blob: 2a29b7d4cd5c1e081798ad375bb19e9ccaced198 [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 {
Victor Stinnerd9491262013-04-14 02:06:32 +020013 PyObject *inobj;
14 Py_ssize_t inpos, inlen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000015 unsigned char *outbuf, *outbuf_end;
16 PyObject *excobj, *outobj;
17} MultibyteEncodeBuffer;
18
19typedef struct {
20 const unsigned char *inbuf, *inbuf_top, *inbuf_end;
Victor Stinnera0dd0212013-04-11 22:09:04 +020021 PyObject *excobj;
22 _PyUnicodeWriter writer;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000023} MultibyteDecodeBuffer;
24
25PyDoc_STRVAR(MultibyteCodec_Encode__doc__,
26"I.encode(unicode[, errors]) -> (string, length consumed)\n\
27\n\
28Return an encoded string version of `unicode'. errors may be given to\n\
29set a different error handling scheme. Default is 'strict' meaning that\n\
30encoding errors raise a UnicodeEncodeError. Other possible values are\n\
31'ignore', 'replace' and 'xmlcharrefreplace' as well as any other name\n\
32registered with codecs.register_error that can handle UnicodeEncodeErrors.");
33
34PyDoc_STRVAR(MultibyteCodec_Decode__doc__,
35"I.decode(string[, errors]) -> (unicodeobject, length consumed)\n\
36\n\
37Decodes `string' using I, an MultibyteCodec instance. errors may be given\n\
38to set a different error handling scheme. Default is 'strict' meaning\n\
39that encoding errors raise a UnicodeDecodeError. Other possible values\n\
Benjamin Petersonf10a79a2008-10-11 00:49:57 +000040are 'ignore' and 'replace' as well as any other name registered with\n\
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000041codecs.register_error that is able to handle UnicodeDecodeErrors.");
42
Martin v. Löwis02cbf4a2006-02-27 17:20:04 +000043static char *codeckwarglist[] = {"input", "errors", NULL};
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000044static char *incnewkwarglist[] = {"errors", NULL};
45static char *incrementalkwarglist[] = {"input", "final", NULL};
Martin v. Löwis02cbf4a2006-02-27 17:20:04 +000046static char *streamkwarglist[] = {"stream", "errors", NULL};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000047
48static PyObject *multibytecodec_encode(MultibyteCodec *,
Victor Stinnerd9491262013-04-14 02:06:32 +020049 MultibyteCodec_State *, PyObject *, Py_ssize_t *,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000050 PyObject *, int);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000051
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000052#define MBENC_RESET MBENC_MAX<<1 /* reset after an encoding session */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000053
54static PyObject *
Hye-Shik Chang4b96c132006-03-04 16:08:19 +000055make_tuple(PyObject *object, Py_ssize_t len)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000056{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000057 PyObject *v, *w;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000058
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000059 if (object == NULL)
60 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000061
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000062 v = PyTuple_New(2);
63 if (v == NULL) {
64 Py_DECREF(object);
65 return NULL;
66 }
67 PyTuple_SET_ITEM(v, 0, object);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000068
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000069 w = PyLong_FromSsize_t(len);
70 if (w == NULL) {
71 Py_DECREF(v);
72 return NULL;
73 }
74 PyTuple_SET_ITEM(v, 1, w);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000075
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000076 return v;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000077}
78
79static PyObject *
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000080internal_error_callback(const char *errors)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000081{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000082 if (errors == NULL || strcmp(errors, "strict") == 0)
83 return ERROR_STRICT;
84 else if (strcmp(errors, "ignore") == 0)
85 return ERROR_IGNORE;
86 else if (strcmp(errors, "replace") == 0)
87 return ERROR_REPLACE;
88 else
89 return PyUnicode_FromString(errors);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +000090}
91
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000092static PyObject *
93call_error_callback(PyObject *errors, PyObject *exc)
94{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000095 PyObject *args, *cb, *r;
96 const char *str;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +000097
Antoine Pitrouf95a1b32010-05-09 15:52:27 +000098 assert(PyUnicode_Check(errors));
99 str = _PyUnicode_AsString(errors);
100 if (str == NULL)
101 return NULL;
102 cb = PyCodec_LookupError(str);
103 if (cb == NULL)
104 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000105
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000106 args = PyTuple_New(1);
107 if (args == NULL) {
108 Py_DECREF(cb);
109 return NULL;
110 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000111
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000112 PyTuple_SET_ITEM(args, 0, exc);
113 Py_INCREF(exc);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000114
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000115 r = PyObject_CallObject(cb, args);
116 Py_DECREF(args);
117 Py_DECREF(cb);
118 return r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000119}
120
121static PyObject *
122codecctx_errors_get(MultibyteStatefulCodecContext *self)
123{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000124 const char *errors;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000125
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000126 if (self->errors == ERROR_STRICT)
127 errors = "strict";
128 else if (self->errors == ERROR_IGNORE)
129 errors = "ignore";
130 else if (self->errors == ERROR_REPLACE)
131 errors = "replace";
132 else {
133 Py_INCREF(self->errors);
134 return self->errors;
135 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000136
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000137 return PyUnicode_FromString(errors);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000138}
139
140static int
141codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000142 void *closure)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000143{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000144 PyObject *cb;
145 const char *str;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000146
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000147 if (!PyUnicode_Check(value)) {
148 PyErr_SetString(PyExc_TypeError, "errors must be a string");
149 return -1;
150 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000151
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000152 str = _PyUnicode_AsString(value);
153 if (str == NULL)
154 return -1;
Neal Norwitz6ea45d32007-08-26 04:19:43 +0000155
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000156 cb = internal_error_callback(str);
157 if (cb == NULL)
158 return -1;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000159
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000160 ERROR_DECREF(self->errors);
161 self->errors = cb;
162 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000163}
164
165/* This getset handlers list is used by all the stateful codec objects */
166static PyGetSetDef codecctx_getsets[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000167 {"errors", (getter)codecctx_errors_get,
168 (setter)codecctx_errors_set,
169 PyDoc_STR("how to treat errors")},
170 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000171};
172
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000173static int
Hye-Shik Chang4b96c132006-03-04 16:08:19 +0000174expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000175{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000176 Py_ssize_t orgpos, orgsize, incsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000177
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000178 orgpos = (Py_ssize_t)((char *)buf->outbuf -
179 PyBytes_AS_STRING(buf->outobj));
180 orgsize = PyBytes_GET_SIZE(buf->outobj);
181 incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize);
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000182
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000183 if (orgsize > PY_SSIZE_T_MAX - incsize)
184 return -1;
Amaury Forgeot d'Arc9c74b142008-06-18 00:47:36 +0000185
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000186 if (_PyBytes_Resize(&buf->outobj, orgsize + incsize) == -1)
187 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000188
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000189 buf->outbuf = (unsigned char *)PyBytes_AS_STRING(buf->outobj) +orgpos;
190 buf->outbuf_end = (unsigned char *)PyBytes_AS_STRING(buf->outobj)
191 + PyBytes_GET_SIZE(buf->outobj);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000192
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000193 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000194}
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000195#define REQUIRE_ENCODEBUFFER(buf, s) { \
196 if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end) \
197 if (expand_encodebuffer(buf, s) == -1) \
198 goto errorexit; \
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000199}
200
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000201
202/**
203 * MultibyteCodec object
204 */
205
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000206static int
207multibytecodec_encerror(MultibyteCodec *codec,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000208 MultibyteCodec_State *state,
209 MultibyteEncodeBuffer *buf,
210 PyObject *errors, Py_ssize_t e)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000211{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000212 PyObject *retobj = NULL, *retstr = NULL, *tobj;
213 Py_ssize_t retstrsize, newpos;
214 Py_ssize_t esize, start, end;
215 const char *reason;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000216
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000217 if (e > 0) {
218 reason = "illegal multibyte sequence";
219 esize = e;
220 }
221 else {
222 switch (e) {
223 case MBERR_TOOSMALL:
224 REQUIRE_ENCODEBUFFER(buf, -1);
225 return 0; /* retry it */
226 case MBERR_TOOFEW:
227 reason = "incomplete multibyte sequence";
Victor Stinnerd9491262013-04-14 02:06:32 +0200228 esize = (Py_ssize_t)buf->inpos;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000229 break;
230 case MBERR_INTERNAL:
231 PyErr_SetString(PyExc_RuntimeError,
232 "internal codec error");
233 return -1;
234 default:
235 PyErr_SetString(PyExc_RuntimeError,
236 "unknown runtime error");
237 return -1;
238 }
239 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000240
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000241 if (errors == ERROR_REPLACE) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200242 PyObject *replchar;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000243 Py_ssize_t r;
Victor Stinnerd9491262013-04-14 02:06:32 +0200244 Py_ssize_t inpos;
245 int kind;
246 void *data;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000247
Victor Stinnerd9491262013-04-14 02:06:32 +0200248 replchar = PyUnicode_FromOrdinal('?');
249 if (replchar == NULL)
250 goto errorexit;
251 kind = PyUnicode_KIND(replchar);
252 data = PyUnicode_DATA(replchar);
253
254 inpos = 0;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000255 for (;;) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200256 Py_ssize_t outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000257
Victor Stinnerd9491262013-04-14 02:06:32 +0200258 r = codec->encode(state, codec->config,
259 kind, data, &inpos, 1,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000260 &buf->outbuf, outleft, 0);
261 if (r == MBERR_TOOSMALL) {
262 REQUIRE_ENCODEBUFFER(buf, -1);
263 continue;
264 }
265 else
266 break;
267 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000268
Victor Stinnerd9491262013-04-14 02:06:32 +0200269 Py_DECREF(replchar);
270
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000271 if (r != 0) {
272 REQUIRE_ENCODEBUFFER(buf, 1);
273 *buf->outbuf++ = '?';
274 }
275 }
276 if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200277 buf->inpos += esize;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000278 return 0;
279 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000280
Victor Stinnerd9491262013-04-14 02:06:32 +0200281 start = (Py_ssize_t)buf->inpos;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000282 end = start + esize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000283
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000284 /* use cached exception object if available */
285 if (buf->excobj == NULL) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200286 buf->excobj = PyObject_CallFunction(PyExc_UnicodeEncodeError,
287 "sOnns",
288 codec->encoding, buf->inobj,
289 start, end, reason);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000290 if (buf->excobj == NULL)
291 goto errorexit;
292 }
293 else
294 if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 ||
295 PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 ||
296 PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0)
297 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000298
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000299 if (errors == ERROR_STRICT) {
300 PyCodec_StrictErrors(buf->excobj);
301 goto errorexit;
302 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000303
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000304 retobj = call_error_callback(errors, buf->excobj);
305 if (retobj == NULL)
306 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000307
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000308 if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
Benjamin Peterson47a00f32012-12-02 11:20:28 -0500309 (!PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) && !PyBytes_Check(tobj)) ||
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000310 !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) {
311 PyErr_SetString(PyExc_TypeError,
312 "encoding error handler must return "
Benjamin Petersonaff47232012-12-02 10:53:41 -0500313 "(str, int) tuple");
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000314 goto errorexit;
315 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000316
Benjamin Peterson47a00f32012-12-02 11:20:28 -0500317 if (PyUnicode_Check(tobj)) {
Victor Stinnerd9491262013-04-14 02:06:32 +0200318 Py_ssize_t inpos;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000319
Victor Stinnerd9491262013-04-14 02:06:32 +0200320 retstr = multibytecodec_encode(codec, state, tobj,
321 &inpos, ERROR_STRICT,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000322 MBENC_FLUSH);
323 if (retstr == NULL)
324 goto errorexit;
325 }
Benjamin Peterson47a00f32012-12-02 11:20:28 -0500326 else {
327 Py_INCREF(tobj);
328 retstr = tobj;
329 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000330
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000331 assert(PyBytes_Check(retstr));
332 retstrsize = PyBytes_GET_SIZE(retstr);
333 REQUIRE_ENCODEBUFFER(buf, retstrsize);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000334
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000335 memcpy(buf->outbuf, PyBytes_AS_STRING(retstr), retstrsize);
336 buf->outbuf += retstrsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000337
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000338 newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
339 if (newpos < 0 && !PyErr_Occurred())
Victor Stinnerd9491262013-04-14 02:06:32 +0200340 newpos += (Py_ssize_t)buf->inlen;
341 if (newpos < 0 || newpos > buf->inlen) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000342 PyErr_Clear();
343 PyErr_Format(PyExc_IndexError,
344 "position %zd from error handler out of bounds",
345 newpos);
346 goto errorexit;
347 }
Victor Stinnerd9491262013-04-14 02:06:32 +0200348 buf->inpos = newpos;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000349
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000350 Py_DECREF(retobj);
351 Py_DECREF(retstr);
352 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000353
354errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +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 Pitrouf95a1b32010-05-09 15:52:27 +0000362 MultibyteCodec_State *state,
363 MultibyteDecodeBuffer *buf,
364 PyObject *errors, Py_ssize_t e)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000365{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000366 PyObject *retobj = NULL, *retuni = NULL;
Victor Stinnera0dd0212013-04-11 22:09:04 +0200367 Py_ssize_t newpos;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000368 const char *reason;
369 Py_ssize_t esize, start, end;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000370
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000371 if (e > 0) {
372 reason = "illegal multibyte sequence";
373 esize = e;
374 }
375 else {
376 switch (e) {
377 case MBERR_TOOSMALL:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000378 return 0; /* retry it */
379 case MBERR_TOOFEW:
380 reason = "incomplete multibyte sequence";
381 esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
382 break;
383 case MBERR_INTERNAL:
384 PyErr_SetString(PyExc_RuntimeError,
385 "internal codec error");
386 return -1;
387 default:
388 PyErr_SetString(PyExc_RuntimeError,
389 "unknown runtime error");
390 return -1;
391 }
392 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000393
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000394 if (errors == ERROR_REPLACE) {
Victor Stinnera0dd0212013-04-11 22:09:04 +0200395 if (_PyUnicodeWriter_WriteChar(&buf->writer,
396 Py_UNICODE_REPLACEMENT_CHARACTER) < 0)
397 goto errorexit;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000398 }
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 Pitrouf95a1b32010-05-09 15:52:27 +0000404 start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
405 end = start + esize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000406
Antoine Pitrouf95a1b32010-05-09 15:52:27 +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 Pitrouf95a1b32010-05-09 15:52:27 +0000422 if (errors == ERROR_STRICT) {
423 PyCodec_StrictErrors(buf->excobj);
424 goto errorexit;
425 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000426
Antoine Pitrouf95a1b32010-05-09 15:52:27 +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 Pitrouf95a1b32010-05-09 15:52:27 +0000431 if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
432 !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) ||
433 !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) {
434 PyErr_SetString(PyExc_TypeError,
435 "decoding error handler must return "
Benjamin Petersonaff47232012-12-02 10:53:41 -0500436 "(str, int) tuple");
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000437 goto errorexit;
438 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000439
Victor Stinnera0dd0212013-04-11 22:09:04 +0200440 if (_PyUnicodeWriter_WriteStr(&buf->writer, retuni) < 0)
Victor Stinner4eea8492011-11-21 03:01:27 +0100441 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000442
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000443 newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
444 if (newpos < 0 && !PyErr_Occurred())
445 newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
446 if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
447 PyErr_Clear();
448 PyErr_Format(PyExc_IndexError,
449 "position %zd from error handler out of bounds",
450 newpos);
451 goto errorexit;
452 }
453 buf->inbuf = buf->inbuf_top + newpos;
454 Py_DECREF(retobj);
455 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000456
457errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000458 Py_XDECREF(retobj);
459 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000460}
461
462static PyObject *
463multibytecodec_encode(MultibyteCodec *codec,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000464 MultibyteCodec_State *state,
Victor Stinnerd9491262013-04-14 02:06:32 +0200465 PyObject *text, Py_ssize_t *inpos_t,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000466 PyObject *errors, int flags)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000467{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000468 MultibyteEncodeBuffer buf;
469 Py_ssize_t finalsize, r = 0;
Victor Stinnerd9491262013-04-14 02:06:32 +0200470 Py_ssize_t datalen;
471 int kind;
472 void *data;
473
474 if (PyUnicode_READY(text) < 0)
475 return NULL;
476 datalen = PyUnicode_GET_LENGTH(text);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000477
Victor Stinner6bcbef72011-05-24 22:17:55 +0200478 if (datalen == 0 && !(flags & MBENC_RESET))
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000479 return PyBytes_FromStringAndSize(NULL, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000480
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000481 buf.excobj = NULL;
Brett Cannonb94767f2011-02-22 20:15:44 +0000482 buf.outobj = NULL;
Victor Stinnerd9491262013-04-14 02:06:32 +0200483 buf.inobj = text; /* borrowed reference */
484 buf.inpos = 0;
485 buf.inlen = datalen;
486 kind = PyUnicode_KIND(buf.inobj);
487 data = PyUnicode_DATA(buf.inobj);
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
Victor Stinnerd9491262013-04-14 02:06:32 +0200500 while (buf.inpos < buf.inlen) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000501 /* we don't reuse inleft and outleft here.
502 * error callbacks can relocate the cursor anywhere on buffer*/
Victor Stinnerd9491262013-04-14 02:06:32 +0200503 Py_ssize_t outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
504
505 r = codec->encode(state, codec->config,
506 kind, data,
507 &buf.inpos, buf.inlen,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000508 &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
Victor Stinner6bcbef72011-05-24 22:17:55 +0200517 if (codec->encreset != NULL && (flags & MBENC_RESET))
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000518 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
Victor Stinnerd9491262013-04-14 02:06:32 +0200538 if (inpos_t)
539 *inpos_t = buf.inpos;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000540 Py_XDECREF(buf.excobj);
541 return buf.outobj;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000542
543errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000544 Py_XDECREF(buf.excobj);
545 Py_XDECREF(buf.outobj);
546 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000547}
548
549static PyObject *
550MultibyteCodec_Encode(MultibyteCodecObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000551 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000552{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000553 MultibyteCodec_State state;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000554 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
Victor Stinnerd9491262013-04-14 02:06:32 +0200576 if (PyUnicode_READY(arg) < 0) {
Victor Stinner9a80fab2011-11-21 02:50:14 +0100577 Py_XDECREF(ucvt);
578 return NULL;
579 }
Victor Stinnerd9491262013-04-14 02:06:32 +0200580 datalen = PyUnicode_GET_LENGTH(arg);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000581
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000582 errorcb = internal_error_callback(errors);
583 if (errorcb == NULL) {
584 Py_XDECREF(ucvt);
585 return NULL;
586 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000587
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000588 if (self->codec->encinit != NULL &&
589 self->codec->encinit(&state, self->codec->config) != 0)
590 goto errorexit;
591 r = multibytecodec_encode(self->codec, &state,
Victor Stinnerd9491262013-04-14 02:06:32 +0200592 arg, NULL, errorcb,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000593 MBENC_FLUSH | MBENC_RESET);
594 if (r == NULL)
595 goto errorexit;
Hye-Shik Chang2bb146f2004-07-18 03:06:29 +0000596
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000597 ERROR_DECREF(errorcb);
598 Py_XDECREF(ucvt);
599 return make_tuple(r, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000600
601errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000602 ERROR_DECREF(errorcb);
603 Py_XDECREF(ucvt);
604 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000605}
606
607static PyObject *
608MultibyteCodec_Decode(MultibyteCodecObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000609 PyObject *args, PyObject *kwargs)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000610{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000611 MultibyteCodec_State state;
612 MultibyteDecodeBuffer buf;
Victor Stinnera0dd0212013-04-11 22:09:04 +0200613 PyObject *errorcb, *res;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000614 Py_buffer pdata;
615 const char *data, *errors = NULL;
Victor Stinnera0dd0212013-04-11 22:09:04 +0200616 Py_ssize_t datalen;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000617
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000618 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|z:decode",
619 codeckwarglist, &pdata, &errors))
620 return NULL;
621 data = pdata.buf;
622 datalen = pdata.len;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000623
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000624 errorcb = internal_error_callback(errors);
625 if (errorcb == NULL) {
626 PyBuffer_Release(&pdata);
627 return NULL;
628 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000629
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000630 if (datalen == 0) {
631 PyBuffer_Release(&pdata);
632 ERROR_DECREF(errorcb);
Victor Stinnerb37b1742011-12-01 03:18:59 +0100633 return make_tuple(PyUnicode_New(0, 0), 0);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000634 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000635
Victor Stinner8f674cc2013-04-17 23:02:17 +0200636 _PyUnicodeWriter_Init(&buf.writer);
637 buf.writer.min_length = datalen;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000638 buf.excobj = NULL;
639 buf.inbuf = buf.inbuf_top = (unsigned char *)data;
640 buf.inbuf_end = buf.inbuf_top + datalen;
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) {
Victor Stinnera0dd0212013-04-11 22:09:04 +0200647 Py_ssize_t inleft, 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);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000650
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000651 r = self->codec->decode(&state, self->codec->config,
Victor Stinnera0dd0212013-04-11 22:09:04 +0200652 &buf.inbuf, inleft, &buf.writer);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000653 if (r == 0)
654 break;
655 else if (multibytecodec_decerror(self->codec, &state,
656 &buf, errorcb, r))
657 goto errorexit;
658 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000659
Victor Stinnera0dd0212013-04-11 22:09:04 +0200660 res = _PyUnicodeWriter_Finish(&buf.writer);
661 if (res == NULL)
662 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000663
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000664 PyBuffer_Release(&pdata);
665 Py_XDECREF(buf.excobj);
666 ERROR_DECREF(errorcb);
Victor Stinnera0dd0212013-04-11 22:09:04 +0200667 return make_tuple(res, datalen);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000668
669errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000670 PyBuffer_Release(&pdata);
671 ERROR_DECREF(errorcb);
672 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +0200673 _PyUnicodeWriter_Dealloc(&buf.writer);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +0000674
Antoine Pitrouf95a1b32010-05-09 15:52:27 +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 Pitrouf95a1b32010-05-09 15:52:27 +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 Pitrouf95a1b32010-05-09 15:52:27 +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 Pitrouf95a1b32010-05-09 15:52:27 +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_reserved */
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
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000726
727/**
728 * Utility functions for stateful codec mechanism
729 */
730
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000731#define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o))
732#define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o))
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000733
734static PyObject *
735encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000736 PyObject *unistr, int final)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000737{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000738 PyObject *ucvt, *r = NULL;
Victor Stinnerd9491262013-04-14 02:06:32 +0200739 PyObject *inbuf = NULL;
740 Py_ssize_t inpos, datalen;
741 PyObject *origpending = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000742
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000743 if (PyUnicode_Check(unistr))
744 ucvt = NULL;
745 else {
746 unistr = ucvt = PyObject_Str(unistr);
747 if (unistr == NULL)
748 return NULL;
749 else if (!PyUnicode_Check(unistr)) {
750 PyErr_SetString(PyExc_TypeError,
Benjamin Petersonaff47232012-12-02 10:53:41 -0500751 "couldn't convert the object to str.");
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000752 Py_DECREF(ucvt);
753 return NULL;
754 }
755 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000756
Victor Stinnerd9491262013-04-14 02:06:32 +0200757 if (ctx->pending) {
758 PyObject *inbuf_tmp;
759
760 Py_INCREF(ctx->pending);
761 origpending = ctx->pending;
762
763 Py_INCREF(ctx->pending);
764 inbuf_tmp = ctx->pending;
765 PyUnicode_Append(&inbuf_tmp, unistr);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000766 if (inbuf_tmp == NULL)
767 goto errorexit;
Victor Stinnerd9491262013-04-14 02:06:32 +0200768 Py_CLEAR(ctx->pending);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000769 inbuf = inbuf_tmp;
770 }
Victor Stinnerd9491262013-04-14 02:06:32 +0200771 else {
772 origpending = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000773
Victor Stinnerd9491262013-04-14 02:06:32 +0200774 Py_INCREF(unistr);
775 inbuf = unistr;
776 }
777 if (PyUnicode_READY(inbuf) < 0)
778 goto errorexit;
779 inpos = 0;
780 datalen = PyUnicode_GET_LENGTH(inbuf);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000781
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000782 r = multibytecodec_encode(ctx->codec, &ctx->state,
Victor Stinnerd9491262013-04-14 02:06:32 +0200783 inbuf, &inpos,
784 ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000785 if (r == NULL) {
786 /* recover the original pending buffer */
Victor Stinnerd9491262013-04-14 02:06:32 +0200787 Py_CLEAR(ctx->pending);
788 ctx->pending = origpending;
789 origpending = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000790 goto errorexit;
791 }
Victor Stinner322cc742013-04-14 18:11:41 +0200792 Py_XDECREF(origpending);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000793
Victor Stinnerd9491262013-04-14 02:06:32 +0200794 if (inpos < datalen) {
795 if (datalen - inpos > MAXENCPENDING) {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000796 /* normal codecs can't reach here */
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000797 PyErr_SetString(PyExc_UnicodeError,
798 "pending buffer overflow");
799 goto errorexit;
800 }
Victor Stinnerd9491262013-04-14 02:06:32 +0200801 ctx->pending = PyUnicode_Substring(inbuf, inpos, datalen);
802 if (ctx->pending == NULL) {
803 /* normal codecs can't reach here */
804 goto errorexit;
805 }
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000806 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000807
Victor Stinner322cc742013-04-14 18:11:41 +0200808 Py_DECREF(inbuf);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000809 Py_XDECREF(ucvt);
810 return r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000811
812errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000813 Py_XDECREF(r);
814 Py_XDECREF(ucvt);
Victor Stinnerd9491262013-04-14 02:06:32 +0200815 Py_XDECREF(origpending);
Victor Stinner322cc742013-04-14 18:11:41 +0200816 Py_XDECREF(inbuf);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000817 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000818}
819
820static int
821decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000822 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000823{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000824 Py_ssize_t npendings;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000825
Antoine Pitrouf95a1b32010-05-09 15:52:27 +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;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000835}
836
837static int
838decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000839 Py_ssize_t size)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000840{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000841 buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
842 buf->inbuf_end = buf->inbuf_top + size;
Victor Stinner8f674cc2013-04-17 23:02:17 +0200843 buf->writer.min_length += size;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000844 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000845}
846
847static int
848decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000849 MultibyteDecodeBuffer *buf)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000850{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000851 while (buf->inbuf < buf->inbuf_end) {
Victor Stinnera0dd0212013-04-11 22:09:04 +0200852 Py_ssize_t inleft;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000853 Py_ssize_t r;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000854
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000855 inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000856
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000857 r = ctx->codec->decode(&ctx->state, ctx->codec->config,
Victor Stinnera0dd0212013-04-11 22:09:04 +0200858 &buf->inbuf, inleft, &buf->writer);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000859 if (r == 0 || r == MBERR_TOOFEW)
860 break;
861 else if (multibytecodec_decerror(ctx->codec, &ctx->state,
862 buf, ctx->errors, r))
863 return -1;
864 }
865 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000866}
867
868
869/**
870 * MultibyteIncrementalEncoder object
871 */
872
873static PyObject *
874mbiencoder_encode(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000875 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000876{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000877 PyObject *data;
878 int final = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000879
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000880 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode",
881 incrementalkwarglist, &data, &final))
882 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000883
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000884 return encoder_encode_stateful(STATEFUL_ECTX(self), data, final);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000885}
886
887static PyObject *
888mbiencoder_reset(MultibyteIncrementalEncoderObject *self)
889{
Victor Stinnere15dce32011-05-30 22:56:00 +0200890 /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */
891 unsigned char buffer[4], *outbuf;
892 Py_ssize_t r;
893 if (self->codec->encreset != NULL) {
894 outbuf = buffer;
895 r = self->codec->encreset(&self->state, self->codec->config,
896 &outbuf, sizeof(buffer));
897 if (r != 0)
898 return NULL;
899 }
Victor Stinnerd9491262013-04-14 02:06:32 +0200900 Py_CLEAR(self->pending);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000901 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000902}
903
904static struct PyMethodDef mbiencoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000905 {"encode", (PyCFunction)mbiencoder_encode,
906 METH_VARARGS | METH_KEYWORDS, NULL},
907 {"reset", (PyCFunction)mbiencoder_reset,
908 METH_NOARGS, NULL},
909 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000910};
911
912static PyObject *
913mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
914{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000915 MultibyteIncrementalEncoderObject *self;
916 PyObject *codec = NULL;
917 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000918
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000919 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
920 incnewkwarglist, &errors))
921 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000922
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000923 self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
924 if (self == NULL)
925 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000926
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000927 codec = PyObject_GetAttrString((PyObject *)type, "codec");
928 if (codec == NULL)
929 goto errorexit;
930 if (!MultibyteCodec_Check(codec)) {
931 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
932 goto errorexit;
933 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000934
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000935 self->codec = ((MultibyteCodecObject *)codec)->codec;
Victor Stinnerd9491262013-04-14 02:06:32 +0200936 self->pending = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000937 self->errors = internal_error_callback(errors);
938 if (self->errors == NULL)
939 goto errorexit;
940 if (self->codec->encinit != NULL &&
941 self->codec->encinit(&self->state, self->codec->config) != 0)
942 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000943
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000944 Py_DECREF(codec);
945 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000946
947errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000948 Py_XDECREF(self);
949 Py_XDECREF(codec);
950 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000951}
952
953static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000954mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
955{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000956 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +0000957}
958
959static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000960mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000961 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000962{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000963 if (ERROR_ISCUSTOM(self->errors))
964 Py_VISIT(self->errors);
965 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000966}
967
968static void
969mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
970{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000971 PyObject_GC_UnTrack(self);
972 ERROR_DECREF(self->errors);
973 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +0000974}
975
976static PyTypeObject MultibyteIncrementalEncoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +0000977 PyVarObject_HEAD_INIT(NULL, 0)
978 "MultibyteIncrementalEncoder", /* tp_name */
979 sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
980 0, /* tp_itemsize */
981 /* methods */
982 (destructor)mbiencoder_dealloc, /* tp_dealloc */
983 0, /* tp_print */
984 0, /* tp_getattr */
985 0, /* tp_setattr */
986 0, /* tp_reserved */
987 0, /* tp_repr */
988 0, /* tp_as_number */
989 0, /* tp_as_sequence */
990 0, /* tp_as_mapping */
991 0, /* tp_hash */
992 0, /* tp_call */
993 0, /* tp_str */
994 PyObject_GenericGetAttr, /* tp_getattro */
995 0, /* tp_setattro */
996 0, /* tp_as_buffer */
997 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
998 | Py_TPFLAGS_BASETYPE, /* tp_flags */
999 0, /* tp_doc */
1000 (traverseproc)mbiencoder_traverse, /* tp_traverse */
1001 0, /* tp_clear */
1002 0, /* tp_richcompare */
1003 0, /* tp_weaklistoffset */
1004 0, /* tp_iter */
1005 0, /* tp_iterext */
1006 mbiencoder_methods, /* tp_methods */
1007 0, /* tp_members */
1008 codecctx_getsets, /* tp_getset */
1009 0, /* tp_base */
1010 0, /* tp_dict */
1011 0, /* tp_descr_get */
1012 0, /* tp_descr_set */
1013 0, /* tp_dictoffset */
1014 mbiencoder_init, /* tp_init */
1015 0, /* tp_alloc */
1016 mbiencoder_new, /* tp_new */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001017};
1018
1019
1020/**
1021 * MultibyteIncrementalDecoder object
1022 */
1023
1024static PyObject *
1025mbidecoder_decode(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001026 PyObject *args, PyObject *kwargs)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001027{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001028 MultibyteDecodeBuffer buf;
1029 char *data, *wdata = NULL;
1030 Py_buffer pdata;
Victor Stinnera0dd0212013-04-11 22:09:04 +02001031 Py_ssize_t wsize, size, origpending;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001032 int final = 0;
Victor Stinnera0dd0212013-04-11 22:09:04 +02001033 PyObject *res;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001034
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001035 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|i:decode",
1036 incrementalkwarglist, &pdata, &final))
1037 return NULL;
1038 data = pdata.buf;
1039 size = pdata.len;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001040
Victor Stinner8f674cc2013-04-17 23:02:17 +02001041 _PyUnicodeWriter_Init(&buf.writer);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001042 buf.excobj = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001043 origpending = self->pendingsize;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001044
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001045 if (self->pendingsize == 0) {
1046 wsize = size;
1047 wdata = data;
1048 }
1049 else {
1050 if (size > PY_SSIZE_T_MAX - self->pendingsize) {
1051 PyErr_NoMemory();
1052 goto errorexit;
1053 }
1054 wsize = size + self->pendingsize;
1055 wdata = PyMem_Malloc(wsize);
Victor Stinner33283ba2013-07-15 17:47:39 +02001056 if (wdata == NULL) {
1057 PyErr_NoMemory();
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001058 goto errorexit;
Victor Stinner33283ba2013-07-15 17:47:39 +02001059 }
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001060 memcpy(wdata, self->pending, self->pendingsize);
1061 memcpy(wdata + self->pendingsize, data, size);
1062 self->pendingsize = 0;
1063 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001064
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001065 if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
1066 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001067
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001068 if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
1069 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001070
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001071 if (final && buf.inbuf < buf.inbuf_end) {
1072 if (multibytecodec_decerror(self->codec, &self->state,
1073 &buf, self->errors, MBERR_TOOFEW)) {
1074 /* recover the original pending buffer */
1075 memcpy(self->pending, wdata, origpending);
1076 self->pendingsize = origpending;
1077 goto errorexit;
1078 }
1079 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001080
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001081 if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
1082 if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
1083 goto errorexit;
1084 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001085
Victor Stinnera0dd0212013-04-11 22:09:04 +02001086 res = _PyUnicodeWriter_Finish(&buf.writer);
1087 if (res == NULL)
1088 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001089
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001090 PyBuffer_Release(&pdata);
1091 if (wdata != data)
1092 PyMem_Del(wdata);
1093 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001094 return res;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001095
1096errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001097 PyBuffer_Release(&pdata);
1098 if (wdata != NULL && wdata != data)
1099 PyMem_Del(wdata);
1100 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001101 _PyUnicodeWriter_Dealloc(&buf.writer);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001102 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001103}
1104
1105static PyObject *
1106mbidecoder_reset(MultibyteIncrementalDecoderObject *self)
1107{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001108 if (self->codec->decreset != NULL &&
1109 self->codec->decreset(&self->state, self->codec->config) != 0)
1110 return NULL;
1111 self->pendingsize = 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001112
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001113 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001114}
1115
1116static struct PyMethodDef mbidecoder_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001117 {"decode", (PyCFunction)mbidecoder_decode,
1118 METH_VARARGS | METH_KEYWORDS, NULL},
1119 {"reset", (PyCFunction)mbidecoder_reset,
1120 METH_NOARGS, NULL},
1121 {NULL, NULL},
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001122};
1123
1124static PyObject *
1125mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1126{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001127 MultibyteIncrementalDecoderObject *self;
1128 PyObject *codec = NULL;
1129 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001130
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001131 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
1132 incnewkwarglist, &errors))
1133 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001134
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001135 self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
1136 if (self == NULL)
1137 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001138
Antoine Pitrouf95a1b32010-05-09 15:52:27 +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 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001146
Antoine Pitrouf95a1b32010-05-09 15:52:27 +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;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001155
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001156 Py_DECREF(codec);
1157 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001158
1159errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001160 Py_XDECREF(self);
1161 Py_XDECREF(codec);
1162 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001163}
1164
1165static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001166mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
1167{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001168 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001169}
1170
1171static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001172mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001173 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001174{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001175 if (ERROR_ISCUSTOM(self->errors))
1176 Py_VISIT(self->errors);
1177 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001178}
1179
1180static void
1181mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
1182{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001183 PyObject_GC_UnTrack(self);
1184 ERROR_DECREF(self->errors);
1185 Py_TYPE(self)->tp_free(self);
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001186}
1187
1188static PyTypeObject MultibyteIncrementalDecoder_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +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_reserved */
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 */
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001229};
1230
1231
1232/**
1233 * MultibyteStreamReader object
1234 */
1235
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001236static PyObject *
1237mbstreamreader_iread(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001238 const char *method, Py_ssize_t sizehint)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001239{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001240 MultibyteDecodeBuffer buf;
Victor Stinnera0dd0212013-04-11 22:09:04 +02001241 PyObject *cres, *res;
1242 Py_ssize_t rsize;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001243
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001244 if (sizehint == 0)
Victor Stinnerb37b1742011-12-01 03:18:59 +01001245 return PyUnicode_New(0, 0);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001246
Victor Stinner8f674cc2013-04-17 23:02:17 +02001247 _PyUnicodeWriter_Init(&buf.writer);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001248 buf.excobj = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001249 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001250
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001251 for (;;) {
1252 int endoffile;
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001253
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001254 if (sizehint < 0)
1255 cres = PyObject_CallMethod(self->stream,
1256 (char *)method, NULL);
1257 else
1258 cres = PyObject_CallMethod(self->stream,
1259 (char *)method, "i", sizehint);
1260 if (cres == NULL)
1261 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001262
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001263 if (!PyBytes_Check(cres)) {
1264 PyErr_Format(PyExc_TypeError,
1265 "stream function returned a "
1266 "non-bytes object (%.100s)",
1267 cres->ob_type->tp_name);
1268 goto errorexit;
1269 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001270
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001271 endoffile = (PyBytes_GET_SIZE(cres) == 0);
Guido van Rossumcd16bf62007-06-13 18:07:49 +00001272
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001273 if (self->pendingsize > 0) {
1274 PyObject *ctr;
1275 char *ctrdata;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001276
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001277 if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
1278 PyErr_NoMemory();
1279 goto errorexit;
Victor Stinner064bbdc2013-07-08 22:28:27 +02001280 }
1281 rsize = PyBytes_GET_SIZE(cres) + self->pendingsize;
1282 ctr = PyBytes_FromStringAndSize(NULL, rsize);
1283 if (ctr == NULL)
1284 goto errorexit;
1285 ctrdata = PyBytes_AS_STRING(ctr);
1286 memcpy(ctrdata, self->pending, self->pendingsize);
1287 memcpy(ctrdata + self->pendingsize,
1288 PyBytes_AS_STRING(cres),
1289 PyBytes_GET_SIZE(cres));
1290 Py_DECREF(cres);
1291 cres = ctr;
1292 self->pendingsize = 0;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001293 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001294
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001295 rsize = PyBytes_GET_SIZE(cres);
1296 if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres),
1297 rsize) != 0)
1298 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001299
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001300 if (rsize > 0 && decoder_feed_buffer(
1301 (MultibyteStatefulDecoderContext *)self, &buf))
1302 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001303
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001304 if (endoffile || sizehint < 0) {
1305 if (buf.inbuf < buf.inbuf_end &&
1306 multibytecodec_decerror(self->codec, &self->state,
1307 &buf, self->errors, MBERR_TOOFEW))
1308 goto errorexit;
1309 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001310
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001311 if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
1312 if (decoder_append_pending(STATEFUL_DCTX(self),
1313 &buf) != 0)
1314 goto errorexit;
1315 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001316
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001317 Py_DECREF(cres);
1318 cres = NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001319
Victor Stinnera0dd0212013-04-11 22:09:04 +02001320 if (sizehint < 0 || buf.writer.pos != 0 || rsize == 0)
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001321 break;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001322
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001323 sizehint = 1; /* read 1 more byte and retry */
1324 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001325
Victor Stinnera0dd0212013-04-11 22:09:04 +02001326 res = _PyUnicodeWriter_Finish(&buf.writer);
1327 if (res == NULL)
1328 goto errorexit;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001329
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001330 Py_XDECREF(cres);
1331 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001332 return res;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001333
1334errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001335 Py_XDECREF(cres);
1336 Py_XDECREF(buf.excobj);
Victor Stinnera0dd0212013-04-11 22:09:04 +02001337 _PyUnicodeWriter_Dealloc(&buf.writer);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001338 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001339}
1340
1341static PyObject *
1342mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args)
1343{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001344 PyObject *sizeobj = NULL;
1345 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001346
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001347 if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj))
1348 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001349
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001350 if (sizeobj == Py_None || sizeobj == NULL)
1351 size = -1;
1352 else if (PyLong_Check(sizeobj))
1353 size = PyLong_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 Pitrouf95a1b32010-05-09 15:52:27 +00001359 if (size == -1 && PyErr_Occurred())
1360 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001361
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001362 return mbstreamreader_iread(self, "read", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001363}
1364
1365static PyObject *
1366mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args)
1367{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001368 PyObject *sizeobj = NULL;
1369 Py_ssize_t size;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001370
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001371 if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj))
1372 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001373
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001374 if (sizeobj == Py_None || sizeobj == NULL)
1375 size = -1;
1376 else if (PyLong_Check(sizeobj))
1377 size = PyLong_AsSsize_t(sizeobj);
1378 else {
1379 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1380 return NULL;
1381 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001382
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001383 if (size == -1 && PyErr_Occurred())
1384 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001385
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001386 return mbstreamreader_iread(self, "readline", size);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001387}
1388
1389static PyObject *
1390mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args)
1391{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001392 PyObject *sizehintobj = NULL, *r, *sr;
1393 Py_ssize_t sizehint;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001394
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001395 if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj))
1396 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001397
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001398 if (sizehintobj == Py_None || sizehintobj == NULL)
1399 sizehint = -1;
1400 else if (PyLong_Check(sizehintobj))
1401 sizehint = PyLong_AsSsize_t(sizehintobj);
1402 else {
1403 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1404 return NULL;
1405 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001406
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001407 if (sizehint == -1 && PyErr_Occurred())
1408 return NULL;
Guido van Rossumddefaf32007-01-14 03:31:43 +00001409
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001410 r = mbstreamreader_iread(self, "read", sizehint);
1411 if (r == NULL)
1412 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001413
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001414 sr = PyUnicode_Splitlines(r, 1);
1415 Py_DECREF(r);
1416 return sr;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001417}
1418
1419static PyObject *
1420mbstreamreader_reset(MultibyteStreamReaderObject *self)
1421{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001422 if (self->codec->decreset != NULL &&
1423 self->codec->decreset(&self->state, self->codec->config) != 0)
1424 return NULL;
1425 self->pendingsize = 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001426
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001427 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001428}
1429
1430static struct PyMethodDef mbstreamreader_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001431 {"read", (PyCFunction)mbstreamreader_read,
1432 METH_VARARGS, NULL},
1433 {"readline", (PyCFunction)mbstreamreader_readline,
1434 METH_VARARGS, NULL},
1435 {"readlines", (PyCFunction)mbstreamreader_readlines,
1436 METH_VARARGS, NULL},
1437 {"reset", (PyCFunction)mbstreamreader_reset,
1438 METH_NOARGS, NULL},
1439 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001440};
1441
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001442static PyMemberDef mbstreamreader_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001443 {"stream", T_OBJECT,
1444 offsetof(MultibyteStreamReaderObject, stream),
1445 READONLY, NULL},
1446 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001447};
1448
1449static PyObject *
1450mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1451{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001452 MultibyteStreamReaderObject *self;
1453 PyObject *stream, *codec = NULL;
1454 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001455
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001456 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
1457 streamkwarglist, &stream, &errors))
1458 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001459
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001460 self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
1461 if (self == NULL)
1462 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001463
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001464 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1465 if (codec == NULL)
1466 goto errorexit;
1467 if (!MultibyteCodec_Check(codec)) {
1468 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1469 goto errorexit;
1470 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001471
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001472 self->codec = ((MultibyteCodecObject *)codec)->codec;
1473 self->stream = stream;
1474 Py_INCREF(stream);
1475 self->pendingsize = 0;
1476 self->errors = internal_error_callback(errors);
1477 if (self->errors == NULL)
1478 goto errorexit;
1479 if (self->codec->decinit != NULL &&
1480 self->codec->decinit(&self->state, self->codec->config) != 0)
1481 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001482
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001483 Py_DECREF(codec);
1484 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001485
1486errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001487 Py_XDECREF(self);
1488 Py_XDECREF(codec);
1489 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001490}
1491
1492static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001493mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
1494{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001495 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001496}
1497
1498static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001499mbstreamreader_traverse(MultibyteStreamReaderObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001500 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001501{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001502 if (ERROR_ISCUSTOM(self->errors))
1503 Py_VISIT(self->errors);
1504 Py_VISIT(self->stream);
1505 return 0;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001506}
1507
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001508static void
1509mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
1510{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001511 PyObject_GC_UnTrack(self);
1512 ERROR_DECREF(self->errors);
1513 Py_XDECREF(self->stream);
1514 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001515}
1516
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001517static PyTypeObject MultibyteStreamReader_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001518 PyVarObject_HEAD_INIT(NULL, 0)
1519 "MultibyteStreamReader", /* tp_name */
1520 sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
1521 0, /* tp_itemsize */
1522 /* methods */
1523 (destructor)mbstreamreader_dealloc, /* tp_dealloc */
1524 0, /* tp_print */
1525 0, /* tp_getattr */
1526 0, /* tp_setattr */
1527 0, /* tp_reserved */
1528 0, /* tp_repr */
1529 0, /* tp_as_number */
1530 0, /* tp_as_sequence */
1531 0, /* tp_as_mapping */
1532 0, /* tp_hash */
1533 0, /* tp_call */
1534 0, /* tp_str */
1535 PyObject_GenericGetAttr, /* tp_getattro */
1536 0, /* tp_setattro */
1537 0, /* tp_as_buffer */
1538 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1539 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1540 0, /* tp_doc */
1541 (traverseproc)mbstreamreader_traverse, /* tp_traverse */
1542 0, /* tp_clear */
1543 0, /* tp_richcompare */
1544 0, /* tp_weaklistoffset */
1545 0, /* tp_iter */
1546 0, /* tp_iterext */
1547 mbstreamreader_methods, /* tp_methods */
1548 mbstreamreader_members, /* tp_members */
1549 codecctx_getsets, /* tp_getset */
1550 0, /* tp_base */
1551 0, /* tp_dict */
1552 0, /* tp_descr_get */
1553 0, /* tp_descr_set */
1554 0, /* tp_dictoffset */
1555 mbstreamreader_init, /* tp_init */
1556 0, /* tp_alloc */
1557 mbstreamreader_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001558};
1559
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001560
1561/**
1562 * MultibyteStreamWriter object
1563 */
1564
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001565static int
1566mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001567 PyObject *unistr)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001568{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001569 PyObject *str, *wr;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001570 _Py_IDENTIFIER(write);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001571
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001572 str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
1573 if (str == NULL)
1574 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001575
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001576 wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001577 Py_DECREF(str);
1578 if (wr == NULL)
1579 return -1;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001580
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001581 Py_DECREF(wr);
1582 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001583}
1584
1585static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001586mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001587{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001588 if (mbstreamwriter_iwrite(self, strobj))
1589 return NULL;
1590 else
1591 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001592}
1593
1594static PyObject *
Thomas Wouters4d70c3d2006-06-08 14:42:34 +00001595mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001596{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001597 PyObject *strobj;
1598 int i, r;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001599
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001600 if (!PySequence_Check(lines)) {
1601 PyErr_SetString(PyExc_TypeError,
1602 "arg must be a sequence object");
1603 return NULL;
1604 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001605
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001606 for (i = 0; i < PySequence_Length(lines); i++) {
1607 /* length can be changed even within this loop */
1608 strobj = PySequence_GetItem(lines, i);
1609 if (strobj == NULL)
1610 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001611
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001612 r = mbstreamwriter_iwrite(self, strobj);
1613 Py_DECREF(strobj);
1614 if (r == -1)
1615 return NULL;
1616 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001617
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001618 Py_RETURN_NONE;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001619}
1620
1621static PyObject *
1622mbstreamwriter_reset(MultibyteStreamWriterObject *self)
1623{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001624 PyObject *pwrt;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001625
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001626 pwrt = multibytecodec_encode(self->codec, &self->state,
Victor Stinnerd9491262013-04-14 02:06:32 +02001627 self->pending, NULL, self->errors,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001628 MBENC_FLUSH | MBENC_RESET);
1629 /* some pending buffer can be truncated when UnicodeEncodeError is
1630 * raised on 'strict' mode. but, 'reset' method is designed to
1631 * reset the pending buffer or states so failed string sequence
1632 * ought to be missed */
Victor Stinnerd9491262013-04-14 02:06:32 +02001633 Py_CLEAR(self->pending);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001634 if (pwrt == NULL)
1635 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001636
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001637 assert(PyBytes_Check(pwrt));
1638 if (PyBytes_Size(pwrt) > 0) {
1639 PyObject *wr;
Martin v. Löwisbd928fe2011-10-14 10:20:37 +02001640 _Py_IDENTIFIER(write);
Martin v. Löwisafe55bb2011-10-09 10:38:36 +02001641
1642 wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt);
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001643 if (wr == NULL) {
1644 Py_DECREF(pwrt);
1645 return NULL;
1646 }
1647 }
1648 Py_DECREF(pwrt);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001649
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001650 Py_RETURN_NONE;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001651}
1652
1653static PyObject *
1654mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1655{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001656 MultibyteStreamWriterObject *self;
1657 PyObject *stream, *codec = NULL;
1658 char *errors = NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001659
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001660 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
1661 streamkwarglist, &stream, &errors))
1662 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001663
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001664 self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
1665 if (self == NULL)
1666 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001667
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001668 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1669 if (codec == NULL)
1670 goto errorexit;
1671 if (!MultibyteCodec_Check(codec)) {
1672 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1673 goto errorexit;
1674 }
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001675
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001676 self->codec = ((MultibyteCodecObject *)codec)->codec;
1677 self->stream = stream;
1678 Py_INCREF(stream);
Victor Stinnerd9491262013-04-14 02:06:32 +02001679 self->pending = NULL;
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001680 self->errors = internal_error_callback(errors);
1681 if (self->errors == NULL)
1682 goto errorexit;
1683 if (self->codec->encinit != NULL &&
1684 self->codec->encinit(&self->state, self->codec->config) != 0)
1685 goto errorexit;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001686
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001687 Py_DECREF(codec);
1688 return (PyObject *)self;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001689
1690errorexit:
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001691 Py_XDECREF(self);
1692 Py_XDECREF(codec);
1693 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001694}
1695
1696static int
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001697mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
1698{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001699 return 0;
Hye-Shik Chang13247bf2006-04-21 16:21:44 +00001700}
1701
1702static int
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001703mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001704 visitproc visit, void *arg)
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001705{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001706 if (ERROR_ISCUSTOM(self->errors))
1707 Py_VISIT(self->errors);
1708 Py_VISIT(self->stream);
1709 return 0;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001710}
1711
1712static void
1713mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
1714{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001715 PyObject_GC_UnTrack(self);
1716 ERROR_DECREF(self->errors);
1717 Py_XDECREF(self->stream);
1718 Py_TYPE(self)->tp_free(self);
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001719}
1720
1721static struct PyMethodDef mbstreamwriter_methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001722 {"write", (PyCFunction)mbstreamwriter_write,
1723 METH_O, NULL},
1724 {"writelines", (PyCFunction)mbstreamwriter_writelines,
1725 METH_O, NULL},
1726 {"reset", (PyCFunction)mbstreamwriter_reset,
1727 METH_NOARGS, NULL},
1728 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001729};
1730
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001731static PyMemberDef mbstreamwriter_members[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001732 {"stream", T_OBJECT,
1733 offsetof(MultibyteStreamWriterObject, stream),
1734 READONLY, NULL},
1735 {NULL,}
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001736};
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001737
1738static PyTypeObject MultibyteStreamWriter_Type = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001739 PyVarObject_HEAD_INIT(NULL, 0)
1740 "MultibyteStreamWriter", /* tp_name */
1741 sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
1742 0, /* tp_itemsize */
1743 /* methods */
1744 (destructor)mbstreamwriter_dealloc, /* tp_dealloc */
1745 0, /* tp_print */
1746 0, /* tp_getattr */
1747 0, /* tp_setattr */
1748 0, /* tp_reserved */
1749 0, /* tp_repr */
1750 0, /* tp_as_number */
1751 0, /* tp_as_sequence */
1752 0, /* tp_as_mapping */
1753 0, /* tp_hash */
1754 0, /* tp_call */
1755 0, /* tp_str */
1756 PyObject_GenericGetAttr, /* tp_getattro */
1757 0, /* tp_setattro */
1758 0, /* tp_as_buffer */
1759 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1760 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1761 0, /* tp_doc */
1762 (traverseproc)mbstreamwriter_traverse, /* tp_traverse */
1763 0, /* tp_clear */
1764 0, /* tp_richcompare */
1765 0, /* tp_weaklistoffset */
1766 0, /* tp_iter */
1767 0, /* tp_iterext */
1768 mbstreamwriter_methods, /* tp_methods */
1769 mbstreamwriter_members, /* tp_members */
1770 codecctx_getsets, /* tp_getset */
1771 0, /* tp_base */
1772 0, /* tp_dict */
1773 0, /* tp_descr_get */
1774 0, /* tp_descr_set */
1775 0, /* tp_dictoffset */
1776 mbstreamwriter_init, /* tp_init */
1777 0, /* tp_alloc */
1778 mbstreamwriter_new, /* tp_new */
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001779};
1780
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001781
1782/**
1783 * Exposed factory function
1784 */
1785
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001786static PyObject *
1787__create_codec(PyObject *ignore, PyObject *arg)
1788{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001789 MultibyteCodecObject *self;
1790 MultibyteCodec *codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001791
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001792 if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
1793 PyErr_SetString(PyExc_ValueError, "argument type invalid");
1794 return NULL;
1795 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001796
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001797 codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
1798 if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
1799 return NULL;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001800
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001801 self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
1802 if (self == NULL)
1803 return NULL;
1804 self->codec = codec;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001805
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001806 return (PyObject *)self;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001807}
1808
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001809static struct PyMethodDef __methods[] = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001810 {"__create_codec", (PyCFunction)__create_codec, METH_O},
1811 {NULL, NULL},
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001812};
1813
Martin v. Löwis1a214512008-06-11 05:26:20 +00001814
1815static struct PyModuleDef _multibytecodecmodule = {
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001816 PyModuleDef_HEAD_INIT,
1817 "_multibytecodec",
1818 NULL,
1819 -1,
1820 __methods,
1821 NULL,
1822 NULL,
1823 NULL,
1824 NULL
Martin v. Löwis1a214512008-06-11 05:26:20 +00001825};
1826
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001827PyMODINIT_FUNC
Martin v. Löwis1a214512008-06-11 05:26:20 +00001828PyInit__multibytecodec(void)
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001829{
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001830 int i;
1831 PyObject *m;
1832 PyTypeObject *typelist[] = {
1833 &MultibyteIncrementalEncoder_Type,
1834 &MultibyteIncrementalDecoder_Type,
1835 &MultibyteStreamReader_Type,
1836 &MultibyteStreamWriter_Type,
1837 NULL
1838 };
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001839
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001840 if (PyType_Ready(&MultibyteCodec_Type) < 0)
1841 return NULL;
Thomas Wouters49fd7fa2006-04-21 10:40:58 +00001842
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001843 m = PyModule_Create(&_multibytecodecmodule);
1844 if (m == NULL)
1845 return NULL;
Neal Norwitz058bde12005-09-21 06:44:25 +00001846
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001847 for (i = 0; typelist[i] != NULL; i++) {
1848 if (PyType_Ready(typelist[i]) < 0)
1849 return NULL;
1850 Py_INCREF(typelist[i]);
1851 PyModule_AddObject(m, typelist[i]->tp_name,
1852 (PyObject *)typelist[i]);
1853 }
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001854
Antoine Pitrouf95a1b32010-05-09 15:52:27 +00001855 if (PyErr_Occurred()) {
1856 Py_FatalError("can't initialize the _multibytecodec module");
1857 Py_DECREF(m);
1858 m = NULL;
1859 }
1860 return m;
Hye-Shik Chang3e2a3062004-01-17 14:29:29 +00001861}