blob: 6b2352373c85d64b2e359d3ef3746574dc679a21 [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
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000446 retunisize = PyUnicode_GET_SIZE(retuni);
447 if (retunisize > 0) {
448 REQUIRE_DECODEBUFFER(buf, retunisize);
449 memcpy((char *)buf->outbuf, PyUnicode_AS_DATA(retuni),
450 retunisize * Py_UNICODE_SIZE);
451 buf->outbuf += retunisize;
452 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000453
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000454 newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
455 if (newpos < 0 && !PyErr_Occurred())
456 newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
457 if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
458 PyErr_Clear();
459 PyErr_Format(PyExc_IndexError,
460 "position %zd from error handler out of bounds",
461 newpos);
462 goto errorexit;
463 }
464 buf->inbuf = buf->inbuf_top + newpos;
465 Py_DECREF(retobj);
466 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000467
468errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000469 Py_XDECREF(retobj);
470 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000471}
472
473static PyObject *
474multibytecodec_encode(MultibyteCodec *codec,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000475 MultibyteCodec_State *state,
476 const Py_UNICODE **data, Py_ssize_t datalen,
477 PyObject *errors, int flags)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000478{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000479 MultibyteEncodeBuffer buf;
480 Py_ssize_t finalsize, r = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000481
Victor Stinner6bcbef72011-05-24 22:17:55 +0200482 if (datalen == 0 && !(flags & MBENC_RESET))
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000483 return PyBytes_FromStringAndSize(NULL, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000484
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000485 buf.excobj = NULL;
Brett Cannonb94767f2011-02-22 20:15:44 +0000486 buf.outobj = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000487 buf.inbuf = buf.inbuf_top = *data;
488 buf.inbuf_end = buf.inbuf_top + datalen;
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000489
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000490 if (datalen > (PY_SSIZE_T_MAX - 16) / 2) {
491 PyErr_NoMemory();
492 goto errorexit;
493 }
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000494
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000495 buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16);
496 if (buf.outobj == NULL)
497 goto errorexit;
498 buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj);
499 buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000500
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000501 while (buf.inbuf < buf.inbuf_end) {
502 Py_ssize_t inleft, outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000503
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000504 /* we don't reuse inleft and outleft here.
505 * error callbacks can relocate the cursor anywhere on buffer*/
506 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
507 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
508 r = codec->encode(state, codec->config, &buf.inbuf, inleft,
509 &buf.outbuf, outleft, flags);
510 if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH)))
511 break;
512 else if (multibytecodec_encerror(codec, state, &buf, errors,r))
513 goto errorexit;
514 else if (r == MBERR_TOOFEW)
515 break;
516 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000517
Victor Stinner6bcbef72011-05-24 22:17:55 +0200518 if (codec->encreset != NULL && (flags & MBENC_RESET))
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000519 for (;;) {
520 Py_ssize_t outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000521
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000522 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
523 r = codec->encreset(state, codec->config, &buf.outbuf,
524 outleft);
525 if (r == 0)
526 break;
527 else if (multibytecodec_encerror(codec, state,
528 &buf, errors, r))
529 goto errorexit;
530 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000531
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000532 finalsize = (Py_ssize_t)((char *)buf.outbuf -
533 PyBytes_AS_STRING(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000534
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000535 if (finalsize != PyBytes_GET_SIZE(buf.outobj))
536 if (_PyBytes_Resize(&buf.outobj, finalsize) == -1)
537 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000538
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000539 *data = buf.inbuf;
540 Py_XDECREF(buf.excobj);
541 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000542
543errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000544 Py_XDECREF(buf.excobj);
545 Py_XDECREF(buf.outobj);
546 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000547}
548
549static PyObject *
550MultibyteCodec_Encode(MultibyteCodecObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000551 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000552{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000553 MultibyteCodec_State state;
554 Py_UNICODE *data;
555 PyObject *errorcb, *r, *arg, *ucvt;
556 const char *errors = NULL;
557 Py_ssize_t datalen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000558
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000559 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode",
560 codeckwarglist, &arg, &errors))
561 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000562
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000563 if (PyUnicode_Check(arg))
564 ucvt = NULL;
565 else {
566 arg = ucvt = PyObject_Str(arg);
567 if (arg == NULL)
568 return NULL;
569 else if (!PyUnicode_Check(arg)) {
570 PyErr_SetString(PyExc_TypeError,
571 "couldn't convert the object to unicode.");
572 Py_DECREF(ucvt);
573 return NULL;
574 }
575 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000576
Victor Stinner9a80fab2011-11-21 02:50:14 +0100577 data = PyUnicode_AsUnicodeAndSize(arg, &datalen);
578 if (data == NULL) {
579 Py_XDECREF(ucvt);
580 return NULL;
581 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000582
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000583 errorcb = internal_error_callback(errors);
584 if (errorcb == NULL) {
585 Py_XDECREF(ucvt);
586 return NULL;
587 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000588
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000589 if (self->codec->encinit != NULL &&
590 self->codec->encinit(&state, self->codec->config) != 0)
591 goto errorexit;
592 r = multibytecodec_encode(self->codec, &state,
593 (const Py_UNICODE **)&data, datalen, errorcb,
594 MBENC_FLUSH | MBENC_RESET);
595 if (r == NULL)
596 goto errorexit;
Hye-Shik Chang2bb146f2004-07-18 03:06:29 +0000597
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000598 ERROR_DECREF(errorcb);
599 Py_XDECREF(ucvt);
600 return make_tuple(r, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000601
602errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000603 ERROR_DECREF(errorcb);
604 Py_XDECREF(ucvt);
605 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000606}
607
608static PyObject *
609MultibyteCodec_Decode(MultibyteCodecObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000610 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000611{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000612 MultibyteCodec_State state;
613 MultibyteDecodeBuffer buf;
614 PyObject *errorcb;
615 Py_buffer pdata;
616 const char *data, *errors = NULL;
617 Py_ssize_t datalen, finalsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000618
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000619 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|z:decode",
620 codeckwarglist, &pdata, &errors))
621 return NULL;
622 data = pdata.buf;
623 datalen = pdata.len;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000624
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000625 errorcb = internal_error_callback(errors);
626 if (errorcb == NULL) {
627 PyBuffer_Release(&pdata);
628 return NULL;
629 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000630
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000631 if (datalen == 0) {
632 PyBuffer_Release(&pdata);
633 ERROR_DECREF(errorcb);
634 return make_tuple(PyUnicode_FromUnicode(NULL, 0), 0);
635 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000636
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000637 buf.excobj = NULL;
638 buf.inbuf = buf.inbuf_top = (unsigned char *)data;
639 buf.inbuf_end = buf.inbuf_top + datalen;
640 buf.outobj = PyUnicode_FromUnicode(NULL, datalen);
641 if (buf.outobj == NULL)
642 goto errorexit;
643 buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj);
644 buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000645
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000646 if (self->codec->decinit != NULL &&
647 self->codec->decinit(&state, self->codec->config) != 0)
648 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000649
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000650 while (buf.inbuf < buf.inbuf_end) {
651 Py_ssize_t inleft, outleft, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000652
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000653 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
654 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000655
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000656 r = self->codec->decode(&state, self->codec->config,
657 &buf.inbuf, inleft, &buf.outbuf, outleft);
658 if (r == 0)
659 break;
660 else if (multibytecodec_decerror(self->codec, &state,
661 &buf, errorcb, r))
662 goto errorexit;
663 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000664
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000665 finalsize = (Py_ssize_t)(buf.outbuf -
666 PyUnicode_AS_UNICODE(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000667
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000668 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
669 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
670 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000671
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000672 PyBuffer_Release(&pdata);
673 Py_XDECREF(buf.excobj);
674 ERROR_DECREF(errorcb);
675 return make_tuple(buf.outobj, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000676
677errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000678 PyBuffer_Release(&pdata);
679 ERROR_DECREF(errorcb);
680 Py_XDECREF(buf.excobj);
681 Py_XDECREF(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000682
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000683 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000684}
685
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000686static struct PyMethodDef multibytecodec_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000687 {"encode", (PyCFunction)MultibyteCodec_Encode,
688 METH_VARARGS | METH_KEYWORDS,
689 MultibyteCodec_Encode__doc__},
690 {"decode", (PyCFunction)MultibyteCodec_Decode,
691 METH_VARARGS | METH_KEYWORDS,
692 MultibyteCodec_Decode__doc__},
693 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000694};
695
696static void
697multibytecodec_dealloc(MultibyteCodecObject *self)
698{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000699 PyObject_Del(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000700}
701
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000702static PyTypeObject MultibyteCodec_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000703 PyVarObject_HEAD_INIT(NULL, 0)
704 "MultibyteCodec", /* tp_name */
705 sizeof(MultibyteCodecObject), /* tp_basicsize */
706 0, /* tp_itemsize */
707 /* methods */
708 (destructor)multibytecodec_dealloc, /* tp_dealloc */
709 0, /* tp_print */
710 0, /* tp_getattr */
711 0, /* tp_setattr */
712 0, /* tp_reserved */
713 0, /* tp_repr */
714 0, /* tp_as_number */
715 0, /* tp_as_sequence */
716 0, /* tp_as_mapping */
717 0, /* tp_hash */
718 0, /* tp_call */
719 0, /* tp_str */
720 PyObject_GenericGetAttr, /* tp_getattro */
721 0, /* tp_setattro */
722 0, /* tp_as_buffer */
723 Py_TPFLAGS_DEFAULT, /* tp_flags */
724 0, /* tp_doc */
725 0, /* tp_traverse */
726 0, /* tp_clear */
727 0, /* tp_richcompare */
728 0, /* tp_weaklistoffset */
729 0, /* tp_iter */
730 0, /* tp_iterext */
731 multibytecodec_methods, /* tp_methods */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000732};
733
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000734
735/**
736 * Utility functions for stateful codec mechanism
737 */
738
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000739#define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o))
740#define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o))
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000741
742static PyObject *
743encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000744 PyObject *unistr, int final)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000745{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000746 PyObject *ucvt, *r = NULL;
747 Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL;
748 Py_ssize_t datalen, origpending;
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,
758 "couldn't convert the object to unicode.");
759 Py_DECREF(ucvt);
760 return NULL;
761 }
762 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000763
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000764 datalen = PyUnicode_GET_SIZE(unistr);
765 origpending = ctx->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000766
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000767 if (origpending > 0) {
768 if (datalen > PY_SSIZE_T_MAX - ctx->pendingsize) {
769 PyErr_NoMemory();
770 /* inbuf_tmp == NULL */
771 goto errorexit;
772 }
773 inbuf_tmp = PyMem_New(Py_UNICODE, datalen + ctx->pendingsize);
774 if (inbuf_tmp == NULL)
775 goto errorexit;
776 memcpy(inbuf_tmp, ctx->pending,
777 Py_UNICODE_SIZE * ctx->pendingsize);
778 memcpy(inbuf_tmp + ctx->pendingsize,
779 PyUnicode_AS_UNICODE(unistr),
780 Py_UNICODE_SIZE * datalen);
781 datalen += ctx->pendingsize;
782 ctx->pendingsize = 0;
783 inbuf = inbuf_tmp;
784 }
785 else
786 inbuf = (Py_UNICODE *)PyUnicode_AS_UNICODE(unistr);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000787
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000788 inbuf_end = inbuf + datalen;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000789
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000790 r = multibytecodec_encode(ctx->codec, &ctx->state,
Victor Stinner6bcbef72011-05-24 22:17:55 +0200791 (const Py_UNICODE **)&inbuf, datalen,
792 ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000793 if (r == NULL) {
794 /* recover the original pending buffer */
795 if (origpending > 0)
796 memcpy(ctx->pending, inbuf_tmp,
797 Py_UNICODE_SIZE * origpending);
798 ctx->pendingsize = origpending;
799 goto errorexit;
800 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000801
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000802 if (inbuf < inbuf_end) {
803 ctx->pendingsize = (Py_ssize_t)(inbuf_end - inbuf);
804 if (ctx->pendingsize > MAXENCPENDING) {
805 /* normal codecs can't reach here */
806 ctx->pendingsize = 0;
807 PyErr_SetString(PyExc_UnicodeError,
808 "pending buffer overflow");
809 goto errorexit;
810 }
811 memcpy(ctx->pending, inbuf,
812 ctx->pendingsize * Py_UNICODE_SIZE);
813 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000814
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000815 if (inbuf_tmp != NULL)
816 PyMem_Del(inbuf_tmp);
817 Py_XDECREF(ucvt);
818 return r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000819
820errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000821 if (inbuf_tmp != NULL)
822 PyMem_Del(inbuf_tmp);
823 Py_XDECREF(r);
824 Py_XDECREF(ucvt);
825 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000826}
827
828static int
829decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000830 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000831{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000832 Py_ssize_t npendings;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000833
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000834 npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
835 if (npendings + ctx->pendingsize > MAXDECPENDING ||
836 npendings > PY_SSIZE_T_MAX - ctx->pendingsize) {
837 PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow");
838 return -1;
839 }
840 memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings);
841 ctx->pendingsize += npendings;
842 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000843}
844
845static int
846decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000847 Py_ssize_t size)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000848{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000849 buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
850 buf->inbuf_end = buf->inbuf_top + size;
851 if (buf->outobj == NULL) { /* only if outobj is not allocated yet */
852 buf->outobj = PyUnicode_FromUnicode(NULL, size);
853 if (buf->outobj == NULL)
854 return -1;
855 buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj);
856 buf->outbuf_end = buf->outbuf +
857 PyUnicode_GET_SIZE(buf->outobj);
858 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000859
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000860 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000861}
862
863static int
864decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000865 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000866{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000867 while (buf->inbuf < buf->inbuf_end) {
868 Py_ssize_t inleft, outleft;
869 Py_ssize_t r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000870
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000871 inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
872 outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000873
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000874 r = ctx->codec->decode(&ctx->state, ctx->codec->config,
875 &buf->inbuf, inleft, &buf->outbuf, outleft);
876 if (r == 0 || r == MBERR_TOOFEW)
877 break;
878 else if (multibytecodec_decerror(ctx->codec, &ctx->state,
879 buf, ctx->errors, r))
880 return -1;
881 }
882 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000883}
884
885
886/**
887 * MultibyteIncrementalEncoder object
888 */
889
890static PyObject *
891mbiencoder_encode(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000892 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000893{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000894 PyObject *data;
895 int final = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000896
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000897 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode",
898 incrementalkwarglist, &data, &final))
899 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000900
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000901 return encoder_encode_stateful(STATEFUL_ECTX(self), data, final);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000902}
903
904static PyObject *
905mbiencoder_reset(MultibyteIncrementalEncoderObject *self)
906{
Victor Stinnere15dce32011-05-30 22:56:00 +0200907 /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */
908 unsigned char buffer[4], *outbuf;
909 Py_ssize_t r;
910 if (self->codec->encreset != NULL) {
911 outbuf = buffer;
912 r = self->codec->encreset(&self->state, self->codec->config,
913 &outbuf, sizeof(buffer));
914 if (r != 0)
915 return NULL;
916 }
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000917 self->pendingsize = 0;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000918 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000919}
920
921static struct PyMethodDef mbiencoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000922 {"encode", (PyCFunction)mbiencoder_encode,
923 METH_VARARGS | METH_KEYWORDS, NULL},
924 {"reset", (PyCFunction)mbiencoder_reset,
925 METH_NOARGS, NULL},
926 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000927};
928
929static PyObject *
930mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
931{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000932 MultibyteIncrementalEncoderObject *self;
933 PyObject *codec = NULL;
934 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000935
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000936 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
937 incnewkwarglist, &errors))
938 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000939
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000940 self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
941 if (self == NULL)
942 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000943
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000944 codec = PyObject_GetAttrString((PyObject *)type, "codec");
945 if (codec == NULL)
946 goto errorexit;
947 if (!MultibyteCodec_Check(codec)) {
948 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
949 goto errorexit;
950 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000951
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000952 self->codec = ((MultibyteCodecObject *)codec)->codec;
953 self->pendingsize = 0;
954 self->errors = internal_error_callback(errors);
955 if (self->errors == NULL)
956 goto errorexit;
957 if (self->codec->encinit != NULL &&
958 self->codec->encinit(&self->state, self->codec->config) != 0)
959 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000960
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000961 Py_DECREF(codec);
962 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000963
964errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000965 Py_XDECREF(self);
966 Py_XDECREF(codec);
967 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000968}
969
970static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000971mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
972{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000973 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000974}
975
976static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000977mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000978 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000979{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000980 if (ERROR_ISCUSTOM(self->errors))
981 Py_VISIT(self->errors);
982 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000983}
984
985static void
986mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
987{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000988 PyObject_GC_UnTrack(self);
989 ERROR_DECREF(self->errors);
990 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000991}
992
993static PyTypeObject MultibyteIncrementalEncoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000994 PyVarObject_HEAD_INIT(NULL, 0)
995 "MultibyteIncrementalEncoder", /* tp_name */
996 sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
997 0, /* tp_itemsize */
998 /* methods */
999 (destructor)mbiencoder_dealloc, /* tp_dealloc */
1000 0, /* tp_print */
1001 0, /* tp_getattr */
1002 0, /* tp_setattr */
1003 0, /* tp_reserved */
1004 0, /* tp_repr */
1005 0, /* tp_as_number */
1006 0, /* tp_as_sequence */
1007 0, /* tp_as_mapping */
1008 0, /* tp_hash */
1009 0, /* tp_call */
1010 0, /* tp_str */
1011 PyObject_GenericGetAttr, /* tp_getattro */
1012 0, /* tp_setattro */
1013 0, /* tp_as_buffer */
1014 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1015 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1016 0, /* tp_doc */
1017 (traverseproc)mbiencoder_traverse, /* tp_traverse */
1018 0, /* tp_clear */
1019 0, /* tp_richcompare */
1020 0, /* tp_weaklistoffset */
1021 0, /* tp_iter */
1022 0, /* tp_iterext */
1023 mbiencoder_methods, /* tp_methods */
1024 0, /* tp_members */
1025 codecctx_getsets, /* tp_getset */
1026 0, /* tp_base */
1027 0, /* tp_dict */
1028 0, /* tp_descr_get */
1029 0, /* tp_descr_set */
1030 0, /* tp_dictoffset */
1031 mbiencoder_init, /* tp_init */
1032 0, /* tp_alloc */
1033 mbiencoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001034};
1035
1036
1037/**
1038 * MultibyteIncrementalDecoder object
1039 */
1040
1041static PyObject *
1042mbidecoder_decode(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001043 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001044{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001045 MultibyteDecodeBuffer buf;
1046 char *data, *wdata = NULL;
1047 Py_buffer pdata;
1048 Py_ssize_t wsize, finalsize = 0, size, origpending;
1049 int final = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001050
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001051 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|i:decode",
1052 incrementalkwarglist, &pdata, &final))
1053 return NULL;
1054 data = pdata.buf;
1055 size = pdata.len;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001056
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001057 buf.outobj = buf.excobj = NULL;
1058 origpending = self->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001059
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001060 if (self->pendingsize == 0) {
1061 wsize = size;
1062 wdata = data;
1063 }
1064 else {
1065 if (size > PY_SSIZE_T_MAX - self->pendingsize) {
1066 PyErr_NoMemory();
1067 goto errorexit;
1068 }
1069 wsize = size + self->pendingsize;
1070 wdata = PyMem_Malloc(wsize);
1071 if (wdata == NULL)
1072 goto errorexit;
1073 memcpy(wdata, self->pending, self->pendingsize);
1074 memcpy(wdata + self->pendingsize, data, size);
1075 self->pendingsize = 0;
1076 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001077
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001078 if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
1079 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001080
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001081 if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
1082 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001083
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001084 if (final && buf.inbuf < buf.inbuf_end) {
1085 if (multibytecodec_decerror(self->codec, &self->state,
1086 &buf, self->errors, MBERR_TOOFEW)) {
1087 /* recover the original pending buffer */
1088 memcpy(self->pending, wdata, origpending);
1089 self->pendingsize = origpending;
1090 goto errorexit;
1091 }
1092 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001093
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001094 if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
1095 if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
1096 goto errorexit;
1097 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001098
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001099 finalsize = (Py_ssize_t)(buf.outbuf - PyUnicode_AS_UNICODE(buf.outobj));
1100 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1101 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1102 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001103
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001104 PyBuffer_Release(&pdata);
1105 if (wdata != data)
1106 PyMem_Del(wdata);
1107 Py_XDECREF(buf.excobj);
1108 return buf.outobj;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001109
1110errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001111 PyBuffer_Release(&pdata);
1112 if (wdata != NULL && wdata != data)
1113 PyMem_Del(wdata);
1114 Py_XDECREF(buf.excobj);
1115 Py_XDECREF(buf.outobj);
1116 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001117}
1118
1119static PyObject *
1120mbidecoder_reset(MultibyteIncrementalDecoderObject *self)
1121{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001122 if (self->codec->decreset != NULL &&
1123 self->codec->decreset(&self->state, self->codec->config) != 0)
1124 return NULL;
1125 self->pendingsize = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001126
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001127 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001128}
1129
1130static struct PyMethodDef mbidecoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001131 {"decode", (PyCFunction)mbidecoder_decode,
1132 METH_VARARGS | METH_KEYWORDS, NULL},
1133 {"reset", (PyCFunction)mbidecoder_reset,
1134 METH_NOARGS, NULL},
1135 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001136};
1137
1138static PyObject *
1139mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1140{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001141 MultibyteIncrementalDecoderObject *self;
1142 PyObject *codec = NULL;
1143 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001144
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001145 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
1146 incnewkwarglist, &errors))
1147 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001148
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001149 self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
1150 if (self == NULL)
1151 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001152
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001153 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1154 if (codec == NULL)
1155 goto errorexit;
1156 if (!MultibyteCodec_Check(codec)) {
1157 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1158 goto errorexit;
1159 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001160
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001161 self->codec = ((MultibyteCodecObject *)codec)->codec;
1162 self->pendingsize = 0;
1163 self->errors = internal_error_callback(errors);
1164 if (self->errors == NULL)
1165 goto errorexit;
1166 if (self->codec->decinit != NULL &&
1167 self->codec->decinit(&self->state, self->codec->config) != 0)
1168 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001169
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001170 Py_DECREF(codec);
1171 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001172
1173errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001174 Py_XDECREF(self);
1175 Py_XDECREF(codec);
1176 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001177}
1178
1179static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001180mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
1181{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001182 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001183}
1184
1185static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001186mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001187 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001188{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001189 if (ERROR_ISCUSTOM(self->errors))
1190 Py_VISIT(self->errors);
1191 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001192}
1193
1194static void
1195mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
1196{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001197 PyObject_GC_UnTrack(self);
1198 ERROR_DECREF(self->errors);
1199 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001200}
1201
1202static PyTypeObject MultibyteIncrementalDecoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001203 PyVarObject_HEAD_INIT(NULL, 0)
1204 "MultibyteIncrementalDecoder", /* tp_name */
1205 sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */
1206 0, /* tp_itemsize */
1207 /* methods */
1208 (destructor)mbidecoder_dealloc, /* tp_dealloc */
1209 0, /* tp_print */
1210 0, /* tp_getattr */
1211 0, /* tp_setattr */
1212 0, /* tp_reserved */
1213 0, /* tp_repr */
1214 0, /* tp_as_number */
1215 0, /* tp_as_sequence */
1216 0, /* tp_as_mapping */
1217 0, /* tp_hash */
1218 0, /* tp_call */
1219 0, /* tp_str */
1220 PyObject_GenericGetAttr, /* tp_getattro */
1221 0, /* tp_setattro */
1222 0, /* tp_as_buffer */
1223 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1224 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1225 0, /* tp_doc */
1226 (traverseproc)mbidecoder_traverse, /* tp_traverse */
1227 0, /* tp_clear */
1228 0, /* tp_richcompare */
1229 0, /* tp_weaklistoffset */
1230 0, /* tp_iter */
1231 0, /* tp_iterext */
1232 mbidecoder_methods, /* tp_methods */
1233 0, /* tp_members */
1234 codecctx_getsets, /* tp_getset */
1235 0, /* tp_base */
1236 0, /* tp_dict */
1237 0, /* tp_descr_get */
1238 0, /* tp_descr_set */
1239 0, /* tp_dictoffset */
1240 mbidecoder_init, /* tp_init */
1241 0, /* tp_alloc */
1242 mbidecoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001243};
1244
1245
1246/**
1247 * MultibyteStreamReader object
1248 */
1249
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001250static PyObject *
1251mbstreamreader_iread(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001252 const char *method, Py_ssize_t sizehint)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001253{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001254 MultibyteDecodeBuffer buf;
1255 PyObject *cres;
1256 Py_ssize_t rsize, finalsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001257
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001258 if (sizehint == 0)
1259 return PyUnicode_FromUnicode(NULL, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001260
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001261 buf.outobj = buf.excobj = NULL;
1262 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001263
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001264 for (;;) {
1265 int endoffile;
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001266
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001267 if (sizehint < 0)
1268 cres = PyObject_CallMethod(self->stream,
1269 (char *)method, NULL);
1270 else
1271 cres = PyObject_CallMethod(self->stream,
1272 (char *)method, "i", sizehint);
1273 if (cres == NULL)
1274 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001275
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001276 if (!PyBytes_Check(cres)) {
1277 PyErr_Format(PyExc_TypeError,
1278 "stream function returned a "
1279 "non-bytes object (%.100s)",
1280 cres->ob_type->tp_name);
1281 goto errorexit;
1282 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001283
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001284 endoffile = (PyBytes_GET_SIZE(cres) == 0);
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001285
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001286 if (self->pendingsize > 0) {
1287 PyObject *ctr;
1288 char *ctrdata;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001289
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001290 if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
1291 PyErr_NoMemory();
1292 goto errorexit;
1293 }
1294 rsize = PyBytes_GET_SIZE(cres) + self->pendingsize;
1295 ctr = PyBytes_FromStringAndSize(NULL, rsize);
1296 if (ctr == NULL)
1297 goto errorexit;
1298 ctrdata = PyBytes_AS_STRING(ctr);
1299 memcpy(ctrdata, self->pending, self->pendingsize);
1300 memcpy(ctrdata + self->pendingsize,
1301 PyBytes_AS_STRING(cres),
1302 PyBytes_GET_SIZE(cres));
1303 Py_DECREF(cres);
1304 cres = ctr;
1305 self->pendingsize = 0;
1306 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001307
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001308 rsize = PyBytes_GET_SIZE(cres);
1309 if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres),
1310 rsize) != 0)
1311 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001312
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001313 if (rsize > 0 && decoder_feed_buffer(
1314 (MultibyteStatefulDecoderContext *)self, &buf))
1315 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001316
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001317 if (endoffile || sizehint < 0) {
1318 if (buf.inbuf < buf.inbuf_end &&
1319 multibytecodec_decerror(self->codec, &self->state,
1320 &buf, self->errors, MBERR_TOOFEW))
1321 goto errorexit;
1322 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001323
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001324 if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
1325 if (decoder_append_pending(STATEFUL_DCTX(self),
1326 &buf) != 0)
1327 goto errorexit;
1328 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001329
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001330 finalsize = (Py_ssize_t)(buf.outbuf -
1331 PyUnicode_AS_UNICODE(buf.outobj));
1332 Py_DECREF(cres);
1333 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001334
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001335 if (sizehint < 0 || finalsize != 0 || rsize == 0)
1336 break;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001337
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001338 sizehint = 1; /* read 1 more byte and retry */
1339 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001340
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001341 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1342 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1343 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001344
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001345 Py_XDECREF(cres);
1346 Py_XDECREF(buf.excobj);
1347 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001348
1349errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001350 Py_XDECREF(cres);
1351 Py_XDECREF(buf.excobj);
1352 Py_XDECREF(buf.outobj);
1353 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001354}
1355
1356static PyObject *
1357mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args)
1358{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001359 PyObject *sizeobj = NULL;
1360 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001361
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001362 if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj))
1363 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001364
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001365 if (sizeobj == Py_None || sizeobj == NULL)
1366 size = -1;
1367 else if (PyLong_Check(sizeobj))
1368 size = PyLong_AsSsize_t(sizeobj);
1369 else {
1370 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1371 return NULL;
1372 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001373
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001374 if (size == -1 && PyErr_Occurred())
1375 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001376
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001377 return mbstreamreader_iread(self, "read", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001378}
1379
1380static PyObject *
1381mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args)
1382{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001383 PyObject *sizeobj = NULL;
1384 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001385
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001386 if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj))
1387 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001388
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001389 if (sizeobj == Py_None || sizeobj == NULL)
1390 size = -1;
1391 else if (PyLong_Check(sizeobj))
1392 size = PyLong_AsSsize_t(sizeobj);
1393 else {
1394 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1395 return NULL;
1396 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001397
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001398 if (size == -1 && PyErr_Occurred())
1399 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001400
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001401 return mbstreamreader_iread(self, "readline", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001402}
1403
1404static PyObject *
1405mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args)
1406{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001407 PyObject *sizehintobj = NULL, *r, *sr;
1408 Py_ssize_t sizehint;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001409
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001410 if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj))
1411 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001412
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001413 if (sizehintobj == Py_None || sizehintobj == NULL)
1414 sizehint = -1;
1415 else if (PyLong_Check(sizehintobj))
1416 sizehint = PyLong_AsSsize_t(sizehintobj);
1417 else {
1418 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1419 return NULL;
1420 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001421
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001422 if (sizehint == -1 && PyErr_Occurred())
1423 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001424
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001425 r = mbstreamreader_iread(self, "read", sizehint);
1426 if (r == NULL)
1427 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001428
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001429 sr = PyUnicode_Splitlines(r, 1);
1430 Py_DECREF(r);
1431 return sr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001432}
1433
1434static PyObject *
1435mbstreamreader_reset(MultibyteStreamReaderObject *self)
1436{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001437 if (self->codec->decreset != NULL &&
1438 self->codec->decreset(&self->state, self->codec->config) != 0)
1439 return NULL;
1440 self->pendingsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001441
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001442 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001443}
1444
1445static struct PyMethodDef mbstreamreader_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001446 {"read", (PyCFunction)mbstreamreader_read,
1447 METH_VARARGS, NULL},
1448 {"readline", (PyCFunction)mbstreamreader_readline,
1449 METH_VARARGS, NULL},
1450 {"readlines", (PyCFunction)mbstreamreader_readlines,
1451 METH_VARARGS, NULL},
1452 {"reset", (PyCFunction)mbstreamreader_reset,
1453 METH_NOARGS, NULL},
1454 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001455};
1456
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001457static PyMemberDef mbstreamreader_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001458 {"stream", T_OBJECT,
1459 offsetof(MultibyteStreamReaderObject, stream),
1460 READONLY, NULL},
1461 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001462};
1463
1464static PyObject *
1465mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1466{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001467 MultibyteStreamReaderObject *self;
1468 PyObject *stream, *codec = NULL;
1469 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001470
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001471 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
1472 streamkwarglist, &stream, &errors))
1473 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001474
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001475 self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
1476 if (self == NULL)
1477 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001478
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001479 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1480 if (codec == NULL)
1481 goto errorexit;
1482 if (!MultibyteCodec_Check(codec)) {
1483 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1484 goto errorexit;
1485 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001486
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001487 self->codec = ((MultibyteCodecObject *)codec)->codec;
1488 self->stream = stream;
1489 Py_INCREF(stream);
1490 self->pendingsize = 0;
1491 self->errors = internal_error_callback(errors);
1492 if (self->errors == NULL)
1493 goto errorexit;
1494 if (self->codec->decinit != NULL &&
1495 self->codec->decinit(&self->state, self->codec->config) != 0)
1496 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001497
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001498 Py_DECREF(codec);
1499 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001500
1501errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001502 Py_XDECREF(self);
1503 Py_XDECREF(codec);
1504 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001505}
1506
1507static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001508mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
1509{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001510 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001511}
1512
1513static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001514mbstreamreader_traverse(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001515 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001516{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001517 if (ERROR_ISCUSTOM(self->errors))
1518 Py_VISIT(self->errors);
1519 Py_VISIT(self->stream);
1520 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001521}
1522
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001523static void
1524mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
1525{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001526 PyObject_GC_UnTrack(self);
1527 ERROR_DECREF(self->errors);
1528 Py_XDECREF(self->stream);
1529 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001530}
1531
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001532static PyTypeObject MultibyteStreamReader_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001533 PyVarObject_HEAD_INIT(NULL, 0)
1534 "MultibyteStreamReader", /* tp_name */
1535 sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
1536 0, /* tp_itemsize */
1537 /* methods */
1538 (destructor)mbstreamreader_dealloc, /* tp_dealloc */
1539 0, /* tp_print */
1540 0, /* tp_getattr */
1541 0, /* tp_setattr */
1542 0, /* tp_reserved */
1543 0, /* tp_repr */
1544 0, /* tp_as_number */
1545 0, /* tp_as_sequence */
1546 0, /* tp_as_mapping */
1547 0, /* tp_hash */
1548 0, /* tp_call */
1549 0, /* tp_str */
1550 PyObject_GenericGetAttr, /* tp_getattro */
1551 0, /* tp_setattro */
1552 0, /* tp_as_buffer */
1553 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1554 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1555 0, /* tp_doc */
1556 (traverseproc)mbstreamreader_traverse, /* tp_traverse */
1557 0, /* tp_clear */
1558 0, /* tp_richcompare */
1559 0, /* tp_weaklistoffset */
1560 0, /* tp_iter */
1561 0, /* tp_iterext */
1562 mbstreamreader_methods, /* tp_methods */
1563 mbstreamreader_members, /* tp_members */
1564 codecctx_getsets, /* tp_getset */
1565 0, /* tp_base */
1566 0, /* tp_dict */
1567 0, /* tp_descr_get */
1568 0, /* tp_descr_set */
1569 0, /* tp_dictoffset */
1570 mbstreamreader_init, /* tp_init */
1571 0, /* tp_alloc */
1572 mbstreamreader_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001573};
1574
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001575
1576/**
1577 * MultibyteStreamWriter object
1578 */
1579
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001580static int
1581mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001582 PyObject *unistr)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001583{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001584 PyObject *str, *wr;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001585 _Py_IDENTIFIER(write);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001586
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001587 str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
1588 if (str == NULL)
1589 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001590
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001591 wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001592 Py_DECREF(str);
1593 if (wr == NULL)
1594 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001595
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001596 Py_DECREF(wr);
1597 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001598}
1599
1600static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001601mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001602{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001603 if (mbstreamwriter_iwrite(self, strobj))
1604 return NULL;
1605 else
1606 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001607}
1608
1609static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001610mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001611{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001612 PyObject *strobj;
1613 int i, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001614
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001615 if (!PySequence_Check(lines)) {
1616 PyErr_SetString(PyExc_TypeError,
1617 "arg must be a sequence object");
1618 return NULL;
1619 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001620
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001621 for (i = 0; i < PySequence_Length(lines); i++) {
1622 /* length can be changed even within this loop */
1623 strobj = PySequence_GetItem(lines, i);
1624 if (strobj == NULL)
1625 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001626
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001627 r = mbstreamwriter_iwrite(self, strobj);
1628 Py_DECREF(strobj);
1629 if (r == -1)
1630 return NULL;
1631 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001632
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001633 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001634}
1635
1636static PyObject *
1637mbstreamwriter_reset(MultibyteStreamWriterObject *self)
1638{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001639 const Py_UNICODE *pending;
1640 PyObject *pwrt;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001641
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001642 pending = self->pending;
1643 pwrt = multibytecodec_encode(self->codec, &self->state,
1644 &pending, self->pendingsize, self->errors,
1645 MBENC_FLUSH | MBENC_RESET);
1646 /* some pending buffer can be truncated when UnicodeEncodeError is
1647 * raised on 'strict' mode. but, 'reset' method is designed to
1648 * reset the pending buffer or states so failed string sequence
1649 * ought to be missed */
1650 self->pendingsize = 0;
1651 if (pwrt == NULL)
1652 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001653
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001654 assert(PyBytes_Check(pwrt));
1655 if (PyBytes_Size(pwrt) > 0) {
1656 PyObject *wr;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001657 _Py_IDENTIFIER(write);
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001658
1659 wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001660 if (wr == NULL) {
1661 Py_DECREF(pwrt);
1662 return NULL;
1663 }
1664 }
1665 Py_DECREF(pwrt);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001666
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001667 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001668}
1669
1670static PyObject *
1671mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1672{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001673 MultibyteStreamWriterObject *self;
1674 PyObject *stream, *codec = NULL;
1675 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001676
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001677 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
1678 streamkwarglist, &stream, &errors))
1679 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001680
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001681 self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
1682 if (self == NULL)
1683 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001684
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001685 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1686 if (codec == NULL)
1687 goto errorexit;
1688 if (!MultibyteCodec_Check(codec)) {
1689 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1690 goto errorexit;
1691 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001692
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001693 self->codec = ((MultibyteCodecObject *)codec)->codec;
1694 self->stream = stream;
1695 Py_INCREF(stream);
1696 self->pendingsize = 0;
1697 self->errors = internal_error_callback(errors);
1698 if (self->errors == NULL)
1699 goto errorexit;
1700 if (self->codec->encinit != NULL &&
1701 self->codec->encinit(&self->state, self->codec->config) != 0)
1702 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001703
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001704 Py_DECREF(codec);
1705 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001706
1707errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001708 Py_XDECREF(self);
1709 Py_XDECREF(codec);
1710 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001711}
1712
1713static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001714mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
1715{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001716 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001717}
1718
1719static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001720mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001721 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001722{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001723 if (ERROR_ISCUSTOM(self->errors))
1724 Py_VISIT(self->errors);
1725 Py_VISIT(self->stream);
1726 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001727}
1728
1729static void
1730mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
1731{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001732 PyObject_GC_UnTrack(self);
1733 ERROR_DECREF(self->errors);
1734 Py_XDECREF(self->stream);
1735 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001736}
1737
1738static struct PyMethodDef mbstreamwriter_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001739 {"write", (PyCFunction)mbstreamwriter_write,
1740 METH_O, NULL},
1741 {"writelines", (PyCFunction)mbstreamwriter_writelines,
1742 METH_O, NULL},
1743 {"reset", (PyCFunction)mbstreamwriter_reset,
1744 METH_NOARGS, NULL},
1745 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001746};
1747
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001748static PyMemberDef mbstreamwriter_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001749 {"stream", T_OBJECT,
1750 offsetof(MultibyteStreamWriterObject, stream),
1751 READONLY, NULL},
1752 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001753};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001754
1755static PyTypeObject MultibyteStreamWriter_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001756 PyVarObject_HEAD_INIT(NULL, 0)
1757 "MultibyteStreamWriter", /* tp_name */
1758 sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
1759 0, /* tp_itemsize */
1760 /* methods */
1761 (destructor)mbstreamwriter_dealloc, /* tp_dealloc */
1762 0, /* tp_print */
1763 0, /* tp_getattr */
1764 0, /* tp_setattr */
1765 0, /* tp_reserved */
1766 0, /* tp_repr */
1767 0, /* tp_as_number */
1768 0, /* tp_as_sequence */
1769 0, /* tp_as_mapping */
1770 0, /* tp_hash */
1771 0, /* tp_call */
1772 0, /* tp_str */
1773 PyObject_GenericGetAttr, /* tp_getattro */
1774 0, /* tp_setattro */
1775 0, /* tp_as_buffer */
1776 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1777 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1778 0, /* tp_doc */
1779 (traverseproc)mbstreamwriter_traverse, /* tp_traverse */
1780 0, /* tp_clear */
1781 0, /* tp_richcompare */
1782 0, /* tp_weaklistoffset */
1783 0, /* tp_iter */
1784 0, /* tp_iterext */
1785 mbstreamwriter_methods, /* tp_methods */
1786 mbstreamwriter_members, /* tp_members */
1787 codecctx_getsets, /* tp_getset */
1788 0, /* tp_base */
1789 0, /* tp_dict */
1790 0, /* tp_descr_get */
1791 0, /* tp_descr_set */
1792 0, /* tp_dictoffset */
1793 mbstreamwriter_init, /* tp_init */
1794 0, /* tp_alloc */
1795 mbstreamwriter_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001796};
1797
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001798
1799/**
1800 * Exposed factory function
1801 */
1802
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001803static PyObject *
1804__create_codec(PyObject *ignore, PyObject *arg)
1805{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001806 MultibyteCodecObject *self;
1807 MultibyteCodec *codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001808
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001809 if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
1810 PyErr_SetString(PyExc_ValueError, "argument type invalid");
1811 return NULL;
1812 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001813
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001814 codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
1815 if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
1816 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001817
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001818 self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
1819 if (self == NULL)
1820 return NULL;
1821 self->codec = codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001822
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001823 return (PyObject *)self;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001824}
1825
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001826static struct PyMethodDef __methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001827 {"__create_codec", (PyCFunction)__create_codec, METH_O},
1828 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001829};
1830
Martin v. Löwis1a214512008-06-11 05:26:20 +00001831
1832static struct PyModuleDef _multibytecodecmodule = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001833 PyModuleDef_HEAD_INIT,
1834 "_multibytecodec",
1835 NULL,
1836 -1,
1837 __methods,
1838 NULL,
1839 NULL,
1840 NULL,
1841 NULL
Martin v. Löwis1a214512008-06-11 05:26:20 +00001842};
1843
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001844PyMODINIT_FUNC
Martin v. Löwis1a214512008-06-11 05:26:20 +00001845PyInit__multibytecodec(void)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001846{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001847 int i;
1848 PyObject *m;
1849 PyTypeObject *typelist[] = {
1850 &MultibyteIncrementalEncoder_Type,
1851 &MultibyteIncrementalDecoder_Type,
1852 &MultibyteStreamReader_Type,
1853 &MultibyteStreamWriter_Type,
1854 NULL
1855 };
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001856
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001857 if (PyType_Ready(&MultibyteCodec_Type) < 0)
1858 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001859
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001860 m = PyModule_Create(&_multibytecodecmodule);
1861 if (m == NULL)
1862 return NULL;
Neal Norwitz058bde12005-09-21 06:44:25 +00001863
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001864 for (i = 0; typelist[i] != NULL; i++) {
1865 if (PyType_Ready(typelist[i]) < 0)
1866 return NULL;
1867 Py_INCREF(typelist[i]);
1868 PyModule_AddObject(m, typelist[i]->tp_name,
1869 (PyObject *)typelist[i]);
1870 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001871
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001872 if (PyErr_Occurred()) {
1873 Py_FatalError("can't initialize the _multibytecodec module");
1874 Py_DECREF(m);
1875 m = NULL;
1876 }
1877 return m;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001878}