blob: 1d6db9545efed3f8ca5e5756265f03aaec23ce78 [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"
11
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000012typedef struct {
13 const Py_UNICODE *inbuf, *inbuf_top, *inbuf_end;
14 unsigned char *outbuf, *outbuf_end;
15 PyObject *excobj, *outobj;
16} MultibyteEncodeBuffer;
17
18typedef struct {
19 const unsigned char *inbuf, *inbuf_top, *inbuf_end;
20 Py_UNICODE *outbuf, *outbuf_end;
21 PyObject *excobj, *outobj;
22} MultibyteDecodeBuffer;
23
24PyDoc_STRVAR(MultibyteCodec_Encode__doc__,
25"I.encode(unicode[, errors]) -> (string, length consumed)\n\
26\n\
27Return an encoded string version of `unicode'. errors may be given to\n\
28set a different error handling scheme. Default is 'strict' meaning that\n\
29encoding errors raise a UnicodeEncodeError. Other possible values are\n\
30'ignore', 'replace' and 'xmlcharrefreplace' as well as any other name\n\
31registered with codecs.register_error that can handle UnicodeEncodeErrors.");
32
33PyDoc_STRVAR(MultibyteCodec_Decode__doc__,
34"I.decode(string[, errors]) -> (unicodeobject, length consumed)\n\
35\n\
36Decodes `string' using I, an MultibyteCodec instance. errors may be given\n\
37to set a different error handling scheme. Default is 'strict' meaning\n\
38that encoding errors raise a UnicodeDecodeError. Other possible values\n\
Benjamin Petersonf10a79a2008-10-11 00:49:57 +000039are 'ignore' and 'replace' as well as any other name registered with\n\
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000040codecs.register_error that is able to handle UnicodeDecodeErrors.");
41
Martin v. Löwis02cbf4a2006-02-27 17:20:04 +000042static char *codeckwarglist[] = {"input", "errors", NULL};
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000043static char *incnewkwarglist[] = {"errors", NULL};
44static char *incrementalkwarglist[] = {"input", "final", NULL};
Martin v. Löwis02cbf4a2006-02-27 17:20:04 +000045static char *streamkwarglist[] = {"stream", "errors", NULL};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000046
47static PyObject *multibytecodec_encode(MultibyteCodec *,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000048 MultibyteCodec_State *, const Py_UNICODE **, Py_ssize_t,
49 PyObject *, int);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000050
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000051#define MBENC_RESET MBENC_MAX<<1 /* reset after an encoding session */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000052
53static PyObject *
Hye-Shik Chang4b96c132006-03-04 16:08:19 +000054make_tuple(PyObject *object, Py_ssize_t len)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000055{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000056 PyObject *v, *w;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000057
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000058 if (object == NULL)
59 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000060
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000061 v = PyTuple_New(2);
62 if (v == NULL) {
63 Py_DECREF(object);
64 return NULL;
65 }
66 PyTuple_SET_ITEM(v, 0, object);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000067
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000068 w = PyLong_FromSsize_t(len);
69 if (w == NULL) {
70 Py_DECREF(v);
71 return NULL;
72 }
73 PyTuple_SET_ITEM(v, 1, w);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000074
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000075 return v;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000076}
77
78static PyObject *
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000079internal_error_callback(const char *errors)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000080{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000081 if (errors == NULL || strcmp(errors, "strict") == 0)
82 return ERROR_STRICT;
83 else if (strcmp(errors, "ignore") == 0)
84 return ERROR_IGNORE;
85 else if (strcmp(errors, "replace") == 0)
86 return ERROR_REPLACE;
87 else
88 return PyUnicode_FromString(errors);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000089}
90
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000091static PyObject *
92call_error_callback(PyObject *errors, PyObject *exc)
93{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000094 PyObject *args, *cb, *r;
95 const char *str;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000096
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000097 assert(PyUnicode_Check(errors));
98 str = _PyUnicode_AsString(errors);
99 if (str == NULL)
100 return NULL;
101 cb = PyCodec_LookupError(str);
102 if (cb == NULL)
103 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000104
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000105 args = PyTuple_New(1);
106 if (args == NULL) {
107 Py_DECREF(cb);
108 return NULL;
109 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000110
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000111 PyTuple_SET_ITEM(args, 0, exc);
112 Py_INCREF(exc);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000113
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000114 r = PyObject_CallObject(cb, args);
115 Py_DECREF(args);
116 Py_DECREF(cb);
117 return r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000118}
119
120static PyObject *
121codecctx_errors_get(MultibyteStatefulCodecContext *self)
122{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000123 const char *errors;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000124
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000125 if (self->errors == ERROR_STRICT)
126 errors = "strict";
127 else if (self->errors == ERROR_IGNORE)
128 errors = "ignore";
129 else if (self->errors == ERROR_REPLACE)
130 errors = "replace";
131 else {
132 Py_INCREF(self->errors);
133 return self->errors;
134 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000135
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000136 return PyUnicode_FromString(errors);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000137}
138
139static int
140codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000141 void *closure)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000142{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000143 PyObject *cb;
144 const char *str;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000145
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000146 if (!PyUnicode_Check(value)) {
147 PyErr_SetString(PyExc_TypeError, "errors must be a string");
148 return -1;
149 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000150
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000151 str = _PyUnicode_AsString(value);
152 if (str == NULL)
153 return -1;
Neal Norwitz6ea45d32007-08-26 04:19:43 +0000154
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000155 cb = internal_error_callback(str);
156 if (cb == NULL)
157 return -1;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000158
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000159 ERROR_DECREF(self->errors);
160 self->errors = cb;
161 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000162}
163
164/* This getset handlers list is used by all the stateful codec objects */
165static PyGetSetDef codecctx_getsets[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000166 {"errors", (getter)codecctx_errors_get,
167 (setter)codecctx_errors_set,
168 PyDoc_STR("how to treat errors")},
169 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000170};
171
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000172static int
Hye-Shik Chang4b96c132006-03-04 16:08:19 +0000173expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000174{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000175 Py_ssize_t orgpos, orgsize, incsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000176
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000177 orgpos = (Py_ssize_t)((char *)buf->outbuf -
178 PyBytes_AS_STRING(buf->outobj));
179 orgsize = PyBytes_GET_SIZE(buf->outobj);
180 incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize);
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000181
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000182 if (orgsize > PY_SSIZE_T_MAX - incsize)
183 return -1;
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000184
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000185 if (_PyBytes_Resize(&buf->outobj, orgsize + incsize) == -1)
186 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000187
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000188 buf->outbuf = (unsigned char *)PyBytes_AS_STRING(buf->outobj) +orgpos;
189 buf->outbuf_end = (unsigned char *)PyBytes_AS_STRING(buf->outobj)
190 + PyBytes_GET_SIZE(buf->outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000191
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000192 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000193}
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000194#define REQUIRE_ENCODEBUFFER(buf, s) { \
195 if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end) \
196 if (expand_encodebuffer(buf, s) == -1) \
197 goto errorexit; \
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000198}
199
200static int
Hye-Shik Chang4b96c132006-03-04 16:08:19 +0000201expand_decodebuffer(MultibyteDecodeBuffer *buf, Py_ssize_t esize)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000202{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000203 Py_ssize_t orgpos, orgsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000204
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000205 orgpos = (Py_ssize_t)(buf->outbuf - PyUnicode_AS_UNICODE(buf->outobj));
206 orgsize = PyUnicode_GET_SIZE(buf->outobj);
207 if (PyUnicode_Resize(&buf->outobj, orgsize + (
208 esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize)) == -1)
209 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000210
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000211 buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj) + orgpos;
212 buf->outbuf_end = PyUnicode_AS_UNICODE(buf->outobj)
213 + PyUnicode_GET_SIZE(buf->outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000214
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000215 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000216}
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000217#define REQUIRE_DECODEBUFFER(buf, s) { \
218 if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end) \
219 if (expand_decodebuffer(buf, s) == -1) \
220 goto errorexit; \
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000221}
222
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000223
224/**
225 * MultibyteCodec object
226 */
227
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000228static int
229multibytecodec_encerror(MultibyteCodec *codec,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000230 MultibyteCodec_State *state,
231 MultibyteEncodeBuffer *buf,
232 PyObject *errors, Py_ssize_t e)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000233{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000234 PyObject *retobj = NULL, *retstr = NULL, *tobj;
235 Py_ssize_t retstrsize, newpos;
236 Py_ssize_t esize, start, end;
237 const char *reason;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000238
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000239 if (e > 0) {
240 reason = "illegal multibyte sequence";
241 esize = e;
242 }
243 else {
244 switch (e) {
245 case MBERR_TOOSMALL:
246 REQUIRE_ENCODEBUFFER(buf, -1);
247 return 0; /* retry it */
248 case MBERR_TOOFEW:
249 reason = "incomplete multibyte sequence";
250 esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
251 break;
252 case MBERR_INTERNAL:
253 PyErr_SetString(PyExc_RuntimeError,
254 "internal codec error");
255 return -1;
256 default:
257 PyErr_SetString(PyExc_RuntimeError,
258 "unknown runtime error");
259 return -1;
260 }
261 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000262
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000263 if (errors == ERROR_REPLACE) {
264 const Py_UNICODE replchar = '?', *inbuf = &replchar;
265 Py_ssize_t r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000266
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000267 for (;;) {
268 Py_ssize_t outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000269
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000270 outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
271 r = codec->encode(state, codec->config, &inbuf, 1,
272 &buf->outbuf, outleft, 0);
273 if (r == MBERR_TOOSMALL) {
274 REQUIRE_ENCODEBUFFER(buf, -1);
275 continue;
276 }
277 else
278 break;
279 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000280
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000281 if (r != 0) {
282 REQUIRE_ENCODEBUFFER(buf, 1);
283 *buf->outbuf++ = '?';
284 }
285 }
286 if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
287 buf->inbuf += esize;
288 return 0;
289 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000290
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000291 start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
292 end = start + esize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000293
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000294 /* use cached exception object if available */
295 if (buf->excobj == NULL) {
296 buf->excobj = PyUnicodeEncodeError_Create(codec->encoding,
297 buf->inbuf_top,
298 buf->inbuf_end - buf->inbuf_top,
299 start, end, reason);
300 if (buf->excobj == NULL)
301 goto errorexit;
302 }
303 else
304 if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 ||
305 PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 ||
306 PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0)
307 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000308
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000309 if (errors == ERROR_STRICT) {
310 PyCodec_StrictErrors(buf->excobj);
311 goto errorexit;
312 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000313
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000314 retobj = call_error_callback(errors, buf->excobj);
315 if (retobj == NULL)
316 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000317
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000318 if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
319 !PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) ||
320 !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) {
321 PyErr_SetString(PyExc_TypeError,
322 "encoding error handler must return "
323 "(unicode, int) tuple");
324 goto errorexit;
325 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000326
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000327 {
328 const Py_UNICODE *uraw = PyUnicode_AS_UNICODE(tobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000329
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000330 retstr = multibytecodec_encode(codec, state, &uraw,
331 PyUnicode_GET_SIZE(tobj), ERROR_STRICT,
332 MBENC_FLUSH);
333 if (retstr == NULL)
334 goto errorexit;
335 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000336
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000337 assert(PyBytes_Check(retstr));
338 retstrsize = PyBytes_GET_SIZE(retstr);
339 REQUIRE_ENCODEBUFFER(buf, retstrsize);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000340
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000341 memcpy(buf->outbuf, PyBytes_AS_STRING(retstr), retstrsize);
342 buf->outbuf += retstrsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000343
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000344 newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
345 if (newpos < 0 && !PyErr_Occurred())
346 newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
347 if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
348 PyErr_Clear();
349 PyErr_Format(PyExc_IndexError,
350 "position %zd from error handler out of bounds",
351 newpos);
352 goto errorexit;
353 }
354 buf->inbuf = buf->inbuf_top + newpos;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000355
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000356 Py_DECREF(retobj);
357 Py_DECREF(retstr);
358 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000359
360errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000361 Py_XDECREF(retobj);
362 Py_XDECREF(retstr);
363 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000364}
365
366static int
367multibytecodec_decerror(MultibyteCodec *codec,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000368 MultibyteCodec_State *state,
369 MultibyteDecodeBuffer *buf,
370 PyObject *errors, Py_ssize_t e)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000371{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000372 PyObject *retobj = NULL, *retuni = NULL;
373 Py_ssize_t retunisize, newpos;
374 const char *reason;
375 Py_ssize_t esize, start, end;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000376
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000377 if (e > 0) {
378 reason = "illegal multibyte sequence";
379 esize = e;
380 }
381 else {
382 switch (e) {
383 case MBERR_TOOSMALL:
384 REQUIRE_DECODEBUFFER(buf, -1);
385 return 0; /* retry it */
386 case MBERR_TOOFEW:
387 reason = "incomplete multibyte sequence";
388 esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
389 break;
390 case MBERR_INTERNAL:
391 PyErr_SetString(PyExc_RuntimeError,
392 "internal codec error");
393 return -1;
394 default:
395 PyErr_SetString(PyExc_RuntimeError,
396 "unknown runtime error");
397 return -1;
398 }
399 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000400
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000401 if (errors == ERROR_REPLACE) {
402 REQUIRE_DECODEBUFFER(buf, 1);
403 *buf->outbuf++ = Py_UNICODE_REPLACEMENT_CHARACTER;
404 }
405 if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
406 buf->inbuf += esize;
407 return 0;
408 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000409
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000410 start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
411 end = start + esize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000412
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000413 /* use cached exception object if available */
414 if (buf->excobj == NULL) {
415 buf->excobj = PyUnicodeDecodeError_Create(codec->encoding,
416 (const char *)buf->inbuf_top,
417 (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top),
418 start, end, reason);
419 if (buf->excobj == NULL)
420 goto errorexit;
421 }
422 else
423 if (PyUnicodeDecodeError_SetStart(buf->excobj, start) ||
424 PyUnicodeDecodeError_SetEnd(buf->excobj, end) ||
425 PyUnicodeDecodeError_SetReason(buf->excobj, reason))
426 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000427
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000428 if (errors == ERROR_STRICT) {
429 PyCodec_StrictErrors(buf->excobj);
430 goto errorexit;
431 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000432
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000433 retobj = call_error_callback(errors, buf->excobj);
434 if (retobj == NULL)
435 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000436
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000437 if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
438 !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) ||
439 !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) {
440 PyErr_SetString(PyExc_TypeError,
441 "decoding error handler must return "
442 "(unicode, int) tuple");
443 goto errorexit;
444 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000445
Victor Stinner4eea8492011-11-21 03:01:27 +0100446 if (PyUnicode_AsUnicode(retuni) == NULL)
447 goto errorexit;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000448 retunisize = PyUnicode_GET_SIZE(retuni);
449 if (retunisize > 0) {
450 REQUIRE_DECODEBUFFER(buf, retunisize);
Victor Stinner4eea8492011-11-21 03:01:27 +0100451 memcpy((char *)buf->outbuf, PyUnicode_AS_UNICODE(retuni),
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000452 retunisize * Py_UNICODE_SIZE);
453 buf->outbuf += retunisize;
454 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000455
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000456 newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
457 if (newpos < 0 && !PyErr_Occurred())
458 newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
459 if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
460 PyErr_Clear();
461 PyErr_Format(PyExc_IndexError,
462 "position %zd from error handler out of bounds",
463 newpos);
464 goto errorexit;
465 }
466 buf->inbuf = buf->inbuf_top + newpos;
467 Py_DECREF(retobj);
468 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000469
470errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000471 Py_XDECREF(retobj);
472 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000473}
474
475static PyObject *
476multibytecodec_encode(MultibyteCodec *codec,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000477 MultibyteCodec_State *state,
478 const Py_UNICODE **data, Py_ssize_t datalen,
479 PyObject *errors, int flags)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000480{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000481 MultibyteEncodeBuffer buf;
482 Py_ssize_t finalsize, r = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000483
Victor Stinner6bcbef72011-05-24 22:17:55 +0200484 if (datalen == 0 && !(flags & MBENC_RESET))
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000485 return PyBytes_FromStringAndSize(NULL, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000486
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000487 buf.excobj = NULL;
Brett Cannonb94767f2011-02-22 20:15:44 +0000488 buf.outobj = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000489 buf.inbuf = buf.inbuf_top = *data;
490 buf.inbuf_end = buf.inbuf_top + datalen;
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000491
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000492 if (datalen > (PY_SSIZE_T_MAX - 16) / 2) {
493 PyErr_NoMemory();
494 goto errorexit;
495 }
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000496
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000497 buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16);
498 if (buf.outobj == NULL)
499 goto errorexit;
500 buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj);
501 buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000502
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000503 while (buf.inbuf < buf.inbuf_end) {
504 Py_ssize_t inleft, outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000505
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000506 /* we don't reuse inleft and outleft here.
507 * error callbacks can relocate the cursor anywhere on buffer*/
508 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
509 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
510 r = codec->encode(state, codec->config, &buf.inbuf, inleft,
511 &buf.outbuf, outleft, flags);
512 if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH)))
513 break;
514 else if (multibytecodec_encerror(codec, state, &buf, errors,r))
515 goto errorexit;
516 else if (r == MBERR_TOOFEW)
517 break;
518 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000519
Victor Stinner6bcbef72011-05-24 22:17:55 +0200520 if (codec->encreset != NULL && (flags & MBENC_RESET))
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000521 for (;;) {
522 Py_ssize_t outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000523
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000524 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
525 r = codec->encreset(state, codec->config, &buf.outbuf,
526 outleft);
527 if (r == 0)
528 break;
529 else if (multibytecodec_encerror(codec, state,
530 &buf, errors, r))
531 goto errorexit;
532 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000533
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000534 finalsize = (Py_ssize_t)((char *)buf.outbuf -
535 PyBytes_AS_STRING(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000536
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000537 if (finalsize != PyBytes_GET_SIZE(buf.outobj))
538 if (_PyBytes_Resize(&buf.outobj, finalsize) == -1)
539 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000540
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000541 *data = buf.inbuf;
542 Py_XDECREF(buf.excobj);
543 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000544
545errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000546 Py_XDECREF(buf.excobj);
547 Py_XDECREF(buf.outobj);
548 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000549}
550
551static PyObject *
552MultibyteCodec_Encode(MultibyteCodecObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000553 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000554{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000555 MultibyteCodec_State state;
556 Py_UNICODE *data;
557 PyObject *errorcb, *r, *arg, *ucvt;
558 const char *errors = NULL;
559 Py_ssize_t datalen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000560
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000561 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode",
562 codeckwarglist, &arg, &errors))
563 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000564
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000565 if (PyUnicode_Check(arg))
566 ucvt = NULL;
567 else {
568 arg = ucvt = PyObject_Str(arg);
569 if (arg == NULL)
570 return NULL;
571 else if (!PyUnicode_Check(arg)) {
572 PyErr_SetString(PyExc_TypeError,
573 "couldn't convert the object to unicode.");
574 Py_DECREF(ucvt);
575 return NULL;
576 }
577 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000578
Victor Stinner9a80fab2011-11-21 02:50:14 +0100579 data = PyUnicode_AsUnicodeAndSize(arg, &datalen);
580 if (data == NULL) {
581 Py_XDECREF(ucvt);
582 return NULL;
583 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000584
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000585 errorcb = internal_error_callback(errors);
586 if (errorcb == NULL) {
587 Py_XDECREF(ucvt);
588 return NULL;
589 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000590
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000591 if (self->codec->encinit != NULL &&
592 self->codec->encinit(&state, self->codec->config) != 0)
593 goto errorexit;
594 r = multibytecodec_encode(self->codec, &state,
595 (const Py_UNICODE **)&data, datalen, errorcb,
596 MBENC_FLUSH | MBENC_RESET);
597 if (r == NULL)
598 goto errorexit;
Hye-Shik Chang2bb146f2004-07-18 03:06:29 +0000599
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000600 ERROR_DECREF(errorcb);
601 Py_XDECREF(ucvt);
602 return make_tuple(r, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000603
604errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000605 ERROR_DECREF(errorcb);
606 Py_XDECREF(ucvt);
607 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000608}
609
610static PyObject *
611MultibyteCodec_Decode(MultibyteCodecObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000612 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000613{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000614 MultibyteCodec_State state;
615 MultibyteDecodeBuffer buf;
616 PyObject *errorcb;
617 Py_buffer pdata;
618 const char *data, *errors = NULL;
619 Py_ssize_t datalen, finalsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000620
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000621 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|z:decode",
622 codeckwarglist, &pdata, &errors))
623 return NULL;
624 data = pdata.buf;
625 datalen = pdata.len;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000626
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000627 errorcb = internal_error_callback(errors);
628 if (errorcb == NULL) {
629 PyBuffer_Release(&pdata);
630 return NULL;
631 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000632
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000633 if (datalen == 0) {
634 PyBuffer_Release(&pdata);
635 ERROR_DECREF(errorcb);
636 return make_tuple(PyUnicode_FromUnicode(NULL, 0), 0);
637 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000638
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000639 buf.excobj = NULL;
640 buf.inbuf = buf.inbuf_top = (unsigned char *)data;
641 buf.inbuf_end = buf.inbuf_top + datalen;
642 buf.outobj = PyUnicode_FromUnicode(NULL, datalen);
643 if (buf.outobj == NULL)
644 goto errorexit;
645 buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj);
646 buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000647
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000648 if (self->codec->decinit != NULL &&
649 self->codec->decinit(&state, self->codec->config) != 0)
650 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000651
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000652 while (buf.inbuf < buf.inbuf_end) {
653 Py_ssize_t inleft, outleft, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000654
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000655 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
656 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000657
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000658 r = self->codec->decode(&state, self->codec->config,
659 &buf.inbuf, inleft, &buf.outbuf, outleft);
660 if (r == 0)
661 break;
662 else if (multibytecodec_decerror(self->codec, &state,
663 &buf, errorcb, r))
664 goto errorexit;
665 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000666
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000667 finalsize = (Py_ssize_t)(buf.outbuf -
668 PyUnicode_AS_UNICODE(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000669
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000670 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
671 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
672 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000673
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000674 PyBuffer_Release(&pdata);
675 Py_XDECREF(buf.excobj);
676 ERROR_DECREF(errorcb);
677 return make_tuple(buf.outobj, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000678
679errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000680 PyBuffer_Release(&pdata);
681 ERROR_DECREF(errorcb);
682 Py_XDECREF(buf.excobj);
683 Py_XDECREF(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000684
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000685 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000686}
687
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000688static struct PyMethodDef multibytecodec_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000689 {"encode", (PyCFunction)MultibyteCodec_Encode,
690 METH_VARARGS | METH_KEYWORDS,
691 MultibyteCodec_Encode__doc__},
692 {"decode", (PyCFunction)MultibyteCodec_Decode,
693 METH_VARARGS | METH_KEYWORDS,
694 MultibyteCodec_Decode__doc__},
695 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000696};
697
698static void
699multibytecodec_dealloc(MultibyteCodecObject *self)
700{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000701 PyObject_Del(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000702}
703
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000704static PyTypeObject MultibyteCodec_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000705 PyVarObject_HEAD_INIT(NULL, 0)
706 "MultibyteCodec", /* tp_name */
707 sizeof(MultibyteCodecObject), /* tp_basicsize */
708 0, /* tp_itemsize */
709 /* methods */
710 (destructor)multibytecodec_dealloc, /* tp_dealloc */
711 0, /* tp_print */
712 0, /* tp_getattr */
713 0, /* tp_setattr */
714 0, /* tp_reserved */
715 0, /* tp_repr */
716 0, /* tp_as_number */
717 0, /* tp_as_sequence */
718 0, /* tp_as_mapping */
719 0, /* tp_hash */
720 0, /* tp_call */
721 0, /* tp_str */
722 PyObject_GenericGetAttr, /* tp_getattro */
723 0, /* tp_setattro */
724 0, /* tp_as_buffer */
725 Py_TPFLAGS_DEFAULT, /* tp_flags */
726 0, /* tp_doc */
727 0, /* tp_traverse */
728 0, /* tp_clear */
729 0, /* tp_richcompare */
730 0, /* tp_weaklistoffset */
731 0, /* tp_iter */
732 0, /* tp_iterext */
733 multibytecodec_methods, /* tp_methods */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000734};
735
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000736
737/**
738 * Utility functions for stateful codec mechanism
739 */
740
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000741#define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o))
742#define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o))
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000743
744static PyObject *
745encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000746 PyObject *unistr, int final)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000747{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000748 PyObject *ucvt, *r = NULL;
749 Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL;
750 Py_ssize_t datalen, origpending;
Victor Stinner4eea8492011-11-21 03:01:27 +0100751 wchar_t *data;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000752
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000753 if (PyUnicode_Check(unistr))
754 ucvt = NULL;
755 else {
756 unistr = ucvt = PyObject_Str(unistr);
757 if (unistr == NULL)
758 return NULL;
759 else if (!PyUnicode_Check(unistr)) {
760 PyErr_SetString(PyExc_TypeError,
761 "couldn't convert the object to unicode.");
762 Py_DECREF(ucvt);
763 return NULL;
764 }
765 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000766
Victor Stinner4eea8492011-11-21 03:01:27 +0100767 data = PyUnicode_AsUnicodeAndSize(unistr, &datalen);
768 if (data == NULL)
769 goto errorexit;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000770 origpending = ctx->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000771
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000772 if (origpending > 0) {
773 if (datalen > PY_SSIZE_T_MAX - ctx->pendingsize) {
774 PyErr_NoMemory();
775 /* inbuf_tmp == NULL */
776 goto errorexit;
777 }
778 inbuf_tmp = PyMem_New(Py_UNICODE, datalen + ctx->pendingsize);
779 if (inbuf_tmp == NULL)
780 goto errorexit;
781 memcpy(inbuf_tmp, ctx->pending,
782 Py_UNICODE_SIZE * ctx->pendingsize);
783 memcpy(inbuf_tmp + ctx->pendingsize,
784 PyUnicode_AS_UNICODE(unistr),
785 Py_UNICODE_SIZE * datalen);
786 datalen += ctx->pendingsize;
787 ctx->pendingsize = 0;
788 inbuf = inbuf_tmp;
789 }
790 else
791 inbuf = (Py_UNICODE *)PyUnicode_AS_UNICODE(unistr);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000792
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000793 inbuf_end = inbuf + datalen;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000794
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000795 r = multibytecodec_encode(ctx->codec, &ctx->state,
Victor Stinner6bcbef72011-05-24 22:17:55 +0200796 (const Py_UNICODE **)&inbuf, datalen,
797 ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000798 if (r == NULL) {
799 /* recover the original pending buffer */
800 if (origpending > 0)
801 memcpy(ctx->pending, inbuf_tmp,
802 Py_UNICODE_SIZE * origpending);
803 ctx->pendingsize = origpending;
804 goto errorexit;
805 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000806
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000807 if (inbuf < inbuf_end) {
808 ctx->pendingsize = (Py_ssize_t)(inbuf_end - inbuf);
809 if (ctx->pendingsize > MAXENCPENDING) {
810 /* normal codecs can't reach here */
811 ctx->pendingsize = 0;
812 PyErr_SetString(PyExc_UnicodeError,
813 "pending buffer overflow");
814 goto errorexit;
815 }
816 memcpy(ctx->pending, inbuf,
817 ctx->pendingsize * Py_UNICODE_SIZE);
818 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000819
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000820 if (inbuf_tmp != NULL)
821 PyMem_Del(inbuf_tmp);
822 Py_XDECREF(ucvt);
823 return r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000824
825errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000826 if (inbuf_tmp != NULL)
827 PyMem_Del(inbuf_tmp);
828 Py_XDECREF(r);
829 Py_XDECREF(ucvt);
830 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000831}
832
833static int
834decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000835 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000836{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000837 Py_ssize_t npendings;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000838
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000839 npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
840 if (npendings + ctx->pendingsize > MAXDECPENDING ||
841 npendings > PY_SSIZE_T_MAX - ctx->pendingsize) {
842 PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow");
843 return -1;
844 }
845 memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings);
846 ctx->pendingsize += npendings;
847 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000848}
849
850static int
851decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000852 Py_ssize_t size)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000853{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000854 buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
855 buf->inbuf_end = buf->inbuf_top + size;
856 if (buf->outobj == NULL) { /* only if outobj is not allocated yet */
857 buf->outobj = PyUnicode_FromUnicode(NULL, size);
858 if (buf->outobj == NULL)
859 return -1;
Victor Stinner4eea8492011-11-21 03:01:27 +0100860 buf->outbuf = PyUnicode_AsUnicode(buf->outobj);
861 if (buf->outbuf == NULL)
862 return -1;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000863 buf->outbuf_end = buf->outbuf +
864 PyUnicode_GET_SIZE(buf->outobj);
865 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000866
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000867 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000868}
869
870static int
871decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000872 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000873{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000874 while (buf->inbuf < buf->inbuf_end) {
875 Py_ssize_t inleft, outleft;
876 Py_ssize_t r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000877
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000878 inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
879 outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000880
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000881 r = ctx->codec->decode(&ctx->state, ctx->codec->config,
882 &buf->inbuf, inleft, &buf->outbuf, outleft);
883 if (r == 0 || r == MBERR_TOOFEW)
884 break;
885 else if (multibytecodec_decerror(ctx->codec, &ctx->state,
886 buf, ctx->errors, r))
887 return -1;
888 }
889 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000890}
891
892
893/**
894 * MultibyteIncrementalEncoder object
895 */
896
897static PyObject *
898mbiencoder_encode(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000899 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000900{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000901 PyObject *data;
902 int final = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000903
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000904 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode",
905 incrementalkwarglist, &data, &final))
906 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000907
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000908 return encoder_encode_stateful(STATEFUL_ECTX(self), data, final);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000909}
910
911static PyObject *
912mbiencoder_reset(MultibyteIncrementalEncoderObject *self)
913{
Victor Stinnere15dce32011-05-30 22:56:00 +0200914 /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */
915 unsigned char buffer[4], *outbuf;
916 Py_ssize_t r;
917 if (self->codec->encreset != NULL) {
918 outbuf = buffer;
919 r = self->codec->encreset(&self->state, self->codec->config,
920 &outbuf, sizeof(buffer));
921 if (r != 0)
922 return NULL;
923 }
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000924 self->pendingsize = 0;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000925 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000926}
927
928static struct PyMethodDef mbiencoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000929 {"encode", (PyCFunction)mbiencoder_encode,
930 METH_VARARGS | METH_KEYWORDS, NULL},
931 {"reset", (PyCFunction)mbiencoder_reset,
932 METH_NOARGS, NULL},
933 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000934};
935
936static PyObject *
937mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
938{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000939 MultibyteIncrementalEncoderObject *self;
940 PyObject *codec = NULL;
941 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000942
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000943 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
944 incnewkwarglist, &errors))
945 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000946
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000947 self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
948 if (self == NULL)
949 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000950
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000951 codec = PyObject_GetAttrString((PyObject *)type, "codec");
952 if (codec == NULL)
953 goto errorexit;
954 if (!MultibyteCodec_Check(codec)) {
955 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
956 goto errorexit;
957 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000958
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000959 self->codec = ((MultibyteCodecObject *)codec)->codec;
960 self->pendingsize = 0;
961 self->errors = internal_error_callback(errors);
962 if (self->errors == NULL)
963 goto errorexit;
964 if (self->codec->encinit != NULL &&
965 self->codec->encinit(&self->state, self->codec->config) != 0)
966 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000967
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000968 Py_DECREF(codec);
969 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000970
971errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000972 Py_XDECREF(self);
973 Py_XDECREF(codec);
974 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000975}
976
977static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000978mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
979{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000980 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000981}
982
983static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000984mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000985 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000986{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000987 if (ERROR_ISCUSTOM(self->errors))
988 Py_VISIT(self->errors);
989 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000990}
991
992static void
993mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
994{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000995 PyObject_GC_UnTrack(self);
996 ERROR_DECREF(self->errors);
997 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000998}
999
1000static PyTypeObject MultibyteIncrementalEncoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001001 PyVarObject_HEAD_INIT(NULL, 0)
1002 "MultibyteIncrementalEncoder", /* tp_name */
1003 sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
1004 0, /* tp_itemsize */
1005 /* methods */
1006 (destructor)mbiencoder_dealloc, /* tp_dealloc */
1007 0, /* tp_print */
1008 0, /* tp_getattr */
1009 0, /* tp_setattr */
1010 0, /* tp_reserved */
1011 0, /* tp_repr */
1012 0, /* tp_as_number */
1013 0, /* tp_as_sequence */
1014 0, /* tp_as_mapping */
1015 0, /* tp_hash */
1016 0, /* tp_call */
1017 0, /* tp_str */
1018 PyObject_GenericGetAttr, /* tp_getattro */
1019 0, /* tp_setattro */
1020 0, /* tp_as_buffer */
1021 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1022 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1023 0, /* tp_doc */
1024 (traverseproc)mbiencoder_traverse, /* tp_traverse */
1025 0, /* tp_clear */
1026 0, /* tp_richcompare */
1027 0, /* tp_weaklistoffset */
1028 0, /* tp_iter */
1029 0, /* tp_iterext */
1030 mbiencoder_methods, /* tp_methods */
1031 0, /* tp_members */
1032 codecctx_getsets, /* tp_getset */
1033 0, /* tp_base */
1034 0, /* tp_dict */
1035 0, /* tp_descr_get */
1036 0, /* tp_descr_set */
1037 0, /* tp_dictoffset */
1038 mbiencoder_init, /* tp_init */
1039 0, /* tp_alloc */
1040 mbiencoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001041};
1042
1043
1044/**
1045 * MultibyteIncrementalDecoder object
1046 */
1047
1048static PyObject *
1049mbidecoder_decode(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001050 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001051{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001052 MultibyteDecodeBuffer buf;
1053 char *data, *wdata = NULL;
1054 Py_buffer pdata;
1055 Py_ssize_t wsize, finalsize = 0, size, origpending;
1056 int final = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001057
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001058 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|i:decode",
1059 incrementalkwarglist, &pdata, &final))
1060 return NULL;
1061 data = pdata.buf;
1062 size = pdata.len;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001063
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001064 buf.outobj = buf.excobj = NULL;
1065 origpending = self->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001066
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001067 if (self->pendingsize == 0) {
1068 wsize = size;
1069 wdata = data;
1070 }
1071 else {
1072 if (size > PY_SSIZE_T_MAX - self->pendingsize) {
1073 PyErr_NoMemory();
1074 goto errorexit;
1075 }
1076 wsize = size + self->pendingsize;
1077 wdata = PyMem_Malloc(wsize);
1078 if (wdata == NULL)
1079 goto errorexit;
1080 memcpy(wdata, self->pending, self->pendingsize);
1081 memcpy(wdata + self->pendingsize, data, size);
1082 self->pendingsize = 0;
1083 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001084
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001085 if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
1086 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001087
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001088 if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
1089 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001090
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001091 if (final && buf.inbuf < buf.inbuf_end) {
1092 if (multibytecodec_decerror(self->codec, &self->state,
1093 &buf, self->errors, MBERR_TOOFEW)) {
1094 /* recover the original pending buffer */
1095 memcpy(self->pending, wdata, origpending);
1096 self->pendingsize = origpending;
1097 goto errorexit;
1098 }
1099 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001100
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001101 if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
1102 if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
1103 goto errorexit;
1104 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001105
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001106 finalsize = (Py_ssize_t)(buf.outbuf - PyUnicode_AS_UNICODE(buf.outobj));
1107 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1108 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1109 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001110
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001111 PyBuffer_Release(&pdata);
1112 if (wdata != data)
1113 PyMem_Del(wdata);
1114 Py_XDECREF(buf.excobj);
1115 return buf.outobj;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001116
1117errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001118 PyBuffer_Release(&pdata);
1119 if (wdata != NULL && wdata != data)
1120 PyMem_Del(wdata);
1121 Py_XDECREF(buf.excobj);
1122 Py_XDECREF(buf.outobj);
1123 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001124}
1125
1126static PyObject *
1127mbidecoder_reset(MultibyteIncrementalDecoderObject *self)
1128{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001129 if (self->codec->decreset != NULL &&
1130 self->codec->decreset(&self->state, self->codec->config) != 0)
1131 return NULL;
1132 self->pendingsize = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001133
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001134 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001135}
1136
1137static struct PyMethodDef mbidecoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001138 {"decode", (PyCFunction)mbidecoder_decode,
1139 METH_VARARGS | METH_KEYWORDS, NULL},
1140 {"reset", (PyCFunction)mbidecoder_reset,
1141 METH_NOARGS, NULL},
1142 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001143};
1144
1145static PyObject *
1146mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1147{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001148 MultibyteIncrementalDecoderObject *self;
1149 PyObject *codec = NULL;
1150 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001151
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001152 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
1153 incnewkwarglist, &errors))
1154 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001155
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001156 self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
1157 if (self == NULL)
1158 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001159
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001160 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1161 if (codec == NULL)
1162 goto errorexit;
1163 if (!MultibyteCodec_Check(codec)) {
1164 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1165 goto errorexit;
1166 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001167
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001168 self->codec = ((MultibyteCodecObject *)codec)->codec;
1169 self->pendingsize = 0;
1170 self->errors = internal_error_callback(errors);
1171 if (self->errors == NULL)
1172 goto errorexit;
1173 if (self->codec->decinit != NULL &&
1174 self->codec->decinit(&self->state, self->codec->config) != 0)
1175 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001176
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001177 Py_DECREF(codec);
1178 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001179
1180errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001181 Py_XDECREF(self);
1182 Py_XDECREF(codec);
1183 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001184}
1185
1186static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001187mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
1188{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001189 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001190}
1191
1192static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001193mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001194 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001195{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001196 if (ERROR_ISCUSTOM(self->errors))
1197 Py_VISIT(self->errors);
1198 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001199}
1200
1201static void
1202mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
1203{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001204 PyObject_GC_UnTrack(self);
1205 ERROR_DECREF(self->errors);
1206 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001207}
1208
1209static PyTypeObject MultibyteIncrementalDecoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001210 PyVarObject_HEAD_INIT(NULL, 0)
1211 "MultibyteIncrementalDecoder", /* tp_name */
1212 sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */
1213 0, /* tp_itemsize */
1214 /* methods */
1215 (destructor)mbidecoder_dealloc, /* tp_dealloc */
1216 0, /* tp_print */
1217 0, /* tp_getattr */
1218 0, /* tp_setattr */
1219 0, /* tp_reserved */
1220 0, /* tp_repr */
1221 0, /* tp_as_number */
1222 0, /* tp_as_sequence */
1223 0, /* tp_as_mapping */
1224 0, /* tp_hash */
1225 0, /* tp_call */
1226 0, /* tp_str */
1227 PyObject_GenericGetAttr, /* tp_getattro */
1228 0, /* tp_setattro */
1229 0, /* tp_as_buffer */
1230 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1231 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1232 0, /* tp_doc */
1233 (traverseproc)mbidecoder_traverse, /* tp_traverse */
1234 0, /* tp_clear */
1235 0, /* tp_richcompare */
1236 0, /* tp_weaklistoffset */
1237 0, /* tp_iter */
1238 0, /* tp_iterext */
1239 mbidecoder_methods, /* tp_methods */
1240 0, /* tp_members */
1241 codecctx_getsets, /* tp_getset */
1242 0, /* tp_base */
1243 0, /* tp_dict */
1244 0, /* tp_descr_get */
1245 0, /* tp_descr_set */
1246 0, /* tp_dictoffset */
1247 mbidecoder_init, /* tp_init */
1248 0, /* tp_alloc */
1249 mbidecoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001250};
1251
1252
1253/**
1254 * MultibyteStreamReader object
1255 */
1256
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001257static PyObject *
1258mbstreamreader_iread(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001259 const char *method, Py_ssize_t sizehint)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001260{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001261 MultibyteDecodeBuffer buf;
1262 PyObject *cres;
1263 Py_ssize_t rsize, finalsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001264
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001265 if (sizehint == 0)
1266 return PyUnicode_FromUnicode(NULL, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001267
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001268 buf.outobj = buf.excobj = NULL;
1269 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001270
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001271 for (;;) {
1272 int endoffile;
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001273
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001274 if (sizehint < 0)
1275 cres = PyObject_CallMethod(self->stream,
1276 (char *)method, NULL);
1277 else
1278 cres = PyObject_CallMethod(self->stream,
1279 (char *)method, "i", sizehint);
1280 if (cres == NULL)
1281 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001282
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001283 if (!PyBytes_Check(cres)) {
1284 PyErr_Format(PyExc_TypeError,
1285 "stream function returned a "
1286 "non-bytes object (%.100s)",
1287 cres->ob_type->tp_name);
1288 goto errorexit;
1289 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001290
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001291 endoffile = (PyBytes_GET_SIZE(cres) == 0);
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001292
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001293 if (self->pendingsize > 0) {
1294 PyObject *ctr;
1295 char *ctrdata;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001296
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001297 if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
1298 PyErr_NoMemory();
1299 goto errorexit;
1300 }
1301 rsize = PyBytes_GET_SIZE(cres) + self->pendingsize;
1302 ctr = PyBytes_FromStringAndSize(NULL, rsize);
1303 if (ctr == NULL)
1304 goto errorexit;
1305 ctrdata = PyBytes_AS_STRING(ctr);
1306 memcpy(ctrdata, self->pending, self->pendingsize);
1307 memcpy(ctrdata + self->pendingsize,
1308 PyBytes_AS_STRING(cres),
1309 PyBytes_GET_SIZE(cres));
1310 Py_DECREF(cres);
1311 cres = ctr;
1312 self->pendingsize = 0;
1313 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001314
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001315 rsize = PyBytes_GET_SIZE(cres);
1316 if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres),
1317 rsize) != 0)
1318 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001319
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001320 if (rsize > 0 && decoder_feed_buffer(
1321 (MultibyteStatefulDecoderContext *)self, &buf))
1322 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001323
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001324 if (endoffile || sizehint < 0) {
1325 if (buf.inbuf < buf.inbuf_end &&
1326 multibytecodec_decerror(self->codec, &self->state,
1327 &buf, self->errors, MBERR_TOOFEW))
1328 goto errorexit;
1329 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001330
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001331 if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
1332 if (decoder_append_pending(STATEFUL_DCTX(self),
1333 &buf) != 0)
1334 goto errorexit;
1335 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001336
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001337 finalsize = (Py_ssize_t)(buf.outbuf -
1338 PyUnicode_AS_UNICODE(buf.outobj));
1339 Py_DECREF(cres);
1340 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001341
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001342 if (sizehint < 0 || finalsize != 0 || rsize == 0)
1343 break;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001344
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001345 sizehint = 1; /* read 1 more byte and retry */
1346 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001347
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001348 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1349 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1350 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001351
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001352 Py_XDECREF(cres);
1353 Py_XDECREF(buf.excobj);
1354 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001355
1356errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001357 Py_XDECREF(cres);
1358 Py_XDECREF(buf.excobj);
1359 Py_XDECREF(buf.outobj);
1360 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001361}
1362
1363static PyObject *
1364mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args)
1365{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001366 PyObject *sizeobj = NULL;
1367 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001368
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001369 if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj))
1370 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001371
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001372 if (sizeobj == Py_None || sizeobj == NULL)
1373 size = -1;
1374 else if (PyLong_Check(sizeobj))
1375 size = PyLong_AsSsize_t(sizeobj);
1376 else {
1377 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1378 return NULL;
1379 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001380
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001381 if (size == -1 && PyErr_Occurred())
1382 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001383
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001384 return mbstreamreader_iread(self, "read", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001385}
1386
1387static PyObject *
1388mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args)
1389{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001390 PyObject *sizeobj = NULL;
1391 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001392
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001393 if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj))
1394 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001395
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001396 if (sizeobj == Py_None || sizeobj == NULL)
1397 size = -1;
1398 else if (PyLong_Check(sizeobj))
1399 size = PyLong_AsSsize_t(sizeobj);
1400 else {
1401 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1402 return NULL;
1403 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001404
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001405 if (size == -1 && PyErr_Occurred())
1406 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001407
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001408 return mbstreamreader_iread(self, "readline", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001409}
1410
1411static PyObject *
1412mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args)
1413{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001414 PyObject *sizehintobj = NULL, *r, *sr;
1415 Py_ssize_t sizehint;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001416
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001417 if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj))
1418 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001419
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001420 if (sizehintobj == Py_None || sizehintobj == NULL)
1421 sizehint = -1;
1422 else if (PyLong_Check(sizehintobj))
1423 sizehint = PyLong_AsSsize_t(sizehintobj);
1424 else {
1425 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1426 return NULL;
1427 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001428
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001429 if (sizehint == -1 && PyErr_Occurred())
1430 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001431
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001432 r = mbstreamreader_iread(self, "read", sizehint);
1433 if (r == NULL)
1434 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001435
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001436 sr = PyUnicode_Splitlines(r, 1);
1437 Py_DECREF(r);
1438 return sr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001439}
1440
1441static PyObject *
1442mbstreamreader_reset(MultibyteStreamReaderObject *self)
1443{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001444 if (self->codec->decreset != NULL &&
1445 self->codec->decreset(&self->state, self->codec->config) != 0)
1446 return NULL;
1447 self->pendingsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001448
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001449 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001450}
1451
1452static struct PyMethodDef mbstreamreader_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001453 {"read", (PyCFunction)mbstreamreader_read,
1454 METH_VARARGS, NULL},
1455 {"readline", (PyCFunction)mbstreamreader_readline,
1456 METH_VARARGS, NULL},
1457 {"readlines", (PyCFunction)mbstreamreader_readlines,
1458 METH_VARARGS, NULL},
1459 {"reset", (PyCFunction)mbstreamreader_reset,
1460 METH_NOARGS, NULL},
1461 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001462};
1463
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001464static PyMemberDef mbstreamreader_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001465 {"stream", T_OBJECT,
1466 offsetof(MultibyteStreamReaderObject, stream),
1467 READONLY, NULL},
1468 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001469};
1470
1471static PyObject *
1472mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1473{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001474 MultibyteStreamReaderObject *self;
1475 PyObject *stream, *codec = NULL;
1476 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001477
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001478 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
1479 streamkwarglist, &stream, &errors))
1480 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001481
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001482 self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
1483 if (self == NULL)
1484 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001485
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001486 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1487 if (codec == NULL)
1488 goto errorexit;
1489 if (!MultibyteCodec_Check(codec)) {
1490 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1491 goto errorexit;
1492 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001493
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001494 self->codec = ((MultibyteCodecObject *)codec)->codec;
1495 self->stream = stream;
1496 Py_INCREF(stream);
1497 self->pendingsize = 0;
1498 self->errors = internal_error_callback(errors);
1499 if (self->errors == NULL)
1500 goto errorexit;
1501 if (self->codec->decinit != NULL &&
1502 self->codec->decinit(&self->state, self->codec->config) != 0)
1503 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001504
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001505 Py_DECREF(codec);
1506 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001507
1508errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001509 Py_XDECREF(self);
1510 Py_XDECREF(codec);
1511 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001512}
1513
1514static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001515mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
1516{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001517 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001518}
1519
1520static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001521mbstreamreader_traverse(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001522 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001523{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001524 if (ERROR_ISCUSTOM(self->errors))
1525 Py_VISIT(self->errors);
1526 Py_VISIT(self->stream);
1527 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001528}
1529
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001530static void
1531mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
1532{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001533 PyObject_GC_UnTrack(self);
1534 ERROR_DECREF(self->errors);
1535 Py_XDECREF(self->stream);
1536 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001537}
1538
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001539static PyTypeObject MultibyteStreamReader_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001540 PyVarObject_HEAD_INIT(NULL, 0)
1541 "MultibyteStreamReader", /* tp_name */
1542 sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
1543 0, /* tp_itemsize */
1544 /* methods */
1545 (destructor)mbstreamreader_dealloc, /* tp_dealloc */
1546 0, /* tp_print */
1547 0, /* tp_getattr */
1548 0, /* tp_setattr */
1549 0, /* tp_reserved */
1550 0, /* tp_repr */
1551 0, /* tp_as_number */
1552 0, /* tp_as_sequence */
1553 0, /* tp_as_mapping */
1554 0, /* tp_hash */
1555 0, /* tp_call */
1556 0, /* tp_str */
1557 PyObject_GenericGetAttr, /* tp_getattro */
1558 0, /* tp_setattro */
1559 0, /* tp_as_buffer */
1560 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1561 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1562 0, /* tp_doc */
1563 (traverseproc)mbstreamreader_traverse, /* tp_traverse */
1564 0, /* tp_clear */
1565 0, /* tp_richcompare */
1566 0, /* tp_weaklistoffset */
1567 0, /* tp_iter */
1568 0, /* tp_iterext */
1569 mbstreamreader_methods, /* tp_methods */
1570 mbstreamreader_members, /* tp_members */
1571 codecctx_getsets, /* tp_getset */
1572 0, /* tp_base */
1573 0, /* tp_dict */
1574 0, /* tp_descr_get */
1575 0, /* tp_descr_set */
1576 0, /* tp_dictoffset */
1577 mbstreamreader_init, /* tp_init */
1578 0, /* tp_alloc */
1579 mbstreamreader_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001580};
1581
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001582
1583/**
1584 * MultibyteStreamWriter object
1585 */
1586
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001587static int
1588mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001589 PyObject *unistr)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001590{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001591 PyObject *str, *wr;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001592 _Py_IDENTIFIER(write);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001593
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001594 str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
1595 if (str == NULL)
1596 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001597
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001598 wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001599 Py_DECREF(str);
1600 if (wr == NULL)
1601 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001602
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001603 Py_DECREF(wr);
1604 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001605}
1606
1607static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001608mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001609{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001610 if (mbstreamwriter_iwrite(self, strobj))
1611 return NULL;
1612 else
1613 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001614}
1615
1616static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001617mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001618{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001619 PyObject *strobj;
1620 int i, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001621
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001622 if (!PySequence_Check(lines)) {
1623 PyErr_SetString(PyExc_TypeError,
1624 "arg must be a sequence object");
1625 return NULL;
1626 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001627
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001628 for (i = 0; i < PySequence_Length(lines); i++) {
1629 /* length can be changed even within this loop */
1630 strobj = PySequence_GetItem(lines, i);
1631 if (strobj == NULL)
1632 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001633
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001634 r = mbstreamwriter_iwrite(self, strobj);
1635 Py_DECREF(strobj);
1636 if (r == -1)
1637 return NULL;
1638 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001639
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001640 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001641}
1642
1643static PyObject *
1644mbstreamwriter_reset(MultibyteStreamWriterObject *self)
1645{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001646 const Py_UNICODE *pending;
1647 PyObject *pwrt;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001648
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001649 pending = self->pending;
1650 pwrt = multibytecodec_encode(self->codec, &self->state,
1651 &pending, self->pendingsize, self->errors,
1652 MBENC_FLUSH | MBENC_RESET);
1653 /* some pending buffer can be truncated when UnicodeEncodeError is
1654 * raised on 'strict' mode. but, 'reset' method is designed to
1655 * reset the pending buffer or states so failed string sequence
1656 * ought to be missed */
1657 self->pendingsize = 0;
1658 if (pwrt == NULL)
1659 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001660
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001661 assert(PyBytes_Check(pwrt));
1662 if (PyBytes_Size(pwrt) > 0) {
1663 PyObject *wr;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001664 _Py_IDENTIFIER(write);
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001665
1666 wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001667 if (wr == NULL) {
1668 Py_DECREF(pwrt);
1669 return NULL;
1670 }
1671 }
1672 Py_DECREF(pwrt);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001673
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001674 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001675}
1676
1677static PyObject *
1678mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1679{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001680 MultibyteStreamWriterObject *self;
1681 PyObject *stream, *codec = NULL;
1682 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001683
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001684 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
1685 streamkwarglist, &stream, &errors))
1686 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001687
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001688 self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
1689 if (self == NULL)
1690 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001691
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001692 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1693 if (codec == NULL)
1694 goto errorexit;
1695 if (!MultibyteCodec_Check(codec)) {
1696 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1697 goto errorexit;
1698 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001699
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001700 self->codec = ((MultibyteCodecObject *)codec)->codec;
1701 self->stream = stream;
1702 Py_INCREF(stream);
1703 self->pendingsize = 0;
1704 self->errors = internal_error_callback(errors);
1705 if (self->errors == NULL)
1706 goto errorexit;
1707 if (self->codec->encinit != NULL &&
1708 self->codec->encinit(&self->state, self->codec->config) != 0)
1709 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001710
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001711 Py_DECREF(codec);
1712 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001713
1714errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001715 Py_XDECREF(self);
1716 Py_XDECREF(codec);
1717 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001718}
1719
1720static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001721mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
1722{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001723 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001724}
1725
1726static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001727mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001728 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001729{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001730 if (ERROR_ISCUSTOM(self->errors))
1731 Py_VISIT(self->errors);
1732 Py_VISIT(self->stream);
1733 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001734}
1735
1736static void
1737mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
1738{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001739 PyObject_GC_UnTrack(self);
1740 ERROR_DECREF(self->errors);
1741 Py_XDECREF(self->stream);
1742 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001743}
1744
1745static struct PyMethodDef mbstreamwriter_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001746 {"write", (PyCFunction)mbstreamwriter_write,
1747 METH_O, NULL},
1748 {"writelines", (PyCFunction)mbstreamwriter_writelines,
1749 METH_O, NULL},
1750 {"reset", (PyCFunction)mbstreamwriter_reset,
1751 METH_NOARGS, NULL},
1752 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001753};
1754
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001755static PyMemberDef mbstreamwriter_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001756 {"stream", T_OBJECT,
1757 offsetof(MultibyteStreamWriterObject, stream),
1758 READONLY, NULL},
1759 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001760};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001761
1762static PyTypeObject MultibyteStreamWriter_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001763 PyVarObject_HEAD_INIT(NULL, 0)
1764 "MultibyteStreamWriter", /* tp_name */
1765 sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
1766 0, /* tp_itemsize */
1767 /* methods */
1768 (destructor)mbstreamwriter_dealloc, /* tp_dealloc */
1769 0, /* tp_print */
1770 0, /* tp_getattr */
1771 0, /* tp_setattr */
1772 0, /* tp_reserved */
1773 0, /* tp_repr */
1774 0, /* tp_as_number */
1775 0, /* tp_as_sequence */
1776 0, /* tp_as_mapping */
1777 0, /* tp_hash */
1778 0, /* tp_call */
1779 0, /* tp_str */
1780 PyObject_GenericGetAttr, /* tp_getattro */
1781 0, /* tp_setattro */
1782 0, /* tp_as_buffer */
1783 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1784 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1785 0, /* tp_doc */
1786 (traverseproc)mbstreamwriter_traverse, /* tp_traverse */
1787 0, /* tp_clear */
1788 0, /* tp_richcompare */
1789 0, /* tp_weaklistoffset */
1790 0, /* tp_iter */
1791 0, /* tp_iterext */
1792 mbstreamwriter_methods, /* tp_methods */
1793 mbstreamwriter_members, /* tp_members */
1794 codecctx_getsets, /* tp_getset */
1795 0, /* tp_base */
1796 0, /* tp_dict */
1797 0, /* tp_descr_get */
1798 0, /* tp_descr_set */
1799 0, /* tp_dictoffset */
1800 mbstreamwriter_init, /* tp_init */
1801 0, /* tp_alloc */
1802 mbstreamwriter_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001803};
1804
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001805
1806/**
1807 * Exposed factory function
1808 */
1809
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001810static PyObject *
1811__create_codec(PyObject *ignore, PyObject *arg)
1812{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001813 MultibyteCodecObject *self;
1814 MultibyteCodec *codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001815
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001816 if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
1817 PyErr_SetString(PyExc_ValueError, "argument type invalid");
1818 return NULL;
1819 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001820
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001821 codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
1822 if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
1823 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001824
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001825 self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
1826 if (self == NULL)
1827 return NULL;
1828 self->codec = codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001829
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001830 return (PyObject *)self;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001831}
1832
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001833static struct PyMethodDef __methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001834 {"__create_codec", (PyCFunction)__create_codec, METH_O},
1835 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001836};
1837
Martin v. Löwis1a214512008-06-11 05:26:20 +00001838
1839static struct PyModuleDef _multibytecodecmodule = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001840 PyModuleDef_HEAD_INIT,
1841 "_multibytecodec",
1842 NULL,
1843 -1,
1844 __methods,
1845 NULL,
1846 NULL,
1847 NULL,
1848 NULL
Martin v. Löwis1a214512008-06-11 05:26:20 +00001849};
1850
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001851PyMODINIT_FUNC
Martin v. Löwis1a214512008-06-11 05:26:20 +00001852PyInit__multibytecodec(void)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001853{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001854 int i;
1855 PyObject *m;
1856 PyTypeObject *typelist[] = {
1857 &MultibyteIncrementalEncoder_Type,
1858 &MultibyteIncrementalDecoder_Type,
1859 &MultibyteStreamReader_Type,
1860 &MultibyteStreamWriter_Type,
1861 NULL
1862 };
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001863
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001864 if (PyType_Ready(&MultibyteCodec_Type) < 0)
1865 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001866
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001867 m = PyModule_Create(&_multibytecodecmodule);
1868 if (m == NULL)
1869 return NULL;
Neal Norwitz058bde12005-09-21 06:44:25 +00001870
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001871 for (i = 0; typelist[i] != NULL; i++) {
1872 if (PyType_Ready(typelist[i]) < 0)
1873 return NULL;
1874 Py_INCREF(typelist[i]);
1875 PyModule_AddObject(m, typelist[i]->tp_name,
1876 (PyObject *)typelist[i]);
1877 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001878
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001879 if (PyErr_Occurred()) {
1880 Py_FatalError("can't initialize the _multibytecodec module");
1881 Py_DECREF(m);
1882 m = NULL;
1883 }
1884 return m;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001885}