blob: af7ea5b83a8b5357fd653150c86f16eb4edfc8da [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
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000482 if (datalen == 0)
483 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;
486 buf.inbuf = buf.inbuf_top = *data;
487 buf.inbuf_end = buf.inbuf_top + datalen;
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000488
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000489 if (datalen > (PY_SSIZE_T_MAX - 16) / 2) {
490 PyErr_NoMemory();
491 goto errorexit;
492 }
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000493
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000494 buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16);
495 if (buf.outobj == NULL)
496 goto errorexit;
497 buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj);
498 buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000499
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000500 while (buf.inbuf < buf.inbuf_end) {
501 Py_ssize_t inleft, outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000502
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000503 /* we don't reuse inleft and outleft here.
504 * error callbacks can relocate the cursor anywhere on buffer*/
505 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
506 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
507 r = codec->encode(state, codec->config, &buf.inbuf, inleft,
508 &buf.outbuf, outleft, flags);
509 if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH)))
510 break;
511 else if (multibytecodec_encerror(codec, state, &buf, errors,r))
512 goto errorexit;
513 else if (r == MBERR_TOOFEW)
514 break;
515 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000516
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000517 if (codec->encreset != NULL)
518 for (;;) {
519 Py_ssize_t outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000520
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000521 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
522 r = codec->encreset(state, codec->config, &buf.outbuf,
523 outleft);
524 if (r == 0)
525 break;
526 else if (multibytecodec_encerror(codec, state,
527 &buf, errors, r))
528 goto errorexit;
529 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000530
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000531 finalsize = (Py_ssize_t)((char *)buf.outbuf -
532 PyBytes_AS_STRING(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000533
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000534 if (finalsize != PyBytes_GET_SIZE(buf.outobj))
535 if (_PyBytes_Resize(&buf.outobj, finalsize) == -1)
536 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000537
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000538 *data = buf.inbuf;
539 Py_XDECREF(buf.excobj);
540 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000541
542errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000543 Py_XDECREF(buf.excobj);
544 Py_XDECREF(buf.outobj);
545 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000546}
547
548static PyObject *
549MultibyteCodec_Encode(MultibyteCodecObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000550 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000551{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000552 MultibyteCodec_State state;
553 Py_UNICODE *data;
554 PyObject *errorcb, *r, *arg, *ucvt;
555 const char *errors = NULL;
556 Py_ssize_t datalen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000557
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000558 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode",
559 codeckwarglist, &arg, &errors))
560 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000561
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000562 if (PyUnicode_Check(arg))
563 ucvt = NULL;
564 else {
565 arg = ucvt = PyObject_Str(arg);
566 if (arg == NULL)
567 return NULL;
568 else if (!PyUnicode_Check(arg)) {
569 PyErr_SetString(PyExc_TypeError,
570 "couldn't convert the object to unicode.");
571 Py_DECREF(ucvt);
572 return NULL;
573 }
574 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000575
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000576 data = PyUnicode_AS_UNICODE(arg);
577 datalen = PyUnicode_GET_SIZE(arg);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000578
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000579 errorcb = internal_error_callback(errors);
580 if (errorcb == NULL) {
581 Py_XDECREF(ucvt);
582 return NULL;
583 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000584
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000585 if (self->codec->encinit != NULL &&
586 self->codec->encinit(&state, self->codec->config) != 0)
587 goto errorexit;
588 r = multibytecodec_encode(self->codec, &state,
589 (const Py_UNICODE **)&data, datalen, errorcb,
590 MBENC_FLUSH | MBENC_RESET);
591 if (r == NULL)
592 goto errorexit;
Hye-Shik Chang2bb146f2004-07-18 03:06:29 +0000593
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000594 ERROR_DECREF(errorcb);
595 Py_XDECREF(ucvt);
596 return make_tuple(r, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000597
598errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000599 ERROR_DECREF(errorcb);
600 Py_XDECREF(ucvt);
601 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000602}
603
604static PyObject *
605MultibyteCodec_Decode(MultibyteCodecObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000606 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000607{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000608 MultibyteCodec_State state;
609 MultibyteDecodeBuffer buf;
610 PyObject *errorcb;
611 Py_buffer pdata;
612 const char *data, *errors = NULL;
613 Py_ssize_t datalen, finalsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000614
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000615 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|z:decode",
616 codeckwarglist, &pdata, &errors))
617 return NULL;
618 data = pdata.buf;
619 datalen = pdata.len;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000620
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000621 errorcb = internal_error_callback(errors);
622 if (errorcb == NULL) {
623 PyBuffer_Release(&pdata);
624 return NULL;
625 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000626
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000627 if (datalen == 0) {
628 PyBuffer_Release(&pdata);
629 ERROR_DECREF(errorcb);
630 return make_tuple(PyUnicode_FromUnicode(NULL, 0), 0);
631 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000632
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000633 buf.excobj = NULL;
634 buf.inbuf = buf.inbuf_top = (unsigned char *)data;
635 buf.inbuf_end = buf.inbuf_top + datalen;
636 buf.outobj = PyUnicode_FromUnicode(NULL, datalen);
637 if (buf.outobj == NULL)
638 goto errorexit;
639 buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj);
640 buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000641
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000642 if (self->codec->decinit != NULL &&
643 self->codec->decinit(&state, self->codec->config) != 0)
644 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000645
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000646 while (buf.inbuf < buf.inbuf_end) {
647 Py_ssize_t inleft, outleft, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000648
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000649 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
650 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000651
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000652 r = self->codec->decode(&state, self->codec->config,
653 &buf.inbuf, inleft, &buf.outbuf, outleft);
654 if (r == 0)
655 break;
656 else if (multibytecodec_decerror(self->codec, &state,
657 &buf, errorcb, r))
658 goto errorexit;
659 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000660
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000661 finalsize = (Py_ssize_t)(buf.outbuf -
662 PyUnicode_AS_UNICODE(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000663
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000664 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
665 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
666 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000667
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000668 PyBuffer_Release(&pdata);
669 Py_XDECREF(buf.excobj);
670 ERROR_DECREF(errorcb);
671 return make_tuple(buf.outobj, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000672
673errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000674 PyBuffer_Release(&pdata);
675 ERROR_DECREF(errorcb);
676 Py_XDECREF(buf.excobj);
677 Py_XDECREF(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000678
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000679 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000680}
681
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000682static struct PyMethodDef multibytecodec_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000683 {"encode", (PyCFunction)MultibyteCodec_Encode,
684 METH_VARARGS | METH_KEYWORDS,
685 MultibyteCodec_Encode__doc__},
686 {"decode", (PyCFunction)MultibyteCodec_Decode,
687 METH_VARARGS | METH_KEYWORDS,
688 MultibyteCodec_Decode__doc__},
689 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000690};
691
692static void
693multibytecodec_dealloc(MultibyteCodecObject *self)
694{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000695 PyObject_Del(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000696}
697
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000698static PyTypeObject MultibyteCodec_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000699 PyVarObject_HEAD_INIT(NULL, 0)
700 "MultibyteCodec", /* tp_name */
701 sizeof(MultibyteCodecObject), /* tp_basicsize */
702 0, /* tp_itemsize */
703 /* methods */
704 (destructor)multibytecodec_dealloc, /* tp_dealloc */
705 0, /* tp_print */
706 0, /* tp_getattr */
707 0, /* tp_setattr */
708 0, /* tp_reserved */
709 0, /* tp_repr */
710 0, /* tp_as_number */
711 0, /* tp_as_sequence */
712 0, /* tp_as_mapping */
713 0, /* tp_hash */
714 0, /* tp_call */
715 0, /* tp_str */
716 PyObject_GenericGetAttr, /* tp_getattro */
717 0, /* tp_setattro */
718 0, /* tp_as_buffer */
719 Py_TPFLAGS_DEFAULT, /* tp_flags */
720 0, /* tp_doc */
721 0, /* tp_traverse */
722 0, /* tp_clear */
723 0, /* tp_richcompare */
724 0, /* tp_weaklistoffset */
725 0, /* tp_iter */
726 0, /* tp_iterext */
727 multibytecodec_methods, /* tp_methods */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000728};
729
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000730
731/**
732 * Utility functions for stateful codec mechanism
733 */
734
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000735#define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o))
736#define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o))
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000737
738static PyObject *
739encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000740 PyObject *unistr, int final)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000741{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000742 PyObject *ucvt, *r = NULL;
743 Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL;
744 Py_ssize_t datalen, origpending;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000745
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000746 if (PyUnicode_Check(unistr))
747 ucvt = NULL;
748 else {
749 unistr = ucvt = PyObject_Str(unistr);
750 if (unistr == NULL)
751 return NULL;
752 else if (!PyUnicode_Check(unistr)) {
753 PyErr_SetString(PyExc_TypeError,
754 "couldn't convert the object to unicode.");
755 Py_DECREF(ucvt);
756 return NULL;
757 }
758 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000759
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000760 datalen = PyUnicode_GET_SIZE(unistr);
761 origpending = ctx->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000762
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000763 if (origpending > 0) {
764 if (datalen > PY_SSIZE_T_MAX - ctx->pendingsize) {
765 PyErr_NoMemory();
766 /* inbuf_tmp == NULL */
767 goto errorexit;
768 }
769 inbuf_tmp = PyMem_New(Py_UNICODE, datalen + ctx->pendingsize);
770 if (inbuf_tmp == NULL)
771 goto errorexit;
772 memcpy(inbuf_tmp, ctx->pending,
773 Py_UNICODE_SIZE * ctx->pendingsize);
774 memcpy(inbuf_tmp + ctx->pendingsize,
775 PyUnicode_AS_UNICODE(unistr),
776 Py_UNICODE_SIZE * datalen);
777 datalen += ctx->pendingsize;
778 ctx->pendingsize = 0;
779 inbuf = inbuf_tmp;
780 }
781 else
782 inbuf = (Py_UNICODE *)PyUnicode_AS_UNICODE(unistr);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000783
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000784 inbuf_end = inbuf + datalen;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000785
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000786 r = multibytecodec_encode(ctx->codec, &ctx->state,
787 (const Py_UNICODE **)&inbuf,
788 datalen, ctx->errors, final ? MBENC_FLUSH : 0);
789 if (r == NULL) {
790 /* recover the original pending buffer */
791 if (origpending > 0)
792 memcpy(ctx->pending, inbuf_tmp,
793 Py_UNICODE_SIZE * origpending);
794 ctx->pendingsize = origpending;
795 goto errorexit;
796 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000797
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000798 if (inbuf < inbuf_end) {
799 ctx->pendingsize = (Py_ssize_t)(inbuf_end - inbuf);
800 if (ctx->pendingsize > MAXENCPENDING) {
801 /* normal codecs can't reach here */
802 ctx->pendingsize = 0;
803 PyErr_SetString(PyExc_UnicodeError,
804 "pending buffer overflow");
805 goto errorexit;
806 }
807 memcpy(ctx->pending, inbuf,
808 ctx->pendingsize * Py_UNICODE_SIZE);
809 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000810
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000811 if (inbuf_tmp != NULL)
812 PyMem_Del(inbuf_tmp);
813 Py_XDECREF(ucvt);
814 return r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000815
816errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000817 if (inbuf_tmp != NULL)
818 PyMem_Del(inbuf_tmp);
819 Py_XDECREF(r);
820 Py_XDECREF(ucvt);
821 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000822}
823
824static int
825decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000826 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000827{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000828 Py_ssize_t npendings;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000829
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000830 npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
831 if (npendings + ctx->pendingsize > MAXDECPENDING ||
832 npendings > PY_SSIZE_T_MAX - ctx->pendingsize) {
833 PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow");
834 return -1;
835 }
836 memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings);
837 ctx->pendingsize += npendings;
838 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000839}
840
841static int
842decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000843 Py_ssize_t size)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000844{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000845 buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
846 buf->inbuf_end = buf->inbuf_top + size;
847 if (buf->outobj == NULL) { /* only if outobj is not allocated yet */
848 buf->outobj = PyUnicode_FromUnicode(NULL, size);
849 if (buf->outobj == NULL)
850 return -1;
851 buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj);
852 buf->outbuf_end = buf->outbuf +
853 PyUnicode_GET_SIZE(buf->outobj);
854 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000855
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000856 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000857}
858
859static int
860decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000861 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000862{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000863 while (buf->inbuf < buf->inbuf_end) {
864 Py_ssize_t inleft, outleft;
865 Py_ssize_t r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000866
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000867 inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
868 outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000869
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000870 r = ctx->codec->decode(&ctx->state, ctx->codec->config,
871 &buf->inbuf, inleft, &buf->outbuf, outleft);
872 if (r == 0 || r == MBERR_TOOFEW)
873 break;
874 else if (multibytecodec_decerror(ctx->codec, &ctx->state,
875 buf, ctx->errors, r))
876 return -1;
877 }
878 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000879}
880
881
882/**
883 * MultibyteIncrementalEncoder object
884 */
885
886static PyObject *
887mbiencoder_encode(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000888 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000889{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000890 PyObject *data;
891 int final = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000892
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000893 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode",
894 incrementalkwarglist, &data, &final))
895 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000896
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000897 return encoder_encode_stateful(STATEFUL_ECTX(self), data, final);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000898}
899
900static PyObject *
901mbiencoder_reset(MultibyteIncrementalEncoderObject *self)
902{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000903 if (self->codec->decreset != NULL &&
904 self->codec->decreset(&self->state, self->codec->config) != 0)
905 return NULL;
906 self->pendingsize = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000907
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000908 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000909}
910
911static struct PyMethodDef mbiencoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000912 {"encode", (PyCFunction)mbiencoder_encode,
913 METH_VARARGS | METH_KEYWORDS, NULL},
914 {"reset", (PyCFunction)mbiencoder_reset,
915 METH_NOARGS, NULL},
916 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000917};
918
919static PyObject *
920mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
921{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000922 MultibyteIncrementalEncoderObject *self;
923 PyObject *codec = NULL;
924 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000925
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000926 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
927 incnewkwarglist, &errors))
928 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000929
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000930 self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
931 if (self == NULL)
932 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000933
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000934 codec = PyObject_GetAttrString((PyObject *)type, "codec");
935 if (codec == NULL)
936 goto errorexit;
937 if (!MultibyteCodec_Check(codec)) {
938 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
939 goto errorexit;
940 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000941
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000942 self->codec = ((MultibyteCodecObject *)codec)->codec;
943 self->pendingsize = 0;
944 self->errors = internal_error_callback(errors);
945 if (self->errors == NULL)
946 goto errorexit;
947 if (self->codec->encinit != NULL &&
948 self->codec->encinit(&self->state, self->codec->config) != 0)
949 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000950
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000951 Py_DECREF(codec);
952 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000953
954errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000955 Py_XDECREF(self);
956 Py_XDECREF(codec);
957 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000958}
959
960static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000961mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
962{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000963 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000964}
965
966static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000967mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000968 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000969{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000970 if (ERROR_ISCUSTOM(self->errors))
971 Py_VISIT(self->errors);
972 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000973}
974
975static void
976mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
977{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000978 PyObject_GC_UnTrack(self);
979 ERROR_DECREF(self->errors);
980 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000981}
982
983static PyTypeObject MultibyteIncrementalEncoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000984 PyVarObject_HEAD_INIT(NULL, 0)
985 "MultibyteIncrementalEncoder", /* tp_name */
986 sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
987 0, /* tp_itemsize */
988 /* methods */
989 (destructor)mbiencoder_dealloc, /* tp_dealloc */
990 0, /* tp_print */
991 0, /* tp_getattr */
992 0, /* tp_setattr */
993 0, /* tp_reserved */
994 0, /* tp_repr */
995 0, /* tp_as_number */
996 0, /* tp_as_sequence */
997 0, /* tp_as_mapping */
998 0, /* tp_hash */
999 0, /* tp_call */
1000 0, /* tp_str */
1001 PyObject_GenericGetAttr, /* tp_getattro */
1002 0, /* tp_setattro */
1003 0, /* tp_as_buffer */
1004 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1005 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1006 0, /* tp_doc */
1007 (traverseproc)mbiencoder_traverse, /* tp_traverse */
1008 0, /* tp_clear */
1009 0, /* tp_richcompare */
1010 0, /* tp_weaklistoffset */
1011 0, /* tp_iter */
1012 0, /* tp_iterext */
1013 mbiencoder_methods, /* tp_methods */
1014 0, /* tp_members */
1015 codecctx_getsets, /* tp_getset */
1016 0, /* tp_base */
1017 0, /* tp_dict */
1018 0, /* tp_descr_get */
1019 0, /* tp_descr_set */
1020 0, /* tp_dictoffset */
1021 mbiencoder_init, /* tp_init */
1022 0, /* tp_alloc */
1023 mbiencoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001024};
1025
1026
1027/**
1028 * MultibyteIncrementalDecoder object
1029 */
1030
1031static PyObject *
1032mbidecoder_decode(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001033 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001034{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001035 MultibyteDecodeBuffer buf;
1036 char *data, *wdata = NULL;
1037 Py_buffer pdata;
1038 Py_ssize_t wsize, finalsize = 0, size, origpending;
1039 int final = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001040
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001041 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|i:decode",
1042 incrementalkwarglist, &pdata, &final))
1043 return NULL;
1044 data = pdata.buf;
1045 size = pdata.len;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001046
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001047 buf.outobj = buf.excobj = NULL;
1048 origpending = self->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001049
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001050 if (self->pendingsize == 0) {
1051 wsize = size;
1052 wdata = data;
1053 }
1054 else {
1055 if (size > PY_SSIZE_T_MAX - self->pendingsize) {
1056 PyErr_NoMemory();
1057 goto errorexit;
1058 }
1059 wsize = size + self->pendingsize;
1060 wdata = PyMem_Malloc(wsize);
1061 if (wdata == NULL)
1062 goto errorexit;
1063 memcpy(wdata, self->pending, self->pendingsize);
1064 memcpy(wdata + self->pendingsize, data, size);
1065 self->pendingsize = 0;
1066 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001067
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001068 if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
1069 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001070
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001071 if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
1072 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001073
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001074 if (final && buf.inbuf < buf.inbuf_end) {
1075 if (multibytecodec_decerror(self->codec, &self->state,
1076 &buf, self->errors, MBERR_TOOFEW)) {
1077 /* recover the original pending buffer */
1078 memcpy(self->pending, wdata, origpending);
1079 self->pendingsize = origpending;
1080 goto errorexit;
1081 }
1082 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001083
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001084 if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
1085 if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
1086 goto errorexit;
1087 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001088
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001089 finalsize = (Py_ssize_t)(buf.outbuf - PyUnicode_AS_UNICODE(buf.outobj));
1090 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1091 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1092 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001093
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001094 PyBuffer_Release(&pdata);
1095 if (wdata != data)
1096 PyMem_Del(wdata);
1097 Py_XDECREF(buf.excobj);
1098 return buf.outobj;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001099
1100errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001101 PyBuffer_Release(&pdata);
1102 if (wdata != NULL && wdata != data)
1103 PyMem_Del(wdata);
1104 Py_XDECREF(buf.excobj);
1105 Py_XDECREF(buf.outobj);
1106 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001107}
1108
1109static PyObject *
1110mbidecoder_reset(MultibyteIncrementalDecoderObject *self)
1111{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001112 if (self->codec->decreset != NULL &&
1113 self->codec->decreset(&self->state, self->codec->config) != 0)
1114 return NULL;
1115 self->pendingsize = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001116
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001117 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001118}
1119
1120static struct PyMethodDef mbidecoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001121 {"decode", (PyCFunction)mbidecoder_decode,
1122 METH_VARARGS | METH_KEYWORDS, NULL},
1123 {"reset", (PyCFunction)mbidecoder_reset,
1124 METH_NOARGS, NULL},
1125 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001126};
1127
1128static PyObject *
1129mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1130{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001131 MultibyteIncrementalDecoderObject *self;
1132 PyObject *codec = NULL;
1133 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001134
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001135 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
1136 incnewkwarglist, &errors))
1137 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001138
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001139 self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
1140 if (self == NULL)
1141 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001142
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001143 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1144 if (codec == NULL)
1145 goto errorexit;
1146 if (!MultibyteCodec_Check(codec)) {
1147 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1148 goto errorexit;
1149 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001150
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001151 self->codec = ((MultibyteCodecObject *)codec)->codec;
1152 self->pendingsize = 0;
1153 self->errors = internal_error_callback(errors);
1154 if (self->errors == NULL)
1155 goto errorexit;
1156 if (self->codec->decinit != NULL &&
1157 self->codec->decinit(&self->state, self->codec->config) != 0)
1158 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001159
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001160 Py_DECREF(codec);
1161 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001162
1163errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001164 Py_XDECREF(self);
1165 Py_XDECREF(codec);
1166 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001167}
1168
1169static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001170mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
1171{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001172 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001173}
1174
1175static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001176mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001177 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001178{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001179 if (ERROR_ISCUSTOM(self->errors))
1180 Py_VISIT(self->errors);
1181 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001182}
1183
1184static void
1185mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
1186{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001187 PyObject_GC_UnTrack(self);
1188 ERROR_DECREF(self->errors);
1189 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001190}
1191
1192static PyTypeObject MultibyteIncrementalDecoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001193 PyVarObject_HEAD_INIT(NULL, 0)
1194 "MultibyteIncrementalDecoder", /* tp_name */
1195 sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */
1196 0, /* tp_itemsize */
1197 /* methods */
1198 (destructor)mbidecoder_dealloc, /* tp_dealloc */
1199 0, /* tp_print */
1200 0, /* tp_getattr */
1201 0, /* tp_setattr */
1202 0, /* tp_reserved */
1203 0, /* tp_repr */
1204 0, /* tp_as_number */
1205 0, /* tp_as_sequence */
1206 0, /* tp_as_mapping */
1207 0, /* tp_hash */
1208 0, /* tp_call */
1209 0, /* tp_str */
1210 PyObject_GenericGetAttr, /* tp_getattro */
1211 0, /* tp_setattro */
1212 0, /* tp_as_buffer */
1213 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1214 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1215 0, /* tp_doc */
1216 (traverseproc)mbidecoder_traverse, /* tp_traverse */
1217 0, /* tp_clear */
1218 0, /* tp_richcompare */
1219 0, /* tp_weaklistoffset */
1220 0, /* tp_iter */
1221 0, /* tp_iterext */
1222 mbidecoder_methods, /* tp_methods */
1223 0, /* tp_members */
1224 codecctx_getsets, /* tp_getset */
1225 0, /* tp_base */
1226 0, /* tp_dict */
1227 0, /* tp_descr_get */
1228 0, /* tp_descr_set */
1229 0, /* tp_dictoffset */
1230 mbidecoder_init, /* tp_init */
1231 0, /* tp_alloc */
1232 mbidecoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001233};
1234
1235
1236/**
1237 * MultibyteStreamReader object
1238 */
1239
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001240static PyObject *
1241mbstreamreader_iread(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001242 const char *method, Py_ssize_t sizehint)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001243{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001244 MultibyteDecodeBuffer buf;
1245 PyObject *cres;
1246 Py_ssize_t rsize, finalsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001247
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001248 if (sizehint == 0)
1249 return PyUnicode_FromUnicode(NULL, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001250
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001251 buf.outobj = buf.excobj = NULL;
1252 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001253
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001254 for (;;) {
1255 int endoffile;
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001256
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001257 if (sizehint < 0)
1258 cres = PyObject_CallMethod(self->stream,
1259 (char *)method, NULL);
1260 else
1261 cres = PyObject_CallMethod(self->stream,
1262 (char *)method, "i", sizehint);
1263 if (cres == NULL)
1264 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001265
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001266 if (!PyBytes_Check(cres)) {
1267 PyErr_Format(PyExc_TypeError,
1268 "stream function returned a "
1269 "non-bytes object (%.100s)",
1270 cres->ob_type->tp_name);
1271 goto errorexit;
1272 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001273
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001274 endoffile = (PyBytes_GET_SIZE(cres) == 0);
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001275
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001276 if (self->pendingsize > 0) {
1277 PyObject *ctr;
1278 char *ctrdata;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001279
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001280 if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
1281 PyErr_NoMemory();
1282 goto errorexit;
1283 }
1284 rsize = PyBytes_GET_SIZE(cres) + self->pendingsize;
1285 ctr = PyBytes_FromStringAndSize(NULL, rsize);
1286 if (ctr == NULL)
1287 goto errorexit;
1288 ctrdata = PyBytes_AS_STRING(ctr);
1289 memcpy(ctrdata, self->pending, self->pendingsize);
1290 memcpy(ctrdata + self->pendingsize,
1291 PyBytes_AS_STRING(cres),
1292 PyBytes_GET_SIZE(cres));
1293 Py_DECREF(cres);
1294 cres = ctr;
1295 self->pendingsize = 0;
1296 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001297
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001298 rsize = PyBytes_GET_SIZE(cres);
1299 if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres),
1300 rsize) != 0)
1301 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001302
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001303 if (rsize > 0 && decoder_feed_buffer(
1304 (MultibyteStatefulDecoderContext *)self, &buf))
1305 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001306
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001307 if (endoffile || sizehint < 0) {
1308 if (buf.inbuf < buf.inbuf_end &&
1309 multibytecodec_decerror(self->codec, &self->state,
1310 &buf, self->errors, MBERR_TOOFEW))
1311 goto errorexit;
1312 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001313
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001314 if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
1315 if (decoder_append_pending(STATEFUL_DCTX(self),
1316 &buf) != 0)
1317 goto errorexit;
1318 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001319
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001320 finalsize = (Py_ssize_t)(buf.outbuf -
1321 PyUnicode_AS_UNICODE(buf.outobj));
1322 Py_DECREF(cres);
1323 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001324
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001325 if (sizehint < 0 || finalsize != 0 || rsize == 0)
1326 break;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001327
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001328 sizehint = 1; /* read 1 more byte and retry */
1329 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001330
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001331 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1332 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1333 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001334
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001335 Py_XDECREF(cres);
1336 Py_XDECREF(buf.excobj);
1337 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001338
1339errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001340 Py_XDECREF(cres);
1341 Py_XDECREF(buf.excobj);
1342 Py_XDECREF(buf.outobj);
1343 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001344}
1345
1346static PyObject *
1347mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args)
1348{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001349 PyObject *sizeobj = NULL;
1350 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001351
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001352 if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj))
1353 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001354
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001355 if (sizeobj == Py_None || sizeobj == NULL)
1356 size = -1;
1357 else if (PyLong_Check(sizeobj))
1358 size = PyLong_AsSsize_t(sizeobj);
1359 else {
1360 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1361 return NULL;
1362 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001363
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001364 if (size == -1 && PyErr_Occurred())
1365 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001366
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001367 return mbstreamreader_iread(self, "read", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001368}
1369
1370static PyObject *
1371mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args)
1372{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001373 PyObject *sizeobj = NULL;
1374 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001375
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001376 if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj))
1377 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001378
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001379 if (sizeobj == Py_None || sizeobj == NULL)
1380 size = -1;
1381 else if (PyLong_Check(sizeobj))
1382 size = PyLong_AsSsize_t(sizeobj);
1383 else {
1384 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1385 return NULL;
1386 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001387
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001388 if (size == -1 && PyErr_Occurred())
1389 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001390
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001391 return mbstreamreader_iread(self, "readline", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001392}
1393
1394static PyObject *
1395mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args)
1396{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001397 PyObject *sizehintobj = NULL, *r, *sr;
1398 Py_ssize_t sizehint;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001399
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001400 if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj))
1401 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001402
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001403 if (sizehintobj == Py_None || sizehintobj == NULL)
1404 sizehint = -1;
1405 else if (PyLong_Check(sizehintobj))
1406 sizehint = PyLong_AsSsize_t(sizehintobj);
1407 else {
1408 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1409 return NULL;
1410 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001411
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001412 if (sizehint == -1 && PyErr_Occurred())
1413 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001414
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001415 r = mbstreamreader_iread(self, "read", sizehint);
1416 if (r == NULL)
1417 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001418
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001419 sr = PyUnicode_Splitlines(r, 1);
1420 Py_DECREF(r);
1421 return sr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001422}
1423
1424static PyObject *
1425mbstreamreader_reset(MultibyteStreamReaderObject *self)
1426{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001427 if (self->codec->decreset != NULL &&
1428 self->codec->decreset(&self->state, self->codec->config) != 0)
1429 return NULL;
1430 self->pendingsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001431
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001432 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001433}
1434
1435static struct PyMethodDef mbstreamreader_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001436 {"read", (PyCFunction)mbstreamreader_read,
1437 METH_VARARGS, NULL},
1438 {"readline", (PyCFunction)mbstreamreader_readline,
1439 METH_VARARGS, NULL},
1440 {"readlines", (PyCFunction)mbstreamreader_readlines,
1441 METH_VARARGS, NULL},
1442 {"reset", (PyCFunction)mbstreamreader_reset,
1443 METH_NOARGS, NULL},
1444 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001445};
1446
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001447static PyMemberDef mbstreamreader_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001448 {"stream", T_OBJECT,
1449 offsetof(MultibyteStreamReaderObject, stream),
1450 READONLY, NULL},
1451 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001452};
1453
1454static PyObject *
1455mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1456{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001457 MultibyteStreamReaderObject *self;
1458 PyObject *stream, *codec = NULL;
1459 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001460
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001461 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
1462 streamkwarglist, &stream, &errors))
1463 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001464
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001465 self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
1466 if (self == NULL)
1467 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001468
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001469 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1470 if (codec == NULL)
1471 goto errorexit;
1472 if (!MultibyteCodec_Check(codec)) {
1473 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1474 goto errorexit;
1475 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001476
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001477 self->codec = ((MultibyteCodecObject *)codec)->codec;
1478 self->stream = stream;
1479 Py_INCREF(stream);
1480 self->pendingsize = 0;
1481 self->errors = internal_error_callback(errors);
1482 if (self->errors == NULL)
1483 goto errorexit;
1484 if (self->codec->decinit != NULL &&
1485 self->codec->decinit(&self->state, self->codec->config) != 0)
1486 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001487
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001488 Py_DECREF(codec);
1489 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001490
1491errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001492 Py_XDECREF(self);
1493 Py_XDECREF(codec);
1494 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001495}
1496
1497static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001498mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
1499{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001500 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001501}
1502
1503static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001504mbstreamreader_traverse(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001505 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001506{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001507 if (ERROR_ISCUSTOM(self->errors))
1508 Py_VISIT(self->errors);
1509 Py_VISIT(self->stream);
1510 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001511}
1512
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001513static void
1514mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
1515{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001516 PyObject_GC_UnTrack(self);
1517 ERROR_DECREF(self->errors);
1518 Py_XDECREF(self->stream);
1519 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001520}
1521
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001522static PyTypeObject MultibyteStreamReader_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001523 PyVarObject_HEAD_INIT(NULL, 0)
1524 "MultibyteStreamReader", /* tp_name */
1525 sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
1526 0, /* tp_itemsize */
1527 /* methods */
1528 (destructor)mbstreamreader_dealloc, /* tp_dealloc */
1529 0, /* tp_print */
1530 0, /* tp_getattr */
1531 0, /* tp_setattr */
1532 0, /* tp_reserved */
1533 0, /* tp_repr */
1534 0, /* tp_as_number */
1535 0, /* tp_as_sequence */
1536 0, /* tp_as_mapping */
1537 0, /* tp_hash */
1538 0, /* tp_call */
1539 0, /* tp_str */
1540 PyObject_GenericGetAttr, /* tp_getattro */
1541 0, /* tp_setattro */
1542 0, /* tp_as_buffer */
1543 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1544 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1545 0, /* tp_doc */
1546 (traverseproc)mbstreamreader_traverse, /* tp_traverse */
1547 0, /* tp_clear */
1548 0, /* tp_richcompare */
1549 0, /* tp_weaklistoffset */
1550 0, /* tp_iter */
1551 0, /* tp_iterext */
1552 mbstreamreader_methods, /* tp_methods */
1553 mbstreamreader_members, /* tp_members */
1554 codecctx_getsets, /* tp_getset */
1555 0, /* tp_base */
1556 0, /* tp_dict */
1557 0, /* tp_descr_get */
1558 0, /* tp_descr_set */
1559 0, /* tp_dictoffset */
1560 mbstreamreader_init, /* tp_init */
1561 0, /* tp_alloc */
1562 mbstreamreader_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001563};
1564
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001565
1566/**
1567 * MultibyteStreamWriter object
1568 */
1569
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001570static int
1571mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001572 PyObject *unistr)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001573{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001574 PyObject *str, *wr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001575
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001576 str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
1577 if (str == NULL)
1578 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001579
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001580 wr = PyObject_CallMethod(self->stream, "write", "O", str);
1581 Py_DECREF(str);
1582 if (wr == NULL)
1583 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001584
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001585 Py_DECREF(wr);
1586 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001587}
1588
1589static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001590mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001591{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001592 if (mbstreamwriter_iwrite(self, strobj))
1593 return NULL;
1594 else
1595 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001596}
1597
1598static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001599mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001600{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001601 PyObject *strobj;
1602 int i, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001603
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001604 if (!PySequence_Check(lines)) {
1605 PyErr_SetString(PyExc_TypeError,
1606 "arg must be a sequence object");
1607 return NULL;
1608 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001609
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001610 for (i = 0; i < PySequence_Length(lines); i++) {
1611 /* length can be changed even within this loop */
1612 strobj = PySequence_GetItem(lines, i);
1613 if (strobj == NULL)
1614 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001615
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001616 r = mbstreamwriter_iwrite(self, strobj);
1617 Py_DECREF(strobj);
1618 if (r == -1)
1619 return NULL;
1620 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001621
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001622 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001623}
1624
1625static PyObject *
1626mbstreamwriter_reset(MultibyteStreamWriterObject *self)
1627{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001628 const Py_UNICODE *pending;
1629 PyObject *pwrt;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001630
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001631 pending = self->pending;
1632 pwrt = multibytecodec_encode(self->codec, &self->state,
1633 &pending, self->pendingsize, self->errors,
1634 MBENC_FLUSH | MBENC_RESET);
1635 /* some pending buffer can be truncated when UnicodeEncodeError is
1636 * raised on 'strict' mode. but, 'reset' method is designed to
1637 * reset the pending buffer or states so failed string sequence
1638 * ought to be missed */
1639 self->pendingsize = 0;
1640 if (pwrt == NULL)
1641 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001642
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001643 assert(PyBytes_Check(pwrt));
1644 if (PyBytes_Size(pwrt) > 0) {
1645 PyObject *wr;
1646 wr = PyObject_CallMethod(self->stream, "write", "O", pwrt);
1647 if (wr == NULL) {
1648 Py_DECREF(pwrt);
1649 return NULL;
1650 }
1651 }
1652 Py_DECREF(pwrt);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001653
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001654 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001655}
1656
1657static PyObject *
1658mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1659{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001660 MultibyteStreamWriterObject *self;
1661 PyObject *stream, *codec = NULL;
1662 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001663
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001664 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
1665 streamkwarglist, &stream, &errors))
1666 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001667
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001668 self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
1669 if (self == NULL)
1670 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001671
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001672 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1673 if (codec == NULL)
1674 goto errorexit;
1675 if (!MultibyteCodec_Check(codec)) {
1676 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1677 goto errorexit;
1678 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001679
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001680 self->codec = ((MultibyteCodecObject *)codec)->codec;
1681 self->stream = stream;
1682 Py_INCREF(stream);
1683 self->pendingsize = 0;
1684 self->errors = internal_error_callback(errors);
1685 if (self->errors == NULL)
1686 goto errorexit;
1687 if (self->codec->encinit != NULL &&
1688 self->codec->encinit(&self->state, self->codec->config) != 0)
1689 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001690
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001691 Py_DECREF(codec);
1692 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001693
1694errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001695 Py_XDECREF(self);
1696 Py_XDECREF(codec);
1697 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001698}
1699
1700static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001701mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
1702{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001703 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001704}
1705
1706static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001707mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001708 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001709{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001710 if (ERROR_ISCUSTOM(self->errors))
1711 Py_VISIT(self->errors);
1712 Py_VISIT(self->stream);
1713 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001714}
1715
1716static void
1717mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
1718{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001719 PyObject_GC_UnTrack(self);
1720 ERROR_DECREF(self->errors);
1721 Py_XDECREF(self->stream);
1722 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001723}
1724
1725static struct PyMethodDef mbstreamwriter_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001726 {"write", (PyCFunction)mbstreamwriter_write,
1727 METH_O, NULL},
1728 {"writelines", (PyCFunction)mbstreamwriter_writelines,
1729 METH_O, NULL},
1730 {"reset", (PyCFunction)mbstreamwriter_reset,
1731 METH_NOARGS, NULL},
1732 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001733};
1734
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001735static PyMemberDef mbstreamwriter_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001736 {"stream", T_OBJECT,
1737 offsetof(MultibyteStreamWriterObject, stream),
1738 READONLY, NULL},
1739 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001740};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001741
1742static PyTypeObject MultibyteStreamWriter_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001743 PyVarObject_HEAD_INIT(NULL, 0)
1744 "MultibyteStreamWriter", /* tp_name */
1745 sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
1746 0, /* tp_itemsize */
1747 /* methods */
1748 (destructor)mbstreamwriter_dealloc, /* tp_dealloc */
1749 0, /* tp_print */
1750 0, /* tp_getattr */
1751 0, /* tp_setattr */
1752 0, /* tp_reserved */
1753 0, /* tp_repr */
1754 0, /* tp_as_number */
1755 0, /* tp_as_sequence */
1756 0, /* tp_as_mapping */
1757 0, /* tp_hash */
1758 0, /* tp_call */
1759 0, /* tp_str */
1760 PyObject_GenericGetAttr, /* tp_getattro */
1761 0, /* tp_setattro */
1762 0, /* tp_as_buffer */
1763 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1764 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1765 0, /* tp_doc */
1766 (traverseproc)mbstreamwriter_traverse, /* tp_traverse */
1767 0, /* tp_clear */
1768 0, /* tp_richcompare */
1769 0, /* tp_weaklistoffset */
1770 0, /* tp_iter */
1771 0, /* tp_iterext */
1772 mbstreamwriter_methods, /* tp_methods */
1773 mbstreamwriter_members, /* tp_members */
1774 codecctx_getsets, /* tp_getset */
1775 0, /* tp_base */
1776 0, /* tp_dict */
1777 0, /* tp_descr_get */
1778 0, /* tp_descr_set */
1779 0, /* tp_dictoffset */
1780 mbstreamwriter_init, /* tp_init */
1781 0, /* tp_alloc */
1782 mbstreamwriter_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001783};
1784
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001785
1786/**
1787 * Exposed factory function
1788 */
1789
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001790static PyObject *
1791__create_codec(PyObject *ignore, PyObject *arg)
1792{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001793 MultibyteCodecObject *self;
1794 MultibyteCodec *codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001795
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001796 if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
1797 PyErr_SetString(PyExc_ValueError, "argument type invalid");
1798 return NULL;
1799 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001800
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001801 codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
1802 if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
1803 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001804
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001805 self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
1806 if (self == NULL)
1807 return NULL;
1808 self->codec = codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001809
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001810 return (PyObject *)self;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001811}
1812
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001813static struct PyMethodDef __methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001814 {"__create_codec", (PyCFunction)__create_codec, METH_O},
1815 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001816};
1817
Martin v. Löwis1a214512008-06-11 05:26:20 +00001818
1819static struct PyModuleDef _multibytecodecmodule = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001820 PyModuleDef_HEAD_INIT,
1821 "_multibytecodec",
1822 NULL,
1823 -1,
1824 __methods,
1825 NULL,
1826 NULL,
1827 NULL,
1828 NULL
Martin v. Löwis1a214512008-06-11 05:26:20 +00001829};
1830
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001831PyMODINIT_FUNC
Martin v. Löwis1a214512008-06-11 05:26:20 +00001832PyInit__multibytecodec(void)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001833{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001834 int i;
1835 PyObject *m;
1836 PyTypeObject *typelist[] = {
1837 &MultibyteIncrementalEncoder_Type,
1838 &MultibyteIncrementalDecoder_Type,
1839 &MultibyteStreamReader_Type,
1840 &MultibyteStreamWriter_Type,
1841 NULL
1842 };
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001843
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001844 if (PyType_Ready(&MultibyteCodec_Type) < 0)
1845 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001846
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001847 m = PyModule_Create(&_multibytecodecmodule);
1848 if (m == NULL)
1849 return NULL;
Neal Norwitz058bde12005-09-21 06:44:25 +00001850
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001851 for (i = 0; typelist[i] != NULL; i++) {
1852 if (PyType_Ready(typelist[i]) < 0)
1853 return NULL;
1854 Py_INCREF(typelist[i]);
1855 PyModule_AddObject(m, typelist[i]->tp_name,
1856 (PyObject *)typelist[i]);
1857 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001858
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001859 if (PyErr_Occurred()) {
1860 Py_FatalError("can't initialize the _multibytecodec module");
1861 Py_DECREF(m);
1862 m = NULL;
1863 }
1864 return m;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001865}