blob: 74ecafd3d65fe08dc71669753b5456d03012bd08 [file] [log] [blame]
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001/*
2 * multibytecodec.c: Common Multibyte Codec Implementation
3 *
4 * Written by Hye-Shik Chang <perky@FreeBSD.org>
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00005 */
6
Hye-Shik Chang4b96c132006-03-04 16:08:19 +00007#define PY_SSIZE_T_CLEAN
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00008#include "Python.h"
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00009#include "structmember.h"
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000010#include "multibytecodec.h"
Brett Cannonf2de1fc2014-08-22 11:45:03 -040011#include "clinic/multibytecodec.c.h"
12
13/*[clinic input]
14output preset file
15module _multibytecodec
16[clinic start generated code]*/
17/*[clinic end generated code: output=da39a3ee5e6b4b0d input=e0cf1b7f3c472d17]*/
18
19/*[clinic input]
20class _multibytecodec.MultibyteCodec "MultibyteCodecObject *" "&MultibyteCodec_Type"
21[clinic start generated code]*/
22/*[clinic end generated code: output=da39a3ee5e6b4b0d input=d5b1fc1fec8eb003]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000023
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000024typedef struct {
Victor Stinnerd9491262013-04-14 02:06:32 +020025 PyObject *inobj;
26 Py_ssize_t inpos, inlen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000027 unsigned char *outbuf, *outbuf_end;
28 PyObject *excobj, *outobj;
29} MultibyteEncodeBuffer;
30
31typedef struct {
32 const unsigned char *inbuf, *inbuf_top, *inbuf_end;
Victor Stinnera0dd0212013-04-11 22:09:04 +020033 PyObject *excobj;
34 _PyUnicodeWriter writer;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000035} MultibyteDecodeBuffer;
36
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000037static char *incnewkwarglist[] = {"errors", NULL};
Martin v. Löwis02cbf4a2006-02-27 17:20:04 +000038static char *streamkwarglist[] = {"stream", "errors", NULL};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000039
40static PyObject *multibytecodec_encode(MultibyteCodec *,
Victor Stinnerd9491262013-04-14 02:06:32 +020041 MultibyteCodec_State *, PyObject *, Py_ssize_t *,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000042 PyObject *, int);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000043
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000044#define MBENC_RESET MBENC_MAX<<1 /* reset after an encoding session */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000045
Victor Stinner3f36a572013-11-12 21:39:02 +010046_Py_IDENTIFIER(write);
47
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000048static PyObject *
Hye-Shik Chang4b96c132006-03-04 16:08:19 +000049make_tuple(PyObject *object, Py_ssize_t len)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000050{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000051 PyObject *v, *w;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000052
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000053 if (object == NULL)
54 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000055
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000056 v = PyTuple_New(2);
57 if (v == NULL) {
58 Py_DECREF(object);
59 return NULL;
60 }
61 PyTuple_SET_ITEM(v, 0, object);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000062
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000063 w = PyLong_FromSsize_t(len);
64 if (w == NULL) {
65 Py_DECREF(v);
66 return NULL;
67 }
68 PyTuple_SET_ITEM(v, 1, w);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000069
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000070 return v;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000071}
72
73static PyObject *
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000074internal_error_callback(const char *errors)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000075{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000076 if (errors == NULL || strcmp(errors, "strict") == 0)
77 return ERROR_STRICT;
78 else if (strcmp(errors, "ignore") == 0)
79 return ERROR_IGNORE;
80 else if (strcmp(errors, "replace") == 0)
81 return ERROR_REPLACE;
82 else
83 return PyUnicode_FromString(errors);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000084}
85
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000086static PyObject *
87call_error_callback(PyObject *errors, PyObject *exc)
88{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000089 PyObject *args, *cb, *r;
90 const char *str;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000091
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000092 assert(PyUnicode_Check(errors));
93 str = _PyUnicode_AsString(errors);
94 if (str == NULL)
95 return NULL;
96 cb = PyCodec_LookupError(str);
97 if (cb == NULL)
98 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000099
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000100 args = PyTuple_New(1);
101 if (args == NULL) {
102 Py_DECREF(cb);
103 return NULL;
104 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000105
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000106 PyTuple_SET_ITEM(args, 0, exc);
107 Py_INCREF(exc);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000108
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000109 r = PyObject_CallObject(cb, args);
110 Py_DECREF(args);
111 Py_DECREF(cb);
112 return r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000113}
114
115static PyObject *
116codecctx_errors_get(MultibyteStatefulCodecContext *self)
117{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000118 const char *errors;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000119
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000120 if (self->errors == ERROR_STRICT)
121 errors = "strict";
122 else if (self->errors == ERROR_IGNORE)
123 errors = "ignore";
124 else if (self->errors == ERROR_REPLACE)
125 errors = "replace";
126 else {
127 Py_INCREF(self->errors);
128 return self->errors;
129 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000130
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000131 return PyUnicode_FromString(errors);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000132}
133
134static int
135codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000136 void *closure)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000137{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000138 PyObject *cb;
139 const char *str;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000140
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000141 if (!PyUnicode_Check(value)) {
142 PyErr_SetString(PyExc_TypeError, "errors must be a string");
143 return -1;
144 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000145
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000146 str = _PyUnicode_AsString(value);
147 if (str == NULL)
148 return -1;
Neal Norwitz6ea45d32007-08-26 04:19:43 +0000149
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000150 cb = internal_error_callback(str);
151 if (cb == NULL)
152 return -1;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000153
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000154 ERROR_DECREF(self->errors);
155 self->errors = cb;
156 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000157}
158
159/* This getset handlers list is used by all the stateful codec objects */
160static PyGetSetDef codecctx_getsets[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000161 {"errors", (getter)codecctx_errors_get,
162 (setter)codecctx_errors_set,
163 PyDoc_STR("how to treat errors")},
164 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000165};
166
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000167static int
Hye-Shik Chang4b96c132006-03-04 16:08:19 +0000168expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000169{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000170 Py_ssize_t orgpos, orgsize, incsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000171
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000172 orgpos = (Py_ssize_t)((char *)buf->outbuf -
173 PyBytes_AS_STRING(buf->outobj));
174 orgsize = PyBytes_GET_SIZE(buf->outobj);
175 incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize);
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000176
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000177 if (orgsize > PY_SSIZE_T_MAX - incsize)
178 return -1;
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000179
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000180 if (_PyBytes_Resize(&buf->outobj, orgsize + incsize) == -1)
181 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000182
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000183 buf->outbuf = (unsigned char *)PyBytes_AS_STRING(buf->outobj) +orgpos;
184 buf->outbuf_end = (unsigned char *)PyBytes_AS_STRING(buf->outobj)
185 + PyBytes_GET_SIZE(buf->outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000186
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000187 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000188}
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000189#define REQUIRE_ENCODEBUFFER(buf, s) { \
190 if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end) \
191 if (expand_encodebuffer(buf, s) == -1) \
192 goto errorexit; \
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000193}
194
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000195
196/**
197 * MultibyteCodec object
198 */
199
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000200static int
201multibytecodec_encerror(MultibyteCodec *codec,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000202 MultibyteCodec_State *state,
203 MultibyteEncodeBuffer *buf,
204 PyObject *errors, Py_ssize_t e)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000205{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000206 PyObject *retobj = NULL, *retstr = NULL, *tobj;
207 Py_ssize_t retstrsize, newpos;
208 Py_ssize_t esize, start, end;
209 const char *reason;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000210
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000211 if (e > 0) {
212 reason = "illegal multibyte sequence";
213 esize = e;
214 }
215 else {
216 switch (e) {
217 case MBERR_TOOSMALL:
218 REQUIRE_ENCODEBUFFER(buf, -1);
219 return 0; /* retry it */
220 case MBERR_TOOFEW:
221 reason = "incomplete multibyte sequence";
Victor Stinnerd9491262013-04-14 02:06:32 +0200222 esize = (Py_ssize_t)buf->inpos;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000223 break;
224 case MBERR_INTERNAL:
225 PyErr_SetString(PyExc_RuntimeError,
226 "internal codec error");
227 return -1;
228 default:
229 PyErr_SetString(PyExc_RuntimeError,
230 "unknown runtime error");
231 return -1;
232 }
233 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000234
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000235 if (errors == ERROR_REPLACE) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200236 PyObject *replchar;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000237 Py_ssize_t r;
Victor Stinnerd9491262013-04-14 02:06:32 +0200238 Py_ssize_t inpos;
239 int kind;
240 void *data;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000241
Victor Stinnerd9491262013-04-14 02:06:32 +0200242 replchar = PyUnicode_FromOrdinal('?');
243 if (replchar == NULL)
244 goto errorexit;
245 kind = PyUnicode_KIND(replchar);
246 data = PyUnicode_DATA(replchar);
247
248 inpos = 0;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000249 for (;;) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200250 Py_ssize_t outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000251
Victor Stinnerd9491262013-04-14 02:06:32 +0200252 r = codec->encode(state, codec->config,
253 kind, data, &inpos, 1,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000254 &buf->outbuf, outleft, 0);
255 if (r == MBERR_TOOSMALL) {
256 REQUIRE_ENCODEBUFFER(buf, -1);
257 continue;
258 }
259 else
260 break;
261 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000262
Victor Stinnerd9491262013-04-14 02:06:32 +0200263 Py_DECREF(replchar);
264
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000265 if (r != 0) {
266 REQUIRE_ENCODEBUFFER(buf, 1);
267 *buf->outbuf++ = '?';
268 }
269 }
270 if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200271 buf->inpos += esize;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000272 return 0;
273 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000274
Victor Stinnerd9491262013-04-14 02:06:32 +0200275 start = (Py_ssize_t)buf->inpos;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000276 end = start + esize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000277
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000278 /* use cached exception object if available */
279 if (buf->excobj == NULL) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200280 buf->excobj = PyObject_CallFunction(PyExc_UnicodeEncodeError,
281 "sOnns",
282 codec->encoding, buf->inobj,
283 start, end, reason);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000284 if (buf->excobj == NULL)
285 goto errorexit;
286 }
287 else
288 if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 ||
289 PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 ||
290 PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0)
291 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000292
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000293 if (errors == ERROR_STRICT) {
294 PyCodec_StrictErrors(buf->excobj);
295 goto errorexit;
296 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000297
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000298 retobj = call_error_callback(errors, buf->excobj);
299 if (retobj == NULL)
300 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000301
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000302 if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
Benjamin Peterson47a00f32012-12-02 11:20:28 -0500303 (!PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) && !PyBytes_Check(tobj)) ||
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000304 !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) {
305 PyErr_SetString(PyExc_TypeError,
306 "encoding error handler must return "
Benjamin Petersonaff47232012-12-02 10:53:41 -0500307 "(str, int) tuple");
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000308 goto errorexit;
309 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000310
Benjamin Peterson47a00f32012-12-02 11:20:28 -0500311 if (PyUnicode_Check(tobj)) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200312 Py_ssize_t inpos;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000313
Victor Stinnerd9491262013-04-14 02:06:32 +0200314 retstr = multibytecodec_encode(codec, state, tobj,
315 &inpos, ERROR_STRICT,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000316 MBENC_FLUSH);
317 if (retstr == NULL)
318 goto errorexit;
319 }
Benjamin Peterson47a00f32012-12-02 11:20:28 -0500320 else {
321 Py_INCREF(tobj);
322 retstr = tobj;
323 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000324
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000325 assert(PyBytes_Check(retstr));
326 retstrsize = PyBytes_GET_SIZE(retstr);
327 REQUIRE_ENCODEBUFFER(buf, retstrsize);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000328
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000329 memcpy(buf->outbuf, PyBytes_AS_STRING(retstr), retstrsize);
330 buf->outbuf += retstrsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000331
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000332 newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
333 if (newpos < 0 && !PyErr_Occurred())
Victor Stinnerd9491262013-04-14 02:06:32 +0200334 newpos += (Py_ssize_t)buf->inlen;
335 if (newpos < 0 || newpos > buf->inlen) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000336 PyErr_Clear();
337 PyErr_Format(PyExc_IndexError,
338 "position %zd from error handler out of bounds",
339 newpos);
340 goto errorexit;
341 }
Victor Stinnerd9491262013-04-14 02:06:32 +0200342 buf->inpos = newpos;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000343
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000344 Py_DECREF(retobj);
345 Py_DECREF(retstr);
346 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000347
348errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000349 Py_XDECREF(retobj);
350 Py_XDECREF(retstr);
351 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000352}
353
354static int
355multibytecodec_decerror(MultibyteCodec *codec,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000356 MultibyteCodec_State *state,
357 MultibyteDecodeBuffer *buf,
358 PyObject *errors, Py_ssize_t e)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000359{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000360 PyObject *retobj = NULL, *retuni = NULL;
Victor Stinnera0dd0212013-04-11 22:09:04 +0200361 Py_ssize_t newpos;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000362 const char *reason;
363 Py_ssize_t esize, start, end;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000364
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000365 if (e > 0) {
366 reason = "illegal multibyte sequence";
367 esize = e;
368 }
369 else {
370 switch (e) {
371 case MBERR_TOOSMALL:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000372 return 0; /* retry it */
373 case MBERR_TOOFEW:
374 reason = "incomplete multibyte sequence";
375 esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
376 break;
377 case MBERR_INTERNAL:
378 PyErr_SetString(PyExc_RuntimeError,
379 "internal codec error");
380 return -1;
Victor Stinnerd1f99422013-07-16 21:41:43 +0200381 case MBERR_EXCEPTION:
382 return -1;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000383 default:
384 PyErr_SetString(PyExc_RuntimeError,
385 "unknown runtime error");
386 return -1;
387 }
388 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000389
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000390 if (errors == ERROR_REPLACE) {
Victor Stinnera0dd0212013-04-11 22:09:04 +0200391 if (_PyUnicodeWriter_WriteChar(&buf->writer,
392 Py_UNICODE_REPLACEMENT_CHARACTER) < 0)
393 goto errorexit;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000394 }
395 if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
396 buf->inbuf += esize;
397 return 0;
398 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000399
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000400 start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
401 end = start + esize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000402
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000403 /* use cached exception object if available */
404 if (buf->excobj == NULL) {
405 buf->excobj = PyUnicodeDecodeError_Create(codec->encoding,
406 (const char *)buf->inbuf_top,
407 (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top),
408 start, end, reason);
409 if (buf->excobj == NULL)
410 goto errorexit;
411 }
412 else
413 if (PyUnicodeDecodeError_SetStart(buf->excobj, start) ||
414 PyUnicodeDecodeError_SetEnd(buf->excobj, end) ||
415 PyUnicodeDecodeError_SetReason(buf->excobj, reason))
416 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000417
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000418 if (errors == ERROR_STRICT) {
419 PyCodec_StrictErrors(buf->excobj);
420 goto errorexit;
421 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000422
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000423 retobj = call_error_callback(errors, buf->excobj);
424 if (retobj == NULL)
425 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000426
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000427 if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
428 !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) ||
429 !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) {
430 PyErr_SetString(PyExc_TypeError,
431 "decoding error handler must return "
Benjamin Petersonaff47232012-12-02 10:53:41 -0500432 "(str, int) tuple");
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000433 goto errorexit;
434 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000435
Victor Stinnera0dd0212013-04-11 22:09:04 +0200436 if (_PyUnicodeWriter_WriteStr(&buf->writer, retuni) < 0)
Victor Stinner4eea8492011-11-21 03:01:27 +0100437 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000438
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000439 newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
440 if (newpos < 0 && !PyErr_Occurred())
441 newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
442 if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
443 PyErr_Clear();
444 PyErr_Format(PyExc_IndexError,
445 "position %zd from error handler out of bounds",
446 newpos);
447 goto errorexit;
448 }
449 buf->inbuf = buf->inbuf_top + newpos;
450 Py_DECREF(retobj);
451 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000452
453errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000454 Py_XDECREF(retobj);
455 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000456}
457
458static PyObject *
459multibytecodec_encode(MultibyteCodec *codec,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000460 MultibyteCodec_State *state,
Victor Stinnerd9491262013-04-14 02:06:32 +0200461 PyObject *text, Py_ssize_t *inpos_t,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000462 PyObject *errors, int flags)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000463{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000464 MultibyteEncodeBuffer buf;
465 Py_ssize_t finalsize, r = 0;
Victor Stinnerd9491262013-04-14 02:06:32 +0200466 Py_ssize_t datalen;
467 int kind;
468 void *data;
469
470 if (PyUnicode_READY(text) < 0)
471 return NULL;
472 datalen = PyUnicode_GET_LENGTH(text);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000473
Victor Stinner6bcbef72011-05-24 22:17:55 +0200474 if (datalen == 0 && !(flags & MBENC_RESET))
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000475 return PyBytes_FromStringAndSize(NULL, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000476
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000477 buf.excobj = NULL;
Brett Cannonb94767f2011-02-22 20:15:44 +0000478 buf.outobj = NULL;
Victor Stinnerd9491262013-04-14 02:06:32 +0200479 buf.inobj = text; /* borrowed reference */
480 buf.inpos = 0;
481 buf.inlen = datalen;
482 kind = PyUnicode_KIND(buf.inobj);
483 data = PyUnicode_DATA(buf.inobj);
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000484
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000485 if (datalen > (PY_SSIZE_T_MAX - 16) / 2) {
486 PyErr_NoMemory();
487 goto errorexit;
488 }
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000489
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000490 buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16);
491 if (buf.outobj == NULL)
492 goto errorexit;
493 buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj);
494 buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000495
Victor Stinnerd9491262013-04-14 02:06:32 +0200496 while (buf.inpos < buf.inlen) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000497 /* we don't reuse inleft and outleft here.
498 * error callbacks can relocate the cursor anywhere on buffer*/
Victor Stinnerd9491262013-04-14 02:06:32 +0200499 Py_ssize_t outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
500
501 r = codec->encode(state, codec->config,
502 kind, data,
503 &buf.inpos, buf.inlen,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000504 &buf.outbuf, outleft, flags);
505 if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH)))
506 break;
507 else if (multibytecodec_encerror(codec, state, &buf, errors,r))
508 goto errorexit;
509 else if (r == MBERR_TOOFEW)
510 break;
511 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000512
Victor Stinner6bcbef72011-05-24 22:17:55 +0200513 if (codec->encreset != NULL && (flags & MBENC_RESET))
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000514 for (;;) {
515 Py_ssize_t outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000516
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000517 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
518 r = codec->encreset(state, codec->config, &buf.outbuf,
519 outleft);
520 if (r == 0)
521 break;
522 else if (multibytecodec_encerror(codec, state,
523 &buf, errors, r))
524 goto errorexit;
525 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000526
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000527 finalsize = (Py_ssize_t)((char *)buf.outbuf -
528 PyBytes_AS_STRING(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000529
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000530 if (finalsize != PyBytes_GET_SIZE(buf.outobj))
531 if (_PyBytes_Resize(&buf.outobj, finalsize) == -1)
532 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000533
Victor Stinnerd9491262013-04-14 02:06:32 +0200534 if (inpos_t)
535 *inpos_t = buf.inpos;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000536 Py_XDECREF(buf.excobj);
537 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000538
539errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000540 Py_XDECREF(buf.excobj);
541 Py_XDECREF(buf.outobj);
542 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000543}
544
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400545/*[clinic input]
546_multibytecodec.MultibyteCodec.encode
547
548 input: object
549 errors: str(nullable=True) = NULL
550
551Return an encoded string version of `input'.
552
553'errors' may be given to set a different error handling scheme. Default is
554'strict' meaning that encoding errors raise a UnicodeEncodeError. Other possible
555values are 'ignore', 'replace' and 'xmlcharrefreplace' as well as any other name
556registered with codecs.register_error that can handle UnicodeEncodeErrors.
557[clinic start generated code]*/
558
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000559static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400560_multibytecodec_MultibyteCodec_encode_impl(MultibyteCodecObject *self, PyObject *input, const char *errors)
561/*[clinic end generated code: output=a36bfa08783a0d0b input=252e7ee695867b2d]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000562{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000563 MultibyteCodec_State state;
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400564 PyObject *errorcb, *r, *ucvt;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000565 Py_ssize_t datalen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000566
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400567 if (PyUnicode_Check(input))
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000568 ucvt = NULL;
569 else {
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400570 input = ucvt = PyObject_Str(input);
571 if (input == NULL)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000572 return NULL;
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400573 else if (!PyUnicode_Check(input)) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000574 PyErr_SetString(PyExc_TypeError,
575 "couldn't convert the object to unicode.");
576 Py_DECREF(ucvt);
577 return NULL;
578 }
579 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000580
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400581 if (PyUnicode_READY(input) < 0) {
Victor Stinner9a80fab2011-11-21 02:50:14 +0100582 Py_XDECREF(ucvt);
583 return NULL;
584 }
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400585 datalen = PyUnicode_GET_LENGTH(input);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000586
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000587 errorcb = internal_error_callback(errors);
588 if (errorcb == NULL) {
589 Py_XDECREF(ucvt);
590 return NULL;
591 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000592
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000593 if (self->codec->encinit != NULL &&
594 self->codec->encinit(&state, self->codec->config) != 0)
595 goto errorexit;
596 r = multibytecodec_encode(self->codec, &state,
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400597 input, NULL, errorcb,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000598 MBENC_FLUSH | MBENC_RESET);
599 if (r == NULL)
600 goto errorexit;
Hye-Shik Chang2bb146f2004-07-18 03:06:29 +0000601
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000602 ERROR_DECREF(errorcb);
603 Py_XDECREF(ucvt);
604 return make_tuple(r, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000605
606errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000607 ERROR_DECREF(errorcb);
608 Py_XDECREF(ucvt);
609 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000610}
611
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400612/*[clinic input]
613_multibytecodec.MultibyteCodec.decode
614
615 input: Py_buffer
616 errors: str(nullable=True) = NULL
617
618Decodes 'input'.
619
620'errors' may be given to set a different error handling scheme. Default is
621'strict' meaning that encoding errors raise a UnicodeDecodeError. Other possible
622values are 'ignore' and 'replace' as well as any other name registered with
623codecs.register_error that is able to handle UnicodeDecodeErrors."
624[clinic start generated code]*/
625
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000626static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400627_multibytecodec_MultibyteCodec_decode_impl(MultibyteCodecObject *self, Py_buffer *input, const char *errors)
628/*[clinic end generated code: output=4c8ee8b2931b014e input=37e1d9236e3ce8f3]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000629{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000630 MultibyteCodec_State state;
631 MultibyteDecodeBuffer buf;
Victor Stinnera0dd0212013-04-11 22:09:04 +0200632 PyObject *errorcb, *res;
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400633 const char *data;
Victor Stinnera0dd0212013-04-11 22:09:04 +0200634 Py_ssize_t datalen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000635
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400636 data = input->buf;
637 datalen = input->len;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000638
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000639 errorcb = internal_error_callback(errors);
640 if (errorcb == NULL) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000641 return NULL;
642 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000643
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000644 if (datalen == 0) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000645 ERROR_DECREF(errorcb);
Victor Stinnerb37b1742011-12-01 03:18:59 +0100646 return make_tuple(PyUnicode_New(0, 0), 0);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000647 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000648
Victor Stinner8f674cc2013-04-17 23:02:17 +0200649 _PyUnicodeWriter_Init(&buf.writer);
650 buf.writer.min_length = datalen;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000651 buf.excobj = NULL;
652 buf.inbuf = buf.inbuf_top = (unsigned char *)data;
653 buf.inbuf_end = buf.inbuf_top + datalen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000654
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000655 if (self->codec->decinit != NULL &&
656 self->codec->decinit(&state, self->codec->config) != 0)
657 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000658
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000659 while (buf.inbuf < buf.inbuf_end) {
Victor Stinnera0dd0212013-04-11 22:09:04 +0200660 Py_ssize_t inleft, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000661
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000662 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000663
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000664 r = self->codec->decode(&state, self->codec->config,
Victor Stinnera0dd0212013-04-11 22:09:04 +0200665 &buf.inbuf, inleft, &buf.writer);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000666 if (r == 0)
667 break;
668 else if (multibytecodec_decerror(self->codec, &state,
669 &buf, errorcb, r))
670 goto errorexit;
671 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000672
Victor Stinnera0dd0212013-04-11 22:09:04 +0200673 res = _PyUnicodeWriter_Finish(&buf.writer);
674 if (res == NULL)
675 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000676
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000677 Py_XDECREF(buf.excobj);
678 ERROR_DECREF(errorcb);
Victor Stinnera0dd0212013-04-11 22:09:04 +0200679 return make_tuple(res, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000680
681errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000682 ERROR_DECREF(errorcb);
683 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +0200684 _PyUnicodeWriter_Dealloc(&buf.writer);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000685
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000686 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000687}
688
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000689static struct PyMethodDef multibytecodec_methods[] = {
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400690 _MULTIBYTECODEC_MULTIBYTECODEC_ENCODE_METHODDEF
691 _MULTIBYTECODEC_MULTIBYTECODEC_DECODE_METHODDEF
692 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000693};
694
695static void
696multibytecodec_dealloc(MultibyteCodecObject *self)
697{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000698 PyObject_Del(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000699}
700
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000701static PyTypeObject MultibyteCodec_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000702 PyVarObject_HEAD_INIT(NULL, 0)
703 "MultibyteCodec", /* tp_name */
704 sizeof(MultibyteCodecObject), /* tp_basicsize */
705 0, /* tp_itemsize */
706 /* methods */
707 (destructor)multibytecodec_dealloc, /* tp_dealloc */
708 0, /* tp_print */
709 0, /* tp_getattr */
710 0, /* tp_setattr */
711 0, /* tp_reserved */
712 0, /* tp_repr */
713 0, /* tp_as_number */
714 0, /* tp_as_sequence */
715 0, /* tp_as_mapping */
716 0, /* tp_hash */
717 0, /* tp_call */
718 0, /* tp_str */
719 PyObject_GenericGetAttr, /* tp_getattro */
720 0, /* tp_setattro */
721 0, /* tp_as_buffer */
722 Py_TPFLAGS_DEFAULT, /* tp_flags */
723 0, /* tp_doc */
724 0, /* tp_traverse */
725 0, /* tp_clear */
726 0, /* tp_richcompare */
727 0, /* tp_weaklistoffset */
728 0, /* tp_iter */
729 0, /* tp_iterext */
730 multibytecodec_methods, /* tp_methods */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000731};
732
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000733
734/**
735 * Utility functions for stateful codec mechanism
736 */
737
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000738#define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o))
739#define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o))
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000740
741static PyObject *
742encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000743 PyObject *unistr, int final)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000744{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000745 PyObject *ucvt, *r = NULL;
Victor Stinnerd9491262013-04-14 02:06:32 +0200746 PyObject *inbuf = NULL;
747 Py_ssize_t inpos, datalen;
748 PyObject *origpending = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000749
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000750 if (PyUnicode_Check(unistr))
751 ucvt = NULL;
752 else {
753 unistr = ucvt = PyObject_Str(unistr);
754 if (unistr == NULL)
755 return NULL;
756 else if (!PyUnicode_Check(unistr)) {
757 PyErr_SetString(PyExc_TypeError,
Benjamin Petersonaff47232012-12-02 10:53:41 -0500758 "couldn't convert the object to str.");
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000759 Py_DECREF(ucvt);
760 return NULL;
761 }
762 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000763
Victor Stinnerd9491262013-04-14 02:06:32 +0200764 if (ctx->pending) {
765 PyObject *inbuf_tmp;
766
767 Py_INCREF(ctx->pending);
768 origpending = ctx->pending;
769
770 Py_INCREF(ctx->pending);
771 inbuf_tmp = ctx->pending;
772 PyUnicode_Append(&inbuf_tmp, unistr);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000773 if (inbuf_tmp == NULL)
774 goto errorexit;
Victor Stinnerd9491262013-04-14 02:06:32 +0200775 Py_CLEAR(ctx->pending);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000776 inbuf = inbuf_tmp;
777 }
Victor Stinnerd9491262013-04-14 02:06:32 +0200778 else {
779 origpending = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000780
Victor Stinnerd9491262013-04-14 02:06:32 +0200781 Py_INCREF(unistr);
782 inbuf = unistr;
783 }
784 if (PyUnicode_READY(inbuf) < 0)
785 goto errorexit;
786 inpos = 0;
787 datalen = PyUnicode_GET_LENGTH(inbuf);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000788
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000789 r = multibytecodec_encode(ctx->codec, &ctx->state,
Victor Stinnerd9491262013-04-14 02:06:32 +0200790 inbuf, &inpos,
791 ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000792 if (r == NULL) {
793 /* recover the original pending buffer */
Victor Stinnerd9491262013-04-14 02:06:32 +0200794 Py_CLEAR(ctx->pending);
795 ctx->pending = origpending;
796 origpending = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000797 goto errorexit;
798 }
Victor Stinner322cc742013-04-14 18:11:41 +0200799 Py_XDECREF(origpending);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000800
Victor Stinnerd9491262013-04-14 02:06:32 +0200801 if (inpos < datalen) {
802 if (datalen - inpos > MAXENCPENDING) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000803 /* normal codecs can't reach here */
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000804 PyErr_SetString(PyExc_UnicodeError,
805 "pending buffer overflow");
806 goto errorexit;
807 }
Victor Stinnerd9491262013-04-14 02:06:32 +0200808 ctx->pending = PyUnicode_Substring(inbuf, inpos, datalen);
809 if (ctx->pending == NULL) {
810 /* normal codecs can't reach here */
811 goto errorexit;
812 }
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000813 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000814
Victor Stinner322cc742013-04-14 18:11:41 +0200815 Py_DECREF(inbuf);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000816 Py_XDECREF(ucvt);
817 return r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000818
819errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000820 Py_XDECREF(r);
821 Py_XDECREF(ucvt);
Victor Stinnerd9491262013-04-14 02:06:32 +0200822 Py_XDECREF(origpending);
Victor Stinner322cc742013-04-14 18:11:41 +0200823 Py_XDECREF(inbuf);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000824 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000825}
826
827static int
828decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000829 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000830{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000831 Py_ssize_t npendings;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000832
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000833 npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
834 if (npendings + ctx->pendingsize > MAXDECPENDING ||
835 npendings > PY_SSIZE_T_MAX - ctx->pendingsize) {
836 PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow");
837 return -1;
838 }
839 memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings);
840 ctx->pendingsize += npendings;
841 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000842}
843
844static int
845decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000846 Py_ssize_t size)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000847{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000848 buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
849 buf->inbuf_end = buf->inbuf_top + size;
Victor Stinner8f674cc2013-04-17 23:02:17 +0200850 buf->writer.min_length += size;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000851 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000852}
853
854static int
855decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000856 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000857{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000858 while (buf->inbuf < buf->inbuf_end) {
Victor Stinnera0dd0212013-04-11 22:09:04 +0200859 Py_ssize_t inleft;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000860 Py_ssize_t r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000861
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000862 inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000863
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000864 r = ctx->codec->decode(&ctx->state, ctx->codec->config,
Victor Stinnera0dd0212013-04-11 22:09:04 +0200865 &buf->inbuf, inleft, &buf->writer);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000866 if (r == 0 || r == MBERR_TOOFEW)
867 break;
868 else if (multibytecodec_decerror(ctx->codec, &ctx->state,
869 buf, ctx->errors, r))
870 return -1;
871 }
872 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000873}
874
875
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400876/*[clinic input]
877 class _multibytecodec.MultibyteIncrementalEncoder "MultibyteIncrementalEncoderObject *" "&MultibyteIncrementalEncoder_Type"
878[clinic start generated code]*/
879/*[clinic end generated code: output=da39a3ee5e6b4b0d input=3be82909cd08924d]*/
880
881/*[clinic input]
882_multibytecodec.MultibyteIncrementalEncoder.encode
883
884 input: object
885 final: int = 0
886[clinic start generated code]*/
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000887
888static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400889_multibytecodec_MultibyteIncrementalEncoder_encode_impl(MultibyteIncrementalEncoderObject *self, PyObject *input, int final)
890/*[clinic end generated code: output=3cd8780c8a719bbf input=456b76d73e464661]*/
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000891{
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400892 return encoder_encode_stateful(STATEFUL_ECTX(self), input, final);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000893}
894
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400895/*[clinic input]
896_multibytecodec.MultibyteIncrementalEncoder.reset
897[clinic start generated code]*/
898
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000899static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400900_multibytecodec_MultibyteIncrementalEncoder_reset_impl(MultibyteIncrementalEncoderObject *self)
901/*[clinic end generated code: output=b4125d8f537a253f input=930f06760707b6ea]*/
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000902{
Victor Stinnere15dce32011-05-30 22:56:00 +0200903 /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */
904 unsigned char buffer[4], *outbuf;
905 Py_ssize_t r;
906 if (self->codec->encreset != NULL) {
907 outbuf = buffer;
908 r = self->codec->encreset(&self->state, self->codec->config,
909 &outbuf, sizeof(buffer));
910 if (r != 0)
911 return NULL;
912 }
Victor Stinnerd9491262013-04-14 02:06:32 +0200913 Py_CLEAR(self->pending);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000914 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000915}
916
917static struct PyMethodDef mbiencoder_methods[] = {
Brett Cannonf2de1fc2014-08-22 11:45:03 -0400918 _MULTIBYTECODEC_MULTIBYTEINCREMENTALENCODER_ENCODE_METHODDEF
919 _MULTIBYTECODEC_MULTIBYTEINCREMENTALENCODER_RESET_METHODDEF
920 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000921};
922
923static PyObject *
924mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
925{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000926 MultibyteIncrementalEncoderObject *self;
927 PyObject *codec = NULL;
928 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000929
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000930 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
931 incnewkwarglist, &errors))
932 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000933
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000934 self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
935 if (self == NULL)
936 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000937
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000938 codec = PyObject_GetAttrString((PyObject *)type, "codec");
939 if (codec == NULL)
940 goto errorexit;
941 if (!MultibyteCodec_Check(codec)) {
942 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
943 goto errorexit;
944 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000945
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000946 self->codec = ((MultibyteCodecObject *)codec)->codec;
Victor Stinnerd9491262013-04-14 02:06:32 +0200947 self->pending = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000948 self->errors = internal_error_callback(errors);
949 if (self->errors == NULL)
950 goto errorexit;
951 if (self->codec->encinit != NULL &&
952 self->codec->encinit(&self->state, self->codec->config) != 0)
953 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000954
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000955 Py_DECREF(codec);
956 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000957
958errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000959 Py_XDECREF(self);
960 Py_XDECREF(codec);
961 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000962}
963
964static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000965mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
966{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000967 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000968}
969
970static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000971mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000972 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000973{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000974 if (ERROR_ISCUSTOM(self->errors))
975 Py_VISIT(self->errors);
976 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000977}
978
979static void
980mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
981{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000982 PyObject_GC_UnTrack(self);
983 ERROR_DECREF(self->errors);
984 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000985}
986
987static PyTypeObject MultibyteIncrementalEncoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000988 PyVarObject_HEAD_INIT(NULL, 0)
989 "MultibyteIncrementalEncoder", /* tp_name */
990 sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
991 0, /* tp_itemsize */
992 /* methods */
993 (destructor)mbiencoder_dealloc, /* tp_dealloc */
994 0, /* tp_print */
995 0, /* tp_getattr */
996 0, /* tp_setattr */
997 0, /* tp_reserved */
998 0, /* tp_repr */
999 0, /* tp_as_number */
1000 0, /* tp_as_sequence */
1001 0, /* tp_as_mapping */
1002 0, /* tp_hash */
1003 0, /* tp_call */
1004 0, /* tp_str */
1005 PyObject_GenericGetAttr, /* tp_getattro */
1006 0, /* tp_setattro */
1007 0, /* tp_as_buffer */
1008 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1009 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1010 0, /* tp_doc */
1011 (traverseproc)mbiencoder_traverse, /* tp_traverse */
1012 0, /* tp_clear */
1013 0, /* tp_richcompare */
1014 0, /* tp_weaklistoffset */
1015 0, /* tp_iter */
1016 0, /* tp_iterext */
1017 mbiencoder_methods, /* tp_methods */
1018 0, /* tp_members */
1019 codecctx_getsets, /* tp_getset */
1020 0, /* tp_base */
1021 0, /* tp_dict */
1022 0, /* tp_descr_get */
1023 0, /* tp_descr_set */
1024 0, /* tp_dictoffset */
1025 mbiencoder_init, /* tp_init */
1026 0, /* tp_alloc */
1027 mbiencoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001028};
1029
1030
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001031/*[clinic input]
1032 class _multibytecodec.MultibyteIncrementalDecoder "MultibyteIncrementalDecoderObject *" "&MultibyteIncrementalDecoder_Type"
1033[clinic start generated code]*/
1034/*[clinic end generated code: output=da39a3ee5e6b4b0d input=f6003faaf2cea692]*/
1035
1036/*[clinic input]
1037_multibytecodec.MultibyteIncrementalDecoder.decode
1038
1039 input: Py_buffer
1040 final: int = 0
1041[clinic start generated code]*/
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001042
1043static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001044_multibytecodec_MultibyteIncrementalDecoder_decode_impl(MultibyteIncrementalDecoderObject *self, Py_buffer *input, int final)
1045/*[clinic end generated code: output=a0f3f92aa7303cf7 input=eb18c2f6e83589e1]*/
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001046{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001047 MultibyteDecodeBuffer buf;
1048 char *data, *wdata = NULL;
Victor Stinnera0dd0212013-04-11 22:09:04 +02001049 Py_ssize_t wsize, size, origpending;
Victor Stinnera0dd0212013-04-11 22:09:04 +02001050 PyObject *res;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001051
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001052 data = input->buf;
1053 size = input->len;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001054
Victor Stinner8f674cc2013-04-17 23:02:17 +02001055 _PyUnicodeWriter_Init(&buf.writer);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001056 buf.excobj = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001057 origpending = self->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001058
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001059 if (self->pendingsize == 0) {
1060 wsize = size;
1061 wdata = data;
1062 }
1063 else {
1064 if (size > PY_SSIZE_T_MAX - self->pendingsize) {
1065 PyErr_NoMemory();
1066 goto errorexit;
1067 }
1068 wsize = size + self->pendingsize;
1069 wdata = PyMem_Malloc(wsize);
Victor Stinner33283ba2013-07-15 17:47:39 +02001070 if (wdata == NULL) {
1071 PyErr_NoMemory();
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001072 goto errorexit;
Victor Stinner33283ba2013-07-15 17:47:39 +02001073 }
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001074 memcpy(wdata, self->pending, self->pendingsize);
1075 memcpy(wdata + self->pendingsize, data, size);
1076 self->pendingsize = 0;
1077 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001078
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001079 if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
1080 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001081
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001082 if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
1083 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001084
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001085 if (final && buf.inbuf < buf.inbuf_end) {
1086 if (multibytecodec_decerror(self->codec, &self->state,
1087 &buf, self->errors, MBERR_TOOFEW)) {
1088 /* recover the original pending buffer */
1089 memcpy(self->pending, wdata, origpending);
1090 self->pendingsize = origpending;
1091 goto errorexit;
1092 }
1093 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001094
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001095 if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
1096 if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
1097 goto errorexit;
1098 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001099
Victor Stinnera0dd0212013-04-11 22:09:04 +02001100 res = _PyUnicodeWriter_Finish(&buf.writer);
1101 if (res == NULL)
1102 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001103
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001104 if (wdata != data)
1105 PyMem_Del(wdata);
1106 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001107 return res;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001108
1109errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001110 if (wdata != NULL && wdata != data)
1111 PyMem_Del(wdata);
1112 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001113 _PyUnicodeWriter_Dealloc(&buf.writer);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001114 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001115}
1116
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001117/*[clinic input]
1118_multibytecodec.MultibyteIncrementalDecoder.reset
1119[clinic start generated code]*/
1120
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001121static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001122_multibytecodec_MultibyteIncrementalDecoder_reset_impl(MultibyteIncrementalDecoderObject *self)
1123/*[clinic end generated code: output=da423b1782c23ed1 input=3b63b3be85b2fb45]*/
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001124{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001125 if (self->codec->decreset != NULL &&
1126 self->codec->decreset(&self->state, self->codec->config) != 0)
1127 return NULL;
1128 self->pendingsize = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001129
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001130 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001131}
1132
1133static struct PyMethodDef mbidecoder_methods[] = {
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001134 _MULTIBYTECODEC_MULTIBYTEINCREMENTALDECODER_DECODE_METHODDEF
1135 _MULTIBYTECODEC_MULTIBYTEINCREMENTALDECODER_RESET_METHODDEF
1136 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001137};
1138
1139static PyObject *
1140mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1141{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001142 MultibyteIncrementalDecoderObject *self;
1143 PyObject *codec = NULL;
1144 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001145
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001146 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
1147 incnewkwarglist, &errors))
1148 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001149
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001150 self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
1151 if (self == NULL)
1152 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001153
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001154 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1155 if (codec == NULL)
1156 goto errorexit;
1157 if (!MultibyteCodec_Check(codec)) {
1158 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1159 goto errorexit;
1160 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001161
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001162 self->codec = ((MultibyteCodecObject *)codec)->codec;
1163 self->pendingsize = 0;
1164 self->errors = internal_error_callback(errors);
1165 if (self->errors == NULL)
1166 goto errorexit;
1167 if (self->codec->decinit != NULL &&
1168 self->codec->decinit(&self->state, self->codec->config) != 0)
1169 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001170
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001171 Py_DECREF(codec);
1172 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001173
1174errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001175 Py_XDECREF(self);
1176 Py_XDECREF(codec);
1177 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001178}
1179
1180static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001181mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
1182{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001183 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001184}
1185
1186static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001187mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001188 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001189{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001190 if (ERROR_ISCUSTOM(self->errors))
1191 Py_VISIT(self->errors);
1192 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001193}
1194
1195static void
1196mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
1197{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001198 PyObject_GC_UnTrack(self);
1199 ERROR_DECREF(self->errors);
1200 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001201}
1202
1203static PyTypeObject MultibyteIncrementalDecoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001204 PyVarObject_HEAD_INIT(NULL, 0)
1205 "MultibyteIncrementalDecoder", /* tp_name */
1206 sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */
1207 0, /* tp_itemsize */
1208 /* methods */
1209 (destructor)mbidecoder_dealloc, /* tp_dealloc */
1210 0, /* tp_print */
1211 0, /* tp_getattr */
1212 0, /* tp_setattr */
1213 0, /* tp_reserved */
1214 0, /* tp_repr */
1215 0, /* tp_as_number */
1216 0, /* tp_as_sequence */
1217 0, /* tp_as_mapping */
1218 0, /* tp_hash */
1219 0, /* tp_call */
1220 0, /* tp_str */
1221 PyObject_GenericGetAttr, /* tp_getattro */
1222 0, /* tp_setattro */
1223 0, /* tp_as_buffer */
1224 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1225 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1226 0, /* tp_doc */
1227 (traverseproc)mbidecoder_traverse, /* tp_traverse */
1228 0, /* tp_clear */
1229 0, /* tp_richcompare */
1230 0, /* tp_weaklistoffset */
1231 0, /* tp_iter */
1232 0, /* tp_iterext */
1233 mbidecoder_methods, /* tp_methods */
1234 0, /* tp_members */
1235 codecctx_getsets, /* tp_getset */
1236 0, /* tp_base */
1237 0, /* tp_dict */
1238 0, /* tp_descr_get */
1239 0, /* tp_descr_set */
1240 0, /* tp_dictoffset */
1241 mbidecoder_init, /* tp_init */
1242 0, /* tp_alloc */
1243 mbidecoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001244};
1245
1246
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001247/*[clinic input]
1248 class _multibytecodec.MultibyteStreamReader "MultibyteStreamReaderObject *" "MultibyteStreamReader_Type"
1249[clinic start generated code]*/
1250/*[clinic end generated code: output=da39a3ee5e6b4b0d input=d323634b74976f09]*/
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001251
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001252static PyObject *
1253mbstreamreader_iread(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001254 const char *method, Py_ssize_t sizehint)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001255{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001256 MultibyteDecodeBuffer buf;
Victor Stinnera0dd0212013-04-11 22:09:04 +02001257 PyObject *cres, *res;
1258 Py_ssize_t rsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001259
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001260 if (sizehint == 0)
Victor Stinnerb37b1742011-12-01 03:18:59 +01001261 return PyUnicode_New(0, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001262
Victor Stinner8f674cc2013-04-17 23:02:17 +02001263 _PyUnicodeWriter_Init(&buf.writer);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001264 buf.excobj = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001265 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001266
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001267 for (;;) {
1268 int endoffile;
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001269
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001270 if (sizehint < 0)
1271 cres = PyObject_CallMethod(self->stream,
1272 (char *)method, NULL);
1273 else
1274 cres = PyObject_CallMethod(self->stream,
1275 (char *)method, "i", sizehint);
1276 if (cres == NULL)
1277 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001278
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001279 if (!PyBytes_Check(cres)) {
1280 PyErr_Format(PyExc_TypeError,
1281 "stream function returned a "
1282 "non-bytes object (%.100s)",
1283 cres->ob_type->tp_name);
1284 goto errorexit;
1285 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001286
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001287 endoffile = (PyBytes_GET_SIZE(cres) == 0);
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001288
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001289 if (self->pendingsize > 0) {
1290 PyObject *ctr;
1291 char *ctrdata;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001292
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001293 if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
1294 PyErr_NoMemory();
1295 goto errorexit;
Victor Stinner064bbdc2013-07-08 22:28:27 +02001296 }
1297 rsize = PyBytes_GET_SIZE(cres) + self->pendingsize;
1298 ctr = PyBytes_FromStringAndSize(NULL, rsize);
1299 if (ctr == NULL)
1300 goto errorexit;
1301 ctrdata = PyBytes_AS_STRING(ctr);
1302 memcpy(ctrdata, self->pending, self->pendingsize);
1303 memcpy(ctrdata + self->pendingsize,
1304 PyBytes_AS_STRING(cres),
1305 PyBytes_GET_SIZE(cres));
1306 Py_DECREF(cres);
1307 cres = ctr;
1308 self->pendingsize = 0;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001309 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001310
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001311 rsize = PyBytes_GET_SIZE(cres);
1312 if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres),
1313 rsize) != 0)
1314 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001315
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001316 if (rsize > 0 && decoder_feed_buffer(
1317 (MultibyteStatefulDecoderContext *)self, &buf))
1318 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001319
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001320 if (endoffile || sizehint < 0) {
1321 if (buf.inbuf < buf.inbuf_end &&
1322 multibytecodec_decerror(self->codec, &self->state,
1323 &buf, self->errors, MBERR_TOOFEW))
1324 goto errorexit;
1325 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001326
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001327 if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
1328 if (decoder_append_pending(STATEFUL_DCTX(self),
1329 &buf) != 0)
1330 goto errorexit;
1331 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001332
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001333 Py_DECREF(cres);
1334 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001335
Victor Stinnera0dd0212013-04-11 22:09:04 +02001336 if (sizehint < 0 || buf.writer.pos != 0 || rsize == 0)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001337 break;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001338
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001339 sizehint = 1; /* read 1 more byte and retry */
1340 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001341
Victor Stinnera0dd0212013-04-11 22:09:04 +02001342 res = _PyUnicodeWriter_Finish(&buf.writer);
1343 if (res == NULL)
1344 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001345
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001346 Py_XDECREF(cres);
1347 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001348 return res;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001349
1350errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001351 Py_XDECREF(cres);
1352 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001353 _PyUnicodeWriter_Dealloc(&buf.writer);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001354 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001355}
1356
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001357/*[clinic input]
1358 _multibytecodec.MultibyteStreamReader.read
1359
1360 sizeobj: object = None
1361 /
1362[clinic start generated code]*/
1363
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001364static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001365_multibytecodec_MultibyteStreamReader_read_impl(MultibyteStreamReaderObject *self, PyObject *sizeobj)
1366/*[clinic end generated code: output=f298ea6e1bd2083c input=015b0d3ff2fca485]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001367{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001368 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001369
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001370 if (sizeobj == Py_None)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001371 size = -1;
1372 else if (PyLong_Check(sizeobj))
1373 size = PyLong_AsSsize_t(sizeobj);
1374 else {
1375 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1376 return NULL;
1377 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001378
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001379 if (size == -1 && PyErr_Occurred())
1380 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001381
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001382 return mbstreamreader_iread(self, "read", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001383}
1384
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001385/*[clinic input]
1386 _multibytecodec.MultibyteStreamReader.readline
1387
1388 sizeobj: object = None
1389 /
1390[clinic start generated code]*/
1391
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001392static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001393_multibytecodec_MultibyteStreamReader_readline_impl(MultibyteStreamReaderObject *self, PyObject *sizeobj)
1394/*[clinic end generated code: output=e5ac302a6d0999de input=41ccc64f9bb0cec3]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001395{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001396 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001397
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001398 if (sizeobj == Py_None)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001399 size = -1;
1400 else if (PyLong_Check(sizeobj))
1401 size = PyLong_AsSsize_t(sizeobj);
1402 else {
1403 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1404 return NULL;
1405 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001406
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001407 if (size == -1 && PyErr_Occurred())
1408 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001409
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001410 return mbstreamreader_iread(self, "readline", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001411}
1412
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001413/*[clinic input]
1414 _multibytecodec.MultibyteStreamReader.readlines
1415
1416 sizehintobj: object = None
1417 /
1418[clinic start generated code]*/
1419
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001420static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001421_multibytecodec_MultibyteStreamReader_readlines_impl(MultibyteStreamReaderObject *self, PyObject *sizehintobj)
1422/*[clinic end generated code: output=68f024178b77cb0f input=54932f5d4d88e880]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001423{
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001424 PyObject *r, *sr;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001425 Py_ssize_t sizehint;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001426
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001427 if (sizehintobj == Py_None)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001428 sizehint = -1;
1429 else if (PyLong_Check(sizehintobj))
1430 sizehint = PyLong_AsSsize_t(sizehintobj);
1431 else {
1432 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1433 return NULL;
1434 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001435
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001436 if (sizehint == -1 && PyErr_Occurred())
1437 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001438
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001439 r = mbstreamreader_iread(self, "read", sizehint);
1440 if (r == NULL)
1441 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001442
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001443 sr = PyUnicode_Splitlines(r, 1);
1444 Py_DECREF(r);
1445 return sr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001446}
1447
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001448/*[clinic input]
1449 _multibytecodec.MultibyteStreamReader.reset
1450[clinic start generated code]*/
1451
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001452static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001453_multibytecodec_MultibyteStreamReader_reset_impl(MultibyteStreamReaderObject *self)
1454/*[clinic end generated code: output=138490370a680abc input=5d4140db84b5e1e2]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001455{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001456 if (self->codec->decreset != NULL &&
1457 self->codec->decreset(&self->state, self->codec->config) != 0)
1458 return NULL;
1459 self->pendingsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001460
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001461 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001462}
1463
1464static struct PyMethodDef mbstreamreader_methods[] = {
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001465 _MULTIBYTECODEC_MULTIBYTESTREAMREADER_READ_METHODDEF
1466 _MULTIBYTECODEC_MULTIBYTESTREAMREADER_READLINE_METHODDEF
1467 _MULTIBYTECODEC_MULTIBYTESTREAMREADER_READLINES_METHODDEF
1468 _MULTIBYTECODEC_MULTIBYTESTREAMREADER_RESET_METHODDEF
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001469 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001470};
1471
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001472static PyMemberDef mbstreamreader_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001473 {"stream", T_OBJECT,
1474 offsetof(MultibyteStreamReaderObject, stream),
1475 READONLY, NULL},
1476 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001477};
1478
1479static PyObject *
1480mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1481{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001482 MultibyteStreamReaderObject *self;
1483 PyObject *stream, *codec = NULL;
1484 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001485
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001486 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
1487 streamkwarglist, &stream, &errors))
1488 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001489
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001490 self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
1491 if (self == NULL)
1492 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001493
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001494 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1495 if (codec == NULL)
1496 goto errorexit;
1497 if (!MultibyteCodec_Check(codec)) {
1498 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1499 goto errorexit;
1500 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001501
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001502 self->codec = ((MultibyteCodecObject *)codec)->codec;
1503 self->stream = stream;
1504 Py_INCREF(stream);
1505 self->pendingsize = 0;
1506 self->errors = internal_error_callback(errors);
1507 if (self->errors == NULL)
1508 goto errorexit;
1509 if (self->codec->decinit != NULL &&
1510 self->codec->decinit(&self->state, self->codec->config) != 0)
1511 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001512
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001513 Py_DECREF(codec);
1514 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001515
1516errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001517 Py_XDECREF(self);
1518 Py_XDECREF(codec);
1519 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001520}
1521
1522static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001523mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
1524{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001525 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001526}
1527
1528static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001529mbstreamreader_traverse(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001530 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001531{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001532 if (ERROR_ISCUSTOM(self->errors))
1533 Py_VISIT(self->errors);
1534 Py_VISIT(self->stream);
1535 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001536}
1537
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001538static void
1539mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
1540{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001541 PyObject_GC_UnTrack(self);
1542 ERROR_DECREF(self->errors);
1543 Py_XDECREF(self->stream);
1544 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001545}
1546
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001547static PyTypeObject MultibyteStreamReader_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001548 PyVarObject_HEAD_INIT(NULL, 0)
1549 "MultibyteStreamReader", /* tp_name */
1550 sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
1551 0, /* tp_itemsize */
1552 /* methods */
1553 (destructor)mbstreamreader_dealloc, /* tp_dealloc */
1554 0, /* tp_print */
1555 0, /* tp_getattr */
1556 0, /* tp_setattr */
1557 0, /* tp_reserved */
1558 0, /* tp_repr */
1559 0, /* tp_as_number */
1560 0, /* tp_as_sequence */
1561 0, /* tp_as_mapping */
1562 0, /* tp_hash */
1563 0, /* tp_call */
1564 0, /* tp_str */
1565 PyObject_GenericGetAttr, /* tp_getattro */
1566 0, /* tp_setattro */
1567 0, /* tp_as_buffer */
1568 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1569 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1570 0, /* tp_doc */
1571 (traverseproc)mbstreamreader_traverse, /* tp_traverse */
1572 0, /* tp_clear */
1573 0, /* tp_richcompare */
1574 0, /* tp_weaklistoffset */
1575 0, /* tp_iter */
1576 0, /* tp_iterext */
1577 mbstreamreader_methods, /* tp_methods */
1578 mbstreamreader_members, /* tp_members */
1579 codecctx_getsets, /* tp_getset */
1580 0, /* tp_base */
1581 0, /* tp_dict */
1582 0, /* tp_descr_get */
1583 0, /* tp_descr_set */
1584 0, /* tp_dictoffset */
1585 mbstreamreader_init, /* tp_init */
1586 0, /* tp_alloc */
1587 mbstreamreader_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001588};
1589
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001590
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001591/*[clinic input]
1592 class _multibytecodec.MultibyteStreamWriter "MultibyteStreamWriterObject *" "&MultibyteStreamWriter_Type"
1593[clinic start generated code]*/
1594/*[clinic end generated code: output=da39a3ee5e6b4b0d input=cde22780a215d6ac]*/
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001595
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001596static int
1597mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001598 PyObject *unistr)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001599{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001600 PyObject *str, *wr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001601
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001602 str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
1603 if (str == NULL)
1604 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001605
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001606 wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001607 Py_DECREF(str);
1608 if (wr == NULL)
1609 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001610
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001611 Py_DECREF(wr);
1612 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001613}
1614
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001615/*[clinic input]
1616 _multibytecodec.MultibyteStreamWriter.write
1617
1618 strobj: object
1619 /
1620[clinic start generated code]*/
1621
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001622static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001623_multibytecodec_MultibyteStreamWriter_write(MultibyteStreamWriterObject *self, PyObject *strobj)
1624/*[clinic end generated code: output=44e9eb0db0374cb1 input=551dc4c018c10a2b]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001625{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001626 if (mbstreamwriter_iwrite(self, strobj))
1627 return NULL;
1628 else
1629 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001630}
1631
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001632/*[clinic input]
1633 _multibytecodec.MultibyteStreamWriter.writelines
1634
1635 lines: object
1636 /
1637[clinic start generated code]*/
1638
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001639static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001640_multibytecodec_MultibyteStreamWriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines)
1641/*[clinic end generated code: output=4facbb0638dde172 input=57797fe7008d4e96]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001642{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001643 PyObject *strobj;
1644 int i, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001645
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001646 if (!PySequence_Check(lines)) {
1647 PyErr_SetString(PyExc_TypeError,
1648 "arg must be a sequence object");
1649 return NULL;
1650 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001651
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001652 for (i = 0; i < PySequence_Length(lines); i++) {
1653 /* length can be changed even within this loop */
1654 strobj = PySequence_GetItem(lines, i);
1655 if (strobj == NULL)
1656 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001657
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001658 r = mbstreamwriter_iwrite(self, strobj);
1659 Py_DECREF(strobj);
1660 if (r == -1)
1661 return NULL;
1662 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001663
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001664 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001665}
1666
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001667/*[clinic input]
1668 _multibytecodec.MultibyteStreamWriter.reset
1669[clinic start generated code]*/
1670
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001671static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001672_multibytecodec_MultibyteStreamWriter_reset_impl(MultibyteStreamWriterObject *self)
1673/*[clinic end generated code: output=8f54a4d9b03db5ff input=b56dbcbaf35cc10c]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001674{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001675 PyObject *pwrt;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001676
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001677 pwrt = multibytecodec_encode(self->codec, &self->state,
Victor Stinnerd9491262013-04-14 02:06:32 +02001678 self->pending, NULL, self->errors,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001679 MBENC_FLUSH | MBENC_RESET);
1680 /* some pending buffer can be truncated when UnicodeEncodeError is
1681 * raised on 'strict' mode. but, 'reset' method is designed to
1682 * reset the pending buffer or states so failed string sequence
1683 * ought to be missed */
Victor Stinnerd9491262013-04-14 02:06:32 +02001684 Py_CLEAR(self->pending);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001685 if (pwrt == NULL)
1686 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001687
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001688 assert(PyBytes_Check(pwrt));
1689 if (PyBytes_Size(pwrt) > 0) {
1690 PyObject *wr;
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001691
1692 wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001693 if (wr == NULL) {
1694 Py_DECREF(pwrt);
1695 return NULL;
1696 }
1697 }
1698 Py_DECREF(pwrt);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001699
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001700 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001701}
1702
1703static PyObject *
1704mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1705{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001706 MultibyteStreamWriterObject *self;
1707 PyObject *stream, *codec = NULL;
1708 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001709
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001710 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
1711 streamkwarglist, &stream, &errors))
1712 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001713
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001714 self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
1715 if (self == NULL)
1716 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001717
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001718 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1719 if (codec == NULL)
1720 goto errorexit;
1721 if (!MultibyteCodec_Check(codec)) {
1722 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1723 goto errorexit;
1724 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001725
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001726 self->codec = ((MultibyteCodecObject *)codec)->codec;
1727 self->stream = stream;
1728 Py_INCREF(stream);
Victor Stinnerd9491262013-04-14 02:06:32 +02001729 self->pending = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001730 self->errors = internal_error_callback(errors);
1731 if (self->errors == NULL)
1732 goto errorexit;
1733 if (self->codec->encinit != NULL &&
1734 self->codec->encinit(&self->state, self->codec->config) != 0)
1735 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001736
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001737 Py_DECREF(codec);
1738 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001739
1740errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001741 Py_XDECREF(self);
1742 Py_XDECREF(codec);
1743 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001744}
1745
1746static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001747mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
1748{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001749 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001750}
1751
1752static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001753mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001754 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001755{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001756 if (ERROR_ISCUSTOM(self->errors))
1757 Py_VISIT(self->errors);
1758 Py_VISIT(self->stream);
1759 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001760}
1761
1762static void
1763mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
1764{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001765 PyObject_GC_UnTrack(self);
1766 ERROR_DECREF(self->errors);
1767 Py_XDECREF(self->stream);
1768 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001769}
1770
1771static struct PyMethodDef mbstreamwriter_methods[] = {
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001772 _MULTIBYTECODEC_MULTIBYTESTREAMWRITER_WRITE_METHODDEF
1773 _MULTIBYTECODEC_MULTIBYTESTREAMWRITER_WRITELINES_METHODDEF
1774 _MULTIBYTECODEC_MULTIBYTESTREAMWRITER_RESET_METHODDEF
1775 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001776};
1777
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001778static PyMemberDef mbstreamwriter_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001779 {"stream", T_OBJECT,
1780 offsetof(MultibyteStreamWriterObject, stream),
1781 READONLY, NULL},
1782 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001783};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001784
1785static PyTypeObject MultibyteStreamWriter_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001786 PyVarObject_HEAD_INIT(NULL, 0)
1787 "MultibyteStreamWriter", /* tp_name */
1788 sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
1789 0, /* tp_itemsize */
1790 /* methods */
1791 (destructor)mbstreamwriter_dealloc, /* tp_dealloc */
1792 0, /* tp_print */
1793 0, /* tp_getattr */
1794 0, /* tp_setattr */
1795 0, /* tp_reserved */
1796 0, /* tp_repr */
1797 0, /* tp_as_number */
1798 0, /* tp_as_sequence */
1799 0, /* tp_as_mapping */
1800 0, /* tp_hash */
1801 0, /* tp_call */
1802 0, /* tp_str */
1803 PyObject_GenericGetAttr, /* tp_getattro */
1804 0, /* tp_setattro */
1805 0, /* tp_as_buffer */
1806 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1807 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1808 0, /* tp_doc */
1809 (traverseproc)mbstreamwriter_traverse, /* tp_traverse */
1810 0, /* tp_clear */
1811 0, /* tp_richcompare */
1812 0, /* tp_weaklistoffset */
1813 0, /* tp_iter */
1814 0, /* tp_iterext */
1815 mbstreamwriter_methods, /* tp_methods */
1816 mbstreamwriter_members, /* tp_members */
1817 codecctx_getsets, /* tp_getset */
1818 0, /* tp_base */
1819 0, /* tp_dict */
1820 0, /* tp_descr_get */
1821 0, /* tp_descr_set */
1822 0, /* tp_dictoffset */
1823 mbstreamwriter_init, /* tp_init */
1824 0, /* tp_alloc */
1825 mbstreamwriter_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001826};
1827
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001828
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001829/*[clinic input]
1830_multibytecodec.__create_codec
1831
1832 arg: object
1833 /
1834[clinic start generated code]*/
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001835
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001836static PyObject *
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001837_multibytecodec___create_codec(PyModuleDef *module, PyObject *arg)
1838/*[clinic end generated code: output=fbe74f6510640163 input=6840b2a6b183fcfa]*/
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001839{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001840 MultibyteCodecObject *self;
1841 MultibyteCodec *codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001842
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001843 if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
1844 PyErr_SetString(PyExc_ValueError, "argument type invalid");
1845 return NULL;
1846 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001847
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001848 codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
1849 if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
1850 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001851
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001852 self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
1853 if (self == NULL)
1854 return NULL;
1855 self->codec = codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001856
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001857 return (PyObject *)self;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001858}
1859
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001860static struct PyMethodDef __methods[] = {
Brett Cannonf2de1fc2014-08-22 11:45:03 -04001861 _MULTIBYTECODEC___CREATE_CODEC_METHODDEF
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001862 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001863};
1864
Martin v. Löwis1a214512008-06-11 05:26:20 +00001865
1866static struct PyModuleDef _multibytecodecmodule = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001867 PyModuleDef_HEAD_INIT,
1868 "_multibytecodec",
1869 NULL,
1870 -1,
1871 __methods,
1872 NULL,
1873 NULL,
1874 NULL,
1875 NULL
Martin v. Löwis1a214512008-06-11 05:26:20 +00001876};
1877
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001878PyMODINIT_FUNC
Martin v. Löwis1a214512008-06-11 05:26:20 +00001879PyInit__multibytecodec(void)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001880{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001881 int i;
1882 PyObject *m;
1883 PyTypeObject *typelist[] = {
1884 &MultibyteIncrementalEncoder_Type,
1885 &MultibyteIncrementalDecoder_Type,
1886 &MultibyteStreamReader_Type,
1887 &MultibyteStreamWriter_Type,
1888 NULL
1889 };
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001890
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001891 if (PyType_Ready(&MultibyteCodec_Type) < 0)
1892 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001893
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001894 m = PyModule_Create(&_multibytecodecmodule);
1895 if (m == NULL)
1896 return NULL;
Neal Norwitz058bde12005-09-21 06:44:25 +00001897
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001898 for (i = 0; typelist[i] != NULL; i++) {
1899 if (PyType_Ready(typelist[i]) < 0)
1900 return NULL;
1901 Py_INCREF(typelist[i]);
1902 PyModule_AddObject(m, typelist[i]->tp_name,
1903 (PyObject *)typelist[i]);
1904 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001905
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001906 if (PyErr_Occurred()) {
1907 Py_FatalError("can't initialize the _multibytecodec module");
1908 Py_DECREF(m);
1909 m = NULL;
1910 }
1911 return m;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001912}