blob: 8901b42e41e8745ca22511f309d2ad98403f1461 [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"
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +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\
Andrew M. Kuchling658c45f2008-10-03 12:26:42 +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};
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +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 Pitrouc83ea132010-05-09 14:46:46 +000048 MultibyteCodec_State *, const Py_UNICODE **, Py_ssize_t,
49 PyObject *, int);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000050
Antoine Pitrouc83ea132010-05-09 14:46:46 +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 Pitrouc83ea132010-05-09 14:46:46 +000056 PyObject *v, *w;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000057
Antoine Pitrouc83ea132010-05-09 14:46:46 +000058 if (object == NULL)
59 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000060
Antoine Pitrouc83ea132010-05-09 14:46:46 +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 Pitrouc83ea132010-05-09 14:46:46 +000068 w = PyInt_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 Pitrouc83ea132010-05-09 14:46:46 +000075 return v;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000076}
77
78static PyObject *
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +000079internal_error_callback(const char *errors)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000080{
Antoine Pitrouc83ea132010-05-09 14:46:46 +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 PyString_FromString(errors);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000089}
90
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +000091static PyObject *
92call_error_callback(PyObject *errors, PyObject *exc)
93{
Antoine Pitrouc83ea132010-05-09 14:46:46 +000094 PyObject *args, *cb, *r;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +000095
Antoine Pitrouc83ea132010-05-09 14:46:46 +000096 assert(PyString_Check(errors));
97 cb = PyCodec_LookupError(PyString_AS_STRING(errors));
98 if (cb == NULL)
99 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000100
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000101 args = PyTuple_New(1);
102 if (args == NULL) {
103 Py_DECREF(cb);
104 return NULL;
105 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000106
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000107 PyTuple_SET_ITEM(args, 0, exc);
108 Py_INCREF(exc);
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000109
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000110 r = PyObject_CallObject(cb, args);
111 Py_DECREF(args);
112 Py_DECREF(cb);
113 return r;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000114}
115
116static PyObject *
117codecctx_errors_get(MultibyteStatefulCodecContext *self)
118{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000119 const char *errors;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000120
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000121 if (self->errors == ERROR_STRICT)
122 errors = "strict";
123 else if (self->errors == ERROR_IGNORE)
124 errors = "ignore";
125 else if (self->errors == ERROR_REPLACE)
126 errors = "replace";
127 else {
128 Py_INCREF(self->errors);
129 return self->errors;
130 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000131
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000132 return PyString_FromString(errors);
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000133}
134
135static int
136codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000137 void *closure)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000138{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000139 PyObject *cb;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000140
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000141 if (!PyString_Check(value)) {
142 PyErr_SetString(PyExc_TypeError, "errors must be a string");
143 return -1;
144 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000145
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000146 cb = internal_error_callback(PyString_AS_STRING(value));
147 if (cb == NULL)
148 return -1;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000149
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000150 ERROR_DECREF(self->errors);
151 self->errors = cb;
152 return 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000153}
154
155/* This getset handlers list is used by all the stateful codec objects */
156static PyGetSetDef codecctx_getsets[] = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000157 {"errors", (getter)codecctx_errors_get,
158 (setter)codecctx_errors_set,
159 PyDoc_STR("how to treat errors")},
160 {NULL,}
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000161};
162
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000163static int
Hye-Shik Chang4b96c132006-03-04 16:08:19 +0000164expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000165{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000166 Py_ssize_t orgpos, orgsize, incsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000167
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000168 orgpos = (Py_ssize_t)((char *)buf->outbuf -
169 PyString_AS_STRING(buf->outobj));
170 orgsize = PyString_GET_SIZE(buf->outobj);
171 incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize);
Gregory P. Smith9d534572008-06-11 07:41:16 +0000172
Serhiy Storchaka021d55f2015-02-21 01:19:17 +0200173 if (orgsize > PY_SSIZE_T_MAX - incsize) {
174 PyErr_NoMemory();
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000175 return -1;
Serhiy Storchaka021d55f2015-02-21 01:19:17 +0200176 }
Gregory P. Smith9d534572008-06-11 07:41:16 +0000177
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000178 if (_PyString_Resize(&buf->outobj, orgsize + incsize) == -1)
179 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000180
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000181 buf->outbuf = (unsigned char *)PyString_AS_STRING(buf->outobj) +orgpos;
182 buf->outbuf_end = (unsigned char *)PyString_AS_STRING(buf->outobj)
183 + PyString_GET_SIZE(buf->outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000184
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000185 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000186}
Serhiy Storchaka021d55f2015-02-21 01:19:17 +0200187#define REQUIRE_ENCODEBUFFER(buf, s) do { \
188 if ((s) < 0 || (s) > (buf)->outbuf_end - (buf)->outbuf) \
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000189 if (expand_encodebuffer(buf, s) == -1) \
190 goto errorexit; \
Serhiy Storchaka021d55f2015-02-21 01:19:17 +0200191} while(0)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000192
193static int
Hye-Shik Chang4b96c132006-03-04 16:08:19 +0000194expand_decodebuffer(MultibyteDecodeBuffer *buf, Py_ssize_t esize)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000195{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000196 Py_ssize_t orgpos, orgsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000197
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000198 orgpos = (Py_ssize_t)(buf->outbuf - PyUnicode_AS_UNICODE(buf->outobj));
199 orgsize = PyUnicode_GET_SIZE(buf->outobj);
200 if (PyUnicode_Resize(&buf->outobj, orgsize + (
201 esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize)) == -1)
202 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000203
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000204 buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj) + orgpos;
205 buf->outbuf_end = PyUnicode_AS_UNICODE(buf->outobj)
206 + PyUnicode_GET_SIZE(buf->outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000207
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000208 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000209}
Serhiy Storchaka021d55f2015-02-21 01:19:17 +0200210#define REQUIRE_DECODEBUFFER(buf, s) do { \
211 if ((s) < 0 || (s) > (buf)->outbuf_end - (buf)->outbuf) \
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000212 if (expand_decodebuffer(buf, s) == -1) \
213 goto errorexit; \
Serhiy Storchaka021d55f2015-02-21 01:19:17 +0200214} while(0)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000215
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000216
217/**
218 * MultibyteCodec object
219 */
220
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000221static int
222multibytecodec_encerror(MultibyteCodec *codec,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000223 MultibyteCodec_State *state,
224 MultibyteEncodeBuffer *buf,
225 PyObject *errors, Py_ssize_t e)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000226{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000227 PyObject *retobj = NULL, *retstr = NULL, *tobj;
228 Py_ssize_t retstrsize, newpos;
229 Py_ssize_t esize, start, end;
230 const char *reason;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000231
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000232 if (e > 0) {
233 reason = "illegal multibyte sequence";
234 esize = e;
235 }
236 else {
237 switch (e) {
238 case MBERR_TOOSMALL:
239 REQUIRE_ENCODEBUFFER(buf, -1);
240 return 0; /* retry it */
241 case MBERR_TOOFEW:
242 reason = "incomplete multibyte sequence";
243 esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
244 break;
245 case MBERR_INTERNAL:
246 PyErr_SetString(PyExc_RuntimeError,
247 "internal codec error");
248 return -1;
249 default:
250 PyErr_SetString(PyExc_RuntimeError,
251 "unknown runtime error");
252 return -1;
253 }
254 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000255
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000256 if (errors == ERROR_REPLACE) {
257 const Py_UNICODE replchar = '?', *inbuf = &replchar;
258 Py_ssize_t r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000259
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000260 for (;;) {
261 Py_ssize_t outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000262
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000263 outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
264 r = codec->encode(state, codec->config, &inbuf, 1,
265 &buf->outbuf, outleft, 0);
266 if (r == MBERR_TOOSMALL) {
267 REQUIRE_ENCODEBUFFER(buf, -1);
268 continue;
269 }
270 else
271 break;
272 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000273
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000274 if (r != 0) {
275 REQUIRE_ENCODEBUFFER(buf, 1);
276 *buf->outbuf++ = '?';
277 }
278 }
279 if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
280 buf->inbuf += esize;
281 return 0;
282 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000283
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000284 start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
285 end = start + esize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000286
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000287 /* use cached exception object if available */
288 if (buf->excobj == NULL) {
289 buf->excobj = PyUnicodeEncodeError_Create(codec->encoding,
290 buf->inbuf_top,
291 buf->inbuf_end - buf->inbuf_top,
292 start, end, reason);
293 if (buf->excobj == NULL)
294 goto errorexit;
295 }
296 else
297 if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 ||
298 PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 ||
299 PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0)
300 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000301
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000302 if (errors == ERROR_STRICT) {
303 PyCodec_StrictErrors(buf->excobj);
304 goto errorexit;
305 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000306
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000307 retobj = call_error_callback(errors, buf->excobj);
308 if (retobj == NULL)
309 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000310
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000311 if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
312 !PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) ||
313 !(PyInt_Check(PyTuple_GET_ITEM(retobj, 1)) ||
314 PyLong_Check(PyTuple_GET_ITEM(retobj, 1)))) {
315 PyErr_SetString(PyExc_TypeError,
316 "encoding error handler must return "
317 "(unicode, int) tuple");
318 goto errorexit;
319 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000320
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000321 {
322 const Py_UNICODE *uraw = PyUnicode_AS_UNICODE(tobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000323
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000324 retstr = multibytecodec_encode(codec, state, &uraw,
325 PyUnicode_GET_SIZE(tobj), ERROR_STRICT,
326 MBENC_FLUSH);
327 if (retstr == NULL)
328 goto errorexit;
329 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000330
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000331 retstrsize = PyString_GET_SIZE(retstr);
Serhiy Storchaka021d55f2015-02-21 01:19:17 +0200332 if (retstrsize > 0) {
333 REQUIRE_ENCODEBUFFER(buf, retstrsize);
334 memcpy(buf->outbuf, PyString_AS_STRING(retstr), retstrsize);
335 buf->outbuf += retstrsize;
336 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000337
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000338 newpos = PyInt_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
339 if (newpos < 0 && !PyErr_Occurred())
340 newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
341 if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
342 PyErr_Clear();
343 PyErr_Format(PyExc_IndexError,
344 "position %zd from error handler out of bounds",
345 newpos);
346 goto errorexit;
347 }
348 buf->inbuf = buf->inbuf_top + newpos;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000349
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000350 Py_DECREF(retobj);
351 Py_DECREF(retstr);
352 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000353
354errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000355 Py_XDECREF(retobj);
356 Py_XDECREF(retstr);
357 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000358}
359
360static int
361multibytecodec_decerror(MultibyteCodec *codec,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000362 MultibyteCodec_State *state,
363 MultibyteDecodeBuffer *buf,
364 PyObject *errors, Py_ssize_t e)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000365{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000366 PyObject *retobj = NULL, *retuni = NULL;
367 Py_ssize_t retunisize, newpos;
368 const char *reason;
369 Py_ssize_t esize, start, end;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000370
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000371 if (e > 0) {
372 reason = "illegal multibyte sequence";
373 esize = e;
374 }
375 else {
376 switch (e) {
377 case MBERR_TOOSMALL:
378 REQUIRE_DECODEBUFFER(buf, -1);
379 return 0; /* retry it */
380 case MBERR_TOOFEW:
381 reason = "incomplete multibyte sequence";
382 esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
383 break;
384 case MBERR_INTERNAL:
385 PyErr_SetString(PyExc_RuntimeError,
386 "internal codec error");
387 return -1;
388 default:
389 PyErr_SetString(PyExc_RuntimeError,
390 "unknown runtime error");
391 return -1;
392 }
393 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000394
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000395 if (errors == ERROR_REPLACE) {
396 REQUIRE_DECODEBUFFER(buf, 1);
397 *buf->outbuf++ = Py_UNICODE_REPLACEMENT_CHARACTER;
398 }
399 if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
400 buf->inbuf += esize;
401 return 0;
402 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000403
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000404 start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
405 end = start + esize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000406
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000407 /* use cached exception object if available */
408 if (buf->excobj == NULL) {
409 buf->excobj = PyUnicodeDecodeError_Create(codec->encoding,
410 (const char *)buf->inbuf_top,
411 (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top),
412 start, end, reason);
413 if (buf->excobj == NULL)
414 goto errorexit;
415 }
416 else
417 if (PyUnicodeDecodeError_SetStart(buf->excobj, start) ||
418 PyUnicodeDecodeError_SetEnd(buf->excobj, end) ||
419 PyUnicodeDecodeError_SetReason(buf->excobj, reason))
420 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000421
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000422 if (errors == ERROR_STRICT) {
423 PyCodec_StrictErrors(buf->excobj);
424 goto errorexit;
425 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000426
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000427 retobj = call_error_callback(errors, buf->excobj);
428 if (retobj == NULL)
429 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000430
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000431 if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
432 !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) ||
433 !(PyInt_Check(PyTuple_GET_ITEM(retobj, 1)) ||
434 PyLong_Check(PyTuple_GET_ITEM(retobj, 1)))) {
435 PyErr_SetString(PyExc_TypeError,
436 "decoding error handler must return "
437 "(unicode, int) tuple");
438 goto errorexit;
439 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000440
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000441 retunisize = PyUnicode_GET_SIZE(retuni);
442 if (retunisize > 0) {
443 REQUIRE_DECODEBUFFER(buf, retunisize);
444 memcpy((char *)buf->outbuf, PyUnicode_AS_DATA(retuni),
445 retunisize * Py_UNICODE_SIZE);
446 buf->outbuf += retunisize;
447 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000448
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000449 newpos = PyInt_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
450 if (newpos < 0 && !PyErr_Occurred())
451 newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
452 if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
453 PyErr_Clear();
454 PyErr_Format(PyExc_IndexError,
455 "position %zd from error handler out of bounds",
456 newpos);
457 goto errorexit;
458 }
459 buf->inbuf = buf->inbuf_top + newpos;
460 Py_DECREF(retobj);
461 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000462
463errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000464 Py_XDECREF(retobj);
465 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000466}
467
468static PyObject *
469multibytecodec_encode(MultibyteCodec *codec,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000470 MultibyteCodec_State *state,
471 const Py_UNICODE **data, Py_ssize_t datalen,
472 PyObject *errors, int flags)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000473{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000474 MultibyteEncodeBuffer buf;
475 Py_ssize_t finalsize, r = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000476
Victor Stinner40b41e12011-05-24 22:29:13 +0200477 if (datalen == 0 && !(flags & MBENC_RESET))
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000478 return PyString_FromString("");
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000479
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000480 buf.excobj = NULL;
Serhiy Storchaka81c670d2015-07-24 07:44:42 +0300481 buf.outobj = NULL;
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000482 buf.inbuf = buf.inbuf_top = *data;
483 buf.inbuf_end = buf.inbuf_top + datalen;
Gregory P. Smith9d534572008-06-11 07:41:16 +0000484
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000485 if (datalen > (PY_SSIZE_T_MAX - 16) / 2) {
486 PyErr_NoMemory();
487 goto errorexit;
488 }
Gregory P. Smith9d534572008-06-11 07:41:16 +0000489
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000490 buf.outobj = PyString_FromStringAndSize(NULL, datalen * 2 + 16);
491 if (buf.outobj == NULL)
492 goto errorexit;
493 buf.outbuf = (unsigned char *)PyString_AS_STRING(buf.outobj);
494 buf.outbuf_end = buf.outbuf + PyString_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000495
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000496 while (buf.inbuf < buf.inbuf_end) {
497 Py_ssize_t inleft, outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000498
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000499 /* we don't reuse inleft and outleft here.
500 * error callbacks can relocate the cursor anywhere on buffer*/
501 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
502 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
503 r = codec->encode(state, codec->config, &buf.inbuf, inleft,
504 &buf.outbuf, outleft, flags);
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000505 if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH)))
506 break;
507 else if (multibytecodec_encerror(codec, state, &buf, errors,r))
508 goto errorexit;
509 else if (r == MBERR_TOOFEW)
510 break;
511 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000512
Victor Stinner40b41e12011-05-24 22:29:13 +0200513 if (codec->encreset != NULL && (flags & MBENC_RESET))
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000514 for (;;) {
515 Py_ssize_t outleft;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000516
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000517 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
518 r = codec->encreset(state, codec->config, &buf.outbuf,
519 outleft);
520 if (r == 0)
521 break;
522 else if (multibytecodec_encerror(codec, state,
523 &buf, errors, r))
524 goto errorexit;
525 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000526
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000527 finalsize = (Py_ssize_t)((char *)buf.outbuf -
528 PyString_AS_STRING(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000529
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000530 if (finalsize != PyString_GET_SIZE(buf.outobj))
531 if (_PyString_Resize(&buf.outobj, finalsize) == -1)
532 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000533
Benjamin Petersone4668aa2016-09-13 21:48:57 -0700534 *data = buf.inbuf;
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000535 Py_XDECREF(buf.excobj);
536 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000537
538errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000539 Py_XDECREF(buf.excobj);
540 Py_XDECREF(buf.outobj);
541 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000542}
543
544static PyObject *
545MultibyteCodec_Encode(MultibyteCodecObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000546 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000547{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000548 MultibyteCodec_State state;
549 Py_UNICODE *data;
550 PyObject *errorcb, *r, *arg, *ucvt;
551 const char *errors = NULL;
552 Py_ssize_t datalen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000553
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000554 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode",
555 codeckwarglist, &arg, &errors))
556 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000557
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000558 if (PyUnicode_Check(arg))
559 ucvt = NULL;
560 else {
561 arg = ucvt = PyObject_Unicode(arg);
562 if (arg == NULL)
563 return NULL;
564 else if (!PyUnicode_Check(arg)) {
565 PyErr_SetString(PyExc_TypeError,
566 "couldn't convert the object to unicode.");
567 Py_DECREF(ucvt);
568 return NULL;
569 }
570 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000571
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000572 data = PyUnicode_AS_UNICODE(arg);
573 datalen = PyUnicode_GET_SIZE(arg);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000574
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000575 errorcb = internal_error_callback(errors);
576 if (errorcb == NULL) {
577 Py_XDECREF(ucvt);
578 return NULL;
579 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000580
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000581 if (self->codec->encinit != NULL &&
582 self->codec->encinit(&state, self->codec->config) != 0)
583 goto errorexit;
584 r = multibytecodec_encode(self->codec, &state,
585 (const Py_UNICODE **)&data, datalen, errorcb,
586 MBENC_FLUSH | MBENC_RESET);
587 if (r == NULL)
588 goto errorexit;
Hye-Shik Chang2bb146f2004-07-18 03:06:29 +0000589
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000590 ERROR_DECREF(errorcb);
591 Py_XDECREF(ucvt);
592 return make_tuple(r, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000593
594errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000595 ERROR_DECREF(errorcb);
596 Py_XDECREF(ucvt);
597 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000598}
599
600static PyObject *
601MultibyteCodec_Decode(MultibyteCodecObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000602 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000603{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000604 MultibyteCodec_State state;
605 MultibyteDecodeBuffer buf;
606 PyObject *errorcb;
607 Py_buffer pdata;
608 const char *data, *errors = NULL;
609 Py_ssize_t datalen, finalsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000610
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000611 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "s*|z:decode",
612 codeckwarglist, &pdata, &errors))
613 return NULL;
614 data = pdata.buf;
615 datalen = pdata.len;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000616
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000617 errorcb = internal_error_callback(errors);
618 if (errorcb == NULL) {
619 PyBuffer_Release(&pdata);
620 return NULL;
621 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000622
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000623 if (datalen == 0) {
624 PyBuffer_Release(&pdata);
625 ERROR_DECREF(errorcb);
626 return make_tuple(PyUnicode_FromUnicode(NULL, 0), 0);
627 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000628
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000629 buf.excobj = NULL;
630 buf.inbuf = buf.inbuf_top = (unsigned char *)data;
631 buf.inbuf_end = buf.inbuf_top + datalen;
632 buf.outobj = PyUnicode_FromUnicode(NULL, datalen);
633 if (buf.outobj == NULL)
634 goto errorexit;
635 buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj);
636 buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000637
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000638 if (self->codec->decinit != NULL &&
639 self->codec->decinit(&state, self->codec->config) != 0)
640 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000641
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000642 while (buf.inbuf < buf.inbuf_end) {
643 Py_ssize_t inleft, outleft, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000644
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000645 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
646 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000647
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000648 r = self->codec->decode(&state, self->codec->config,
649 &buf.inbuf, inleft, &buf.outbuf, outleft);
650 if (r == 0)
651 break;
652 else if (multibytecodec_decerror(self->codec, &state,
653 &buf, errorcb, r))
654 goto errorexit;
655 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000656
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000657 finalsize = (Py_ssize_t)(buf.outbuf -
658 PyUnicode_AS_UNICODE(buf.outobj));
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000659
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000660 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
661 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
662 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000663
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000664 PyBuffer_Release(&pdata);
665 Py_XDECREF(buf.excobj);
666 ERROR_DECREF(errorcb);
667 return make_tuple(buf.outobj, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000668
669errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000670 PyBuffer_Release(&pdata);
671 ERROR_DECREF(errorcb);
672 Py_XDECREF(buf.excobj);
673 Py_XDECREF(buf.outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000674
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000675 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000676}
677
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000678static struct PyMethodDef multibytecodec_methods[] = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000679 {"encode", (PyCFunction)MultibyteCodec_Encode,
680 METH_VARARGS | METH_KEYWORDS,
681 MultibyteCodec_Encode__doc__},
682 {"decode", (PyCFunction)MultibyteCodec_Decode,
683 METH_VARARGS | METH_KEYWORDS,
684 MultibyteCodec_Decode__doc__},
685 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000686};
687
688static void
689multibytecodec_dealloc(MultibyteCodecObject *self)
690{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000691 PyObject_Del(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000692}
693
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000694static PyTypeObject MultibyteCodec_Type = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000695 PyVarObject_HEAD_INIT(NULL, 0)
696 "MultibyteCodec", /* tp_name */
697 sizeof(MultibyteCodecObject), /* tp_basicsize */
698 0, /* tp_itemsize */
699 /* methods */
700 (destructor)multibytecodec_dealloc, /* tp_dealloc */
701 0, /* tp_print */
702 0, /* tp_getattr */
703 0, /* tp_setattr */
704 0, /* tp_compare */
705 0, /* tp_repr */
706 0, /* tp_as_number */
707 0, /* tp_as_sequence */
708 0, /* tp_as_mapping */
709 0, /* tp_hash */
710 0, /* tp_call */
711 0, /* tp_str */
712 PyObject_GenericGetAttr, /* tp_getattro */
713 0, /* tp_setattro */
714 0, /* tp_as_buffer */
715 Py_TPFLAGS_DEFAULT, /* tp_flags */
716 0, /* tp_doc */
717 0, /* tp_traverse */
718 0, /* tp_clear */
719 0, /* tp_richcompare */
720 0, /* tp_weaklistoffset */
721 0, /* tp_iter */
722 0, /* tp_iterext */
723 multibytecodec_methods, /* tp_methods */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000724};
725
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000726
727/**
728 * Utility functions for stateful codec mechanism
729 */
730
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000731#define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o))
732#define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o))
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000733
734static PyObject *
735encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000736 PyObject *unistr, int final)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000737{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000738 PyObject *ucvt, *r = NULL;
739 Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL;
740 Py_ssize_t datalen, origpending;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000741
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000742 if (PyUnicode_Check(unistr))
743 ucvt = NULL;
744 else {
745 unistr = ucvt = PyObject_Unicode(unistr);
746 if (unistr == NULL)
747 return NULL;
748 else if (!PyUnicode_Check(unistr)) {
749 PyErr_SetString(PyExc_TypeError,
750 "couldn't convert the object to unicode.");
751 Py_DECREF(ucvt);
752 return NULL;
753 }
754 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000755
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000756 datalen = PyUnicode_GET_SIZE(unistr);
757 origpending = ctx->pendingsize;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000758
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000759 if (origpending > 0) {
760 if (datalen > PY_SSIZE_T_MAX - ctx->pendingsize) {
761 PyErr_NoMemory();
762 /* inbuf_tmp == NULL */
763 goto errorexit;
764 }
765 inbuf_tmp = PyMem_New(Py_UNICODE, datalen + ctx->pendingsize);
766 if (inbuf_tmp == NULL)
767 goto errorexit;
768 memcpy(inbuf_tmp, ctx->pending,
769 Py_UNICODE_SIZE * ctx->pendingsize);
770 memcpy(inbuf_tmp + ctx->pendingsize,
771 PyUnicode_AS_UNICODE(unistr),
772 Py_UNICODE_SIZE * datalen);
773 datalen += ctx->pendingsize;
774 ctx->pendingsize = 0;
775 inbuf = inbuf_tmp;
776 }
777 else
778 inbuf = (Py_UNICODE *)PyUnicode_AS_UNICODE(unistr);
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000779
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000780 inbuf_end = inbuf + datalen;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000781
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000782 r = multibytecodec_encode(ctx->codec, &ctx->state,
Victor Stinner40b41e12011-05-24 22:29:13 +0200783 (const Py_UNICODE **)&inbuf, datalen,
784 ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0);
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000785 if (r == NULL) {
786 /* recover the original pending buffer */
787 if (origpending > 0)
788 memcpy(ctx->pending, inbuf_tmp,
789 Py_UNICODE_SIZE * origpending);
790 ctx->pendingsize = origpending;
791 goto errorexit;
792 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000793
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000794 if (inbuf < inbuf_end) {
795 ctx->pendingsize = (Py_ssize_t)(inbuf_end - inbuf);
796 if (ctx->pendingsize > MAXENCPENDING) {
797 /* normal codecs can't reach here */
798 ctx->pendingsize = 0;
799 PyErr_SetString(PyExc_UnicodeError,
800 "pending buffer overflow");
801 goto errorexit;
802 }
803 memcpy(ctx->pending, inbuf,
804 ctx->pendingsize * Py_UNICODE_SIZE);
805 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000806
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000807 if (inbuf_tmp != NULL)
808 PyMem_Del(inbuf_tmp);
809 Py_XDECREF(ucvt);
810 return r;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000811
812errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000813 if (inbuf_tmp != NULL)
814 PyMem_Del(inbuf_tmp);
815 Py_XDECREF(r);
816 Py_XDECREF(ucvt);
817 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000818}
819
820static int
821decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000822 MultibyteDecodeBuffer *buf)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000823{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000824 Py_ssize_t npendings;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000825
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000826 npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
827 if (npendings + ctx->pendingsize > MAXDECPENDING ||
828 npendings > PY_SSIZE_T_MAX - ctx->pendingsize) {
829 PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow");
830 return -1;
831 }
832 memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings);
833 ctx->pendingsize += npendings;
834 return 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000835}
836
837static int
838decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000839 Py_ssize_t size)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000840{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000841 buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
842 buf->inbuf_end = buf->inbuf_top + size;
843 if (buf->outobj == NULL) { /* only if outobj is not allocated yet */
844 buf->outobj = PyUnicode_FromUnicode(NULL, size);
845 if (buf->outobj == NULL)
846 return -1;
847 buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj);
848 buf->outbuf_end = buf->outbuf +
849 PyUnicode_GET_SIZE(buf->outobj);
850 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000851
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000852 return 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000853}
854
855static int
856decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000857 MultibyteDecodeBuffer *buf)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000858{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000859 while (buf->inbuf < buf->inbuf_end) {
860 Py_ssize_t inleft, outleft;
861 Py_ssize_t r;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000862
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000863 inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
864 outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000865
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000866 r = ctx->codec->decode(&ctx->state, ctx->codec->config,
867 &buf->inbuf, inleft, &buf->outbuf, outleft);
868 if (r == 0 || r == MBERR_TOOFEW)
869 break;
870 else if (multibytecodec_decerror(ctx->codec, &ctx->state,
871 buf, ctx->errors, r))
872 return -1;
873 }
874 return 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000875}
876
877
878/**
879 * MultibyteIncrementalEncoder object
880 */
881
882static PyObject *
883mbiencoder_encode(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000884 PyObject *args, PyObject *kwargs)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000885{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000886 PyObject *data;
887 int final = 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000888
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000889 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode",
890 incrementalkwarglist, &data, &final))
891 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000892
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000893 return encoder_encode_stateful(STATEFUL_ECTX(self), data, final);
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000894}
895
896static PyObject *
897mbiencoder_reset(MultibyteIncrementalEncoderObject *self)
898{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000899 if (self->codec->decreset != NULL &&
900 self->codec->decreset(&self->state, self->codec->config) != 0)
901 return NULL;
902 self->pendingsize = 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000903
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000904 Py_RETURN_NONE;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000905}
906
907static struct PyMethodDef mbiencoder_methods[] = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000908 {"encode", (PyCFunction)mbiencoder_encode,
909 METH_VARARGS | METH_KEYWORDS, NULL},
910 {"reset", (PyCFunction)mbiencoder_reset,
911 METH_NOARGS, NULL},
912 {NULL, NULL},
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000913};
914
915static PyObject *
916mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
917{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000918 MultibyteIncrementalEncoderObject *self;
919 PyObject *codec = NULL;
920 char *errors = NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000921
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000922 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
923 incnewkwarglist, &errors))
924 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000925
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000926 self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
927 if (self == NULL)
928 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000929
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000930 codec = PyObject_GetAttrString((PyObject *)type, "codec");
931 if (codec == NULL)
932 goto errorexit;
933 if (!MultibyteCodec_Check(codec)) {
934 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
935 goto errorexit;
936 }
Hye-Shik Changb9c03e92006-03-27 08:24:54 +0000937
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000938 self->codec = ((MultibyteCodecObject *)codec)->codec;
939 self->pendingsize = 0;
940 self->errors = internal_error_callback(errors);
941 if (self->errors == NULL)
942 goto errorexit;
943 if (self->codec->encinit != NULL &&
944 self->codec->encinit(&self->state, self->codec->config) != 0)
945 goto errorexit;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000946
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000947 Py_DECREF(codec);
948 return (PyObject *)self;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000949
950errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000951 Py_XDECREF(self);
952 Py_XDECREF(codec);
953 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000954}
955
956static int
Hye-Shik Changdb8955b2006-04-21 16:28:34 +0000957mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
958{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000959 return 0;
Hye-Shik Changdb8955b2006-04-21 16:28:34 +0000960}
961
962static int
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000963mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000964 visitproc visit, void *arg)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000965{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000966 if (ERROR_ISCUSTOM(self->errors))
967 Py_VISIT(self->errors);
968 return 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000969}
970
971static void
972mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
973{
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000974 PyObject_GC_UnTrack(self);
975 ERROR_DECREF(self->errors);
976 Py_TYPE(self)->tp_free(self);
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +0000977}
978
979static PyTypeObject MultibyteIncrementalEncoder_Type = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +0000980 PyVarObject_HEAD_INIT(NULL, 0)
981 "MultibyteIncrementalEncoder", /* tp_name */
982 sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
983 0, /* tp_itemsize */
984 /* methods */
985 (destructor)mbiencoder_dealloc, /* tp_dealloc */
986 0, /* tp_print */
987 0, /* tp_getattr */
988 0, /* tp_setattr */
989 0, /* tp_compare */
990 0, /* tp_repr */
991 0, /* tp_as_number */
992 0, /* tp_as_sequence */
993 0, /* tp_as_mapping */
994 0, /* tp_hash */
995 0, /* tp_call */
996 0, /* tp_str */
997 PyObject_GenericGetAttr, /* tp_getattro */
998 0, /* tp_setattro */
999 0, /* tp_as_buffer */
1000 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1001 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1002 0, /* tp_doc */
1003 (traverseproc)mbiencoder_traverse, /* tp_traverse */
1004 0, /* tp_clear */
1005 0, /* tp_richcompare */
1006 0, /* tp_weaklistoffset */
1007 0, /* tp_iter */
1008 0, /* tp_iterext */
1009 mbiencoder_methods, /* tp_methods */
1010 0, /* tp_members */
1011 codecctx_getsets, /* tp_getset */
1012 0, /* tp_base */
1013 0, /* tp_dict */
1014 0, /* tp_descr_get */
1015 0, /* tp_descr_set */
1016 0, /* tp_dictoffset */
1017 mbiencoder_init, /* tp_init */
1018 0, /* tp_alloc */
1019 mbiencoder_new, /* tp_new */
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001020};
1021
1022
1023/**
1024 * MultibyteIncrementalDecoder object
1025 */
1026
1027static PyObject *
1028mbidecoder_decode(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001029 PyObject *args, PyObject *kwargs)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001030{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001031 MultibyteDecodeBuffer buf;
1032 char *data, *wdata = NULL;
1033 Py_buffer pdata;
1034 Py_ssize_t wsize, finalsize = 0, size, origpending;
1035 int final = 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001036
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001037 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "s*|i:decode",
1038 incrementalkwarglist, &pdata, &final))
1039 return NULL;
1040 data = pdata.buf;
1041 size = pdata.len;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001042
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001043 buf.outobj = buf.excobj = NULL;
1044 origpending = self->pendingsize;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001045
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001046 if (self->pendingsize == 0) {
1047 wsize = size;
1048 wdata = data;
1049 }
1050 else {
1051 if (size > PY_SSIZE_T_MAX - self->pendingsize) {
1052 PyErr_NoMemory();
1053 goto errorexit;
1054 }
1055 wsize = size + self->pendingsize;
1056 wdata = PyMem_Malloc(wsize);
1057 if (wdata == NULL)
1058 goto errorexit;
1059 memcpy(wdata, self->pending, self->pendingsize);
1060 memcpy(wdata + self->pendingsize, data, size);
1061 self->pendingsize = 0;
1062 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001063
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001064 if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
1065 goto errorexit;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001066
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001067 if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
1068 goto errorexit;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001069
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001070 if (final && buf.inbuf < buf.inbuf_end) {
1071 if (multibytecodec_decerror(self->codec, &self->state,
1072 &buf, self->errors, MBERR_TOOFEW)) {
1073 /* recover the original pending buffer */
1074 memcpy(self->pending, wdata, origpending);
1075 self->pendingsize = origpending;
1076 goto errorexit;
1077 }
1078 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001079
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001080 if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
1081 if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
1082 goto errorexit;
1083 }
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001084
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001085 finalsize = (Py_ssize_t)(buf.outbuf - PyUnicode_AS_UNICODE(buf.outobj));
1086 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1087 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1088 goto errorexit;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001089
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001090 PyBuffer_Release(&pdata);
1091 if (wdata != data)
1092 PyMem_Del(wdata);
1093 Py_XDECREF(buf.excobj);
1094 return buf.outobj;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001095
1096errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001097 PyBuffer_Release(&pdata);
1098 if (wdata != NULL && wdata != data)
1099 PyMem_Del(wdata);
1100 Py_XDECREF(buf.excobj);
1101 Py_XDECREF(buf.outobj);
1102 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001103}
1104
1105static PyObject *
1106mbidecoder_reset(MultibyteIncrementalDecoderObject *self)
1107{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001108 if (self->codec->decreset != NULL &&
1109 self->codec->decreset(&self->state, self->codec->config) != 0)
1110 return NULL;
1111 self->pendingsize = 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001112
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001113 Py_RETURN_NONE;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001114}
1115
1116static struct PyMethodDef mbidecoder_methods[] = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001117 {"decode", (PyCFunction)mbidecoder_decode,
1118 METH_VARARGS | METH_KEYWORDS, NULL},
1119 {"reset", (PyCFunction)mbidecoder_reset,
1120 METH_NOARGS, NULL},
1121 {NULL, NULL},
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001122};
1123
1124static PyObject *
1125mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1126{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001127 MultibyteIncrementalDecoderObject *self;
1128 PyObject *codec = NULL;
1129 char *errors = NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001130
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001131 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
1132 incnewkwarglist, &errors))
1133 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001134
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001135 self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
1136 if (self == NULL)
1137 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001138
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001139 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1140 if (codec == NULL)
1141 goto errorexit;
1142 if (!MultibyteCodec_Check(codec)) {
1143 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1144 goto errorexit;
1145 }
Hye-Shik Changb9c03e92006-03-27 08:24:54 +00001146
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001147 self->codec = ((MultibyteCodecObject *)codec)->codec;
1148 self->pendingsize = 0;
1149 self->errors = internal_error_callback(errors);
1150 if (self->errors == NULL)
1151 goto errorexit;
1152 if (self->codec->decinit != NULL &&
1153 self->codec->decinit(&self->state, self->codec->config) != 0)
1154 goto errorexit;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001155
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001156 Py_DECREF(codec);
1157 return (PyObject *)self;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001158
1159errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001160 Py_XDECREF(self);
1161 Py_XDECREF(codec);
1162 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001163}
1164
1165static int
Hye-Shik Changdb8955b2006-04-21 16:28:34 +00001166mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
1167{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001168 return 0;
Hye-Shik Changdb8955b2006-04-21 16:28:34 +00001169}
1170
1171static int
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001172mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001173 visitproc visit, void *arg)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001174{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001175 if (ERROR_ISCUSTOM(self->errors))
1176 Py_VISIT(self->errors);
1177 return 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001178}
1179
1180static void
1181mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
1182{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001183 PyObject_GC_UnTrack(self);
1184 ERROR_DECREF(self->errors);
1185 Py_TYPE(self)->tp_free(self);
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001186}
1187
1188static PyTypeObject MultibyteIncrementalDecoder_Type = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001189 PyVarObject_HEAD_INIT(NULL, 0)
1190 "MultibyteIncrementalDecoder", /* tp_name */
1191 sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */
1192 0, /* tp_itemsize */
1193 /* methods */
1194 (destructor)mbidecoder_dealloc, /* tp_dealloc */
1195 0, /* tp_print */
1196 0, /* tp_getattr */
1197 0, /* tp_setattr */
1198 0, /* tp_compare */
1199 0, /* tp_repr */
1200 0, /* tp_as_number */
1201 0, /* tp_as_sequence */
1202 0, /* tp_as_mapping */
1203 0, /* tp_hash */
1204 0, /* tp_call */
1205 0, /* tp_str */
1206 PyObject_GenericGetAttr, /* tp_getattro */
1207 0, /* tp_setattro */
1208 0, /* tp_as_buffer */
1209 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1210 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1211 0, /* tp_doc */
1212 (traverseproc)mbidecoder_traverse, /* tp_traverse */
1213 0, /* tp_clear */
1214 0, /* tp_richcompare */
1215 0, /* tp_weaklistoffset */
1216 0, /* tp_iter */
1217 0, /* tp_iterext */
1218 mbidecoder_methods, /* tp_methods */
1219 0, /* tp_members */
1220 codecctx_getsets, /* tp_getset */
1221 0, /* tp_base */
1222 0, /* tp_dict */
1223 0, /* tp_descr_get */
1224 0, /* tp_descr_set */
1225 0, /* tp_dictoffset */
1226 mbidecoder_init, /* tp_init */
1227 0, /* tp_alloc */
1228 mbidecoder_new, /* tp_new */
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001229};
1230
1231
1232/**
1233 * MultibyteStreamReader object
1234 */
1235
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001236static PyObject *
1237mbstreamreader_iread(MultibyteStreamReaderObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001238 const char *method, Py_ssize_t sizehint)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001239{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001240 MultibyteDecodeBuffer buf;
1241 PyObject *cres;
1242 Py_ssize_t rsize, finalsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001243
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001244 if (sizehint == 0)
1245 return PyUnicode_FromUnicode(NULL, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001246
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001247 buf.outobj = buf.excobj = NULL;
1248 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001249
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001250 for (;;) {
1251 int endoffile;
Hye-Shik Chang9b541402007-06-05 18:58:51 +00001252
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001253 if (sizehint < 0)
1254 cres = PyObject_CallMethod(self->stream,
1255 (char *)method, NULL);
1256 else
1257 cres = PyObject_CallMethod(self->stream,
1258 (char *)method, "i", sizehint);
1259 if (cres == NULL)
1260 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001261
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001262 if (!PyString_Check(cres)) {
1263 PyErr_SetString(PyExc_TypeError,
1264 "stream function returned a "
1265 "non-string object");
1266 goto errorexit;
1267 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001268
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001269 endoffile = (PyString_GET_SIZE(cres) == 0);
Hye-Shik Chang9b541402007-06-05 18:58:51 +00001270
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001271 if (self->pendingsize > 0) {
1272 PyObject *ctr;
1273 char *ctrdata;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001274
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001275 if (PyString_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
1276 PyErr_NoMemory();
1277 goto errorexit;
Martin Panter5f755022016-09-15 01:50:53 +00001278 }
1279 rsize = PyString_GET_SIZE(cres) + self->pendingsize;
1280 ctr = PyString_FromStringAndSize(NULL, rsize);
1281 if (ctr == NULL)
1282 goto errorexit;
1283 ctrdata = PyString_AS_STRING(ctr);
1284 memcpy(ctrdata, self->pending, self->pendingsize);
1285 memcpy(ctrdata + self->pendingsize,
1286 PyString_AS_STRING(cres),
1287 PyString_GET_SIZE(cres));
1288 Py_DECREF(cres);
1289 cres = ctr;
1290 self->pendingsize = 0;
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001291 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001292
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001293 rsize = PyString_GET_SIZE(cres);
1294 if (decoder_prepare_buffer(&buf, PyString_AS_STRING(cres),
1295 rsize) != 0)
1296 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001297
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001298 if (rsize > 0 && decoder_feed_buffer(
1299 (MultibyteStatefulDecoderContext *)self, &buf))
1300 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001301
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001302 if (endoffile || sizehint < 0) {
1303 if (buf.inbuf < buf.inbuf_end &&
1304 multibytecodec_decerror(self->codec, &self->state,
1305 &buf, self->errors, MBERR_TOOFEW))
1306 goto errorexit;
1307 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001308
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001309 if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
1310 if (decoder_append_pending(STATEFUL_DCTX(self),
1311 &buf) != 0)
1312 goto errorexit;
1313 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001314
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001315 finalsize = (Py_ssize_t)(buf.outbuf -
1316 PyUnicode_AS_UNICODE(buf.outobj));
1317 Py_DECREF(cres);
1318 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001319
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001320 if (sizehint < 0 || finalsize != 0 || rsize == 0)
1321 break;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001322
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001323 sizehint = 1; /* read 1 more byte and retry */
1324 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001325
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001326 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1327 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1328 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001329
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001330 Py_XDECREF(cres);
1331 Py_XDECREF(buf.excobj);
1332 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001333
1334errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001335 Py_XDECREF(cres);
1336 Py_XDECREF(buf.excobj);
1337 Py_XDECREF(buf.outobj);
1338 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001339}
1340
1341static PyObject *
1342mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args)
1343{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001344 PyObject *sizeobj = NULL;
1345 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001346
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001347 if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj))
1348 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001349
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001350 if (sizeobj == Py_None || sizeobj == NULL)
1351 size = -1;
1352 else if (PyInt_Check(sizeobj))
1353 size = PyInt_AsSsize_t(sizeobj);
1354 else {
1355 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1356 return NULL;
1357 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001358
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001359 return mbstreamreader_iread(self, "read", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001360}
1361
1362static PyObject *
1363mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args)
1364{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001365 PyObject *sizeobj = NULL;
1366 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001367
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001368 if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj))
1369 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001370
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001371 if (sizeobj == Py_None || sizeobj == NULL)
1372 size = -1;
1373 else if (PyInt_Check(sizeobj))
1374 size = PyInt_AsSsize_t(sizeobj);
1375 else {
1376 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1377 return NULL;
1378 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001379
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001380 return mbstreamreader_iread(self, "readline", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001381}
1382
1383static PyObject *
1384mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args)
1385{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001386 PyObject *sizehintobj = NULL, *r, *sr;
1387 Py_ssize_t sizehint;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001388
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001389 if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj))
1390 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001391
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001392 if (sizehintobj == Py_None || sizehintobj == NULL)
1393 sizehint = -1;
1394 else if (PyInt_Check(sizehintobj))
1395 sizehint = PyInt_AsSsize_t(sizehintobj);
1396 else {
1397 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1398 return NULL;
1399 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001400
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001401 r = mbstreamreader_iread(self, "read", sizehint);
1402 if (r == NULL)
1403 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001404
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001405 sr = PyUnicode_Splitlines(r, 1);
1406 Py_DECREF(r);
1407 return sr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001408}
1409
1410static PyObject *
1411mbstreamreader_reset(MultibyteStreamReaderObject *self)
1412{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001413 if (self->codec->decreset != NULL &&
1414 self->codec->decreset(&self->state, self->codec->config) != 0)
1415 return NULL;
1416 self->pendingsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001417
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001418 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001419}
1420
1421static struct PyMethodDef mbstreamreader_methods[] = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001422 {"read", (PyCFunction)mbstreamreader_read,
1423 METH_VARARGS, NULL},
1424 {"readline", (PyCFunction)mbstreamreader_readline,
1425 METH_VARARGS, NULL},
1426 {"readlines", (PyCFunction)mbstreamreader_readlines,
1427 METH_VARARGS, NULL},
1428 {"reset", (PyCFunction)mbstreamreader_reset,
1429 METH_NOARGS, NULL},
1430 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001431};
1432
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001433static PyMemberDef mbstreamreader_members[] = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001434 {"stream", T_OBJECT,
1435 offsetof(MultibyteStreamReaderObject, stream),
1436 READONLY, NULL},
1437 {NULL,}
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001438};
1439
1440static PyObject *
1441mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1442{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001443 MultibyteStreamReaderObject *self;
1444 PyObject *stream, *codec = NULL;
1445 char *errors = NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001446
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001447 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
1448 streamkwarglist, &stream, &errors))
1449 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001450
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001451 self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
1452 if (self == NULL)
1453 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001454
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001455 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1456 if (codec == NULL)
1457 goto errorexit;
1458 if (!MultibyteCodec_Check(codec)) {
1459 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1460 goto errorexit;
1461 }
Hye-Shik Changb9c03e92006-03-27 08:24:54 +00001462
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001463 self->codec = ((MultibyteCodecObject *)codec)->codec;
1464 self->stream = stream;
1465 Py_INCREF(stream);
1466 self->pendingsize = 0;
1467 self->errors = internal_error_callback(errors);
1468 if (self->errors == NULL)
1469 goto errorexit;
1470 if (self->codec->decinit != NULL &&
1471 self->codec->decinit(&self->state, self->codec->config) != 0)
1472 goto errorexit;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001473
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001474 Py_DECREF(codec);
1475 return (PyObject *)self;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001476
1477errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001478 Py_XDECREF(self);
1479 Py_XDECREF(codec);
1480 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001481}
1482
1483static int
Hye-Shik Changdb8955b2006-04-21 16:28:34 +00001484mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
1485{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001486 return 0;
Hye-Shik Changdb8955b2006-04-21 16:28:34 +00001487}
1488
1489static int
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001490mbstreamreader_traverse(MultibyteStreamReaderObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001491 visitproc visit, void *arg)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001492{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001493 if (ERROR_ISCUSTOM(self->errors))
1494 Py_VISIT(self->errors);
1495 Py_VISIT(self->stream);
1496 return 0;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001497}
1498
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001499static void
1500mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
1501{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001502 PyObject_GC_UnTrack(self);
1503 ERROR_DECREF(self->errors);
1504 Py_XDECREF(self->stream);
1505 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001506}
1507
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001508static PyTypeObject MultibyteStreamReader_Type = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001509 PyVarObject_HEAD_INIT(NULL, 0)
1510 "MultibyteStreamReader", /* tp_name */
1511 sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
1512 0, /* tp_itemsize */
1513 /* methods */
1514 (destructor)mbstreamreader_dealloc, /* tp_dealloc */
1515 0, /* tp_print */
1516 0, /* tp_getattr */
1517 0, /* tp_setattr */
1518 0, /* tp_compare */
1519 0, /* tp_repr */
1520 0, /* tp_as_number */
1521 0, /* tp_as_sequence */
1522 0, /* tp_as_mapping */
1523 0, /* tp_hash */
1524 0, /* tp_call */
1525 0, /* tp_str */
1526 PyObject_GenericGetAttr, /* tp_getattro */
1527 0, /* tp_setattro */
1528 0, /* tp_as_buffer */
1529 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1530 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1531 0, /* tp_doc */
1532 (traverseproc)mbstreamreader_traverse, /* tp_traverse */
1533 0, /* tp_clear */
1534 0, /* tp_richcompare */
1535 0, /* tp_weaklistoffset */
1536 0, /* tp_iter */
1537 0, /* tp_iterext */
1538 mbstreamreader_methods, /* tp_methods */
1539 mbstreamreader_members, /* tp_members */
1540 codecctx_getsets, /* tp_getset */
1541 0, /* tp_base */
1542 0, /* tp_dict */
1543 0, /* tp_descr_get */
1544 0, /* tp_descr_set */
1545 0, /* tp_dictoffset */
1546 mbstreamreader_init, /* tp_init */
1547 0, /* tp_alloc */
1548 mbstreamreader_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001549};
1550
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001551
1552/**
1553 * MultibyteStreamWriter object
1554 */
1555
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001556static int
1557mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001558 PyObject *unistr)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001559{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001560 PyObject *str, *wr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001561
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001562 str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
1563 if (str == NULL)
1564 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001565
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001566 wr = PyObject_CallMethod(self->stream, "write", "O", str);
1567 Py_DECREF(str);
1568 if (wr == NULL)
1569 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001570
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001571 Py_DECREF(wr);
1572 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001573}
1574
1575static PyObject *
Georg Brandl96a8c392006-05-29 21:04:52 +00001576mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001577{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001578 if (mbstreamwriter_iwrite(self, strobj))
1579 return NULL;
1580 else
1581 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001582}
1583
1584static PyObject *
Georg Brandl96a8c392006-05-29 21:04:52 +00001585mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001586{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001587 PyObject *strobj;
1588 int i, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001589
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001590 if (!PySequence_Check(lines)) {
1591 PyErr_SetString(PyExc_TypeError,
1592 "arg must be a sequence object");
1593 return NULL;
1594 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001595
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001596 for (i = 0; i < PySequence_Length(lines); i++) {
1597 /* length can be changed even within this loop */
1598 strobj = PySequence_GetItem(lines, i);
1599 if (strobj == NULL)
1600 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001601
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001602 r = mbstreamwriter_iwrite(self, strobj);
1603 Py_DECREF(strobj);
1604 if (r == -1)
1605 return NULL;
1606 }
Serhiy Storchaka64aa4df2017-04-19 22:34:58 +03001607 /* PySequence_Length() can fail */
1608 if (PyErr_Occurred())
1609 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001610
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001611 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001612}
1613
1614static PyObject *
1615mbstreamwriter_reset(MultibyteStreamWriterObject *self)
1616{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001617 const Py_UNICODE *pending;
1618 PyObject *pwrt;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001619
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001620 pending = self->pending;
1621 pwrt = multibytecodec_encode(self->codec, &self->state,
1622 &pending, self->pendingsize, self->errors,
1623 MBENC_FLUSH | MBENC_RESET);
1624 /* some pending buffer can be truncated when UnicodeEncodeError is
1625 * raised on 'strict' mode. but, 'reset' method is designed to
1626 * reset the pending buffer or states so failed string sequence
1627 * ought to be missed */
1628 self->pendingsize = 0;
1629 if (pwrt == NULL)
1630 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001631
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001632 if (PyString_Size(pwrt) > 0) {
1633 PyObject *wr;
1634 wr = PyObject_CallMethod(self->stream, "write", "O", pwrt);
1635 if (wr == NULL) {
1636 Py_DECREF(pwrt);
1637 return NULL;
1638 }
1639 }
1640 Py_DECREF(pwrt);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001641
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001642 Py_RETURN_NONE;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001643}
1644
1645static PyObject *
1646mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1647{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001648 MultibyteStreamWriterObject *self;
1649 PyObject *stream, *codec = NULL;
1650 char *errors = NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001651
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001652 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
1653 streamkwarglist, &stream, &errors))
1654 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001655
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001656 self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
1657 if (self == NULL)
1658 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001659
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001660 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1661 if (codec == NULL)
1662 goto errorexit;
1663 if (!MultibyteCodec_Check(codec)) {
1664 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1665 goto errorexit;
1666 }
Hye-Shik Changb9c03e92006-03-27 08:24:54 +00001667
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001668 self->codec = ((MultibyteCodecObject *)codec)->codec;
1669 self->stream = stream;
1670 Py_INCREF(stream);
1671 self->pendingsize = 0;
1672 self->errors = internal_error_callback(errors);
1673 if (self->errors == NULL)
1674 goto errorexit;
1675 if (self->codec->encinit != NULL &&
1676 self->codec->encinit(&self->state, self->codec->config) != 0)
1677 goto errorexit;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001678
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001679 Py_DECREF(codec);
1680 return (PyObject *)self;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001681
1682errorexit:
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001683 Py_XDECREF(self);
1684 Py_XDECREF(codec);
1685 return NULL;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001686}
1687
1688static int
Hye-Shik Changdb8955b2006-04-21 16:28:34 +00001689mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
1690{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001691 return 0;
Hye-Shik Changdb8955b2006-04-21 16:28:34 +00001692}
1693
1694static int
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001695mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001696 visitproc visit, void *arg)
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001697{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001698 if (ERROR_ISCUSTOM(self->errors))
1699 Py_VISIT(self->errors);
1700 Py_VISIT(self->stream);
1701 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001702}
1703
1704static void
1705mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
1706{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001707 PyObject_GC_UnTrack(self);
1708 ERROR_DECREF(self->errors);
1709 Py_XDECREF(self->stream);
1710 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001711}
1712
1713static struct PyMethodDef mbstreamwriter_methods[] = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001714 {"write", (PyCFunction)mbstreamwriter_write,
1715 METH_O, NULL},
1716 {"writelines", (PyCFunction)mbstreamwriter_writelines,
1717 METH_O, NULL},
1718 {"reset", (PyCFunction)mbstreamwriter_reset,
1719 METH_NOARGS, NULL},
1720 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001721};
1722
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001723static PyMemberDef mbstreamwriter_members[] = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001724 {"stream", T_OBJECT,
1725 offsetof(MultibyteStreamWriterObject, stream),
1726 READONLY, NULL},
1727 {NULL,}
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001728};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001729
1730static PyTypeObject MultibyteStreamWriter_Type = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001731 PyVarObject_HEAD_INIT(NULL, 0)
1732 "MultibyteStreamWriter", /* tp_name */
1733 sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
1734 0, /* tp_itemsize */
1735 /* methods */
1736 (destructor)mbstreamwriter_dealloc, /* tp_dealloc */
1737 0, /* tp_print */
1738 0, /* tp_getattr */
1739 0, /* tp_setattr */
1740 0, /* tp_compare */
1741 0, /* tp_repr */
1742 0, /* tp_as_number */
1743 0, /* tp_as_sequence */
1744 0, /* tp_as_mapping */
1745 0, /* tp_hash */
1746 0, /* tp_call */
1747 0, /* tp_str */
1748 PyObject_GenericGetAttr, /* tp_getattro */
1749 0, /* tp_setattro */
1750 0, /* tp_as_buffer */
1751 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1752 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1753 0, /* tp_doc */
1754 (traverseproc)mbstreamwriter_traverse, /* tp_traverse */
1755 0, /* tp_clear */
1756 0, /* tp_richcompare */
1757 0, /* tp_weaklistoffset */
1758 0, /* tp_iter */
1759 0, /* tp_iterext */
1760 mbstreamwriter_methods, /* tp_methods */
1761 mbstreamwriter_members, /* tp_members */
1762 codecctx_getsets, /* tp_getset */
1763 0, /* tp_base */
1764 0, /* tp_dict */
1765 0, /* tp_descr_get */
1766 0, /* tp_descr_set */
1767 0, /* tp_dictoffset */
1768 mbstreamwriter_init, /* tp_init */
1769 0, /* tp_alloc */
1770 mbstreamwriter_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001771};
1772
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001773
1774/**
1775 * Exposed factory function
1776 */
1777
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001778static PyObject *
1779__create_codec(PyObject *ignore, PyObject *arg)
1780{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001781 MultibyteCodecObject *self;
1782 MultibyteCodec *codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001783
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001784 if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
1785 PyErr_SetString(PyExc_ValueError, "argument type invalid");
1786 return NULL;
1787 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001788
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001789 codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
1790 if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
1791 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001792
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001793 self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
1794 if (self == NULL)
1795 return NULL;
1796 self->codec = codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001797
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001798 return (PyObject *)self;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001799}
1800
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001801static struct PyMethodDef __methods[] = {
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001802 {"__create_codec", (PyCFunction)__create_codec, METH_O},
1803 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001804};
1805
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001806PyMODINIT_FUNC
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001807init_multibytecodec(void)
1808{
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001809 int i;
1810 PyObject *m;
1811 PyTypeObject *typelist[] = {
1812 &MultibyteIncrementalEncoder_Type,
1813 &MultibyteIncrementalDecoder_Type,
1814 &MultibyteStreamReader_Type,
1815 &MultibyteStreamWriter_Type,
1816 NULL
1817 };
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001818
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001819 if (PyType_Ready(&MultibyteCodec_Type) < 0)
1820 return;
Hye-Shik Change2ac4ab2006-03-26 02:34:59 +00001821
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001822 m = Py_InitModule("_multibytecodec", __methods);
1823 if (m == NULL)
1824 return;
Neal Norwitz058bde12005-09-21 06:44:25 +00001825
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001826 for (i = 0; typelist[i] != NULL; i++) {
1827 if (PyType_Ready(typelist[i]) < 0)
1828 return;
1829 Py_INCREF(typelist[i]);
1830 PyModule_AddObject(m, typelist[i]->tp_name,
1831 (PyObject *)typelist[i]);
1832 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001833
Antoine Pitrouc83ea132010-05-09 14:46:46 +00001834 if (PyErr_Occurred())
1835 Py_FatalError("can't initialize the _multibytecodec module");
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001836}