blob: 1b37845285f2ce248ae8901cf0d0701cac37ff7d [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
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000577 data = PyUnicode_AS_UNICODE(arg);
578 datalen = PyUnicode_GET_SIZE(arg);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000579
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000580 errorcb = internal_error_callback(errors);
581 if (errorcb == NULL) {
582 Py_XDECREF(ucvt);
583 return NULL;
584 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000585
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000586 if (self->codec->encinit != NULL &&
587 self->codec->encinit(&state, self->codec->config) != 0)
588 goto errorexit;
589 r = multibytecodec_encode(self->codec, &state,
590 (const Py_UNICODE **)&data, datalen, errorcb,
591 MBENC_FLUSH | MBENC_RESET);
592 if (r == NULL)
593 goto errorexit;
Hye-Shik Chang2bb146f2004-07-18 03:06:29 +0000594
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000595 ERROR_DECREF(errorcb);
596 Py_XDECREF(ucvt);
597 return make_tuple(r, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000598
599errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000600 ERROR_DECREF(errorcb);
601 Py_XDECREF(ucvt);
602 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000603}
604
605static PyObject *
606MultibyteCodec_Decode(MultibyteCodecObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000607 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000608{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000609 MultibyteCodec_State state;
610 MultibyteDecodeBuffer buf;
611 PyObject *errorcb;
612 Py_buffer pdata;
613 const char *data, *errors = NULL;
614 Py_ssize_t datalen, finalsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000615
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000616 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|z:decode",
617 codeckwarglist, &pdata, &errors))
618 return NULL;
619 data = pdata.buf;
620 datalen = pdata.len;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000621
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000622 errorcb = internal_error_callback(errors);
623 if (errorcb == NULL) {
624 PyBuffer_Release(&pdata);
625 return NULL;
626 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000627
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000628 if (datalen == 0) {
629 PyBuffer_Release(&pdata);
630 ERROR_DECREF(errorcb);
631 return make_tuple(PyUnicode_FromUnicode(NULL, 0), 0);
632 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000633
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000634 buf.excobj = NULL;
635 buf.inbuf = buf.inbuf_top = (unsigned char *)data;
636 buf.inbuf_end = buf.inbuf_top + datalen;
637 buf.outobj = PyUnicode_FromUnicode(NULL, datalen);
638 if (buf.outobj == NULL)
639 goto errorexit;
640 buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj);
641 buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000642
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000643 if (self->codec->decinit != NULL &&
644 self->codec->decinit(&state, self->codec->config) != 0)
645 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000646
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000647 while (buf.inbuf < buf.inbuf_end) {
648 Py_ssize_t inleft, outleft, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000649
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000650 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
651 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000652
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000653 r = self->codec->decode(&state, self->codec->config,
654 &buf.inbuf, inleft, &buf.outbuf, outleft);
655 if (r == 0)
656 break;
657 else if (multibytecodec_decerror(self->codec, &state,
658 &buf, errorcb, r))
659 goto errorexit;
660 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000661
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000662 finalsize = (Py_ssize_t)(buf.outbuf -
663 PyUnicode_AS_UNICODE(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000664
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000665 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
666 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
667 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000668
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000669 PyBuffer_Release(&pdata);
670 Py_XDECREF(buf.excobj);
671 ERROR_DECREF(errorcb);
672 return make_tuple(buf.outobj, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000673
674errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000675 PyBuffer_Release(&pdata);
676 ERROR_DECREF(errorcb);
677 Py_XDECREF(buf.excobj);
678 Py_XDECREF(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000679
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000680 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000681}
682
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000683static struct PyMethodDef multibytecodec_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000684 {"encode", (PyCFunction)MultibyteCodec_Encode,
685 METH_VARARGS | METH_KEYWORDS,
686 MultibyteCodec_Encode__doc__},
687 {"decode", (PyCFunction)MultibyteCodec_Decode,
688 METH_VARARGS | METH_KEYWORDS,
689 MultibyteCodec_Decode__doc__},
690 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000691};
692
693static void
694multibytecodec_dealloc(MultibyteCodecObject *self)
695{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000696 PyObject_Del(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000697}
698
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000699static PyTypeObject MultibyteCodec_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000700 PyVarObject_HEAD_INIT(NULL, 0)
701 "MultibyteCodec", /* tp_name */
702 sizeof(MultibyteCodecObject), /* tp_basicsize */
703 0, /* tp_itemsize */
704 /* methods */
705 (destructor)multibytecodec_dealloc, /* tp_dealloc */
706 0, /* tp_print */
707 0, /* tp_getattr */
708 0, /* tp_setattr */
709 0, /* tp_reserved */
710 0, /* tp_repr */
711 0, /* tp_as_number */
712 0, /* tp_as_sequence */
713 0, /* tp_as_mapping */
714 0, /* tp_hash */
715 0, /* tp_call */
716 0, /* tp_str */
717 PyObject_GenericGetAttr, /* tp_getattro */
718 0, /* tp_setattro */
719 0, /* tp_as_buffer */
720 Py_TPFLAGS_DEFAULT, /* tp_flags */
721 0, /* tp_doc */
722 0, /* tp_traverse */
723 0, /* tp_clear */
724 0, /* tp_richcompare */
725 0, /* tp_weaklistoffset */
726 0, /* tp_iter */
727 0, /* tp_iterext */
728 multibytecodec_methods, /* tp_methods */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000729};
730
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000731
732/**
733 * Utility functions for stateful codec mechanism
734 */
735
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000736#define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o))
737#define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o))
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000738
739static PyObject *
740encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000741 PyObject *unistr, int final)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000742{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000743 PyObject *ucvt, *r = NULL;
744 Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL;
745 Py_ssize_t datalen, origpending;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000746
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000747 if (PyUnicode_Check(unistr))
748 ucvt = NULL;
749 else {
750 unistr = ucvt = PyObject_Str(unistr);
751 if (unistr == NULL)
752 return NULL;
753 else if (!PyUnicode_Check(unistr)) {
754 PyErr_SetString(PyExc_TypeError,
755 "couldn't convert the object to unicode.");
756 Py_DECREF(ucvt);
757 return NULL;
758 }
759 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000760
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000761 datalen = PyUnicode_GET_SIZE(unistr);
762 origpending = ctx->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000763
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000764 if (origpending > 0) {
765 if (datalen > PY_SSIZE_T_MAX - ctx->pendingsize) {
766 PyErr_NoMemory();
767 /* inbuf_tmp == NULL */
768 goto errorexit;
769 }
770 inbuf_tmp = PyMem_New(Py_UNICODE, datalen + ctx->pendingsize);
771 if (inbuf_tmp == NULL)
772 goto errorexit;
773 memcpy(inbuf_tmp, ctx->pending,
774 Py_UNICODE_SIZE * ctx->pendingsize);
775 memcpy(inbuf_tmp + ctx->pendingsize,
776 PyUnicode_AS_UNICODE(unistr),
777 Py_UNICODE_SIZE * datalen);
778 datalen += ctx->pendingsize;
779 ctx->pendingsize = 0;
780 inbuf = inbuf_tmp;
781 }
782 else
783 inbuf = (Py_UNICODE *)PyUnicode_AS_UNICODE(unistr);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000784
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000785 inbuf_end = inbuf + datalen;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000786
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000787 r = multibytecodec_encode(ctx->codec, &ctx->state,
Victor Stinner6bcbef72011-05-24 22:17:55 +0200788 (const Py_UNICODE **)&inbuf, datalen,
789 ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000790 if (r == NULL) {
791 /* recover the original pending buffer */
792 if (origpending > 0)
793 memcpy(ctx->pending, inbuf_tmp,
794 Py_UNICODE_SIZE * origpending);
795 ctx->pendingsize = origpending;
796 goto errorexit;
797 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000798
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000799 if (inbuf < inbuf_end) {
800 ctx->pendingsize = (Py_ssize_t)(inbuf_end - inbuf);
801 if (ctx->pendingsize > MAXENCPENDING) {
802 /* normal codecs can't reach here */
803 ctx->pendingsize = 0;
804 PyErr_SetString(PyExc_UnicodeError,
805 "pending buffer overflow");
806 goto errorexit;
807 }
808 memcpy(ctx->pending, inbuf,
809 ctx->pendingsize * Py_UNICODE_SIZE);
810 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000811
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000812 if (inbuf_tmp != NULL)
813 PyMem_Del(inbuf_tmp);
814 Py_XDECREF(ucvt);
815 return r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000816
817errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000818 if (inbuf_tmp != NULL)
819 PyMem_Del(inbuf_tmp);
820 Py_XDECREF(r);
821 Py_XDECREF(ucvt);
822 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000823}
824
825static int
826decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000827 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000828{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000829 Py_ssize_t npendings;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000830
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000831 npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
832 if (npendings + ctx->pendingsize > MAXDECPENDING ||
833 npendings > PY_SSIZE_T_MAX - ctx->pendingsize) {
834 PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow");
835 return -1;
836 }
837 memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings);
838 ctx->pendingsize += npendings;
839 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000840}
841
842static int
843decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000844 Py_ssize_t size)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000845{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000846 buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
847 buf->inbuf_end = buf->inbuf_top + size;
848 if (buf->outobj == NULL) { /* only if outobj is not allocated yet */
849 buf->outobj = PyUnicode_FromUnicode(NULL, size);
850 if (buf->outobj == NULL)
851 return -1;
852 buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj);
853 buf->outbuf_end = buf->outbuf +
854 PyUnicode_GET_SIZE(buf->outobj);
855 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000856
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000857 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000858}
859
860static int
861decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000862 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000863{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000864 while (buf->inbuf < buf->inbuf_end) {
865 Py_ssize_t inleft, outleft;
866 Py_ssize_t r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000867
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000868 inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
869 outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000870
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000871 r = ctx->codec->decode(&ctx->state, ctx->codec->config,
872 &buf->inbuf, inleft, &buf->outbuf, outleft);
873 if (r == 0 || r == MBERR_TOOFEW)
874 break;
875 else if (multibytecodec_decerror(ctx->codec, &ctx->state,
876 buf, ctx->errors, r))
877 return -1;
878 }
879 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000880}
881
882
883/**
884 * MultibyteIncrementalEncoder object
885 */
886
887static PyObject *
888mbiencoder_encode(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000889 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000890{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000891 PyObject *data;
892 int final = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000893
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000894 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode",
895 incrementalkwarglist, &data, &final))
896 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000897
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000898 return encoder_encode_stateful(STATEFUL_ECTX(self), data, final);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000899}
900
901static PyObject *
902mbiencoder_reset(MultibyteIncrementalEncoderObject *self)
903{
Victor Stinnere15dce32011-05-30 22:56:00 +0200904 /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */
905 unsigned char buffer[4], *outbuf;
906 Py_ssize_t r;
907 if (self->codec->encreset != NULL) {
908 outbuf = buffer;
909 r = self->codec->encreset(&self->state, self->codec->config,
910 &outbuf, sizeof(buffer));
911 if (r != 0)
912 return NULL;
913 }
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000914 self->pendingsize = 0;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000915 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000916}
917
918static struct PyMethodDef mbiencoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000919 {"encode", (PyCFunction)mbiencoder_encode,
920 METH_VARARGS | METH_KEYWORDS, NULL},
921 {"reset", (PyCFunction)mbiencoder_reset,
922 METH_NOARGS, NULL},
923 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000924};
925
926static PyObject *
927mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
928{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000929 MultibyteIncrementalEncoderObject *self;
930 PyObject *codec = NULL;
931 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000932
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000933 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
934 incnewkwarglist, &errors))
935 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000936
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000937 self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
938 if (self == NULL)
939 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000940
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000941 codec = PyObject_GetAttrString((PyObject *)type, "codec");
942 if (codec == NULL)
943 goto errorexit;
944 if (!MultibyteCodec_Check(codec)) {
945 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
946 goto errorexit;
947 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000948
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000949 self->codec = ((MultibyteCodecObject *)codec)->codec;
950 self->pendingsize = 0;
951 self->errors = internal_error_callback(errors);
952 if (self->errors == NULL)
953 goto errorexit;
954 if (self->codec->encinit != NULL &&
955 self->codec->encinit(&self->state, self->codec->config) != 0)
956 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000957
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000958 Py_DECREF(codec);
959 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000960
961errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000962 Py_XDECREF(self);
963 Py_XDECREF(codec);
964 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000965}
966
967static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000968mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
969{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000970 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000971}
972
973static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000974mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000975 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000976{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000977 if (ERROR_ISCUSTOM(self->errors))
978 Py_VISIT(self->errors);
979 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000980}
981
982static void
983mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
984{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000985 PyObject_GC_UnTrack(self);
986 ERROR_DECREF(self->errors);
987 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000988}
989
990static PyTypeObject MultibyteIncrementalEncoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000991 PyVarObject_HEAD_INIT(NULL, 0)
992 "MultibyteIncrementalEncoder", /* tp_name */
993 sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
994 0, /* tp_itemsize */
995 /* methods */
996 (destructor)mbiencoder_dealloc, /* tp_dealloc */
997 0, /* tp_print */
998 0, /* tp_getattr */
999 0, /* tp_setattr */
1000 0, /* tp_reserved */
1001 0, /* tp_repr */
1002 0, /* tp_as_number */
1003 0, /* tp_as_sequence */
1004 0, /* tp_as_mapping */
1005 0, /* tp_hash */
1006 0, /* tp_call */
1007 0, /* tp_str */
1008 PyObject_GenericGetAttr, /* tp_getattro */
1009 0, /* tp_setattro */
1010 0, /* tp_as_buffer */
1011 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1012 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1013 0, /* tp_doc */
1014 (traverseproc)mbiencoder_traverse, /* tp_traverse */
1015 0, /* tp_clear */
1016 0, /* tp_richcompare */
1017 0, /* tp_weaklistoffset */
1018 0, /* tp_iter */
1019 0, /* tp_iterext */
1020 mbiencoder_methods, /* tp_methods */
1021 0, /* tp_members */
1022 codecctx_getsets, /* tp_getset */
1023 0, /* tp_base */
1024 0, /* tp_dict */
1025 0, /* tp_descr_get */
1026 0, /* tp_descr_set */
1027 0, /* tp_dictoffset */
1028 mbiencoder_init, /* tp_init */
1029 0, /* tp_alloc */
1030 mbiencoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001031};
1032
1033
1034/**
1035 * MultibyteIncrementalDecoder object
1036 */
1037
1038static PyObject *
1039mbidecoder_decode(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001040 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001041{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001042 MultibyteDecodeBuffer buf;
1043 char *data, *wdata = NULL;
1044 Py_buffer pdata;
1045 Py_ssize_t wsize, finalsize = 0, size, origpending;
1046 int final = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001047
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001048 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|i:decode",
1049 incrementalkwarglist, &pdata, &final))
1050 return NULL;
1051 data = pdata.buf;
1052 size = pdata.len;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001053
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001054 buf.outobj = buf.excobj = NULL;
1055 origpending = self->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001056
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001057 if (self->pendingsize == 0) {
1058 wsize = size;
1059 wdata = data;
1060 }
1061 else {
1062 if (size > PY_SSIZE_T_MAX - self->pendingsize) {
1063 PyErr_NoMemory();
1064 goto errorexit;
1065 }
1066 wsize = size + self->pendingsize;
1067 wdata = PyMem_Malloc(wsize);
1068 if (wdata == NULL)
1069 goto errorexit;
1070 memcpy(wdata, self->pending, self->pendingsize);
1071 memcpy(wdata + self->pendingsize, data, size);
1072 self->pendingsize = 0;
1073 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001074
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001075 if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
1076 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001077
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001078 if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
1079 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001080
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001081 if (final && buf.inbuf < buf.inbuf_end) {
1082 if (multibytecodec_decerror(self->codec, &self->state,
1083 &buf, self->errors, MBERR_TOOFEW)) {
1084 /* recover the original pending buffer */
1085 memcpy(self->pending, wdata, origpending);
1086 self->pendingsize = origpending;
1087 goto errorexit;
1088 }
1089 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001090
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001091 if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
1092 if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
1093 goto errorexit;
1094 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001095
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001096 finalsize = (Py_ssize_t)(buf.outbuf - PyUnicode_AS_UNICODE(buf.outobj));
1097 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1098 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1099 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001100
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001101 PyBuffer_Release(&pdata);
1102 if (wdata != data)
1103 PyMem_Del(wdata);
1104 Py_XDECREF(buf.excobj);
1105 return buf.outobj;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001106
1107errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001108 PyBuffer_Release(&pdata);
1109 if (wdata != NULL && wdata != data)
1110 PyMem_Del(wdata);
1111 Py_XDECREF(buf.excobj);
1112 Py_XDECREF(buf.outobj);
1113 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001114}
1115
1116static PyObject *
1117mbidecoder_reset(MultibyteIncrementalDecoderObject *self)
1118{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001119 if (self->codec->decreset != NULL &&
1120 self->codec->decreset(&self->state, self->codec->config) != 0)
1121 return NULL;
1122 self->pendingsize = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001123
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001124 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001125}
1126
1127static struct PyMethodDef mbidecoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001128 {"decode", (PyCFunction)mbidecoder_decode,
1129 METH_VARARGS | METH_KEYWORDS, NULL},
1130 {"reset", (PyCFunction)mbidecoder_reset,
1131 METH_NOARGS, NULL},
1132 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001133};
1134
1135static PyObject *
1136mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1137{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001138 MultibyteIncrementalDecoderObject *self;
1139 PyObject *codec = NULL;
1140 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001141
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001142 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
1143 incnewkwarglist, &errors))
1144 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001145
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001146 self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
1147 if (self == NULL)
1148 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001149
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001150 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1151 if (codec == NULL)
1152 goto errorexit;
1153 if (!MultibyteCodec_Check(codec)) {
1154 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1155 goto errorexit;
1156 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001157
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001158 self->codec = ((MultibyteCodecObject *)codec)->codec;
1159 self->pendingsize = 0;
1160 self->errors = internal_error_callback(errors);
1161 if (self->errors == NULL)
1162 goto errorexit;
1163 if (self->codec->decinit != NULL &&
1164 self->codec->decinit(&self->state, self->codec->config) != 0)
1165 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001166
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001167 Py_DECREF(codec);
1168 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001169
1170errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001171 Py_XDECREF(self);
1172 Py_XDECREF(codec);
1173 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001174}
1175
1176static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001177mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
1178{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001179 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001180}
1181
1182static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001183mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001184 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001185{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001186 if (ERROR_ISCUSTOM(self->errors))
1187 Py_VISIT(self->errors);
1188 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001189}
1190
1191static void
1192mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
1193{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001194 PyObject_GC_UnTrack(self);
1195 ERROR_DECREF(self->errors);
1196 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001197}
1198
1199static PyTypeObject MultibyteIncrementalDecoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001200 PyVarObject_HEAD_INIT(NULL, 0)
1201 "MultibyteIncrementalDecoder", /* tp_name */
1202 sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */
1203 0, /* tp_itemsize */
1204 /* methods */
1205 (destructor)mbidecoder_dealloc, /* tp_dealloc */
1206 0, /* tp_print */
1207 0, /* tp_getattr */
1208 0, /* tp_setattr */
1209 0, /* tp_reserved */
1210 0, /* tp_repr */
1211 0, /* tp_as_number */
1212 0, /* tp_as_sequence */
1213 0, /* tp_as_mapping */
1214 0, /* tp_hash */
1215 0, /* tp_call */
1216 0, /* tp_str */
1217 PyObject_GenericGetAttr, /* tp_getattro */
1218 0, /* tp_setattro */
1219 0, /* tp_as_buffer */
1220 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1221 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1222 0, /* tp_doc */
1223 (traverseproc)mbidecoder_traverse, /* tp_traverse */
1224 0, /* tp_clear */
1225 0, /* tp_richcompare */
1226 0, /* tp_weaklistoffset */
1227 0, /* tp_iter */
1228 0, /* tp_iterext */
1229 mbidecoder_methods, /* tp_methods */
1230 0, /* tp_members */
1231 codecctx_getsets, /* tp_getset */
1232 0, /* tp_base */
1233 0, /* tp_dict */
1234 0, /* tp_descr_get */
1235 0, /* tp_descr_set */
1236 0, /* tp_dictoffset */
1237 mbidecoder_init, /* tp_init */
1238 0, /* tp_alloc */
1239 mbidecoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001240};
1241
1242
1243/**
1244 * MultibyteStreamReader object
1245 */
1246
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001247static PyObject *
1248mbstreamreader_iread(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001249 const char *method, Py_ssize_t sizehint)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001250{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001251 MultibyteDecodeBuffer buf;
1252 PyObject *cres;
1253 Py_ssize_t rsize, finalsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001254
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001255 if (sizehint == 0)
1256 return PyUnicode_FromUnicode(NULL, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001257
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001258 buf.outobj = buf.excobj = NULL;
1259 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001260
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001261 for (;;) {
1262 int endoffile;
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001263
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001264 if (sizehint < 0)
1265 cres = PyObject_CallMethod(self->stream,
1266 (char *)method, NULL);
1267 else
1268 cres = PyObject_CallMethod(self->stream,
1269 (char *)method, "i", sizehint);
1270 if (cres == NULL)
1271 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001272
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001273 if (!PyBytes_Check(cres)) {
1274 PyErr_Format(PyExc_TypeError,
1275 "stream function returned a "
1276 "non-bytes object (%.100s)",
1277 cres->ob_type->tp_name);
1278 goto errorexit;
1279 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001280
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001281 endoffile = (PyBytes_GET_SIZE(cres) == 0);
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001282
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001283 if (self->pendingsize > 0) {
1284 PyObject *ctr;
1285 char *ctrdata;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001286
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001287 if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
1288 PyErr_NoMemory();
1289 goto errorexit;
1290 }
1291 rsize = PyBytes_GET_SIZE(cres) + self->pendingsize;
1292 ctr = PyBytes_FromStringAndSize(NULL, rsize);
1293 if (ctr == NULL)
1294 goto errorexit;
1295 ctrdata = PyBytes_AS_STRING(ctr);
1296 memcpy(ctrdata, self->pending, self->pendingsize);
1297 memcpy(ctrdata + self->pendingsize,
1298 PyBytes_AS_STRING(cres),
1299 PyBytes_GET_SIZE(cres));
1300 Py_DECREF(cres);
1301 cres = ctr;
1302 self->pendingsize = 0;
1303 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001304
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001305 rsize = PyBytes_GET_SIZE(cres);
1306 if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres),
1307 rsize) != 0)
1308 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001309
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001310 if (rsize > 0 && decoder_feed_buffer(
1311 (MultibyteStatefulDecoderContext *)self, &buf))
1312 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001313
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001314 if (endoffile || sizehint < 0) {
1315 if (buf.inbuf < buf.inbuf_end &&
1316 multibytecodec_decerror(self->codec, &self->state,
1317 &buf, self->errors, MBERR_TOOFEW))
1318 goto errorexit;
1319 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001320
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001321 if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
1322 if (decoder_append_pending(STATEFUL_DCTX(self),
1323 &buf) != 0)
1324 goto errorexit;
1325 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001326
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001327 finalsize = (Py_ssize_t)(buf.outbuf -
1328 PyUnicode_AS_UNICODE(buf.outobj));
1329 Py_DECREF(cres);
1330 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001331
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001332 if (sizehint < 0 || finalsize != 0 || rsize == 0)
1333 break;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001334
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001335 sizehint = 1; /* read 1 more byte and retry */
1336 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001337
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001338 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1339 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1340 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001341
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001342 Py_XDECREF(cres);
1343 Py_XDECREF(buf.excobj);
1344 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001345
1346errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001347 Py_XDECREF(cres);
1348 Py_XDECREF(buf.excobj);
1349 Py_XDECREF(buf.outobj);
1350 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001351}
1352
1353static PyObject *
1354mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args)
1355{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001356 PyObject *sizeobj = NULL;
1357 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001358
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001359 if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj))
1360 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001361
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001362 if (sizeobj == Py_None || sizeobj == NULL)
1363 size = -1;
1364 else if (PyLong_Check(sizeobj))
1365 size = PyLong_AsSsize_t(sizeobj);
1366 else {
1367 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1368 return NULL;
1369 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001370
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001371 if (size == -1 && PyErr_Occurred())
1372 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001373
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001374 return mbstreamreader_iread(self, "read", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001375}
1376
1377static PyObject *
1378mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args)
1379{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001380 PyObject *sizeobj = NULL;
1381 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001382
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001383 if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj))
1384 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001385
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001386 if (sizeobj == Py_None || sizeobj == NULL)
1387 size = -1;
1388 else if (PyLong_Check(sizeobj))
1389 size = PyLong_AsSsize_t(sizeobj);
1390 else {
1391 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1392 return NULL;
1393 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001394
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001395 if (size == -1 && PyErr_Occurred())
1396 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001397
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001398 return mbstreamreader_iread(self, "readline", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001399}
1400
1401static PyObject *
1402mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args)
1403{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001404 PyObject *sizehintobj = NULL, *r, *sr;
1405 Py_ssize_t sizehint;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001406
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001407 if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj))
1408 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001409
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001410 if (sizehintobj == Py_None || sizehintobj == NULL)
1411 sizehint = -1;
1412 else if (PyLong_Check(sizehintobj))
1413 sizehint = PyLong_AsSsize_t(sizehintobj);
1414 else {
1415 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1416 return NULL;
1417 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001418
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001419 if (sizehint == -1 && PyErr_Occurred())
1420 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001421
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001422 r = mbstreamreader_iread(self, "read", sizehint);
1423 if (r == NULL)
1424 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001425
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001426 sr = PyUnicode_Splitlines(r, 1);
1427 Py_DECREF(r);
1428 return sr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001429}
1430
1431static PyObject *
1432mbstreamreader_reset(MultibyteStreamReaderObject *self)
1433{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001434 if (self->codec->decreset != NULL &&
1435 self->codec->decreset(&self->state, self->codec->config) != 0)
1436 return NULL;
1437 self->pendingsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001438
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001439 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001440}
1441
1442static struct PyMethodDef mbstreamreader_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001443 {"read", (PyCFunction)mbstreamreader_read,
1444 METH_VARARGS, NULL},
1445 {"readline", (PyCFunction)mbstreamreader_readline,
1446 METH_VARARGS, NULL},
1447 {"readlines", (PyCFunction)mbstreamreader_readlines,
1448 METH_VARARGS, NULL},
1449 {"reset", (PyCFunction)mbstreamreader_reset,
1450 METH_NOARGS, NULL},
1451 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001452};
1453
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001454static PyMemberDef mbstreamreader_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001455 {"stream", T_OBJECT,
1456 offsetof(MultibyteStreamReaderObject, stream),
1457 READONLY, NULL},
1458 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001459};
1460
1461static PyObject *
1462mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1463{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001464 MultibyteStreamReaderObject *self;
1465 PyObject *stream, *codec = NULL;
1466 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001467
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001468 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
1469 streamkwarglist, &stream, &errors))
1470 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001471
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001472 self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
1473 if (self == NULL)
1474 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001475
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001476 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1477 if (codec == NULL)
1478 goto errorexit;
1479 if (!MultibyteCodec_Check(codec)) {
1480 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1481 goto errorexit;
1482 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001483
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001484 self->codec = ((MultibyteCodecObject *)codec)->codec;
1485 self->stream = stream;
1486 Py_INCREF(stream);
1487 self->pendingsize = 0;
1488 self->errors = internal_error_callback(errors);
1489 if (self->errors == NULL)
1490 goto errorexit;
1491 if (self->codec->decinit != NULL &&
1492 self->codec->decinit(&self->state, self->codec->config) != 0)
1493 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001494
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001495 Py_DECREF(codec);
1496 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001497
1498errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001499 Py_XDECREF(self);
1500 Py_XDECREF(codec);
1501 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001502}
1503
1504static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001505mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
1506{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001507 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001508}
1509
1510static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001511mbstreamreader_traverse(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001512 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001513{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001514 if (ERROR_ISCUSTOM(self->errors))
1515 Py_VISIT(self->errors);
1516 Py_VISIT(self->stream);
1517 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001518}
1519
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001520static void
1521mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
1522{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001523 PyObject_GC_UnTrack(self);
1524 ERROR_DECREF(self->errors);
1525 Py_XDECREF(self->stream);
1526 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001527}
1528
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001529static PyTypeObject MultibyteStreamReader_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001530 PyVarObject_HEAD_INIT(NULL, 0)
1531 "MultibyteStreamReader", /* tp_name */
1532 sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
1533 0, /* tp_itemsize */
1534 /* methods */
1535 (destructor)mbstreamreader_dealloc, /* tp_dealloc */
1536 0, /* tp_print */
1537 0, /* tp_getattr */
1538 0, /* tp_setattr */
1539 0, /* tp_reserved */
1540 0, /* tp_repr */
1541 0, /* tp_as_number */
1542 0, /* tp_as_sequence */
1543 0, /* tp_as_mapping */
1544 0, /* tp_hash */
1545 0, /* tp_call */
1546 0, /* tp_str */
1547 PyObject_GenericGetAttr, /* tp_getattro */
1548 0, /* tp_setattro */
1549 0, /* tp_as_buffer */
1550 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1551 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1552 0, /* tp_doc */
1553 (traverseproc)mbstreamreader_traverse, /* tp_traverse */
1554 0, /* tp_clear */
1555 0, /* tp_richcompare */
1556 0, /* tp_weaklistoffset */
1557 0, /* tp_iter */
1558 0, /* tp_iterext */
1559 mbstreamreader_methods, /* tp_methods */
1560 mbstreamreader_members, /* tp_members */
1561 codecctx_getsets, /* tp_getset */
1562 0, /* tp_base */
1563 0, /* tp_dict */
1564 0, /* tp_descr_get */
1565 0, /* tp_descr_set */
1566 0, /* tp_dictoffset */
1567 mbstreamreader_init, /* tp_init */
1568 0, /* tp_alloc */
1569 mbstreamreader_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001570};
1571
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001572
1573/**
1574 * MultibyteStreamWriter object
1575 */
1576
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001577static int
1578mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001579 PyObject *unistr)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001580{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001581 PyObject *str, *wr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001582
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001583 str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
1584 if (str == NULL)
1585 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001586
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001587 wr = PyObject_CallMethod(self->stream, "write", "O", str);
1588 Py_DECREF(str);
1589 if (wr == NULL)
1590 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001591
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001592 Py_DECREF(wr);
1593 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001594}
1595
1596static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001597mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001598{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001599 if (mbstreamwriter_iwrite(self, strobj))
1600 return NULL;
1601 else
1602 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001603}
1604
1605static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001606mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001607{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001608 PyObject *strobj;
1609 int i, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001610
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001611 if (!PySequence_Check(lines)) {
1612 PyErr_SetString(PyExc_TypeError,
1613 "arg must be a sequence object");
1614 return NULL;
1615 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001616
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001617 for (i = 0; i < PySequence_Length(lines); i++) {
1618 /* length can be changed even within this loop */
1619 strobj = PySequence_GetItem(lines, i);
1620 if (strobj == NULL)
1621 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001622
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001623 r = mbstreamwriter_iwrite(self, strobj);
1624 Py_DECREF(strobj);
1625 if (r == -1)
1626 return NULL;
1627 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001628
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001629 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001630}
1631
1632static PyObject *
1633mbstreamwriter_reset(MultibyteStreamWriterObject *self)
1634{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001635 const Py_UNICODE *pending;
1636 PyObject *pwrt;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001637
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001638 pending = self->pending;
1639 pwrt = multibytecodec_encode(self->codec, &self->state,
1640 &pending, self->pendingsize, self->errors,
1641 MBENC_FLUSH | MBENC_RESET);
1642 /* some pending buffer can be truncated when UnicodeEncodeError is
1643 * raised on 'strict' mode. but, 'reset' method is designed to
1644 * reset the pending buffer or states so failed string sequence
1645 * ought to be missed */
1646 self->pendingsize = 0;
1647 if (pwrt == NULL)
1648 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001649
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001650 assert(PyBytes_Check(pwrt));
1651 if (PyBytes_Size(pwrt) > 0) {
1652 PyObject *wr;
1653 wr = PyObject_CallMethod(self->stream, "write", "O", pwrt);
1654 if (wr == NULL) {
1655 Py_DECREF(pwrt);
1656 return NULL;
1657 }
1658 }
1659 Py_DECREF(pwrt);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001660
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001661 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001662}
1663
1664static PyObject *
1665mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1666{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001667 MultibyteStreamWriterObject *self;
1668 PyObject *stream, *codec = NULL;
1669 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001670
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001671 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
1672 streamkwarglist, &stream, &errors))
1673 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001674
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001675 self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
1676 if (self == NULL)
1677 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001678
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001679 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1680 if (codec == NULL)
1681 goto errorexit;
1682 if (!MultibyteCodec_Check(codec)) {
1683 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1684 goto errorexit;
1685 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001686
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001687 self->codec = ((MultibyteCodecObject *)codec)->codec;
1688 self->stream = stream;
1689 Py_INCREF(stream);
1690 self->pendingsize = 0;
1691 self->errors = internal_error_callback(errors);
1692 if (self->errors == NULL)
1693 goto errorexit;
1694 if (self->codec->encinit != NULL &&
1695 self->codec->encinit(&self->state, self->codec->config) != 0)
1696 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001697
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001698 Py_DECREF(codec);
1699 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001700
1701errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001702 Py_XDECREF(self);
1703 Py_XDECREF(codec);
1704 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001705}
1706
1707static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001708mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
1709{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001710 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001711}
1712
1713static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001714mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001715 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001716{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001717 if (ERROR_ISCUSTOM(self->errors))
1718 Py_VISIT(self->errors);
1719 Py_VISIT(self->stream);
1720 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001721}
1722
1723static void
1724mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
1725{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001726 PyObject_GC_UnTrack(self);
1727 ERROR_DECREF(self->errors);
1728 Py_XDECREF(self->stream);
1729 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001730}
1731
1732static struct PyMethodDef mbstreamwriter_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001733 {"write", (PyCFunction)mbstreamwriter_write,
1734 METH_O, NULL},
1735 {"writelines", (PyCFunction)mbstreamwriter_writelines,
1736 METH_O, NULL},
1737 {"reset", (PyCFunction)mbstreamwriter_reset,
1738 METH_NOARGS, NULL},
1739 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001740};
1741
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001742static PyMemberDef mbstreamwriter_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001743 {"stream", T_OBJECT,
1744 offsetof(MultibyteStreamWriterObject, stream),
1745 READONLY, NULL},
1746 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001747};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001748
1749static PyTypeObject MultibyteStreamWriter_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001750 PyVarObject_HEAD_INIT(NULL, 0)
1751 "MultibyteStreamWriter", /* tp_name */
1752 sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
1753 0, /* tp_itemsize */
1754 /* methods */
1755 (destructor)mbstreamwriter_dealloc, /* tp_dealloc */
1756 0, /* tp_print */
1757 0, /* tp_getattr */
1758 0, /* tp_setattr */
1759 0, /* tp_reserved */
1760 0, /* tp_repr */
1761 0, /* tp_as_number */
1762 0, /* tp_as_sequence */
1763 0, /* tp_as_mapping */
1764 0, /* tp_hash */
1765 0, /* tp_call */
1766 0, /* tp_str */
1767 PyObject_GenericGetAttr, /* tp_getattro */
1768 0, /* tp_setattro */
1769 0, /* tp_as_buffer */
1770 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1771 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1772 0, /* tp_doc */
1773 (traverseproc)mbstreamwriter_traverse, /* tp_traverse */
1774 0, /* tp_clear */
1775 0, /* tp_richcompare */
1776 0, /* tp_weaklistoffset */
1777 0, /* tp_iter */
1778 0, /* tp_iterext */
1779 mbstreamwriter_methods, /* tp_methods */
1780 mbstreamwriter_members, /* tp_members */
1781 codecctx_getsets, /* tp_getset */
1782 0, /* tp_base */
1783 0, /* tp_dict */
1784 0, /* tp_descr_get */
1785 0, /* tp_descr_set */
1786 0, /* tp_dictoffset */
1787 mbstreamwriter_init, /* tp_init */
1788 0, /* tp_alloc */
1789 mbstreamwriter_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001790};
1791
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001792
1793/**
1794 * Exposed factory function
1795 */
1796
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001797static PyObject *
1798__create_codec(PyObject *ignore, PyObject *arg)
1799{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001800 MultibyteCodecObject *self;
1801 MultibyteCodec *codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001802
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001803 if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
1804 PyErr_SetString(PyExc_ValueError, "argument type invalid");
1805 return NULL;
1806 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001807
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001808 codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
1809 if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
1810 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001811
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001812 self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
1813 if (self == NULL)
1814 return NULL;
1815 self->codec = codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001816
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001817 return (PyObject *)self;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001818}
1819
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001820static struct PyMethodDef __methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001821 {"__create_codec", (PyCFunction)__create_codec, METH_O},
1822 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001823};
1824
Martin v. Löwis1a214512008-06-11 05:26:20 +00001825
1826static struct PyModuleDef _multibytecodecmodule = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001827 PyModuleDef_HEAD_INIT,
1828 "_multibytecodec",
1829 NULL,
1830 -1,
1831 __methods,
1832 NULL,
1833 NULL,
1834 NULL,
1835 NULL
Martin v. Löwis1a214512008-06-11 05:26:20 +00001836};
1837
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001838PyMODINIT_FUNC
Martin v. Löwis1a214512008-06-11 05:26:20 +00001839PyInit__multibytecodec(void)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001840{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001841 int i;
1842 PyObject *m;
1843 PyTypeObject *typelist[] = {
1844 &MultibyteIncrementalEncoder_Type,
1845 &MultibyteIncrementalDecoder_Type,
1846 &MultibyteStreamReader_Type,
1847 &MultibyteStreamWriter_Type,
1848 NULL
1849 };
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001850
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001851 if (PyType_Ready(&MultibyteCodec_Type) < 0)
1852 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001853
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001854 m = PyModule_Create(&_multibytecodecmodule);
1855 if (m == NULL)
1856 return NULL;
Neal Norwitz058bde12005-09-21 06:44:25 +00001857
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001858 for (i = 0; typelist[i] != NULL; i++) {
1859 if (PyType_Ready(typelist[i]) < 0)
1860 return NULL;
1861 Py_INCREF(typelist[i]);
1862 PyModule_AddObject(m, typelist[i]->tp_name,
1863 (PyObject *)typelist[i]);
1864 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001865
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001866 if (PyErr_Occurred()) {
1867 Py_FatalError("can't initialize the _multibytecodec module");
1868 Py_DECREF(m);
1869 m = NULL;
1870 }
1871 return m;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001872}