blob: 40b29a7d69368cde4251667703477b5a97f1b9cf [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
Alexandre Vassalottica2d6102008-06-12 18:26:05 +000023/* Bump this when new opcodes are added to the pickle protocol. */
24enum {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +010025 HIGHEST_PROTOCOL = 4,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +000026 DEFAULT_PROTOCOL = 3
27};
28
Alexandre Vassalottica2d6102008-06-12 18:26:05 +000029/* Pickle opcodes. These must be kept updated with pickle.py.
30 Extensive docs are in pickletools.py. */
31enum opcode {
32 MARK = '(',
33 STOP = '.',
34 POP = '0',
35 POP_MARK = '1',
36 DUP = '2',
37 FLOAT = 'F',
38 INT = 'I',
39 BININT = 'J',
40 BININT1 = 'K',
41 LONG = 'L',
42 BININT2 = 'M',
43 NONE = 'N',
44 PERSID = 'P',
45 BINPERSID = 'Q',
46 REDUCE = 'R',
47 STRING = 'S',
48 BINSTRING = 'T',
49 SHORT_BINSTRING = 'U',
50 UNICODE = 'V',
51 BINUNICODE = 'X',
52 APPEND = 'a',
53 BUILD = 'b',
54 GLOBAL = 'c',
55 DICT = 'd',
56 EMPTY_DICT = '}',
57 APPENDS = 'e',
58 GET = 'g',
59 BINGET = 'h',
60 INST = 'i',
61 LONG_BINGET = 'j',
62 LIST = 'l',
63 EMPTY_LIST = ']',
64 OBJ = 'o',
65 PUT = 'p',
66 BINPUT = 'q',
67 LONG_BINPUT = 'r',
68 SETITEM = 's',
69 TUPLE = 't',
70 EMPTY_TUPLE = ')',
71 SETITEMS = 'u',
72 BINFLOAT = 'G',
73
74 /* Protocol 2. */
75 PROTO = '\x80',
76 NEWOBJ = '\x81',
77 EXT1 = '\x82',
78 EXT2 = '\x83',
79 EXT4 = '\x84',
80 TUPLE1 = '\x85',
81 TUPLE2 = '\x86',
82 TUPLE3 = '\x87',
83 NEWTRUE = '\x88',
84 NEWFALSE = '\x89',
85 LONG1 = '\x8a',
86 LONG4 = '\x8b',
87
88 /* Protocol 3 (Python 3.x) */
89 BINBYTES = 'B',
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +010090 SHORT_BINBYTES = 'C',
91
92 /* Protocol 4 */
93 SHORT_BINUNICODE = '\x8c',
94 BINUNICODE8 = '\x8d',
95 BINBYTES8 = '\x8e',
96 EMPTY_SET = '\x8f',
97 ADDITEMS = '\x90',
98 FROZENSET = '\x91',
99 NEWOBJ_EX = '\x92',
100 STACK_GLOBAL = '\x93',
101 MEMOIZE = '\x94',
102 FRAME = '\x95'
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000103};
104
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000105enum {
106 /* Keep in synch with pickle.Pickler._BATCHSIZE. This is how many elements
107 batch_list/dict() pumps out before doing APPENDS/SETITEMS. Nothing will
108 break if this gets out of synch with pickle.py, but it's unclear that would
109 help anything either. */
110 BATCHSIZE = 1000,
111
112 /* Nesting limit until Pickler, when running in "fast mode", starts
113 checking for self-referential data-structures. */
114 FAST_NESTING_LIMIT = 50,
115
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000116 /* Initial size of the write buffer of Pickler. */
117 WRITE_BUF_SIZE = 4096,
118
Antoine Pitrou04248a82010-10-12 20:51:21 +0000119 /* Prefetch size when unpickling (disabled on unpeekable streams) */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100120 PREFETCH = 8192 * 16,
121
Serhiy Storchaka1211c9a2018-01-20 16:42:44 +0200122 FRAME_SIZE_MIN = 4,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100123 FRAME_SIZE_TARGET = 64 * 1024,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100124 FRAME_HEADER_SIZE = 9
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000125};
126
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800127/*************************************************************************/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000128
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800129/* State of the pickle module, per PEP 3121. */
130typedef struct {
131 /* Exception classes for pickle. */
132 PyObject *PickleError;
133 PyObject *PicklingError;
134 PyObject *UnpicklingError;
Larry Hastings61272b72014-01-07 12:41:53 -0800135
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800136 /* copyreg.dispatch_table, {type_object: pickling_function} */
137 PyObject *dispatch_table;
Antoine Pitroud9dfaa92009-06-04 20:32:06 +0000138
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800139 /* For the extension opcodes EXT1, EXT2 and EXT4. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000140
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800141 /* copyreg._extension_registry, {(module_name, function_name): code} */
142 PyObject *extension_registry;
143 /* copyreg._extension_cache, {code: object} */
144 PyObject *extension_cache;
145 /* copyreg._inverted_registry, {code: (module_name, function_name)} */
146 PyObject *inverted_registry;
147
148 /* Import mappings for compatibility with Python 2.x */
149
150 /* _compat_pickle.NAME_MAPPING,
151 {(oldmodule, oldname): (newmodule, newname)} */
152 PyObject *name_mapping_2to3;
153 /* _compat_pickle.IMPORT_MAPPING, {oldmodule: newmodule} */
154 PyObject *import_mapping_2to3;
155 /* Same, but with REVERSE_NAME_MAPPING / REVERSE_IMPORT_MAPPING */
156 PyObject *name_mapping_3to2;
157 PyObject *import_mapping_3to2;
158
159 /* codecs.encode, used for saving bytes in older protocols */
160 PyObject *codecs_encode;
Serhiy Storchaka58e41342015-03-31 14:07:24 +0300161 /* builtins.getattr, used for saving nested names with protocol < 4 */
162 PyObject *getattr;
Serhiy Storchaka0d554d72015-10-10 22:42:18 +0300163 /* functools.partial, used for implementing __newobj_ex__ with protocols
164 2 and 3 */
165 PyObject *partial;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800166} PickleState;
167
168/* Forward declaration of the _pickle module definition. */
169static struct PyModuleDef _picklemodule;
170
171/* Given a module object, get its per-module state. */
172static PickleState *
173_Pickle_GetState(PyObject *module)
174{
175 return (PickleState *)PyModule_GetState(module);
176}
177
178/* Find the module instance imported in the currently running sub-interpreter
179 and get its state. */
180static PickleState *
181_Pickle_GetGlobalState(void)
182{
183 return _Pickle_GetState(PyState_FindModule(&_picklemodule));
184}
185
186/* Clear the given pickle module state. */
187static void
188_Pickle_ClearState(PickleState *st)
189{
190 Py_CLEAR(st->PickleError);
191 Py_CLEAR(st->PicklingError);
192 Py_CLEAR(st->UnpicklingError);
193 Py_CLEAR(st->dispatch_table);
194 Py_CLEAR(st->extension_registry);
195 Py_CLEAR(st->extension_cache);
196 Py_CLEAR(st->inverted_registry);
197 Py_CLEAR(st->name_mapping_2to3);
198 Py_CLEAR(st->import_mapping_2to3);
199 Py_CLEAR(st->name_mapping_3to2);
200 Py_CLEAR(st->import_mapping_3to2);
201 Py_CLEAR(st->codecs_encode);
Serhiy Storchaka58e41342015-03-31 14:07:24 +0300202 Py_CLEAR(st->getattr);
Victor Stinner9ba97df2015-11-17 12:15:07 +0100203 Py_CLEAR(st->partial);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800204}
205
206/* Initialize the given pickle module state. */
207static int
208_Pickle_InitState(PickleState *st)
209{
Serhiy Storchaka58e41342015-03-31 14:07:24 +0300210 PyObject *builtins;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800211 PyObject *copyreg = NULL;
212 PyObject *compat_pickle = NULL;
213 PyObject *codecs = NULL;
Serhiy Storchaka0d554d72015-10-10 22:42:18 +0300214 PyObject *functools = NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800215
Serhiy Storchaka58e41342015-03-31 14:07:24 +0300216 builtins = PyEval_GetBuiltins();
217 if (builtins == NULL)
218 goto error;
219 st->getattr = PyDict_GetItemString(builtins, "getattr");
220 if (st->getattr == NULL)
221 goto error;
222 Py_INCREF(st->getattr);
223
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800224 copyreg = PyImport_ImportModule("copyreg");
225 if (!copyreg)
226 goto error;
227 st->dispatch_table = PyObject_GetAttrString(copyreg, "dispatch_table");
228 if (!st->dispatch_table)
229 goto error;
230 if (!PyDict_CheckExact(st->dispatch_table)) {
231 PyErr_Format(PyExc_RuntimeError,
232 "copyreg.dispatch_table should be a dict, not %.200s",
233 Py_TYPE(st->dispatch_table)->tp_name);
234 goto error;
235 }
236 st->extension_registry = \
237 PyObject_GetAttrString(copyreg, "_extension_registry");
238 if (!st->extension_registry)
239 goto error;
240 if (!PyDict_CheckExact(st->extension_registry)) {
241 PyErr_Format(PyExc_RuntimeError,
242 "copyreg._extension_registry should be a dict, "
243 "not %.200s", Py_TYPE(st->extension_registry)->tp_name);
244 goto error;
245 }
246 st->inverted_registry = \
247 PyObject_GetAttrString(copyreg, "_inverted_registry");
248 if (!st->inverted_registry)
249 goto error;
250 if (!PyDict_CheckExact(st->inverted_registry)) {
251 PyErr_Format(PyExc_RuntimeError,
252 "copyreg._inverted_registry should be a dict, "
253 "not %.200s", Py_TYPE(st->inverted_registry)->tp_name);
254 goto error;
255 }
256 st->extension_cache = PyObject_GetAttrString(copyreg, "_extension_cache");
257 if (!st->extension_cache)
258 goto error;
259 if (!PyDict_CheckExact(st->extension_cache)) {
260 PyErr_Format(PyExc_RuntimeError,
261 "copyreg._extension_cache should be a dict, "
262 "not %.200s", Py_TYPE(st->extension_cache)->tp_name);
263 goto error;
264 }
265 Py_CLEAR(copyreg);
266
267 /* Load the 2.x -> 3.x stdlib module mapping tables */
268 compat_pickle = PyImport_ImportModule("_compat_pickle");
269 if (!compat_pickle)
270 goto error;
271 st->name_mapping_2to3 = \
272 PyObject_GetAttrString(compat_pickle, "NAME_MAPPING");
273 if (!st->name_mapping_2to3)
274 goto error;
275 if (!PyDict_CheckExact(st->name_mapping_2to3)) {
276 PyErr_Format(PyExc_RuntimeError,
277 "_compat_pickle.NAME_MAPPING should be a dict, not %.200s",
278 Py_TYPE(st->name_mapping_2to3)->tp_name);
279 goto error;
280 }
281 st->import_mapping_2to3 = \
282 PyObject_GetAttrString(compat_pickle, "IMPORT_MAPPING");
283 if (!st->import_mapping_2to3)
284 goto error;
285 if (!PyDict_CheckExact(st->import_mapping_2to3)) {
286 PyErr_Format(PyExc_RuntimeError,
287 "_compat_pickle.IMPORT_MAPPING should be a dict, "
288 "not %.200s", Py_TYPE(st->import_mapping_2to3)->tp_name);
289 goto error;
290 }
291 /* ... and the 3.x -> 2.x mapping tables */
292 st->name_mapping_3to2 = \
293 PyObject_GetAttrString(compat_pickle, "REVERSE_NAME_MAPPING");
294 if (!st->name_mapping_3to2)
295 goto error;
296 if (!PyDict_CheckExact(st->name_mapping_3to2)) {
297 PyErr_Format(PyExc_RuntimeError,
298 "_compat_pickle.REVERSE_NAME_MAPPING should be a dict, "
299 "not %.200s", Py_TYPE(st->name_mapping_3to2)->tp_name);
300 goto error;
301 }
302 st->import_mapping_3to2 = \
303 PyObject_GetAttrString(compat_pickle, "REVERSE_IMPORT_MAPPING");
304 if (!st->import_mapping_3to2)
305 goto error;
306 if (!PyDict_CheckExact(st->import_mapping_3to2)) {
307 PyErr_Format(PyExc_RuntimeError,
308 "_compat_pickle.REVERSE_IMPORT_MAPPING should be a dict, "
309 "not %.200s", Py_TYPE(st->import_mapping_3to2)->tp_name);
310 goto error;
311 }
312 Py_CLEAR(compat_pickle);
313
314 codecs = PyImport_ImportModule("codecs");
315 if (codecs == NULL)
316 goto error;
317 st->codecs_encode = PyObject_GetAttrString(codecs, "encode");
318 if (st->codecs_encode == NULL) {
319 goto error;
320 }
321 if (!PyCallable_Check(st->codecs_encode)) {
322 PyErr_Format(PyExc_RuntimeError,
323 "codecs.encode should be a callable, not %.200s",
324 Py_TYPE(st->codecs_encode)->tp_name);
325 goto error;
326 }
327 Py_CLEAR(codecs);
328
Serhiy Storchaka0d554d72015-10-10 22:42:18 +0300329 functools = PyImport_ImportModule("functools");
330 if (!functools)
331 goto error;
332 st->partial = PyObject_GetAttrString(functools, "partial");
333 if (!st->partial)
334 goto error;
335 Py_CLEAR(functools);
336
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800337 return 0;
338
339 error:
340 Py_CLEAR(copyreg);
341 Py_CLEAR(compat_pickle);
342 Py_CLEAR(codecs);
Serhiy Storchaka0d554d72015-10-10 22:42:18 +0300343 Py_CLEAR(functools);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800344 _Pickle_ClearState(st);
345 return -1;
346}
347
348/* Helper for calling a function with a single argument quickly.
349
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800350 This function steals the reference of the given argument. */
351static PyObject *
352_Pickle_FastCall(PyObject *func, PyObject *obj)
353{
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800354 PyObject *result;
355
Victor Stinner7bfb42d2016-12-05 17:04:32 +0100356 result = PyObject_CallFunctionObjArgs(func, obj, NULL);
Victor Stinner75210692016-08-19 18:59:15 +0200357 Py_DECREF(obj);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -0800358 return result;
359}
360
361/*************************************************************************/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000362
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200363/* Retrieve and deconstruct a method for avoiding a reference cycle
364 (pickler -> bound method of pickler -> pickler) */
365static int
366init_method_ref(PyObject *self, _Py_Identifier *name,
367 PyObject **method_func, PyObject **method_self)
368{
369 PyObject *func, *func2;
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200370 int ret;
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200371
372 /* *method_func and *method_self should be consistent. All refcount decrements
373 should be occurred after setting *method_self and *method_func. */
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200374 ret = _PyObject_LookupAttrId(self, name, &func);
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200375 if (func == NULL) {
376 *method_self = NULL;
377 Py_CLEAR(*method_func);
Serhiy Storchakaf320be72018-01-25 10:49:40 +0200378 return ret;
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200379 }
380
381 if (PyMethod_Check(func) && PyMethod_GET_SELF(func) == self) {
382 /* Deconstruct a bound Python method */
383 func2 = PyMethod_GET_FUNCTION(func);
384 Py_INCREF(func2);
385 *method_self = self; /* borrowed */
386 Py_XSETREF(*method_func, func2);
387 Py_DECREF(func);
388 return 0;
389 }
390 else {
391 *method_self = NULL;
392 Py_XSETREF(*method_func, func);
393 return 0;
394 }
395}
396
397/* Bind a method if it was deconstructed */
398static PyObject *
399reconstruct_method(PyObject *func, PyObject *self)
400{
401 if (self) {
402 return PyMethod_New(func, self);
403 }
404 else {
405 Py_INCREF(func);
406 return func;
407 }
408}
409
410static PyObject *
411call_method(PyObject *func, PyObject *self, PyObject *obj)
412{
413 if (self) {
414 return PyObject_CallFunctionObjArgs(func, self, obj, NULL);
415 }
416 else {
417 return PyObject_CallFunctionObjArgs(func, obj, NULL);
418 }
419}
420
421/*************************************************************************/
422
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000423/* Internal data type used as the unpickling stack. */
424typedef struct {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000425 PyObject_VAR_HEAD
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000426 PyObject **data;
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200427 int mark_set; /* is MARK set? */
428 Py_ssize_t fence; /* position of top MARK or 0 */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000429 Py_ssize_t allocated; /* number of slots in data allocated */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000430} Pdata;
431
432static void
433Pdata_dealloc(Pdata *self)
434{
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200435 Py_ssize_t i = Py_SIZE(self);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000436 while (--i >= 0) {
437 Py_DECREF(self->data[i]);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000438 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000439 PyMem_FREE(self->data);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000440 PyObject_Del(self);
441}
442
443static PyTypeObject Pdata_Type = {
444 PyVarObject_HEAD_INIT(NULL, 0)
445 "_pickle.Pdata", /*tp_name*/
446 sizeof(Pdata), /*tp_basicsize*/
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +0200447 sizeof(PyObject *), /*tp_itemsize*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000448 (destructor)Pdata_dealloc, /*tp_dealloc*/
449};
450
451static PyObject *
452Pdata_New(void)
453{
454 Pdata *self;
455
456 if (!(self = PyObject_New(Pdata, &Pdata_Type)))
457 return NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000458 Py_SIZE(self) = 0;
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200459 self->mark_set = 0;
460 self->fence = 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000461 self->allocated = 8;
462 self->data = PyMem_MALLOC(self->allocated * sizeof(PyObject *));
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000463 if (self->data)
464 return (PyObject *)self;
465 Py_DECREF(self);
466 return PyErr_NoMemory();
467}
468
469
470/* Retain only the initial clearto items. If clearto >= the current
471 * number of items, this is a (non-erroneous) NOP.
472 */
473static int
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200474Pdata_clear(Pdata *self, Py_ssize_t clearto)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000475{
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200476 Py_ssize_t i = Py_SIZE(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000477
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200478 assert(clearto >= self->fence);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000479 if (clearto >= i)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000480 return 0;
481
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000482 while (--i >= clearto) {
483 Py_CLEAR(self->data[i]);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000484 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000485 Py_SIZE(self) = clearto;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000486 return 0;
487}
488
489static int
490Pdata_grow(Pdata *self)
491{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000492 PyObject **data = self->data;
Victor Stinnerf13c46c2014-08-17 21:05:55 +0200493 size_t allocated = (size_t)self->allocated;
494 size_t new_allocated;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000495
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000496 new_allocated = (allocated >> 3) + 6;
497 /* check for integer overflow */
Victor Stinnerf13c46c2014-08-17 21:05:55 +0200498 if (new_allocated > (size_t)PY_SSIZE_T_MAX - allocated)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000499 goto nomemory;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000500 new_allocated += allocated;
Benjamin Peterson59b08c12015-06-27 13:41:33 -0500501 PyMem_RESIZE(data, PyObject *, new_allocated);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000502 if (data == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000503 goto nomemory;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000504
505 self->data = data;
Victor Stinnerf13c46c2014-08-17 21:05:55 +0200506 self->allocated = (Py_ssize_t)new_allocated;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000507 return 0;
508
509 nomemory:
510 PyErr_NoMemory();
511 return -1;
512}
513
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200514static int
515Pdata_stack_underflow(Pdata *self)
516{
517 PickleState *st = _Pickle_GetGlobalState();
518 PyErr_SetString(st->UnpicklingError,
519 self->mark_set ?
520 "unexpected MARK found" :
521 "unpickling stack underflow");
522 return -1;
523}
524
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000525/* D is a Pdata*. Pop the topmost element and store it into V, which
526 * must be an lvalue holding PyObject*. On stack underflow, UnpicklingError
527 * is raised and V is set to NULL.
528 */
529static PyObject *
530Pdata_pop(Pdata *self)
531{
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200532 if (Py_SIZE(self) <= self->fence) {
533 Pdata_stack_underflow(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000534 return NULL;
535 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000536 return self->data[--Py_SIZE(self)];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000537}
538#define PDATA_POP(D, V) do { (V) = Pdata_pop((D)); } while (0)
539
540static int
541Pdata_push(Pdata *self, PyObject *obj)
542{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000543 if (Py_SIZE(self) == self->allocated && Pdata_grow(self) < 0) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000544 return -1;
545 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000546 self->data[Py_SIZE(self)++] = obj;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000547 return 0;
548}
549
550/* Push an object on stack, transferring its ownership to the stack. */
551#define PDATA_PUSH(D, O, ER) do { \
552 if (Pdata_push((D), (O)) < 0) return (ER); } while(0)
553
554/* Push an object on stack, adding a new reference to the object. */
555#define PDATA_APPEND(D, O, ER) do { \
556 Py_INCREF((O)); \
557 if (Pdata_push((D), (O)) < 0) return (ER); } while(0)
558
559static PyObject *
560Pdata_poptuple(Pdata *self, Py_ssize_t start)
561{
562 PyObject *tuple;
563 Py_ssize_t len, i, j;
564
Serhiy Storchaka59fb6342015-12-06 22:01:35 +0200565 if (start < self->fence) {
566 Pdata_stack_underflow(self);
567 return NULL;
568 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000569 len = Py_SIZE(self) - start;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000570 tuple = PyTuple_New(len);
571 if (tuple == NULL)
572 return NULL;
573 for (i = start, j = 0; j < len; i++, j++)
574 PyTuple_SET_ITEM(tuple, j, self->data[i]);
575
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000576 Py_SIZE(self) = start;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000577 return tuple;
578}
579
580static PyObject *
581Pdata_poplist(Pdata *self, Py_ssize_t start)
582{
583 PyObject *list;
584 Py_ssize_t len, i, j;
585
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000586 len = Py_SIZE(self) - start;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000587 list = PyList_New(len);
588 if (list == NULL)
589 return NULL;
590 for (i = start, j = 0; j < len; i++, j++)
591 PyList_SET_ITEM(list, j, self->data[i]);
592
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000593 Py_SIZE(self) = start;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000594 return list;
595}
596
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000597typedef struct {
598 PyObject *me_key;
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200599 Py_ssize_t me_value;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000600} PyMemoEntry;
601
602typedef struct {
603 Py_ssize_t mt_mask;
604 Py_ssize_t mt_used;
605 Py_ssize_t mt_allocated;
606 PyMemoEntry *mt_table;
607} PyMemoTable;
608
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000609typedef struct PicklerObject {
610 PyObject_HEAD
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000611 PyMemoTable *memo; /* Memo table, keep track of the seen
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000612 objects to support self-referential objects
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000613 pickling. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000614 PyObject *pers_func; /* persistent_id() method, can be NULL */
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200615 PyObject *pers_func_self; /* borrowed reference to self if pers_func
616 is an unbound method, NULL otherwise */
Antoine Pitrou8d3c2902012-03-04 18:31:48 +0100617 PyObject *dispatch_table; /* private dispatch_table, can be NULL */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000618
619 PyObject *write; /* write() method of the output stream. */
620 PyObject *output_buffer; /* Write into a local bytearray buffer before
621 flushing to the stream. */
622 Py_ssize_t output_len; /* Length of output_buffer. */
623 Py_ssize_t max_output_len; /* Allocation size of output_buffer. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000624 int proto; /* Pickle protocol number, >= 0 */
625 int bin; /* Boolean, true if proto > 0 */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100626 int framing; /* True when framing is enabled, proto >= 4 */
627 Py_ssize_t frame_start; /* Position in output_buffer where the
Martin Pantera90a4a92016-05-30 04:04:50 +0000628 current frame begins. -1 if there
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100629 is no frame currently open. */
630
631 Py_ssize_t buf_size; /* Size of the current buffered pickle data */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000632 int fast; /* Enable fast mode if set to a true value.
633 The fast mode disable the usage of memo,
634 therefore speeding the pickling process by
635 not generating superfluous PUT opcodes. It
636 should not be used if with self-referential
637 objects. */
638 int fast_nesting;
Antoine Pitroud9dfaa92009-06-04 20:32:06 +0000639 int fix_imports; /* Indicate whether Pickler should fix
640 the name of globals for Python 2.x. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000641 PyObject *fast_memo;
642} PicklerObject;
643
644typedef struct UnpicklerObject {
645 PyObject_HEAD
646 Pdata *stack; /* Pickle data stack, store unpickled objects. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000647
648 /* The unpickler memo is just an array of PyObject *s. Using a dict
649 is unnecessary, since the keys are contiguous ints. */
650 PyObject **memo;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100651 Py_ssize_t memo_size; /* Capacity of the memo array */
652 Py_ssize_t memo_len; /* Number of objects in the memo */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000653
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000654 PyObject *pers_func; /* persistent_load() method, can be NULL. */
Serhiy Storchaka986375e2017-11-30 22:48:31 +0200655 PyObject *pers_func_self; /* borrowed reference to self if pers_func
656 is an unbound method, NULL otherwise */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000657
658 Py_buffer buffer;
659 char *input_buffer;
660 char *input_line;
661 Py_ssize_t input_len;
662 Py_ssize_t next_read_idx;
Antoine Pitrou04248a82010-10-12 20:51:21 +0000663 Py_ssize_t prefetched_idx; /* index of first prefetched byte */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100664
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000665 PyObject *read; /* read() method of the input stream. */
666 PyObject *readline; /* readline() method of the input stream. */
Antoine Pitrou04248a82010-10-12 20:51:21 +0000667 PyObject *peek; /* peek() method of the input stream, or NULL */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000668
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000669 char *encoding; /* Name of the encoding to be used for
670 decoding strings pickled using Python
671 2.x. The default value is "ASCII" */
672 char *errors; /* Name of errors handling scheme to used when
673 decoding strings. The default value is
674 "strict". */
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -0500675 Py_ssize_t *marks; /* Mark stack, used for unpickling container
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000676 objects. */
677 Py_ssize_t num_marks; /* Number of marks in the mark stack. */
678 Py_ssize_t marks_size; /* Current allocated size of the mark stack. */
Antoine Pitroud9dfaa92009-06-04 20:32:06 +0000679 int proto; /* Protocol of the pickle loaded. */
680 int fix_imports; /* Indicate whether Unpickler should fix
681 the name of globals pickled by Python 2.x. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000682} UnpicklerObject;
683
Serhiy Storchaka3c1f0f12014-01-27 10:34:22 +0200684typedef struct {
685 PyObject_HEAD
686 PicklerObject *pickler; /* Pickler whose memo table we're proxying. */
687} PicklerMemoProxyObject;
688
689typedef struct {
690 PyObject_HEAD
691 UnpicklerObject *unpickler;
692} UnpicklerMemoProxyObject;
693
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000694/* Forward declarations */
695static int save(PicklerObject *, PyObject *, int);
696static int save_reduce(PicklerObject *, PyObject *, PyObject *);
697static PyTypeObject Pickler_Type;
698static PyTypeObject Unpickler_Type;
699
Serhiy Storchaka3c1f0f12014-01-27 10:34:22 +0200700#include "clinic/_pickle.c.h"
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000701
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000702/*************************************************************************
Serhiy Storchaka95949422013-08-27 19:40:23 +0300703 A custom hashtable mapping void* to Python ints. This is used by the pickler
704 for memoization. Using a custom hashtable rather than PyDict allows us to skip
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000705 a bunch of unnecessary object creation. This makes a huge performance
706 difference. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000707
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000708#define MT_MINSIZE 8
709#define PERTURB_SHIFT 5
710
711
712static PyMemoTable *
713PyMemoTable_New(void)
714{
715 PyMemoTable *memo = PyMem_MALLOC(sizeof(PyMemoTable));
716 if (memo == NULL) {
717 PyErr_NoMemory();
718 return NULL;
719 }
720
721 memo->mt_used = 0;
722 memo->mt_allocated = MT_MINSIZE;
723 memo->mt_mask = MT_MINSIZE - 1;
724 memo->mt_table = PyMem_MALLOC(MT_MINSIZE * sizeof(PyMemoEntry));
725 if (memo->mt_table == NULL) {
726 PyMem_FREE(memo);
727 PyErr_NoMemory();
728 return NULL;
729 }
730 memset(memo->mt_table, 0, MT_MINSIZE * sizeof(PyMemoEntry));
731
732 return memo;
733}
734
735static PyMemoTable *
736PyMemoTable_Copy(PyMemoTable *self)
737{
738 Py_ssize_t i;
739 PyMemoTable *new = PyMemoTable_New();
740 if (new == NULL)
741 return NULL;
742
743 new->mt_used = self->mt_used;
744 new->mt_allocated = self->mt_allocated;
745 new->mt_mask = self->mt_mask;
746 /* The table we get from _New() is probably smaller than we wanted.
747 Free it and allocate one that's the right size. */
748 PyMem_FREE(new->mt_table);
Benjamin Peterson59b08c12015-06-27 13:41:33 -0500749 new->mt_table = PyMem_NEW(PyMemoEntry, self->mt_allocated);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000750 if (new->mt_table == NULL) {
751 PyMem_FREE(new);
Victor Stinner42024562013-07-12 00:53:57 +0200752 PyErr_NoMemory();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000753 return NULL;
754 }
755 for (i = 0; i < self->mt_allocated; i++) {
756 Py_XINCREF(self->mt_table[i].me_key);
757 }
758 memcpy(new->mt_table, self->mt_table,
759 sizeof(PyMemoEntry) * self->mt_allocated);
760
761 return new;
762}
763
764static Py_ssize_t
765PyMemoTable_Size(PyMemoTable *self)
766{
767 return self->mt_used;
768}
769
770static int
771PyMemoTable_Clear(PyMemoTable *self)
772{
773 Py_ssize_t i = self->mt_allocated;
774
775 while (--i >= 0) {
776 Py_XDECREF(self->mt_table[i].me_key);
777 }
778 self->mt_used = 0;
779 memset(self->mt_table, 0, self->mt_allocated * sizeof(PyMemoEntry));
780 return 0;
781}
782
783static void
784PyMemoTable_Del(PyMemoTable *self)
785{
786 if (self == NULL)
787 return;
788 PyMemoTable_Clear(self);
789
790 PyMem_FREE(self->mt_table);
791 PyMem_FREE(self);
792}
793
794/* Since entries cannot be deleted from this hashtable, _PyMemoTable_Lookup()
795 can be considerably simpler than dictobject.c's lookdict(). */
796static PyMemoEntry *
797_PyMemoTable_Lookup(PyMemoTable *self, PyObject *key)
798{
799 size_t i;
800 size_t perturb;
801 size_t mask = (size_t)self->mt_mask;
802 PyMemoEntry *table = self->mt_table;
803 PyMemoEntry *entry;
Benjamin Peterson8f67d082010-10-17 20:54:53 +0000804 Py_hash_t hash = (Py_hash_t)key >> 3;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000805
806 i = hash & mask;
807 entry = &table[i];
808 if (entry->me_key == NULL || entry->me_key == key)
809 return entry;
810
811 for (perturb = hash; ; perturb >>= PERTURB_SHIFT) {
812 i = (i << 2) + i + perturb + 1;
813 entry = &table[i & mask];
814 if (entry->me_key == NULL || entry->me_key == key)
815 return entry;
816 }
Barry Warsawb2e57942017-09-14 18:13:16 -0700817 Py_UNREACHABLE();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000818}
819
820/* Returns -1 on failure, 0 on success. */
821static int
822_PyMemoTable_ResizeTable(PyMemoTable *self, Py_ssize_t min_size)
823{
824 PyMemoEntry *oldtable = NULL;
825 PyMemoEntry *oldentry, *newentry;
826 Py_ssize_t new_size = MT_MINSIZE;
827 Py_ssize_t to_process;
828
829 assert(min_size > 0);
830
831 /* Find the smallest valid table size >= min_size. */
832 while (new_size < min_size && new_size > 0)
833 new_size <<= 1;
834 if (new_size <= 0) {
835 PyErr_NoMemory();
836 return -1;
837 }
838 /* new_size needs to be a power of two. */
839 assert((new_size & (new_size - 1)) == 0);
840
841 /* Allocate new table. */
842 oldtable = self->mt_table;
Benjamin Peterson59b08c12015-06-27 13:41:33 -0500843 self->mt_table = PyMem_NEW(PyMemoEntry, new_size);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000844 if (self->mt_table == NULL) {
Victor Stinner8ca72e22013-07-12 00:53:26 +0200845 self->mt_table = oldtable;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000846 PyErr_NoMemory();
847 return -1;
848 }
849 self->mt_allocated = new_size;
850 self->mt_mask = new_size - 1;
851 memset(self->mt_table, 0, sizeof(PyMemoEntry) * new_size);
852
853 /* Copy entries from the old table. */
854 to_process = self->mt_used;
855 for (oldentry = oldtable; to_process > 0; oldentry++) {
856 if (oldentry->me_key != NULL) {
857 to_process--;
858 /* newentry is a pointer to a chunk of the new
859 mt_table, so we're setting the key:value pair
860 in-place. */
861 newentry = _PyMemoTable_Lookup(self, oldentry->me_key);
862 newentry->me_key = oldentry->me_key;
863 newentry->me_value = oldentry->me_value;
864 }
865 }
866
867 /* Deallocate the old table. */
868 PyMem_FREE(oldtable);
869 return 0;
870}
871
872/* Returns NULL on failure, a pointer to the value otherwise. */
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200873static Py_ssize_t *
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000874PyMemoTable_Get(PyMemoTable *self, PyObject *key)
875{
876 PyMemoEntry *entry = _PyMemoTable_Lookup(self, key);
877 if (entry->me_key == NULL)
878 return NULL;
879 return &entry->me_value;
880}
881
882/* Returns -1 on failure, 0 on success. */
883static int
Antoine Pitrou82be19f2011-08-29 23:09:33 +0200884PyMemoTable_Set(PyMemoTable *self, PyObject *key, Py_ssize_t value)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000885{
886 PyMemoEntry *entry;
887
888 assert(key != NULL);
889
890 entry = _PyMemoTable_Lookup(self, key);
891 if (entry->me_key != NULL) {
892 entry->me_value = value;
893 return 0;
894 }
895 Py_INCREF(key);
896 entry->me_key = key;
897 entry->me_value = value;
898 self->mt_used++;
899
900 /* If we added a key, we can safely resize. Otherwise just return!
901 * If used >= 2/3 size, adjust size. Normally, this quaduples the size.
902 *
903 * Quadrupling the size improves average table sparseness
904 * (reducing collisions) at the cost of some memory. It also halves
905 * the number of expensive resize operations in a growing memo table.
906 *
907 * Very large memo tables (over 50K items) use doubling instead.
908 * This may help applications with severe memory constraints.
909 */
910 if (!(self->mt_used * 3 >= (self->mt_mask + 1) * 2))
911 return 0;
912 return _PyMemoTable_ResizeTable(self,
913 (self->mt_used > 50000 ? 2 : 4) * self->mt_used);
914}
915
916#undef MT_MINSIZE
917#undef PERTURB_SHIFT
918
919/*************************************************************************/
920
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000921
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000922static int
923_Pickler_ClearBuffer(PicklerObject *self)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +0000924{
Serhiy Storchaka48842712016-04-06 09:45:48 +0300925 Py_XSETREF(self->output_buffer,
Serhiy Storchaka4a1e70f2015-12-27 12:36:18 +0200926 PyBytes_FromStringAndSize(NULL, self->max_output_len));
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000927 if (self->output_buffer == NULL)
Amaury Forgeot d'Arc87eee632008-10-17 20:15:53 +0000928 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000929 self->output_len = 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100930 self->frame_start = -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000931 return 0;
932}
Amaury Forgeot d'Arc87eee632008-10-17 20:15:53 +0000933
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100934static void
Antoine Pitrou8f2ee6e2013-11-23 21:05:08 +0100935_write_size64(char *out, size_t value)
936{
Victor Stinnerf13c46c2014-08-17 21:05:55 +0200937 size_t i;
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -0800938
Serhiy Storchakafad85aa2015-11-07 15:42:38 +0200939 Py_BUILD_ASSERT(sizeof(size_t) <= 8);
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -0800940
941 for (i = 0; i < sizeof(size_t); i++) {
942 out[i] = (unsigned char)((value >> (8 * i)) & 0xff);
943 }
944 for (i = sizeof(size_t); i < 8; i++) {
945 out[i] = 0;
Alexandre Vassalottided929b2013-11-24 22:41:13 -0800946 }
Antoine Pitrou8f2ee6e2013-11-23 21:05:08 +0100947}
948
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100949static int
950_Pickler_CommitFrame(PicklerObject *self)
951{
952 size_t frame_len;
953 char *qdata;
954
955 if (!self->framing || self->frame_start == -1)
956 return 0;
957 frame_len = self->output_len - self->frame_start - FRAME_HEADER_SIZE;
958 qdata = PyBytes_AS_STRING(self->output_buffer) + self->frame_start;
Serhiy Storchaka1211c9a2018-01-20 16:42:44 +0200959 if (frame_len >= FRAME_SIZE_MIN) {
960 qdata[0] = FRAME;
961 _write_size64(qdata + 1, frame_len);
962 }
963 else {
964 memmove(qdata, qdata + FRAME_HEADER_SIZE, frame_len);
965 self->output_len -= FRAME_HEADER_SIZE;
966 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100967 self->frame_start = -1;
968 return 0;
969}
970
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000971static PyObject *
972_Pickler_GetString(PicklerObject *self)
973{
974 PyObject *output_buffer = self->output_buffer;
975
976 assert(self->output_buffer != NULL);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100977
978 if (_Pickler_CommitFrame(self))
979 return NULL;
980
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000981 self->output_buffer = NULL;
982 /* Resize down to exact size */
983 if (_PyBytes_Resize(&output_buffer, self->output_len) < 0)
984 return NULL;
985 return output_buffer;
986}
987
988static int
989_Pickler_FlushToFile(PicklerObject *self)
990{
991 PyObject *output, *result;
992
993 assert(self->write != NULL);
994
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +0100995 /* This will commit the frame first */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +0000996 output = _Pickler_GetString(self);
997 if (output == NULL)
998 return -1;
999
Alexandre Vassalotti20c28c12013-11-27 02:26:54 -08001000 result = _Pickle_FastCall(self->write, output);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001001 Py_XDECREF(result);
1002 return (result == NULL) ? -1 : 0;
1003}
1004
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01001005static int
1006_Pickler_OpcodeBoundary(PicklerObject *self)
1007{
1008 Py_ssize_t frame_len;
1009
1010 if (!self->framing || self->frame_start == -1) {
1011 return 0;
1012 }
1013 frame_len = self->output_len - self->frame_start - FRAME_HEADER_SIZE;
1014 if (frame_len >= FRAME_SIZE_TARGET) {
1015 if(_Pickler_CommitFrame(self)) {
1016 return -1;
1017 }
Miss Islington (bot)e86db342018-02-03 17:41:43 -08001018 /* Flush the content of the committed frame to the underlying
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01001019 * file and reuse the pickler buffer for the next frame so as
1020 * to limit memory usage when dumping large complex objects to
1021 * a file.
1022 *
1023 * self->write is NULL when called via dumps.
1024 */
1025 if (self->write != NULL) {
1026 if (_Pickler_FlushToFile(self) < 0) {
1027 return -1;
1028 }
1029 if (_Pickler_ClearBuffer(self) < 0) {
1030 return -1;
1031 }
1032 }
1033 }
1034 return 0;
1035}
1036
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001037static Py_ssize_t
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001038_Pickler_Write(PicklerObject *self, const char *s, Py_ssize_t data_len)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001039{
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001040 Py_ssize_t i, n, required;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001041 char *buffer;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001042 int need_new_frame;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001043
1044 assert(s != NULL);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001045 need_new_frame = (self->framing && self->frame_start == -1);
1046
1047 if (need_new_frame)
1048 n = data_len + FRAME_HEADER_SIZE;
1049 else
1050 n = data_len;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001051
1052 required = self->output_len + n;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001053 if (required > self->max_output_len) {
1054 /* Make place in buffer for the pickle chunk */
1055 if (self->output_len >= PY_SSIZE_T_MAX / 2 - n) {
1056 PyErr_NoMemory();
1057 return -1;
1058 }
1059 self->max_output_len = (self->output_len + n) / 2 * 3;
1060 if (_PyBytes_Resize(&self->output_buffer, self->max_output_len) < 0)
1061 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001062 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001063 buffer = PyBytes_AS_STRING(self->output_buffer);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001064 if (need_new_frame) {
1065 /* Setup new frame */
1066 Py_ssize_t frame_start = self->output_len;
1067 self->frame_start = frame_start;
1068 for (i = 0; i < FRAME_HEADER_SIZE; i++) {
1069 /* Write an invalid value, for debugging */
1070 buffer[frame_start + i] = 0xFE;
1071 }
1072 self->output_len += FRAME_HEADER_SIZE;
1073 }
1074 if (data_len < 8) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001075 /* This is faster than memcpy when the string is short. */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001076 for (i = 0; i < data_len; i++) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001077 buffer[self->output_len + i] = s[i];
1078 }
1079 }
1080 else {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001081 memcpy(buffer + self->output_len, s, data_len);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001082 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001083 self->output_len += data_len;
1084 return data_len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001085}
1086
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001087static PicklerObject *
1088_Pickler_New(void)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001089{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001090 PicklerObject *self;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001091
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001092 self = PyObject_GC_New(PicklerObject, &Pickler_Type);
1093 if (self == NULL)
1094 return NULL;
1095
1096 self->pers_func = NULL;
Antoine Pitrou8d3c2902012-03-04 18:31:48 +01001097 self->dispatch_table = NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001098 self->write = NULL;
1099 self->proto = 0;
1100 self->bin = 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001101 self->framing = 0;
1102 self->frame_start = -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001103 self->fast = 0;
1104 self->fast_nesting = 0;
1105 self->fix_imports = 0;
1106 self->fast_memo = NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001107 self->max_output_len = WRITE_BUF_SIZE;
1108 self->output_len = 0;
Victor Stinner68c8ea22013-07-11 22:56:25 +02001109
1110 self->memo = PyMemoTable_New();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001111 self->output_buffer = PyBytes_FromStringAndSize(NULL,
1112 self->max_output_len);
Victor Stinner68c8ea22013-07-11 22:56:25 +02001113
1114 if (self->memo == NULL || self->output_buffer == NULL) {
Victor Stinnerc31df042013-07-12 00:08:59 +02001115 Py_DECREF(self);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001116 return NULL;
1117 }
1118 return self;
1119}
1120
1121static int
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001122_Pickler_SetProtocol(PicklerObject *self, PyObject *protocol, int fix_imports)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001123{
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001124 long proto;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001125
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001126 if (protocol == NULL || protocol == Py_None) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001127 proto = DEFAULT_PROTOCOL;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001128 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001129 else {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001130 proto = PyLong_AsLong(protocol);
1131 if (proto < 0) {
1132 if (proto == -1 && PyErr_Occurred())
1133 return -1;
1134 proto = HIGHEST_PROTOCOL;
1135 }
1136 else if (proto > HIGHEST_PROTOCOL) {
1137 PyErr_Format(PyExc_ValueError, "pickle protocol must be <= %d",
1138 HIGHEST_PROTOCOL);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001139 return -1;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001140 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001141 }
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08001142 self->proto = (int)proto;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001143 self->bin = proto > 0;
1144 self->fix_imports = fix_imports && proto < 3;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001145 return 0;
1146}
1147
1148/* Returns -1 (with an exception set) on failure, 0 on success. This may
1149 be called once on a freshly created Pickler. */
1150static int
1151_Pickler_SetOutputStream(PicklerObject *self, PyObject *file)
1152{
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001153 _Py_IDENTIFIER(write);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001154 assert(file != NULL);
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001155 if (_PyObject_LookupAttrId(file, &PyId_write, &self->write) < 0) {
1156 return -1;
1157 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001158 if (self->write == NULL) {
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001159 PyErr_SetString(PyExc_TypeError,
1160 "file must have a 'write' attribute");
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001161 return -1;
1162 }
1163
1164 return 0;
1165}
1166
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001167/* Returns the size of the input on success, -1 on failure. This takes its
1168 own reference to `input`. */
1169static Py_ssize_t
1170_Unpickler_SetStringInput(UnpicklerObject *self, PyObject *input)
1171{
1172 if (self->buffer.buf != NULL)
1173 PyBuffer_Release(&self->buffer);
1174 if (PyObject_GetBuffer(input, &self->buffer, PyBUF_CONTIG_RO) < 0)
1175 return -1;
1176 self->input_buffer = self->buffer.buf;
1177 self->input_len = self->buffer.len;
1178 self->next_read_idx = 0;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001179 self->prefetched_idx = self->input_len;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001180 return self->input_len;
1181}
1182
Antoine Pitrou04248a82010-10-12 20:51:21 +00001183static int
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001184bad_readline(void)
1185{
1186 PickleState *st = _Pickle_GetGlobalState();
1187 PyErr_SetString(st->UnpicklingError, "pickle data was truncated");
1188 return -1;
1189}
1190
1191static int
Antoine Pitrou04248a82010-10-12 20:51:21 +00001192_Unpickler_SkipConsumed(UnpicklerObject *self)
1193{
Victor Stinnerb43ad1d2013-10-31 13:38:42 +01001194 Py_ssize_t consumed;
1195 PyObject *r;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001196
Victor Stinnerb43ad1d2013-10-31 13:38:42 +01001197 consumed = self->next_read_idx - self->prefetched_idx;
1198 if (consumed <= 0)
1199 return 0;
1200
1201 assert(self->peek); /* otherwise we did something wrong */
Martin Panter6245cb32016-04-15 02:14:19 +00001202 /* This makes a useless copy... */
Victor Stinnerb43ad1d2013-10-31 13:38:42 +01001203 r = PyObject_CallFunction(self->read, "n", consumed);
1204 if (r == NULL)
1205 return -1;
1206 Py_DECREF(r);
1207
1208 self->prefetched_idx = self->next_read_idx;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001209 return 0;
1210}
1211
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001212static const Py_ssize_t READ_WHOLE_LINE = -1;
1213
1214/* If reading from a file, we need to only pull the bytes we need, since there
1215 may be multiple pickle objects arranged contiguously in the same input
1216 buffer.
1217
1218 If `n` is READ_WHOLE_LINE, read a whole line. Otherwise, read up to `n`
1219 bytes from the input stream/buffer.
1220
1221 Update the unpickler's input buffer with the newly-read data. Returns -1 on
1222 failure; on success, returns the number of bytes read from the file.
1223
1224 On success, self->input_len will be 0; this is intentional so that when
1225 unpickling from a file, the "we've run out of data" code paths will trigger,
1226 causing the Unpickler to go back to the file for more data. Use the returned
1227 size to tell you how much data you can process. */
1228static Py_ssize_t
1229_Unpickler_ReadFromFile(UnpicklerObject *self, Py_ssize_t n)
1230{
1231 PyObject *data;
Serhiy Storchaka6fe39b72013-11-30 23:15:38 +02001232 Py_ssize_t read_size;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001233
1234 assert(self->read != NULL);
Victor Stinner121aab42011-09-29 23:40:53 +02001235
Antoine Pitrou04248a82010-10-12 20:51:21 +00001236 if (_Unpickler_SkipConsumed(self) < 0)
1237 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001238
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08001239 if (n == READ_WHOLE_LINE) {
Victor Stinner559bb6a2016-08-22 22:48:54 +02001240 data = _PyObject_CallNoArg(self->readline);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08001241 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001242 else {
Serhiy Storchaka6fe39b72013-11-30 23:15:38 +02001243 PyObject *len;
1244 /* Prefetch some data without advancing the file pointer, if possible */
1245 if (self->peek && n < PREFETCH) {
1246 len = PyLong_FromSsize_t(PREFETCH);
1247 if (len == NULL)
1248 return -1;
1249 data = _Pickle_FastCall(self->peek, len);
1250 if (data == NULL) {
1251 if (!PyErr_ExceptionMatches(PyExc_NotImplementedError))
1252 return -1;
1253 /* peek() is probably not supported by the given file object */
1254 PyErr_Clear();
1255 Py_CLEAR(self->peek);
1256 }
1257 else {
1258 read_size = _Unpickler_SetStringInput(self, data);
1259 Py_DECREF(data);
1260 self->prefetched_idx = 0;
1261 if (n <= read_size)
1262 return n;
1263 }
1264 }
1265 len = PyLong_FromSsize_t(n);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001266 if (len == NULL)
1267 return -1;
Alexandre Vassalotti20c28c12013-11-27 02:26:54 -08001268 data = _Pickle_FastCall(self->read, len);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001269 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001270 if (data == NULL)
1271 return -1;
1272
Serhiy Storchaka6fe39b72013-11-30 23:15:38 +02001273 read_size = _Unpickler_SetStringInput(self, data);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001274 Py_DECREF(data);
1275 return read_size;
1276}
1277
Victor Stinner19ed27e2016-05-20 11:42:37 +02001278/* Don't call it directly: use _Unpickler_Read() */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001279static Py_ssize_t
Victor Stinner19ed27e2016-05-20 11:42:37 +02001280_Unpickler_ReadImpl(UnpicklerObject *self, char **s, Py_ssize_t n)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001281{
Antoine Pitrou04248a82010-10-12 20:51:21 +00001282 Py_ssize_t num_read;
1283
Benjamin Peterson6aa15642015-09-27 01:16:03 -07001284 *s = NULL;
Benjamin Petersone48cf7e2015-09-26 00:08:34 -07001285 if (self->next_read_idx > PY_SSIZE_T_MAX - n) {
1286 PickleState *st = _Pickle_GetGlobalState();
1287 PyErr_SetString(st->UnpicklingError,
1288 "read would overflow (invalid bytecode)");
1289 return -1;
1290 }
Victor Stinner19ed27e2016-05-20 11:42:37 +02001291
1292 /* This case is handled by the _Unpickler_Read() macro for efficiency */
1293 assert(self->next_read_idx + n > self->input_len);
1294
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001295 if (!self->read)
1296 return bad_readline();
1297
Antoine Pitrou04248a82010-10-12 20:51:21 +00001298 num_read = _Unpickler_ReadFromFile(self, n);
1299 if (num_read < 0)
1300 return -1;
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001301 if (num_read < n)
1302 return bad_readline();
Antoine Pitrou04248a82010-10-12 20:51:21 +00001303 *s = self->input_buffer;
1304 self->next_read_idx = n;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001305 return n;
1306}
1307
Victor Stinner19ed27e2016-05-20 11:42:37 +02001308/* Read `n` bytes from the unpickler's data source, storing the result in `*s`.
1309
1310 This should be used for all data reads, rather than accessing the unpickler's
1311 input buffer directly. This method deals correctly with reading from input
1312 streams, which the input buffer doesn't deal with.
1313
1314 Note that when reading from a file-like object, self->next_read_idx won't
1315 be updated (it should remain at 0 for the entire unpickling process). You
1316 should use this function's return value to know how many bytes you can
1317 consume.
1318
1319 Returns -1 (with an exception set) on failure. On success, return the
1320 number of chars read. */
1321#define _Unpickler_Read(self, s, n) \
Victor Stinnerda230562016-05-20 21:16:59 +02001322 (((n) <= (self)->input_len - (self)->next_read_idx) \
Victor Stinner19ed27e2016-05-20 11:42:37 +02001323 ? (*(s) = (self)->input_buffer + (self)->next_read_idx, \
1324 (self)->next_read_idx += (n), \
1325 (n)) \
1326 : _Unpickler_ReadImpl(self, (s), (n)))
1327
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001328static Py_ssize_t
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001329_Unpickler_CopyLine(UnpicklerObject *self, char *line, Py_ssize_t len,
1330 char **result)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001331{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001332 char *input_line = PyMem_Realloc(self->input_line, len + 1);
Victor Stinner42024562013-07-12 00:53:57 +02001333 if (input_line == NULL) {
1334 PyErr_NoMemory();
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001335 return -1;
Victor Stinner42024562013-07-12 00:53:57 +02001336 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001337
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001338 memcpy(input_line, line, len);
1339 input_line[len] = '\0';
1340 self->input_line = input_line;
1341 *result = self->input_line;
1342 return len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001343}
1344
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001345/* Read a line from the input stream/buffer. If we run off the end of the input
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001346 before hitting \n, raise an error.
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001347
1348 Returns the number of chars read, or -1 on failure. */
1349static Py_ssize_t
1350_Unpickler_Readline(UnpicklerObject *self, char **result)
1351{
1352 Py_ssize_t i, num_read;
1353
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001354 for (i = self->next_read_idx; i < self->input_len; i++) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001355 if (self->input_buffer[i] == '\n') {
1356 char *line_start = self->input_buffer + self->next_read_idx;
1357 num_read = i - self->next_read_idx + 1;
1358 self->next_read_idx = i + 1;
1359 return _Unpickler_CopyLine(self, line_start, num_read, result);
1360 }
1361 }
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001362 if (!self->read)
1363 return bad_readline();
Victor Stinner121aab42011-09-29 23:40:53 +02001364
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03001365 num_read = _Unpickler_ReadFromFile(self, READ_WHOLE_LINE);
1366 if (num_read < 0)
1367 return -1;
1368 if (num_read == 0 || self->input_buffer[num_read - 1] != '\n')
1369 return bad_readline();
1370 self->next_read_idx = num_read;
1371 return _Unpickler_CopyLine(self, self->input_buffer, num_read, result);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001372}
1373
1374/* Returns -1 (with an exception set) on failure, 0 on success. The memo array
1375 will be modified in place. */
1376static int
1377_Unpickler_ResizeMemoList(UnpicklerObject *self, Py_ssize_t new_size)
1378{
1379 Py_ssize_t i;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001380
1381 assert(new_size > self->memo_size);
1382
Miss Islington (bot)962051e2018-08-16 00:53:00 -04001383 PyObject **memo_new = self->memo;
1384 PyMem_RESIZE(memo_new, PyObject *, new_size);
1385 if (memo_new == NULL) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001386 PyErr_NoMemory();
1387 return -1;
1388 }
Miss Islington (bot)962051e2018-08-16 00:53:00 -04001389 self->memo = memo_new;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001390 for (i = self->memo_size; i < new_size; i++)
1391 self->memo[i] = NULL;
1392 self->memo_size = new_size;
1393 return 0;
1394}
1395
1396/* Returns NULL if idx is out of bounds. */
1397static PyObject *
1398_Unpickler_MemoGet(UnpicklerObject *self, Py_ssize_t idx)
1399{
1400 if (idx < 0 || idx >= self->memo_size)
1401 return NULL;
1402
1403 return self->memo[idx];
1404}
1405
1406/* Returns -1 (with an exception set) on failure, 0 on success.
1407 This takes its own reference to `value`. */
1408static int
1409_Unpickler_MemoPut(UnpicklerObject *self, Py_ssize_t idx, PyObject *value)
1410{
1411 PyObject *old_item;
1412
1413 if (idx >= self->memo_size) {
1414 if (_Unpickler_ResizeMemoList(self, idx * 2) < 0)
1415 return -1;
1416 assert(idx < self->memo_size);
1417 }
1418 Py_INCREF(value);
1419 old_item = self->memo[idx];
1420 self->memo[idx] = value;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001421 if (old_item != NULL) {
1422 Py_DECREF(old_item);
1423 }
1424 else {
1425 self->memo_len++;
1426 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001427 return 0;
1428}
1429
1430static PyObject **
1431_Unpickler_NewMemo(Py_ssize_t new_size)
1432{
Benjamin Peterson59b08c12015-06-27 13:41:33 -05001433 PyObject **memo = PyMem_NEW(PyObject *, new_size);
Victor Stinner42024562013-07-12 00:53:57 +02001434 if (memo == NULL) {
1435 PyErr_NoMemory();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001436 return NULL;
Victor Stinner42024562013-07-12 00:53:57 +02001437 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001438 memset(memo, 0, new_size * sizeof(PyObject *));
1439 return memo;
1440}
1441
1442/* Free the unpickler's memo, taking care to decref any items left in it. */
1443static void
1444_Unpickler_MemoCleanup(UnpicklerObject *self)
1445{
1446 Py_ssize_t i;
1447 PyObject **memo = self->memo;
1448
1449 if (self->memo == NULL)
1450 return;
1451 self->memo = NULL;
1452 i = self->memo_size;
1453 while (--i >= 0) {
1454 Py_XDECREF(memo[i]);
1455 }
1456 PyMem_FREE(memo);
1457}
1458
1459static UnpicklerObject *
1460_Unpickler_New(void)
1461{
1462 UnpicklerObject *self;
1463
1464 self = PyObject_GC_New(UnpicklerObject, &Unpickler_Type);
1465 if (self == NULL)
1466 return NULL;
1467
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001468 self->pers_func = NULL;
1469 self->input_buffer = NULL;
1470 self->input_line = NULL;
1471 self->input_len = 0;
1472 self->next_read_idx = 0;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001473 self->prefetched_idx = 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001474 self->read = NULL;
1475 self->readline = NULL;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001476 self->peek = NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001477 self->encoding = NULL;
1478 self->errors = NULL;
1479 self->marks = NULL;
1480 self->num_marks = 0;
1481 self->marks_size = 0;
1482 self->proto = 0;
1483 self->fix_imports = 0;
Victor Stinner68c8ea22013-07-11 22:56:25 +02001484 memset(&self->buffer, 0, sizeof(Py_buffer));
1485 self->memo_size = 32;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001486 self->memo_len = 0;
Victor Stinner68c8ea22013-07-11 22:56:25 +02001487 self->memo = _Unpickler_NewMemo(self->memo_size);
1488 self->stack = (Pdata *)Pdata_New();
1489
1490 if (self->memo == NULL || self->stack == NULL) {
1491 Py_DECREF(self);
1492 return NULL;
1493 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001494
1495 return self;
1496}
1497
1498/* Returns -1 (with an exception set) on failure, 0 on success. This may
1499 be called once on a freshly created Pickler. */
1500static int
1501_Unpickler_SetInputStream(UnpicklerObject *self, PyObject *file)
1502{
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001503 _Py_IDENTIFIER(peek);
1504 _Py_IDENTIFIER(read);
1505 _Py_IDENTIFIER(readline);
Martin v. Löwis1ee1b6f2011-10-10 18:11:30 +02001506
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001507 if (_PyObject_LookupAttrId(file, &PyId_peek, &self->peek) < 0) {
1508 return -1;
Antoine Pitrou04248a82010-10-12 20:51:21 +00001509 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001510 (void)_PyObject_LookupAttrId(file, &PyId_read, &self->read);
1511 (void)_PyObject_LookupAttrId(file, &PyId_readline, &self->readline);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001512 if (self->readline == NULL || self->read == NULL) {
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001513 if (!PyErr_Occurred()) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001514 PyErr_SetString(PyExc_TypeError,
1515 "file must have 'read' and 'readline' attributes");
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001516 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001517 Py_CLEAR(self->read);
1518 Py_CLEAR(self->readline);
Antoine Pitrou04248a82010-10-12 20:51:21 +00001519 Py_CLEAR(self->peek);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001520 return -1;
1521 }
1522 return 0;
1523}
1524
1525/* Returns -1 (with an exception set) on failure, 0 on success. This may
1526 be called once on a freshly created Pickler. */
1527static int
1528_Unpickler_SetInputEncoding(UnpicklerObject *self,
1529 const char *encoding,
1530 const char *errors)
1531{
1532 if (encoding == NULL)
1533 encoding = "ASCII";
1534 if (errors == NULL)
1535 errors = "strict";
1536
Victor Stinner49fc8ec2013-07-07 23:30:24 +02001537 self->encoding = _PyMem_Strdup(encoding);
1538 self->errors = _PyMem_Strdup(errors);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001539 if (self->encoding == NULL || self->errors == NULL) {
1540 PyErr_NoMemory();
1541 return -1;
1542 }
1543 return 0;
1544}
1545
1546/* Generate a GET opcode for an object stored in the memo. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001547static int
1548memo_get(PicklerObject *self, PyObject *key)
1549{
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001550 Py_ssize_t *value;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001551 char pdata[30];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001552 Py_ssize_t len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001553
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001554 value = PyMemoTable_Get(self->memo, key);
1555 if (value == NULL) {
1556 PyErr_SetObject(PyExc_KeyError, key);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001557 return -1;
1558 }
1559
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001560 if (!self->bin) {
1561 pdata[0] = GET;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001562 PyOS_snprintf(pdata + 1, sizeof(pdata) - 1,
1563 "%" PY_FORMAT_SIZE_T "d\n", *value);
1564 len = strlen(pdata);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001565 }
1566 else {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001567 if (*value < 256) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001568 pdata[0] = BINGET;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001569 pdata[1] = (unsigned char)(*value & 0xff);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001570 len = 2;
1571 }
Serhiy Storchaka67c719b2014-09-05 10:10:23 +03001572 else if ((size_t)*value <= 0xffffffffUL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001573 pdata[0] = LONG_BINGET;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001574 pdata[1] = (unsigned char)(*value & 0xff);
1575 pdata[2] = (unsigned char)((*value >> 8) & 0xff);
1576 pdata[3] = (unsigned char)((*value >> 16) & 0xff);
1577 pdata[4] = (unsigned char)((*value >> 24) & 0xff);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001578 len = 5;
1579 }
1580 else { /* unlikely */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08001581 PickleState *st = _Pickle_GetGlobalState();
1582 PyErr_SetString(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001583 "memo id too large for LONG_BINGET");
1584 return -1;
1585 }
1586 }
1587
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001588 if (_Pickler_Write(self, pdata, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001589 return -1;
1590
1591 return 0;
1592}
1593
1594/* Store an object in the memo, assign it a new unique ID based on the number
1595 of objects currently stored in the memo and generate a PUT opcode. */
1596static int
1597memo_put(PicklerObject *self, PyObject *obj)
1598{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001599 char pdata[30];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001600 Py_ssize_t len;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001601 Py_ssize_t idx;
1602
1603 const char memoize_op = MEMOIZE;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001604
1605 if (self->fast)
1606 return 0;
1607
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001608 idx = PyMemoTable_Size(self->memo);
1609 if (PyMemoTable_Set(self->memo, obj, idx) < 0)
1610 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001611
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001612 if (self->proto >= 4) {
1613 if (_Pickler_Write(self, &memoize_op, 1) < 0)
1614 return -1;
1615 return 0;
1616 }
1617 else if (!self->bin) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001618 pdata[0] = PUT;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02001619 PyOS_snprintf(pdata + 1, sizeof(pdata) - 1,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001620 "%" PY_FORMAT_SIZE_T "d\n", idx);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001621 len = strlen(pdata);
1622 }
1623 else {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001624 if (idx < 256) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001625 pdata[0] = BINPUT;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001626 pdata[1] = (unsigned char)idx;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001627 len = 2;
1628 }
Serhiy Storchaka67c719b2014-09-05 10:10:23 +03001629 else if ((size_t)idx <= 0xffffffffUL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001630 pdata[0] = LONG_BINPUT;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001631 pdata[1] = (unsigned char)(idx & 0xff);
1632 pdata[2] = (unsigned char)((idx >> 8) & 0xff);
1633 pdata[3] = (unsigned char)((idx >> 16) & 0xff);
1634 pdata[4] = (unsigned char)((idx >> 24) & 0xff);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001635 len = 5;
1636 }
1637 else { /* unlikely */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08001638 PickleState *st = _Pickle_GetGlobalState();
1639 PyErr_SetString(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001640 "memo id too large for LONG_BINPUT");
1641 return -1;
1642 }
1643 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001644 if (_Pickler_Write(self, pdata, len) < 0)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001645 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001646
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001647 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001648}
1649
1650static PyObject *
Serhiy Storchaka9937d902017-01-09 10:04:34 +02001651get_dotted_path(PyObject *obj, PyObject *name)
1652{
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001653 _Py_static_string(PyId_dot, ".");
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001654 PyObject *dotted_path;
1655 Py_ssize_t i, n;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001656
1657 dotted_path = PyUnicode_Split(name, _PyUnicode_FromId(&PyId_dot), -1);
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001658 if (dotted_path == NULL)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001659 return NULL;
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001660 n = PyList_GET_SIZE(dotted_path);
1661 assert(n >= 1);
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001662 for (i = 0; i < n; i++) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001663 PyObject *subpath = PyList_GET_ITEM(dotted_path, i);
Serhiy Storchaka9937d902017-01-09 10:04:34 +02001664 if (_PyUnicode_EqualToASCIIString(subpath, "<locals>")) {
Antoine Pitrou6cd5eda2014-12-02 00:20:03 +01001665 if (obj == NULL)
1666 PyErr_Format(PyExc_AttributeError,
1667 "Can't pickle local object %R", name);
1668 else
1669 PyErr_Format(PyExc_AttributeError,
1670 "Can't pickle local attribute %R on %R", name, obj);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001671 Py_DECREF(dotted_path);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001672 return NULL;
1673 }
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001674 }
1675 return dotted_path;
1676}
1677
1678static PyObject *
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001679get_deep_attribute(PyObject *obj, PyObject *names, PyObject **pparent)
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001680{
1681 Py_ssize_t i, n;
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001682 PyObject *parent = NULL;
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001683
1684 assert(PyList_CheckExact(names));
1685 Py_INCREF(obj);
1686 n = PyList_GET_SIZE(names);
1687 for (i = 0; i < n; i++) {
1688 PyObject *name = PyList_GET_ITEM(names, i);
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001689 Py_XDECREF(parent);
1690 parent = obj;
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001691 (void)_PyObject_LookupAttr(parent, name, &obj);
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001692 if (obj == NULL) {
1693 Py_DECREF(parent);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001694 return NULL;
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001695 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001696 }
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001697 if (pparent != NULL)
1698 *pparent = parent;
1699 else
1700 Py_XDECREF(parent);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001701 return obj;
1702}
1703
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001704
1705static PyObject *
1706getattribute(PyObject *obj, PyObject *name, int allow_qualname)
1707{
1708 PyObject *dotted_path, *attr;
1709
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001710 if (allow_qualname) {
1711 dotted_path = get_dotted_path(obj, name);
1712 if (dotted_path == NULL)
1713 return NULL;
1714 attr = get_deep_attribute(obj, dotted_path, NULL);
1715 Py_DECREF(dotted_path);
1716 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001717 else {
1718 (void)_PyObject_LookupAttr(obj, name, &attr);
1719 }
1720 if (attr == NULL && !PyErr_Occurred()) {
1721 PyErr_Format(PyExc_AttributeError,
1722 "Can't get attribute %R on %R", name, obj);
1723 }
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001724 return attr;
1725}
1726
Eric Snow3f9eee62017-09-15 16:35:20 -06001727static int
1728_checkmodule(PyObject *module_name, PyObject *module,
1729 PyObject *global, PyObject *dotted_path)
1730{
1731 if (module == Py_None) {
1732 return -1;
1733 }
1734 if (PyUnicode_Check(module_name) &&
1735 _PyUnicode_EqualToASCIIString(module_name, "__main__")) {
1736 return -1;
1737 }
1738
1739 PyObject *candidate = get_deep_attribute(module, dotted_path, NULL);
1740 if (candidate == NULL) {
Eric Snow3f9eee62017-09-15 16:35:20 -06001741 return -1;
1742 }
1743 if (candidate != global) {
1744 Py_DECREF(candidate);
1745 return -1;
1746 }
1747 Py_DECREF(candidate);
1748 return 0;
1749}
1750
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001751static PyObject *
Serhiy Storchaka58e41342015-03-31 14:07:24 +03001752whichmodule(PyObject *global, PyObject *dotted_path)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001753{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001754 PyObject *module_name;
Eric Snow3f9eee62017-09-15 16:35:20 -06001755 PyObject *module = NULL;
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001756 Py_ssize_t i;
Eric Snow3f9eee62017-09-15 16:35:20 -06001757 PyObject *modules;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001758 _Py_IDENTIFIER(__module__);
1759 _Py_IDENTIFIER(modules);
1760 _Py_IDENTIFIER(__main__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001761
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001762 if (_PyObject_LookupAttrId(global, &PyId___module__, &module_name) < 0) {
1763 return NULL;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001764 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02001765 if (module_name) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001766 /* In some rare cases (e.g., bound methods of extension types),
1767 __module__ can be None. If it is so, then search sys.modules for
1768 the module of global. */
1769 if (module_name != Py_None)
1770 return module_name;
1771 Py_CLEAR(module_name);
1772 }
1773 assert(module_name == NULL);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001774
Antoine Pitroufce60ea2014-10-23 22:47:50 +02001775 /* Fallback on walking sys.modules */
Eric Snow3f9eee62017-09-15 16:35:20 -06001776 modules = _PySys_GetObjectId(&PyId_modules);
1777 if (modules == NULL) {
Victor Stinner1e53bba2013-07-16 22:26:05 +02001778 PyErr_SetString(PyExc_RuntimeError, "unable to get sys.modules");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001779 return NULL;
Victor Stinner1e53bba2013-07-16 22:26:05 +02001780 }
Eric Snow3f9eee62017-09-15 16:35:20 -06001781 if (PyDict_CheckExact(modules)) {
1782 i = 0;
1783 while (PyDict_Next(modules, &i, &module_name, &module)) {
1784 if (_checkmodule(module_name, module, global, dotted_path) == 0) {
1785 Py_INCREF(module_name);
1786 return module_name;
1787 }
1788 if (PyErr_Occurred()) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001789 return NULL;
Eric Snow3f9eee62017-09-15 16:35:20 -06001790 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001791 }
Eric Snow3f9eee62017-09-15 16:35:20 -06001792 }
1793 else {
1794 PyObject *iterator = PyObject_GetIter(modules);
1795 if (iterator == NULL) {
1796 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001797 }
Eric Snow3f9eee62017-09-15 16:35:20 -06001798 while ((module_name = PyIter_Next(iterator))) {
1799 module = PyObject_GetItem(modules, module_name);
1800 if (module == NULL) {
1801 Py_DECREF(module_name);
1802 Py_DECREF(iterator);
1803 return NULL;
1804 }
1805 if (_checkmodule(module_name, module, global, dotted_path) == 0) {
1806 Py_DECREF(module);
1807 Py_DECREF(iterator);
1808 return module_name;
1809 }
1810 Py_DECREF(module);
1811 Py_DECREF(module_name);
1812 if (PyErr_Occurred()) {
1813 Py_DECREF(iterator);
1814 return NULL;
1815 }
1816 }
1817 Py_DECREF(iterator);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001818 }
1819
1820 /* If no module is found, use __main__. */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01001821 module_name = _PyUnicode_FromId(&PyId___main__);
Victor Stinneraf46eb82017-09-05 23:30:16 +02001822 Py_XINCREF(module_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001823 return module_name;
1824}
1825
1826/* fast_save_enter() and fast_save_leave() are guards against recursive
1827 objects when Pickler is used with the "fast mode" (i.e., with object
1828 memoization disabled). If the nesting of a list or dict object exceed
1829 FAST_NESTING_LIMIT, these guards will start keeping an internal
1830 reference to the seen list or dict objects and check whether these objects
1831 are recursive. These are not strictly necessary, since save() has a
1832 hard-coded recursion limit, but they give a nicer error message than the
1833 typical RuntimeError. */
1834static int
1835fast_save_enter(PicklerObject *self, PyObject *obj)
1836{
1837 /* if fast_nesting < 0, we're doing an error exit. */
1838 if (++self->fast_nesting >= FAST_NESTING_LIMIT) {
1839 PyObject *key = NULL;
1840 if (self->fast_memo == NULL) {
1841 self->fast_memo = PyDict_New();
1842 if (self->fast_memo == NULL) {
1843 self->fast_nesting = -1;
1844 return 0;
1845 }
1846 }
1847 key = PyLong_FromVoidPtr(obj);
Mat Mf76231f2017-11-13 02:50:16 -05001848 if (key == NULL) {
1849 self->fast_nesting = -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001850 return 0;
Mat Mf76231f2017-11-13 02:50:16 -05001851 }
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08001852 if (PyDict_GetItemWithError(self->fast_memo, key)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001853 Py_DECREF(key);
1854 PyErr_Format(PyExc_ValueError,
1855 "fast mode: can't pickle cyclic objects "
1856 "including object type %.200s at %p",
1857 obj->ob_type->tp_name, obj);
1858 self->fast_nesting = -1;
1859 return 0;
1860 }
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08001861 if (PyErr_Occurred()) {
Mat Mf76231f2017-11-13 02:50:16 -05001862 Py_DECREF(key);
1863 self->fast_nesting = -1;
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08001864 return 0;
1865 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001866 if (PyDict_SetItem(self->fast_memo, key, Py_None) < 0) {
1867 Py_DECREF(key);
1868 self->fast_nesting = -1;
1869 return 0;
1870 }
1871 Py_DECREF(key);
1872 }
1873 return 1;
1874}
1875
1876static int
1877fast_save_leave(PicklerObject *self, PyObject *obj)
1878{
1879 if (self->fast_nesting-- >= FAST_NESTING_LIMIT) {
1880 PyObject *key = PyLong_FromVoidPtr(obj);
1881 if (key == NULL)
1882 return 0;
1883 if (PyDict_DelItem(self->fast_memo, key) < 0) {
1884 Py_DECREF(key);
1885 return 0;
1886 }
1887 Py_DECREF(key);
1888 }
1889 return 1;
1890}
1891
1892static int
1893save_none(PicklerObject *self, PyObject *obj)
1894{
1895 const char none_op = NONE;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001896 if (_Pickler_Write(self, &none_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001897 return -1;
1898
1899 return 0;
1900}
1901
1902static int
1903save_bool(PicklerObject *self, PyObject *obj)
1904{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001905 if (self->proto >= 2) {
Alexandre Vassalotti8a67f522013-11-24 21:40:18 -08001906 const char bool_op = (obj == Py_True) ? NEWTRUE : NEWFALSE;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001907 if (_Pickler_Write(self, &bool_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001908 return -1;
1909 }
Alexandre Vassalotti8a67f522013-11-24 21:40:18 -08001910 else {
1911 /* These aren't opcodes -- they're ways to pickle bools before protocol 2
1912 * so that unpicklers written before bools were introduced unpickle them
1913 * as ints, but unpicklers after can recognize that bools were intended.
1914 * Note that protocol 2 added direct ways to pickle bools.
1915 */
1916 const char *bool_str = (obj == Py_True) ? "I01\n" : "I00\n";
1917 if (_Pickler_Write(self, bool_str, strlen(bool_str)) < 0)
1918 return -1;
1919 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001920 return 0;
1921}
1922
1923static int
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001924save_long(PicklerObject *self, PyObject *obj)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001925{
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001926 PyObject *repr = NULL;
1927 Py_ssize_t size;
1928 long val;
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001929 int overflow;
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001930 int status = 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001931
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001932 val= PyLong_AsLongAndOverflow(obj, &overflow);
1933 if (!overflow && (sizeof(long) <= 4 ||
1934 (val <= 0x7fffffffL && val >= (-0x7fffffffL - 1))))
1935 {
Larry Hastings61272b72014-01-07 12:41:53 -08001936 /* result fits in a signed 4-byte integer.
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -08001937
1938 Note: we can't use -0x80000000L in the above condition because some
1939 compilers (e.g., MSVC) will promote 0x80000000L to an unsigned type
1940 before applying the unary minus when sizeof(long) <= 4. The
1941 resulting value stays unsigned which is commonly not what we want,
1942 so MSVC happily warns us about it. However, that result would have
1943 been fine because we guard for sizeof(long) <= 4 which turns the
1944 condition true in that particular case. */
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001945 char pdata[32];
1946 Py_ssize_t len = 0;
1947
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001948 if (self->bin) {
1949 pdata[1] = (unsigned char)(val & 0xff);
1950 pdata[2] = (unsigned char)((val >> 8) & 0xff);
1951 pdata[3] = (unsigned char)((val >> 16) & 0xff);
1952 pdata[4] = (unsigned char)((val >> 24) & 0xff);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001953
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001954 if ((pdata[4] != 0) || (pdata[3] != 0)) {
1955 pdata[0] = BININT;
1956 len = 5;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001957 }
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001958 else if (pdata[2] != 0) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001959 pdata[0] = BININT2;
1960 len = 3;
1961 }
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001962 else {
1963 pdata[0] = BININT1;
1964 len = 2;
1965 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001966 }
1967 else {
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001968 sprintf(pdata, "%c%ld\n", INT, val);
1969 len = strlen(pdata);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001970 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001971 if (_Pickler_Write(self, pdata, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001972 return -1;
Alexandre Vassalottided929b2013-11-24 22:41:13 -08001973
1974 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001975 }
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02001976 assert(!PyErr_Occurred());
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001977
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001978 if (self->proto >= 2) {
1979 /* Linear-time pickling. */
1980 size_t nbits;
1981 size_t nbytes;
1982 unsigned char *pdata;
1983 char header[5];
1984 int i;
1985 int sign = _PyLong_Sign(obj);
1986
1987 if (sign == 0) {
1988 header[0] = LONG1;
1989 header[1] = 0; /* It's 0 -- an empty bytestring. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00001990 if (_Pickler_Write(self, header, 2) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00001991 goto error;
1992 return 0;
1993 }
1994 nbits = _PyLong_NumBits(obj);
1995 if (nbits == (size_t)-1 && PyErr_Occurred())
1996 goto error;
1997 /* How many bytes do we need? There are nbits >> 3 full
1998 * bytes of data, and nbits & 7 leftover bits. If there
1999 * are any leftover bits, then we clearly need another
2000 * byte. Wnat's not so obvious is that we *probably*
2001 * need another byte even if there aren't any leftovers:
2002 * the most-significant bit of the most-significant byte
2003 * acts like a sign bit, and it's usually got a sense
Serhiy Storchaka95949422013-08-27 19:40:23 +03002004 * opposite of the one we need. The exception is ints
2005 * of the form -(2**(8*j-1)) for j > 0. Such an int is
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002006 * its own 256's-complement, so has the right sign bit
2007 * even without the extra byte. That's a pain to check
2008 * for in advance, though, so we always grab an extra
2009 * byte at the start, and cut it back later if possible.
2010 */
2011 nbytes = (nbits >> 3) + 1;
Antoine Pitroubf6ecf92012-11-24 20:40:21 +01002012 if (nbytes > 0x7fffffffL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002013 PyErr_SetString(PyExc_OverflowError,
Serhiy Storchaka95949422013-08-27 19:40:23 +03002014 "int too large to pickle");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002015 goto error;
2016 }
Neal Norwitz6ae2eb22008-08-24 23:50:08 +00002017 repr = PyBytes_FromStringAndSize(NULL, (Py_ssize_t)nbytes);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002018 if (repr == NULL)
2019 goto error;
Neal Norwitz6ae2eb22008-08-24 23:50:08 +00002020 pdata = (unsigned char *)PyBytes_AS_STRING(repr);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002021 i = _PyLong_AsByteArray((PyLongObject *)obj,
2022 pdata, nbytes,
2023 1 /* little endian */ , 1 /* signed */ );
2024 if (i < 0)
2025 goto error;
Serhiy Storchaka95949422013-08-27 19:40:23 +03002026 /* If the int is negative, this may be a byte more than
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002027 * needed. This is so iff the MSB is all redundant sign
2028 * bits.
2029 */
2030 if (sign < 0 &&
Victor Stinner121aab42011-09-29 23:40:53 +02002031 nbytes > 1 &&
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002032 pdata[nbytes - 1] == 0xff &&
2033 (pdata[nbytes - 2] & 0x80) != 0) {
2034 nbytes--;
2035 }
2036
2037 if (nbytes < 256) {
2038 header[0] = LONG1;
2039 header[1] = (unsigned char)nbytes;
2040 size = 2;
2041 }
2042 else {
2043 header[0] = LONG4;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002044 size = (Py_ssize_t) nbytes;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002045 for (i = 1; i < 5; i++) {
2046 header[i] = (unsigned char)(size & 0xff);
2047 size >>= 8;
2048 }
2049 size = 5;
2050 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002051 if (_Pickler_Write(self, header, size) < 0 ||
2052 _Pickler_Write(self, (char *)pdata, (int)nbytes) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002053 goto error;
2054 }
2055 else {
Serhiy Storchaka3daaafb2017-11-16 09:44:43 +02002056 const char long_op = LONG;
Serhiy Storchaka85b0f5b2016-11-20 10:16:47 +02002057 const char *string;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002058
Mark Dickinson8dd05142009-01-20 20:43:58 +00002059 /* proto < 2: write the repr and newline. This is quadratic-time (in
2060 the number of digits), in both directions. We add a trailing 'L'
2061 to the repr, for compatibility with Python 2.x. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002062
2063 repr = PyObject_Repr(obj);
2064 if (repr == NULL)
2065 goto error;
2066
Serhiy Storchaka06515832016-11-20 09:13:07 +02002067 string = PyUnicode_AsUTF8AndSize(repr, &size);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002068 if (string == NULL)
2069 goto error;
2070
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002071 if (_Pickler_Write(self, &long_op, 1) < 0 ||
2072 _Pickler_Write(self, string, size) < 0 ||
2073 _Pickler_Write(self, "L\n", 2) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002074 goto error;
2075 }
2076
2077 if (0) {
2078 error:
2079 status = -1;
2080 }
2081 Py_XDECREF(repr);
2082
2083 return status;
2084}
2085
2086static int
2087save_float(PicklerObject *self, PyObject *obj)
2088{
2089 double x = PyFloat_AS_DOUBLE((PyFloatObject *)obj);
2090
2091 if (self->bin) {
2092 char pdata[9];
2093 pdata[0] = BINFLOAT;
2094 if (_PyFloat_Pack8(x, (unsigned char *)&pdata[1], 0) < 0)
2095 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002096 if (_Pickler_Write(self, pdata, 9) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002097 return -1;
Victor Stinner121aab42011-09-29 23:40:53 +02002098 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002099 else {
Eric Smith0923d1d2009-04-16 20:16:10 +00002100 int result = -1;
2101 char *buf = NULL;
2102 char op = FLOAT;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002103
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002104 if (_Pickler_Write(self, &op, 1) < 0)
Eric Smith0923d1d2009-04-16 20:16:10 +00002105 goto done;
2106
Serhiy Storchakac86ca262015-02-15 14:18:32 +02002107 buf = PyOS_double_to_string(x, 'r', 0, Py_DTSF_ADD_DOT_0, NULL);
Eric Smith0923d1d2009-04-16 20:16:10 +00002108 if (!buf) {
2109 PyErr_NoMemory();
2110 goto done;
2111 }
2112
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002113 if (_Pickler_Write(self, buf, strlen(buf)) < 0)
Eric Smith0923d1d2009-04-16 20:16:10 +00002114 goto done;
2115
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002116 if (_Pickler_Write(self, "\n", 1) < 0)
Eric Smith0923d1d2009-04-16 20:16:10 +00002117 goto done;
2118
2119 result = 0;
2120done:
2121 PyMem_Free(buf);
2122 return result;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002123 }
2124
2125 return 0;
2126}
2127
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002128/* Perform direct write of the header and payload of the binary object.
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002129
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002130 The large contiguous data is written directly into the underlying file
2131 object, bypassing the output_buffer of the Pickler. We intentionally
2132 do not insert a protocol 4 frame opcode to make it possible to optimize
2133 file.read calls in the loader.
2134 */
2135static int
2136_Pickler_write_bytes(PicklerObject *self,
2137 const char *header, Py_ssize_t header_size,
2138 const char *data, Py_ssize_t data_size,
2139 PyObject *payload)
2140{
2141 int bypass_buffer = (data_size >= FRAME_SIZE_TARGET);
2142 int framing = self->framing;
2143
2144 if (bypass_buffer) {
2145 assert(self->output_buffer != NULL);
2146 /* Commit the previous frame. */
2147 if (_Pickler_CommitFrame(self)) {
2148 return -1;
2149 }
2150 /* Disable framing temporarily */
2151 self->framing = 0;
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002152 }
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002153
2154 if (_Pickler_Write(self, header, header_size) < 0) {
2155 return -1;
2156 }
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002157
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002158 if (bypass_buffer && self->write != NULL) {
2159 /* Bypass the in-memory buffer to directly stream large data
2160 into the underlying file object. */
2161 PyObject *result, *mem = NULL;
2162 /* Dump the output buffer to the file. */
2163 if (_Pickler_FlushToFile(self) < 0) {
2164 return -1;
2165 }
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002166
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002167 /* Stream write the payload into the file without going through the
2168 output buffer. */
2169 if (payload == NULL) {
Serhiy Storchaka5b76bdb2018-01-13 00:28:31 +02002170 /* TODO: It would be better to use a memoryview with a linked
2171 original string if this is possible. */
2172 payload = mem = PyBytes_FromStringAndSize(data, data_size);
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002173 if (payload == NULL) {
2174 return -1;
2175 }
2176 }
2177 result = PyObject_CallFunctionObjArgs(self->write, payload, NULL);
2178 Py_XDECREF(mem);
2179 if (result == NULL) {
2180 return -1;
2181 }
2182 Py_DECREF(result);
2183
2184 /* Reinitialize the buffer for subsequent calls to _Pickler_Write. */
2185 if (_Pickler_ClearBuffer(self) < 0) {
2186 return -1;
2187 }
2188 }
2189 else {
2190 if (_Pickler_Write(self, data, data_size) < 0) {
2191 return -1;
2192 }
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002193 }
2194
2195 /* Re-enable framing for subsequent calls to _Pickler_Write. */
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002196 self->framing = framing;
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002197
2198 return 0;
2199}
2200
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002201static int
2202save_bytes(PicklerObject *self, PyObject *obj)
2203{
2204 if (self->proto < 3) {
2205 /* Older pickle protocols do not have an opcode for pickling bytes
2206 objects. Therefore, we need to fake the copy protocol (i.e.,
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002207 the __reduce__ method) to permit bytes object unpickling.
2208
2209 Here we use a hack to be compatible with Python 2. Since in Python
2210 2 'bytes' is just an alias for 'str' (which has different
2211 parameters than the actual bytes object), we use codecs.encode
2212 to create the appropriate 'str' object when unpickled using
2213 Python 2 *and* the appropriate 'bytes' object when unpickled
2214 using Python 3. Again this is a hack and we don't need to do this
2215 with newer protocols. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002216 PyObject *reduce_value = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002217 int status;
2218
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002219 if (PyBytes_GET_SIZE(obj) == 0) {
2220 reduce_value = Py_BuildValue("(O())", (PyObject*)&PyBytes_Type);
2221 }
2222 else {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08002223 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002224 PyObject *unicode_str =
2225 PyUnicode_DecodeLatin1(PyBytes_AS_STRING(obj),
2226 PyBytes_GET_SIZE(obj),
2227 "strict");
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002228 _Py_IDENTIFIER(latin1);
2229
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002230 if (unicode_str == NULL)
2231 return -1;
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002232 reduce_value = Py_BuildValue("(O(OO))",
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08002233 st->codecs_encode, unicode_str,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002234 _PyUnicode_FromId(&PyId_latin1));
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002235 Py_DECREF(unicode_str);
2236 }
2237
2238 if (reduce_value == NULL)
2239 return -1;
2240
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002241 /* save_reduce() will memoize the object automatically. */
2242 status = save_reduce(self, reduce_value, obj);
2243 Py_DECREF(reduce_value);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002244 return status;
2245 }
2246 else {
2247 Py_ssize_t size;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002248 char header[9];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002249 Py_ssize_t len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002250
Alexandre Vassalotti3bfc65a2011-12-13 13:08:09 -05002251 size = PyBytes_GET_SIZE(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002252 if (size < 0)
2253 return -1;
2254
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002255 if (size <= 0xff) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002256 header[0] = SHORT_BINBYTES;
2257 header[1] = (unsigned char)size;
2258 len = 2;
2259 }
Serhiy Storchaka67c719b2014-09-05 10:10:23 +03002260 else if ((size_t)size <= 0xffffffffUL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002261 header[0] = BINBYTES;
2262 header[1] = (unsigned char)(size & 0xff);
2263 header[2] = (unsigned char)((size >> 8) & 0xff);
2264 header[3] = (unsigned char)((size >> 16) & 0xff);
2265 header[4] = (unsigned char)((size >> 24) & 0xff);
2266 len = 5;
2267 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002268 else if (self->proto >= 4) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002269 header[0] = BINBYTES8;
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -08002270 _write_size64(header + 1, size);
Alexandre Vassalotti6e73ff12013-12-05 19:29:32 -08002271 len = 9;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002272 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002273 else {
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002274 PyErr_SetString(PyExc_OverflowError,
Serhiy Storchakaf8def282013-02-16 17:29:56 +02002275 "cannot serialize a bytes object larger than 4 GiB");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002276 return -1; /* string too large */
2277 }
2278
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002279 if (_Pickler_write_bytes(self, header, len,
2280 PyBytes_AS_STRING(obj), size, obj) < 0)
2281 {
2282 return -1;
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002283 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002284
2285 if (memo_put(self, obj) < 0)
2286 return -1;
2287
2288 return 0;
2289 }
2290}
2291
2292/* A copy of PyUnicode_EncodeRawUnicodeEscape() that also translates
2293 backslash and newline characters to \uXXXX escapes. */
2294static PyObject *
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002295raw_unicode_escape(PyObject *obj)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002296{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002297 char *p;
Victor Stinner049e5092014-08-17 22:20:00 +02002298 Py_ssize_t i, size;
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002299 void *data;
2300 unsigned int kind;
Victor Stinner358af132015-10-12 22:36:57 +02002301 _PyBytesWriter writer;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002302
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002303 if (PyUnicode_READY(obj))
2304 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002305
Victor Stinner358af132015-10-12 22:36:57 +02002306 _PyBytesWriter_Init(&writer);
2307
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002308 size = PyUnicode_GET_LENGTH(obj);
2309 data = PyUnicode_DATA(obj);
2310 kind = PyUnicode_KIND(obj);
Victor Stinner121aab42011-09-29 23:40:53 +02002311
Victor Stinner358af132015-10-12 22:36:57 +02002312 p = _PyBytesWriter_Alloc(&writer, size);
2313 if (p == NULL)
2314 goto error;
2315 writer.overallocate = 1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002316
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002317 for (i=0; i < size; i++) {
2318 Py_UCS4 ch = PyUnicode_READ(kind, data, i);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002319 /* Map 32-bit characters to '\Uxxxxxxxx' */
2320 if (ch >= 0x10000) {
Raymond Hettinger15f44ab2016-08-30 10:47:49 -07002321 /* -1: subtract 1 preallocated byte */
Victor Stinner358af132015-10-12 22:36:57 +02002322 p = _PyBytesWriter_Prepare(&writer, p, 10-1);
2323 if (p == NULL)
2324 goto error;
2325
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002326 *p++ = '\\';
2327 *p++ = 'U';
Victor Stinnerf5cff562011-10-14 02:13:11 +02002328 *p++ = Py_hexdigits[(ch >> 28) & 0xf];
2329 *p++ = Py_hexdigits[(ch >> 24) & 0xf];
2330 *p++ = Py_hexdigits[(ch >> 20) & 0xf];
2331 *p++ = Py_hexdigits[(ch >> 16) & 0xf];
2332 *p++ = Py_hexdigits[(ch >> 12) & 0xf];
2333 *p++ = Py_hexdigits[(ch >> 8) & 0xf];
2334 *p++ = Py_hexdigits[(ch >> 4) & 0xf];
2335 *p++ = Py_hexdigits[ch & 15];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002336 }
Victor Stinner358af132015-10-12 22:36:57 +02002337 /* Map 16-bit characters, '\\' and '\n' to '\uxxxx' */
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002338 else if (ch >= 256 || ch == '\\' || ch == '\n') {
Raymond Hettinger15f44ab2016-08-30 10:47:49 -07002339 /* -1: subtract 1 preallocated byte */
Victor Stinner358af132015-10-12 22:36:57 +02002340 p = _PyBytesWriter_Prepare(&writer, p, 6-1);
2341 if (p == NULL)
2342 goto error;
2343
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002344 *p++ = '\\';
2345 *p++ = 'u';
Victor Stinnerf5cff562011-10-14 02:13:11 +02002346 *p++ = Py_hexdigits[(ch >> 12) & 0xf];
2347 *p++ = Py_hexdigits[(ch >> 8) & 0xf];
2348 *p++ = Py_hexdigits[(ch >> 4) & 0xf];
2349 *p++ = Py_hexdigits[ch & 15];
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002350 }
Alexandre Vassalotti554d8782008-12-27 07:32:41 +00002351 /* Copy everything else as-is */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002352 else
2353 *p++ = (char) ch;
2354 }
Victor Stinner358af132015-10-12 22:36:57 +02002355
2356 return _PyBytesWriter_Finish(&writer, p);
2357
2358error:
2359 _PyBytesWriter_Dealloc(&writer);
2360 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002361}
2362
2363static int
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002364write_unicode_binary(PicklerObject *self, PyObject *obj)
Antoine Pitrou299978d2013-04-07 17:38:11 +02002365{
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002366 char header[9];
2367 Py_ssize_t len;
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002368 PyObject *encoded = NULL;
2369 Py_ssize_t size;
2370 const char *data;
2371
2372 if (PyUnicode_READY(obj))
2373 return -1;
2374
2375 data = PyUnicode_AsUTF8AndSize(obj, &size);
2376 if (data == NULL) {
2377 /* Issue #8383: for strings with lone surrogates, fallback on the
2378 "surrogatepass" error handler. */
2379 PyErr_Clear();
2380 encoded = PyUnicode_AsEncodedString(obj, "utf-8", "surrogatepass");
2381 if (encoded == NULL)
2382 return -1;
2383
2384 data = PyBytes_AS_STRING(encoded);
2385 size = PyBytes_GET_SIZE(encoded);
2386 }
Antoine Pitrou299978d2013-04-07 17:38:11 +02002387
Victor Stinnerf13c46c2014-08-17 21:05:55 +02002388 assert(size >= 0);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002389 if (size <= 0xff && self->proto >= 4) {
2390 header[0] = SHORT_BINUNICODE;
2391 header[1] = (unsigned char)(size & 0xff);
2392 len = 2;
2393 }
Victor Stinnerf13c46c2014-08-17 21:05:55 +02002394 else if ((size_t)size <= 0xffffffffUL) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002395 header[0] = BINUNICODE;
2396 header[1] = (unsigned char)(size & 0xff);
2397 header[2] = (unsigned char)((size >> 8) & 0xff);
2398 header[3] = (unsigned char)((size >> 16) & 0xff);
2399 header[4] = (unsigned char)((size >> 24) & 0xff);
2400 len = 5;
2401 }
2402 else if (self->proto >= 4) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002403 header[0] = BINUNICODE8;
Alexandre Vassalotti1048fb52013-11-25 11:35:46 -08002404 _write_size64(header + 1, size);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01002405 len = 9;
2406 }
2407 else {
Antoine Pitrou299978d2013-04-07 17:38:11 +02002408 PyErr_SetString(PyExc_OverflowError,
Antoine Pitrou4b7b0f02013-04-07 23:46:52 +02002409 "cannot serialize a string larger than 4GiB");
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002410 Py_XDECREF(encoded);
Antoine Pitrou299978d2013-04-07 17:38:11 +02002411 return -1;
2412 }
Antoine Pitrou299978d2013-04-07 17:38:11 +02002413
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002414 if (_Pickler_write_bytes(self, header, len, data, size, encoded) < 0) {
2415 Py_XDECREF(encoded);
2416 return -1;
Olivier Grisel3cd7c6e2018-01-06 16:18:54 +01002417 }
Serhiy Storchaka0a2da502018-01-11 13:03:20 +02002418 Py_XDECREF(encoded);
Antoine Pitrou299978d2013-04-07 17:38:11 +02002419 return 0;
2420}
2421
2422static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002423save_unicode(PicklerObject *self, PyObject *obj)
2424{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002425 if (self->bin) {
Antoine Pitrou299978d2013-04-07 17:38:11 +02002426 if (write_unicode_binary(self, obj) < 0)
2427 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002428 }
2429 else {
Antoine Pitrou299978d2013-04-07 17:38:11 +02002430 PyObject *encoded;
2431 Py_ssize_t size;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002432 const char unicode_op = UNICODE;
2433
Victor Stinnerc806fdc2011-09-29 23:50:23 +02002434 encoded = raw_unicode_escape(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002435 if (encoded == NULL)
Antoine Pitrou299978d2013-04-07 17:38:11 +02002436 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002437
Antoine Pitrou299978d2013-04-07 17:38:11 +02002438 if (_Pickler_Write(self, &unicode_op, 1) < 0) {
2439 Py_DECREF(encoded);
2440 return -1;
2441 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002442
2443 size = PyBytes_GET_SIZE(encoded);
Antoine Pitrou299978d2013-04-07 17:38:11 +02002444 if (_Pickler_Write(self, PyBytes_AS_STRING(encoded), size) < 0) {
2445 Py_DECREF(encoded);
2446 return -1;
2447 }
2448 Py_DECREF(encoded);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002449
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002450 if (_Pickler_Write(self, "\n", 1) < 0)
Antoine Pitrou299978d2013-04-07 17:38:11 +02002451 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002452 }
2453 if (memo_put(self, obj) < 0)
Antoine Pitrou299978d2013-04-07 17:38:11 +02002454 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002455
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002456 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002457}
2458
2459/* A helper for save_tuple. Push the len elements in tuple t on the stack. */
2460static int
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002461store_tuple_elements(PicklerObject *self, PyObject *t, Py_ssize_t len)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002462{
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002463 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002464
2465 assert(PyTuple_Size(t) == len);
2466
2467 for (i = 0; i < len; i++) {
2468 PyObject *element = PyTuple_GET_ITEM(t, i);
2469
2470 if (element == NULL)
2471 return -1;
2472 if (save(self, element, 0) < 0)
2473 return -1;
2474 }
2475
2476 return 0;
2477}
2478
2479/* Tuples are ubiquitous in the pickle protocols, so many techniques are
2480 * used across protocols to minimize the space needed to pickle them.
2481 * Tuples are also the only builtin immutable type that can be recursive
2482 * (a tuple can be reached from itself), and that requires some subtle
2483 * magic so that it works in all cases. IOW, this is a long routine.
2484 */
2485static int
2486save_tuple(PicklerObject *self, PyObject *obj)
2487{
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002488 Py_ssize_t len, i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002489
2490 const char mark_op = MARK;
2491 const char tuple_op = TUPLE;
2492 const char pop_op = POP;
2493 const char pop_mark_op = POP_MARK;
2494 const char len2opcode[] = {EMPTY_TUPLE, TUPLE1, TUPLE2, TUPLE3};
2495
2496 if ((len = PyTuple_Size(obj)) < 0)
2497 return -1;
2498
2499 if (len == 0) {
2500 char pdata[2];
2501
2502 if (self->proto) {
2503 pdata[0] = EMPTY_TUPLE;
2504 len = 1;
2505 }
2506 else {
2507 pdata[0] = MARK;
2508 pdata[1] = TUPLE;
2509 len = 2;
2510 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002511 if (_Pickler_Write(self, pdata, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002512 return -1;
2513 return 0;
2514 }
2515
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002516 /* The tuple isn't in the memo now. If it shows up there after
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002517 * saving the tuple elements, the tuple must be recursive, in
2518 * which case we'll pop everything we put on the stack, and fetch
2519 * its value from the memo.
2520 */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002521 if (len <= 3 && self->proto >= 2) {
2522 /* Use TUPLE{1,2,3} opcodes. */
2523 if (store_tuple_elements(self, obj, len) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002524 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002525
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002526 if (PyMemoTable_Get(self->memo, obj)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002527 /* pop the len elements */
2528 for (i = 0; i < len; i++)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002529 if (_Pickler_Write(self, &pop_op, 1) < 0)
2530 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002531 /* fetch from memo */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002532 if (memo_get(self, obj) < 0)
2533 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002534
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002535 return 0;
2536 }
2537 else { /* Not recursive. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002538 if (_Pickler_Write(self, len2opcode + len, 1) < 0)
2539 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002540 }
2541 goto memoize;
2542 }
2543
2544 /* proto < 2 and len > 0, or proto >= 2 and len > 3.
2545 * Generate MARK e1 e2 ... TUPLE
2546 */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002547 if (_Pickler_Write(self, &mark_op, 1) < 0)
2548 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002549
2550 if (store_tuple_elements(self, obj, len) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002551 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002552
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002553 if (PyMemoTable_Get(self->memo, obj)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002554 /* pop the stack stuff we pushed */
2555 if (self->bin) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002556 if (_Pickler_Write(self, &pop_mark_op, 1) < 0)
2557 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002558 }
2559 else {
2560 /* Note that we pop one more than len, to remove
2561 * the MARK too.
2562 */
2563 for (i = 0; i <= len; i++)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002564 if (_Pickler_Write(self, &pop_op, 1) < 0)
2565 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002566 }
2567 /* fetch from memo */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002568 if (memo_get(self, obj) < 0)
2569 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002570
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002571 return 0;
2572 }
2573 else { /* Not recursive. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002574 if (_Pickler_Write(self, &tuple_op, 1) < 0)
2575 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002576 }
2577
2578 memoize:
2579 if (memo_put(self, obj) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002580 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002581
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002582 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002583}
2584
2585/* iter is an iterator giving items, and we batch up chunks of
2586 * MARK item item ... item APPENDS
2587 * opcode sequences. Calling code should have arranged to first create an
2588 * empty list, or list-like object, for the APPENDS to operate on.
2589 * Returns 0 on success, <0 on error.
2590 */
2591static int
2592batch_list(PicklerObject *self, PyObject *iter)
2593{
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002594 PyObject *obj = NULL;
2595 PyObject *firstitem = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002596 int i, n;
2597
2598 const char mark_op = MARK;
2599 const char append_op = APPEND;
2600 const char appends_op = APPENDS;
2601
2602 assert(iter != NULL);
2603
2604 /* XXX: I think this function could be made faster by avoiding the
2605 iterator interface and fetching objects directly from list using
2606 PyList_GET_ITEM.
2607 */
2608
2609 if (self->proto == 0) {
2610 /* APPENDS isn't available; do one at a time. */
2611 for (;;) {
2612 obj = PyIter_Next(iter);
2613 if (obj == NULL) {
2614 if (PyErr_Occurred())
2615 return -1;
2616 break;
2617 }
2618 i = save(self, obj, 0);
2619 Py_DECREF(obj);
2620 if (i < 0)
2621 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002622 if (_Pickler_Write(self, &append_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002623 return -1;
2624 }
2625 return 0;
2626 }
2627
2628 /* proto > 0: write in batches of BATCHSIZE. */
2629 do {
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002630 /* Get first item */
2631 firstitem = PyIter_Next(iter);
2632 if (firstitem == NULL) {
2633 if (PyErr_Occurred())
2634 goto error;
2635
2636 /* nothing more to add */
2637 break;
2638 }
2639
2640 /* Try to get a second item */
2641 obj = PyIter_Next(iter);
2642 if (obj == NULL) {
2643 if (PyErr_Occurred())
2644 goto error;
2645
2646 /* Only one item to write */
2647 if (save(self, firstitem, 0) < 0)
2648 goto error;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002649 if (_Pickler_Write(self, &append_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002650 goto error;
2651 Py_CLEAR(firstitem);
2652 break;
2653 }
2654
2655 /* More than one item to write */
2656
2657 /* Pump out MARK, items, APPENDS. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002658 if (_Pickler_Write(self, &mark_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002659 goto error;
2660
2661 if (save(self, firstitem, 0) < 0)
2662 goto error;
2663 Py_CLEAR(firstitem);
2664 n = 1;
2665
2666 /* Fetch and save up to BATCHSIZE items */
2667 while (obj) {
2668 if (save(self, obj, 0) < 0)
2669 goto error;
2670 Py_CLEAR(obj);
2671 n += 1;
2672
2673 if (n == BATCHSIZE)
2674 break;
2675
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002676 obj = PyIter_Next(iter);
2677 if (obj == NULL) {
2678 if (PyErr_Occurred())
2679 goto error;
2680 break;
2681 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002682 }
2683
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002684 if (_Pickler_Write(self, &appends_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002685 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002686
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002687 } while (n == BATCHSIZE);
2688 return 0;
2689
2690 error:
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002691 Py_XDECREF(firstitem);
2692 Py_XDECREF(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002693 return -1;
2694}
2695
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002696/* This is a variant of batch_list() above, specialized for lists (with no
2697 * support for list subclasses). Like batch_list(), we batch up chunks of
2698 * MARK item item ... item APPENDS
2699 * opcode sequences. Calling code should have arranged to first create an
2700 * empty list, or list-like object, for the APPENDS to operate on.
2701 * Returns 0 on success, -1 on error.
2702 *
2703 * This version is considerably faster than batch_list(), if less general.
2704 *
2705 * Note that this only works for protocols > 0.
2706 */
2707static int
2708batch_list_exact(PicklerObject *self, PyObject *obj)
2709{
2710 PyObject *item = NULL;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002711 Py_ssize_t this_batch, total;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002712
2713 const char append_op = APPEND;
2714 const char appends_op = APPENDS;
2715 const char mark_op = MARK;
2716
2717 assert(obj != NULL);
2718 assert(self->proto > 0);
2719 assert(PyList_CheckExact(obj));
2720
2721 if (PyList_GET_SIZE(obj) == 1) {
2722 item = PyList_GET_ITEM(obj, 0);
2723 if (save(self, item, 0) < 0)
2724 return -1;
2725 if (_Pickler_Write(self, &append_op, 1) < 0)
2726 return -1;
2727 return 0;
2728 }
2729
2730 /* Write in batches of BATCHSIZE. */
2731 total = 0;
2732 do {
2733 this_batch = 0;
2734 if (_Pickler_Write(self, &mark_op, 1) < 0)
2735 return -1;
2736 while (total < PyList_GET_SIZE(obj)) {
2737 item = PyList_GET_ITEM(obj, total);
2738 if (save(self, item, 0) < 0)
2739 return -1;
2740 total++;
2741 if (++this_batch == BATCHSIZE)
2742 break;
2743 }
2744 if (_Pickler_Write(self, &appends_op, 1) < 0)
2745 return -1;
2746
2747 } while (total < PyList_GET_SIZE(obj));
2748
2749 return 0;
2750}
2751
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002752static int
2753save_list(PicklerObject *self, PyObject *obj)
2754{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002755 char header[3];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02002756 Py_ssize_t len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002757 int status = 0;
2758
2759 if (self->fast && !fast_save_enter(self, obj))
2760 goto error;
2761
2762 /* Create an empty list. */
2763 if (self->bin) {
2764 header[0] = EMPTY_LIST;
2765 len = 1;
2766 }
2767 else {
2768 header[0] = MARK;
2769 header[1] = LIST;
2770 len = 2;
2771 }
2772
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002773 if (_Pickler_Write(self, header, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002774 goto error;
2775
2776 /* Get list length, and bow out early if empty. */
2777 if ((len = PyList_Size(obj)) < 0)
2778 goto error;
2779
2780 if (memo_put(self, obj) < 0)
2781 goto error;
2782
2783 if (len != 0) {
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002784 /* Materialize the list elements. */
2785 if (PyList_CheckExact(obj) && self->proto > 0) {
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00002786 if (Py_EnterRecursiveCall(" while pickling an object"))
2787 goto error;
2788 status = batch_list_exact(self, obj);
2789 Py_LeaveRecursiveCall();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002790 } else {
2791 PyObject *iter = PyObject_GetIter(obj);
2792 if (iter == NULL)
2793 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002794
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00002795 if (Py_EnterRecursiveCall(" while pickling an object")) {
2796 Py_DECREF(iter);
2797 goto error;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002798 }
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00002799 status = batch_list(self, iter);
2800 Py_LeaveRecursiveCall();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002801 Py_DECREF(iter);
2802 }
2803 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002804 if (0) {
2805 error:
2806 status = -1;
2807 }
2808
2809 if (self->fast && !fast_save_leave(self, obj))
2810 status = -1;
2811
2812 return status;
2813}
2814
2815/* iter is an iterator giving (key, value) pairs, and we batch up chunks of
2816 * MARK key value ... key value SETITEMS
2817 * opcode sequences. Calling code should have arranged to first create an
2818 * empty dict, or dict-like object, for the SETITEMS to operate on.
2819 * Returns 0 on success, <0 on error.
2820 *
2821 * This is very much like batch_list(). The difference between saving
2822 * elements directly, and picking apart two-tuples, is so long-winded at
2823 * the C level, though, that attempts to combine these routines were too
2824 * ugly to bear.
2825 */
2826static int
2827batch_dict(PicklerObject *self, PyObject *iter)
2828{
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002829 PyObject *obj = NULL;
2830 PyObject *firstitem = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002831 int i, n;
2832
2833 const char mark_op = MARK;
2834 const char setitem_op = SETITEM;
2835 const char setitems_op = SETITEMS;
2836
2837 assert(iter != NULL);
2838
2839 if (self->proto == 0) {
2840 /* SETITEMS isn't available; do one at a time. */
2841 for (;;) {
2842 obj = PyIter_Next(iter);
2843 if (obj == NULL) {
2844 if (PyErr_Occurred())
2845 return -1;
2846 break;
2847 }
2848 if (!PyTuple_Check(obj) || PyTuple_Size(obj) != 2) {
2849 PyErr_SetString(PyExc_TypeError, "dict items "
2850 "iterator must return 2-tuples");
2851 return -1;
2852 }
2853 i = save(self, PyTuple_GET_ITEM(obj, 0), 0);
2854 if (i >= 0)
2855 i = save(self, PyTuple_GET_ITEM(obj, 1), 0);
2856 Py_DECREF(obj);
2857 if (i < 0)
2858 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002859 if (_Pickler_Write(self, &setitem_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002860 return -1;
2861 }
2862 return 0;
2863 }
2864
2865 /* proto > 0: write in batches of BATCHSIZE. */
2866 do {
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002867 /* Get first item */
2868 firstitem = PyIter_Next(iter);
2869 if (firstitem == NULL) {
2870 if (PyErr_Occurred())
2871 goto error;
2872
2873 /* nothing more to add */
2874 break;
2875 }
2876 if (!PyTuple_Check(firstitem) || PyTuple_Size(firstitem) != 2) {
2877 PyErr_SetString(PyExc_TypeError, "dict items "
2878 "iterator must return 2-tuples");
2879 goto error;
2880 }
2881
2882 /* Try to get a second item */
2883 obj = PyIter_Next(iter);
2884 if (obj == NULL) {
2885 if (PyErr_Occurred())
2886 goto error;
2887
2888 /* Only one item to write */
2889 if (save(self, PyTuple_GET_ITEM(firstitem, 0), 0) < 0)
2890 goto error;
2891 if (save(self, PyTuple_GET_ITEM(firstitem, 1), 0) < 0)
2892 goto error;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002893 if (_Pickler_Write(self, &setitem_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002894 goto error;
2895 Py_CLEAR(firstitem);
2896 break;
2897 }
2898
2899 /* More than one item to write */
2900
2901 /* Pump out MARK, items, SETITEMS. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002902 if (_Pickler_Write(self, &mark_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002903 goto error;
2904
2905 if (save(self, PyTuple_GET_ITEM(firstitem, 0), 0) < 0)
2906 goto error;
2907 if (save(self, PyTuple_GET_ITEM(firstitem, 1), 0) < 0)
2908 goto error;
2909 Py_CLEAR(firstitem);
2910 n = 1;
2911
2912 /* Fetch and save up to BATCHSIZE items */
2913 while (obj) {
2914 if (!PyTuple_Check(obj) || PyTuple_Size(obj) != 2) {
2915 PyErr_SetString(PyExc_TypeError, "dict items "
2916 "iterator must return 2-tuples");
2917 goto error;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002918 }
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002919 if (save(self, PyTuple_GET_ITEM(obj, 0), 0) < 0 ||
2920 save(self, PyTuple_GET_ITEM(obj, 1), 0) < 0)
2921 goto error;
2922 Py_CLEAR(obj);
2923 n += 1;
2924
2925 if (n == BATCHSIZE)
2926 break;
2927
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002928 obj = PyIter_Next(iter);
2929 if (obj == NULL) {
2930 if (PyErr_Occurred())
2931 goto error;
2932 break;
2933 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002934 }
2935
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002936 if (_Pickler_Write(self, &setitems_op, 1) < 0)
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002937 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002938
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002939 } while (n == BATCHSIZE);
2940 return 0;
2941
2942 error:
Amaury Forgeot d'Arcfb1a5eb2008-09-11 21:03:37 +00002943 Py_XDECREF(firstitem);
2944 Py_XDECREF(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00002945 return -1;
2946}
2947
Collin Winter5c9b02d2009-05-25 05:43:30 +00002948/* This is a variant of batch_dict() above that specializes for dicts, with no
2949 * support for dict subclasses. Like batch_dict(), we batch up chunks of
2950 * MARK key value ... key value SETITEMS
2951 * opcode sequences. Calling code should have arranged to first create an
2952 * empty dict, or dict-like object, for the SETITEMS to operate on.
2953 * Returns 0 on success, -1 on error.
2954 *
2955 * Note that this currently doesn't work for protocol 0.
2956 */
2957static int
2958batch_dict_exact(PicklerObject *self, PyObject *obj)
2959{
2960 PyObject *key = NULL, *value = NULL;
2961 int i;
2962 Py_ssize_t dict_size, ppos = 0;
2963
Alexandre Vassalottif70b1292009-05-25 18:00:52 +00002964 const char mark_op = MARK;
2965 const char setitem_op = SETITEM;
2966 const char setitems_op = SETITEMS;
Collin Winter5c9b02d2009-05-25 05:43:30 +00002967
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02002968 assert(obj != NULL && PyDict_CheckExact(obj));
Collin Winter5c9b02d2009-05-25 05:43:30 +00002969 assert(self->proto > 0);
2970
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02002971 dict_size = PyDict_GET_SIZE(obj);
Collin Winter5c9b02d2009-05-25 05:43:30 +00002972
2973 /* Special-case len(d) == 1 to save space. */
2974 if (dict_size == 1) {
2975 PyDict_Next(obj, &ppos, &key, &value);
2976 if (save(self, key, 0) < 0)
2977 return -1;
2978 if (save(self, value, 0) < 0)
2979 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002980 if (_Pickler_Write(self, &setitem_op, 1) < 0)
Collin Winter5c9b02d2009-05-25 05:43:30 +00002981 return -1;
2982 return 0;
2983 }
2984
2985 /* Write in batches of BATCHSIZE. */
2986 do {
2987 i = 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002988 if (_Pickler_Write(self, &mark_op, 1) < 0)
Collin Winter5c9b02d2009-05-25 05:43:30 +00002989 return -1;
2990 while (PyDict_Next(obj, &ppos, &key, &value)) {
2991 if (save(self, key, 0) < 0)
2992 return -1;
2993 if (save(self, value, 0) < 0)
2994 return -1;
2995 if (++i == BATCHSIZE)
2996 break;
2997 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00002998 if (_Pickler_Write(self, &setitems_op, 1) < 0)
Collin Winter5c9b02d2009-05-25 05:43:30 +00002999 return -1;
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02003000 if (PyDict_GET_SIZE(obj) != dict_size) {
Collin Winter5c9b02d2009-05-25 05:43:30 +00003001 PyErr_Format(
3002 PyExc_RuntimeError,
3003 "dictionary changed size during iteration");
3004 return -1;
3005 }
3006
3007 } while (i == BATCHSIZE);
3008 return 0;
3009}
3010
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003011static int
3012save_dict(PicklerObject *self, PyObject *obj)
3013{
3014 PyObject *items, *iter;
3015 char header[3];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02003016 Py_ssize_t len;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003017 int status = 0;
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02003018 assert(PyDict_Check(obj));
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003019
3020 if (self->fast && !fast_save_enter(self, obj))
3021 goto error;
3022
3023 /* Create an empty dict. */
3024 if (self->bin) {
3025 header[0] = EMPTY_DICT;
3026 len = 1;
3027 }
3028 else {
3029 header[0] = MARK;
3030 header[1] = DICT;
3031 len = 2;
3032 }
3033
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003034 if (_Pickler_Write(self, header, len) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003035 goto error;
3036
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003037 if (memo_put(self, obj) < 0)
3038 goto error;
3039
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02003040 if (PyDict_GET_SIZE(obj)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003041 /* Save the dict items. */
Collin Winter5c9b02d2009-05-25 05:43:30 +00003042 if (PyDict_CheckExact(obj) && self->proto > 0) {
3043 /* We can take certain shortcuts if we know this is a dict and
3044 not a dict subclass. */
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00003045 if (Py_EnterRecursiveCall(" while pickling an object"))
3046 goto error;
3047 status = batch_dict_exact(self, obj);
3048 Py_LeaveRecursiveCall();
Collin Winter5c9b02d2009-05-25 05:43:30 +00003049 } else {
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02003050 _Py_IDENTIFIER(items);
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02003051
Victor Stinnerad8c83a2016-09-05 17:53:15 -07003052 items = _PyObject_CallMethodId(obj, &PyId_items, NULL);
Collin Winter5c9b02d2009-05-25 05:43:30 +00003053 if (items == NULL)
3054 goto error;
3055 iter = PyObject_GetIter(items);
3056 Py_DECREF(items);
3057 if (iter == NULL)
3058 goto error;
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00003059 if (Py_EnterRecursiveCall(" while pickling an object")) {
3060 Py_DECREF(iter);
3061 goto error;
3062 }
Collin Winter5c9b02d2009-05-25 05:43:30 +00003063 status = batch_dict(self, iter);
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00003064 Py_LeaveRecursiveCall();
Collin Winter5c9b02d2009-05-25 05:43:30 +00003065 Py_DECREF(iter);
3066 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003067 }
3068
3069 if (0) {
3070 error:
3071 status = -1;
3072 }
3073
3074 if (self->fast && !fast_save_leave(self, obj))
3075 status = -1;
3076
3077 return status;
3078}
3079
3080static int
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003081save_set(PicklerObject *self, PyObject *obj)
3082{
3083 PyObject *item;
3084 int i;
3085 Py_ssize_t set_size, ppos = 0;
3086 Py_hash_t hash;
3087
3088 const char empty_set_op = EMPTY_SET;
3089 const char mark_op = MARK;
3090 const char additems_op = ADDITEMS;
3091
3092 if (self->proto < 4) {
3093 PyObject *items;
3094 PyObject *reduce_value;
3095 int status;
3096
3097 items = PySequence_List(obj);
3098 if (items == NULL) {
3099 return -1;
3100 }
3101 reduce_value = Py_BuildValue("(O(O))", (PyObject*)&PySet_Type, items);
3102 Py_DECREF(items);
3103 if (reduce_value == NULL) {
3104 return -1;
3105 }
3106 /* save_reduce() will memoize the object automatically. */
3107 status = save_reduce(self, reduce_value, obj);
3108 Py_DECREF(reduce_value);
3109 return status;
3110 }
3111
3112 if (_Pickler_Write(self, &empty_set_op, 1) < 0)
3113 return -1;
3114
3115 if (memo_put(self, obj) < 0)
3116 return -1;
3117
3118 set_size = PySet_GET_SIZE(obj);
3119 if (set_size == 0)
3120 return 0; /* nothing to do */
3121
3122 /* Write in batches of BATCHSIZE. */
3123 do {
3124 i = 0;
3125 if (_Pickler_Write(self, &mark_op, 1) < 0)
3126 return -1;
3127 while (_PySet_NextEntry(obj, &ppos, &item, &hash)) {
3128 if (save(self, item, 0) < 0)
3129 return -1;
3130 if (++i == BATCHSIZE)
3131 break;
3132 }
3133 if (_Pickler_Write(self, &additems_op, 1) < 0)
3134 return -1;
3135 if (PySet_GET_SIZE(obj) != set_size) {
3136 PyErr_Format(
3137 PyExc_RuntimeError,
3138 "set changed size during iteration");
3139 return -1;
3140 }
3141 } while (i == BATCHSIZE);
3142
3143 return 0;
3144}
3145
3146static int
3147save_frozenset(PicklerObject *self, PyObject *obj)
3148{
3149 PyObject *iter;
3150
3151 const char mark_op = MARK;
3152 const char frozenset_op = FROZENSET;
3153
3154 if (self->fast && !fast_save_enter(self, obj))
3155 return -1;
3156
3157 if (self->proto < 4) {
3158 PyObject *items;
3159 PyObject *reduce_value;
3160 int status;
3161
3162 items = PySequence_List(obj);
3163 if (items == NULL) {
3164 return -1;
3165 }
3166 reduce_value = Py_BuildValue("(O(O))", (PyObject*)&PyFrozenSet_Type,
3167 items);
3168 Py_DECREF(items);
3169 if (reduce_value == NULL) {
3170 return -1;
3171 }
3172 /* save_reduce() will memoize the object automatically. */
3173 status = save_reduce(self, reduce_value, obj);
3174 Py_DECREF(reduce_value);
3175 return status;
3176 }
3177
3178 if (_Pickler_Write(self, &mark_op, 1) < 0)
3179 return -1;
3180
3181 iter = PyObject_GetIter(obj);
Christian Heimesb3d3ee42013-11-23 21:01:40 +01003182 if (iter == NULL) {
Christian Heimes74d8d632013-11-23 21:05:31 +01003183 return -1;
Christian Heimesb3d3ee42013-11-23 21:01:40 +01003184 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003185 for (;;) {
3186 PyObject *item;
3187
3188 item = PyIter_Next(iter);
3189 if (item == NULL) {
3190 if (PyErr_Occurred()) {
3191 Py_DECREF(iter);
3192 return -1;
3193 }
3194 break;
3195 }
3196 if (save(self, item, 0) < 0) {
3197 Py_DECREF(item);
3198 Py_DECREF(iter);
3199 return -1;
3200 }
3201 Py_DECREF(item);
3202 }
3203 Py_DECREF(iter);
3204
3205 /* If the object is already in the memo, this means it is
3206 recursive. In this case, throw away everything we put on the
3207 stack, and fetch the object back from the memo. */
3208 if (PyMemoTable_Get(self->memo, obj)) {
3209 const char pop_mark_op = POP_MARK;
3210
3211 if (_Pickler_Write(self, &pop_mark_op, 1) < 0)
3212 return -1;
3213 if (memo_get(self, obj) < 0)
3214 return -1;
3215 return 0;
3216 }
3217
3218 if (_Pickler_Write(self, &frozenset_op, 1) < 0)
3219 return -1;
3220 if (memo_put(self, obj) < 0)
3221 return -1;
3222
3223 return 0;
3224}
3225
3226static int
3227fix_imports(PyObject **module_name, PyObject **global_name)
3228{
3229 PyObject *key;
3230 PyObject *item;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003231 PickleState *st = _Pickle_GetGlobalState();
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003232
3233 key = PyTuple_Pack(2, *module_name, *global_name);
3234 if (key == NULL)
3235 return -1;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003236 item = PyDict_GetItemWithError(st->name_mapping_3to2, key);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003237 Py_DECREF(key);
3238 if (item) {
3239 PyObject *fixed_module_name;
3240 PyObject *fixed_global_name;
3241
3242 if (!PyTuple_Check(item) || PyTuple_GET_SIZE(item) != 2) {
3243 PyErr_Format(PyExc_RuntimeError,
3244 "_compat_pickle.REVERSE_NAME_MAPPING values "
3245 "should be 2-tuples, not %.200s",
3246 Py_TYPE(item)->tp_name);
3247 return -1;
3248 }
3249 fixed_module_name = PyTuple_GET_ITEM(item, 0);
3250 fixed_global_name = PyTuple_GET_ITEM(item, 1);
3251 if (!PyUnicode_Check(fixed_module_name) ||
3252 !PyUnicode_Check(fixed_global_name)) {
3253 PyErr_Format(PyExc_RuntimeError,
3254 "_compat_pickle.REVERSE_NAME_MAPPING values "
3255 "should be pairs of str, not (%.200s, %.200s)",
3256 Py_TYPE(fixed_module_name)->tp_name,
3257 Py_TYPE(fixed_global_name)->tp_name);
3258 return -1;
3259 }
3260
3261 Py_CLEAR(*module_name);
3262 Py_CLEAR(*global_name);
3263 Py_INCREF(fixed_module_name);
3264 Py_INCREF(fixed_global_name);
3265 *module_name = fixed_module_name;
3266 *global_name = fixed_global_name;
Serhiy Storchakabfe18242015-03-31 13:12:37 +03003267 return 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003268 }
3269 else if (PyErr_Occurred()) {
3270 return -1;
3271 }
3272
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003273 item = PyDict_GetItemWithError(st->import_mapping_3to2, *module_name);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003274 if (item) {
3275 if (!PyUnicode_Check(item)) {
3276 PyErr_Format(PyExc_RuntimeError,
3277 "_compat_pickle.REVERSE_IMPORT_MAPPING values "
3278 "should be strings, not %.200s",
3279 Py_TYPE(item)->tp_name);
3280 return -1;
3281 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003282 Py_INCREF(item);
Serhiy Storchaka48842712016-04-06 09:45:48 +03003283 Py_XSETREF(*module_name, item);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003284 }
3285 else if (PyErr_Occurred()) {
3286 return -1;
3287 }
3288
3289 return 0;
3290}
3291
3292static int
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003293save_global(PicklerObject *self, PyObject *obj, PyObject *name)
3294{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003295 PyObject *global_name = NULL;
3296 PyObject *module_name = NULL;
3297 PyObject *module = NULL;
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003298 PyObject *parent = NULL;
3299 PyObject *dotted_path = NULL;
3300 PyObject *lastname = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003301 PyObject *cls;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003302 PickleState *st = _Pickle_GetGlobalState();
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003303 int status = 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003304 _Py_IDENTIFIER(__name__);
3305 _Py_IDENTIFIER(__qualname__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003306
3307 const char global_op = GLOBAL;
3308
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003309 if (name) {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003310 Py_INCREF(name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003311 global_name = name;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003312 }
3313 else {
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003314 if (_PyObject_LookupAttrId(obj, &PyId___qualname__, &global_name) < 0)
3315 goto error;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003316 if (global_name == NULL) {
3317 global_name = _PyObject_GetAttrId(obj, &PyId___name__);
3318 if (global_name == NULL)
3319 goto error;
3320 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003321 }
3322
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003323 dotted_path = get_dotted_path(module, global_name);
3324 if (dotted_path == NULL)
3325 goto error;
3326 module_name = whichmodule(obj, dotted_path);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003327 if (module_name == NULL)
3328 goto error;
3329
3330 /* XXX: Change to use the import C API directly with level=0 to disallow
3331 relative imports.
3332
3333 XXX: PyImport_ImportModuleLevel could be used. However, this bypasses
3334 builtins.__import__. Therefore, _pickle, unlike pickle.py, will ignore
3335 custom import functions (IMHO, this would be a nice security
3336 feature). The import C API would need to be extended to support the
3337 extra parameters of __import__ to fix that. */
3338 module = PyImport_Import(module_name);
3339 if (module == NULL) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003340 PyErr_Format(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003341 "Can't pickle %R: import of module %R failed",
3342 obj, module_name);
3343 goto error;
3344 }
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003345 lastname = PyList_GET_ITEM(dotted_path, PyList_GET_SIZE(dotted_path)-1);
3346 Py_INCREF(lastname);
3347 cls = get_deep_attribute(module, dotted_path, &parent);
3348 Py_CLEAR(dotted_path);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003349 if (cls == NULL) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003350 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003351 "Can't pickle %R: attribute lookup %S on %S failed",
3352 obj, global_name, module_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003353 goto error;
3354 }
3355 if (cls != obj) {
3356 Py_DECREF(cls);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003357 PyErr_Format(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003358 "Can't pickle %R: it's not the same object as %S.%S",
3359 obj, module_name, global_name);
3360 goto error;
3361 }
3362 Py_DECREF(cls);
3363
3364 if (self->proto >= 2) {
3365 /* See whether this is in the extension registry, and if
3366 * so generate an EXT opcode.
3367 */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003368 PyObject *extension_key;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003369 PyObject *code_obj; /* extension code as Python object */
3370 long code; /* extension code as C value */
3371 char pdata[5];
Antoine Pitrou82be19f2011-08-29 23:09:33 +02003372 Py_ssize_t n;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003373
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003374 extension_key = PyTuple_Pack(2, module_name, global_name);
3375 if (extension_key == NULL) {
3376 goto error;
3377 }
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08003378 code_obj = PyDict_GetItemWithError(st->extension_registry,
3379 extension_key);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003380 Py_DECREF(extension_key);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003381 /* The object is not registered in the extension registry.
3382 This is the most likely code path. */
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08003383 if (code_obj == NULL) {
3384 if (PyErr_Occurred()) {
3385 goto error;
3386 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003387 goto gen_global;
Alexandre Vassalotti567eba12013-11-28 17:09:16 -08003388 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003389
3390 /* XXX: pickle.py doesn't check neither the type, nor the range
3391 of the value returned by the extension_registry. It should for
3392 consistency. */
3393
3394 /* Verify code_obj has the right type and value. */
3395 if (!PyLong_Check(code_obj)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003396 PyErr_Format(st->PicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003397 "Can't pickle %R: extension code %R isn't an integer",
3398 obj, code_obj);
3399 goto error;
3400 }
3401 code = PyLong_AS_LONG(code_obj);
3402 if (code <= 0 || code > 0x7fffffffL) {
Antoine Pitrou82be19f2011-08-29 23:09:33 +02003403 if (!PyErr_Occurred())
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003404 PyErr_Format(st->PicklingError, "Can't pickle %R: extension "
3405 "code %ld is out of range", obj, code);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003406 goto error;
3407 }
3408
3409 /* Generate an EXT opcode. */
3410 if (code <= 0xff) {
3411 pdata[0] = EXT1;
3412 pdata[1] = (unsigned char)code;
3413 n = 2;
3414 }
3415 else if (code <= 0xffff) {
3416 pdata[0] = EXT2;
3417 pdata[1] = (unsigned char)(code & 0xff);
3418 pdata[2] = (unsigned char)((code >> 8) & 0xff);
3419 n = 3;
3420 }
3421 else {
3422 pdata[0] = EXT4;
3423 pdata[1] = (unsigned char)(code & 0xff);
3424 pdata[2] = (unsigned char)((code >> 8) & 0xff);
3425 pdata[3] = (unsigned char)((code >> 16) & 0xff);
3426 pdata[4] = (unsigned char)((code >> 24) & 0xff);
3427 n = 5;
3428 }
3429
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003430 if (_Pickler_Write(self, pdata, n) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003431 goto error;
3432 }
3433 else {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003434 gen_global:
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003435 if (parent == module) {
3436 Py_INCREF(lastname);
3437 Py_DECREF(global_name);
3438 global_name = lastname;
3439 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003440 if (self->proto >= 4) {
3441 const char stack_global_op = STACK_GLOBAL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003442
Christian Heimese8b1ba12013-11-23 21:13:39 +01003443 if (save(self, module_name, 0) < 0)
3444 goto error;
3445 if (save(self, global_name, 0) < 0)
3446 goto error;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003447
3448 if (_Pickler_Write(self, &stack_global_op, 1) < 0)
3449 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003450 }
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003451 else if (parent != module) {
3452 PickleState *st = _Pickle_GetGlobalState();
3453 PyObject *reduce_value = Py_BuildValue("(O(OO))",
3454 st->getattr, parent, lastname);
3455 status = save_reduce(self, reduce_value, NULL);
3456 Py_DECREF(reduce_value);
3457 if (status < 0)
3458 goto error;
3459 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003460 else {
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003461 /* Generate a normal global opcode if we are using a pickle
3462 protocol < 4, or if the object is not registered in the
3463 extension registry. */
3464 PyObject *encoded;
3465 PyObject *(*unicode_encoder)(PyObject *);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003466
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003467 if (_Pickler_Write(self, &global_op, 1) < 0)
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003468 goto error;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003469
3470 /* For protocol < 3 and if the user didn't request against doing
3471 so, we convert module names to the old 2.x module names. */
3472 if (self->proto < 3 && self->fix_imports) {
3473 if (fix_imports(&module_name, &global_name) < 0) {
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003474 goto error;
3475 }
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003476 }
3477
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003478 /* Since Python 3.0 now supports non-ASCII identifiers, we encode
3479 both the module name and the global name using UTF-8. We do so
3480 only when we are using the pickle protocol newer than version
3481 3. This is to ensure compatibility with older Unpickler running
3482 on Python 2.x. */
3483 if (self->proto == 3) {
3484 unicode_encoder = PyUnicode_AsUTF8String;
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003485 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003486 else {
3487 unicode_encoder = PyUnicode_AsASCIIString;
3488 }
3489 encoded = unicode_encoder(module_name);
3490 if (encoded == NULL) {
3491 if (PyErr_ExceptionMatches(PyExc_UnicodeEncodeError))
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003492 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003493 "can't pickle module identifier '%S' using "
3494 "pickle protocol %i",
3495 module_name, self->proto);
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00003496 goto error;
3497 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003498 if (_Pickler_Write(self, PyBytes_AS_STRING(encoded),
3499 PyBytes_GET_SIZE(encoded)) < 0) {
3500 Py_DECREF(encoded);
3501 goto error;
3502 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003503 Py_DECREF(encoded);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003504 if(_Pickler_Write(self, "\n", 1) < 0)
3505 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003506
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003507 /* Save the name of the module. */
3508 encoded = unicode_encoder(global_name);
3509 if (encoded == NULL) {
3510 if (PyErr_ExceptionMatches(PyExc_UnicodeEncodeError))
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003511 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003512 "can't pickle global identifier '%S' using "
3513 "pickle protocol %i",
3514 global_name, self->proto);
3515 goto error;
3516 }
3517 if (_Pickler_Write(self, PyBytes_AS_STRING(encoded),
3518 PyBytes_GET_SIZE(encoded)) < 0) {
3519 Py_DECREF(encoded);
3520 goto error;
3521 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003522 Py_DECREF(encoded);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003523 if (_Pickler_Write(self, "\n", 1) < 0)
3524 goto error;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003525 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003526 /* Memoize the object. */
3527 if (memo_put(self, obj) < 0)
3528 goto error;
3529 }
3530
3531 if (0) {
3532 error:
3533 status = -1;
3534 }
3535 Py_XDECREF(module_name);
3536 Py_XDECREF(global_name);
3537 Py_XDECREF(module);
Serhiy Storchaka58e41342015-03-31 14:07:24 +03003538 Py_XDECREF(parent);
3539 Py_XDECREF(dotted_path);
3540 Py_XDECREF(lastname);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003541
3542 return status;
3543}
3544
3545static int
Alexandre Vassalotti19b6fa62013-11-30 16:06:39 -08003546save_singleton_type(PicklerObject *self, PyObject *obj, PyObject *singleton)
3547{
3548 PyObject *reduce_value;
3549 int status;
3550
3551 reduce_value = Py_BuildValue("O(O)", &PyType_Type, singleton);
3552 if (reduce_value == NULL) {
3553 return -1;
3554 }
3555 status = save_reduce(self, reduce_value, obj);
3556 Py_DECREF(reduce_value);
3557 return status;
3558}
3559
3560static int
3561save_type(PicklerObject *self, PyObject *obj)
3562{
Alexandre Vassalotti65846c62013-11-30 17:55:48 -08003563 if (obj == (PyObject *)&_PyNone_Type) {
Alexandre Vassalotti19b6fa62013-11-30 16:06:39 -08003564 return save_singleton_type(self, obj, Py_None);
3565 }
3566 else if (obj == (PyObject *)&PyEllipsis_Type) {
3567 return save_singleton_type(self, obj, Py_Ellipsis);
3568 }
Alexandre Vassalotti65846c62013-11-30 17:55:48 -08003569 else if (obj == (PyObject *)&_PyNotImplemented_Type) {
Alexandre Vassalotti19b6fa62013-11-30 16:06:39 -08003570 return save_singleton_type(self, obj, Py_NotImplemented);
3571 }
3572 return save_global(self, obj, NULL);
3573}
3574
3575static int
Serhiy Storchaka986375e2017-11-30 22:48:31 +02003576save_pers(PicklerObject *self, PyObject *obj)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003577{
3578 PyObject *pid = NULL;
3579 int status = 0;
3580
3581 const char persid_op = PERSID;
3582 const char binpersid_op = BINPERSID;
3583
Serhiy Storchaka986375e2017-11-30 22:48:31 +02003584 pid = call_method(self->pers_func, self->pers_func_self, obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003585 if (pid == NULL)
3586 return -1;
3587
3588 if (pid != Py_None) {
3589 if (self->bin) {
3590 if (save(self, pid, 1) < 0 ||
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003591 _Pickler_Write(self, &binpersid_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003592 goto error;
3593 }
3594 else {
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003595 PyObject *pid_str;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003596
3597 pid_str = PyObject_Str(pid);
3598 if (pid_str == NULL)
3599 goto error;
3600
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003601 /* XXX: Should it check whether the pid contains embedded
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003602 newlines? */
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003603 if (!PyUnicode_IS_ASCII(pid_str)) {
3604 PyErr_SetString(_Pickle_GetGlobalState()->PicklingError,
3605 "persistent IDs in protocol 0 must be "
3606 "ASCII strings");
3607 Py_DECREF(pid_str);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003608 goto error;
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003609 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003610
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003611 if (_Pickler_Write(self, &persid_op, 1) < 0 ||
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003612 _Pickler_Write(self, PyUnicode_DATA(pid_str),
3613 PyUnicode_GET_LENGTH(pid_str)) < 0 ||
3614 _Pickler_Write(self, "\n", 1) < 0) {
3615 Py_DECREF(pid_str);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003616 goto error;
Serhiy Storchakadec25af2016-07-17 11:24:17 +03003617 }
3618 Py_DECREF(pid_str);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003619 }
3620 status = 1;
3621 }
3622
3623 if (0) {
3624 error:
3625 status = -1;
3626 }
3627 Py_XDECREF(pid);
3628
3629 return status;
3630}
3631
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003632static PyObject *
3633get_class(PyObject *obj)
3634{
3635 PyObject *cls;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003636 _Py_IDENTIFIER(__class__);
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003637
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003638 if (_PyObject_LookupAttrId(obj, &PyId___class__, &cls) == 0) {
3639 cls = (PyObject *) Py_TYPE(obj);
3640 Py_INCREF(cls);
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003641 }
3642 return cls;
3643}
3644
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003645/* We're saving obj, and args is the 2-thru-5 tuple returned by the
3646 * appropriate __reduce__ method for obj.
3647 */
3648static int
3649save_reduce(PicklerObject *self, PyObject *args, PyObject *obj)
3650{
3651 PyObject *callable;
3652 PyObject *argtup;
3653 PyObject *state = NULL;
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003654 PyObject *listitems = Py_None;
3655 PyObject *dictitems = Py_None;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003656 PickleState *st = _Pickle_GetGlobalState();
Hirokazu Yamamotob46a6332008-11-04 00:35:10 +00003657 Py_ssize_t size;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003658 int use_newobj = 0, use_newobj_ex = 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003659
3660 const char reduce_op = REDUCE;
3661 const char build_op = BUILD;
3662 const char newobj_op = NEWOBJ;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003663 const char newobj_ex_op = NEWOBJ_EX;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003664
Hirokazu Yamamotob46a6332008-11-04 00:35:10 +00003665 size = PyTuple_Size(args);
3666 if (size < 2 || size > 5) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003667 PyErr_SetString(st->PicklingError, "tuple returned by "
Hirokazu Yamamotob46a6332008-11-04 00:35:10 +00003668 "__reduce__ must contain 2 through 5 elements");
3669 return -1;
3670 }
3671
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003672 if (!PyArg_UnpackTuple(args, "save_reduce", 2, 5,
3673 &callable, &argtup, &state, &listitems, &dictitems))
3674 return -1;
3675
3676 if (!PyCallable_Check(callable)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003677 PyErr_SetString(st->PicklingError, "first item of the tuple "
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003678 "returned by __reduce__ must be callable");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003679 return -1;
3680 }
3681 if (!PyTuple_Check(argtup)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003682 PyErr_SetString(st->PicklingError, "second item of the tuple "
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003683 "returned by __reduce__ must be a tuple");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003684 return -1;
3685 }
3686
3687 if (state == Py_None)
3688 state = NULL;
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003689
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003690 if (listitems == Py_None)
3691 listitems = NULL;
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003692 else if (!PyIter_Check(listitems)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003693 PyErr_Format(st->PicklingError, "fourth element of the tuple "
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003694 "returned by __reduce__ must be an iterator, not %s",
3695 Py_TYPE(listitems)->tp_name);
3696 return -1;
3697 }
3698
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003699 if (dictitems == Py_None)
3700 dictitems = NULL;
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003701 else if (!PyIter_Check(dictitems)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003702 PyErr_Format(st->PicklingError, "fifth element of the tuple "
Amaury Forgeot d'Arc424b4812008-10-30 22:25:31 +00003703 "returned by __reduce__ must be an iterator, not %s",
3704 Py_TYPE(dictitems)->tp_name);
3705 return -1;
3706 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003707
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003708 if (self->proto >= 2) {
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003709 PyObject *name;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003710 _Py_IDENTIFIER(__name__);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003711
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003712 if (_PyObject_LookupAttrId(callable, &PyId___name__, &name) < 0) {
3713 return -1;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003714 }
Serhiy Storchakaf320be72018-01-25 10:49:40 +02003715 if (name != NULL && PyUnicode_Check(name)) {
Serhiy Storchaka0d554d72015-10-10 22:42:18 +03003716 _Py_IDENTIFIER(__newobj_ex__);
Serhiy Storchakaf0f35a62017-01-09 10:09:43 +02003717 use_newobj_ex = _PyUnicode_EqualToASCIIId(
3718 name, &PyId___newobj_ex__);
Serhiy Storchaka707b5cc2014-12-16 19:43:46 +02003719 if (!use_newobj_ex) {
3720 _Py_IDENTIFIER(__newobj__);
Serhiy Storchaka9937d902017-01-09 10:04:34 +02003721 use_newobj = _PyUnicode_EqualToASCIIId(name, &PyId___newobj__);
Serhiy Storchaka707b5cc2014-12-16 19:43:46 +02003722 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003723 }
Serhiy Storchaka707b5cc2014-12-16 19:43:46 +02003724 Py_XDECREF(name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003725 }
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003726
3727 if (use_newobj_ex) {
3728 PyObject *cls;
3729 PyObject *args;
3730 PyObject *kwargs;
3731
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003732 if (PyTuple_GET_SIZE(argtup) != 3) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003733 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003734 "length of the NEWOBJ_EX argument tuple must be "
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003735 "exactly 3, not %zd", PyTuple_GET_SIZE(argtup));
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003736 return -1;
3737 }
3738
3739 cls = PyTuple_GET_ITEM(argtup, 0);
3740 if (!PyType_Check(cls)) {
Larry Hastings61272b72014-01-07 12:41:53 -08003741 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003742 "first item from NEWOBJ_EX argument tuple must "
3743 "be a class, not %.200s", Py_TYPE(cls)->tp_name);
3744 return -1;
3745 }
3746 args = PyTuple_GET_ITEM(argtup, 1);
3747 if (!PyTuple_Check(args)) {
Larry Hastings61272b72014-01-07 12:41:53 -08003748 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003749 "second item from NEWOBJ_EX argument tuple must "
3750 "be a tuple, not %.200s", Py_TYPE(args)->tp_name);
3751 return -1;
3752 }
3753 kwargs = PyTuple_GET_ITEM(argtup, 2);
3754 if (!PyDict_Check(kwargs)) {
Larry Hastings61272b72014-01-07 12:41:53 -08003755 PyErr_Format(st->PicklingError,
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003756 "third item from NEWOBJ_EX argument tuple must "
3757 "be a dict, not %.200s", Py_TYPE(kwargs)->tp_name);
3758 return -1;
3759 }
3760
Serhiy Storchaka0d554d72015-10-10 22:42:18 +03003761 if (self->proto >= 4) {
3762 if (save(self, cls, 0) < 0 ||
3763 save(self, args, 0) < 0 ||
3764 save(self, kwargs, 0) < 0 ||
3765 _Pickler_Write(self, &newobj_ex_op, 1) < 0) {
3766 return -1;
3767 }
3768 }
3769 else {
3770 PyObject *newargs;
3771 PyObject *cls_new;
3772 Py_ssize_t i;
3773 _Py_IDENTIFIER(__new__);
3774
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003775 newargs = PyTuple_New(PyTuple_GET_SIZE(args) + 2);
Serhiy Storchaka0d554d72015-10-10 22:42:18 +03003776 if (newargs == NULL)
3777 return -1;
3778
3779 cls_new = _PyObject_GetAttrId(cls, &PyId___new__);
3780 if (cls_new == NULL) {
3781 Py_DECREF(newargs);
3782 return -1;
3783 }
3784 PyTuple_SET_ITEM(newargs, 0, cls_new);
3785 Py_INCREF(cls);
3786 PyTuple_SET_ITEM(newargs, 1, cls);
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003787 for (i = 0; i < PyTuple_GET_SIZE(args); i++) {
Serhiy Storchaka0d554d72015-10-10 22:42:18 +03003788 PyObject *item = PyTuple_GET_ITEM(args, i);
3789 Py_INCREF(item);
3790 PyTuple_SET_ITEM(newargs, i + 2, item);
3791 }
3792
3793 callable = PyObject_Call(st->partial, newargs, kwargs);
3794 Py_DECREF(newargs);
3795 if (callable == NULL)
3796 return -1;
3797
3798 newargs = PyTuple_New(0);
3799 if (newargs == NULL) {
3800 Py_DECREF(callable);
3801 return -1;
3802 }
3803
3804 if (save(self, callable, 0) < 0 ||
3805 save(self, newargs, 0) < 0 ||
3806 _Pickler_Write(self, &reduce_op, 1) < 0) {
3807 Py_DECREF(newargs);
3808 Py_DECREF(callable);
3809 return -1;
3810 }
3811 Py_DECREF(newargs);
3812 Py_DECREF(callable);
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003813 }
3814 }
3815 else if (use_newobj) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003816 PyObject *cls;
3817 PyObject *newargtup;
3818 PyObject *obj_class;
3819 int p;
3820
3821 /* Sanity checks. */
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003822 if (PyTuple_GET_SIZE(argtup) < 1) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003823 PyErr_SetString(st->PicklingError, "__newobj__ arglist is empty");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003824 return -1;
3825 }
3826
3827 cls = PyTuple_GET_ITEM(argtup, 0);
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003828 if (!PyType_Check(cls)) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003829 PyErr_SetString(st->PicklingError, "args[0] from "
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003830 "__newobj__ args is not a type");
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003831 return -1;
3832 }
3833
3834 if (obj != NULL) {
Antoine Pitrou16c4ce12011-03-11 21:30:43 +01003835 obj_class = get_class(obj);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003836 p = obj_class != cls; /* true iff a problem */
3837 Py_DECREF(obj_class);
3838 if (p) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08003839 PyErr_SetString(st->PicklingError, "args[0] from "
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003840 "__newobj__ args has the wrong class");
3841 return -1;
3842 }
3843 }
3844 /* XXX: These calls save() are prone to infinite recursion. Imagine
3845 what happen if the value returned by the __reduce__() method of
3846 some extension type contains another object of the same type. Ouch!
3847
3848 Here is a quick example, that I ran into, to illustrate what I
3849 mean:
3850
3851 >>> import pickle, copyreg
3852 >>> copyreg.dispatch_table.pop(complex)
3853 >>> pickle.dumps(1+2j)
3854 Traceback (most recent call last):
3855 ...
Yury Selivanovf488fb42015-07-03 01:04:23 -04003856 RecursionError: maximum recursion depth exceeded
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003857
3858 Removing the complex class from copyreg.dispatch_table made the
3859 __reduce_ex__() method emit another complex object:
3860
3861 >>> (1+1j).__reduce_ex__(2)
3862 (<function __newobj__ at 0xb7b71c3c>,
3863 (<class 'complex'>, (1+1j)), None, None, None)
3864
3865 Thus when save() was called on newargstup (the 2nd item) recursion
3866 ensued. Of course, the bug was in the complex class which had a
3867 broken __getnewargs__() that emitted another complex object. But,
3868 the point, here, is it is quite easy to end up with a broken reduce
3869 function. */
3870
3871 /* Save the class and its __new__ arguments. */
3872 if (save(self, cls, 0) < 0)
3873 return -1;
3874
Serhiy Storchakafff9a312017-03-21 08:53:25 +02003875 newargtup = PyTuple_GetSlice(argtup, 1, PyTuple_GET_SIZE(argtup));
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003876 if (newargtup == NULL)
3877 return -1;
3878
3879 p = save(self, newargtup, 0);
3880 Py_DECREF(newargtup);
3881 if (p < 0)
3882 return -1;
3883
3884 /* Add NEWOBJ opcode. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003885 if (_Pickler_Write(self, &newobj_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003886 return -1;
3887 }
3888 else { /* Not using NEWOBJ. */
3889 if (save(self, callable, 0) < 0 ||
3890 save(self, argtup, 0) < 0 ||
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003891 _Pickler_Write(self, &reduce_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003892 return -1;
3893 }
3894
3895 /* obj can be NULL when save_reduce() is used directly. A NULL obj means
3896 the caller do not want to memoize the object. Not particularly useful,
3897 but that is to mimic the behavior save_reduce() in pickle.py when
3898 obj is None. */
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01003899 if (obj != NULL) {
3900 /* If the object is already in the memo, this means it is
3901 recursive. In this case, throw away everything we put on the
3902 stack, and fetch the object back from the memo. */
3903 if (PyMemoTable_Get(self->memo, obj)) {
3904 const char pop_op = POP;
3905
3906 if (_Pickler_Write(self, &pop_op, 1) < 0)
3907 return -1;
3908 if (memo_get(self, obj) < 0)
3909 return -1;
3910
3911 return 0;
3912 }
3913 else if (memo_put(self, obj) < 0)
3914 return -1;
3915 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003916
3917 if (listitems && batch_list(self, listitems) < 0)
3918 return -1;
3919
3920 if (dictitems && batch_dict(self, dictitems) < 0)
3921 return -1;
3922
3923 if (state) {
Victor Stinner121aab42011-09-29 23:40:53 +02003924 if (save(self, state, 0) < 0 ||
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003925 _Pickler_Write(self, &build_op, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003926 return -1;
3927 }
3928
3929 return 0;
3930}
3931
3932static int
3933save(PicklerObject *self, PyObject *obj, int pers_save)
3934{
3935 PyTypeObject *type;
3936 PyObject *reduce_func = NULL;
3937 PyObject *reduce_value = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003938 int status = 0;
3939
Alexandre Vassalottib6a2f2a2013-11-23 20:30:03 -08003940 if (_Pickler_OpcodeBoundary(self) < 0)
3941 return -1;
3942
Antoine Pitroue6d4c5b2011-01-23 17:12:25 +00003943 if (Py_EnterRecursiveCall(" while pickling an object"))
Alexandre Vassalottidff18342008-07-13 18:48:30 +00003944 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003945
3946 /* The extra pers_save argument is necessary to avoid calling save_pers()
3947 on its returned object. */
3948 if (!pers_save && self->pers_func) {
3949 /* save_pers() returns:
3950 -1 to signal an error;
3951 0 if it did nothing successfully;
3952 1 if a persistent id was saved.
3953 */
Serhiy Storchaka986375e2017-11-30 22:48:31 +02003954 if ((status = save_pers(self, obj)) != 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003955 goto done;
3956 }
3957
3958 type = Py_TYPE(obj);
3959
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003960 /* The old cPickle had an optimization that used switch-case statement
3961 dispatching on the first letter of the type name. This has was removed
3962 since benchmarks shown that this optimization was actually slowing
3963 things down. */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003964
3965 /* Atom types; these aren't memoized, so don't check the memo. */
3966
3967 if (obj == Py_None) {
3968 status = save_none(self, obj);
3969 goto done;
3970 }
3971 else if (obj == Py_False || obj == Py_True) {
3972 status = save_bool(self, obj);
3973 goto done;
3974 }
3975 else if (type == &PyLong_Type) {
3976 status = save_long(self, obj);
3977 goto done;
3978 }
3979 else if (type == &PyFloat_Type) {
3980 status = save_float(self, obj);
3981 goto done;
3982 }
3983
3984 /* Check the memo to see if it has the object. If so, generate
3985 a GET (or BINGET) opcode, instead of pickling the object
3986 once again. */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00003987 if (PyMemoTable_Get(self->memo, obj)) {
3988 if (memo_get(self, obj) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00003989 goto error;
3990 goto done;
3991 }
3992
3993 if (type == &PyBytes_Type) {
3994 status = save_bytes(self, obj);
3995 goto done;
3996 }
3997 else if (type == &PyUnicode_Type) {
3998 status = save_unicode(self, obj);
3999 goto done;
4000 }
4001 else if (type == &PyDict_Type) {
4002 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 ||
Miss Islington (bot)74735e22018-04-03 14:35:11 -07004157 _Pickler_Write(self, &stop_op, 1) < 0 ||
4158 _Pickler_CommitFrame(self) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00004159 return -1;
Miss Islington (bot)74735e22018-04-03 14:35:11 -07004160 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
6291 if ((self->num_marks + 1) >= self->marks_size) {
6292 size_t alloc;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006293
6294 /* Use the size_t type to check for overflow. */
6295 alloc = ((size_t)self->num_marks << 1) + 20;
Antoine Pitrou82be19f2011-08-29 23:09:33 +02006296 if (alloc > (PY_SSIZE_T_MAX / sizeof(Py_ssize_t)) ||
Alexandre Vassalotti7634ff52008-06-13 02:16:06 +00006297 alloc <= ((size_t)self->num_marks + 1)) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006298 PyErr_NoMemory();
6299 return -1;
6300 }
6301
Miss Islington (bot)962051e2018-08-16 00:53:00 -04006302 Py_ssize_t *marks_old = self->marks;
6303 PyMem_RESIZE(self->marks, Py_ssize_t, alloc);
Benjamin Peterson59b08c12015-06-27 13:41:33 -05006304 if (self->marks == NULL) {
Miss Islington (bot)962051e2018-08-16 00:53:00 -04006305 PyMem_FREE(marks_old);
Benjamin Peterson59b08c12015-06-27 13:41:33 -05006306 self->marks_size = 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006307 PyErr_NoMemory();
6308 return -1;
6309 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006310 self->marks_size = (Py_ssize_t)alloc;
6311 }
6312
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02006313 self->stack->mark_set = 1;
6314 self->marks[self->num_marks++] = self->stack->fence = Py_SIZE(self->stack);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006315
6316 return 0;
6317}
6318
6319static int
6320load_reduce(UnpicklerObject *self)
6321{
6322 PyObject *callable = NULL;
6323 PyObject *argtup = NULL;
6324 PyObject *obj = NULL;
6325
6326 PDATA_POP(self->stack, argtup);
6327 if (argtup == NULL)
6328 return -1;
6329 PDATA_POP(self->stack, callable);
6330 if (callable) {
Alexander Belopolskyd92f0402010-07-17 22:50:45 +00006331 obj = PyObject_CallObject(callable, argtup);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006332 Py_DECREF(callable);
6333 }
6334 Py_DECREF(argtup);
6335
6336 if (obj == NULL)
6337 return -1;
6338
6339 PDATA_PUSH(self->stack, obj, -1);
6340 return 0;
6341}
6342
6343/* Just raises an error if we don't know the protocol specified. PROTO
6344 * is the first opcode for protocols >= 2.
6345 */
6346static int
6347load_proto(UnpicklerObject *self)
6348{
6349 char *s;
6350 int i;
6351
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006352 if (_Unpickler_Read(self, &s, 1) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006353 return -1;
6354
6355 i = (unsigned char)s[0];
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006356 if (i <= HIGHEST_PROTOCOL) {
6357 self->proto = i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006358 return 0;
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006359 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006360
6361 PyErr_Format(PyExc_ValueError, "unsupported pickle protocol: %d", i);
6362 return -1;
6363}
6364
Alexandre Vassalottib6a2f2a2013-11-23 20:30:03 -08006365static int
6366load_frame(UnpicklerObject *self)
6367{
6368 char *s;
6369 Py_ssize_t frame_len;
6370
6371 if (_Unpickler_Read(self, &s, 8) < 0)
6372 return -1;
6373
6374 frame_len = calc_binsize(s, 8);
6375 if (frame_len < 0) {
6376 PyErr_Format(PyExc_OverflowError,
6377 "FRAME length exceeds system's maximum of %zd bytes",
6378 PY_SSIZE_T_MAX);
6379 return -1;
6380 }
6381
6382 if (_Unpickler_Read(self, &s, frame_len) < 0)
6383 return -1;
6384
6385 /* Rewind to start of frame */
6386 self->next_read_idx -= frame_len;
6387 return 0;
6388}
6389
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006390static PyObject *
6391load(UnpicklerObject *self)
6392{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006393 PyObject *value = NULL;
Christian Heimes27ea78b2014-01-27 01:03:53 +01006394 char *s = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006395
6396 self->num_marks = 0;
Serhiy Storchaka59fb6342015-12-06 22:01:35 +02006397 self->stack->mark_set = 0;
6398 self->stack->fence = 0;
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006399 self->proto = 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006400 if (Py_SIZE(self->stack))
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006401 Pdata_clear(self->stack, 0);
6402
6403 /* Convenient macros for the dispatch while-switch loop just below. */
6404#define OP(opcode, load_func) \
6405 case opcode: if (load_func(self) < 0) break; continue;
6406
6407#define OP_ARG(opcode, load_func, arg) \
6408 case opcode: if (load_func(self, (arg)) < 0) break; continue;
6409
6410 while (1) {
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03006411 if (_Unpickler_Read(self, &s, 1) < 0) {
6412 PickleState *st = _Pickle_GetGlobalState();
6413 if (PyErr_ExceptionMatches(st->UnpicklingError)) {
6414 PyErr_Format(PyExc_EOFError, "Ran out of input");
6415 }
6416 return NULL;
6417 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006418
6419 switch ((enum opcode)s[0]) {
6420 OP(NONE, load_none)
6421 OP(BININT, load_binint)
6422 OP(BININT1, load_binint1)
6423 OP(BININT2, load_binint2)
6424 OP(INT, load_int)
6425 OP(LONG, load_long)
6426 OP_ARG(LONG1, load_counted_long, 1)
6427 OP_ARG(LONG4, load_counted_long, 4)
6428 OP(FLOAT, load_float)
6429 OP(BINFLOAT, load_binfloat)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006430 OP_ARG(SHORT_BINBYTES, load_counted_binbytes, 1)
6431 OP_ARG(BINBYTES, load_counted_binbytes, 4)
6432 OP_ARG(BINBYTES8, load_counted_binbytes, 8)
6433 OP_ARG(SHORT_BINSTRING, load_counted_binstring, 1)
6434 OP_ARG(BINSTRING, load_counted_binstring, 4)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006435 OP(STRING, load_string)
6436 OP(UNICODE, load_unicode)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006437 OP_ARG(SHORT_BINUNICODE, load_counted_binunicode, 1)
6438 OP_ARG(BINUNICODE, load_counted_binunicode, 4)
6439 OP_ARG(BINUNICODE8, load_counted_binunicode, 8)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006440 OP_ARG(EMPTY_TUPLE, load_counted_tuple, 0)
6441 OP_ARG(TUPLE1, load_counted_tuple, 1)
6442 OP_ARG(TUPLE2, load_counted_tuple, 2)
6443 OP_ARG(TUPLE3, load_counted_tuple, 3)
6444 OP(TUPLE, load_tuple)
6445 OP(EMPTY_LIST, load_empty_list)
6446 OP(LIST, load_list)
6447 OP(EMPTY_DICT, load_empty_dict)
6448 OP(DICT, load_dict)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006449 OP(EMPTY_SET, load_empty_set)
6450 OP(ADDITEMS, load_additems)
6451 OP(FROZENSET, load_frozenset)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006452 OP(OBJ, load_obj)
6453 OP(INST, load_inst)
6454 OP(NEWOBJ, load_newobj)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006455 OP(NEWOBJ_EX, load_newobj_ex)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006456 OP(GLOBAL, load_global)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006457 OP(STACK_GLOBAL, load_stack_global)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006458 OP(APPEND, load_append)
6459 OP(APPENDS, load_appends)
6460 OP(BUILD, load_build)
6461 OP(DUP, load_dup)
6462 OP(BINGET, load_binget)
6463 OP(LONG_BINGET, load_long_binget)
6464 OP(GET, load_get)
6465 OP(MARK, load_mark)
6466 OP(BINPUT, load_binput)
6467 OP(LONG_BINPUT, load_long_binput)
6468 OP(PUT, load_put)
Antoine Pitrouc9dc4a22013-11-23 18:59:12 +01006469 OP(MEMOIZE, load_memoize)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006470 OP(POP, load_pop)
6471 OP(POP_MARK, load_pop_mark)
6472 OP(SETITEM, load_setitem)
6473 OP(SETITEMS, load_setitems)
6474 OP(PERSID, load_persid)
6475 OP(BINPERSID, load_binpersid)
6476 OP(REDUCE, load_reduce)
6477 OP(PROTO, load_proto)
Alexandre Vassalottib6a2f2a2013-11-23 20:30:03 -08006478 OP(FRAME, load_frame)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006479 OP_ARG(EXT1, load_extension, 1)
6480 OP_ARG(EXT2, load_extension, 2)
6481 OP_ARG(EXT4, load_extension, 4)
6482 OP_ARG(NEWTRUE, load_bool, Py_True)
6483 OP_ARG(NEWFALSE, load_bool, Py_False)
6484
6485 case STOP:
6486 break;
6487
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006488 default:
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03006489 {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006490 PickleState *st = _Pickle_GetGlobalState();
Serhiy Storchaka90493ab2016-09-06 23:55:11 +03006491 unsigned char c = (unsigned char) *s;
6492 if (0x20 <= c && c <= 0x7e && c != '\'' && c != '\\') {
6493 PyErr_Format(st->UnpicklingError,
6494 "invalid load key, '%c'.", c);
6495 }
6496 else {
6497 PyErr_Format(st->UnpicklingError,
6498 "invalid load key, '\\x%02x'.", c);
6499 }
6500 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006501 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006502 }
6503
6504 break; /* and we are done! */
6505 }
6506
Alexandre Vassalottib6a2f2a2013-11-23 20:30:03 -08006507 if (PyErr_Occurred()) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006508 return NULL;
6509 }
6510
Victor Stinner2ae57e32013-10-31 13:39:23 +01006511 if (_Unpickler_SkipConsumed(self) < 0)
6512 return NULL;
6513
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006514 PDATA_POP(self->stack, value);
6515 return value;
6516}
6517
Larry Hastings61272b72014-01-07 12:41:53 -08006518/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006519
6520_pickle.Unpickler.load
6521
6522Load a pickle.
6523
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006524Read a pickled object representation from the open file object given
6525in the constructor, and return the reconstituted object hierarchy
6526specified therein.
Larry Hastings61272b72014-01-07 12:41:53 -08006527[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006528
Larry Hastings3cceb382014-01-04 11:09:09 -08006529static PyObject *
Larry Hastingsc2047262014-01-25 20:43:29 -08006530_pickle_Unpickler_load_impl(UnpicklerObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08006531/*[clinic end generated code: output=fdcc488aad675b14 input=acbb91a42fa9b7b9]*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006532{
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006533 UnpicklerObject *unpickler = (UnpicklerObject*)self;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006534
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006535 /* Check whether the Unpickler was initialized correctly. This prevents
6536 segfaulting if a subclass overridden __init__ with a function that does
6537 not call Unpickler.__init__(). Here, we simply ensure that self->read
6538 is not NULL. */
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006539 if (unpickler->read == NULL) {
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006540 PickleState *st = _Pickle_GetGlobalState();
6541 PyErr_Format(st->UnpicklingError,
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006542 "Unpickler.__init__() was not called by %s.__init__()",
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006543 Py_TYPE(unpickler)->tp_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006544 return NULL;
6545 }
6546
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006547 return load(unpickler);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006548}
6549
6550/* The name of find_class() is misleading. In newer pickle protocols, this
6551 function is used for loading any global (i.e., functions), not just
6552 classes. The name is kept only for backward compatibility. */
6553
Larry Hastings61272b72014-01-07 12:41:53 -08006554/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006555
6556_pickle.Unpickler.find_class
6557
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006558 module_name: object
6559 global_name: object
6560 /
6561
6562Return an object from a specified module.
6563
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006564If necessary, the module will be imported. Subclasses may override
6565this method (e.g. to restrict unpickling of arbitrary classes and
6566functions).
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006567
6568This method is called whenever a class or a function object is
6569needed. Both arguments passed are str objects.
Larry Hastings61272b72014-01-07 12:41:53 -08006570[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006571
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006572static PyObject *
Larry Hastings89964c42015-04-14 18:07:59 -04006573_pickle_Unpickler_find_class_impl(UnpicklerObject *self,
6574 PyObject *module_name,
6575 PyObject *global_name)
6576/*[clinic end generated code: output=becc08d7f9ed41e3 input=e2e6a865de093ef4]*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006577{
6578 PyObject *global;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006579 PyObject *module;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006580
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006581 /* Try to map the old names used in Python 2.x to the new ones used in
6582 Python 3.x. We do this only with old pickle protocols and when the
6583 user has not disabled the feature. */
6584 if (self->proto < 3 && self->fix_imports) {
6585 PyObject *key;
6586 PyObject *item;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006587 PickleState *st = _Pickle_GetGlobalState();
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006588
6589 /* Check if the global (i.e., a function or a class) was renamed
6590 or moved to another module. */
6591 key = PyTuple_Pack(2, module_name, global_name);
6592 if (key == NULL)
6593 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08006594 item = PyDict_GetItemWithError(st->name_mapping_2to3, key);
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006595 Py_DECREF(key);
6596 if (item) {
6597 if (!PyTuple_Check(item) || PyTuple_GET_SIZE(item) != 2) {
6598 PyErr_Format(PyExc_RuntimeError,
6599 "_compat_pickle.NAME_MAPPING values should be "
6600 "2-tuples, not %.200s", Py_TYPE(item)->tp_name);
6601 return NULL;
6602 }
6603 module_name = PyTuple_GET_ITEM(item, 0);
6604 global_name = PyTuple_GET_ITEM(item, 1);
6605 if (!PyUnicode_Check(module_name) ||
6606 !PyUnicode_Check(global_name)) {
6607 PyErr_Format(PyExc_RuntimeError,
6608 "_compat_pickle.NAME_MAPPING values should be "
6609 "pairs of str, not (%.200s, %.200s)",
6610 Py_TYPE(module_name)->tp_name,
6611 Py_TYPE(global_name)->tp_name);
6612 return NULL;
6613 }
6614 }
6615 else if (PyErr_Occurred()) {
6616 return NULL;
6617 }
Serhiy Storchakabfe18242015-03-31 13:12:37 +03006618 else {
6619 /* Check if the module was renamed. */
6620 item = PyDict_GetItemWithError(st->import_mapping_2to3, module_name);
6621 if (item) {
6622 if (!PyUnicode_Check(item)) {
6623 PyErr_Format(PyExc_RuntimeError,
6624 "_compat_pickle.IMPORT_MAPPING values should be "
6625 "strings, not %.200s", Py_TYPE(item)->tp_name);
6626 return NULL;
6627 }
6628 module_name = item;
6629 }
6630 else if (PyErr_Occurred()) {
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006631 return NULL;
6632 }
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006633 }
6634 }
6635
Eric Snow3f9eee62017-09-15 16:35:20 -06006636 module = PyImport_GetModule(module_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006637 if (module == NULL) {
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006638 if (PyErr_Occurred())
6639 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006640 module = PyImport_Import(module_name);
6641 if (module == NULL)
6642 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006643 }
Eric Snow3f9eee62017-09-15 16:35:20 -06006644 global = getattribute(module, global_name, self->proto >= 4);
6645 Py_DECREF(module);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006646 return global;
6647}
6648
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +02006649/*[clinic input]
6650
6651_pickle.Unpickler.__sizeof__ -> Py_ssize_t
6652
6653Returns size in memory, in bytes.
6654[clinic start generated code]*/
6655
6656static Py_ssize_t
6657_pickle_Unpickler___sizeof___impl(UnpicklerObject *self)
6658/*[clinic end generated code: output=119d9d03ad4c7651 input=13333471fdeedf5e]*/
6659{
6660 Py_ssize_t res;
6661
Serhiy Storchaka5c4064e2015-12-19 20:05:25 +02006662 res = _PyObject_SIZE(Py_TYPE(self));
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +02006663 if (self->memo != NULL)
6664 res += self->memo_size * sizeof(PyObject *);
6665 if (self->marks != NULL)
6666 res += self->marks_size * sizeof(Py_ssize_t);
6667 if (self->input_line != NULL)
6668 res += strlen(self->input_line) + 1;
6669 if (self->encoding != NULL)
6670 res += strlen(self->encoding) + 1;
6671 if (self->errors != NULL)
6672 res += strlen(self->errors) + 1;
6673 return res;
6674}
6675
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006676static struct PyMethodDef Unpickler_methods[] = {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006677 _PICKLE_UNPICKLER_LOAD_METHODDEF
6678 _PICKLE_UNPICKLER_FIND_CLASS_METHODDEF
Serhiy Storchaka5bbd2312014-12-16 19:39:08 +02006679 _PICKLE_UNPICKLER___SIZEOF___METHODDEF
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006680 {NULL, NULL} /* sentinel */
6681};
6682
6683static void
6684Unpickler_dealloc(UnpicklerObject *self)
6685{
6686 PyObject_GC_UnTrack((PyObject *)self);
6687 Py_XDECREF(self->readline);
6688 Py_XDECREF(self->read);
Antoine Pitrou04248a82010-10-12 20:51:21 +00006689 Py_XDECREF(self->peek);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006690 Py_XDECREF(self->stack);
6691 Py_XDECREF(self->pers_func);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006692 if (self->buffer.buf != NULL) {
6693 PyBuffer_Release(&self->buffer);
6694 self->buffer.buf = NULL;
6695 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006696
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006697 _Unpickler_MemoCleanup(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006698 PyMem_Free(self->marks);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006699 PyMem_Free(self->input_line);
Victor Stinner49fc8ec2013-07-07 23:30:24 +02006700 PyMem_Free(self->encoding);
6701 PyMem_Free(self->errors);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006702
6703 Py_TYPE(self)->tp_free((PyObject *)self);
6704}
6705
6706static int
6707Unpickler_traverse(UnpicklerObject *self, visitproc visit, void *arg)
6708{
6709 Py_VISIT(self->readline);
6710 Py_VISIT(self->read);
Antoine Pitrou04248a82010-10-12 20:51:21 +00006711 Py_VISIT(self->peek);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006712 Py_VISIT(self->stack);
6713 Py_VISIT(self->pers_func);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006714 return 0;
6715}
6716
6717static int
6718Unpickler_clear(UnpicklerObject *self)
6719{
6720 Py_CLEAR(self->readline);
6721 Py_CLEAR(self->read);
Antoine Pitrou04248a82010-10-12 20:51:21 +00006722 Py_CLEAR(self->peek);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006723 Py_CLEAR(self->stack);
6724 Py_CLEAR(self->pers_func);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006725 if (self->buffer.buf != NULL) {
6726 PyBuffer_Release(&self->buffer);
6727 self->buffer.buf = NULL;
6728 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006729
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006730 _Unpickler_MemoCleanup(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006731 PyMem_Free(self->marks);
6732 self->marks = NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006733 PyMem_Free(self->input_line);
6734 self->input_line = NULL;
Victor Stinner49fc8ec2013-07-07 23:30:24 +02006735 PyMem_Free(self->encoding);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006736 self->encoding = NULL;
Victor Stinner49fc8ec2013-07-07 23:30:24 +02006737 PyMem_Free(self->errors);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006738 self->errors = NULL;
6739
6740 return 0;
6741}
6742
Larry Hastings61272b72014-01-07 12:41:53 -08006743/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006744
6745_pickle.Unpickler.__init__
6746
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006747 file: object
6748 *
6749 fix_imports: bool = True
6750 encoding: str = 'ASCII'
6751 errors: str = 'strict'
6752
6753This takes a binary file for reading a pickle data stream.
6754
6755The protocol version of the pickle is detected automatically, so no
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006756protocol argument is needed. Bytes past the pickled object's
6757representation are ignored.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006758
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006759The argument *file* must have two methods, a read() method that takes
6760an integer argument, and a readline() method that requires no
6761arguments. Both methods should return bytes. Thus *file* can be a
Martin Panter7462b6492015-11-02 03:37:02 +00006762binary file object opened for reading, an io.BytesIO object, or any
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006763other custom object that meets this interface.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006764
6765Optional keyword arguments are *fix_imports*, *encoding* and *errors*,
Martin Panter46f50722016-05-26 05:35:26 +00006766which are used to control compatibility support for pickle stream
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006767generated by Python 2. If *fix_imports* is True, pickle will try to
6768map the old Python 2 names to the new names used in Python 3. The
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006769*encoding* and *errors* tell pickle how to decode 8-bit string
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08006770instances pickled by Python 2; these default to 'ASCII' and 'strict',
6771respectively. The *encoding* can be 'bytes' to read these 8-bit
6772string instances as bytes objects.
Larry Hastings61272b72014-01-07 12:41:53 -08006773[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006774
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006775static int
Larry Hastings89964c42015-04-14 18:07:59 -04006776_pickle_Unpickler___init___impl(UnpicklerObject *self, PyObject *file,
6777 int fix_imports, const char *encoding,
6778 const char *errors)
Martin Panter46f50722016-05-26 05:35:26 +00006779/*[clinic end generated code: output=e2c8ce748edc57b0 input=f9b7da04f5f4f335]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006780{
Martin v. Löwis1c67dd92011-10-14 15:16:45 +02006781 _Py_IDENTIFIER(persistent_load);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006782
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006783 /* In case of multiple __init__() calls, clear previous content. */
6784 if (self->read != NULL)
6785 (void)Unpickler_clear(self);
6786
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006787 if (_Unpickler_SetInputStream(self, file) < 0)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006788 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006789
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006790 if (_Unpickler_SetInputEncoding(self, encoding, errors) < 0)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006791 return -1;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006792
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006793 self->fix_imports = fix_imports;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006794
Serhiy Storchaka986375e2017-11-30 22:48:31 +02006795 if (init_method_ref((PyObject *)self, &PyId_persistent_load,
6796 &self->pers_func, &self->pers_func_self) < 0)
6797 {
6798 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006799 }
6800
6801 self->stack = (Pdata *)Pdata_New();
6802 if (self->stack == NULL)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006803 return 1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006804
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006805 self->memo_size = 32;
6806 self->memo = _Unpickler_NewMemo(self->memo_size);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006807 if (self->memo == NULL)
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006808 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006809
Antoine Pitroud9dfaa92009-06-04 20:32:06 +00006810 self->proto = 0;
Alexandre Vassalotti0e7aa8c2009-04-03 04:17:41 +00006811
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006812 return 0;
6813}
6814
Larry Hastingsb7ccb202014-01-18 23:50:21 -08006815
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006816/* Define a proxy object for the Unpickler's internal memo object. This is to
6817 * avoid breaking code like:
6818 * unpickler.memo.clear()
6819 * and
6820 * unpickler.memo = saved_memo
6821 * Is this a good idea? Not really, but we don't want to break code that uses
6822 * it. Note that we don't implement the entire mapping API here. This is
6823 * intentional, as these should be treated as black-box implementation details.
6824 *
6825 * We do, however, have to implement pickling/unpickling support because of
Victor Stinner121aab42011-09-29 23:40:53 +02006826 * real-world code like cvs2svn.
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006827 */
6828
Larry Hastings61272b72014-01-07 12:41:53 -08006829/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006830_pickle.UnpicklerMemoProxy.clear
6831
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006832Remove all items from memo.
Larry Hastings61272b72014-01-07 12:41:53 -08006833[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006834
Larry Hastings3cceb382014-01-04 11:09:09 -08006835static PyObject *
6836_pickle_UnpicklerMemoProxy_clear_impl(UnpicklerMemoProxyObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08006837/*[clinic end generated code: output=d20cd43f4ba1fb1f input=b1df7c52e7afd9bd]*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006838{
6839 _Unpickler_MemoCleanup(self->unpickler);
6840 self->unpickler->memo = _Unpickler_NewMemo(self->unpickler->memo_size);
6841 if (self->unpickler->memo == NULL)
6842 return NULL;
6843 Py_RETURN_NONE;
6844}
6845
Larry Hastings61272b72014-01-07 12:41:53 -08006846/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006847_pickle.UnpicklerMemoProxy.copy
6848
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006849Copy the memo to a new object.
Larry Hastings61272b72014-01-07 12:41:53 -08006850[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006851
Larry Hastings3cceb382014-01-04 11:09:09 -08006852static PyObject *
6853_pickle_UnpicklerMemoProxy_copy_impl(UnpicklerMemoProxyObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08006854/*[clinic end generated code: output=e12af7e9bc1e4c77 input=97769247ce032c1d]*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006855{
6856 Py_ssize_t i;
6857 PyObject *new_memo = PyDict_New();
6858 if (new_memo == NULL)
6859 return NULL;
6860
6861 for (i = 0; i < self->unpickler->memo_size; i++) {
6862 int status;
6863 PyObject *key, *value;
6864
6865 value = self->unpickler->memo[i];
6866 if (value == NULL)
6867 continue;
6868
6869 key = PyLong_FromSsize_t(i);
6870 if (key == NULL)
6871 goto error;
6872 status = PyDict_SetItem(new_memo, key, value);
6873 Py_DECREF(key);
6874 if (status < 0)
6875 goto error;
6876 }
6877 return new_memo;
6878
6879error:
6880 Py_DECREF(new_memo);
6881 return NULL;
6882}
6883
Larry Hastings61272b72014-01-07 12:41:53 -08006884/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006885_pickle.UnpicklerMemoProxy.__reduce__
6886
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006887Implement pickling support.
Larry Hastings61272b72014-01-07 12:41:53 -08006888[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006889
Larry Hastings3cceb382014-01-04 11:09:09 -08006890static PyObject *
6891_pickle_UnpicklerMemoProxy___reduce___impl(UnpicklerMemoProxyObject *self)
Larry Hastings581ee362014-01-28 05:00:08 -08006892/*[clinic end generated code: output=6da34ac048d94cca input=6920862413407199]*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006893{
6894 PyObject *reduce_value;
6895 PyObject *constructor_args;
Larry Hastings3cceb382014-01-04 11:09:09 -08006896 PyObject *contents = _pickle_UnpicklerMemoProxy_copy_impl(self);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006897 if (contents == NULL)
6898 return NULL;
6899
6900 reduce_value = PyTuple_New(2);
6901 if (reduce_value == NULL) {
6902 Py_DECREF(contents);
6903 return NULL;
6904 }
6905 constructor_args = PyTuple_New(1);
6906 if (constructor_args == NULL) {
6907 Py_DECREF(contents);
6908 Py_DECREF(reduce_value);
6909 return NULL;
6910 }
6911 PyTuple_SET_ITEM(constructor_args, 0, contents);
6912 Py_INCREF((PyObject *)&PyDict_Type);
6913 PyTuple_SET_ITEM(reduce_value, 0, (PyObject *)&PyDict_Type);
6914 PyTuple_SET_ITEM(reduce_value, 1, constructor_args);
6915 return reduce_value;
6916}
6917
6918static PyMethodDef unpicklerproxy_methods[] = {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08006919 _PICKLE_UNPICKLERMEMOPROXY_CLEAR_METHODDEF
6920 _PICKLE_UNPICKLERMEMOPROXY_COPY_METHODDEF
6921 _PICKLE_UNPICKLERMEMOPROXY___REDUCE___METHODDEF
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006922 {NULL, NULL} /* sentinel */
6923};
6924
6925static void
6926UnpicklerMemoProxy_dealloc(UnpicklerMemoProxyObject *self)
6927{
6928 PyObject_GC_UnTrack(self);
6929 Py_XDECREF(self->unpickler);
6930 PyObject_GC_Del((PyObject *)self);
6931}
6932
6933static int
6934UnpicklerMemoProxy_traverse(UnpicklerMemoProxyObject *self,
6935 visitproc visit, void *arg)
6936{
6937 Py_VISIT(self->unpickler);
6938 return 0;
6939}
6940
6941static int
6942UnpicklerMemoProxy_clear(UnpicklerMemoProxyObject *self)
6943{
6944 Py_CLEAR(self->unpickler);
6945 return 0;
6946}
6947
6948static PyTypeObject UnpicklerMemoProxyType = {
6949 PyVarObject_HEAD_INIT(NULL, 0)
6950 "_pickle.UnpicklerMemoProxy", /*tp_name*/
6951 sizeof(UnpicklerMemoProxyObject), /*tp_basicsize*/
6952 0,
6953 (destructor)UnpicklerMemoProxy_dealloc, /* tp_dealloc */
6954 0, /* tp_print */
6955 0, /* tp_getattr */
6956 0, /* tp_setattr */
6957 0, /* tp_compare */
6958 0, /* tp_repr */
6959 0, /* tp_as_number */
6960 0, /* tp_as_sequence */
6961 0, /* tp_as_mapping */
Georg Brandlf038b322010-10-18 07:35:09 +00006962 PyObject_HashNotImplemented, /* tp_hash */
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00006963 0, /* tp_call */
6964 0, /* tp_str */
6965 PyObject_GenericGetAttr, /* tp_getattro */
6966 PyObject_GenericSetAttr, /* tp_setattro */
6967 0, /* tp_as_buffer */
6968 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC,
6969 0, /* tp_doc */
6970 (traverseproc)UnpicklerMemoProxy_traverse, /* tp_traverse */
6971 (inquiry)UnpicklerMemoProxy_clear, /* tp_clear */
6972 0, /* tp_richcompare */
6973 0, /* tp_weaklistoffset */
6974 0, /* tp_iter */
6975 0, /* tp_iternext */
6976 unpicklerproxy_methods, /* tp_methods */
6977};
6978
6979static PyObject *
6980UnpicklerMemoProxy_New(UnpicklerObject *unpickler)
6981{
6982 UnpicklerMemoProxyObject *self;
6983
6984 self = PyObject_GC_New(UnpicklerMemoProxyObject,
6985 &UnpicklerMemoProxyType);
6986 if (self == NULL)
6987 return NULL;
6988 Py_INCREF(unpickler);
6989 self->unpickler = unpickler;
6990 PyObject_GC_Track(self);
6991 return (PyObject *)self;
6992}
6993
6994/*****************************************************************************/
6995
6996
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00006997static PyObject *
6998Unpickler_get_memo(UnpicklerObject *self)
6999{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007000 return UnpicklerMemoProxy_New(self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007001}
7002
7003static int
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007004Unpickler_set_memo(UnpicklerObject *self, PyObject *obj)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007005{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007006 PyObject **new_memo;
7007 Py_ssize_t new_memo_size = 0;
7008 Py_ssize_t i;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007009
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007010 if (obj == NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007011 PyErr_SetString(PyExc_TypeError,
7012 "attribute deletion is not supported");
7013 return -1;
7014 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007015
7016 if (Py_TYPE(obj) == &UnpicklerMemoProxyType) {
7017 UnpicklerObject *unpickler =
7018 ((UnpicklerMemoProxyObject *)obj)->unpickler;
7019
7020 new_memo_size = unpickler->memo_size;
7021 new_memo = _Unpickler_NewMemo(new_memo_size);
7022 if (new_memo == NULL)
7023 return -1;
7024
7025 for (i = 0; i < new_memo_size; i++) {
7026 Py_XINCREF(unpickler->memo[i]);
7027 new_memo[i] = unpickler->memo[i];
7028 }
7029 }
7030 else if (PyDict_Check(obj)) {
7031 Py_ssize_t i = 0;
7032 PyObject *key, *value;
7033
Serhiy Storchaka5ab81d72016-12-16 16:18:57 +02007034 new_memo_size = PyDict_GET_SIZE(obj);
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007035 new_memo = _Unpickler_NewMemo(new_memo_size);
7036 if (new_memo == NULL)
7037 return -1;
7038
7039 while (PyDict_Next(obj, &i, &key, &value)) {
7040 Py_ssize_t idx;
7041 if (!PyLong_Check(key)) {
7042 PyErr_SetString(PyExc_TypeError,
7043 "memo key must be integers");
7044 goto error;
7045 }
7046 idx = PyLong_AsSsize_t(key);
7047 if (idx == -1 && PyErr_Occurred())
7048 goto error;
Christian Heimesa24b4d22013-07-01 15:17:45 +02007049 if (idx < 0) {
7050 PyErr_SetString(PyExc_ValueError,
Christian Heimes80878792013-07-01 15:23:39 +02007051 "memo key must be positive integers.");
Christian Heimesa24b4d22013-07-01 15:17:45 +02007052 goto error;
7053 }
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007054 if (_Unpickler_MemoPut(self, idx, value) < 0)
7055 goto error;
7056 }
7057 }
7058 else {
7059 PyErr_Format(PyExc_TypeError,
7060 "'memo' attribute must be an UnpicklerMemoProxy object"
7061 "or dict, not %.200s", Py_TYPE(obj)->tp_name);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007062 return -1;
7063 }
7064
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007065 _Unpickler_MemoCleanup(self);
7066 self->memo_size = new_memo_size;
7067 self->memo = new_memo;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007068
7069 return 0;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007070
7071 error:
7072 if (new_memo_size) {
7073 i = new_memo_size;
7074 while (--i >= 0) {
7075 Py_XDECREF(new_memo[i]);
7076 }
7077 PyMem_FREE(new_memo);
7078 }
7079 return -1;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007080}
7081
7082static PyObject *
7083Unpickler_get_persload(UnpicklerObject *self)
7084{
Serhiy Storchaka986375e2017-11-30 22:48:31 +02007085 if (self->pers_func == NULL) {
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007086 PyErr_SetString(PyExc_AttributeError, "persistent_load");
Serhiy Storchaka986375e2017-11-30 22:48:31 +02007087 return NULL;
7088 }
7089 return reconstruct_method(self->pers_func, self->pers_func_self);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007090}
7091
7092static int
7093Unpickler_set_persload(UnpicklerObject *self, PyObject *value)
7094{
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007095 if (value == NULL) {
7096 PyErr_SetString(PyExc_TypeError,
7097 "attribute deletion is not supported");
7098 return -1;
7099 }
7100 if (!PyCallable_Check(value)) {
7101 PyErr_SetString(PyExc_TypeError,
7102 "persistent_load must be a callable taking "
7103 "one argument");
7104 return -1;
7105 }
7106
Serhiy Storchaka986375e2017-11-30 22:48:31 +02007107 self->pers_func_self = NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007108 Py_INCREF(value);
Serhiy Storchakaec397562016-04-06 09:50:03 +03007109 Py_XSETREF(self->pers_func, value);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007110
7111 return 0;
7112}
7113
7114static PyGetSetDef Unpickler_getsets[] = {
7115 {"memo", (getter)Unpickler_get_memo, (setter)Unpickler_set_memo},
7116 {"persistent_load", (getter)Unpickler_get_persload,
7117 (setter)Unpickler_set_persload},
7118 {NULL}
7119};
7120
7121static PyTypeObject Unpickler_Type = {
7122 PyVarObject_HEAD_INIT(NULL, 0)
7123 "_pickle.Unpickler", /*tp_name*/
7124 sizeof(UnpicklerObject), /*tp_basicsize*/
7125 0, /*tp_itemsize*/
7126 (destructor)Unpickler_dealloc, /*tp_dealloc*/
7127 0, /*tp_print*/
7128 0, /*tp_getattr*/
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007129 0, /*tp_setattr*/
Mark Dickinsone94c6792009-02-02 20:36:42 +00007130 0, /*tp_reserved*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007131 0, /*tp_repr*/
7132 0, /*tp_as_number*/
7133 0, /*tp_as_sequence*/
7134 0, /*tp_as_mapping*/
7135 0, /*tp_hash*/
7136 0, /*tp_call*/
7137 0, /*tp_str*/
7138 0, /*tp_getattro*/
7139 0, /*tp_setattro*/
7140 0, /*tp_as_buffer*/
7141 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_BASETYPE | Py_TPFLAGS_HAVE_GC,
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007142 _pickle_Unpickler___init____doc__, /*tp_doc*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007143 (traverseproc)Unpickler_traverse, /*tp_traverse*/
7144 (inquiry)Unpickler_clear, /*tp_clear*/
7145 0, /*tp_richcompare*/
7146 0, /*tp_weaklistoffset*/
7147 0, /*tp_iter*/
7148 0, /*tp_iternext*/
7149 Unpickler_methods, /*tp_methods*/
7150 0, /*tp_members*/
7151 Unpickler_getsets, /*tp_getset*/
7152 0, /*tp_base*/
7153 0, /*tp_dict*/
7154 0, /*tp_descr_get*/
7155 0, /*tp_descr_set*/
7156 0, /*tp_dictoffset*/
Larry Hastingsb7ccb202014-01-18 23:50:21 -08007157 _pickle_Unpickler___init__, /*tp_init*/
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007158 PyType_GenericAlloc, /*tp_alloc*/
7159 PyType_GenericNew, /*tp_new*/
7160 PyObject_GC_Del, /*tp_free*/
7161 0, /*tp_is_gc*/
7162};
7163
Larry Hastings61272b72014-01-07 12:41:53 -08007164/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007165
7166_pickle.dump
7167
7168 obj: object
7169 file: object
7170 protocol: object = NULL
7171 *
7172 fix_imports: bool = True
7173
7174Write a pickled representation of obj to the open file object file.
7175
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007176This is equivalent to ``Pickler(file, protocol).dump(obj)``, but may
7177be more efficient.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007178
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007179The optional *protocol* argument tells the pickler to use the given
7180protocol supported protocols are 0, 1, 2, 3 and 4. The default
7181protocol is 3; a backward-incompatible protocol designed for Python 3.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007182
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007183Specifying a negative protocol version selects the highest protocol
7184version supported. The higher the protocol used, the more recent the
7185version of Python needed to read the pickle produced.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007186
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007187The *file* argument must have a write() method that accepts a single
7188bytes argument. It can thus be a file object opened for binary
Martin Panter7462b6492015-11-02 03:37:02 +00007189writing, an io.BytesIO instance, or any other custom object that meets
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007190this interface.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007191
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007192If *fix_imports* is True and protocol is less than 3, pickle will try
7193to map the new Python 3 names to the old module names used in Python
71942, so that the pickle data stream is readable with Python 2.
Larry Hastings61272b72014-01-07 12:41:53 -08007195[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007196
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007197static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007198_pickle_dump_impl(PyObject *module, PyObject *obj, PyObject *file,
Larry Hastings89964c42015-04-14 18:07:59 -04007199 PyObject *protocol, int fix_imports)
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007200/*[clinic end generated code: output=a4774d5fde7d34de input=830f8a64cef6f042]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007201{
7202 PicklerObject *pickler = _Pickler_New();
7203
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007204 if (pickler == NULL)
7205 return NULL;
7206
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007207 if (_Pickler_SetProtocol(pickler, protocol, fix_imports) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007208 goto error;
7209
7210 if (_Pickler_SetOutputStream(pickler, file) < 0)
7211 goto error;
7212
7213 if (dump(pickler, obj) < 0)
7214 goto error;
7215
7216 if (_Pickler_FlushToFile(pickler) < 0)
7217 goto error;
7218
7219 Py_DECREF(pickler);
7220 Py_RETURN_NONE;
7221
7222 error:
7223 Py_XDECREF(pickler);
7224 return NULL;
7225}
7226
Larry Hastings61272b72014-01-07 12:41:53 -08007227/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007228
7229_pickle.dumps
7230
7231 obj: object
7232 protocol: object = NULL
7233 *
7234 fix_imports: bool = True
7235
7236Return the pickled representation of the object as a bytes object.
7237
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007238The optional *protocol* argument tells the pickler to use the given
7239protocol; supported protocols are 0, 1, 2, 3 and 4. The default
7240protocol is 3; a backward-incompatible protocol designed for Python 3.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007241
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007242Specifying a negative protocol version selects the highest protocol
7243version supported. The higher the protocol used, the more recent the
7244version of Python needed to read the pickle produced.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007245
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007246If *fix_imports* is True and *protocol* is less than 3, pickle will
7247try to map the new Python 3 names to the old module names used in
7248Python 2, so that the pickle data stream is readable with Python 2.
Larry Hastings61272b72014-01-07 12:41:53 -08007249[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007250
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007251static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007252_pickle_dumps_impl(PyObject *module, PyObject *obj, PyObject *protocol,
Larry Hastings89964c42015-04-14 18:07:59 -04007253 int fix_imports)
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007254/*[clinic end generated code: output=d75d5cda456fd261 input=293dbeda181580b7]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007255{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007256 PyObject *result;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007257 PicklerObject *pickler = _Pickler_New();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007258
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007259 if (pickler == NULL)
7260 return NULL;
7261
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007262 if (_Pickler_SetProtocol(pickler, protocol, fix_imports) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007263 goto error;
7264
7265 if (dump(pickler, obj) < 0)
7266 goto error;
7267
7268 result = _Pickler_GetString(pickler);
7269 Py_DECREF(pickler);
7270 return result;
7271
7272 error:
7273 Py_XDECREF(pickler);
7274 return NULL;
7275}
7276
Larry Hastings61272b72014-01-07 12:41:53 -08007277/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007278
7279_pickle.load
7280
7281 file: object
7282 *
7283 fix_imports: bool = True
7284 encoding: str = 'ASCII'
7285 errors: str = 'strict'
7286
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007287Read and return an object from the pickle data stored in a file.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007288
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007289This is equivalent to ``Unpickler(file).load()``, but may be more
7290efficient.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007291
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007292The protocol version of the pickle is detected automatically, so no
7293protocol argument is needed. Bytes past the pickled object's
7294representation are ignored.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007295
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007296The argument *file* must have two methods, a read() method that takes
7297an integer argument, and a readline() method that requires no
7298arguments. Both methods should return bytes. Thus *file* can be a
Martin Panter7462b6492015-11-02 03:37:02 +00007299binary file object opened for reading, an io.BytesIO object, or any
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007300other custom object that meets this interface.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007301
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007302Optional keyword arguments are *fix_imports*, *encoding* and *errors*,
Martin Panter46f50722016-05-26 05:35:26 +00007303which are used to control compatibility support for pickle stream
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007304generated by Python 2. If *fix_imports* is True, pickle will try to
7305map the old Python 2 names to the new names used in Python 3. The
7306*encoding* and *errors* tell pickle how to decode 8-bit string
7307instances pickled by Python 2; these default to 'ASCII' and 'strict',
7308respectively. The *encoding* can be 'bytes' to read these 8-bit
7309string instances as bytes objects.
Larry Hastings61272b72014-01-07 12:41:53 -08007310[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007311
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007312static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007313_pickle_load_impl(PyObject *module, PyObject *file, int fix_imports,
Larry Hastings89964c42015-04-14 18:07:59 -04007314 const char *encoding, const char *errors)
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007315/*[clinic end generated code: output=69e298160285199e input=01b44dd3fc07afa7]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007316{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007317 PyObject *result;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007318 UnpicklerObject *unpickler = _Unpickler_New();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007319
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007320 if (unpickler == NULL)
7321 return NULL;
7322
7323 if (_Unpickler_SetInputStream(unpickler, file) < 0)
7324 goto error;
7325
7326 if (_Unpickler_SetInputEncoding(unpickler, encoding, errors) < 0)
7327 goto error;
7328
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007329 unpickler->fix_imports = fix_imports;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007330
7331 result = load(unpickler);
7332 Py_DECREF(unpickler);
7333 return result;
7334
7335 error:
7336 Py_XDECREF(unpickler);
7337 return NULL;
7338}
7339
Larry Hastings61272b72014-01-07 12:41:53 -08007340/*[clinic input]
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007341
7342_pickle.loads
7343
7344 data: object
7345 *
7346 fix_imports: bool = True
7347 encoding: str = 'ASCII'
7348 errors: str = 'strict'
7349
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007350Read and return an object from the given pickle data.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007351
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007352The protocol version of the pickle is detected automatically, so no
7353protocol argument is needed. Bytes past the pickled object's
7354representation are ignored.
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007355
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007356Optional keyword arguments are *fix_imports*, *encoding* and *errors*,
Martin Panter46f50722016-05-26 05:35:26 +00007357which are used to control compatibility support for pickle stream
Alexandre Vassalottid05c9ff2013-12-07 01:09:27 -08007358generated by Python 2. If *fix_imports* is True, pickle will try to
7359map the old Python 2 names to the new names used in Python 3. The
7360*encoding* and *errors* tell pickle how to decode 8-bit string
7361instances pickled by Python 2; these default to 'ASCII' and 'strict',
7362respectively. The *encoding* can be 'bytes' to read these 8-bit
7363string instances as bytes objects.
Larry Hastings61272b72014-01-07 12:41:53 -08007364[clinic start generated code]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007365
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007366static PyObject *
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007367_pickle_loads_impl(PyObject *module, PyObject *data, int fix_imports,
Larry Hastings89964c42015-04-14 18:07:59 -04007368 const char *encoding, const char *errors)
Serhiy Storchaka1a2b24f2016-07-07 17:35:15 +03007369/*[clinic end generated code: output=1e7cb2343f2c440f input=70605948a719feb9]*/
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007370{
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007371 PyObject *result;
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007372 UnpicklerObject *unpickler = _Unpickler_New();
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007373
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007374 if (unpickler == NULL)
7375 return NULL;
7376
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007377 if (_Unpickler_SetStringInput(unpickler, data) < 0)
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007378 goto error;
7379
7380 if (_Unpickler_SetInputEncoding(unpickler, encoding, errors) < 0)
7381 goto error;
7382
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007383 unpickler->fix_imports = fix_imports;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007384
7385 result = load(unpickler);
7386 Py_DECREF(unpickler);
7387 return result;
7388
7389 error:
7390 Py_XDECREF(unpickler);
7391 return NULL;
7392}
7393
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007394static struct PyMethodDef pickle_methods[] = {
Alexandre Vassalottied8c9062013-11-24 12:25:48 -08007395 _PICKLE_DUMP_METHODDEF
7396 _PICKLE_DUMPS_METHODDEF
7397 _PICKLE_LOAD_METHODDEF
7398 _PICKLE_LOADS_METHODDEF
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007399 {NULL, NULL} /* sentinel */
7400};
7401
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007402static int
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007403pickle_clear(PyObject *m)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007404{
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007405 _Pickle_ClearState(_Pickle_GetState(m));
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007406 return 0;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007407}
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007408
Stefan Krahf483b0f2013-12-14 13:43:10 +01007409static void
7410pickle_free(PyObject *m)
7411{
7412 _Pickle_ClearState(_Pickle_GetState(m));
7413}
7414
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007415static int
7416pickle_traverse(PyObject *m, visitproc visit, void *arg)
7417{
7418 PickleState *st = _Pickle_GetState(m);
7419 Py_VISIT(st->PickleError);
7420 Py_VISIT(st->PicklingError);
7421 Py_VISIT(st->UnpicklingError);
7422 Py_VISIT(st->dispatch_table);
7423 Py_VISIT(st->extension_registry);
7424 Py_VISIT(st->extension_cache);
7425 Py_VISIT(st->inverted_registry);
7426 Py_VISIT(st->name_mapping_2to3);
7427 Py_VISIT(st->import_mapping_2to3);
7428 Py_VISIT(st->name_mapping_3to2);
7429 Py_VISIT(st->import_mapping_3to2);
7430 Py_VISIT(st->codecs_encode);
Serhiy Storchaka58e41342015-03-31 14:07:24 +03007431 Py_VISIT(st->getattr);
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007432 return 0;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007433}
7434
7435static struct PyModuleDef _picklemodule = {
7436 PyModuleDef_HEAD_INIT,
Stefan Krahf483b0f2013-12-14 13:43:10 +01007437 "_pickle", /* m_name */
7438 pickle_module_doc, /* m_doc */
7439 sizeof(PickleState), /* m_size */
7440 pickle_methods, /* m_methods */
7441 NULL, /* m_reload */
7442 pickle_traverse, /* m_traverse */
7443 pickle_clear, /* m_clear */
7444 (freefunc)pickle_free /* m_free */
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007445};
7446
7447PyMODINIT_FUNC
7448PyInit__pickle(void)
7449{
7450 PyObject *m;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007451 PickleState *st;
7452
7453 m = PyState_FindModule(&_picklemodule);
7454 if (m) {
7455 Py_INCREF(m);
7456 return m;
7457 }
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007458
7459 if (PyType_Ready(&Unpickler_Type) < 0)
7460 return NULL;
7461 if (PyType_Ready(&Pickler_Type) < 0)
7462 return NULL;
7463 if (PyType_Ready(&Pdata_Type) < 0)
7464 return NULL;
Antoine Pitrouea99c5c2010-09-09 18:33:21 +00007465 if (PyType_Ready(&PicklerMemoProxyType) < 0)
7466 return NULL;
7467 if (PyType_Ready(&UnpicklerMemoProxyType) < 0)
7468 return NULL;
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007469
7470 /* Create the module and add the functions. */
7471 m = PyModule_Create(&_picklemodule);
7472 if (m == NULL)
7473 return NULL;
7474
Antoine Pitrou8391cf42011-07-15 21:01:21 +02007475 Py_INCREF(&Pickler_Type);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007476 if (PyModule_AddObject(m, "Pickler", (PyObject *)&Pickler_Type) < 0)
7477 return NULL;
Antoine Pitrou8391cf42011-07-15 21:01:21 +02007478 Py_INCREF(&Unpickler_Type);
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007479 if (PyModule_AddObject(m, "Unpickler", (PyObject *)&Unpickler_Type) < 0)
7480 return NULL;
7481
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007482 st = _Pickle_GetState(m);
7483
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007484 /* Initialize the exceptions. */
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007485 st->PickleError = PyErr_NewException("_pickle.PickleError", NULL, NULL);
7486 if (st->PickleError == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007487 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007488 st->PicklingError = \
7489 PyErr_NewException("_pickle.PicklingError", st->PickleError, NULL);
7490 if (st->PicklingError == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007491 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007492 st->UnpicklingError = \
7493 PyErr_NewException("_pickle.UnpicklingError", st->PickleError, NULL);
7494 if (st->UnpicklingError == NULL)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007495 return NULL;
7496
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007497 Py_INCREF(st->PickleError);
7498 if (PyModule_AddObject(m, "PickleError", st->PickleError) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007499 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007500 Py_INCREF(st->PicklingError);
7501 if (PyModule_AddObject(m, "PicklingError", st->PicklingError) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007502 return NULL;
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007503 Py_INCREF(st->UnpicklingError);
7504 if (PyModule_AddObject(m, "UnpicklingError", st->UnpicklingError) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007505 return NULL;
7506
Alexandre Vassalotti23bdd832013-11-27 19:36:52 -08007507 if (_Pickle_InitState(st) < 0)
Alexandre Vassalottica2d6102008-06-12 18:26:05 +00007508 return NULL;
7509
7510 return m;
7511}