnpicklerObject *self)
1447{
1448 Py_ssize_t i;
1449 PyObject **memo = self->memo;
1450
1451 if (self->memo == NULL)
1452 return;
1453 self->memo = NULL;
1454 i = self->memo_size;
1455 while (--i >= 0) {
1456 Py_XDECREF(memo[i]);
1457 }
1458 PyMem_FREE(memo);
1459}
1460
1461static UnpicklerObject *
1462_Unpickler_New(void)
1463{
1464 UnpicklerObject *self;
1465
1466 self = PyObject_GC_New(UnpicklerObject, &Unpickler_Type);
1467 if (self == NULL)
1468 return NULL;
1469
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001470 self->pers_func = NULL;
1471 self->input_buffer = NULL;
1472 self->input_line = NULL;
1473 self->input_len = 0;
1474 self->next_read_idx = 0;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001475 self->prefetched_idx = 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001476 self->read = NULL;
1477 self->readline = NULL;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001478 self->peek = NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001479 self->encoding = NULL;
1480 self->errors = NULL;
1481 self->marks = NULL;
1482 self->num_marks = 0;
1483 self->marks_size = 0;
1484 self->proto = 0;
1485 self->fix_imports = 0;
Victor Stinner68c8ea22013-07-11 22:56:25 +02001486 memset(&self->buffer, 0, sizeof(Py_buffer));
1487 self->memo_size = 32;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001488 self->memo_len = 0;
Victor Stinner68c8ea22013-07-11 22:56:25 +02001489 self->memo = _Unpickler_NewMemo(self->memo_size);
1490 self->stack = (Pdata *)Pdata_New();
1491
1492 if (self->memo == NULL || self->stack == NULL) {
1493 Py_DECREF(self);
1494 return NULL;
1495 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001496
1497 return self;
1498}
1499
1500/* Returns -1 (with an exception set) on failure, 0 on success. This may
1501 be called once on a freshly created Pickler. */
1502static int
1503_Unpickler_SetInputStream(UnpicklerObject *self, PyObject *file)
1504{
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001505 _Py_IDENTIFIER(peek);
1506 _Py_IDENTIFIER(read);
1507 _Py_IDENTIFIER(readline);
Martin v. Löwis1ee1b6f2011-10-10 18:11:30 +02001508
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001509 if (_PyObject_LookupAttrId(file, &PyId_peek, &self->peek) < 0) {
1510 return -1;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001511 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001512 (void)_PyObject_LookupAttrId(file, &PyId_read, &self->read);
1513 (void)_PyObject_LookupAttrId(file, &PyId_readline, &self->readline);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001514 if (self->readline == NULL || self->read == NULL) {
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001515 if (!PyErr_Occurred()) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001516 PyErr_SetString(PyExc_TypeError,
1517 "file must have 'read' and 'readline' attributes");
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001518 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001519 Py_CLEAR(self->read);
1520 Py_CLEAR(self->readline);
Antoine Pitrou04248a82010-10-12 20:51:21 +00001521 Py_CLEAR(self->peek);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001522 return -1;
1523 }
1524 return 0;
1525}
1526
1527/* Returns -1 (with an exception set) on failure, 0 on success. This may
1528 be called once on a freshly created Pickler. */
1529static int
1530_Unpickler_SetInputEncoding(UnpicklerObject *self,
1531 const char *encoding,
1532 const char *errors)
1533{
1534 if (encoding == NULL)
1535 encoding = "ASCII";
1536 if (errors == NULL)
1537 errors = "strict";
1538
Victor Stinner49fc8ec2013-07-07 23:30:24 +02001539 self->encoding = _PyMem_Strdup(encoding);
1540 self->errors = _PyMem_Strdup(errors);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001541 if (self->encoding == NULL || self->errors == NULL) {
1542 PyErr_NoMemory();
1543 return -1;
1544 }
1545 return 0;
1546}
1547
1548/* Generate a GET opcode for an object stored in the memo. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001549static int
1550memo_get(PicklerObject *self, PyObject *key)
1551{
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001552 Py_ssize_t *value;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001553 char pdata[30];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001554 Py_ssize_t len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001555
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001556 value = PyMemoTable_Get(self->memo, key);
1557 if (value == NULL) {
1558 PyErr_SetObject(PyExc_KeyError, key);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001559 return -1;
1560 }
1561
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001562 if (!self->bin) {
1563 pdata[0] = GET;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001564 PyOS_snprintf(pdata + 1, sizeof(pdata) - 1,
1565 "%" PY_FORMAT_SIZE_T "d\n", *value);
1566 len = strlen(pdata);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001567 }
1568 else {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001569 if (*value < 256) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001570 pdata[0] = BINGET;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001571 pdata[1] = (unsigned char)(*value & 0xff);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001572 len = 2;
1573 }
Serhiy Storchaka67c719b2014-09-05 10:10:23 +03001574 else if ((size_t)*value <= 0xffffffffUL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001575 pdata[0] = LONG_BINGET;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001576 pdata[1] = (unsigned char)(*value & 0xff);
1577 pdata[2] = (unsigned char)((*value >> 8) & 0xff);
1578 pdata[3] = (unsigned char)((*value >> 16) & 0xff);
1579 pdata[4] = (unsigned char)((*value >> 24) & 0xff);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001580 len = 5;
1581 }
1582 else { /* unlikely */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08001583 PickleState *st = _Pickle_GetGlobalState();
1584 PyErr_SetString(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001585 "memo id too large for LONG_BINGET");
1586 return -1;
1587 }
1588 }
1589
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001590 if (_Pickler_Write(self, pdata, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001591 return -1;
1592
1593 return 0;
1594}
1595
1596/* Store an object in the memo, assign it a new unique ID based on the number
1597 of objects currently stored in the memo and generate a PUT opcode. */
1598static int
1599memo_put(PicklerObject *self, PyObject *obj)
1600{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001601 char pdata[30];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001602 Py_ssize_t len;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001603 Py_ssize_t idx;
1604
1605 const char memoize_op = MEMOIZE;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001606
1607 if (self->fast)
1608 return 0;
1609
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001610 idx = PyMemoTable_Size(self->memo);
1611 if (PyMemoTable_Set(self->memo, obj, idx) < 0)
1612 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001613
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001614 if (self->proto >= 4) {
1615 if (_Pickler_Write(self, &memoize_op, 1) < 0)
1616 return -1;
1617 return 0;
1618 }
1619 else if (!self->bin) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001620 pdata[0] = PUT;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001621 PyOS_snprintf(pdata + 1, sizeof(pdata) - 1,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001622 "%" PY_FORMAT_SIZE_T "d\n", idx);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001623 len = strlen(pdata);
1624 }
1625 else {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001626 if (idx < 256) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001627 pdata[0] = BINPUT;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001628 pdata[1] = (unsigned char)idx;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001629 len = 2;
1630 }
Serhiy Storchaka67c719b2014-09-05 10:10:23 +03001631 else if ((size_t)idx <= 0xffffffffUL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001632 pdata[0] = LONG_BINPUT;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001633 pdata[1] = (unsigned char)(idx & 0xff);
1634 pdata[2] = (unsigned char)((idx >> 8) & 0xff);
1635 pdata[3] = (unsigned char)((idx >> 16) & 0xff);
1636 pdata[4] = (unsigned char)((idx >> 24) & 0xff);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001637 len = 5;
1638 }
1639 else { /* unlikely */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08001640 PickleState *st = _Pickle_GetGlobalState();
1641 PyErr_SetString(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001642 "memo id too large for LONG_BINPUT");
1643 return -1;
1644 }
1645 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001646 if (_Pickler_Write(self, pdata, len) < 0)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001647 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001648
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001649 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001650}
1651
1652static PyObject *
Serhiy Storchaka9937d902017-01-09 10:04:34 +02001653get_dotted_path(PyObject *obj, PyObject *name)
1654{
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001655 _Py_static_string(PyId_dot, ".");
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001656 PyObject *dotted_path;
1657 Py_ssize_t i, n;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001658
1659 dotted_path = PyUnicode_Split(name, _PyUnicode_FromId(&PyId_dot), -1);
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001660 if (dotted_path == NULL)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001661 return NULL;
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001662 n = PyList_GET_SIZE(dotted_path);
1663 assert(n >= 1);
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001664 for (i = 0; i < n; i++) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001665 PyObject *subpath = PyList_GET_ITEM(dotted_path, i);
Serhiy Storchaka9937d902017-01-09 10:04:34 +02001666 if (_PyUnicode_EqualToASCIIString(subpath, "<locals>")) {
Antoine Pitrou6cd5eda2014-12-02 00:20:03 +01001667 if (obj == NULL)
1668 PyErr_Format(PyExc_AttributeError,
1669 "Can't pickle local object %R", name);
1670 else
1671 PyErr_Format(PyExc_AttributeError,
1672 "Can't pickle local attribute %R on %R", name, obj);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001673 Py_DECREF(dotted_path);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001674 return NULL;
1675 }
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001676 }
1677 return dotted_path;
1678}
1679
1680static PyObject *
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001681get_deep_attribute(PyObject *obj, PyObject *names, PyObject **pparent)
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001682{
1683 Py_ssize_t i, n;
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001684 PyObject *parent = NULL;
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001685
1686 assert(PyList_CheckExact(names));
1687 Py_INCREF(obj);
1688 n = PyList_GET_SIZE(names);
1689 for (i = 0; i < n; i++) {
1690 PyObject *name = PyList_GET_ITEM(names, i);
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001691 Py_XDECREF(parent);
1692 parent = obj;
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001693 (void)_PyObject_LookupAttr(parent, name, &obj);
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001694 if (obj == NULL) {
1695 Py_DECREF(parent);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001696 return NULL;
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001697 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001698 }
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001699 if (pparent != NULL)
1700 *pparent = parent;
1701 else
1702 Py_XDECREF(parent);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001703 return obj;
1704}
1705
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001706
1707static PyObject *
1708getattribute(PyObject *obj, PyObject *name, int allow_qualname)
1709{
1710 PyObject *dotted_path, *attr;
1711
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001712 if (allow_qualname) {
1713 dotted_path = get_dotted_path(obj, name);
1714 if (dotted_path == NULL)
1715 return NULL;
1716 attr = get_deep_attribute(obj, dotted_path, NULL);
1717 Py_DECREF(dotted_path);
1718 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001719 else {
1720 (void)_PyObject_LookupAttr(obj, name, &attr);
1721 }
1722 if (attr == NULL && !PyErr_Occurred()) {
1723 PyErr_Format(PyExc_AttributeError,
1724 "Can't get attribute %R on %R", name, obj);
1725 }
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001726 return attr;
1727}
1728
Eric Snow3f9eee62017-09-15 16:35:20 -06001729static int
1730_checkmodule(PyObject *module_name, PyObject *module,
1731 PyObject *global, PyObject *dotted_path)
1732{
1733 if (module == Py_None) {
1734 return -1;
1735 }
1736 if (PyUnicode_Check(module_name) &&
1737 _PyUnicode_EqualToASCIIString(module_name, "__main__")) {
1738 return -1;
1739 }
1740
1741 PyObject *candidate = get_deep_attribute(module, dotted_path, NULL);
1742 if (candidate == NULL) {
Eric Snow3f9eee62017-09-15 16:35:20 -06001743 return -1;
1744 }
1745 if (candidate != global) {
1746 Py_DECREF(candidate);
1747 return -1;
1748 }
1749 Py_DECREF(candidate);
1750 return 0;
1751}
1752
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001753static PyObject *
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001754whichmodule(PyObject *global, PyObject *dotted_path)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001755{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001756 PyObject *module_name;
Eric Snow3f9eee62017-09-15 16:35:20 -06001757 PyObject *module = NULL;
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001758 Py_ssize_t i;
Eric Snow3f9eee62017-09-15 16:35:20 -06001759 PyObject *modules;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001760 _Py_IDENTIFIER(__module__);
1761 _Py_IDENTIFIER(modules);
1762 _Py_IDENTIFIER(__main__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001763
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001764 if (_PyObject_LookupAttrId(global, &PyId___module__, &module_name) < 0) {
1765 return NULL;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001766 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001767 if (module_name) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001768 /* In some rare cases (e.g., bound methods of extension types),
1769 __module__ can be None. If it is so, then search sys.modules for
1770 the module of global. */
1771 if (module_name != Py_None)
1772 return module_name;
1773 Py_CLEAR(module_name);
1774 }
1775 assert(module_name == NULL);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001776
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001777 /* Fallback on walking sys.modules */
Eric Snow3f9eee62017-09-15 16:35:20 -06001778 modules = _PySys_GetObjectId(&PyId_modules);
1779 if (modules == NULL) {
Victor Stinner1e53bba2013-07-16 22:26:05 +02001780 PyErr_SetString(PyExc_RuntimeError, "unable to get sys.modules");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001781 return NULL;
Victor Stinner1e53bba2013-07-16 22:26:05 +02001782 }
Eric Snow3f9eee62017-09-15 16:35:20 -06001783 if (PyDict_CheckExact(modules)) {
1784 i = 0;
1785 while (PyDict_Next(modules, &i, &module_name, &module)) {
1786 if (_checkmodule(module_name, module, global, dotted_path) == 0) {
1787 Py_INCREF(module_name);
1788 return module_name;
1789 }
1790 if (PyErr_Occurred()) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001791 return NULL;
Eric Snow3f9eee62017-09-15 16:35:20 -06001792 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001793 }
Eric Snow3f9eee62017-09-15 16:35:20 -06001794 }
1795 else {
1796 PyObject *iterator = PyObject_GetIter(modules);
1797 if (iterator == NULL) {
1798 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001799 }
Eric Snow3f9eee62017-09-15 16:35:20 -06001800 while ((module_name = PyIter_Next(iterator))) {
1801 module = PyObject_GetItem(modules, module_name);
1802 if (module == NULL) {
1803 Py_DECREF(module_name);
1804 Py_DECREF(iterator);
1805 return NULL;
1806 }
1807 if (_checkmodule(module_name, module, global, dotted_path) == 0) {
1808 Py_DECREF(module);
1809 Py_DECREF(iterator);
1810 return module_name;
1811 }
1812 Py_DECREF(module);
1813 Py_DECREF(module_name);
1814 if (PyErr_Occurred()) {
1815 Py_DECREF(iterator);
1816 return NULL;
1817 }
1818 }
1819 Py_DECREF(iterator);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001820 }
1821
1822 /* If no module is found, use __main__. */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001823 module_name = _PyUnicode_FromId(&PyId___main__);
Victor Stinneraf46eb82017-09-05 23:30:16 +02001824 Py_XINCREF(module_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001825 return module_name;
1826}
1827
1828/* fast_save_enter() and fast_save_leave() are guards against recursive
1829 objects when Pickler is used with the "fast mode" (i.e., with object
1830 memoization disabled). If the nesting of a list or dict object exceed
1831 FAST_NESTING_LIMIT, these guards will start keeping an internal
1832 reference to the seen list or dict objects and check whether these objects
1833 are recursive. These are not strictly necessary, since save() has a
1834 hard-coded recursion limit, but they give a nicer error message than the
1835 typical RuntimeError. */
1836static int
1837fast_save_enter(PicklerObject *self, PyObject *obj)
1838{
1839 /* if fast_nesting < 0, we're doing an error exit. */
1840 if (++self->fast_nesting >= FAST_NESTING_LIMIT) {
1841 PyObject *key = NULL;
1842 if (self->fast_memo == NULL) {
1843 self->fast_memo = PyDict_New();
1844 if (self->fast_memo == NULL) {
1845 self->fast_nesting = -1;
1846 return 0;
1847 }
1848 }
1849 key = PyLong_FromVoidPtr(obj);
Mat Mf76231f2017-11-13 02:50:16 -05001850 if (key == NULL) {
1851 self->fast_nesting = -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001852 return 0;
Mat Mf76231f2017-11-13 02:50:16 -05001853 }
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08001854 if (PyDict_GetItemWithError(self->fast_memo, key)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001855 Py_DECREF(key);
1856 PyErr_Format(PyExc_ValueError,
1857 "fast mode: can't pickle cyclic objects "
1858 "including object type %.200s at %p",
1859 obj->ob_type->tp_name, obj);
1860 self->fast_nesting = -1;
1861 return 0;
1862 }
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08001863 if (PyErr_Occurred()) {
Mat Mf76231f2017-11-13 02:50:16 -05001864 Py_DECREF(key);
1865 self->fast_nesting = -1;
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08001866 return 0;
1867 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001868 if (PyDict_SetItem(self->fast_memo, key, Py_None) < 0) {
1869 Py_DECREF(key);
1870 self->fast_nesting = -1;
1871 return 0;
1872 }
1873 Py_DECREF(key);
1874 }
1875 return 1;
1876}
1877
1878static int
1879fast_save_leave(PicklerObject *self, PyObject *obj)
1880{
1881 if (self->fast_nesting-- >= FAST_NESTING_LIMIT) {
1882 PyObject *key = PyLong_FromVoidPtr(obj);
1883 if (key == NULL)
1884 return 0;
1885 if (PyDict_DelItem(self->fast_memo, key) < 0) {
1886 Py_DECREF(key);
1887 return 0;
1888 }
1889 Py_DECREF(key);
1890 }
1891 return 1;
1892}
1893
1894static int
1895save_none(PicklerObject *self, PyObject *obj)
1896{
1897 const char none_op = NONE;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001898 if (_Pickler_Write(self, &none_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001899 return -1;
1900
1901 return 0;
1902}
1903
1904static int
1905save_bool(PicklerObject *self, PyObject *obj)
1906{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001907 if (self->proto >= 2) {
Alexandre Vassalotti8a67f522013-11-24 21:40:18 -08001908 const char bool_op = (obj == Py_True) ? NEWTRUE : NEWFALSE;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001909 if (_Pickler_Write(self, &bool_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001910 return -1;
1911 }
Alexandre Vassalotti8a67f522013-11-24 21:40:18 -08001912 else {
1913 /* These aren't opcodes -- they're ways to pickle bools before protocol 2
1914 * so that unpicklers written before bools were introduced unpickle them
1915 * as ints, but unpicklers after can recognize that bools were intended.
1916 * Note that protocol 2 added direct ways to pickle bools.
1917 */
1918 const char *bool_str = (obj == Py_True) ? "I01\n" : "I00\n";
1919 if (_Pickler_Write(self, bool_str, strlen(bool_str)) < 0)
1920 return -1;
1921 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001922 return 0;
1923}
1924
1925static int
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001926save_long(PicklerObject *self, PyObject *obj)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001927{
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001928 PyObject *repr = NULL;
1929 Py_ssize_t size;
1930 long val;
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001931 int overflow;
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001932 int status = 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001933
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001934 val= PyLong_AsLongAndOverflow(obj, &overflow);
1935 if (!overflow && (sizeof(long) <= 4 ||
1936 (val <= 0x7fffffffL && val >= (-0x7fffffffL - 1))))
1937 {
Larry Hastings61272b72014-01-07 12:41:53 -08001938 /* result fits in a signed 4-byte integer.
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -08001939
1940 Note: we can't use -0x80000000L in the above condition because some
1941 compilers (e.g., MSVC) will promote 0x80000000L to an unsigned type
1942 before applying the unary minus when sizeof(long) <= 4. The
1943 resulting value stays unsigned which is commonly not what we want,
1944 so MSVC happily warns us about it. However, that result would have
1945 been fine because we guard for sizeof(long) <= 4 which turns the
1946 condition true in that particular case. */
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001947 char pdata[32];
1948 Py_ssize_t len = 0;
1949
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001950 if (self->bin) {
1951 pdata[1] = (unsigned char)(val & 0xff);
1952 pdata[2] = (unsigned char)((val >> 8) & 0xff);
1953 pdata[3] = (unsigned char)((val >> 16) & 0xff);
1954 pdata[4] = (unsigned char)((val >> 24) & 0xff);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001955
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001956 if ((pdata[4] != 0) || (pdata[3] != 0)) {
1957 pdata[0] = BININT;
1958 len = 5;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001959 }
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001960 else if (pdata[2] != 0) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001961 pdata[0] = BININT2;
1962 len = 3;
1963 }
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001964 else {
1965 pdata[0] = BININT1;
1966 len = 2;
1967 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001968 }
1969 else {
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001970 sprintf(pdata, "%c%ld\n", INT, val);
1971 len = strlen(pdata);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001972 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001973 if (_Pickler_Write(self, pdata, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001974 return -1;
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001975
1976 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001977 }
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001978 assert(!PyErr_Occurred());
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001979
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001980 if (self->proto >= 2) {
1981 /* Linear-time pickling. */
1982 size_t nbits;
1983 size_t nbytes;
1984 unsigned char *pdata;
1985 char header[5];
1986 int i;
1987 int sign = _PyLong_Sign(obj);
1988
1989 if (sign == 0) {
1990 header[0] = LONG1;
1991 header[1] = 0; /* It's 0 -- an empty bytestring. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001992 if (_Pickler_Write(self, header, 2) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001993 goto error;
1994 return 0;
1995 }
1996 nbits = _PyLong_NumBits(obj);
1997 if (nbits == (size_t)-1 && PyErr_Occurred())
1998 goto error;
1999 /* How many bytes do we need? There are nbits >> 3 full
2000 * bytes of data, and nbits & 7 leftover bits. If there
2001 * are any leftover bits, then we clearly need another
2002 * byte. Wnat's not so obvious is that we *probably*
2003 * need another byte even if there aren't any leftovers:
2004 * the most-significant bit of the most-significant byte
2005 * acts like a sign bit, and it's usually got a sense
Serhiy Storchaka95949422013-08-27 19:40:23 +03002006 * opposite of the one we need. The exception is ints
2007 * of the form -(2**(8*j-1)) for j > 0. Such an int is
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002008 * its own 256's-complement, so has the right sign bit
2009 * even without the extra byte. That's a pain to check
2010 * for in advance, though, so we always grab an extra
2011 * byte at the start, and cut it back later if possible.
2012 */
2013 nbytes = (nbits >> 3) + 1;
Antoine Pitroubf6ecf92012-11-24 20:40:21 +01002014 if (nbytes > 0x7fffffffL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002015 PyErr_SetString(PyExc_OverflowError,
Serhiy Storchaka95949422013-08-27 19:40:23 +03002016 "int too large to pickle");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002017 goto error;
2018 }
Neal Norwitz6ae2eb22008-08-24 23:50:08 +00002019 repr = PyBytes_FromStringAndSize(NULL, (Py_ssize_t)nbytes);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002020 if (repr == NULL)
2021 goto error;
Neal Norwitz6ae2eb22008-08-24 23:50:08 +00002022 pdata = (unsigned char *)PyBytes_AS_STRING(repr);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002023 i = _PyLong_AsByteArray((PyLongObject *)obj,
2024 pdata, nbytes,
2025 1 /* little endian */ , 1 /* signed */ );
2026 if (i < 0)
2027 goto error;
Serhiy Storchaka95949422013-08-27 19:40:23 +03002028 /* If the int is negative, this may be a byte more than
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002029 * needed. This is so iff the MSB is all redundant sign
2030 * bits.
2031 */
2032 if (sign < 0 &&
Victor Stinner121aab42011-09-29 23:40:53 +02002033 nbytes > 1 &&
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002034 pdata[nbytes - 1] == 0xff &&
2035 (pdata[nbytes - 2] & 0x80) != 0) {
2036 nbytes--;
2037 }
2038
2039 if (nbytes < 256) {
2040 header[0] = LONG1;
2041 header[1] = (unsigned char)nbytes;
2042 size = 2;
2043 }
2044 else {
2045 header[0] = LONG4;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002046 size = (Py_ssize_t) nbytes;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002047 for (i = 1; i < 5; i++) {
2048 header[i] = (unsigned char)(size & 0xff);
2049 size >>= 8;
2050 }
2051 size = 5;
2052 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002053 if (_Pickler_Write(self, header, size) < 0 ||
2054 _Pickler_Write(self, (char *)pdata, (int)nbytes) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002055 goto error;
2056 }
2057 else {
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02002058 const char long_op = LONG;
Serhiy Storchaka85b0f5b2016-11-20 10:16:47 +02002059 const char *string;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002060
Mark Dickinson8dd05142009-01-20 20:43:58 +00002061 /* proto < 2: write the repr and newline. This is quadratic-time (in
2062 the number of digits), in both directions. We add a trailing 'L'
2063 to the repr, for compatibility with Python 2.x. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002064
2065 repr = PyObject_Repr(obj);
2066 if (repr == NULL)
2067 goto error;
2068
Serhiy Storchaka06515832016-11-20 09:13:07 +02002069 string = PyUnicode_AsUTF8AndSize(repr, &size);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002070 if (string == NULL)
2071 goto error;
2072
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002073 if (_Pickler_Write(self, &long_op, 1) < 0 ||
2074 _Pickler_Write(self, string, size) < 0 ||
2075 _Pickler_Write(self, "L\n", 2) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002076 goto error;
2077 }
2078
2079 if (0) {
2080 error:
2081 status = -1;
2082 }
2083 Py_XDECREF(repr);
2084
2085 return status;
2086}
2087
2088static int
2089save_float(PicklerObject *self, PyObject *obj)
2090{
2091 double x = PyFloat_AS_DOUBLE((PyFloatObject *)obj);
2092
2093 if (self->bin) {
2094 char pdata[9];
2095 pdata[0] = BINFLOAT;
2096 if (_PyFloat_Pack8(x, (unsigned char *)&pdata[1], 0) < 0)
2097 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002098 if (_Pickler_Write(self, pdata, 9) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002099 return -1;
Victor Stinner121aab42011-09-29 23:40:53 +02002100 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002101 else {
Eric Smith0923d1d2009-04-16 20:16:10 +00002102 int result = -1;
2103 char *buf = NULL;
2104 char op = FLOAT;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002105
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002106 if (_Pickler_Write(self, &op, 1) < 0)
Eric Smith0923d1d2009-04-16 20:16:10 +00002107 goto done;
2108
Serhiy Storchakac86ca262015-02-15 14:18:32 +02002109 buf = PyOS_double_to_string(x, 'r', 0, Py_DTSF_ADD_DOT_0, NULL);
Eric Smith0923d1d2009-04-16 20:16:10 +00002110 if (!buf) {
2111 PyErr_NoMemory();
2112 goto done;
2113 }
2114
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002115 if (_Pickler_Write(self, buf, strlen(buf)) < 0)
Eric Smith0923d1d2009-04-16 20:16:10 +00002116 goto done;
2117
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002118 if (_Pickler_Write(self, "\n", 1) < 0)
Eric Smith0923d1d2009-04-16 20:16:10 +00002119 goto done;
2120
2121 result = 0;
2122done:
2123 PyMem_Free(buf);
2124 return result;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002125 }
2126
2127 return 0;
2128}
2129
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002130/* Perform direct write of the header and payload of the binary object.
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002131
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002132 The large contiguous data is written directly into the underlying file
2133 object, bypassing the output_buffer of the Pickler. We intentionally
2134 do not insert a protocol 4 frame opcode to make it possible to optimize
2135 file.read calls in the loader.
2136 */
2137static int
2138_Pickler_write_bytes(PicklerObject *self,
2139 const char *header, Py_ssize_t header_size,
2140 const char *data, Py_ssize_t data_size,
2141 PyObject *payload)
2142{
2143 int bypass_buffer = (data_size >= FRAME_SIZE_TARGET);
2144 int framing = self->framing;
2145
2146 if (bypass_buffer) {
2147 assert(self->output_buffer != NULL);
2148 /* Commit the previous frame. */
2149 if (_Pickler_CommitFrame(self)) {
2150 return -1;
2151 }
2152 /* Disable framing temporarily */
2153 self->framing = 0;
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002154 }
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002155
2156 if (_Pickler_Write(self, header, header_size) < 0) {
2157 return -1;
2158 }
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002159
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002160 if (bypass_buffer && self->write != NULL) {
2161 /* Bypass the in-memory buffer to directly stream large data
2162 into the underlying file object. */
2163 PyObject *result, *mem = NULL;
2164 /* Dump the output buffer to the file. */
2165 if (_Pickler_FlushToFile(self) < 0) {
2166 return -1;
2167 }
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002168
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002169 /* Stream write the payload into the file without going through the
2170 output buffer. */
2171 if (payload == NULL) {
Serhiy Storchaka5b76bdb2018-01-13 00:28:31 +02002172 /* TODO: It would be better to use a memoryview with a linked
2173 original string if this is possible. */
2174 payload = mem = PyBytes_FromStringAndSize(data, data_size);
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002175 if (payload == NULL) {
2176 return -1;
2177 }
2178 }
2179 result = PyObject_CallFunctionObjArgs(self->write, payload, NULL);
2180 Py_XDECREF(mem);
2181 if (result == NULL) {
2182 return -1;
2183 }
2184 Py_DECREF(result);
2185
2186 /* Reinitialize the buffer for subsequent calls to _Pickler_Write. */
2187 if (_Pickler_ClearBuffer(self) < 0) {
2188 return -1;
2189 }
2190 }
2191 else {
2192 if (_Pickler_Write(self, data, data_size) < 0) {
2193 return -1;
2194 }
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002195 }
2196
2197 /* Re-enable framing for subsequent calls to _Pickler_Write. */
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002198 self->framing = framing;
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002199
2200 return 0;
2201}
2202
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002203static int
2204save_bytes(PicklerObject *self, PyObject *obj)
2205{
2206 if (self->proto < 3) {
2207 /* Older pickle protocols do not have an opcode for pickling bytes
2208 objects. Therefore, we need to fake the copy protocol (i.e.,
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002209 the __reduce__ method) to permit bytes object unpickling.
2210
2211 Here we use a hack to be compatible with Python 2. Since in Python
2212 2 'bytes' is just an alias for 'str' (which has different
2213 parameters than the actual bytes object), we use codecs.encode
2214 to create the appropriate 'str' object when unpickled using
2215 Python 2 *and* the appropriate 'bytes' object when unpickled
2216 using Python 3. Again this is a hack and we don't need to do this
2217 with newer protocols. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002218 PyObject *reduce_value = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002219 int status;
2220
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002221 if (PyBytes_GET_SIZE(obj) == 0) {
2222 reduce_value = Py_BuildValue("(O())", (PyObject*)&PyBytes_Type);
2223 }
2224 else {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08002225 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002226 PyObject *unicode_str =
2227 PyUnicode_DecodeLatin1(PyBytes_AS_STRING(obj),
2228 PyBytes_GET_SIZE(obj),
2229 "strict");
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002230 _Py_IDENTIFIER(latin1);
2231
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002232 if (unicode_str == NULL)
2233 return -1;
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002234 reduce_value = Py_BuildValue("(O(OO))",
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08002235 st->codecs_encode, unicode_str,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002236 _PyUnicode_FromId(&PyId_latin1));
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002237 Py_DECREF(unicode_str);
2238 }
2239
2240 if (reduce_value == NULL)
2241 return -1;
2242
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002243 /* save_reduce() will memoize the object automatically. */
2244 status = save_reduce(self, reduce_value, obj);
2245 Py_DECREF(reduce_value);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002246 return status;
2247 }
2248 else {
2249 Py_ssize_t size;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002250 char header[9];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002251 Py_ssize_t len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002252
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002253 size = PyBytes_GET_SIZE(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002254 if (size < 0)
2255 return -1;
2256
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002257 if (size <= 0xff) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002258 header[0] = SHORT_BINBYTES;
2259 header[1] = (unsigned char)size;
2260 len = 2;
2261 }
Serhiy Storchaka67c719b2014-09-05 10:10:23 +03002262 else if ((size_t)size <= 0xffffffffUL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002263 header[0] = BINBYTES;
2264 header[1] = (unsigned char)(size & 0xff);
2265 header[2] = (unsigned char)((size >> 8) & 0xff);
2266 header[3] = (unsigned char)((size >> 16) & 0xff);
2267 header[4] = (unsigned char)((size >> 24) & 0xff);
2268 len = 5;
2269 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002270 else if (self->proto >= 4) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002271 header[0] = BINBYTES8;
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -08002272 _write_size64(header + 1, size);
Alexandre Vassalotti6e73ff12013-12-05 19:29:32 -08002273 len = 9;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002274 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002275 else {
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002276 PyErr_SetString(PyExc_OverflowError,
Serhiy Storchakaf8def282013-02-16 17:29:56 +02002277 "cannot serialize a bytes object larger than 4 GiB");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002278 return -1; /* string too large */
2279 }
2280
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002281 if (_Pickler_write_bytes(self, header, len,
2282 PyBytes_AS_STRING(obj), size, obj) < 0)
2283 {
2284 return -1;
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002285 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002286
2287 if (memo_put(self, obj) < 0)
2288 return -1;
2289
2290 return 0;
2291 }
2292}
2293
2294/* A copy of PyUnicode_EncodeRawUnicodeEscape() that also translates
2295 backslash and newline characters to \uXXXX escapes. */
2296static PyObject *
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002297raw_unicode_escape(PyObject *obj)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002298{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002299 char *p;
Victor Stinner049e5092014-08-17 22:20:00 +02002300 Py_ssize_t i, size;
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002301 void *data;
2302 unsigned int kind;
Victor Stinner358af132015-10-12 22:36:57 +02002303 _PyBytesWriter writer;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002304
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002305 if (PyUnicode_READY(obj))
2306 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002307
Victor Stinner358af132015-10-12 22:36:57 +02002308 _PyBytesWriter_Init(&writer);
2309
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002310 size = PyUnicode_GET_LENGTH(obj);
2311 data = PyUnicode_DATA(obj);
2312 kind = PyUnicode_KIND(obj);
Victor Stinner121aab42011-09-29 23:40:53 +02002313
Victor Stinner358af132015-10-12 22:36:57 +02002314 p = _PyBytesWriter_Alloc(&writer, size);
2315 if (p == NULL)
2316 goto error;
2317 writer.overallocate = 1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002318
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002319 for (i=0; i < size; i++) {
2320 Py_UCS4 ch = PyUnicode_READ(kind, data, i);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002321 /* Map 32-bit characters to '\Uxxxxxxxx' */
2322 if (ch >= 0x10000) {
Raymond Hettinger15f44ab2016-08-30 10:47:49 -07002323 /* -1: subtract 1 preallocated byte */
Victor Stinner358af132015-10-12 22:36:57 +02002324 p = _PyBytesWriter_Prepare(&writer, p, 10-1);
2325 if (p == NULL)
2326 goto error;
2327
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002328 *p++ = '\\';
2329 *p++ = 'U';
Victor Stinnerf5cff562011-10-14 02:13:11 +02002330 *p++ = Py_hexdigits[(ch >> 28) & 0xf];
2331 *p++ = Py_hexdigits[(ch >> 24) & 0xf];
2332 *p++ = Py_hexdigits[(ch >> 20) & 0xf];
2333 *p++ = Py_hexdigits[(ch >> 16) & 0xf];
2334 *p++ = Py_hexdigits[(ch >> 12) & 0xf];
2335 *p++ = Py_hexdigits[(ch >> 8) & 0xf];
2336 *p++ = Py_hexdigits[(ch >> 4) & 0xf];
2337 *p++ = Py_hexdigits[ch & 15];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002338 }
Victor Stinner358af132015-10-12 22:36:57 +02002339 /* Map 16-bit characters, '\\' and '\n' to '\uxxxx' */
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002340 else if (ch >= 256 || ch == '\\' || ch == '\n') {
Raymond Hettinger15f44ab2016-08-30 10:47:49 -07002341 /* -1: subtract 1 preallocated byte */
Victor Stinner358af132015-10-12 22:36:57 +02002342 p = _PyBytesWriter_Prepare(&writer, p, 6-1);
2343 if (p == NULL)
2344 goto error;
2345
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002346 *p++ = '\\';
2347 *p++ = 'u';
Victor Stinnerf5cff562011-10-14 02:13:11 +02002348 *p++ = Py_hexdigits[(ch >> 12) & 0xf];
2349 *p++ = Py_hexdigits[(ch >> 8) & 0xf];
2350 *p++ = Py_hexdigits[(ch >> 4) & 0xf];
2351 *p++ = Py_hexdigits[ch & 15];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002352 }
Alexandre Vassalotti554d8782008-12-27 07:32:41 +00002353 /* Copy everything else as-is */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002354 else
2355 *p++ = (char) ch;
2356 }
Victor Stinner358af132015-10-12 22:36:57 +02002357
2358 return _PyBytesWriter_Finish(&writer, p);
2359
2360error:
2361 _PyBytesWriter_Dealloc(&writer);
2362 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002363}
2364
2365static int
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002366write_unicode_binary(PicklerObject *self, PyObject *obj)
Antoine Pitrou299978d2013-04-07 17:38:11 +02002367{
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002368 char header[9];
2369 Py_ssize_t len;
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002370 PyObject *encoded = NULL;
2371 Py_ssize_t size;
2372 const char *data;
2373
2374 if (PyUnicode_READY(obj))
2375 return -1;
2376
2377 data = PyUnicode_AsUTF8AndSize(obj, &size);
2378 if (data == NULL) {
2379 /* Issue #8383: for strings with lone surrogates, fallback on the
2380 "surrogatepass" error handler. */
2381 PyErr_Clear();
2382 encoded = PyUnicode_AsEncodedString(obj, "utf-8", "surrogatepass");
2383 if (encoded == NULL)
2384 return -1;
2385
2386 data = PyBytes_AS_STRING(encoded);
2387 size = PyBytes_GET_SIZE(encoded);
2388 }
Antoine Pitrou299978d2013-04-07 17:38:11 +02002389
Victor Stinnerf13c46c2014-08-17 21:05:55 +02002390 assert(size >= 0);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002391 if (size <= 0xff && self->proto >= 4) {
2392 header[0] = SHORT_BINUNICODE;
2393 header[1] = (unsigned char)(size & 0xff);
2394 len = 2;
2395 }
Victor Stinnerf13c46c2014-08-17 21:05:55 +02002396 else if ((size_t)size <= 0xffffffffUL) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002397 header[0] = BINUNICODE;
2398 header[1] = (unsigned char)(size & 0xff);
2399 header[2] = (unsigned char)((size >> 8) & 0xff);
2400 header[3] = (unsigned char)((size >> 16) & 0xff);
2401 header[4] = (unsigned char)((size >> 24) & 0xff);
2402 len = 5;
2403 }
2404 else if (self->proto >= 4) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002405 header[0] = BINUNICODE8;
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -08002406 _write_size64(header + 1, size);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002407 len = 9;
2408 }
2409 else {
Antoine Pitrou299978d2013-04-07 17:38:11 +02002410 PyErr_SetString(PyExc_OverflowError,
Antoine Pitrou4b7b0f02013-04-07 23:46:52 +02002411 "cannot serialize a string larger than 4GiB");
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002412 Py_XDECREF(encoded);
Antoine Pitrou299978d2013-04-07 17:38:11 +02002413 return -1;
2414 }
Antoine Pitrou299978d2013-04-07 17:38:11 +02002415
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002416 if (_Pickler_write_bytes(self, header, len, data, size, encoded) < 0) {
2417 Py_XDECREF(encoded);
2418 return -1;
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002419 }
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002420 Py_XDECREF(encoded);
Antoine Pitrou299978d2013-04-07 17:38:11 +02002421 return 0;
2422}
2423
2424static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002425save_unicode(PicklerObject *self, PyObject *obj)
2426{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002427 if (self->bin) {
Antoine Pitrou299978d2013-04-07 17:38:11 +02002428 if (write_unicode_binary(self, obj) < 0)
2429 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002430 }
2431 else {
Antoine Pitrou299978d2013-04-07 17:38:11 +02002432 PyObject *encoded;
2433 Py_ssize_t size;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002434 const char unicode_op = UNICODE;
2435
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002436 encoded = raw_unicode_escape(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002437 if (encoded == NULL)
Antoine Pitrou299978d2013-04-07 17:38:11 +02002438 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002439
Antoine Pitrou299978d2013-04-07 17:38:11 +02002440 if (_Pickler_Write(self, &unicode_op, 1) < 0) {
2441 Py_DECREF(encoded);
2442 return -1;
2443 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002444
2445 size = PyBytes_GET_SIZE(encoded);
Antoine Pitrou299978d2013-04-07 17:38:11 +02002446 if (_Pickler_Write(self, PyBytes_AS_STRING(encoded), size) < 0) {
2447 Py_DECREF(encoded);
2448 return -1;
2449 }
2450 Py_DECREF(encoded);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002451
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002452 if (_Pickler_Write(self, "\n", 1) < 0)
Antoine Pitrou299978d2013-04-07 17:38:11 +02002453 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002454 }
2455 if (memo_put(self, obj) < 0)
Antoine Pitrou299978d2013-04-07 17:38:11 +02002456 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002457
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002458 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002459}
2460
2461/* A helper for save_tuple. Push the len elements in tuple t on the stack. */
2462static int
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002463store_tuple_elements(PicklerObject *self, PyObject *t, Py_ssize_t len)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002464{
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002465 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002466
2467 assert(PyTuple_Size(t) == len);
2468
2469 for (i = 0; i < len; i++) {
2470 PyObject *element = PyTuple_GET_ITEM(t, i);
2471
2472 if (element == NULL)
2473 return -1;
2474 if (save(self, element, 0) < 0)
2475 return -1;
2476 }
2477
2478 return 0;
2479}
2480
2481/* Tuples are ubiquitous in the pickle protocols, so many techniques are
2482 * used across protocols to minimize the space needed to pickle them.
2483 * Tuples are also the only builtin immutable type that can be recursive
2484 * (a tuple can be reached from itself), and that requires some subtle
2485 * magic so that it works in all cases. IOW, this is a long routine.
2486 */
2487static int
2488save_tuple(PicklerObject *self, PyObject *obj)
2489{
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002490 Py_ssize_t len, i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002491
2492 const char mark_op = MARK;
2493 const char tuple_op = TUPLE;
2494 const char pop_op = POP;
2495 const char pop_mark_op = POP_MARK;
2496 const char len2opcode[] = {EMPTY_TUPLE, TUPLE1, TUPLE2, TUPLE3};
2497
2498 if ((len = PyTuple_Size(obj)) < 0)
2499 return -1;
2500
2501 if (len == 0) {
2502 char pdata[2];
2503
2504 if (self->proto) {
2505 pdata[0] = EMPTY_TUPLE;
2506 len = 1;
2507 }
2508 else {
2509 pdata[0] = MARK;
2510 pdata[1] = TUPLE;
2511 len = 2;
2512 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002513 if (_Pickler_Write(self, pdata, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002514 return -1;
2515 return 0;
2516 }
2517
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002518 /* The tuple isn't in the memo now. If it shows up there after
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002519 * saving the tuple elements, the tuple must be recursive, in
2520 * which case we'll pop everything we put on the stack, and fetch
2521 * its value from the memo.
2522 */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002523 if (len <= 3 && self->proto >= 2) {
2524 /* Use TUPLE{1,2,3} opcodes. */
2525 if (store_tuple_elements(self, obj, len) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002526 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002527
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002528 if (PyMemoTable_Get(self->memo, obj)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002529 /* pop the len elements */
2530 for (i = 0; i < len; i++)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002531 if (_Pickler_Write(self, &pop_op, 1) < 0)
2532 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002533 /* fetch from memo */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002534 if (memo_get(self, obj) < 0)
2535 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002536
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002537 return 0;
2538 }
2539 else { /* Not recursive. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002540 if (_Pickler_Write(self, len2opcode + len, 1) < 0)
2541 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002542 }
2543 goto memoize;
2544 }
2545
2546 /* proto < 2 and len > 0, or proto >= 2 and len > 3.
2547 * Generate MARK e1 e2 ... TUPLE
2548 */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002549 if (_Pickler_Write(self, &mark_op, 1) < 0)
2550 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002551
2552 if (store_tuple_elements(self, obj, len) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002553 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002554
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002555 if (PyMemoTable_Get(self->memo, obj)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002556 /* pop the stack stuff we pushed */
2557 if (self->bin) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002558 if (_Pickler_Write(self, &pop_mark_op, 1) < 0)
2559 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002560 }
2561 else {
2562 /* Note that we pop one more than len, to remove
2563 * the MARK too.
2564 */
2565 for (i = 0; i <= len; i++)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002566 if (_Pickler_Write(self, &pop_op, 1) < 0)
2567 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002568 }
2569 /* fetch from memo */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002570 if (memo_get(self, obj) < 0)
2571 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002572
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002573 return 0;
2574 }
2575 else { /* Not recursive. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002576 if (_Pickler_Write(self, &tuple_op, 1) < 0)
2577 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002578 }
2579
2580 memoize:
2581 if (memo_put(self, obj) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002582 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002583
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002584 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002585}
2586
2587/* iter is an iterator giving items, and we batch up chunks of
2588 * MARK item item ... item APPENDS
2589 * opcode sequences. Calling code should have arranged to first create an
2590 * empty list, or list-like object, for the APPENDS to operate on.
2591 * Returns 0 on success, <0 on error.
2592 */
2593static int
2594batch_list(PicklerObject *self, PyObject *iter)
2595{
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002596 PyObject *obj = NULL;
2597 PyObject *firstitem = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002598 int i, n;
2599
2600 const char mark_op = MARK;
2601 const char append_op = APPEND;
2602 const char appends_op = APPENDS;
2603
2604 assert(iter != NULL);
2605
2606 /* XXX: I think this function could be made faster by avoiding the
2607 iterator interface and fetching objects directly from list using
2608 PyList_GET_ITEM.
2609 */
2610
2611 if (self->proto == 0) {
2612 /* APPENDS isn't available; do one at a time. */
2613 for (;;) {
2614 obj = PyIter_Next(iter);
2615 if (obj == NULL) {
2616 if (PyErr_Occurred())
2617 return -1;
2618 break;
2619 }
2620 i = save(self, obj, 0);
2621 Py_DECREF(obj);
2622 if (i < 0)
2623 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002624 if (_Pickler_Write(self, &append_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002625 return -1;
2626 }
2627 return 0;
2628 }
2629
2630 /* proto > 0: write in batches of BATCHSIZE. */
2631 do {
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002632 /* Get first item */
2633 firstitem = PyIter_Next(iter);
2634 if (firstitem == NULL) {
2635 if (PyErr_Occurred())
2636 goto error;
2637
2638 /* nothing more to add */
2639 break;
2640 }
2641
2642 /* Try to get a second item */
2643 obj = PyIter_Next(iter);
2644 if (obj == NULL) {
2645 if (PyErr_Occurred())
2646 goto error;
2647
2648 /* Only one item to write */
2649 if (save(self, firstitem, 0) < 0)
2650 goto error;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002651 if (_Pickler_Write(self, &append_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002652 goto error;
2653 Py_CLEAR(firstitem);
2654 break;
2655 }
2656
2657 /* More than one item to write */
2658
2659 /* Pump out MARK, items, APPENDS. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002660 if (_Pickler_Write(self, &mark_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002661 goto error;
2662
2663 if (save(self, firstitem, 0) < 0)
2664 goto error;
2665 Py_CLEAR(firstitem);
2666 n = 1;
2667
2668 /* Fetch and save up to BATCHSIZE items */
2669 while (obj) {
2670 if (save(self, obj, 0) < 0)
2671 goto error;
2672 Py_CLEAR(obj);
2673 n += 1;
2674
2675 if (n == BATCHSIZE)
2676 break;
2677
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002678 obj = PyIter_Next(iter);
2679 if (obj == NULL) {
2680 if (PyErr_Occurred())
2681 goto error;
2682 break;
2683 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002684 }
2685
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002686 if (_Pickler_Write(self, &appends_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002687 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002688
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002689 } while (n == BATCHSIZE);
2690 return 0;
2691
2692 error:
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002693 Py_XDECREF(firstitem);
2694 Py_XDECREF(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002695 return -1;
2696}
2697
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002698/* This is a variant of batch_list() above, specialized for lists (with no
2699 * support for list subclasses). Like batch_list(), we batch up chunks of
2700 * MARK item item ... item APPENDS
2701 * opcode sequences. Calling code should have arranged to first create an
2702 * empty list, or list-like object, for the APPENDS to operate on.
2703 * Returns 0 on success, -1 on error.
2704 *
2705 * This version is considerably faster than batch_list(), if less general.
2706 *
2707 * Note that this only works for protocols > 0.
2708 */
2709static int
2710batch_list_exact(PicklerObject *self, PyObject *obj)
2711{
2712 PyObject *item = NULL;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002713 Py_ssize_t this_batch, total;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002714
2715 const char append_op = APPEND;
2716 const char appends_op = APPENDS;
2717 const char mark_op = MARK;
2718
2719 assert(obj != NULL);
2720 assert(self->proto > 0);
2721 assert(PyList_CheckExact(obj));
2722
2723 if (PyList_GET_SIZE(obj) == 1) {
2724 item = PyList_GET_ITEM(obj, 0);
2725 if (save(self, item, 0) < 0)
2726 return -1;
2727 if (_Pickler_Write(self, &append_op, 1) < 0)
2728 return -1;
2729 return 0;
2730 }
2731
2732 /* Write in batches of BATCHSIZE. */
2733 total = 0;
2734 do {
2735 this_batch = 0;
2736 if (_Pickler_Write(self, &mark_op, 1) < 0)
2737 return -1;
2738 while (total < PyList_GET_SIZE(obj)) {
2739 item = PyList_GET_ITEM(obj, total);
2740 if (save(self, item, 0) < 0)
2741 return -1;
2742 total++;
2743 if (++this_batch == BATCHSIZE)
2744 break;
2745 }
2746 if (_Pickler_Write(self, &appends_op, 1) < 0)
2747 return -1;
2748
2749 } while (total < PyList_GET_SIZE(obj));
2750
2751 return 0;
2752}
2753
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002754static int
2755save_list(PicklerObject *self, PyObject *obj)
2756{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002757 char header[3];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002758 Py_ssize_t len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002759 int status = 0;
2760
2761 if (self->fast && !fast_save_enter(self, obj))
2762 goto error;
2763
2764 /* Create an empty list. */
2765 if (self->bin) {
2766 header[0] = EMPTY_LIST;
2767 len = 1;
2768 }
2769 else {
2770 header[0] = MARK;
2771 header[1] = LIST;
2772 len = 2;
2773 }
2774
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002775 if (_Pickler_Write(self, header, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002776 goto error;
2777
2778 /* Get list length, and bow out early if empty. */
2779 if ((len = PyList_Size(obj)) < 0)
2780 goto error;
2781
2782 if (memo_put(self, obj) < 0)
2783 goto error;
2784
2785 if (len != 0) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002786 /* Materialize the list elements. */
2787 if (PyList_CheckExact(obj) && self->proto > 0) {
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00002788 if (Py_EnterRecursiveCall(" while pickling an object"))
2789 goto error;
2790 status = batch_list_exact(self, obj);
2791 Py_LeaveRecursiveCall();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002792 } else {
2793 PyObject *iter = PyObject_GetIter(obj);
2794 if (iter == NULL)
2795 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002796
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00002797 if (Py_EnterRecursiveCall(" while pickling an object")) {
2798 Py_DECREF(iter);
2799 goto error;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002800 }
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00002801 status = batch_list(self, iter);
2802 Py_LeaveRecursiveCall();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002803 Py_DECREF(iter);
2804 }
2805 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002806 if (0) {
2807 error:
2808 status = -1;
2809 }
2810
2811 if (self->fast && !fast_save_leave(self, obj))
2812 status = -1;
2813
2814 return status;
2815}
2816
2817/* iter is an iterator giving (key, value) pairs, and we batch up chunks of
2818 * MARK key value ... key value SETITEMS
2819 * opcode sequences. Calling code should have arranged to first create an
2820 * empty dict, or dict-like object, for the SETITEMS to operate on.
2821 * Returns 0 on success, <0 on error.
2822 *
2823 * This is very much like batch_list(). The difference between saving
2824 * elements directly, and picking apart two-tuples, is so long-winded at
2825 * the C level, though, that attempts to combine these routines were too
2826 * ugly to bear.
2827 */
2828static int
2829batch_dict(PicklerObject *self, PyObject *iter)
2830{
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002831 PyObject *obj = NULL;
2832 PyObject *firstitem = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002833 int i, n;
2834
2835 const char mark_op = MARK;
2836 const char setitem_op = SETITEM;
2837 const char setitems_op = SETITEMS;
2838
2839 assert(iter != NULL);
2840
2841 if (self->proto == 0) {
2842 /* SETITEMS isn't available; do one at a time. */
2843 for (;;) {
2844 obj = PyIter_Next(iter);
2845 if (obj == NULL) {
2846 if (PyErr_Occurred())
2847 return -1;
2848 break;
2849 }
2850 if (!PyTuple_Check(obj) || PyTuple_Size(obj) != 2) {
2851 PyErr_SetString(PyExc_TypeError, "dict items "
2852 "iterator must return 2-tuples");
2853 return -1;
2854 }
2855 i = save(self, PyTuple_GET_ITEM(obj, 0), 0);
2856 if (i >= 0)
2857 i = save(self, PyTuple_GET_ITEM(obj, 1), 0);
2858 Py_DECREF(obj);
2859 if (i < 0)
2860 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002861 if (_Pickler_Write(self, &setitem_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002862 return -1;
2863 }
2864 return 0;
2865 }
2866
2867 /* proto > 0: write in batches of BATCHSIZE. */
2868 do {
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002869 /* Get first item */
2870 firstitem = PyIter_Next(iter);
2871 if (firstitem == NULL) {
2872 if (PyErr_Occurred())
2873 goto error;
2874
2875 /* nothing more to add */
2876 break;
2877 }
2878 if (!PyTuple_Check(firstitem) || PyTuple_Size(firstitem) != 2) {
2879 PyErr_SetString(PyExc_TypeError, "dict items "
2880 "iterator must return 2-tuples");
2881 goto error;
2882 }
2883
2884 /* Try to get a second item */
2885 obj = PyIter_Next(iter);
2886 if (obj == NULL) {
2887 if (PyErr_Occurred())
2888 goto error;
2889
2890 /* Only one item to write */
2891 if (save(self, PyTuple_GET_ITEM(firstitem, 0), 0) < 0)
2892 goto error;
2893 if (save(self, PyTuple_GET_ITEM(firstitem, 1), 0) < 0)
2894 goto error;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002895 if (_Pickler_Write(self, &setitem_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002896 goto error;
2897 Py_CLEAR(firstitem);
2898 break;
2899 }
2900
2901 /* More than one item to write */
2902
2903 /* Pump out MARK, items, SETITEMS. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002904 if (_Pickler_Write(self, &mark_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002905 goto error;
2906
2907 if (save(self, PyTuple_GET_ITEM(firstitem, 0), 0) < 0)
2908 goto error;
2909 if (save(self, PyTuple_GET_ITEM(firstitem, 1), 0) < 0)
2910 goto error;
2911 Py_CLEAR(firstitem);
2912 n = 1;
2913
2914 /* Fetch and save up to BATCHSIZE items */
2915 while (obj) {
2916 if (!PyTuple_Check(obj) || PyTuple_Size(obj) != 2) {
2917 PyErr_SetString(PyExc_TypeError, "dict items "
2918 "iterator must return 2-tuples");
2919 goto error;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002920 }
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002921 if (save(self, PyTuple_GET_ITEM(obj, 0), 0) < 0 ||
2922 save(self, PyTuple_GET_ITEM(obj, 1), 0) < 0)
2923 goto error;
2924 Py_CLEAR(obj);
2925 n += 1;
2926
2927 if (n == BATCHSIZE)
2928 break;
2929
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002930 obj = PyIter_Next(iter);
2931 if (obj == NULL) {
2932 if (PyErr_Occurred())
2933 goto error;
2934 break;
2935 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002936 }
2937
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002938 if (_Pickler_Write(self, &setitems_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002939 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002940
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002941 } while (n == BATCHSIZE);
2942 return 0;
2943
2944 error:
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002945 Py_XDECREF(firstitem);
2946 Py_XDECREF(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002947 return -1;
2948}
2949
Collin Winter5c9b02d2009-05-25 05:43:30 +00002950/* This is a variant of batch_dict() above that specializes for dicts, with no
2951 * support for dict subclasses. Like batch_dict(), we batch up chunks of
2952 * MARK key value ... key value SETITEMS
2953 * opcode sequences. Calling code should have arranged to first create an
2954 * empty dict, or dict-like object, for the SETITEMS to operate on.
2955 * Returns 0 on success, -1 on error.
2956 *
2957 * Note that this currently doesn't work for protocol 0.
2958 */
2959static int
2960batch_dict_exact(PicklerObject *self, PyObject *obj)
2961{
2962 PyObject *key = NULL, *value = NULL;
2963 int i;
2964 Py_ssize_t dict_size, ppos = 0;
2965
Alexandre Vassalottif70b1292009-05-25 18:00:52 +00002966 const char mark_op = MARK;
2967 const char setitem_op = SETITEM;
2968 const char setitems_op = SETITEMS;
Collin Winter5c9b02d2009-05-25 05:43:30 +00002969
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02002970 assert(obj != NULL && PyDict_CheckExact(obj));
Collin Winter5c9b02d2009-05-25 05:43:30 +00002971 assert(self->proto > 0);
2972
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02002973 dict_size = PyDict_GET_SIZE(obj);
Collin Winter5c9b02d2009-05-25 05:43:30 +00002974
2975 /* Special-case len(d) == 1 to save space. */
2976 if (dict_size == 1) {
2977 PyDict_Next(obj, &ppos, &key, &value);
2978 if (save(self, key, 0) < 0)
2979 return -1;
2980 if (save(self, value, 0) < 0)
2981 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002982 if (_Pickler_Write(self, &setitem_op, 1) < 0)
Collin Winter5c9b02d2009-05-25 05:43:30 +00002983 return -1;
2984 return 0;
2985 }
2986
2987 /* Write in batches of BATCHSIZE. */
2988 do {
2989 i = 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002990 if (_Pickler_Write(self, &mark_op, 1) < 0)
Collin Winter5c9b02d2009-05-25 05:43:30 +00002991 return -1;
2992 while (PyDict_Next(obj, &ppos, &key, &value)) {
2993 if (save(self, key, 0) < 0)
2994 return -1;
2995 if (save(self, value, 0) < 0)
2996 return -1;
2997 if (++i == BATCHSIZE)
2998 break;
2999 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003000 if (_Pickler_Write(self, &setitems_op, 1) < 0)
Collin Winter5c9b02d2009-05-25 05:43:30 +00003001 return -1;
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02003002 if (PyDict_GET_SIZE(obj) != dict_size) {
Collin Winter5c9b02d2009-05-25 05:43:30 +00003003 PyErr_Format(
3004 PyExc_RuntimeError,
3005 "dictionary changed size during iteration");
3006 return -1;
3007 }
3008
3009 } while (i == BATCHSIZE);
3010 return 0;
3011}
3012
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003013static int
3014save_dict(PicklerObject *self, PyObject *obj)
3015{
3016 PyObject *items, *iter;
3017 char header[3];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02003018 Py_ssize_t len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003019 int status = 0;
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02003020 assert(PyDict_Check(obj));
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003021
3022 if (self->fast && !fast_save_enter(self, obj))
3023 goto error;
3024
3025 /* Create an empty dict. */
3026 if (self->bin) {
3027 header[0] = EMPTY_DICT;
3028 len = 1;
3029 }
3030 else {
3031 header[0] = MARK;
3032 header[1] = DICT;
3033 len = 2;
3034 }
3035
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003036 if (_Pickler_Write(self, header, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003037 goto error;
3038
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003039 if (memo_put(self, obj) < 0)
3040 goto error;
3041
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02003042 if (PyDict_GET_SIZE(obj)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003043 /* Save the dict items. */
Collin Winter5c9b02d2009-05-25 05:43:30 +00003044 if (PyDict_CheckExact(obj) && self->proto > 0) {
3045 /* We can take certain shortcuts if we know this is a dict and
3046 not a dict subclass. */
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00003047 if (Py_EnterRecursiveCall(" while pickling an object"))
3048 goto error;
3049 status = batch_dict_exact(self, obj);
3050 Py_LeaveRecursiveCall();
Collin Winter5c9b02d2009-05-25 05:43:30 +00003051 } else {
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02003052 _Py_IDENTIFIER(items);
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02003053
Victor Stinnerad8c83a2016-09-05 17:53:15 -07003054 items = _PyObject_CallMethodId(obj, &PyId_items, NULL);
Collin Winter5c9b02d2009-05-25 05:43:30 +00003055 if (items == NULL)
3056 goto error;
3057 iter = PyObject_GetIter(items);
3058 Py_DECREF(items);
3059 if (iter == NULL)
3060 goto error;
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00003061 if (Py_EnterRecursiveCall(" while pickling an object")) {
3062 Py_DECREF(iter);
3063 goto error;
3064 }
Collin Winter5c9b02d2009-05-25 05:43:30 +00003065 status = batch_dict(self, iter);
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00003066 Py_LeaveRecursiveCall();
Collin Winter5c9b02d2009-05-25 05:43:30 +00003067 Py_DECREF(iter);
3068 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003069 }
3070
3071 if (0) {
3072 error:
3073 status = -1;
3074 }
3075
3076 if (self->fast && !fast_save_leave(self, obj))
3077 status = -1;
3078
3079 return status;
3080}
3081
3082static int
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003083save_set(PicklerObject *self, PyObject *obj)
3084{
3085 PyObject *item;
3086 int i;
3087 Py_ssize_t set_size, ppos = 0;
3088 Py_hash_t hash;
3089
3090 const char empty_set_op = EMPTY_SET;
3091 const char mark_op = MARK;
3092 const char additems_op = ADDITEMS;
3093
3094 if (self->proto < 4) {
3095 PyObject *items;
3096 PyObject *reduce_value;
3097 int status;
3098
3099 items = PySequence_List(obj);
3100 if (items == NULL) {
3101 return -1;
3102 }
3103 reduce_value = Py_BuildValue("(O(O))", (PyObject*)&PySet_Type, items);
3104 Py_DECREF(items);
3105 if (reduce_value == NULL) {
3106 return -1;
3107 }
3108 /* save_reduce() will memoize the object automatically. */
3109 status = save_reduce(self, reduce_value, obj);
3110 Py_DECREF(reduce_value);
3111 return status;
3112 }
3113
3114 if (_Pickler_Write(self, &empty_set_op, 1) < 0)
3115 return -1;
3116
3117 if (memo_put(self, obj) < 0)
3118 return -1;
3119
3120 set_size = PySet_GET_SIZE(obj);
3121 if (set_size == 0)
3122 return 0; /* nothing to do */
3123
3124 /* Write in batches of BATCHSIZE. */
3125 do {
3126 i = 0;
3127 if (_Pickler_Write(self, &mark_op, 1) < 0)
3128 return -1;
3129 while (_PySet_NextEntry(obj, &ppos, &item, &hash)) {
3130 if (save(self, item, 0) < 0)
3131 return -1;
3132 if (++i == BATCHSIZE)
3133 break;
3134 }
3135 if (_Pickler_Write(self, &additems_op, 1) < 0)
3136 return -1;
3137 if (PySet_GET_SIZE(obj) != set_size) {
3138 PyErr_Format(
3139 PyExc_RuntimeError,
3140 "set changed size during iteration");
3141 return -1;
3142 }
3143 } while (i == BATCHSIZE);
3144
3145 return 0;
3146}
3147
3148static int
3149save_frozenset(PicklerObject *self, PyObject *obj)
3150{
3151 PyObject *iter;
3152
3153 const char mark_op = MARK;
3154 const char frozenset_op = FROZENSET;
3155
3156 if (self->fast && !fast_save_enter(self, obj))
3157 return -1;
3158
3159 if (self->proto < 4) {
3160 PyObject *items;
3161 PyObject *reduce_value;
3162 int status;
3163
3164 items = PySequence_List(obj);
3165 if (items == NULL) {
3166 return -1;
3167 }
3168 reduce_value = Py_BuildValue("(O(O))", (PyObject*)&PyFrozenSet_Type,
3169 items);
3170 Py_DECREF(items);
3171 if (reduce_value == NULL) {
3172 return -1;
3173 }
3174 /* save_reduce() will memoize the object automatically. */
3175 status = save_reduce(self, reduce_value, obj);
3176 Py_DECREF(reduce_value);
3177 return status;
3178 }
3179
3180 if (_Pickler_Write(self, &mark_op, 1) < 0)
3181 return -1;
3182
3183 iter = PyObject_GetIter(obj);
Christian Heimesb3d3ee42013-11-23 21:01:40 +01003184 if (iter == NULL) {
Christian Heimes74d8d632013-11-23 21:05:31 +01003185 return -1;
Christian Heimesb3d3ee42013-11-23 21:01:40 +01003186 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003187 for (;;) {
3188 PyObject *item;
3189
3190 item = PyIter_Next(iter);
3191 if (item == NULL) {
3192 if (PyErr_Occurred()) {
3193 Py_DECREF(iter);
3194 return -1;
3195 }
3196 break;
3197 }
3198 if (save(self, item, 0) < 0) {
3199 Py_DECREF(item);
3200 Py_DECREF(iter);
3201 return -1;
3202 }
3203 Py_DECREF(item);
3204 }
3205 Py_DECREF(iter);
3206
3207 /* If the object is already in the memo, this means it is
3208 recursive. In this case, throw away everything we put on the
3209 stack, and fetch the object back from the memo. */
3210 if (PyMemoTable_Get(self->memo, obj)) {
3211 const char pop_mark_op = POP_MARK;
3212
3213 if (_Pickler_Write(self, &pop_mark_op, 1) < 0)
3214 return -1;
3215 if (memo_get(self, obj) < 0)
3216 return -1;
3217 return 0;
3218 }
3219
3220 if (_Pickler_Write(self, &frozenset_op, 1) < 0)
3221 return -1;
3222 if (memo_put(self, obj) < 0)
3223 return -1;
3224
3225 return 0;
3226}
3227
3228static int
3229fix_imports(PyObject **module_name, PyObject **global_name)
3230{
3231 PyObject *key;
3232 PyObject *item;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003233 PickleState *st = _Pickle_GetGlobalState();
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003234
3235 key = PyTuple_Pack(2, *module_name, *global_name);
3236 if (key == NULL)
3237 return -1;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003238 item = PyDict_GetItemWithError(st->name_mapping_3to2, key);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003239 Py_DECREF(key);
3240 if (item) {
3241 PyObject *fixed_module_name;
3242 PyObject *fixed_global_name;
3243
3244 if (!PyTuple_Check(item) || PyTuple_GET_SIZE(item) != 2) {
3245 PyErr_Format(PyExc_RuntimeError,
3246 "_compat_pickle.REVERSE_NAME_MAPPING values "
3247 "should be 2-tuples, not %.200s",
3248 Py_TYPE(item)->tp_name);
3249 return -1;
3250 }
3251 fixed_module_name = PyTuple_GET_ITEM(item, 0);
3252 fixed_global_name = PyTuple_GET_ITEM(item, 1);
3253 if (!PyUnicode_Check(fixed_module_name) ||
3254 !PyUnicode_Check(fixed_global_name)) {
3255 PyErr_Format(PyExc_RuntimeError,
3256 "_compat_pickle.REVERSE_NAME_MAPPING values "
3257 "should be pairs of str, not (%.200s, %.200s)",
3258 Py_TYPE(fixed_module_name)->tp_name,
3259 Py_TYPE(fixed_global_name)->tp_name);
3260 return -1;
3261 }
3262
3263 Py_CLEAR(*module_name);
3264 Py_CLEAR(*global_name);
3265 Py_INCREF(fixed_module_name);
3266 Py_INCREF(fixed_global_name);
3267 *module_name = fixed_module_name;
3268 *global_name = fixed_global_name;
Serhiy Storchakabfe18242015-03-31 13:12:37 +03003269 return 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003270 }
3271 else if (PyErr_Occurred()) {
3272 return -1;
3273 }
3274
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003275 item = PyDict_GetItemWithError(st->import_mapping_3to2, *module_name);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003276 if (item) {
3277 if (!PyUnicode_Check(item)) {
3278 PyErr_Format(PyExc_RuntimeError,
3279 "_compat_pickle.REVERSE_IMPORT_MAPPING values "
3280 "should be strings, not %.200s",
3281 Py_TYPE(item)->tp_name);
3282 return -1;
3283 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003284 Py_INCREF(item);
Serhiy Storchaka48842712016-04-06 09:45:48 +03003285 Py_XSETREF(*module_name, item);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003286 }
3287 else if (PyErr_Occurred()) {
3288 return -1;
3289 }
3290
3291 return 0;
3292}
3293
3294static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003295save_global(PicklerObject *self, PyObject *obj, PyObject *name)
3296{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003297 PyObject *global_name = NULL;
3298 PyObject *module_name = NULL;
3299 PyObject *module = NULL;
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003300 PyObject *parent = NULL;
3301 PyObject *dotted_path = NULL;
3302 PyObject *lastname = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003303 PyObject *cls;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003304 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003305 int status = 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003306 _Py_IDENTIFIER(__name__);
3307 _Py_IDENTIFIER(__qualname__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003308
3309 const char global_op = GLOBAL;
3310
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003311 if (name) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003312 Py_INCREF(name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003313 global_name = name;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003314 }
3315 else {
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003316 if (_PyObject_LookupAttrId(obj, &PyId___qualname__, &global_name) < 0)
3317 goto error;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003318 if (global_name == NULL) {
3319 global_name = _PyObject_GetAttrId(obj, &PyId___name__);
3320 if (global_name == NULL)
3321 goto error;
3322 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003323 }
3324
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003325 dotted_path = get_dotted_path(module, global_name);
3326 if (dotted_path == NULL)
3327 goto error;
3328 module_name = whichmodule(obj, dotted_path);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003329 if (module_name == NULL)
3330 goto error;
3331
3332 /* XXX: Change to use the import C API directly with level=0 to disallow
3333 relative imports.
3334
3335 XXX: PyImport_ImportModuleLevel could be used. However, this bypasses
3336 builtins.__import__. Therefore, _pickle, unlike pickle.py, will ignore
3337 custom import functions (IMHO, this would be a nice security
3338 feature). The import C API would need to be extended to support the
3339 extra parameters of __import__ to fix that. */
3340 module = PyImport_Import(module_name);
3341 if (module == NULL) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003342 PyErr_Format(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003343 "Can't pickle %R: import of module %R failed",
3344 obj, module_name);
3345 goto error;
3346 }
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003347 lastname = PyList_GET_ITEM(dotted_path, PyList_GET_SIZE(dotted_path)-1);
3348 Py_INCREF(lastname);
3349 cls = get_deep_attribute(module, dotted_path, &parent);
3350 Py_CLEAR(dotted_path);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003351 if (cls == NULL) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003352 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003353 "Can't pickle %R: attribute lookup %S on %S failed",
3354 obj, global_name, module_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003355 goto error;
3356 }
3357 if (cls != obj) {
3358 Py_DECREF(cls);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003359 PyErr_Format(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003360 "Can't pickle %R: it's not the same object as %S.%S",
3361 obj, module_name, global_name);
3362 goto error;
3363 }
3364 Py_DECREF(cls);
3365
3366 if (self->proto >= 2) {
3367 /* See whether this is in the extension registry, and if
3368 * so generate an EXT opcode.
3369 */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003370 PyObject *extension_key;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003371 PyObject *code_obj; /* extension code as Python object */
3372 long code; /* extension code as C value */
3373 char pdata[5];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02003374 Py_ssize_t n;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003375
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003376 extension_key = PyTuple_Pack(2, module_name, global_name);
3377 if (extension_key == NULL) {
3378 goto error;
3379 }
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08003380 code_obj = PyDict_GetItemWithError(st->extension_registry,
3381 extension_key);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003382 Py_DECREF(extension_key);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003383 /* The object is not registered in the extension registry.
3384 This is the most likely code path. */
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08003385 if (code_obj == NULL) {
3386 if (PyErr_Occurred()) {
3387 goto error;
3388 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003389 goto gen_global;
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08003390 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003391
3392 /* XXX: pickle.py doesn't check neither the type, nor the range
3393 of the value returned by the extension_registry. It should for
3394 consistency. */
3395
3396 /* Verify code_obj has the right type and value. */
3397 if (!PyLong_Check(code_obj)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003398 PyErr_Format(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003399 "Can't pickle %R: extension code %R isn't an integer",
3400 obj, code_obj);
3401 goto error;
3402 }
3403 code = PyLong_AS_LONG(code_obj);
3404 if (code <= 0 || code > 0x7fffffffL) {
Antoine Pitrou82be19f2011-08-29 23:09:33 +02003405 if (!PyErr_Occurred())
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003406 PyErr_Format(st->PicklingError, "Can't pickle %R: extension "
3407 "code %ld is out of range", obj, code);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003408 goto error;
3409 }
3410
3411 /* Generate an EXT opcode. */
3412 if (code <= 0xff) {
3413 pdata[0] = EXT1;
3414 pdata[1] = (unsigned char)code;
3415 n = 2;
3416 }
3417 else if (code <= 0xffff) {
3418 pdata[0] = EXT2;
3419 pdata[1] = (unsigned char)(code & 0xff);
3420 pdata[2] = (unsigned char)((code >> 8) & 0xff);
3421 n = 3;
3422 }
3423 else {
3424 pdata[0] = EXT4;
3425 pdata[1] = (unsigned char)(code & 0xff);
3426 pdata[2] = (unsigned char)((code >> 8) & 0xff);
3427 pdata[3] = (unsigned char)((code >> 16) & 0xff);
3428 pdata[4] = (unsigned char)((code >> 24) & 0xff);
3429 n = 5;
3430 }
3431
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003432 if (_Pickler_Write(self, pdata, n) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003433 goto error;
3434 }
3435 else {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003436 gen_global:
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003437 if (parent == module) {
3438 Py_INCREF(lastname);
3439 Py_DECREF(global_name);
3440 global_name = lastname;
3441 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003442 if (self->proto >= 4) {
3443 const char stack_global_op = STACK_GLOBAL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003444
Christian Heimese8b1ba12013-11-23 21:13:39 +01003445 if (save(self, module_name, 0) < 0)
3446 goto error;
3447 if (save(self, global_name, 0) < 0)
3448 goto error;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003449
3450 if (_Pickler_Write(self, &stack_global_op, 1) < 0)
3451 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003452 }
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003453 else if (parent != module) {
3454 PickleState *st = _Pickle_GetGlobalState();
3455 PyObject *reduce_value = Py_BuildValue("(O(OO))",
3456 st->getattr, parent, lastname);
Alexey Izbyshevf8c06b02018-08-22 07:51:25 +03003457 if (reduce_value == NULL)
3458 goto error;
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003459 status = save_reduce(self, reduce_value, NULL);
3460 Py_DECREF(reduce_value);
3461 if (status < 0)
3462 goto error;
3463 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003464 else {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003465 /* Generate a normal global opcode if we are using a pickle
3466 protocol < 4, or if the object is not registered in the
3467 extension registry. */
3468 PyObject *encoded;
3469 PyObject *(*unicode_encoder)(PyObject *);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003470
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003471 if (_Pickler_Write(self, &global_op, 1) < 0)
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003472 goto error;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003473
3474 /* For protocol < 3 and if the user didn't request against doing
3475 so, we convert module names to the old 2.x module names. */
3476 if (self->proto < 3 && self->fix_imports) {
3477 if (fix_imports(&module_name, &global_name) < 0) {
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003478 goto error;
3479 }
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003480 }
3481
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003482 /* Since Python 3.0 now supports non-ASCII identifiers, we encode
3483 both the module name and the global name using UTF-8. We do so
3484 only when we are using the pickle protocol newer than version
3485 3. This is to ensure compatibility with older Unpickler running
3486 on Python 2.x. */
3487 if (self->proto == 3) {
3488 unicode_encoder = PyUnicode_AsUTF8String;
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003489 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003490 else {
3491 unicode_encoder = PyUnicode_AsASCIIString;
3492 }
3493 encoded = unicode_encoder(module_name);
3494 if (encoded == NULL) {
3495 if (PyErr_ExceptionMatches(PyExc_UnicodeEncodeError))
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003496 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003497 "can't pickle module identifier '%S' using "
3498 "pickle protocol %i",
3499 module_name, self->proto);
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003500 goto error;
3501 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003502 if (_Pickler_Write(self, PyBytes_AS_STRING(encoded),
3503 PyBytes_GET_SIZE(encoded)) < 0) {
3504 Py_DECREF(encoded);
3505 goto error;
3506 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003507 Py_DECREF(encoded);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003508 if(_Pickler_Write(self, "\n", 1) < 0)
3509 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003510
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003511 /* Save the name of the module. */
3512 encoded = unicode_encoder(global_name);
3513 if (encoded == NULL) {
3514 if (PyErr_ExceptionMatches(PyExc_UnicodeEncodeError))
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003515 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003516 "can't pickle global identifier '%S' using "
3517 "pickle protocol %i",
3518 global_name, self->proto);
3519 goto error;
3520 }
3521 if (_Pickler_Write(self, PyBytes_AS_STRING(encoded),
3522 PyBytes_GET_SIZE(encoded)) < 0) {
3523 Py_DECREF(encoded);
3524 goto error;
3525 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003526 Py_DECREF(encoded);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003527 if (_Pickler_Write(self, "\n", 1) < 0)
3528 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003529 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003530 /* Memoize the object. */
3531 if (memo_put(self, obj) < 0)
3532 goto error;
3533 }
3534
3535 if (0) {
3536 error:
3537 status = -1;
3538 }
3539 Py_XDECREF(module_name);
3540 Py_XDECREF(global_name);
3541 Py_XDECREF(module);
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003542 Py_XDECREF(parent);
3543 Py_XDECREF(dotted_path);
3544 Py_XDECREF(lastname);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003545
3546 return status;
3547}
3548
3549static int
Alexandre Vassalotti19b6fa62013-11-30 16:06:39 -08003550save_singleton_type(PicklerObject *self, PyObject *obj, PyObject *singleton)
3551{
3552 PyObject *reduce_value;
3553 int status;
3554
3555 reduce_value = Py_BuildValue("O(O)", &PyType_Type, singleton);
3556 if (reduce_value == NULL) {
3557 return -1;
3558 }
3559 status = save_reduce(self, reduce_value, obj);
3560 Py_DECREF(reduce_value);
3561 return status;
3562}
3563
3564static int
3565save_type(PicklerObject *self, PyObject *obj)
3566{
Alexandre Vassalotti65846c62013-11-30 17:55:48 -08003567 if (obj == (PyObject *)&_PyNone_Type) {
Alexandre Vassalotti19b6fa62013-11-30 16:06:39 -08003568 return save_singleton_type(self, obj, Py_None);
3569 }
3570 else if (obj == (PyObject *)&PyEllipsis_Type) {
3571 return save_singleton_type(self, obj, Py_Ellipsis);
3572 }
Alexandre Vassalotti65846c62013-11-30 17:55:48 -08003573 else if (obj == (PyObject *)&_PyNotImplemented_Type) {
Alexandre Vassalotti19b6fa62013-11-30 16:06:39 -08003574 return save_singleton_type(self, obj, Py_NotImplemented);
3575 }
3576 return save_global(self, obj, NULL);
3577}
3578
3579static int
Serhiy Storchaka986375e2017-11-30 22:48:31 +02003580save_pers(PicklerObject *self, PyObject *obj)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003581{
3582 PyObject *pid = NULL;
3583 int status = 0;
3584
3585 const char persid_op = PERSID;
3586 const char binpersid_op = BINPERSID;
3587
Serhiy Storchaka986375e2017-11-30 22:48:31 +02003588 pid = call_method(self->pers_func, self->pers_func_self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003589 if (pid == NULL)
3590 return -1;
3591
3592 if (pid != Py_None) {
3593 if (self->bin) {
3594 if (save(self, pid, 1) < 0 ||
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003595 _Pickler_Write(self, &binpersid_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003596 goto error;
3597 }
3598 else {
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003599 PyObject *pid_str;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003600
3601 pid_str = PyObject_Str(pid);
3602 if (pid_str == NULL)
3603 goto error;
3604
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003605 /* XXX: Should it check whether the pid contains embedded
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003606 newlines? */
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003607 if (!PyUnicode_IS_ASCII(pid_str)) {
3608 PyErr_SetString(_Pickle_GetGlobalState()->PicklingError,
3609 "persistent IDs in protocol 0 must be "
3610 "ASCII strings");
3611 Py_DECREF(pid_str);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003612 goto error;
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003613 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003614
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003615 if (_Pickler_Write(self, &persid_op, 1) < 0 ||
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003616 _Pickler_Write(self, PyUnicode_DATA(pid_str),
3617 PyUnicode_GET_LENGTH(pid_str)) < 0 ||
3618 _Pickler_Write(self, "\n", 1) < 0) {
3619 Py_DECREF(pid_str);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003620 goto error;
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003621 }
3622 Py_DECREF(pid_str);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003623 }
3624 status = 1;
3625 }
3626
3627 if (0) {
3628 error:
3629 status = -1;
3630 }
3631 Py_XDECREF(pid);
3632
3633 return status;
3634}
3635
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003636static PyObject *
3637get_class(PyObject *obj)
3638{
3639 PyObject *cls;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003640 _Py_IDENTIFIER(__class__);
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003641
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003642 if (_PyObject_LookupAttrId(obj, &PyId___class__, &cls) == 0) {
3643 cls = (PyObject *) Py_TYPE(obj);
3644 Py_INCREF(cls);
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003645 }
3646 return cls;
3647}
3648
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003649/* We're saving obj, and args is the 2-thru-5 tuple returned by the
3650 * appropriate __reduce__ method for obj.
3651 */
3652static int
3653save_reduce(PicklerObject *self, PyObject *args, PyObject *obj)
3654{
3655 PyObject *callable;
3656 PyObject *argtup;
3657 PyObject *state = NULL;
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003658 PyObject *listitems = Py_None;
3659 PyObject *dictitems = Py_None;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003660 PickleState *st = _Pickle_GetGlobalState();
Hirokazu Yamamotob46a6332008-11-04 00:35:10 +00003661 Py_ssize_t size;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003662 int use_newobj = 0, use_newobj_ex = 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003663
3664 const char reduce_op = REDUCE;
3665 const char build_op = BUILD;
3666 const char newobj_op = NEWOBJ;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003667 const char newobj_ex_op = NEWOBJ_EX;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003668
Hirokazu Yamamotob46a6332008-11-04 00:35:10 +00003669 size = PyTuple_Size(args);
3670 if (size < 2 || size > 5) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003671 PyErr_SetString(st->PicklingError, "tuple returned by "
Hirokazu Yamamotob46a6332008-11-04 00:35:10 +00003672 "__reduce__ must contain 2 through 5 elements");
3673 return -1;
3674 }
3675
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003676 if (!PyArg_UnpackTuple(args, "save_reduce", 2, 5,
3677 &callable, &argtup, &state, &listitems, &dictitems))
3678 return -1;
3679
3680 if (!PyCallable_Check(callable)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003681 PyErr_SetString(st->PicklingError, "first item of the tuple "
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003682 "returned by __reduce__ must be callable");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003683 return -1;
3684 }
3685 if (!PyTuple_Check(argtup)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003686 PyErr_SetString(st->PicklingError, "second item of the tuple "
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003687 "returned by __reduce__ must be a tuple");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003688 return -1;
3689 }
3690
3691 if (state == Py_None)
3692 state = NULL;
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003693
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003694 if (listitems == Py_None)
3695 listitems = NULL;
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003696 else if (!PyIter_Check(listitems)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003697 PyErr_Format(st->PicklingError, "fourth element of the tuple "
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003698 "returned by __reduce__ must be an iterator, not %s",
3699 Py_TYPE(listitems)->tp_name);
3700 return -1;
3701 }
3702
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003703 if (dictitems == Py_None)
3704 dictitems = NULL;
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003705 else if (!PyIter_Check(dictitems)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003706 PyErr_Format(st->PicklingError, "fifth element of the tuple "
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003707 "returned by __reduce__ must be an iterator, not %s",
3708 Py_TYPE(dictitems)->tp_name);
3709 return -1;
3710 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003711
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003712 if (self->proto >= 2) {
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003713 PyObject *name;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003714 _Py_IDENTIFIER(__name__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003715
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003716 if (_PyObject_LookupAttrId(callable, &PyId___name__, &name) < 0) {
3717 return -1;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003718 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003719 if (name != NULL && PyUnicode_Check(name)) {
Serhiy Storchaka0d554d72015-10-10 22:42:18 +03003720 _Py_IDENTIFIER(__newobj_ex__);
Serhiy Storchakaf0f35a62017-01-09 10:09:43 +02003721 use_newobj_ex = _PyUnicode_EqualToASCIIId(
3722 name, &PyId___newobj_ex__);
Serhiy Storchaka707b5cc2014-12-16 19:43:46 +02003723 if (!use_newobj_ex) {
3724 _Py_IDENTIFIER(__newobj__);
Serhiy Storchaka9937d902017-01-09 10:04:34 +02003725 use_newobj = _PyUnicode_EqualToASCIIId(name, &PyId___newobj__);
Serhiy Storchaka707b5cc2014-12-16 19:43:46 +02003726 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003727 }
Serhiy Storchaka707b5cc2014-12-16 19:43:46 +02003728 Py_XDECREF(name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003729 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003730
3731 if (use_newobj_ex) {
3732 PyObject *cls;
3733 PyObject *args;
3734 PyObject *kwargs;
3735
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003736 if (PyTuple_GET_SIZE(argtup) != 3) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003737 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003738 "length of the NEWOBJ_EX argument tuple must be "
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003739 "exactly 3, not %zd", PyTuple_GET_SIZE(argtup));
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003740 return -1;
3741 }
3742
3743 cls = PyTuple_GET_ITEM(argtup, 0);
3744 if (!PyType_Check(cls)) {
Larry Hastings61272b72014-01-07 12:41:53 -08003745 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003746 "first item from NEWOBJ_EX argument tuple must "
3747 "be a class, not %.200s", Py_TYPE(cls)->tp_name);
3748 return -1;
3749 }
3750 args = PyTuple_GET_ITEM(argtup, 1);
3751 if (!PyTuple_Check(args)) {
Larry Hastings61272b72014-01-07 12:41:53 -08003752 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003753 "second item from NEWOBJ_EX argument tuple must "
3754 "be a tuple, not %.200s", Py_TYPE(args)->tp_name);
3755 return -1;
3756 }
3757 kwargs = PyTuple_GET_ITEM(argtup, 2);
3758 if (!PyDict_Check(kwargs)) {
Larry Hastings61272b72014-01-07 12:41:53 -08003759 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003760 "third item from NEWOBJ_EX argument tuple must "
3761 "be a dict, not %.200s", Py_TYPE(kwargs)->tp_name);
3762 return -1;
3763 }
3764
Serhiy Storchaka0d554d72015-10-10 22:42:18 +03003765 if (self->proto >= 4) {
3766 if (save(self, cls, 0) < 0 ||
3767 save(self, args, 0) < 0 ||
3768 save(self, kwargs, 0) < 0 ||
3769 _Pickler_Write(self, &newobj_ex_op, 1) < 0) {
3770 return -1;
3771 }
3772 }
3773 else {
3774 PyObject *newargs;
3775 PyObject *cls_new;
3776 Py_ssize_t i;
3777 _Py_IDENTIFIER(__new__);
3778
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003779 newargs = PyTuple_New(PyTuple_GET_SIZE(args) + 2);
Serhiy Storchaka0d554d72015-10-10 22:42:18 +03003780 if (newargs == NULL)
3781 return -1;
3782
3783 cls_new = _PyObject_GetAttrId(cls, &PyId___new__);
3784 if (cls_new == NULL) {
3785 Py_DECREF(newargs);
3786 return -1;
3787 }
3788 PyTuple_SET_ITEM(newargs, 0, cls_new);
3789 Py_INCREF(cls);
3790 PyTuple_SET_ITEM(newargs, 1, cls);
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003791 for (i = 0; i < PyTuple_GET_SIZE(args); i++) {
Serhiy Storchaka0d554d72015-10-10 22:42:18 +03003792 PyObject *item = PyTuple_GET_ITEM(args, i);
3793 Py_INCREF(item);
3794 PyTuple_SET_ITEM(newargs, i + 2, item);
3795 }
3796
3797 callable = PyObject_Call(st->partial, newargs, kwargs);
3798 Py_DECREF(newargs);
3799 if (callable == NULL)
3800 return -1;
3801
3802 newargs = PyTuple_New(0);
3803 if (newargs == NULL) {
3804 Py_DECREF(callable);
3805 return -1;
3806 }
3807
3808 if (save(self, callable, 0) < 0 ||
3809 save(self, newargs, 0) < 0 ||
3810 _Pickler_Write(self, &reduce_op, 1) < 0) {
3811 Py_DECREF(newargs);
3812 Py_DECREF(callable);
3813 return -1;
3814 }
3815 Py_DECREF(newargs);
3816 Py_DECREF(callable);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003817 }
3818 }
3819 else if (use_newobj) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003820 PyObject *cls;
3821 PyObject *newargtup;
3822 PyObject *obj_class;
3823 int p;
3824
3825 /* Sanity checks. */
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003826 if (PyTuple_GET_SIZE(argtup) < 1) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003827 PyErr_SetString(st->PicklingError, "__newobj__ arglist is empty");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003828 return -1;
3829 }
3830
3831 cls = PyTuple_GET_ITEM(argtup, 0);
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003832 if (!PyType_Check(cls)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003833 PyErr_SetString(st->PicklingError, "args[0] from "
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003834 "__newobj__ args is not a type");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003835 return -1;
3836 }
3837
3838 if (obj != NULL) {
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003839 obj_class = get_class(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003840 p = obj_class != cls; /* true iff a problem */
3841 Py_DECREF(obj_class);
3842 if (p) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003843 PyErr_SetString(st->PicklingError, "args[0] from "
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003844 "__newobj__ args has the wrong class");
3845 return -1;
3846 }
3847 }
3848 /* XXX: These calls save() are prone to infinite recursion. Imagine
3849 what happen if the value returned by the __reduce__() method of
3850 some extension type contains another object of the same type. Ouch!
3851
3852 Here is a quick example, that I ran into, to illustrate what I
3853 mean:
3854
3855 >>> import pickle, copyreg
3856 >>> copyreg.dispatch_table.pop(complex)
3857 >>> pickle.dumps(1+2j)
3858 Traceback (most recent call last):
3859 ...
Yury Selivanovf488fb42015-07-03 01:04:23 -04003860 RecursionError: maximum recursion depth exceeded
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003861
3862 Removing the complex class from copyreg.dispatch_table made the
3863 __reduce_ex__() method emit another complex object:
3864
3865 >>> (1+1j).__reduce_ex__(2)
3866 (<function __newobj__ at 0xb7b71c3c>,
3867 (<class 'complex'>, (1+1j)), None, None, None)
3868
3869 Thus when save() was called on newargstup (the 2nd item) recursion
3870 ensued. Of course, the bug was in the complex class which had a
3871 broken __getnewargs__() that emitted another complex object. But,
3872 the point, here, is it is quite easy to end up with a broken reduce
3873 function. */
3874
3875 /* Save the class and its __new__ arguments. */
3876 if (save(self, cls, 0) < 0)
3877 return -1;
3878
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003879 newargtup = PyTuple_GetSlice(argtup, 1, PyTuple_GET_SIZE(argtup));
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003880 if (newargtup == NULL)
3881 return -1;
3882
3883 p = save(self, newargtup, 0);
3884 Py_DECREF(newargtup);
3885 if (p < 0)
3886 return -1;
3887
3888 /* Add NEWOBJ opcode. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003889 if (_Pickler_Write(self, &newobj_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003890 return -1;
3891 }
3892 else { /* Not using NEWOBJ. */
3893 if (save(self, callable, 0) < 0 ||
3894 save(self, argtup, 0) < 0 ||
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003895 _Pickler_Write(self, &reduce_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003896 return -1;
3897 }
3898
3899 /* obj can be NULL when save_reduce() is used directly. A NULL obj means
3900 the caller do not want to memoize the object. Not particularly useful,
3901 but that is to mimic the behavior save_reduce() in pickle.py when
3902 obj is None. */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003903 if (obj != NULL) {
3904 /* If the object is already in the memo, this means it is
3905 recursive. In this case, throw away everything we put on the
3906 stack, and fetch the object back from the memo. */
3907 if (PyMemoTable_Get(self->memo, obj)) {
3908 const char pop_op = POP;
3909
3910 if (_Pickler_Write(self, &pop_op, 1) < 0)
3911 return -1;
3912 if (memo_get(self, obj) < 0)
3913 return -1;
3914
3915 return 0;
3916 }
3917 else if (memo_put(self, obj) < 0)
3918 return -1;
3919 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003920
3921 if (listitems && batch_list(self, listitems) < 0)
3922 return -1;
3923
3924 if (dictitems && batch_dict(self, dictitems) < 0)
3925 return -1;
3926
3927 if (state) {
Victor Stinner121aab42011-09-29 23:40:53 +02003928 if (save(self, state, 0) < 0 ||
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003929 _Pickler_Write(self, &build_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003930 return -1;
3931 }
3932
3933 return 0;
3934}
3935
3936static int
3937save(PicklerObject *self, PyObject *obj, int pers_save)
3938{
3939 PyTypeObject *type;
3940 PyObject *reduce_func = NULL;
3941 PyObject *reduce_value = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003942 int status = 0;
3943
Alexandre Vassalottib6a2f2a2013-11-23 20:30:03 -08003944 if (_Pickler_OpcodeBoundary(self) < 0)
3945 return -1;
3946
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003947 /* The extra pers_save argument is necessary to avoid calling save_pers()
3948 on its returned object. */
3949 if (!pers_save && self->pers_func) {
3950 /* save_pers() returns:
3951 -1 to signal an error;
3952 0 if it did nothing successfully;
3953 1 if a persistent id was saved.
3954 */
Serhiy Storchaka986375e2017-11-30 22:48:31 +02003955 if ((status = save_pers(self, obj)) != 0)
Serhiy Storchaka5d4cb542018-07-18 10:10:49 +03003956 return status;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003957 }
3958
3959 type = Py_TYPE(obj);
3960
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003961 /* The old cPickle had an optimization that used switch-case statement
3962 dispatching on the first letter of the type name. This has was removed
3963 since benchmarks shown that this optimization was actually slowing
3964 things down. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003965
3966 /* Atom types; these aren't memoized, so don't check the memo. */
3967
3968 if (obj == Py_None) {
Serhiy Storchaka5d4cb542018-07-18 10:10:49 +03003969 return save_none(self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003970 }
3971 else if (obj == Py_False || obj == Py_True) {
Serhiy Storchaka5d4cb542018-07-18 10:10:49 +03003972 return save_bool(self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003973 }
3974 else if (type == &PyLong_Type) {
Serhiy Storchaka5d4cb542018-07-18 10:10:49 +03003975 return save_long(self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003976 }
3977 else if (type == &PyFloat_Type) {
Serhiy Storchaka5d4cb542018-07-18 10:10:49 +03003978 return save_float(self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003979 }
3980
3981 /* Check the memo to see if it has the object. If so, generate
3982 a GET (or BINGET) opcode, instead of pickling the object
3983 once again. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003984 if (PyMemoTable_Get(self->memo, obj)) {
Serhiy Storchaka5d4cb542018-07-18 10:10:49 +03003985 return memo_get(self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003986 }
3987
3988 if (type == &PyBytes_Type) {
Serhiy Storchaka5d4cb542018-07-18 10:10:49 +03003989 return save_bytes(self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003990 }
3991 else if (type == &PyUnicode_Type) {
Serhiy Storchaka5d4cb542018-07-18 10:10:49 +03003992 return save_unicode(self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003993 }
Serhiy Storchaka5d4cb542018-07-18 10:10:49 +03003994
3995 /* We're only calling Py_EnterRecursiveCall here so that atomic
3996 types above are pickled faster. */
3997 if (Py_EnterRecursiveCall(" while pickling an object")) {
3998 return -1;
3999 }
4000
4001 if (type == &PyDict_Type) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004002 status = save_dict(self, obj);
4003 goto done;
4004 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004005 else if (type == &PySet_Type) {
4006 status = save_set(self, obj);
4007 goto done;
4008 }
4009 else if (type == &PyFrozenSet_Type) {
4010 status = save_frozenset(self, obj);
4011 goto done;
4012 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004013 else if (type == &PyList_Type) {
4014 status = save_list(self, obj);
4015 goto done;
4016 }
4017 else if (type == &PyTuple_Type) {
4018 status = save_tuple(self, obj);
4019 goto done;
4020 }
4021 else if (type == &PyType_Type) {
Alexandre Vassalotti19b6fa62013-11-30 16:06:39 -08004022 status = save_type(self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004023 goto done;
4024 }
4025 else if (type == &PyFunction_Type) {
4026 status = save_global(self, obj, NULL);
Alexandre Vassalottifc912852013-11-24 03:07:35 -08004027 goto done;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004028 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004029
4030 /* XXX: This part needs some unit tests. */
4031
4032 /* Get a reduction callable, and call it. This may come from
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01004033 * self.dispatch_table, copyreg.dispatch_table, the object's
4034 * __reduce_ex__ method, or the object's __reduce__ method.
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004035 */
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01004036 if (self->dispatch_table == NULL) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08004037 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08004038 reduce_func = PyDict_GetItemWithError(st->dispatch_table,
4039 (PyObject *)type);
4040 if (reduce_func == NULL) {
4041 if (PyErr_Occurred()) {
4042 goto error;
4043 }
4044 } else {
4045 /* PyDict_GetItemWithError() returns a borrowed reference.
4046 Increase the reference count to be consistent with
4047 PyObject_GetItem and _PyObject_GetAttrId used below. */
4048 Py_INCREF(reduce_func);
4049 }
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01004050 } else {
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08004051 reduce_func = PyObject_GetItem(self->dispatch_table,
4052 (PyObject *)type);
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01004053 if (reduce_func == NULL) {
4054 if (PyErr_ExceptionMatches(PyExc_KeyError))
4055 PyErr_Clear();
4056 else
4057 goto error;
4058 }
4059 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004060 if (reduce_func != NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004061 Py_INCREF(obj);
Alexandre Vassalotti20c28c12013-11-27 02:26:54 -08004062 reduce_value = _Pickle_FastCall(reduce_func, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004063 }
Antoine Pitrouffd41d92011-10-04 09:23:04 +02004064 else if (PyType_IsSubtype(type, &PyType_Type)) {
4065 status = save_global(self, obj, NULL);
4066 goto done;
4067 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004068 else {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004069 _Py_IDENTIFIER(__reduce__);
4070 _Py_IDENTIFIER(__reduce_ex__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004071
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004072
4073 /* XXX: If the __reduce__ method is defined, __reduce_ex__ is
4074 automatically defined as __reduce__. While this is convenient, this
4075 make it impossible to know which method was actually called. Of
4076 course, this is not a big deal. But still, it would be nice to let
4077 the user know which method was called when something go
4078 wrong. Incidentally, this means if __reduce_ex__ is not defined, we
4079 don't actually have to check for a __reduce__ method. */
4080
4081 /* Check for a __reduce_ex__ method. */
Serhiy Storchakaf320be72018-01-25 10:49:40 +02004082 if (_PyObject_LookupAttrId(obj, &PyId___reduce_ex__, &reduce_func) < 0) {
4083 goto error;
4084 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004085 if (reduce_func != NULL) {
4086 PyObject *proto;
4087 proto = PyLong_FromLong(self->proto);
4088 if (proto != NULL) {
Alexandre Vassalotti20c28c12013-11-27 02:26:54 -08004089 reduce_value = _Pickle_FastCall(reduce_func, proto);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004090 }
4091 }
4092 else {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08004093 PickleState *st = _Pickle_GetGlobalState();
4094
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004095 /* Check for a __reduce__ method. */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004096 reduce_func = _PyObject_GetAttrId(obj, &PyId___reduce__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004097 if (reduce_func != NULL) {
Victor Stinner559bb6a2016-08-22 22:48:54 +02004098 reduce_value = _PyObject_CallNoArg(reduce_func);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004099 }
4100 else {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08004101 PyErr_Format(st->PicklingError,
4102 "can't pickle '%.200s' object: %R",
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004103 type->tp_name, obj);
4104 goto error;
4105 }
4106 }
4107 }
4108
4109 if (reduce_value == NULL)
4110 goto error;
4111
4112 if (PyUnicode_Check(reduce_value)) {
4113 status = save_global(self, obj, reduce_value);
4114 goto done;
4115 }
4116
4117 if (!PyTuple_Check(reduce_value)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08004118 PickleState *st = _Pickle_GetGlobalState();
4119 PyErr_SetString(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004120 "__reduce__ must return a string or tuple");
4121 goto error;
4122 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004123
4124 status = save_reduce(self, reduce_value, obj);
4125
4126 if (0) {
4127 error:
4128 status = -1;
4129 }
4130 done:
Alexandre Vassalottib6a2f2a2013-11-23 20:30:03 -08004131
Alexandre Vassalottidff18342008-07-13 18:48:30 +00004132 Py_LeaveRecursiveCall();
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004133 Py_XDECREF(reduce_func);
4134 Py_XDECREF(reduce_value);
4135
4136 return status;
4137}
4138
4139static int
4140dump(PicklerObject *self, PyObject *obj)
4141{
4142 const char stop_op = STOP;
4143
4144 if (self->proto >= 2) {
4145 char header[2];
4146
4147 header[0] = PROTO;
4148 assert(self->proto >= 0 && self->proto < 256);
4149 header[1] = (unsigned char)self->proto;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004150 if (_Pickler_Write(self, header, 2) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004151 return -1;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004152 if (self->proto >= 4)
4153 self->framing = 1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004154 }
4155
4156 if (save(self, obj, 0) < 0 ||
Serhiy Storchakac8695292018-04-04 00:11:27 +03004157 _Pickler_Write(self, &stop_op, 1) < 0 ||
4158 _Pickler_CommitFrame(self) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004159 return -1;
Serhiy Storchakac8695292018-04-04 00:11:27 +03004160 self->framing = 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004161 return 0;
4162}
4163
Larry Hastings61272b72014-01-07 12:41:53 -08004164/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004165
4166_pickle.Pickler.clear_memo
4167
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004168Clears the pickler's "memo".
4169
4170The memo is the data structure that remembers which objects the
4171pickler has already seen, so that shared or recursive objects are
4172pickled by reference and not by value. This method is useful when
4173re-using picklers.
Larry Hastings61272b72014-01-07 12:41:53 -08004174[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004175
Larry Hastings3cceb382014-01-04 11:09:09 -08004176static PyObject *
4177_pickle_Pickler_clear_memo_impl(PicklerObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08004178/*[clinic end generated code: output=8665c8658aaa094b input=01bdad52f3d93e56]*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004179{
4180 if (self->memo)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004181 PyMemoTable_Clear(self->memo);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004182
4183 Py_RETURN_NONE;
4184}
4185
Larry Hastings61272b72014-01-07 12:41:53 -08004186/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004187
4188_pickle.Pickler.dump
4189
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004190 obj: object
4191 /
4192
4193Write a pickled representation of the given object to the open file.
Larry Hastings61272b72014-01-07 12:41:53 -08004194[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004195
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004196static PyObject *
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004197_pickle_Pickler_dump(PicklerObject *self, PyObject *obj)
Larry Hastings581ee362014-01-28 05:00:08 -08004198/*[clinic end generated code: output=87ecad1261e02ac7 input=552eb1c0f52260d9]*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004199{
Amaury Forgeot d'Arc87eee632008-10-17 20:15:53 +00004200 /* Check whether the Pickler was initialized correctly (issue3664).
4201 Developers often forget to call __init__() in their subclasses, which
4202 would trigger a segfault without this check. */
4203 if (self->write == NULL) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08004204 PickleState *st = _Pickle_GetGlobalState();
4205 PyErr_Format(st->PicklingError,
Amaury Forgeot d'Arc87eee632008-10-17 20:15:53 +00004206 "Pickler.__init__() was not called by %s.__init__()",
4207 Py_TYPE(self)->tp_name);
4208 return NULL;
4209 }
4210
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004211 if (_Pickler_ClearBuffer(self) < 0)
4212 return NULL;
4213
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004214 if (dump(self, obj) < 0)
4215 return NULL;
4216
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004217 if (_Pickler_FlushToFile(self) < 0)
4218 return NULL;
4219
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004220 Py_RETURN_NONE;
4221}
4222
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +02004223/*[clinic input]
4224
4225_pickle.Pickler.__sizeof__ -> Py_ssize_t
4226
4227Returns size in memory, in bytes.
4228[clinic start generated code]*/
4229
4230static Py_ssize_t
4231_pickle_Pickler___sizeof___impl(PicklerObject *self)
4232/*[clinic end generated code: output=106edb3123f332e1 input=8cbbec9bd5540d42]*/
4233{
4234 Py_ssize_t res, s;
4235
Serhiy Storchaka5c4064e2015-12-19 20:05:25 +02004236 res = _PyObject_SIZE(Py_TYPE(self));
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +02004237 if (self->memo != NULL) {
4238 res += sizeof(PyMemoTable);
4239 res += self->memo->mt_allocated * sizeof(PyMemoEntry);
4240 }
4241 if (self->output_buffer != NULL) {
4242 s = _PySys_GetSizeOf(self->output_buffer);
4243 if (s == -1)
4244 return -1;
4245 res += s;
4246 }
4247 return res;
4248}
4249
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004250static struct PyMethodDef Pickler_methods[] = {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004251 _PICKLE_PICKLER_DUMP_METHODDEF
4252 _PICKLE_PICKLER_CLEAR_MEMO_METHODDEF
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +02004253 _PICKLE_PICKLER___SIZEOF___METHODDEF
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004254 {NULL, NULL} /* sentinel */
4255};
4256
4257static void
4258Pickler_dealloc(PicklerObject *self)
4259{
4260 PyObject_GC_UnTrack(self);
4261
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004262 Py_XDECREF(self->output_buffer);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004263 Py_XDECREF(self->write);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004264 Py_XDECREF(self->pers_func);
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01004265 Py_XDECREF(self->dispatch_table);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004266 Py_XDECREF(self->fast_memo);
4267
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004268 PyMemoTable_Del(self->memo);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004269
4270 Py_TYPE(self)->tp_free((PyObject *)self);
4271}
4272
4273static int
4274Pickler_traverse(PicklerObject *self, visitproc visit, void *arg)
4275{
4276 Py_VISIT(self->write);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004277 Py_VISIT(self->pers_func);
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01004278 Py_VISIT(self->dispatch_table);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004279 Py_VISIT(self->fast_memo);
4280 return 0;
4281}
4282
4283static int
4284Pickler_clear(PicklerObject *self)
4285{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004286 Py_CLEAR(self->output_buffer);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004287 Py_CLEAR(self->write);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004288 Py_CLEAR(self->pers_func);
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01004289 Py_CLEAR(self->dispatch_table);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004290 Py_CLEAR(self->fast_memo);
4291
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004292 if (self->memo != NULL) {
4293 PyMemoTable *memo = self->memo;
4294 self->memo = NULL;
4295 PyMemoTable_Del(memo);
4296 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004297 return 0;
4298}
4299
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004300
Larry Hastings61272b72014-01-07 12:41:53 -08004301/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004302
4303_pickle.Pickler.__init__
4304
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004305 file: object
4306 protocol: object = NULL
4307 fix_imports: bool = True
4308
4309This takes a binary file for writing a pickle data stream.
4310
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08004311The optional *protocol* argument tells the pickler to use the given
4312protocol; supported protocols are 0, 1, 2, 3 and 4. The default
4313protocol is 3; a backward-incompatible protocol designed for Python 3.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004314
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08004315Specifying a negative protocol version selects the highest protocol
4316version supported. The higher the protocol used, the more recent the
4317version of Python needed to read the pickle produced.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004318
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08004319The *file* argument must have a write() method that accepts a single
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004320bytes argument. It can thus be a file object opened for binary
Martin Panter7462b6492015-11-02 03:37:02 +00004321writing, an io.BytesIO instance, or any other custom object that meets
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08004322this interface.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004323
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08004324If *fix_imports* is True and protocol is less than 3, pickle will try
4325to map the new Python 3 names to the old module names used in Python
43262, so that the pickle data stream is readable with Python 2.
Larry Hastings61272b72014-01-07 12:41:53 -08004327[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004328
Larry Hastingsb7ccb202014-01-18 23:50:21 -08004329static int
Larry Hastings89964c42015-04-14 18:07:59 -04004330_pickle_Pickler___init___impl(PicklerObject *self, PyObject *file,
4331 PyObject *protocol, int fix_imports)
Martin Panter2eb819f2015-11-02 04:04:57 +00004332/*[clinic end generated code: output=b5f31078dab17fb0 input=4faabdbc763c2389]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004333{
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02004334 _Py_IDENTIFIER(persistent_id);
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01004335 _Py_IDENTIFIER(dispatch_table);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004336
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004337 /* In case of multiple __init__() calls, clear previous content. */
4338 if (self->write != NULL)
4339 (void)Pickler_clear(self);
4340
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004341 if (_Pickler_SetProtocol(self, protocol, fix_imports) < 0)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08004342 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004343
4344 if (_Pickler_SetOutputStream(self, file) < 0)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08004345 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004346
4347 /* memo and output_buffer may have already been created in _Pickler_New */
4348 if (self->memo == NULL) {
4349 self->memo = PyMemoTable_New();
4350 if (self->memo == NULL)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08004351 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004352 }
4353 self->output_len = 0;
4354 if (self->output_buffer == NULL) {
4355 self->max_output_len = WRITE_BUF_SIZE;
4356 self->output_buffer = PyBytes_FromStringAndSize(NULL,
4357 self->max_output_len);
4358 if (self->output_buffer == NULL)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08004359 return -1;
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00004360 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004361
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00004362 self->fast = 0;
4363 self->fast_nesting = 0;
4364 self->fast_memo = NULL;
Serhiy Storchaka04e36af2017-10-22 21:31:34 +03004365
Serhiy Storchaka986375e2017-11-30 22:48:31 +02004366 if (init_method_ref((PyObject *)self, &PyId_persistent_id,
4367 &self->pers_func, &self->pers_func_self) < 0)
4368 {
4369 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004370 }
Serhiy Storchaka04e36af2017-10-22 21:31:34 +03004371
Serhiy Storchakaf320be72018-01-25 10:49:40 +02004372 if (_PyObject_LookupAttrId((PyObject *)self,
4373 &PyId_dispatch_table, &self->dispatch_table) < 0) {
4374 return -1;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004375 }
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08004376
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004377 return 0;
4378}
4379
Larry Hastingsb7ccb202014-01-18 23:50:21 -08004380
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004381/* Define a proxy object for the Pickler's internal memo object. This is to
4382 * avoid breaking code like:
4383 * pickler.memo.clear()
4384 * and
4385 * pickler.memo = saved_memo
4386 * Is this a good idea? Not really, but we don't want to break code that uses
4387 * it. Note that we don't implement the entire mapping API here. This is
4388 * intentional, as these should be treated as black-box implementation details.
4389 */
4390
Larry Hastings61272b72014-01-07 12:41:53 -08004391/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004392_pickle.PicklerMemoProxy.clear
4393
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004394Remove all items from memo.
Larry Hastings61272b72014-01-07 12:41:53 -08004395[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004396
Larry Hastings3cceb382014-01-04 11:09:09 -08004397static PyObject *
4398_pickle_PicklerMemoProxy_clear_impl(PicklerMemoProxyObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08004399/*[clinic end generated code: output=5fb9370d48ae8b05 input=ccc186dacd0f1405]*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004400{
4401 if (self->pickler->memo)
4402 PyMemoTable_Clear(self->pickler->memo);
4403 Py_RETURN_NONE;
4404}
4405
Larry Hastings61272b72014-01-07 12:41:53 -08004406/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004407_pickle.PicklerMemoProxy.copy
4408
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004409Copy the memo to a new object.
Larry Hastings61272b72014-01-07 12:41:53 -08004410[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004411
Larry Hastings3cceb382014-01-04 11:09:09 -08004412static PyObject *
4413_pickle_PicklerMemoProxy_copy_impl(PicklerMemoProxyObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08004414/*[clinic end generated code: output=bb83a919d29225ef input=b73043485ac30b36]*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004415{
4416 Py_ssize_t i;
4417 PyMemoTable *memo;
4418 PyObject *new_memo = PyDict_New();
4419 if (new_memo == NULL)
4420 return NULL;
4421
4422 memo = self->pickler->memo;
4423 for (i = 0; i < memo->mt_allocated; ++i) {
4424 PyMemoEntry entry = memo->mt_table[i];
4425 if (entry.me_key != NULL) {
4426 int status;
4427 PyObject *key, *value;
4428
4429 key = PyLong_FromVoidPtr(entry.me_key);
Antoine Pitrou82be19f2011-08-29 23:09:33 +02004430 value = Py_BuildValue("nO", entry.me_value, entry.me_key);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004431
4432 if (key == NULL || value == NULL) {
4433 Py_XDECREF(key);
4434 Py_XDECREF(value);
4435 goto error;
4436 }
4437 status = PyDict_SetItem(new_memo, key, value);
4438 Py_DECREF(key);
4439 Py_DECREF(value);
4440 if (status < 0)
4441 goto error;
4442 }
4443 }
4444 return new_memo;
4445
4446 error:
4447 Py_XDECREF(new_memo);
4448 return NULL;
4449}
4450
Larry Hastings61272b72014-01-07 12:41:53 -08004451/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004452_pickle.PicklerMemoProxy.__reduce__
4453
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004454Implement pickle support.
Larry Hastings61272b72014-01-07 12:41:53 -08004455[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004456
Larry Hastings3cceb382014-01-04 11:09:09 -08004457static PyObject *
4458_pickle_PicklerMemoProxy___reduce___impl(PicklerMemoProxyObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08004459/*[clinic end generated code: output=bebba1168863ab1d input=2f7c540e24b7aae4]*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004460{
4461 PyObject *reduce_value, *dict_args;
Larry Hastings3cceb382014-01-04 11:09:09 -08004462 PyObject *contents = _pickle_PicklerMemoProxy_copy_impl(self);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004463 if (contents == NULL)
4464 return NULL;
4465
4466 reduce_value = PyTuple_New(2);
4467 if (reduce_value == NULL) {
4468 Py_DECREF(contents);
4469 return NULL;
4470 }
4471 dict_args = PyTuple_New(1);
4472 if (dict_args == NULL) {
4473 Py_DECREF(contents);
4474 Py_DECREF(reduce_value);
4475 return NULL;
4476 }
4477 PyTuple_SET_ITEM(dict_args, 0, contents);
4478 Py_INCREF((PyObject *)&PyDict_Type);
4479 PyTuple_SET_ITEM(reduce_value, 0, (PyObject *)&PyDict_Type);
4480 PyTuple_SET_ITEM(reduce_value, 1, dict_args);
4481 return reduce_value;
4482}
4483
4484static PyMethodDef picklerproxy_methods[] = {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004485 _PICKLE_PICKLERMEMOPROXY_CLEAR_METHODDEF
4486 _PICKLE_PICKLERMEMOPROXY_COPY_METHODDEF
4487 _PICKLE_PICKLERMEMOPROXY___REDUCE___METHODDEF
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004488 {NULL, NULL} /* sentinel */
4489};
4490
4491static void
4492PicklerMemoProxy_dealloc(PicklerMemoProxyObject *self)
4493{
4494 PyObject_GC_UnTrack(self);
4495 Py_XDECREF(self->pickler);
4496 PyObject_GC_Del((PyObject *)self);
4497}
4498
4499static int
4500PicklerMemoProxy_traverse(PicklerMemoProxyObject *self,
4501 visitproc visit, void *arg)
4502{
4503 Py_VISIT(self->pickler);
4504 return 0;
4505}
4506
4507static int
4508PicklerMemoProxy_clear(PicklerMemoProxyObject *self)
4509{
4510 Py_CLEAR(self->pickler);
4511 return 0;
4512}
4513
4514static PyTypeObject PicklerMemoProxyType = {
4515 PyVarObject_HEAD_INIT(NULL, 0)
4516 "_pickle.PicklerMemoProxy", /*tp_name*/
4517 sizeof(PicklerMemoProxyObject), /*tp_basicsize*/
4518 0,
4519 (destructor)PicklerMemoProxy_dealloc, /* tp_dealloc */
4520 0, /* tp_print */
4521 0, /* tp_getattr */
4522 0, /* tp_setattr */
4523 0, /* tp_compare */
4524 0, /* tp_repr */
4525 0, /* tp_as_number */
4526 0, /* tp_as_sequence */
4527 0, /* tp_as_mapping */
Georg Brandlf038b322010-10-18 07:35:09 +00004528 PyObject_HashNotImplemented, /* tp_hash */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004529 0, /* tp_call */
4530 0, /* tp_str */
4531 PyObject_GenericGetAttr, /* tp_getattro */
4532 PyObject_GenericSetAttr, /* tp_setattro */
4533 0, /* tp_as_buffer */
4534 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC,
4535 0, /* tp_doc */
4536 (traverseproc)PicklerMemoProxy_traverse, /* tp_traverse */
4537 (inquiry)PicklerMemoProxy_clear, /* tp_clear */
4538 0, /* tp_richcompare */
4539 0, /* tp_weaklistoffset */
4540 0, /* tp_iter */
4541 0, /* tp_iternext */
4542 picklerproxy_methods, /* tp_methods */
4543};
4544
4545static PyObject *
4546PicklerMemoProxy_New(PicklerObject *pickler)
4547{
4548 PicklerMemoProxyObject *self;
4549
4550 self = PyObject_GC_New(PicklerMemoProxyObject, &PicklerMemoProxyType);
4551 if (self == NULL)
4552 return NULL;
4553 Py_INCREF(pickler);
4554 self->pickler = pickler;
4555 PyObject_GC_Track(self);
4556 return (PyObject *)self;
4557}
4558
4559/*****************************************************************************/
4560
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004561static PyObject *
4562Pickler_get_memo(PicklerObject *self)
4563{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004564 return PicklerMemoProxy_New(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004565}
4566
4567static int
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004568Pickler_set_memo(PicklerObject *self, PyObject *obj)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004569{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004570 PyMemoTable *new_memo = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004571
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004572 if (obj == NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004573 PyErr_SetString(PyExc_TypeError,
4574 "attribute deletion is not supported");
4575 return -1;
4576 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004577
4578 if (Py_TYPE(obj) == &PicklerMemoProxyType) {
4579 PicklerObject *pickler =
4580 ((PicklerMemoProxyObject *)obj)->pickler;
4581
4582 new_memo = PyMemoTable_Copy(pickler->memo);
4583 if (new_memo == NULL)
4584 return -1;
4585 }
4586 else if (PyDict_Check(obj)) {
4587 Py_ssize_t i = 0;
4588 PyObject *key, *value;
4589
4590 new_memo = PyMemoTable_New();
4591 if (new_memo == NULL)
4592 return -1;
4593
4594 while (PyDict_Next(obj, &i, &key, &value)) {
Antoine Pitrou82be19f2011-08-29 23:09:33 +02004595 Py_ssize_t memo_id;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004596 PyObject *memo_obj;
4597
Serhiy Storchakafff9a312017-03-21 08:53:25 +02004598 if (!PyTuple_Check(value) || PyTuple_GET_SIZE(value) != 2) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004599 PyErr_SetString(PyExc_TypeError,
4600 "'memo' values must be 2-item tuples");
4601 goto error;
4602 }
Antoine Pitrou82be19f2011-08-29 23:09:33 +02004603 memo_id = PyLong_AsSsize_t(PyTuple_GET_ITEM(value, 0));
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004604 if (memo_id == -1 && PyErr_Occurred())
4605 goto error;
4606 memo_obj = PyTuple_GET_ITEM(value, 1);
4607 if (PyMemoTable_Set(new_memo, memo_obj, memo_id) < 0)
4608 goto error;
4609 }
4610 }
4611 else {
4612 PyErr_Format(PyExc_TypeError,
Serhiy Storchakab6a9c972016-04-17 09:39:28 +03004613 "'memo' attribute must be a PicklerMemoProxy object"
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004614 "or dict, not %.200s", Py_TYPE(obj)->tp_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004615 return -1;
4616 }
4617
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004618 PyMemoTable_Del(self->memo);
4619 self->memo = new_memo;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004620
4621 return 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004622
4623 error:
4624 if (new_memo)
4625 PyMemoTable_Del(new_memo);
4626 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004627}
4628
4629static PyObject *
4630Pickler_get_persid(PicklerObject *self)
4631{
Serhiy Storchaka986375e2017-11-30 22:48:31 +02004632 if (self->pers_func == NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004633 PyErr_SetString(PyExc_AttributeError, "persistent_id");
Serhiy Storchaka986375e2017-11-30 22:48:31 +02004634 return NULL;
4635 }
4636 return reconstruct_method(self->pers_func, self->pers_func_self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004637}
4638
4639static int
4640Pickler_set_persid(PicklerObject *self, PyObject *value)
4641{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004642 if (value == NULL) {
4643 PyErr_SetString(PyExc_TypeError,
4644 "attribute deletion is not supported");
4645 return -1;
4646 }
4647 if (!PyCallable_Check(value)) {
4648 PyErr_SetString(PyExc_TypeError,
4649 "persistent_id must be a callable taking one argument");
4650 return -1;
4651 }
4652
Serhiy Storchaka986375e2017-11-30 22:48:31 +02004653 self->pers_func_self = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004654 Py_INCREF(value);
Serhiy Storchakaec397562016-04-06 09:50:03 +03004655 Py_XSETREF(self->pers_func, value);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004656
4657 return 0;
4658}
4659
4660static PyMemberDef Pickler_members[] = {
4661 {"bin", T_INT, offsetof(PicklerObject, bin)},
4662 {"fast", T_INT, offsetof(PicklerObject, fast)},
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01004663 {"dispatch_table", T_OBJECT_EX, offsetof(PicklerObject, dispatch_table)},
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004664 {NULL}
4665};
4666
4667static PyGetSetDef Pickler_getsets[] = {
4668 {"memo", (getter)Pickler_get_memo,
4669 (setter)Pickler_set_memo},
4670 {"persistent_id", (getter)Pickler_get_persid,
4671 (setter)Pickler_set_persid},
4672 {NULL}
4673};
4674
4675static PyTypeObject Pickler_Type = {
4676 PyVarObject_HEAD_INIT(NULL, 0)
4677 "_pickle.Pickler" , /*tp_name*/
4678 sizeof(PicklerObject), /*tp_basicsize*/
4679 0, /*tp_itemsize*/
4680 (destructor)Pickler_dealloc, /*tp_dealloc*/
4681 0, /*tp_print*/
4682 0, /*tp_getattr*/
4683 0, /*tp_setattr*/
Mark Dickinsone94c6792009-02-02 20:36:42 +00004684 0, /*tp_reserved*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004685 0, /*tp_repr*/
4686 0, /*tp_as_number*/
4687 0, /*tp_as_sequence*/
4688 0, /*tp_as_mapping*/
4689 0, /*tp_hash*/
4690 0, /*tp_call*/
4691 0, /*tp_str*/
4692 0, /*tp_getattro*/
4693 0, /*tp_setattro*/
4694 0, /*tp_as_buffer*/
4695 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC,
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08004696 _pickle_Pickler___init____doc__, /*tp_doc*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004697 (traverseproc)Pickler_traverse, /*tp_traverse*/
4698 (inquiry)Pickler_clear, /*tp_clear*/
4699 0, /*tp_richcompare*/
4700 0, /*tp_weaklistoffset*/
4701 0, /*tp_iter*/
4702 0, /*tp_iternext*/
4703 Pickler_methods, /*tp_methods*/
4704 Pickler_members, /*tp_members*/
4705 Pickler_getsets, /*tp_getset*/
4706 0, /*tp_base*/
4707 0, /*tp_dict*/
4708 0, /*tp_descr_get*/
4709 0, /*tp_descr_set*/
4710 0, /*tp_dictoffset*/
Larry Hastingsb7ccb202014-01-18 23:50:21 -08004711 _pickle_Pickler___init__, /*tp_init*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004712 PyType_GenericAlloc, /*tp_alloc*/
4713 PyType_GenericNew, /*tp_new*/
4714 PyObject_GC_Del, /*tp_free*/
4715 0, /*tp_is_gc*/
4716};
4717
Victor Stinner121aab42011-09-29 23:40:53 +02004718/* Temporary helper for calling self.find_class().
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004719
4720 XXX: It would be nice to able to avoid Python function call overhead, by
4721 using directly the C version of find_class(), when find_class() is not
4722 overridden by a subclass. Although, this could become rather hackish. A
4723 simpler optimization would be to call the C function when self is not a
4724 subclass instance. */
4725static PyObject *
4726find_class(UnpicklerObject *self, PyObject *module_name, PyObject *global_name)
4727{
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02004728 _Py_IDENTIFIER(find_class);
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02004729
Victor Stinner55ba38a2016-12-09 16:09:30 +01004730 return _PyObject_CallMethodIdObjArgs((PyObject *)self, &PyId_find_class,
4731 module_name, global_name, NULL);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004732}
4733
Antoine Pitrou82be19f2011-08-29 23:09:33 +02004734static Py_ssize_t
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004735marker(UnpicklerObject *self)
4736{
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02004737 Py_ssize_t mark;
4738
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004739 if (self->num_marks < 1) {
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02004740 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08004741 PyErr_SetString(st->UnpicklingError, "could not find MARK");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004742 return -1;
4743 }
4744
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02004745 mark = self->marks[--self->num_marks];
4746 self->stack->mark_set = self->num_marks != 0;
4747 self->stack->fence = self->num_marks ?
4748 self->marks[self->num_marks - 1] : 0;
4749 return mark;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004750}
4751
4752static int
4753load_none(UnpicklerObject *self)
4754{
4755 PDATA_APPEND(self->stack, Py_None, -1);
4756 return 0;
4757}
4758
4759static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004760load_int(UnpicklerObject *self)
4761{
4762 PyObject *value;
4763 char *endptr, *s;
4764 Py_ssize_t len;
4765 long x;
4766
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004767 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004768 return -1;
4769 if (len < 2)
4770 return bad_readline();
4771
4772 errno = 0;
Victor Stinner121aab42011-09-29 23:40:53 +02004773 /* XXX: Should the base argument of strtol() be explicitly set to 10?
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004774 XXX(avassalotti): Should this uses PyOS_strtol()? */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004775 x = strtol(s, &endptr, 0);
4776
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004777 if (errno || (*endptr != '\n' && *endptr != '\0')) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004778 /* Hm, maybe we've got something long. Let's try reading
Serhiy Storchaka95949422013-08-27 19:40:23 +03004779 * it as a Python int object. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004780 errno = 0;
4781 /* XXX: Same thing about the base here. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004782 value = PyLong_FromString(s, NULL, 0);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004783 if (value == NULL) {
4784 PyErr_SetString(PyExc_ValueError,
4785 "could not convert string to int");
4786 return -1;
4787 }
4788 }
4789 else {
4790 if (len == 3 && (x == 0 || x == 1)) {
4791 if ((value = PyBool_FromLong(x)) == NULL)
4792 return -1;
4793 }
4794 else {
4795 if ((value = PyLong_FromLong(x)) == NULL)
4796 return -1;
4797 }
4798 }
4799
4800 PDATA_PUSH(self->stack, value, -1);
4801 return 0;
4802}
4803
4804static int
4805load_bool(UnpicklerObject *self, PyObject *boolean)
4806{
4807 assert(boolean == Py_True || boolean == Py_False);
4808 PDATA_APPEND(self->stack, boolean, -1);
4809 return 0;
4810}
4811
Antoine Pitrou82be19f2011-08-29 23:09:33 +02004812/* s contains x bytes of an unsigned little-endian integer. Return its value
4813 * as a C Py_ssize_t, or -1 if it's higher than PY_SSIZE_T_MAX.
4814 */
4815static Py_ssize_t
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004816calc_binsize(char *bytes, int nbytes)
Antoine Pitrou82be19f2011-08-29 23:09:33 +02004817{
4818 unsigned char *s = (unsigned char *)bytes;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004819 int i;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02004820 size_t x = 0;
4821
Serhiy Storchakae0606192015-09-29 22:10:07 +03004822 if (nbytes > (int)sizeof(size_t)) {
4823 /* Check for integer overflow. BINBYTES8 and BINUNICODE8 opcodes
4824 * have 64-bit size that can't be represented on 32-bit platform.
4825 */
4826 for (i = (int)sizeof(size_t); i < nbytes; i++) {
4827 if (s[i])
4828 return -1;
4829 }
4830 nbytes = (int)sizeof(size_t);
4831 }
4832 for (i = 0; i < nbytes; i++) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004833 x |= (size_t) s[i] << (8 * i);
4834 }
Antoine Pitrou82be19f2011-08-29 23:09:33 +02004835
4836 if (x > PY_SSIZE_T_MAX)
4837 return -1;
4838 else
4839 return (Py_ssize_t) x;
4840}
4841
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004842/* s contains x bytes of a little-endian integer. Return its value as a
4843 * C int. Obscure: when x is 1 or 2, this is an unsigned little-endian
Serhiy Storchaka6a7b3a72016-04-17 08:32:47 +03004844 * int, but when x is 4 it's a signed one. This is a historical source
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004845 * of x-platform bugs.
4846 */
4847static long
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004848calc_binint(char *bytes, int nbytes)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004849{
4850 unsigned char *s = (unsigned char *)bytes;
Victor Stinnerf13c46c2014-08-17 21:05:55 +02004851 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004852 long x = 0;
4853
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004854 for (i = 0; i < nbytes; i++) {
4855 x |= (long)s[i] << (8 * i);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004856 }
4857
4858 /* Unlike BININT1 and BININT2, BININT (more accurately BININT4)
4859 * is signed, so on a box with longs bigger than 4 bytes we need
4860 * to extend a BININT's sign bit to the full width.
4861 */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01004862 if (SIZEOF_LONG > 4 && nbytes == 4) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004863 x |= -(x & (1L << 31));
4864 }
4865
4866 return x;
4867}
4868
4869static int
4870load_binintx(UnpicklerObject *self, char *s, int size)
4871{
4872 PyObject *value;
4873 long x;
4874
4875 x = calc_binint(s, size);
4876
4877 if ((value = PyLong_FromLong(x)) == NULL)
4878 return -1;
4879
4880 PDATA_PUSH(self->stack, value, -1);
4881 return 0;
4882}
4883
4884static int
4885load_binint(UnpicklerObject *self)
4886{
4887 char *s;
4888
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004889 if (_Unpickler_Read(self, &s, 4) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004890 return -1;
4891
4892 return load_binintx(self, s, 4);
4893}
4894
4895static int
4896load_binint1(UnpicklerObject *self)
4897{
4898 char *s;
4899
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004900 if (_Unpickler_Read(self, &s, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004901 return -1;
4902
4903 return load_binintx(self, s, 1);
4904}
4905
4906static int
4907load_binint2(UnpicklerObject *self)
4908{
4909 char *s;
4910
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004911 if (_Unpickler_Read(self, &s, 2) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004912 return -1;
4913
4914 return load_binintx(self, s, 2);
4915}
4916
4917static int
4918load_long(UnpicklerObject *self)
4919{
4920 PyObject *value;
Victor Stinnerb110dad2016-12-09 17:06:43 +01004921 char *s = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004922 Py_ssize_t len;
4923
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004924 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004925 return -1;
4926 if (len < 2)
4927 return bad_readline();
4928
Mark Dickinson8dd05142009-01-20 20:43:58 +00004929 /* s[len-2] will usually be 'L' (and s[len-1] is '\n'); we need to remove
4930 the 'L' before calling PyLong_FromString. In order to maintain
4931 compatibility with Python 3.0.0, we don't actually *require*
4932 the 'L' to be present. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004933 if (s[len-2] == 'L')
Alexandre Vassalotti446f7ff2009-01-23 04:43:46 +00004934 s[len-2] = '\0';
Alexandre Vassalottie4bccb72009-01-24 01:47:57 +00004935 /* XXX: Should the base argument explicitly set to 10? */
4936 value = PyLong_FromString(s, NULL, 0);
Mark Dickinson8dd05142009-01-20 20:43:58 +00004937 if (value == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004938 return -1;
4939
4940 PDATA_PUSH(self->stack, value, -1);
4941 return 0;
4942}
4943
4944/* 'size' bytes contain the # of bytes of little-endian 256's-complement
4945 * data following.
4946 */
4947static int
4948load_counted_long(UnpicklerObject *self, int size)
4949{
4950 PyObject *value;
4951 char *nbytes;
4952 char *pdata;
4953
4954 assert(size == 1 || size == 4);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004955 if (_Unpickler_Read(self, &nbytes, size) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004956 return -1;
4957
4958 size = calc_binint(nbytes, size);
4959 if (size < 0) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08004960 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004961 /* Corrupt or hostile pickle -- we never write one like this */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08004962 PyErr_SetString(st->UnpicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004963 "LONG pickle has negative byte count");
4964 return -1;
4965 }
4966
4967 if (size == 0)
4968 value = PyLong_FromLong(0L);
4969 else {
4970 /* Read the raw little-endian bytes and convert. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004971 if (_Unpickler_Read(self, &pdata, size) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004972 return -1;
4973 value = _PyLong_FromByteArray((unsigned char *)pdata, (size_t)size,
4974 1 /* little endian */ , 1 /* signed */ );
4975 }
4976 if (value == NULL)
4977 return -1;
4978 PDATA_PUSH(self->stack, value, -1);
4979 return 0;
4980}
4981
4982static int
4983load_float(UnpicklerObject *self)
4984{
4985 PyObject *value;
4986 char *endptr, *s;
4987 Py_ssize_t len;
4988 double d;
4989
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004990 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004991 return -1;
4992 if (len < 2)
4993 return bad_readline();
4994
4995 errno = 0;
Mark Dickinson725bfd82009-05-03 20:33:40 +00004996 d = PyOS_string_to_double(s, &endptr, PyExc_OverflowError);
4997 if (d == -1.0 && PyErr_Occurred())
4998 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00004999 if ((endptr[0] != '\n') && (endptr[0] != '\0')) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005000 PyErr_SetString(PyExc_ValueError, "could not convert string to float");
5001 return -1;
5002 }
Mark Dickinson725bfd82009-05-03 20:33:40 +00005003 value = PyFloat_FromDouble(d);
5004 if (value == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005005 return -1;
5006
5007 PDATA_PUSH(self->stack, value, -1);
5008 return 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005009}
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005010
5011static int
5012load_binfloat(UnpicklerObject *self)
5013{
5014 PyObject *value;
5015 double x;
5016 char *s;
5017
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005018 if (_Unpickler_Read(self, &s, 8) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005019 return -1;
5020
5021 x = _PyFloat_Unpack8((unsigned char *)s, 0);
5022 if (x == -1.0 && PyErr_Occurred())
5023 return -1;
5024
5025 if ((value = PyFloat_FromDouble(x)) == NULL)
5026 return -1;
5027
5028 PDATA_PUSH(self->stack, value, -1);
5029 return 0;
5030}
5031
5032static int
5033load_string(UnpicklerObject *self)
5034{
5035 PyObject *bytes;
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08005036 PyObject *obj;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005037 Py_ssize_t len;
5038 char *s, *p;
5039
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005040 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005041 return -1;
Alexandre Vassalotti7c5e0942013-04-15 23:14:55 -07005042 /* Strip the newline */
5043 len--;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005044 /* Strip outermost quotes */
Alexandre Vassalotti7c5e0942013-04-15 23:14:55 -07005045 if (len >= 2 && s[0] == s[len - 1] && (s[0] == '\'' || s[0] == '"')) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005046 p = s + 1;
5047 len -= 2;
5048 }
5049 else {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005050 PickleState *st = _Pickle_GetGlobalState();
5051 PyErr_SetString(st->UnpicklingError,
Alexandre Vassalotti7c5e0942013-04-15 23:14:55 -07005052 "the STRING opcode argument must be quoted");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005053 return -1;
5054 }
Alexandre Vassalotti7c5e0942013-04-15 23:14:55 -07005055 assert(len >= 0);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005056
5057 /* Use the PyBytes API to decode the string, since that is what is used
5058 to encode, and then coerce the result to Unicode. */
5059 bytes = PyBytes_DecodeEscape(p, len, NULL, 0, NULL);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005060 if (bytes == NULL)
5061 return -1;
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08005062
5063 /* Leave the Python 2.x strings as bytes if the *encoding* given to the
5064 Unpickler was 'bytes'. Otherwise, convert them to unicode. */
5065 if (strcmp(self->encoding, "bytes") == 0) {
5066 obj = bytes;
5067 }
5068 else {
5069 obj = PyUnicode_FromEncodedObject(bytes, self->encoding, self->errors);
5070 Py_DECREF(bytes);
5071 if (obj == NULL) {
5072 return -1;
5073 }
5074 }
5075
5076 PDATA_PUSH(self->stack, obj, -1);
5077 return 0;
5078}
5079
5080static int
5081load_counted_binstring(UnpicklerObject *self, int nbytes)
5082{
5083 PyObject *obj;
5084 Py_ssize_t size;
5085 char *s;
5086
5087 if (_Unpickler_Read(self, &s, nbytes) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005088 return -1;
5089
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08005090 size = calc_binsize(s, nbytes);
5091 if (size < 0) {
5092 PickleState *st = _Pickle_GetGlobalState();
5093 PyErr_Format(st->UnpicklingError,
5094 "BINSTRING exceeds system's maximum size of %zd bytes",
5095 PY_SSIZE_T_MAX);
5096 return -1;
5097 }
5098
5099 if (_Unpickler_Read(self, &s, size) < 0)
5100 return -1;
5101
5102 /* Convert Python 2.x strings to bytes if the *encoding* given to the
5103 Unpickler was 'bytes'. Otherwise, convert them to unicode. */
5104 if (strcmp(self->encoding, "bytes") == 0) {
5105 obj = PyBytes_FromStringAndSize(s, size);
5106 }
5107 else {
5108 obj = PyUnicode_Decode(s, size, self->encoding, self->errors);
5109 }
5110 if (obj == NULL) {
5111 return -1;
5112 }
5113
5114 PDATA_PUSH(self->stack, obj, -1);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005115 return 0;
5116}
5117
5118static int
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005119load_counted_binbytes(UnpicklerObject *self, int nbytes)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005120{
5121 PyObject *bytes;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005122 Py_ssize_t size;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005123 char *s;
5124
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005125 if (_Unpickler_Read(self, &s, nbytes) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005126 return -1;
5127
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005128 size = calc_binsize(s, nbytes);
5129 if (size < 0) {
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005130 PyErr_Format(PyExc_OverflowError,
5131 "BINBYTES exceeds system's maximum size of %zd bytes",
Alexandre Vassalotticc757172013-04-14 02:25:10 -07005132 PY_SSIZE_T_MAX);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005133 return -1;
5134 }
5135
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005136 if (_Unpickler_Read(self, &s, size) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005137 return -1;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005138
5139 bytes = PyBytes_FromStringAndSize(s, size);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005140 if (bytes == NULL)
5141 return -1;
5142
5143 PDATA_PUSH(self->stack, bytes, -1);
5144 return 0;
5145}
5146
5147static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005148load_unicode(UnpicklerObject *self)
5149{
5150 PyObject *str;
5151 Py_ssize_t len;
Victor Stinnerb110dad2016-12-09 17:06:43 +01005152 char *s = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005153
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005154 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005155 return -1;
5156 if (len < 1)
5157 return bad_readline();
5158
5159 str = PyUnicode_DecodeRawUnicodeEscape(s, len - 1, NULL);
5160 if (str == NULL)
5161 return -1;
5162
5163 PDATA_PUSH(self->stack, str, -1);
5164 return 0;
5165}
5166
5167static int
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005168load_counted_binunicode(UnpicklerObject *self, int nbytes)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005169{
5170 PyObject *str;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005171 Py_ssize_t size;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005172 char *s;
5173
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005174 if (_Unpickler_Read(self, &s, nbytes) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005175 return -1;
5176
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005177 size = calc_binsize(s, nbytes);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005178 if (size < 0) {
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005179 PyErr_Format(PyExc_OverflowError,
5180 "BINUNICODE exceeds system's maximum size of %zd bytes",
Alexandre Vassalotticc757172013-04-14 02:25:10 -07005181 PY_SSIZE_T_MAX);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005182 return -1;
5183 }
5184
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005185 if (_Unpickler_Read(self, &s, size) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005186 return -1;
5187
Victor Stinner485fb562010-04-13 11:07:24 +00005188 str = PyUnicode_DecodeUTF8(s, size, "surrogatepass");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005189 if (str == NULL)
5190 return -1;
5191
5192 PDATA_PUSH(self->stack, str, -1);
5193 return 0;
5194}
5195
5196static int
Victor Stinner21b47112016-03-14 18:09:39 +01005197load_counted_tuple(UnpicklerObject *self, Py_ssize_t len)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005198{
5199 PyObject *tuple;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005200
Serhiy Storchakaa49de6b2015-11-25 15:01:53 +02005201 if (Py_SIZE(self->stack) < len)
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005202 return Pdata_stack_underflow(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005203
Serhiy Storchakaa49de6b2015-11-25 15:01:53 +02005204 tuple = Pdata_poptuple(self->stack, Py_SIZE(self->stack) - len);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005205 if (tuple == NULL)
5206 return -1;
5207 PDATA_PUSH(self->stack, tuple, -1);
5208 return 0;
5209}
5210
5211static int
Serhiy Storchakaa49de6b2015-11-25 15:01:53 +02005212load_tuple(UnpicklerObject *self)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005213{
Serhiy Storchakaa49de6b2015-11-25 15:01:53 +02005214 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005215
Serhiy Storchakaa49de6b2015-11-25 15:01:53 +02005216 if ((i = marker(self)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005217 return -1;
5218
Serhiy Storchakaa49de6b2015-11-25 15:01:53 +02005219 return load_counted_tuple(self, Py_SIZE(self->stack) - i);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005220}
5221
5222static int
5223load_empty_list(UnpicklerObject *self)
5224{
5225 PyObject *list;
5226
5227 if ((list = PyList_New(0)) == NULL)
5228 return -1;
5229 PDATA_PUSH(self->stack, list, -1);
5230 return 0;
5231}
5232
5233static int
5234load_empty_dict(UnpicklerObject *self)
5235{
5236 PyObject *dict;
5237
5238 if ((dict = PyDict_New()) == NULL)
5239 return -1;
5240 PDATA_PUSH(self->stack, dict, -1);
5241 return 0;
5242}
5243
5244static int
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005245load_empty_set(UnpicklerObject *self)
5246{
5247 PyObject *set;
5248
5249 if ((set = PySet_New(NULL)) == NULL)
5250 return -1;
5251 PDATA_PUSH(self->stack, set, -1);
5252 return 0;
5253}
5254
5255static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005256load_list(UnpicklerObject *self)
5257{
5258 PyObject *list;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005259 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005260
5261 if ((i = marker(self)) < 0)
5262 return -1;
5263
5264 list = Pdata_poplist(self->stack, i);
5265 if (list == NULL)
5266 return -1;
5267 PDATA_PUSH(self->stack, list, -1);
5268 return 0;
5269}
5270
5271static int
5272load_dict(UnpicklerObject *self)
5273{
5274 PyObject *dict, *key, *value;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005275 Py_ssize_t i, j, k;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005276
5277 if ((i = marker(self)) < 0)
5278 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005279 j = Py_SIZE(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005280
5281 if ((dict = PyDict_New()) == NULL)
5282 return -1;
5283
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005284 if ((j - i) % 2 != 0) {
5285 PickleState *st = _Pickle_GetGlobalState();
5286 PyErr_SetString(st->UnpicklingError, "odd number of items for DICT");
Serhiy Storchaka3ac53802015-12-07 11:32:00 +02005287 Py_DECREF(dict);
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005288 return -1;
5289 }
5290
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005291 for (k = i + 1; k < j; k += 2) {
5292 key = self->stack->data[k - 1];
5293 value = self->stack->data[k];
5294 if (PyDict_SetItem(dict, key, value) < 0) {
5295 Py_DECREF(dict);
5296 return -1;
5297 }
5298 }
5299 Pdata_clear(self->stack, i);
5300 PDATA_PUSH(self->stack, dict, -1);
5301 return 0;
5302}
5303
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005304static int
5305load_frozenset(UnpicklerObject *self)
5306{
5307 PyObject *items;
5308 PyObject *frozenset;
5309 Py_ssize_t i;
5310
5311 if ((i = marker(self)) < 0)
5312 return -1;
5313
5314 items = Pdata_poptuple(self->stack, i);
5315 if (items == NULL)
5316 return -1;
5317
5318 frozenset = PyFrozenSet_New(items);
5319 Py_DECREF(items);
5320 if (frozenset == NULL)
5321 return -1;
5322
5323 PDATA_PUSH(self->stack, frozenset, -1);
5324 return 0;
5325}
5326
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005327static PyObject *
5328instantiate(PyObject *cls, PyObject *args)
5329{
Alexander Belopolskyd92f0402010-07-17 22:50:45 +00005330 /* Caller must assure args are a tuple. Normally, args come from
5331 Pdata_poptuple which packs objects from the top of the stack
5332 into a newly created tuple. */
5333 assert(PyTuple_Check(args));
Serhiy Storchaka04e36af2017-10-22 21:31:34 +03005334 if (!PyTuple_GET_SIZE(args) && PyType_Check(cls)) {
5335 _Py_IDENTIFIER(__getinitargs__);
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02005336 _Py_IDENTIFIER(__new__);
Serhiy Storchakaf320be72018-01-25 10:49:40 +02005337 PyObject *func;
5338 if (_PyObject_LookupAttrId(cls, &PyId___getinitargs__, &func) < 0) {
5339 return NULL;
5340 }
Serhiy Storchaka04e36af2017-10-22 21:31:34 +03005341 if (func == NULL) {
Serhiy Storchaka04e36af2017-10-22 21:31:34 +03005342 return _PyObject_CallMethodIdObjArgs(cls, &PyId___new__, cls, NULL);
5343 }
5344 Py_DECREF(func);
Alexander Belopolskyd92f0402010-07-17 22:50:45 +00005345 }
Serhiy Storchaka04e36af2017-10-22 21:31:34 +03005346 return PyObject_CallObject(cls, args);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005347}
5348
5349static int
5350load_obj(UnpicklerObject *self)
5351{
5352 PyObject *cls, *args, *obj = NULL;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005353 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005354
5355 if ((i = marker(self)) < 0)
5356 return -1;
5357
Serhiy Storchakae9b30742015-11-23 15:17:43 +02005358 if (Py_SIZE(self->stack) - i < 1)
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005359 return Pdata_stack_underflow(self->stack);
Serhiy Storchakae9b30742015-11-23 15:17:43 +02005360
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005361 args = Pdata_poptuple(self->stack, i + 1);
5362 if (args == NULL)
5363 return -1;
5364
5365 PDATA_POP(self->stack, cls);
5366 if (cls) {
5367 obj = instantiate(cls, args);
5368 Py_DECREF(cls);
5369 }
5370 Py_DECREF(args);
5371 if (obj == NULL)
5372 return -1;
5373
5374 PDATA_PUSH(self->stack, obj, -1);
5375 return 0;
5376}
5377
5378static int
5379load_inst(UnpicklerObject *self)
5380{
5381 PyObject *cls = NULL;
5382 PyObject *args = NULL;
5383 PyObject *obj = NULL;
5384 PyObject *module_name;
5385 PyObject *class_name;
5386 Py_ssize_t len;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005387 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005388 char *s;
5389
5390 if ((i = marker(self)) < 0)
5391 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005392 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005393 return -1;
5394 if (len < 2)
5395 return bad_readline();
5396
5397 /* Here it is safe to use PyUnicode_DecodeASCII(), even though non-ASCII
5398 identifiers are permitted in Python 3.0, since the INST opcode is only
5399 supported by older protocols on Python 2.x. */
5400 module_name = PyUnicode_DecodeASCII(s, len - 1, "strict");
5401 if (module_name == NULL)
5402 return -1;
5403
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005404 if ((len = _Unpickler_Readline(self, &s)) >= 0) {
Serhiy Storchakaca28eba2015-12-01 00:18:23 +02005405 if (len < 2) {
5406 Py_DECREF(module_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005407 return bad_readline();
Serhiy Storchakaca28eba2015-12-01 00:18:23 +02005408 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005409 class_name = PyUnicode_DecodeASCII(s, len - 1, "strict");
Alexander Belopolskyd92f0402010-07-17 22:50:45 +00005410 if (class_name != NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005411 cls = find_class(self, module_name, class_name);
5412 Py_DECREF(class_name);
5413 }
5414 }
5415 Py_DECREF(module_name);
5416
5417 if (cls == NULL)
5418 return -1;
5419
5420 if ((args = Pdata_poptuple(self->stack, i)) != NULL) {
5421 obj = instantiate(cls, args);
5422 Py_DECREF(args);
5423 }
5424 Py_DECREF(cls);
5425
5426 if (obj == NULL)
5427 return -1;
5428
5429 PDATA_PUSH(self->stack, obj, -1);
5430 return 0;
5431}
5432
5433static int
5434load_newobj(UnpicklerObject *self)
5435{
5436 PyObject *args = NULL;
5437 PyObject *clsraw = NULL;
5438 PyTypeObject *cls; /* clsraw cast to its true type */
5439 PyObject *obj;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005440 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005441
5442 /* Stack is ... cls argtuple, and we want to call
5443 * cls.__new__(cls, *argtuple).
5444 */
5445 PDATA_POP(self->stack, args);
5446 if (args == NULL)
5447 goto error;
5448 if (!PyTuple_Check(args)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005449 PyErr_SetString(st->UnpicklingError,
5450 "NEWOBJ expected an arg " "tuple.");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005451 goto error;
5452 }
5453
5454 PDATA_POP(self->stack, clsraw);
5455 cls = (PyTypeObject *)clsraw;
5456 if (cls == NULL)
5457 goto error;
5458 if (!PyType_Check(cls)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005459 PyErr_SetString(st->UnpicklingError, "NEWOBJ class argument "
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005460 "isn't a type object");
5461 goto error;
5462 }
5463 if (cls->tp_new == NULL) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005464 PyErr_SetString(st->UnpicklingError, "NEWOBJ class argument "
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005465 "has NULL tp_new");
5466 goto error;
5467 }
5468
5469 /* Call __new__. */
5470 obj = cls->tp_new(cls, args, NULL);
5471 if (obj == NULL)
5472 goto error;
5473
5474 Py_DECREF(args);
5475 Py_DECREF(clsraw);
5476 PDATA_PUSH(self->stack, obj, -1);
5477 return 0;
5478
5479 error:
5480 Py_XDECREF(args);
5481 Py_XDECREF(clsraw);
5482 return -1;
5483}
5484
5485static int
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005486load_newobj_ex(UnpicklerObject *self)
5487{
5488 PyObject *cls, *args, *kwargs;
5489 PyObject *obj;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005490 PickleState *st = _Pickle_GetGlobalState();
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005491
5492 PDATA_POP(self->stack, kwargs);
5493 if (kwargs == NULL) {
5494 return -1;
5495 }
5496 PDATA_POP(self->stack, args);
5497 if (args == NULL) {
5498 Py_DECREF(kwargs);
5499 return -1;
5500 }
5501 PDATA_POP(self->stack, cls);
5502 if (cls == NULL) {
5503 Py_DECREF(kwargs);
5504 Py_DECREF(args);
5505 return -1;
5506 }
Larry Hastings61272b72014-01-07 12:41:53 -08005507
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005508 if (!PyType_Check(cls)) {
5509 Py_DECREF(kwargs);
5510 Py_DECREF(args);
Larry Hastings61272b72014-01-07 12:41:53 -08005511 PyErr_Format(st->UnpicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005512 "NEWOBJ_EX class argument must be a type, not %.200s",
5513 Py_TYPE(cls)->tp_name);
Benjamin Peterson80f78a32015-07-02 16:18:38 -05005514 Py_DECREF(cls);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005515 return -1;
5516 }
5517
5518 if (((PyTypeObject *)cls)->tp_new == NULL) {
5519 Py_DECREF(kwargs);
5520 Py_DECREF(args);
5521 Py_DECREF(cls);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005522 PyErr_SetString(st->UnpicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005523 "NEWOBJ_EX class argument doesn't have __new__");
5524 return -1;
5525 }
5526 obj = ((PyTypeObject *)cls)->tp_new((PyTypeObject *)cls, args, kwargs);
5527 Py_DECREF(kwargs);
5528 Py_DECREF(args);
5529 Py_DECREF(cls);
5530 if (obj == NULL) {
5531 return -1;
5532 }
5533 PDATA_PUSH(self->stack, obj, -1);
5534 return 0;
5535}
5536
5537static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005538load_global(UnpicklerObject *self)
5539{
5540 PyObject *global = NULL;
5541 PyObject *module_name;
5542 PyObject *global_name;
5543 Py_ssize_t len;
5544 char *s;
5545
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005546 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005547 return -1;
5548 if (len < 2)
5549 return bad_readline();
5550 module_name = PyUnicode_DecodeUTF8(s, len - 1, "strict");
5551 if (!module_name)
5552 return -1;
5553
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005554 if ((len = _Unpickler_Readline(self, &s)) >= 0) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005555 if (len < 2) {
5556 Py_DECREF(module_name);
5557 return bad_readline();
5558 }
5559 global_name = PyUnicode_DecodeUTF8(s, len - 1, "strict");
5560 if (global_name) {
5561 global = find_class(self, module_name, global_name);
5562 Py_DECREF(global_name);
5563 }
5564 }
5565 Py_DECREF(module_name);
5566
5567 if (global == NULL)
5568 return -1;
5569 PDATA_PUSH(self->stack, global, -1);
5570 return 0;
5571}
5572
5573static int
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005574load_stack_global(UnpicklerObject *self)
5575{
5576 PyObject *global;
5577 PyObject *module_name;
5578 PyObject *global_name;
5579
5580 PDATA_POP(self->stack, global_name);
5581 PDATA_POP(self->stack, module_name);
5582 if (module_name == NULL || !PyUnicode_CheckExact(module_name) ||
5583 global_name == NULL || !PyUnicode_CheckExact(global_name)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005584 PickleState *st = _Pickle_GetGlobalState();
5585 PyErr_SetString(st->UnpicklingError, "STACK_GLOBAL requires str");
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005586 Py_XDECREF(global_name);
5587 Py_XDECREF(module_name);
5588 return -1;
5589 }
5590 global = find_class(self, module_name, global_name);
5591 Py_DECREF(global_name);
5592 Py_DECREF(module_name);
5593 if (global == NULL)
5594 return -1;
5595 PDATA_PUSH(self->stack, global, -1);
5596 return 0;
5597}
5598
5599static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005600load_persid(UnpicklerObject *self)
5601{
Serhiy Storchaka986375e2017-11-30 22:48:31 +02005602 PyObject *pid, *obj;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005603 Py_ssize_t len;
5604 char *s;
5605
5606 if (self->pers_func) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005607 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005608 return -1;
Alexandre Vassalotti896414f2013-11-30 13:52:35 -08005609 if (len < 1)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005610 return bad_readline();
5611
Serhiy Storchakadec25af2016-07-17 11:24:17 +03005612 pid = PyUnicode_DecodeASCII(s, len - 1, "strict");
5613 if (pid == NULL) {
5614 if (PyErr_ExceptionMatches(PyExc_UnicodeDecodeError)) {
5615 PyErr_SetString(_Pickle_GetGlobalState()->UnpicklingError,
5616 "persistent IDs in protocol 0 must be "
5617 "ASCII strings");
5618 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005619 return -1;
Serhiy Storchakadec25af2016-07-17 11:24:17 +03005620 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005621
Serhiy Storchaka986375e2017-11-30 22:48:31 +02005622 obj = call_method(self->pers_func, self->pers_func_self, pid);
5623 Py_DECREF(pid);
5624 if (obj == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005625 return -1;
5626
Serhiy Storchaka986375e2017-11-30 22:48:31 +02005627 PDATA_PUSH(self->stack, obj, -1);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005628 return 0;
5629 }
5630 else {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005631 PickleState *st = _Pickle_GetGlobalState();
5632 PyErr_SetString(st->UnpicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005633 "A load persistent id instruction was encountered,\n"
5634 "but no persistent_load function was specified.");
5635 return -1;
5636 }
5637}
5638
5639static int
5640load_binpersid(UnpicklerObject *self)
5641{
Serhiy Storchaka986375e2017-11-30 22:48:31 +02005642 PyObject *pid, *obj;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005643
5644 if (self->pers_func) {
5645 PDATA_POP(self->stack, pid);
5646 if (pid == NULL)
5647 return -1;
5648
Serhiy Storchaka986375e2017-11-30 22:48:31 +02005649 obj = call_method(self->pers_func, self->pers_func_self, pid);
5650 Py_DECREF(pid);
5651 if (obj == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005652 return -1;
5653
Serhiy Storchaka986375e2017-11-30 22:48:31 +02005654 PDATA_PUSH(self->stack, obj, -1);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005655 return 0;
5656 }
5657 else {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005658 PickleState *st = _Pickle_GetGlobalState();
5659 PyErr_SetString(st->UnpicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005660 "A load persistent id instruction was encountered,\n"
5661 "but no persistent_load function was specified.");
5662 return -1;
5663 }
5664}
5665
5666static int
5667load_pop(UnpicklerObject *self)
5668{
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005669 Py_ssize_t len = Py_SIZE(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005670
5671 /* Note that we split the (pickle.py) stack into two stacks,
5672 * an object stack and a mark stack. We have to be clever and
5673 * pop the right one. We do this by looking at the top of the
Collin Winter8ca69de2009-05-26 16:53:41 +00005674 * mark stack first, and only signalling a stack underflow if
5675 * the object stack is empty and the mark stack doesn't match
5676 * our expectations.
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005677 */
Collin Winter8ca69de2009-05-26 16:53:41 +00005678 if (self->num_marks > 0 && self->marks[self->num_marks - 1] == len) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005679 self->num_marks--;
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005680 self->stack->mark_set = self->num_marks != 0;
5681 self->stack->fence = self->num_marks ?
5682 self->marks[self->num_marks - 1] : 0;
5683 } else if (len <= self->stack->fence)
5684 return Pdata_stack_underflow(self->stack);
5685 else {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005686 len--;
5687 Py_DECREF(self->stack->data[len]);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005688 Py_SIZE(self->stack) = len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005689 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005690 return 0;
5691}
5692
5693static int
5694load_pop_mark(UnpicklerObject *self)
5695{
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005696 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005697
5698 if ((i = marker(self)) < 0)
5699 return -1;
5700
5701 Pdata_clear(self->stack, i);
5702
5703 return 0;
5704}
5705
5706static int
5707load_dup(UnpicklerObject *self)
5708{
5709 PyObject *last;
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005710 Py_ssize_t len = Py_SIZE(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005711
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005712 if (len <= self->stack->fence)
5713 return Pdata_stack_underflow(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005714 last = self->stack->data[len - 1];
5715 PDATA_APPEND(self->stack, last, -1);
5716 return 0;
5717}
5718
5719static int
5720load_get(UnpicklerObject *self)
5721{
5722 PyObject *key, *value;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005723 Py_ssize_t idx;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005724 Py_ssize_t len;
5725 char *s;
5726
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005727 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005728 return -1;
5729 if (len < 2)
5730 return bad_readline();
5731
5732 key = PyLong_FromString(s, NULL, 10);
5733 if (key == NULL)
5734 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005735 idx = PyLong_AsSsize_t(key);
5736 if (idx == -1 && PyErr_Occurred()) {
5737 Py_DECREF(key);
5738 return -1;
5739 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005740
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005741 value = _Unpickler_MemoGet(self, idx);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005742 if (value == NULL) {
5743 if (!PyErr_Occurred())
5744 PyErr_SetObject(PyExc_KeyError, key);
5745 Py_DECREF(key);
5746 return -1;
5747 }
5748 Py_DECREF(key);
5749
5750 PDATA_APPEND(self->stack, value, -1);
5751 return 0;
5752}
5753
5754static int
5755load_binget(UnpicklerObject *self)
5756{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005757 PyObject *value;
5758 Py_ssize_t idx;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005759 char *s;
5760
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005761 if (_Unpickler_Read(self, &s, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005762 return -1;
5763
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005764 idx = Py_CHARMASK(s[0]);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005765
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005766 value = _Unpickler_MemoGet(self, idx);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005767 if (value == NULL) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005768 PyObject *key = PyLong_FromSsize_t(idx);
Christian Heimes9ee5c372013-07-26 22:45:00 +02005769 if (key != NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005770 PyErr_SetObject(PyExc_KeyError, key);
Christian Heimes9ee5c372013-07-26 22:45:00 +02005771 Py_DECREF(key);
5772 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005773 return -1;
5774 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005775
5776 PDATA_APPEND(self->stack, value, -1);
5777 return 0;
5778}
5779
5780static int
5781load_long_binget(UnpicklerObject *self)
5782{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005783 PyObject *value;
5784 Py_ssize_t idx;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005785 char *s;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005786
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005787 if (_Unpickler_Read(self, &s, 4) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005788 return -1;
5789
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005790 idx = calc_binsize(s, 4);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005791
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005792 value = _Unpickler_MemoGet(self, idx);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005793 if (value == NULL) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005794 PyObject *key = PyLong_FromSsize_t(idx);
Christian Heimes9ee5c372013-07-26 22:45:00 +02005795 if (key != NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005796 PyErr_SetObject(PyExc_KeyError, key);
Christian Heimes9ee5c372013-07-26 22:45:00 +02005797 Py_DECREF(key);
5798 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005799 return -1;
5800 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005801
5802 PDATA_APPEND(self->stack, value, -1);
5803 return 0;
5804}
5805
5806/* Push an object from the extension registry (EXT[124]). nbytes is
5807 * the number of bytes following the opcode, holding the index (code) value.
5808 */
5809static int
5810load_extension(UnpicklerObject *self, int nbytes)
5811{
5812 char *codebytes; /* the nbytes bytes after the opcode */
5813 long code; /* calc_binint returns long */
5814 PyObject *py_code; /* code as a Python int */
5815 PyObject *obj; /* the object to push */
5816 PyObject *pair; /* (module_name, class_name) */
5817 PyObject *module_name, *class_name;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005818 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005819
5820 assert(nbytes == 1 || nbytes == 2 || nbytes == 4);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005821 if (_Unpickler_Read(self, &codebytes, nbytes) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005822 return -1;
5823 code = calc_binint(codebytes, nbytes);
5824 if (code <= 0) { /* note that 0 is forbidden */
5825 /* Corrupt or hostile pickle. */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005826 PyErr_SetString(st->UnpicklingError, "EXT specifies code <= 0");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005827 return -1;
5828 }
5829
5830 /* Look for the code in the cache. */
5831 py_code = PyLong_FromLong(code);
5832 if (py_code == NULL)
5833 return -1;
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08005834 obj = PyDict_GetItemWithError(st->extension_cache, py_code);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005835 if (obj != NULL) {
5836 /* Bingo. */
5837 Py_DECREF(py_code);
5838 PDATA_APPEND(self->stack, obj, -1);
5839 return 0;
5840 }
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08005841 if (PyErr_Occurred()) {
5842 Py_DECREF(py_code);
5843 return -1;
5844 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005845
5846 /* Look up the (module_name, class_name) pair. */
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08005847 pair = PyDict_GetItemWithError(st->inverted_registry, py_code);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005848 if (pair == NULL) {
5849 Py_DECREF(py_code);
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08005850 if (!PyErr_Occurred()) {
5851 PyErr_Format(PyExc_ValueError, "unregistered extension "
5852 "code %ld", code);
5853 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005854 return -1;
5855 }
5856 /* Since the extension registry is manipulable via Python code,
5857 * confirm that pair is really a 2-tuple of strings.
5858 */
5859 if (!PyTuple_Check(pair) || PyTuple_Size(pair) != 2 ||
5860 !PyUnicode_Check(module_name = PyTuple_GET_ITEM(pair, 0)) ||
5861 !PyUnicode_Check(class_name = PyTuple_GET_ITEM(pair, 1))) {
5862 Py_DECREF(py_code);
5863 PyErr_Format(PyExc_ValueError, "_inverted_registry[%ld] "
5864 "isn't a 2-tuple of strings", code);
5865 return -1;
5866 }
5867 /* Load the object. */
5868 obj = find_class(self, module_name, class_name);
5869 if (obj == NULL) {
5870 Py_DECREF(py_code);
5871 return -1;
5872 }
5873 /* Cache code -> obj. */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08005874 code = PyDict_SetItem(st->extension_cache, py_code, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005875 Py_DECREF(py_code);
5876 if (code < 0) {
5877 Py_DECREF(obj);
5878 return -1;
5879 }
5880 PDATA_PUSH(self->stack, obj, -1);
5881 return 0;
5882}
5883
5884static int
5885load_put(UnpicklerObject *self)
5886{
5887 PyObject *key, *value;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005888 Py_ssize_t idx;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005889 Py_ssize_t len;
Victor Stinnerb110dad2016-12-09 17:06:43 +01005890 char *s = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005891
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005892 if ((len = _Unpickler_Readline(self, &s)) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005893 return -1;
5894 if (len < 2)
5895 return bad_readline();
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005896 if (Py_SIZE(self->stack) <= self->stack->fence)
5897 return Pdata_stack_underflow(self->stack);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005898 value = self->stack->data[Py_SIZE(self->stack) - 1];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005899
5900 key = PyLong_FromString(s, NULL, 10);
5901 if (key == NULL)
5902 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005903 idx = PyLong_AsSsize_t(key);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005904 Py_DECREF(key);
Antoine Pitrou55549ec2011-08-30 00:27:10 +02005905 if (idx < 0) {
5906 if (!PyErr_Occurred())
5907 PyErr_SetString(PyExc_ValueError,
5908 "negative PUT argument");
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005909 return -1;
Antoine Pitrou55549ec2011-08-30 00:27:10 +02005910 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005911
5912 return _Unpickler_MemoPut(self, idx, value);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005913}
5914
5915static int
5916load_binput(UnpicklerObject *self)
5917{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005918 PyObject *value;
5919 Py_ssize_t idx;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005920 char *s;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005921
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005922 if (_Unpickler_Read(self, &s, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005923 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005924
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005925 if (Py_SIZE(self->stack) <= self->stack->fence)
5926 return Pdata_stack_underflow(self->stack);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005927 value = self->stack->data[Py_SIZE(self->stack) - 1];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005928
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005929 idx = Py_CHARMASK(s[0]);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005930
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005931 return _Unpickler_MemoPut(self, idx, value);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005932}
5933
5934static int
5935load_long_binput(UnpicklerObject *self)
5936{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005937 PyObject *value;
5938 Py_ssize_t idx;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005939 char *s;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005940
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005941 if (_Unpickler_Read(self, &s, 4) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005942 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005943
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005944 if (Py_SIZE(self->stack) <= self->stack->fence)
5945 return Pdata_stack_underflow(self->stack);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005946 value = self->stack->data[Py_SIZE(self->stack) - 1];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005947
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005948 idx = calc_binsize(s, 4);
Antoine Pitrou55549ec2011-08-30 00:27:10 +02005949 if (idx < 0) {
5950 PyErr_SetString(PyExc_ValueError,
5951 "negative LONG_BINPUT argument");
5952 return -1;
5953 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005954
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005955 return _Unpickler_MemoPut(self, idx, value);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005956}
5957
5958static int
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005959load_memoize(UnpicklerObject *self)
5960{
5961 PyObject *value;
5962
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005963 if (Py_SIZE(self->stack) <= self->stack->fence)
5964 return Pdata_stack_underflow(self->stack);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01005965 value = self->stack->data[Py_SIZE(self->stack) - 1];
5966
5967 return _Unpickler_MemoPut(self, self->memo_len, value);
5968}
5969
5970static int
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005971do_append(UnpicklerObject *self, Py_ssize_t x)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005972{
5973 PyObject *value;
Serhiy Storchakabee09ae2017-02-02 11:12:47 +02005974 PyObject *slice;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005975 PyObject *list;
Serhiy Storchakabee09ae2017-02-02 11:12:47 +02005976 PyObject *result;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005977 Py_ssize_t len, i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005978
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00005979 len = Py_SIZE(self->stack);
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02005980 if (x > len || x <= self->stack->fence)
5981 return Pdata_stack_underflow(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005982 if (len == x) /* nothing to do */
5983 return 0;
5984
5985 list = self->stack->data[x - 1];
5986
Serhiy Storchakabee09ae2017-02-02 11:12:47 +02005987 if (PyList_CheckExact(list)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005988 Py_ssize_t list_len;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005989 int ret;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005990
5991 slice = Pdata_poplist(self->stack, x);
5992 if (!slice)
5993 return -1;
5994 list_len = PyList_GET_SIZE(list);
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005995 ret = PyList_SetSlice(list, list_len, list_len, slice);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005996 Py_DECREF(slice);
Antoine Pitrou82be19f2011-08-29 23:09:33 +02005997 return ret;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00005998 }
5999 else {
Serhiy Storchakabee09ae2017-02-02 11:12:47 +02006000 PyObject *extend_func;
6001 _Py_IDENTIFIER(extend);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006002
Serhiy Storchakabee09ae2017-02-02 11:12:47 +02006003 extend_func = _PyObject_GetAttrId(list, &PyId_extend);
6004 if (extend_func != NULL) {
6005 slice = Pdata_poplist(self->stack, x);
6006 if (!slice) {
6007 Py_DECREF(extend_func);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006008 return -1;
6009 }
Serhiy Storchakabee09ae2017-02-02 11:12:47 +02006010 result = _Pickle_FastCall(extend_func, slice);
Serhiy Storchakabee09ae2017-02-02 11:12:47 +02006011 Py_DECREF(extend_func);
6012 if (result == NULL)
6013 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006014 Py_DECREF(result);
6015 }
Serhiy Storchakabee09ae2017-02-02 11:12:47 +02006016 else {
6017 PyObject *append_func;
6018 _Py_IDENTIFIER(append);
6019
6020 /* Even if the PEP 307 requires extend() and append() methods,
6021 fall back on append() if the object has no extend() method
6022 for backward compatibility. */
6023 PyErr_Clear();
6024 append_func = _PyObject_GetAttrId(list, &PyId_append);
6025 if (append_func == NULL)
6026 return -1;
6027 for (i = x; i < len; i++) {
6028 value = self->stack->data[i];
6029 result = _Pickle_FastCall(append_func, value);
6030 if (result == NULL) {
6031 Pdata_clear(self->stack, i + 1);
6032 Py_SIZE(self->stack) = x;
6033 Py_DECREF(append_func);
6034 return -1;
6035 }
6036 Py_DECREF(result);
6037 }
6038 Py_SIZE(self->stack) = x;
6039 Py_DECREF(append_func);
6040 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006041 }
6042
6043 return 0;
6044}
6045
6046static int
6047load_append(UnpicklerObject *self)
6048{
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02006049 if (Py_SIZE(self->stack) - 1 <= self->stack->fence)
6050 return Pdata_stack_underflow(self->stack);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006051 return do_append(self, Py_SIZE(self->stack) - 1);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006052}
6053
6054static int
6055load_appends(UnpicklerObject *self)
6056{
Serhiy Storchakae9b30742015-11-23 15:17:43 +02006057 Py_ssize_t i = marker(self);
6058 if (i < 0)
6059 return -1;
6060 return do_append(self, i);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006061}
6062
6063static int
Antoine Pitrou82be19f2011-08-29 23:09:33 +02006064do_setitems(UnpicklerObject *self, Py_ssize_t x)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006065{
6066 PyObject *value, *key;
6067 PyObject *dict;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02006068 Py_ssize_t len, i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006069 int status = 0;
6070
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006071 len = Py_SIZE(self->stack);
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02006072 if (x > len || x <= self->stack->fence)
6073 return Pdata_stack_underflow(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006074 if (len == x) /* nothing to do */
6075 return 0;
Victor Stinner121aab42011-09-29 23:40:53 +02006076 if ((len - x) % 2 != 0) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006077 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006078 /* Currupt or hostile pickle -- we never write one like this. */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006079 PyErr_SetString(st->UnpicklingError,
6080 "odd number of items for SETITEMS");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006081 return -1;
6082 }
6083
6084 /* Here, dict does not actually need to be a PyDict; it could be anything
6085 that supports the __setitem__ attribute. */
6086 dict = self->stack->data[x - 1];
6087
6088 for (i = x + 1; i < len; i += 2) {
6089 key = self->stack->data[i - 1];
6090 value = self->stack->data[i];
6091 if (PyObject_SetItem(dict, key, value) < 0) {
6092 status = -1;
6093 break;
6094 }
6095 }
6096
6097 Pdata_clear(self->stack, x);
6098 return status;
6099}
6100
6101static int
6102load_setitem(UnpicklerObject *self)
6103{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006104 return do_setitems(self, Py_SIZE(self->stack) - 2);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006105}
6106
6107static int
6108load_setitems(UnpicklerObject *self)
6109{
Serhiy Storchakae9b30742015-11-23 15:17:43 +02006110 Py_ssize_t i = marker(self);
6111 if (i < 0)
6112 return -1;
6113 return do_setitems(self, i);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006114}
6115
6116static int
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006117load_additems(UnpicklerObject *self)
6118{
6119 PyObject *set;
6120 Py_ssize_t mark, len, i;
6121
6122 mark = marker(self);
Serhiy Storchakae9b30742015-11-23 15:17:43 +02006123 if (mark < 0)
6124 return -1;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006125 len = Py_SIZE(self->stack);
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02006126 if (mark > len || mark <= self->stack->fence)
6127 return Pdata_stack_underflow(self->stack);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006128 if (len == mark) /* nothing to do */
6129 return 0;
6130
6131 set = self->stack->data[mark - 1];
6132
6133 if (PySet_Check(set)) {
6134 PyObject *items;
6135 int status;
6136
6137 items = Pdata_poptuple(self->stack, mark);
6138 if (items == NULL)
6139 return -1;
6140
6141 status = _PySet_Update(set, items);
6142 Py_DECREF(items);
6143 return status;
6144 }
6145 else {
6146 PyObject *add_func;
6147 _Py_IDENTIFIER(add);
6148
6149 add_func = _PyObject_GetAttrId(set, &PyId_add);
6150 if (add_func == NULL)
6151 return -1;
6152 for (i = mark; i < len; i++) {
6153 PyObject *result;
6154 PyObject *item;
6155
6156 item = self->stack->data[i];
Alexandre Vassalotti20c28c12013-11-27 02:26:54 -08006157 result = _Pickle_FastCall(add_func, item);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006158 if (result == NULL) {
6159 Pdata_clear(self->stack, i + 1);
6160 Py_SIZE(self->stack) = mark;
6161 return -1;
6162 }
6163 Py_DECREF(result);
6164 }
6165 Py_SIZE(self->stack) = mark;
6166 }
6167
6168 return 0;
6169}
6170
6171static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006172load_build(UnpicklerObject *self)
6173{
6174 PyObject *state, *inst, *slotstate;
6175 PyObject *setstate;
6176 int status = 0;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02006177 _Py_IDENTIFIER(__setstate__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006178
6179 /* Stack is ... instance, state. We want to leave instance at
6180 * the stack top, possibly mutated via instance.__setstate__(state).
6181 */
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02006182 if (Py_SIZE(self->stack) - 2 < self->stack->fence)
6183 return Pdata_stack_underflow(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006184
6185 PDATA_POP(self->stack, state);
6186 if (state == NULL)
6187 return -1;
6188
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006189 inst = self->stack->data[Py_SIZE(self->stack) - 1];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006190
Serhiy Storchakaf320be72018-01-25 10:49:40 +02006191 if (_PyObject_LookupAttrId(inst, &PyId___setstate__, &setstate) < 0) {
6192 Py_DECREF(state);
6193 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006194 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02006195 if (setstate != NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006196 PyObject *result;
6197
6198 /* The explicit __setstate__ is responsible for everything. */
Alexandre Vassalotti20c28c12013-11-27 02:26:54 -08006199 result = _Pickle_FastCall(setstate, state);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006200 Py_DECREF(setstate);
6201 if (result == NULL)
6202 return -1;
6203 Py_DECREF(result);
6204 return 0;
6205 }
6206
6207 /* A default __setstate__. First see whether state embeds a
6208 * slot state dict too (a proto 2 addition).
6209 */
Serhiy Storchakafff9a312017-03-21 08:53:25 +02006210 if (PyTuple_Check(state) && PyTuple_GET_SIZE(state) == 2) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006211 PyObject *tmp = state;
6212
6213 state = PyTuple_GET_ITEM(tmp, 0);
6214 slotstate = PyTuple_GET_ITEM(tmp, 1);
6215 Py_INCREF(state);
6216 Py_INCREF(slotstate);
6217 Py_DECREF(tmp);
6218 }
6219 else
6220 slotstate = NULL;
6221
6222 /* Set inst.__dict__ from the state dict (if any). */
6223 if (state != Py_None) {
6224 PyObject *dict;
Antoine Pitroua9f48a02009-05-02 21:41:14 +00006225 PyObject *d_key, *d_value;
6226 Py_ssize_t i;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02006227 _Py_IDENTIFIER(__dict__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006228
6229 if (!PyDict_Check(state)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006230 PickleState *st = _Pickle_GetGlobalState();
6231 PyErr_SetString(st->UnpicklingError, "state is not a dictionary");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006232 goto error;
6233 }
Martin v. Löwis1ee1b6f2011-10-10 18:11:30 +02006234 dict = _PyObject_GetAttrId(inst, &PyId___dict__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006235 if (dict == NULL)
6236 goto error;
6237
Antoine Pitroua9f48a02009-05-02 21:41:14 +00006238 i = 0;
6239 while (PyDict_Next(state, &i, &d_key, &d_value)) {
6240 /* normally the keys for instance attributes are
6241 interned. we should try to do that here. */
6242 Py_INCREF(d_key);
6243 if (PyUnicode_CheckExact(d_key))
6244 PyUnicode_InternInPlace(&d_key);
6245 if (PyObject_SetItem(dict, d_key, d_value) < 0) {
6246 Py_DECREF(d_key);
6247 goto error;
6248 }
6249 Py_DECREF(d_key);
6250 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006251 Py_DECREF(dict);
6252 }
6253
6254 /* Also set instance attributes from the slotstate dict (if any). */
6255 if (slotstate != NULL) {
6256 PyObject *d_key, *d_value;
6257 Py_ssize_t i;
6258
6259 if (!PyDict_Check(slotstate)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006260 PickleState *st = _Pickle_GetGlobalState();
6261 PyErr_SetString(st->UnpicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006262 "slot state is not a dictionary");
6263 goto error;
6264 }
6265 i = 0;
6266 while (PyDict_Next(slotstate, &i, &d_key, &d_value)) {
6267 if (PyObject_SetAttr(inst, d_key, d_value) < 0)
6268 goto error;
6269 }
6270 }
6271
6272 if (0) {
6273 error:
6274 status = -1;
6275 }
6276
6277 Py_DECREF(state);
6278 Py_XDECREF(slotstate);
6279 return status;
6280}
6281
6282static int
6283load_mark(UnpicklerObject *self)
6284{
6285
6286 /* Note that we split the (pickle.py) stack into two stacks, an
6287 * object stack and a mark stack. Here we push a mark onto the
6288 * mark stack.
6289 */
6290
Sergey Fedoseev86b89912018-08-25 12:54:40 +05006291 if (self->num_marks >= self->marks_size) {
Sergey Fedoseev90555ec2018-08-25 15:41:58 +05006292 size_t alloc = ((size_t)self->num_marks << 1) + 20;
6293 Py_ssize_t *marks_new = self->marks;
6294 PyMem_RESIZE(marks_new, Py_ssize_t, alloc);
6295 if (marks_new == NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006296 PyErr_NoMemory();
6297 return -1;
6298 }
Sergey Fedoseev90555ec2018-08-25 15:41:58 +05006299 self->marks = marks_new;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006300 self->marks_size = (Py_ssize_t)alloc;
6301 }
6302
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02006303 self->stack->mark_set = 1;
6304 self->marks[self->num_marks++] = self->stack->fence = Py_SIZE(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006305
6306 return 0;
6307}
6308
6309static int
6310load_reduce(UnpicklerObject *self)
6311{
6312 PyObject *callable = NULL;
6313 PyObject *argtup = NULL;
6314 PyObject *obj = NULL;
6315
6316 PDATA_POP(self->stack, argtup);
6317 if (argtup == NULL)
6318 return -1;
6319 PDATA_POP(self->stack, callable);
6320 if (callable) {
Alexander Belopolskyd92f0402010-07-17 22:50:45 +00006321 obj = PyObject_CallObject(callable, argtup);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006322 Py_DECREF(callable);
6323 }
6324 Py_DECREF(argtup);
6325
6326 if (obj == NULL)
6327 return -1;
6328
6329 PDATA_PUSH(self->stack, obj, -1);
6330 return 0;
6331}
6332
6333/* Just raises an error if we don't know the protocol specified. PROTO
6334 * is the first opcode for protocols >= 2.
6335 */
6336static int
6337load_proto(UnpicklerObject *self)
6338{
6339 char *s;
6340 int i;
6341
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006342 if (_Unpickler_Read(self, &s, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006343 return -1;
6344
6345 i = (unsigned char)s[0];
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006346 if (i <= HIGHEST_PROTOCOL) {
6347 self->proto = i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006348 return 0;
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006349 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006350
6351 PyErr_Format(PyExc_ValueError, "unsupported pickle protocol: %d", i);
6352 return -1;
6353}
6354
Alexandre Vassalottib6a2f2a2013-11-23 20:30:03 -08006355static int
6356load_frame(UnpicklerObject *self)
6357{
6358 char *s;
6359 Py_ssize_t frame_len;
6360
6361 if (_Unpickler_Read(self, &s, 8) < 0)
6362 return -1;
6363
6364 frame_len = calc_binsize(s, 8);
6365 if (frame_len < 0) {
6366 PyErr_Format(PyExc_OverflowError,
6367 "FRAME length exceeds system's maximum of %zd bytes",
6368 PY_SSIZE_T_MAX);
6369 return -1;
6370 }
6371
6372 if (_Unpickler_Read(self, &s, frame_len) < 0)
6373 return -1;
6374
6375 /* Rewind to start of frame */
6376 self->next_read_idx -= frame_len;
6377 return 0;
6378}
6379
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006380static PyObject *
6381load(UnpicklerObject *self)
6382{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006383 PyObject *value = NULL;
Christian Heimes27ea78b2014-01-27 01:03:53 +01006384 char *s = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006385
6386 self->num_marks = 0;
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02006387 self->stack->mark_set = 0;
6388 self->stack->fence = 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006389 self->proto = 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006390 if (Py_SIZE(self->stack))
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006391 Pdata_clear(self->stack, 0);
6392
6393 /* Convenient macros for the dispatch while-switch loop just below. */
6394#define OP(opcode, load_func) \
6395 case opcode: if (load_func(self) < 0) break; continue;
6396
6397#define OP_ARG(opcode, load_func, arg) \
6398 case opcode: if (load_func(self, (arg)) < 0) break; continue;
6399
6400 while (1) {
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03006401 if (_Unpickler_Read(self, &s, 1) < 0) {
6402 PickleState *st = _Pickle_GetGlobalState();
6403 if (PyErr_ExceptionMatches(st->UnpicklingError)) {
6404 PyErr_Format(PyExc_EOFError, "Ran out of input");
6405 }
6406 return NULL;
6407 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006408
6409 switch ((enum opcode)s[0]) {
6410 OP(NONE, load_none)
6411 OP(BININT, load_binint)
6412 OP(BININT1, load_binint1)
6413 OP(BININT2, load_binint2)
6414 OP(INT, load_int)
6415 OP(LONG, load_long)
6416 OP_ARG(LONG1, load_counted_long, 1)
6417 OP_ARG(LONG4, load_counted_long, 4)
6418 OP(FLOAT, load_float)
6419 OP(BINFLOAT, load_binfloat)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006420 OP_ARG(SHORT_BINBYTES, load_counted_binbytes, 1)
6421 OP_ARG(BINBYTES, load_counted_binbytes, 4)
6422 OP_ARG(BINBYTES8, load_counted_binbytes, 8)
6423 OP_ARG(SHORT_BINSTRING, load_counted_binstring, 1)
6424 OP_ARG(BINSTRING, load_counted_binstring, 4)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006425 OP(STRING, load_string)
6426 OP(UNICODE, load_unicode)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006427 OP_ARG(SHORT_BINUNICODE, load_counted_binunicode, 1)
6428 OP_ARG(BINUNICODE, load_counted_binunicode, 4)
6429 OP_ARG(BINUNICODE8, load_counted_binunicode, 8)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006430 OP_ARG(EMPTY_TUPLE, load_counted_tuple, 0)
6431 OP_ARG(TUPLE1, load_counted_tuple, 1)
6432 OP_ARG(TUPLE2, load_counted_tuple, 2)
6433 OP_ARG(TUPLE3, load_counted_tuple, 3)
6434 OP(TUPLE, load_tuple)
6435 OP(EMPTY_LIST, load_empty_list)
6436 OP(LIST, load_list)
6437 OP(EMPTY_DICT, load_empty_dict)
6438 OP(DICT, load_dict)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006439 OP(EMPTY_SET, load_empty_set)
6440 OP(ADDITEMS, load_additems)
6441 OP(FROZENSET, load_frozenset)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006442 OP(OBJ, load_obj)
6443 OP(INST, load_inst)
6444 OP(NEWOBJ, load_newobj)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006445 OP(NEWOBJ_EX, load_newobj_ex)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006446 OP(GLOBAL, load_global)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006447 OP(STACK_GLOBAL, load_stack_global)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006448 OP(APPEND, load_append)
6449 OP(APPENDS, load_appends)
6450 OP(BUILD, load_build)
6451 OP(DUP, load_dup)
6452 OP(BINGET, load_binget)
6453 OP(LONG_BINGET, load_long_binget)
6454 OP(GET, load_get)
6455 OP(MARK, load_mark)
6456 OP(BINPUT, load_binput)
6457 OP(LONG_BINPUT, load_long_binput)
6458 OP(PUT, load_put)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006459 OP(MEMOIZE, load_memoize)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006460 OP(POP, load_pop)
6461 OP(POP_MARK, load_pop_mark)
6462 OP(SETITEM, load_setitem)
6463 OP(SETITEMS, load_setitems)
6464 OP(PERSID, load_persid)
6465 OP(BINPERSID, load_binpersid)
6466 OP(REDUCE, load_reduce)
6467 OP(PROTO, load_proto)
Alexandre Vassalottib6a2f2a2013-11-23 20:30:03 -08006468 OP(FRAME, load_frame)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006469 OP_ARG(EXT1, load_extension, 1)
6470 OP_ARG(EXT2, load_extension, 2)
6471 OP_ARG(EXT4, load_extension, 4)
6472 OP_ARG(NEWTRUE, load_bool, Py_True)
6473 OP_ARG(NEWFALSE, load_bool, Py_False)
6474
6475 case STOP:
6476 break;
6477
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006478 default:
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03006479 {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006480 PickleState *st = _Pickle_GetGlobalState();
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03006481 unsigned char c = (unsigned char) *s;
6482 if (0x20 <= c && c <= 0x7e && c != '\'' && c != '\\') {
6483 PyErr_Format(st->UnpicklingError,
6484 "invalid load key, '%c'.", c);
6485 }
6486 else {
6487 PyErr_Format(st->UnpicklingError,
6488 "invalid load key, '\\x%02x'.", c);
6489 }
6490 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006491 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006492 }
6493
6494 break; /* and we are done! */
6495 }
6496
Alexandre Vassalottib6a2f2a2013-11-23 20:30:03 -08006497 if (PyErr_Occurred()) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006498 return NULL;
6499 }
6500
Victor Stinner2ae57e32013-10-31 13:39:23 +01006501 if (_Unpickler_SkipConsumed(self) < 0)
6502 return NULL;
6503
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006504 PDATA_POP(self->stack, value);
6505 return value;
6506}
6507
Larry Hastings61272b72014-01-07 12:41:53 -08006508/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006509
6510_pickle.Unpickler.load
6511
6512Load a pickle.
6513
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006514Read a pickled object representation from the open file object given
6515in the constructor, and return the reconstituted object hierarchy
6516specified therein.
Larry Hastings61272b72014-01-07 12:41:53 -08006517[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006518
Larry Hastings3cceb382014-01-04 11:09:09 -08006519static PyObject *
Larry Hastingsc2047262014-01-25 20:43:29 -08006520_pickle_Unpickler_load_impl(UnpicklerObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08006521/*[clinic end generated code: output=fdcc488aad675b14 input=acbb91a42fa9b7b9]*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006522{
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006523 UnpicklerObject *unpickler = (UnpicklerObject*)self;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006524
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006525 /* Check whether the Unpickler was initialized correctly. This prevents
6526 segfaulting if a subclass overridden __init__ with a function that does
6527 not call Unpickler.__init__(). Here, we simply ensure that self->read
6528 is not NULL. */
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006529 if (unpickler->read == NULL) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006530 PickleState *st = _Pickle_GetGlobalState();
6531 PyErr_Format(st->UnpicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006532 "Unpickler.__init__() was not called by %s.__init__()",
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006533 Py_TYPE(unpickler)->tp_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006534 return NULL;
6535 }
6536
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006537 return load(unpickler);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006538}
6539
6540/* The name of find_class() is misleading. In newer pickle protocols, this
6541 function is used for loading any global (i.e., functions), not just
6542 classes. The name is kept only for backward compatibility. */
6543
Larry Hastings61272b72014-01-07 12:41:53 -08006544/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006545
6546_pickle.Unpickler.find_class
6547
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006548 module_name: object
6549 global_name: object
6550 /
6551
6552Return an object from a specified module.
6553
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006554If necessary, the module will be imported. Subclasses may override
6555this method (e.g. to restrict unpickling of arbitrary classes and
6556functions).
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006557
6558This method is called whenever a class or a function object is
6559needed. Both arguments passed are str objects.
Larry Hastings61272b72014-01-07 12:41:53 -08006560[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006561
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006562static PyObject *
Larry Hastings89964c42015-04-14 18:07:59 -04006563_pickle_Unpickler_find_class_impl(UnpicklerObject *self,
6564 PyObject *module_name,
6565 PyObject *global_name)
6566/*[clinic end generated code: output=becc08d7f9ed41e3 input=e2e6a865de093ef4]*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006567{
6568 PyObject *global;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006569 PyObject *module;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006570
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006571 /* Try to map the old names used in Python 2.x to the new ones used in
6572 Python 3.x. We do this only with old pickle protocols and when the
6573 user has not disabled the feature. */
6574 if (self->proto < 3 && self->fix_imports) {
6575 PyObject *key;
6576 PyObject *item;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006577 PickleState *st = _Pickle_GetGlobalState();
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006578
6579 /* Check if the global (i.e., a function or a class) was renamed
6580 or moved to another module. */
6581 key = PyTuple_Pack(2, module_name, global_name);
6582 if (key == NULL)
6583 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006584 item = PyDict_GetItemWithError(st->name_mapping_2to3, key);
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006585 Py_DECREF(key);
6586 if (item) {
6587 if (!PyTuple_Check(item) || PyTuple_GET_SIZE(item) != 2) {
6588 PyErr_Format(PyExc_RuntimeError,
6589 "_compat_pickle.NAME_MAPPING values should be "
6590 "2-tuples, not %.200s", Py_TYPE(item)->tp_name);
6591 return NULL;
6592 }
6593 module_name = PyTuple_GET_ITEM(item, 0);
6594 global_name = PyTuple_GET_ITEM(item, 1);
6595 if (!PyUnicode_Check(module_name) ||
6596 !PyUnicode_Check(global_name)) {
6597 PyErr_Format(PyExc_RuntimeError,
6598 "_compat_pickle.NAME_MAPPING values should be "
6599 "pairs of str, not (%.200s, %.200s)",
6600 Py_TYPE(module_name)->tp_name,
6601 Py_TYPE(global_name)->tp_name);
6602 return NULL;
6603 }
6604 }
6605 else if (PyErr_Occurred()) {
6606 return NULL;
6607 }
Serhiy Storchakabfe18242015-03-31 13:12:37 +03006608 else {
6609 /* Check if the module was renamed. */
6610 item = PyDict_GetItemWithError(st->import_mapping_2to3, module_name);
6611 if (item) {
6612 if (!PyUnicode_Check(item)) {
6613 PyErr_Format(PyExc_RuntimeError,
6614 "_compat_pickle.IMPORT_MAPPING values should be "
6615 "strings, not %.200s", Py_TYPE(item)->tp_name);
6616 return NULL;
6617 }
6618 module_name = item;
6619 }
6620 else if (PyErr_Occurred()) {
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006621 return NULL;
6622 }
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006623 }
6624 }
6625
Eric Snow3f9eee62017-09-15 16:35:20 -06006626 module = PyImport_GetModule(module_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006627 if (module == NULL) {
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006628 if (PyErr_Occurred())
6629 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006630 module = PyImport_Import(module_name);
6631 if (module == NULL)
6632 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006633 }
Eric Snow3f9eee62017-09-15 16:35:20 -06006634 global = getattribute(module, global_name, self->proto >= 4);
6635 Py_DECREF(module);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006636 return global;
6637}
6638
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +02006639/*[clinic input]
6640
6641_pickle.Unpickler.__sizeof__ -> Py_ssize_t
6642
6643Returns size in memory, in bytes.
6644[clinic start generated code]*/
6645
6646static Py_ssize_t
6647_pickle_Unpickler___sizeof___impl(UnpicklerObject *self)
6648/*[clinic end generated code: output=119d9d03ad4c7651 input=13333471fdeedf5e]*/
6649{
6650 Py_ssize_t res;
6651
Serhiy Storchaka5c4064e2015-12-19 20:05:25 +02006652 res = _PyObject_SIZE(Py_TYPE(self));
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +02006653 if (self->memo != NULL)
6654 res += self->memo_size * sizeof(PyObject *);
6655 if (self->marks != NULL)
6656 res += self->marks_size * sizeof(Py_ssize_t);
6657 if (self->input_line != NULL)
6658 res += strlen(self->input_line) + 1;
6659 if (self->encoding != NULL)
6660 res += strlen(self->encoding) + 1;
6661 if (self->errors != NULL)
6662 res += strlen(self->errors) + 1;
6663 return res;
6664}
6665
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006666static struct PyMethodDef Unpickler_methods[] = {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006667 _PICKLE_UNPICKLER_LOAD_METHODDEF
6668 _PICKLE_UNPICKLER_FIND_CLASS_METHODDEF
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +02006669 _PICKLE_UNPICKLER___SIZEOF___METHODDEF
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006670 {NULL, NULL} /* sentinel */
6671};
6672
6673static void
6674Unpickler_dealloc(UnpicklerObject *self)
6675{
6676 PyObject_GC_UnTrack((PyObject *)self);
6677 Py_XDECREF(self->readline);
6678 Py_XDECREF(self->read);
Antoine Pitrou04248a82010-10-12 20:51:21 +00006679 Py_XDECREF(self->peek);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006680 Py_XDECREF(self->stack);
6681 Py_XDECREF(self->pers_func);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006682 if (self->buffer.buf != NULL) {
6683 PyBuffer_Release(&self->buffer);
6684 self->buffer.buf = NULL;
6685 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006686
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006687 _Unpickler_MemoCleanup(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006688 PyMem_Free(self->marks);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006689 PyMem_Free(self->input_line);
Victor Stinner49fc8ec2013-07-07 23:30:24 +02006690 PyMem_Free(self->encoding);
6691 PyMem_Free(self->errors);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006692
6693 Py_TYPE(self)->tp_free((PyObject *)self);
6694}
6695
6696static int
6697Unpickler_traverse(UnpicklerObject *self, visitproc visit, void *arg)
6698{
6699 Py_VISIT(self->readline);
6700 Py_VISIT(self->read);
Antoine Pitrou04248a82010-10-12 20:51:21 +00006701 Py_VISIT(self->peek);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006702 Py_VISIT(self->stack);
6703 Py_VISIT(self->pers_func);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006704 return 0;
6705}
6706
6707static int
6708Unpickler_clear(UnpicklerObject *self)
6709{
6710 Py_CLEAR(self->readline);
6711 Py_CLEAR(self->read);
Antoine Pitrou04248a82010-10-12 20:51:21 +00006712 Py_CLEAR(self->peek);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006713 Py_CLEAR(self->stack);
6714 Py_CLEAR(self->pers_func);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006715 if (self->buffer.buf != NULL) {
6716 PyBuffer_Release(&self->buffer);
6717 self->buffer.buf = NULL;
6718 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006719
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006720 _Unpickler_MemoCleanup(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006721 PyMem_Free(self->marks);
6722 self->marks = NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006723 PyMem_Free(self->input_line);
6724 self->input_line = NULL;
Victor Stinner49fc8ec2013-07-07 23:30:24 +02006725 PyMem_Free(self->encoding);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006726 self->encoding = NULL;
Victor Stinner49fc8ec2013-07-07 23:30:24 +02006727 PyMem_Free(self->errors);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006728 self->errors = NULL;
6729
6730 return 0;
6731}
6732
Larry Hastings61272b72014-01-07 12:41:53 -08006733/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006734
6735_pickle.Unpickler.__init__
6736
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006737 file: object
6738 *
6739 fix_imports: bool = True
6740 encoding: str = 'ASCII'
6741 errors: str = 'strict'
6742
6743This takes a binary file for reading a pickle data stream.
6744
6745The protocol version of the pickle is detected automatically, so no
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006746protocol argument is needed. Bytes past the pickled object's
6747representation are ignored.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006748
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006749The argument *file* must have two methods, a read() method that takes
6750an integer argument, and a readline() method that requires no
6751arguments. Both methods should return bytes. Thus *file* can be a
Martin Panter7462b6492015-11-02 03:37:02 +00006752binary file object opened for reading, an io.BytesIO object, or any
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006753other custom object that meets this interface.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006754
6755Optional keyword arguments are *fix_imports*, *encoding* and *errors*,
Martin Panter46f50722016-05-26 05:35:26 +00006756which are used to control compatibility support for pickle stream
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006757generated by Python 2. If *fix_imports* is True, pickle will try to
6758map the old Python 2 names to the new names used in Python 3. The
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006759*encoding* and *errors* tell pickle how to decode 8-bit string
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006760instances pickled by Python 2; these default to 'ASCII' and 'strict',
6761respectively. The *encoding* can be 'bytes' to read these 8-bit
6762string instances as bytes objects.
Larry Hastings61272b72014-01-07 12:41:53 -08006763[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006764
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006765static int
Larry Hastings89964c42015-04-14 18:07:59 -04006766_pickle_Unpickler___init___impl(UnpicklerObject *self, PyObject *file,
6767 int fix_imports, const char *encoding,
6768 const char *errors)
Martin Panter46f50722016-05-26 05:35:26 +00006769/*[clinic end generated code: output=e2c8ce748edc57b0 input=f9b7da04f5f4f335]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006770{
Martin v. Löwis1c67dd92011-10-14 15:16:45 +02006771 _Py_IDENTIFIER(persistent_load);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006772
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006773 /* In case of multiple __init__() calls, clear previous content. */
6774 if (self->read != NULL)
6775 (void)Unpickler_clear(self);
6776
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006777 if (_Unpickler_SetInputStream(self, file) < 0)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006778 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006779
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006780 if (_Unpickler_SetInputEncoding(self, encoding, errors) < 0)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006781 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006782
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006783 self->fix_imports = fix_imports;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006784
Serhiy Storchaka986375e2017-11-30 22:48:31 +02006785 if (init_method_ref((PyObject *)self, &PyId_persistent_load,
6786 &self->pers_func, &self->pers_func_self) < 0)
6787 {
6788 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006789 }
6790
6791 self->stack = (Pdata *)Pdata_New();
6792 if (self->stack == NULL)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006793 return 1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006794
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006795 self->memo_size = 32;
6796 self->memo = _Unpickler_NewMemo(self->memo_size);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006797 if (self->memo == NULL)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006798 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006799
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006800 self->proto = 0;
Alexandre Vassalotti0e7aa8c2009-04-03 04:17:41 +00006801
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006802 return 0;
6803}
6804
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006805
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006806/* Define a proxy object for the Unpickler's internal memo object. This is to
6807 * avoid breaking code like:
6808 * unpickler.memo.clear()
6809 * and
6810 * unpickler.memo = saved_memo
6811 * Is this a good idea? Not really, but we don't want to break code that uses
6812 * it. Note that we don't implement the entire mapping API here. This is
6813 * intentional, as these should be treated as black-box implementation details.
6814 *
6815 * We do, however, have to implement pickling/unpickling support because of
Victor Stinner121aab42011-09-29 23:40:53 +02006816 * real-world code like cvs2svn.
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006817 */
6818
Larry Hastings61272b72014-01-07 12:41:53 -08006819/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006820_pickle.UnpicklerMemoProxy.clear
6821
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006822Remove all items from memo.
Larry Hastings61272b72014-01-07 12:41:53 -08006823[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006824
Larry Hastings3cceb382014-01-04 11:09:09 -08006825static PyObject *
6826_pickle_UnpicklerMemoProxy_clear_impl(UnpicklerMemoProxyObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08006827/*[clinic end generated code: output=d20cd43f4ba1fb1f input=b1df7c52e7afd9bd]*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006828{
6829 _Unpickler_MemoCleanup(self->unpickler);
6830 self->unpickler->memo = _Unpickler_NewMemo(self->unpickler->memo_size);
6831 if (self->unpickler->memo == NULL)
6832 return NULL;
6833 Py_RETURN_NONE;
6834}
6835
Larry Hastings61272b72014-01-07 12:41:53 -08006836/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006837_pickle.UnpicklerMemoProxy.copy
6838
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006839Copy the memo to a new object.
Larry Hastings61272b72014-01-07 12:41:53 -08006840[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006841
Larry Hastings3cceb382014-01-04 11:09:09 -08006842static PyObject *
6843_pickle_UnpicklerMemoProxy_copy_impl(UnpicklerMemoProxyObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08006844/*[clinic end generated code: output=e12af7e9bc1e4c77 input=97769247ce032c1d]*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006845{
6846 Py_ssize_t i;
6847 PyObject *new_memo = PyDict_New();
6848 if (new_memo == NULL)
6849 return NULL;
6850
6851 for (i = 0; i < self->unpickler->memo_size; i++) {
6852 int status;
6853 PyObject *key, *value;
6854
6855 value = self->unpickler->memo[i];
6856 if (value == NULL)
6857 continue;
6858
6859 key = PyLong_FromSsize_t(i);
6860 if (key == NULL)
6861 goto error;
6862 status = PyDict_SetItem(new_memo, key, value);
6863 Py_DECREF(key);
6864 if (status < 0)
6865 goto error;
6866 }
6867 return new_memo;
6868
6869error:
6870 Py_DECREF(new_memo);
6871 return NULL;
6872}
6873
Larry Hastings61272b72014-01-07 12:41:53 -08006874/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006875_pickle.UnpicklerMemoProxy.__reduce__
6876
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006877Implement pickling support.
Larry Hastings61272b72014-01-07 12:41:53 -08006878[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006879
Larry Hastings3cceb382014-01-04 11:09:09 -08006880static PyObject *
6881_pickle_UnpicklerMemoProxy___reduce___impl(UnpicklerMemoProxyObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08006882/*[clinic end generated code: output=6da34ac048d94cca input=6920862413407199]*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006883{
6884 PyObject *reduce_value;
6885 PyObject *constructor_args;
Larry Hastings3cceb382014-01-04 11:09:09 -08006886 PyObject *contents = _pickle_UnpicklerMemoProxy_copy_impl(self);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006887 if (contents == NULL)
6888 return NULL;
6889
6890 reduce_value = PyTuple_New(2);
6891 if (reduce_value == NULL) {
6892 Py_DECREF(contents);
6893 return NULL;
6894 }
6895 constructor_args = PyTuple_New(1);
6896 if (constructor_args == NULL) {
6897 Py_DECREF(contents);
6898 Py_DECREF(reduce_value);
6899 return NULL;
6900 }
6901 PyTuple_SET_ITEM(constructor_args, 0, contents);
6902 Py_INCREF((PyObject *)&PyDict_Type);
6903 PyTuple_SET_ITEM(reduce_value, 0, (PyObject *)&PyDict_Type);
6904 PyTuple_SET_ITEM(reduce_value, 1, constructor_args);
6905 return reduce_value;
6906}
6907
6908static PyMethodDef unpicklerproxy_methods[] = {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006909 _PICKLE_UNPICKLERMEMOPROXY_CLEAR_METHODDEF
6910 _PICKLE_UNPICKLERMEMOPROXY_COPY_METHODDEF
6911 _PICKLE_UNPICKLERMEMOPROXY___REDUCE___METHODDEF
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006912 {NULL, NULL} /* sentinel */
6913};
6914
6915static void
6916UnpicklerMemoProxy_dealloc(UnpicklerMemoProxyObject *self)
6917{
6918 PyObject_GC_UnTrack(self);
6919 Py_XDECREF(self->unpickler);
6920 PyObject_GC_Del((PyObject *)self);
6921}
6922
6923static int
6924UnpicklerMemoProxy_traverse(UnpicklerMemoProxyObject *self,
6925 visitproc visit, void *arg)
6926{
6927 Py_VISIT(self->unpickler);
6928 return 0;
6929}
6930
6931static int
6932UnpicklerMemoProxy_clear(UnpicklerMemoProxyObject *self)
6933{
6934 Py_CLEAR(self->unpickler);
6935 return 0;
6936}
6937
6938static PyTypeObject UnpicklerMemoProxyType = {
6939 PyVarObject_HEAD_INIT(NULL, 0)
6940 "_pickle.UnpicklerMemoProxy", /*tp_name*/
6941 sizeof(UnpicklerMemoProxyObject), /*tp_basicsize*/
6942 0,
6943 (destructor)UnpicklerMemoProxy_dealloc, /* tp_dealloc */
6944 0, /* tp_print */
6945 0, /* tp_getattr */
6946 0, /* tp_setattr */
6947 0, /* tp_compare */
6948 0, /* tp_repr */
6949 0, /* tp_as_number */
6950 0, /* tp_as_sequence */
6951 0, /* tp_as_mapping */
Georg Brandlf038b322010-10-18 07:35:09 +00006952 PyObject_HashNotImplemented, /* tp_hash */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006953 0, /* tp_call */
6954 0, /* tp_str */
6955 PyObject_GenericGetAttr, /* tp_getattro */
6956 PyObject_GenericSetAttr, /* tp_setattro */
6957 0, /* tp_as_buffer */
6958 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC,
6959 0, /* tp_doc */
6960 (traverseproc)UnpicklerMemoProxy_traverse, /* tp_traverse */
6961 (inquiry)UnpicklerMemoProxy_clear, /* tp_clear */
6962 0, /* tp_richcompare */
6963 0, /* tp_weaklistoffset */
6964 0, /* tp_iter */
6965 0, /* tp_iternext */
6966 unpicklerproxy_methods, /* tp_methods */
6967};
6968
6969static PyObject *
6970UnpicklerMemoProxy_New(UnpicklerObject *unpickler)
6971{
6972 UnpicklerMemoProxyObject *self;
6973
6974 self = PyObject_GC_New(UnpicklerMemoProxyObject,
6975 &UnpicklerMemoProxyType);
6976 if (self == NULL)
6977 return NULL;
6978 Py_INCREF(unpickler);
6979 self->unpickler = unpickler;
6980 PyObject_GC_Track(self);
6981 return (PyObject *)self;
6982}
6983
6984/*****************************************************************************/
6985
6986
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006987static PyObject *
6988Unpickler_get_memo(UnpicklerObject *self)
6989{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006990 return UnpicklerMemoProxy_New(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006991}
6992
6993static int
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006994Unpickler_set_memo(UnpicklerObject *self, PyObject *obj)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006995{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006996 PyObject **new_memo;
6997 Py_ssize_t new_memo_size = 0;
6998 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006999
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007000 if (obj == NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007001 PyErr_SetString(PyExc_TypeError,
7002 "attribute deletion is not supported");
7003 return -1;
7004 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007005
7006 if (Py_TYPE(obj) == &UnpicklerMemoProxyType) {
7007 UnpicklerObject *unpickler =
7008 ((UnpicklerMemoProxyObject *)obj)->unpickler;
7009
7010 new_memo_size = unpickler->memo_size;
7011 new_memo = _Unpickler_NewMemo(new_memo_size);
7012 if (new_memo == NULL)
7013 return -1;
7014
7015 for (i = 0; i < new_memo_size; i++) {
7016 Py_XINCREF(unpickler->memo[i]);
7017 new_memo[i] = unpickler->memo[i];
7018 }
7019 }
7020 else if (PyDict_Check(obj)) {
7021 Py_ssize_t i = 0;
7022 PyObject *key, *value;
7023
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02007024 new_memo_size = PyDict_GET_SIZE(obj);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007025 new_memo = _Unpickler_NewMemo(new_memo_size);
7026 if (new_memo == NULL)
7027 return -1;
7028
7029 while (PyDict_Next(obj, &i, &key, &value)) {
7030 Py_ssize_t idx;
7031 if (!PyLong_Check(key)) {
7032 PyErr_SetString(PyExc_TypeError,
7033 "memo key must be integers");
7034 goto error;
7035 }
7036 idx = PyLong_AsSsize_t(key);
7037 if (idx == -1 && PyErr_Occurred())
7038 goto error;
Christian Heimesa24b4d22013-07-01 15:17:45 +02007039 if (idx < 0) {
7040 PyErr_SetString(PyExc_ValueError,
Christian Heimes80878792013-07-01 15:23:39 +02007041 "memo key must be positive integers.");
Christian Heimesa24b4d22013-07-01 15:17:45 +02007042 goto error;
7043 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007044 if (_Unpickler_MemoPut(self, idx, value) < 0)
7045 goto error;
7046 }
7047 }
7048 else {
7049 PyErr_Format(PyExc_TypeError,
7050 "'memo' attribute must be an UnpicklerMemoProxy object"
7051 "or dict, not %.200s", Py_TYPE(obj)->tp_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007052 return -1;
7053 }
7054
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007055 _Unpickler_MemoCleanup(self);
7056 self->memo_size = new_memo_size;
7057 self->memo = new_memo;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007058
7059 return 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007060
7061 error:
7062 if (new_memo_size) {
7063 i = new_memo_size;
7064 while (--i >= 0) {
7065 Py_XDECREF(new_memo[i]);
7066 }
7067 PyMem_FREE(new_memo);
7068 }
7069 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007070}
7071
7072static PyObject *
7073Unpickler_get_persload(UnpicklerObject *self)
7074{
Serhiy Storchaka986375e2017-11-30 22:48:31 +02007075 if (self->pers_func == NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007076 PyErr_SetString(PyExc_AttributeError, "persistent_load");
Serhiy Storchaka986375e2017-11-30 22:48:31 +02007077 return NULL;
7078 }
7079 return reconstruct_method(self->pers_func, self->pers_func_self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007080}
7081
7082static int
7083Unpickler_set_persload(UnpicklerObject *self, PyObject *value)
7084{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007085 if (value == NULL) {
7086 PyErr_SetString(PyExc_TypeError,
7087 "attribute deletion is not supported");
7088 return -1;
7089 }
7090 if (!PyCallable_Check(value)) {
7091 PyErr_SetString(PyExc_TypeError,
7092 "persistent_load must be a callable taking "
7093 "one argument");
7094 return -1;
7095 }
7096
Serhiy Storchaka986375e2017-11-30 22:48:31 +02007097 self->pers_func_self = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007098 Py_INCREF(value);
Serhiy Storchakaec397562016-04-06 09:50:03 +03007099 Py_XSETREF(self->pers_func, value);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007100
7101 return 0;
7102}
7103
7104static PyGetSetDef Unpickler_getsets[] = {
7105 {"memo", (getter)Unpickler_get_memo, (setter)Unpickler_set_memo},
7106 {"persistent_load", (getter)Unpickler_get_persload,
7107 (setter)Unpickler_set_persload},
7108 {NULL}
7109};
7110
7111static PyTypeObject Unpickler_Type = {
7112 PyVarObject_HEAD_INIT(NULL, 0)
7113 "_pickle.Unpickler", /*tp_name*/
7114 sizeof(UnpicklerObject), /*tp_basicsize*/
7115 0, /*tp_itemsize*/
7116 (destructor)Unpickler_dealloc, /*tp_dealloc*/
7117 0, /*tp_print*/
7118 0, /*tp_getattr*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007119 0, /*tp_setattr*/
Mark Dickinsone94c6792009-02-02 20:36:42 +00007120 0, /*tp_reserved*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007121 0, /*tp_repr*/
7122 0, /*tp_as_number*/
7123 0, /*tp_as_sequence*/
7124 0, /*tp_as_mapping*/
7125 0, /*tp_hash*/
7126 0, /*tp_call*/
7127 0, /*tp_str*/
7128 0, /*tp_getattro*/
7129 0, /*tp_setattro*/
7130 0, /*tp_as_buffer*/
7131 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC,
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007132 _pickle_Unpickler___init____doc__, /*tp_doc*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007133 (traverseproc)Unpickler_traverse, /*tp_traverse*/
7134 (inquiry)Unpickler_clear, /*tp_clear*/
7135 0, /*tp_richcompare*/
7136 0, /*tp_weaklistoffset*/
7137 0, /*tp_iter*/
7138 0, /*tp_iternext*/
7139 Unpickler_methods, /*tp_methods*/
7140 0, /*tp_members*/
7141 Unpickler_getsets, /*tp_getset*/
7142 0, /*tp_base*/
7143 0, /*tp_dict*/
7144 0, /*tp_descr_get*/
7145 0, /*tp_descr_set*/
7146 0, /*tp_dictoffset*/
Larry Hastingsb7ccb202014-01-18 23:50:21 -08007147 _pickle_Unpickler___init__, /*tp_init*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007148 PyType_GenericAlloc, /*tp_alloc*/
7149 PyType_GenericNew, /*tp_new*/
7150 PyObject_GC_Del, /*tp_free*/
7151 0, /*tp_is_gc*/
7152};
7153
Larry Hastings61272b72014-01-07 12:41:53 -08007154/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007155
7156_pickle.dump
7157
7158 obj: object
7159 file: object
7160 protocol: object = NULL
7161 *
7162 fix_imports: bool = True
7163
7164Write a pickled representation of obj to the open file object file.
7165
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007166This is equivalent to ``Pickler(file, protocol).dump(obj)``, but may
7167be more efficient.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007168
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007169The optional *protocol* argument tells the pickler to use the given
Łukasz Langac51d8c92018-04-03 23:06:53 -07007170protocol; supported protocols are 0, 1, 2, 3 and 4. The default
7171protocol is 4. It was introduced in Python 3.4, it is incompatible
7172with previous versions.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007173
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007174Specifying a negative protocol version selects the highest protocol
7175version supported. The higher the protocol used, the more recent the
7176version of Python needed to read the pickle produced.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007177
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007178The *file* argument must have a write() method that accepts a single
7179bytes argument. It can thus be a file object opened for binary
Martin Panter7462b6492015-11-02 03:37:02 +00007180writing, an io.BytesIO instance, or any other custom object that meets
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007181this interface.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007182
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007183If *fix_imports* is True and protocol is less than 3, pickle will try
7184to map the new Python 3 names to the old module names used in Python
71852, so that the pickle data stream is readable with Python 2.
Larry Hastings61272b72014-01-07 12:41:53 -08007186[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007187
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007188static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007189_pickle_dump_impl(PyObject *module, PyObject *obj, PyObject *file,
Larry Hastings89964c42015-04-14 18:07:59 -04007190 PyObject *protocol, int fix_imports)
Łukasz Langac51d8c92018-04-03 23:06:53 -07007191/*[clinic end generated code: output=a4774d5fde7d34de input=93f1408489a87472]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007192{
7193 PicklerObject *pickler = _Pickler_New();
7194
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007195 if (pickler == NULL)
7196 return NULL;
7197
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007198 if (_Pickler_SetProtocol(pickler, protocol, fix_imports) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007199 goto error;
7200
7201 if (_Pickler_SetOutputStream(pickler, file) < 0)
7202 goto error;
7203
7204 if (dump(pickler, obj) < 0)
7205 goto error;
7206
7207 if (_Pickler_FlushToFile(pickler) < 0)
7208 goto error;
7209
7210 Py_DECREF(pickler);
7211 Py_RETURN_NONE;
7212
7213 error:
7214 Py_XDECREF(pickler);
7215 return NULL;
7216}
7217
Larry Hastings61272b72014-01-07 12:41:53 -08007218/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007219
7220_pickle.dumps
7221
7222 obj: object
7223 protocol: object = NULL
7224 *
7225 fix_imports: bool = True
7226
7227Return the pickled representation of the object as a bytes object.
7228
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007229The optional *protocol* argument tells the pickler to use the given
7230protocol; supported protocols are 0, 1, 2, 3 and 4. The default
Łukasz Langac51d8c92018-04-03 23:06:53 -07007231protocol is 4. It was introduced in Python 3.4, it is incompatible
7232with previous versions.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007233
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007234Specifying a negative protocol version selects the highest protocol
7235version supported. The higher the protocol used, the more recent the
7236version of Python needed to read the pickle produced.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007237
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007238If *fix_imports* is True and *protocol* is less than 3, pickle will
7239try to map the new Python 3 names to the old module names used in
7240Python 2, so that the pickle data stream is readable with Python 2.
Larry Hastings61272b72014-01-07 12:41:53 -08007241[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007242
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007243static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007244_pickle_dumps_impl(PyObject *module, PyObject *obj, PyObject *protocol,
Larry Hastings89964c42015-04-14 18:07:59 -04007245 int fix_imports)
Łukasz Langac51d8c92018-04-03 23:06:53 -07007246/*[clinic end generated code: output=d75d5cda456fd261 input=b6efb45a7d19b5ab]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007247{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007248 PyObject *result;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007249 PicklerObject *pickler = _Pickler_New();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007250
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007251 if (pickler == NULL)
7252 return NULL;
7253
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007254 if (_Pickler_SetProtocol(pickler, protocol, fix_imports) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007255 goto error;
7256
7257 if (dump(pickler, obj) < 0)
7258 goto error;
7259
7260 result = _Pickler_GetString(pickler);
7261 Py_DECREF(pickler);
7262 return result;
7263
7264 error:
7265 Py_XDECREF(pickler);
7266 return NULL;
7267}
7268
Larry Hastings61272b72014-01-07 12:41:53 -08007269/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007270
7271_pickle.load
7272
7273 file: object
7274 *
7275 fix_imports: bool = True
7276 encoding: str = 'ASCII'
7277 errors: str = 'strict'
7278
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007279Read and return an object from the pickle data stored in a file.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007280
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007281This is equivalent to ``Unpickler(file).load()``, but may be more
7282efficient.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007283
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007284The protocol version of the pickle is detected automatically, so no
7285protocol argument is needed. Bytes past the pickled object's
7286representation are ignored.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007287
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007288The argument *file* must have two methods, a read() method that takes
7289an integer argument, and a readline() method that requires no
7290arguments. Both methods should return bytes. Thus *file* can be a
Martin Panter7462b6492015-11-02 03:37:02 +00007291binary file object opened for reading, an io.BytesIO object, or any
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007292other custom object that meets this interface.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007293
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007294Optional keyword arguments are *fix_imports*, *encoding* and *errors*,
Martin Panter46f50722016-05-26 05:35:26 +00007295which are used to control compatibility support for pickle stream
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007296generated by Python 2. If *fix_imports* is True, pickle will try to
7297map the old Python 2 names to the new names used in Python 3. The
7298*encoding* and *errors* tell pickle how to decode 8-bit string
7299instances pickled by Python 2; these default to 'ASCII' and 'strict',
7300respectively. The *encoding* can be 'bytes' to read these 8-bit
7301string instances as bytes objects.
Larry Hastings61272b72014-01-07 12:41:53 -08007302[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007303
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007304static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007305_pickle_load_impl(PyObject *module, PyObject *file, int fix_imports,
Larry Hastings89964c42015-04-14 18:07:59 -04007306 const char *encoding, const char *errors)
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007307/*[clinic end generated code: output=69e298160285199e input=01b44dd3fc07afa7]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007308{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007309 PyObject *result;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007310 UnpicklerObject *unpickler = _Unpickler_New();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007311
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007312 if (unpickler == NULL)
7313 return NULL;
7314
7315 if (_Unpickler_SetInputStream(unpickler, file) < 0)
7316 goto error;
7317
7318 if (_Unpickler_SetInputEncoding(unpickler, encoding, errors) < 0)
7319 goto error;
7320
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007321 unpickler->fix_imports = fix_imports;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007322
7323 result = load(unpickler);
7324 Py_DECREF(unpickler);
7325 return result;
7326
7327 error:
7328 Py_XDECREF(unpickler);
7329 return NULL;
7330}
7331
Larry Hastings61272b72014-01-07 12:41:53 -08007332/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007333
7334_pickle.loads
7335
7336 data: object
7337 *
7338 fix_imports: bool = True
7339 encoding: str = 'ASCII'
7340 errors: str = 'strict'
7341
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007342Read and return an object from the given pickle data.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007343
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007344The protocol version of the pickle is detected automatically, so no
7345protocol argument is needed. Bytes past the pickled object's
7346representation are ignored.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007347
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007348Optional keyword arguments are *fix_imports*, *encoding* and *errors*,
Martin Panter46f50722016-05-26 05:35:26 +00007349which are used to control compatibility support for pickle stream
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007350generated by Python 2. If *fix_imports* is True, pickle will try to
7351map the old Python 2 names to the new names used in Python 3. The
7352*encoding* and *errors* tell pickle how to decode 8-bit string
7353instances pickled by Python 2; these default to 'ASCII' and 'strict',
7354respectively. The *encoding* can be 'bytes' to read these 8-bit
7355string instances as bytes objects.
Larry Hastings61272b72014-01-07 12:41:53 -08007356[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007357
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007358static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007359_pickle_loads_impl(PyObject *module, PyObject *data, int fix_imports,
Larry Hastings89964c42015-04-14 18:07:59 -04007360 const char *encoding, const char *errors)
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007361/*[clinic end generated code: output=1e7cb2343f2c440f input=70605948a719feb9]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007362{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007363 PyObject *result;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007364 UnpicklerObject *unpickler = _Unpickler_New();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007365
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007366 if (unpickler == NULL)
7367 return NULL;
7368
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007369 if (_Unpickler_SetStringInput(unpickler, data) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007370 goto error;
7371
7372 if (_Unpickler_SetInputEncoding(unpickler, encoding, errors) < 0)
7373 goto error;
7374
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007375 unpickler->fix_imports = fix_imports;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007376
7377 result = load(unpickler);
7378 Py_DECREF(unpickler);
7379 return result;
7380
7381 error:
7382 Py_XDECREF(unpickler);
7383 return NULL;
7384}
7385
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007386static struct PyMethodDef pickle_methods[] = {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007387 _PICKLE_DUMP_METHODDEF
7388 _PICKLE_DUMPS_METHODDEF
7389 _PICKLE_LOAD_METHODDEF
7390 _PICKLE_LOADS_METHODDEF
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007391 {NULL, NULL} /* sentinel */
7392};
7393
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007394static int
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007395pickle_clear(PyObject *m)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007396{
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007397 _Pickle_ClearState(_Pickle_GetState(m));
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007398 return 0;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007399}
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007400
Stefan Krahf483b0f2013-12-14 13:43:10 +01007401static void
7402pickle_free(PyObject *m)
7403{
7404 _Pickle_ClearState(_Pickle_GetState(m));
7405}
7406
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007407static int
7408pickle_traverse(PyObject *m, visitproc visit, void *arg)
7409{
7410 PickleState *st = _Pickle_GetState(m);
7411 Py_VISIT(st->PickleError);
7412 Py_VISIT(st->PicklingError);
7413 Py_VISIT(st->UnpicklingError);
7414 Py_VISIT(st->dispatch_table);
7415 Py_VISIT(st->extension_registry);
7416 Py_VISIT(st->extension_cache);
7417 Py_VISIT(st->inverted_registry);
7418 Py_VISIT(st->name_mapping_2to3);
7419 Py_VISIT(st->import_mapping_2to3);
7420 Py_VISIT(st->name_mapping_3to2);
7421 Py_VISIT(st->import_mapping_3to2);
7422 Py_VISIT(st->codecs_encode);
Serhiy Storchaka58e41342015-03-31 14:07:24 +03007423 Py_VISIT(st->getattr);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007424 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007425}
7426
7427static struct PyModuleDef _picklemodule = {
7428 PyModuleDef_HEAD_INIT,
Stefan Krahf483b0f2013-12-14 13:43:10 +01007429 "_pickle", /* m_name */
7430 pickle_module_doc, /* m_doc */
7431 sizeof(PickleState), /* m_size */
7432 pickle_methods, /* m_methods */
7433 NULL, /* m_reload */
7434 pickle_traverse, /* m_traverse */
7435 pickle_clear, /* m_clear */
7436 (freefunc)pickle_free /* m_free */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007437};
7438
7439PyMODINIT_FUNC
7440PyInit__pickle(void)
7441{
7442 PyObject *m;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007443 PickleState *st;
7444
7445 m = PyState_FindModule(&_picklemodule);
7446 if (m) {
7447 Py_INCREF(m);
7448 return m;
7449 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007450
7451 if (PyType_Ready(&Unpickler_Type) < 0)
7452 return NULL;
7453 if (PyType_Ready(&Pickler_Type) < 0)
7454 return NULL;
7455 if (PyType_Ready(&Pdata_Type) < 0)
7456 return NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007457 if (PyType_Ready(&PicklerMemoProxyType) < 0)
7458 return NULL;
7459 if (PyType_Ready(&UnpicklerMemoProxyType) < 0)
7460 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007461
7462 /* Create the module and add the functions. */
7463 m = PyModule_Create(&_picklemodule);
7464 if (m == NULL)
7465 return NULL;
7466
Antoine Pitrou8391cf42011-07-15 21:01:21 +02007467 Py_INCREF(&Pickler_Type);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007468 if (PyModule_AddObject(m, "Pickler", (PyObject *)&Pickler_Type) < 0)
7469 return NULL;
Antoine Pitrou8391cf42011-07-15 21:01:21 +02007470 Py_INCREF(&Unpickler_Type);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007471 if (PyModule_AddObject(m, "Unpickler", (PyObject *)&Unpickler_Type) < 0)
7472 return NULL;
7473
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007474 st = _Pickle_GetState(m);
7475
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007476 /* Initialize the exceptions. */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007477 st->PickleError = PyErr_NewException("_pickle.PickleError", NULL, NULL);
7478 if (st->PickleError == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007479 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007480 st->PicklingError = \
7481 PyErr_NewException("_pickle.PicklingError", st->PickleError, NULL);
7482 if (st->PicklingError == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007483 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007484 st->UnpicklingError = \
7485 PyErr_NewException("_pickle.UnpicklingError", st->PickleError, NULL);
7486 if (st->UnpicklingError == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007487 return NULL;
7488
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007489 Py_INCREF(st->PickleError);
7490 if (PyModule_AddObject(m, "PickleError", st->PickleError) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007491 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007492 Py_INCREF(st->PicklingError);
7493 if (PyModule_AddObject(m, "PicklingError", st->PicklingError) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007494 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007495 Py_INCREF(st->UnpicklingError);
7496 if (PyModule_AddObject(m, "UnpicklingError", st->UnpicklingError) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007497 return NULL;
7498
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007499 if (_Pickle_InitState(st) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007500 return NULL;
7501
7502 return m;
7503}