1 /*
2 * multibytecodec.c: Common Multibyte Codec Implementation
3 *
4 * Written by Hye-Shik Chang <perky@FreeBSD.org>
5 */
6
7 #define PY_SSIZE_T_CLEAN
8 #include "Python.h"
9 #include "structmember.h"
10 #include "multibytecodec.h"
11
12 typedef struct {
13 const Py_UNICODE *inbuf, *inbuf_top, *inbuf_end;
14 unsigned char *outbuf, *outbuf_end;
15 PyObject *excobj, *outobj;
16 } MultibyteEncodeBuffer;
17
18 typedef struct {
19 const unsigned char *inbuf, *inbuf_top, *inbuf_end;
20 Py_UNICODE *outbuf, *outbuf_end;
21 PyObject *excobj, *outobj;
22 } MultibyteDecodeBuffer;
23
24 PyDoc_STRVAR(MultibyteCodec_Encode__doc__,
25 "I.encode(unicode[, errors]) -> (string, length consumed)\n\
26 \n\
27 Return an encoded string version of `unicode'. errors may be given to\n\
28 set a different error handling scheme. Default is 'strict' meaning that\n\
29 encoding errors raise a UnicodeEncodeError. Other possible values are\n\
30 'ignore', 'replace' and 'xmlcharrefreplace' as well as any other name\n\
31 registered with codecs.register_error that can handle UnicodeEncodeErrors.");
32
33 PyDoc_STRVAR(MultibyteCodec_Decode__doc__,
34 "I.decode(string[, errors]) -> (unicodeobject, length consumed)\n\
35 \n\
36 Decodes `string' using I, an MultibyteCodec instance. errors may be given\n\
37 to set a different error handling scheme. Default is 'strict' meaning\n\
38 that encoding errors raise a UnicodeDecodeError. Other possible values\n\
39 are 'ignore' and 'replace' as well as any other name registered with\n\
40 codecs.register_error that is able to handle UnicodeDecodeErrors.");
41
42 static char *codeckwarglist[] = {"input", "errors", NULL};
43 static char *incnewkwarglist[] = {"errors", NULL};
44 static char *incrementalkwarglist[] = {"input", "final", NULL};
45 static char *streamkwarglist[] = {"stream", "errors", NULL};
46
47 static PyObject *multibytecodec_encode(MultibyteCodec *,
48 MultibyteCodec_State *, const Py_UNICODE **, Py_ssize_t,
49 PyObject *, int);
50
51 #define MBENC_RESET MBENC_MAX<<1 /* reset after an encoding session */
52
53 static PyObject *
make_tuple(PyObject * object,Py_ssize_t len)54 make_tuple(PyObject *object, Py_ssize_t len)
55 {
56 PyObject *v, *w;
57
58 if (object == NULL)
59 return NULL;
60
61 v = PyTuple_New(2);
62 if (v == NULL) {
63 Py_DECREF(object);
64 return NULL;
65 }
66 PyTuple_SET_ITEM(v, 0, object);
67
68 w = PyInt_FromSsize_t(len);
69 if (w == NULL) {
70 Py_DECREF(v);
71 return NULL;
72 }
73 PyTuple_SET_ITEM(v, 1, w);
74
75 return v;
76 }
77
78 static PyObject *
internal_error_callback(const char * errors)79 internal_error_callback(const char *errors)
80 {
81 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);
89 }
90
91 static PyObject *
call_error_callback(PyObject * errors,PyObject * exc)92 call_error_callback(PyObject *errors, PyObject *exc)
93 {
94 PyObject *args, *cb, *r;
95
96 assert(PyString_Check(errors));
97 cb = PyCodec_LookupError(PyString_AS_STRING(errors));
98 if (cb == NULL)
99 return NULL;
100
101 args = PyTuple_New(1);
102 if (args == NULL) {
103 Py_DECREF(cb);
104 return NULL;
105 }
106
107 PyTuple_SET_ITEM(args, 0, exc);
108 Py_INCREF(exc);
109
110 r = PyObject_CallObject(cb, args);
111 Py_DECREF(args);
112 Py_DECREF(cb);
113 return r;
114 }
115
116 static PyObject *
codecctx_errors_get(MultibyteStatefulCodecContext * self)117 codecctx_errors_get(MultibyteStatefulCodecContext *self)
118 {
119 const char *errors;
120
121 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 }
131
132 return PyString_FromString(errors);
133 }
134
135 static int
codecctx_errors_set(MultibyteStatefulCodecContext * self,PyObject * value,void * closure)136 codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value,
137 void *closure)
138 {
139 PyObject *cb;
140
141 if (!PyString_Check(value)) {
142 PyErr_SetString(PyExc_TypeError, "errors must be a string");
143 return -1;
144 }
145
146 cb = internal_error_callback(PyString_AS_STRING(value));
147 if (cb == NULL)
148 return -1;
149
150 ERROR_DECREF(self->errors);
151 self->errors = cb;
152 return 0;
153 }
154
155 /* This getset handlers list is used by all the stateful codec objects */
156 static PyGetSetDef codecctx_getsets[] = {
157 {"errors", (getter)codecctx_errors_get,
158 (setter)codecctx_errors_set,
159 PyDoc_STR("how to treat errors")},
160 {NULL,}
161 };
162
163 static int
expand_encodebuffer(MultibyteEncodeBuffer * buf,Py_ssize_t esize)164 expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize)
165 {
166 Py_ssize_t orgpos, orgsize, incsize;
167
168 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);
172
173 if (orgsize > PY_SSIZE_T_MAX - incsize) {
174 PyErr_NoMemory();
175 return -1;
176 }
177
178 if (_PyString_Resize(&buf->outobj, orgsize + incsize) == -1)
179 return -1;
180
181 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);
184
185 return 0;
186 }
187 #define REQUIRE_ENCODEBUFFER(buf, s) do { \
188 if ((s) < 0 || (s) > (buf)->outbuf_end - (buf)->outbuf) \
189 if (expand_encodebuffer(buf, s) == -1) \
190 goto errorexit; \
191 } while(0)
192
193 static int
expand_decodebuffer(MultibyteDecodeBuffer * buf,Py_ssize_t esize)194 expand_decodebuffer(MultibyteDecodeBuffer *buf, Py_ssize_t esize)
195 {
196 Py_ssize_t orgpos, orgsize;
197
198 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;
203
204 buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj) + orgpos;
205 buf->outbuf_end = PyUnicode_AS_UNICODE(buf->outobj)
206 + PyUnicode_GET_SIZE(buf->outobj);
207
208 return 0;
209 }
210 #define REQUIRE_DECODEBUFFER(buf, s) do { \
211 if ((s) < 0 || (s) > (buf)->outbuf_end - (buf)->outbuf) \
212 if (expand_decodebuffer(buf, s) == -1) \
213 goto errorexit; \
214 } while(0)
215
216
217 /**
218 * MultibyteCodec object
219 */
220
221 static int
multibytecodec_encerror(MultibyteCodec * codec,MultibyteCodec_State * state,MultibyteEncodeBuffer * buf,PyObject * errors,Py_ssize_t e)222 multibytecodec_encerror(MultibyteCodec *codec,
223 MultibyteCodec_State *state,
224 MultibyteEncodeBuffer *buf,
225 PyObject *errors, Py_ssize_t e)
226 {
227 PyObject *retobj = NULL, *retstr = NULL, *tobj;
228 Py_ssize_t retstrsize, newpos;
229 Py_ssize_t esize, start, end;
230 const char *reason;
231
232 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 }
255
256 if (errors == ERROR_REPLACE) {
257 const Py_UNICODE replchar = '?', *inbuf = &replchar;
258 Py_ssize_t r;
259
260 for (;;) {
261 Py_ssize_t outleft;
262
263 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 }
273
274 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 }
283
284 start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
285 end = start + esize;
286
287 /* 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;
301
302 if (errors == ERROR_STRICT) {
303 PyCodec_StrictErrors(buf->excobj);
304 goto errorexit;
305 }
306
307 retobj = call_error_callback(errors, buf->excobj);
308 if (retobj == NULL)
309 goto errorexit;
310
311 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 }
320
321 {
322 const Py_UNICODE *uraw = PyUnicode_AS_UNICODE(tobj);
323
324 retstr = multibytecodec_encode(codec, state, &uraw,
325 PyUnicode_GET_SIZE(tobj), ERROR_STRICT,
326 MBENC_FLUSH);
327 if (retstr == NULL)
328 goto errorexit;
329 }
330
331 retstrsize = PyString_GET_SIZE(retstr);
332 if (retstrsize > 0) {
333 REQUIRE_ENCODEBUFFER(buf, retstrsize);
334 memcpy(buf->outbuf, PyString_AS_STRING(retstr), retstrsize);
335 buf->outbuf += retstrsize;
336 }
337
338 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;
349
350 Py_DECREF(retobj);
351 Py_DECREF(retstr);
352 return 0;
353
354 errorexit:
355 Py_XDECREF(retobj);
356 Py_XDECREF(retstr);
357 return -1;
358 }
359
360 static int
multibytecodec_decerror(MultibyteCodec * codec,MultibyteCodec_State * state,MultibyteDecodeBuffer * buf,PyObject * errors,Py_ssize_t e)361 multibytecodec_decerror(MultibyteCodec *codec,
362 MultibyteCodec_State *state,
363 MultibyteDecodeBuffer *buf,
364 PyObject *errors, Py_ssize_t e)
365 {
366 PyObject *retobj = NULL, *retuni = NULL;
367 Py_ssize_t retunisize, newpos;
368 const char *reason;
369 Py_ssize_t esize, start, end;
370
371 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 }
394
395 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 }
403
404 start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
405 end = start + esize;
406
407 /* 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;
421
422 if (errors == ERROR_STRICT) {
423 PyCodec_StrictErrors(buf->excobj);
424 goto errorexit;
425 }
426
427 retobj = call_error_callback(errors, buf->excobj);
428 if (retobj == NULL)
429 goto errorexit;
430
431 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 }
440
441 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 }
448
449 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;
462
463 errorexit:
464 Py_XDECREF(retobj);
465 return -1;
466 }
467
468 static PyObject *
multibytecodec_encode(MultibyteCodec * codec,MultibyteCodec_State * state,const Py_UNICODE ** data,Py_ssize_t datalen,PyObject * errors,int flags)469 multibytecodec_encode(MultibyteCodec *codec,
470 MultibyteCodec_State *state,
471 const Py_UNICODE **data, Py_ssize_t datalen,
472 PyObject *errors, int flags)
473 {
474 MultibyteEncodeBuffer buf;
475 Py_ssize_t finalsize, r = 0;
476
477 if (datalen == 0 && !(flags & MBENC_RESET))
478 return PyString_FromString("");
479
480 buf.excobj = NULL;
481 buf.inbuf = buf.inbuf_top = *data;
482 buf.inbuf_end = buf.inbuf_top + datalen;
483
484 if (datalen > (PY_SSIZE_T_MAX - 16) / 2) {
485 PyErr_NoMemory();
486 goto errorexit;
487 }
488
489 buf.outobj = PyString_FromStringAndSize(NULL, datalen * 2 + 16);
490 if (buf.outobj == NULL)
491 goto errorexit;
492 buf.outbuf = (unsigned char *)PyString_AS_STRING(buf.outobj);
493 buf.outbuf_end = buf.outbuf + PyString_GET_SIZE(buf.outobj);
494
495 while (buf.inbuf < buf.inbuf_end) {
496 Py_ssize_t inleft, outleft;
497
498 /* we don't reuse inleft and outleft here.
499 * error callbacks can relocate the cursor anywhere on buffer*/
500 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
501 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
502 r = codec->encode(state, codec->config, &buf.inbuf, inleft,
503 &buf.outbuf, outleft, flags);
504 if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH)))
505 break;
506 else if (multibytecodec_encerror(codec, state, &buf, errors,r))
507 goto errorexit;
508 else if (r == MBERR_TOOFEW)
509 break;
510 }
511
512 if (codec->encreset != NULL && (flags & MBENC_RESET))
513 for (;;) {
514 Py_ssize_t outleft;
515
516 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
517 r = codec->encreset(state, codec->config, &buf.outbuf,
518 outleft);
519 if (r == 0)
520 break;
521 else if (multibytecodec_encerror(codec, state,
522 &buf, errors, r))
523 goto errorexit;
524 }
525
526 finalsize = (Py_ssize_t)((char *)buf.outbuf -
527 PyString_AS_STRING(buf.outobj));
528
529 if (finalsize != PyString_GET_SIZE(buf.outobj))
530 if (_PyString_Resize(&buf.outobj, finalsize) == -1)
531 goto errorexit;
532
533 *data = buf.inbuf;
534 Py_XDECREF(buf.excobj);
535 return buf.outobj;
536
537 errorexit:
538 Py_XDECREF(buf.excobj);
539 Py_XDECREF(buf.outobj);
540 return NULL;
541 }
542
543 static PyObject *
MultibyteCodec_Encode(MultibyteCodecObject * self,PyObject * args,PyObject * kwargs)544 MultibyteCodec_Encode(MultibyteCodecObject *self,
545 PyObject *args, PyObject *kwargs)
546 {
547 MultibyteCodec_State state;
548 Py_UNICODE *data;
549 PyObject *errorcb, *r, *arg, *ucvt;
550 const char *errors = NULL;
551 Py_ssize_t datalen;
552
553 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode",
554 codeckwarglist, &arg, &errors))
555 return NULL;
556
557 if (PyUnicode_Check(arg))
558 ucvt = NULL;
559 else {
560 arg = ucvt = PyObject_Unicode(arg);
561 if (arg == NULL)
562 return NULL;
563 else if (!PyUnicode_Check(arg)) {
564 PyErr_SetString(PyExc_TypeError,
565 "couldn't convert the object to unicode.");
566 Py_DECREF(ucvt);
567 return NULL;
568 }
569 }
570
571 data = PyUnicode_AS_UNICODE(arg);
572 datalen = PyUnicode_GET_SIZE(arg);
573
574 errorcb = internal_error_callback(errors);
575 if (errorcb == NULL) {
576 Py_XDECREF(ucvt);
577 return NULL;
578 }
579
580 if (self->codec->encinit != NULL &&
581 self->codec->encinit(&state, self->codec->config) != 0)
582 goto errorexit;
583 r = multibytecodec_encode(self->codec, &state,
584 (const Py_UNICODE **)&data, datalen, errorcb,
585 MBENC_FLUSH | MBENC_RESET);
586 if (r == NULL)
587 goto errorexit;
588
589 ERROR_DECREF(errorcb);
590 Py_XDECREF(ucvt);
591 return make_tuple(r, datalen);
592
593 errorexit:
594 ERROR_DECREF(errorcb);
595 Py_XDECREF(ucvt);
596 return NULL;
597 }
598
599 static PyObject *
MultibyteCodec_Decode(MultibyteCodecObject * self,PyObject * args,PyObject * kwargs)600 MultibyteCodec_Decode(MultibyteCodecObject *self,
601 PyObject *args, PyObject *kwargs)
602 {
603 MultibyteCodec_State state;
604 MultibyteDecodeBuffer buf;
605 PyObject *errorcb;
606 Py_buffer pdata;
607 const char *data, *errors = NULL;
608 Py_ssize_t datalen, finalsize;
609
610 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "s*|z:decode",
611 codeckwarglist, &pdata, &errors))
612 return NULL;
613 data = pdata.buf;
614 datalen = pdata.len;
615
616 errorcb = internal_error_callback(errors);
617 if (errorcb == NULL) {
618 PyBuffer_Release(&pdata);
619 return NULL;
620 }
621
622 if (datalen == 0) {
623 PyBuffer_Release(&pdata);
624 ERROR_DECREF(errorcb);
625 return make_tuple(PyUnicode_FromUnicode(NULL, 0), 0);
626 }
627
628 buf.excobj = NULL;
629 buf.inbuf = buf.inbuf_top = (unsigned char *)data;
630 buf.inbuf_end = buf.inbuf_top + datalen;
631 buf.outobj = PyUnicode_FromUnicode(NULL, datalen);
632 if (buf.outobj == NULL)
633 goto errorexit;
634 buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj);
635 buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj);
636
637 if (self->codec->decinit != NULL &&
638 self->codec->decinit(&state, self->codec->config) != 0)
639 goto errorexit;
640
641 while (buf.inbuf < buf.inbuf_end) {
642 Py_ssize_t inleft, outleft, r;
643
644 inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
645 outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
646
647 r = self->codec->decode(&state, self->codec->config,
648 &buf.inbuf, inleft, &buf.outbuf, outleft);
649 if (r == 0)
650 break;
651 else if (multibytecodec_decerror(self->codec, &state,
652 &buf, errorcb, r))
653 goto errorexit;
654 }
655
656 finalsize = (Py_ssize_t)(buf.outbuf -
657 PyUnicode_AS_UNICODE(buf.outobj));
658
659 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
660 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
661 goto errorexit;
662
663 PyBuffer_Release(&pdata);
664 Py_XDECREF(buf.excobj);
665 ERROR_DECREF(errorcb);
666 return make_tuple(buf.outobj, datalen);
667
668 errorexit:
669 PyBuffer_Release(&pdata);
670 ERROR_DECREF(errorcb);
671 Py_XDECREF(buf.excobj);
672 Py_XDECREF(buf.outobj);
673
674 return NULL;
675 }
676
677 static struct PyMethodDef multibytecodec_methods[] = {
678 {"encode", (PyCFunction)MultibyteCodec_Encode,
679 METH_VARARGS | METH_KEYWORDS,
680 MultibyteCodec_Encode__doc__},
681 {"decode", (PyCFunction)MultibyteCodec_Decode,
682 METH_VARARGS | METH_KEYWORDS,
683 MultibyteCodec_Decode__doc__},
684 {NULL, NULL},
685 };
686
687 static void
multibytecodec_dealloc(MultibyteCodecObject * self)688 multibytecodec_dealloc(MultibyteCodecObject *self)
689 {
690 PyObject_Del(self);
691 }
692
693 static PyTypeObject MultibyteCodec_Type = {
694 PyVarObject_HEAD_INIT(NULL, 0)
695 "MultibyteCodec", /* tp_name */
696 sizeof(MultibyteCodecObject), /* tp_basicsize */
697 0, /* tp_itemsize */
698 /* methods */
699 (destructor)multibytecodec_dealloc, /* tp_dealloc */
700 0, /* tp_print */
701 0, /* tp_getattr */
702 0, /* tp_setattr */
703 0, /* tp_compare */
704 0, /* tp_repr */
705 0, /* tp_as_number */
706 0, /* tp_as_sequence */
707 0, /* tp_as_mapping */
708 0, /* tp_hash */
709 0, /* tp_call */
710 0, /* tp_str */
711 PyObject_GenericGetAttr, /* tp_getattro */
712 0, /* tp_setattro */
713 0, /* tp_as_buffer */
714 Py_TPFLAGS_DEFAULT, /* tp_flags */
715 0, /* tp_doc */
716 0, /* tp_traverse */
717 0, /* tp_clear */
718 0, /* tp_richcompare */
719 0, /* tp_weaklistoffset */
720 0, /* tp_iter */
721 0, /* tp_iterext */
722 multibytecodec_methods, /* tp_methods */
723 };
724
725
726 /**
727 * Utility functions for stateful codec mechanism
728 */
729
730 #define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o))
731 #define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o))
732
733 static PyObject *
encoder_encode_stateful(MultibyteStatefulEncoderContext * ctx,PyObject * unistr,int final)734 encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
735 PyObject *unistr, int final)
736 {
737 PyObject *ucvt, *r = NULL;
738 Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL;
739 Py_ssize_t datalen, origpending;
740
741 if (PyUnicode_Check(unistr))
742 ucvt = NULL;
743 else {
744 unistr = ucvt = PyObject_Unicode(unistr);
745 if (unistr == NULL)
746 return NULL;
747 else if (!PyUnicode_Check(unistr)) {
748 PyErr_SetString(PyExc_TypeError,
749 "couldn't convert the object to unicode.");
750 Py_DECREF(ucvt);
751 return NULL;
752 }
753 }
754
755 datalen = PyUnicode_GET_SIZE(unistr);
756 origpending = ctx->pendingsize;
757
758 if (origpending > 0) {
759 if (datalen > PY_SSIZE_T_MAX - ctx->pendingsize) {
760 PyErr_NoMemory();
761 /* inbuf_tmp == NULL */
762 goto errorexit;
763 }
764 inbuf_tmp = PyMem_New(Py_UNICODE, datalen + ctx->pendingsize);
765 if (inbuf_tmp == NULL)
766 goto errorexit;
767 memcpy(inbuf_tmp, ctx->pending,
768 Py_UNICODE_SIZE * ctx->pendingsize);
769 memcpy(inbuf_tmp + ctx->pendingsize,
770 PyUnicode_AS_UNICODE(unistr),
771 Py_UNICODE_SIZE * datalen);
772 datalen += ctx->pendingsize;
773 ctx->pendingsize = 0;
774 inbuf = inbuf_tmp;
775 }
776 else
777 inbuf = (Py_UNICODE *)PyUnicode_AS_UNICODE(unistr);
778
779 inbuf_end = inbuf + datalen;
780
781 r = multibytecodec_encode(ctx->codec, &ctx->state,
782 (const Py_UNICODE **)&inbuf, datalen,
783 ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0);
784 if (r == NULL) {
785 /* recover the original pending buffer */
786 if (origpending > 0)
787 memcpy(ctx->pending, inbuf_tmp,
788 Py_UNICODE_SIZE * origpending);
789 ctx->pendingsize = origpending;
790 goto errorexit;
791 }
792
793 if (inbuf < inbuf_end) {
794 ctx->pendingsize = (Py_ssize_t)(inbuf_end - inbuf);
795 if (ctx->pendingsize > MAXENCPENDING) {
796 /* normal codecs can't reach here */
797 ctx->pendingsize = 0;
798 PyErr_SetString(PyExc_UnicodeError,
799 "pending buffer overflow");
800 goto errorexit;
801 }
802 memcpy(ctx->pending, inbuf,
803 ctx->pendingsize * Py_UNICODE_SIZE);
804 }
805
806 if (inbuf_tmp != NULL)
807 PyMem_Del(inbuf_tmp);
808 Py_XDECREF(ucvt);
809 return r;
810
811 errorexit:
812 if (inbuf_tmp != NULL)
813 PyMem_Del(inbuf_tmp);
814 Py_XDECREF(r);
815 Py_XDECREF(ucvt);
816 return NULL;
817 }
818
819 static int
decoder_append_pending(MultibyteStatefulDecoderContext * ctx,MultibyteDecodeBuffer * buf)820 decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
821 MultibyteDecodeBuffer *buf)
822 {
823 Py_ssize_t npendings;
824
825 npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
826 if (npendings + ctx->pendingsize > MAXDECPENDING ||
827 npendings > PY_SSIZE_T_MAX - ctx->pendingsize) {
828 PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow");
829 return -1;
830 }
831 memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings);
832 ctx->pendingsize += npendings;
833 return 0;
834 }
835
836 static int
decoder_prepare_buffer(MultibyteDecodeBuffer * buf,const char * data,Py_ssize_t size)837 decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
838 Py_ssize_t size)
839 {
840 buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
841 buf->inbuf_end = buf->inbuf_top + size;
842 if (buf->outobj == NULL) { /* only if outobj is not allocated yet */
843 buf->outobj = PyUnicode_FromUnicode(NULL, size);
844 if (buf->outobj == NULL)
845 return -1;
846 buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj);
847 buf->outbuf_end = buf->outbuf +
848 PyUnicode_GET_SIZE(buf->outobj);
849 }
850
851 return 0;
852 }
853
854 static int
decoder_feed_buffer(MultibyteStatefulDecoderContext * ctx,MultibyteDecodeBuffer * buf)855 decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
856 MultibyteDecodeBuffer *buf)
857 {
858 while (buf->inbuf < buf->inbuf_end) {
859 Py_ssize_t inleft, outleft;
860 Py_ssize_t r;
861
862 inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
863 outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
864
865 r = ctx->codec->decode(&ctx->state, ctx->codec->config,
866 &buf->inbuf, inleft, &buf->outbuf, outleft);
867 if (r == 0 || r == MBERR_TOOFEW)
868 break;
869 else if (multibytecodec_decerror(ctx->codec, &ctx->state,
870 buf, ctx->errors, r))
871 return -1;
872 }
873 return 0;
874 }
875
876
877 /**
878 * MultibyteIncrementalEncoder object
879 */
880
881 static PyObject *
mbiencoder_encode(MultibyteIncrementalEncoderObject * self,PyObject * args,PyObject * kwargs)882 mbiencoder_encode(MultibyteIncrementalEncoderObject *self,
883 PyObject *args, PyObject *kwargs)
884 {
885 PyObject *data;
886 int final = 0;
887
888 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode",
889 incrementalkwarglist, &data, &final))
890 return NULL;
891
892 return encoder_encode_stateful(STATEFUL_ECTX(self), data, final);
893 }
894
895 static PyObject *
mbiencoder_reset(MultibyteIncrementalEncoderObject * self)896 mbiencoder_reset(MultibyteIncrementalEncoderObject *self)
897 {
898 if (self->codec->decreset != NULL &&
899 self->codec->decreset(&self->state, self->codec->config) != 0)
900 return NULL;
901 self->pendingsize = 0;
902
903 Py_RETURN_NONE;
904 }
905
906 static struct PyMethodDef mbiencoder_methods[] = {
907 {"encode", (PyCFunction)mbiencoder_encode,
908 METH_VARARGS | METH_KEYWORDS, NULL},
909 {"reset", (PyCFunction)mbiencoder_reset,
910 METH_NOARGS, NULL},
911 {NULL, NULL},
912 };
913
914 static PyObject *
mbiencoder_new(PyTypeObject * type,PyObject * args,PyObject * kwds)915 mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
916 {
917 MultibyteIncrementalEncoderObject *self;
918 PyObject *codec = NULL;
919 char *errors = NULL;
920
921 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
922 incnewkwarglist, &errors))
923 return NULL;
924
925 self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
926 if (self == NULL)
927 return NULL;
928
929 codec = PyObject_GetAttrString((PyObject *)type, "codec");
930 if (codec == NULL)
931 goto errorexit;
932 if (!MultibyteCodec_Check(codec)) {
933 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
934 goto errorexit;
935 }
936
937 self->codec = ((MultibyteCodecObject *)codec)->codec;
938 self->pendingsize = 0;
939 self->errors = internal_error_callback(errors);
940 if (self->errors == NULL)
941 goto errorexit;
942 if (self->codec->encinit != NULL &&
943 self->codec->encinit(&self->state, self->codec->config) != 0)
944 goto errorexit;
945
946 Py_DECREF(codec);
947 return (PyObject *)self;
948
949 errorexit:
950 Py_XDECREF(self);
951 Py_XDECREF(codec);
952 return NULL;
953 }
954
955 static int
mbiencoder_init(PyObject * self,PyObject * args,PyObject * kwds)956 mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
957 {
958 return 0;
959 }
960
961 static int
mbiencoder_traverse(MultibyteIncrementalEncoderObject * self,visitproc visit,void * arg)962 mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
963 visitproc visit, void *arg)
964 {
965 if (ERROR_ISCUSTOM(self->errors))
966 Py_VISIT(self->errors);
967 return 0;
968 }
969
970 static void
mbiencoder_dealloc(MultibyteIncrementalEncoderObject * self)971 mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
972 {
973 PyObject_GC_UnTrack(self);
974 ERROR_DECREF(self->errors);
975 Py_TYPE(self)->tp_free(self);
976 }
977
978 static PyTypeObject MultibyteIncrementalEncoder_Type = {
979 PyVarObject_HEAD_INIT(NULL, 0)
980 "MultibyteIncrementalEncoder", /* tp_name */
981 sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
982 0, /* tp_itemsize */
983 /* methods */
984 (destructor)mbiencoder_dealloc, /* tp_dealloc */
985 0, /* tp_print */
986 0, /* tp_getattr */
987 0, /* tp_setattr */
988 0, /* tp_compare */
989 0, /* tp_repr */
990 0, /* tp_as_number */
991 0, /* tp_as_sequence */
992 0, /* tp_as_mapping */
993 0, /* tp_hash */
994 0, /* tp_call */
995 0, /* tp_str */
996 PyObject_GenericGetAttr, /* tp_getattro */
997 0, /* tp_setattro */
998 0, /* tp_as_buffer */
999 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1000 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1001 0, /* tp_doc */
1002 (traverseproc)mbiencoder_traverse, /* tp_traverse */
1003 0, /* tp_clear */
1004 0, /* tp_richcompare */
1005 0, /* tp_weaklistoffset */
1006 0, /* tp_iter */
1007 0, /* tp_iterext */
1008 mbiencoder_methods, /* tp_methods */
1009 0, /* tp_members */
1010 codecctx_getsets, /* tp_getset */
1011 0, /* tp_base */
1012 0, /* tp_dict */
1013 0, /* tp_descr_get */
1014 0, /* tp_descr_set */
1015 0, /* tp_dictoffset */
1016 mbiencoder_init, /* tp_init */
1017 0, /* tp_alloc */
1018 mbiencoder_new, /* tp_new */
1019 };
1020
1021
1022 /**
1023 * MultibyteIncrementalDecoder object
1024 */
1025
1026 static PyObject *
mbidecoder_decode(MultibyteIncrementalDecoderObject * self,PyObject * args,PyObject * kwargs)1027 mbidecoder_decode(MultibyteIncrementalDecoderObject *self,
1028 PyObject *args, PyObject *kwargs)
1029 {
1030 MultibyteDecodeBuffer buf;
1031 char *data, *wdata = NULL;
1032 Py_buffer pdata;
1033 Py_ssize_t wsize, finalsize = 0, size, origpending;
1034 int final = 0;
1035
1036 if (!PyArg_ParseTupleAndKeywords(args, kwargs, "s*|i:decode",
1037 incrementalkwarglist, &pdata, &final))
1038 return NULL;
1039 data = pdata.buf;
1040 size = pdata.len;
1041
1042 buf.outobj = buf.excobj = NULL;
1043 origpending = self->pendingsize;
1044
1045 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);
1056 if (wdata == NULL)
1057 goto errorexit;
1058 memcpy(wdata, self->pending, self->pendingsize);
1059 memcpy(wdata + self->pendingsize, data, size);
1060 self->pendingsize = 0;
1061 }
1062
1063 if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
1064 goto errorexit;
1065
1066 if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
1067 goto errorexit;
1068
1069 if (final && buf.inbuf < buf.inbuf_end) {
1070 if (multibytecodec_decerror(self->codec, &self->state,
1071 &buf, self->errors, MBERR_TOOFEW)) {
1072 /* recover the original pending buffer */
1073 memcpy(self->pending, wdata, origpending);
1074 self->pendingsize = origpending;
1075 goto errorexit;
1076 }
1077 }
1078
1079 if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
1080 if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
1081 goto errorexit;
1082 }
1083
1084 finalsize = (Py_ssize_t)(buf.outbuf - PyUnicode_AS_UNICODE(buf.outobj));
1085 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1086 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1087 goto errorexit;
1088
1089 PyBuffer_Release(&pdata);
1090 if (wdata != data)
1091 PyMem_Del(wdata);
1092 Py_XDECREF(buf.excobj);
1093 return buf.outobj;
1094
1095 errorexit:
1096 PyBuffer_Release(&pdata);
1097 if (wdata != NULL && wdata != data)
1098 PyMem_Del(wdata);
1099 Py_XDECREF(buf.excobj);
1100 Py_XDECREF(buf.outobj);
1101 return NULL;
1102 }
1103
1104 static PyObject *
mbidecoder_reset(MultibyteIncrementalDecoderObject * self)1105 mbidecoder_reset(MultibyteIncrementalDecoderObject *self)
1106 {
1107 if (self->codec->decreset != NULL &&
1108 self->codec->decreset(&self->state, self->codec->config) != 0)
1109 return NULL;
1110 self->pendingsize = 0;
1111
1112 Py_RETURN_NONE;
1113 }
1114
1115 static struct PyMethodDef mbidecoder_methods[] = {
1116 {"decode", (PyCFunction)mbidecoder_decode,
1117 METH_VARARGS | METH_KEYWORDS, NULL},
1118 {"reset", (PyCFunction)mbidecoder_reset,
1119 METH_NOARGS, NULL},
1120 {NULL, NULL},
1121 };
1122
1123 static PyObject *
mbidecoder_new(PyTypeObject * type,PyObject * args,PyObject * kwds)1124 mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1125 {
1126 MultibyteIncrementalDecoderObject *self;
1127 PyObject *codec = NULL;
1128 char *errors = NULL;
1129
1130 if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
1131 incnewkwarglist, &errors))
1132 return NULL;
1133
1134 self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
1135 if (self == NULL)
1136 return NULL;
1137
1138 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1139 if (codec == NULL)
1140 goto errorexit;
1141 if (!MultibyteCodec_Check(codec)) {
1142 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1143 goto errorexit;
1144 }
1145
1146 self->codec = ((MultibyteCodecObject *)codec)->codec;
1147 self->pendingsize = 0;
1148 self->errors = internal_error_callback(errors);
1149 if (self->errors == NULL)
1150 goto errorexit;
1151 if (self->codec->decinit != NULL &&
1152 self->codec->decinit(&self->state, self->codec->config) != 0)
1153 goto errorexit;
1154
1155 Py_DECREF(codec);
1156 return (PyObject *)self;
1157
1158 errorexit:
1159 Py_XDECREF(self);
1160 Py_XDECREF(codec);
1161 return NULL;
1162 }
1163
1164 static int
mbidecoder_init(PyObject * self,PyObject * args,PyObject * kwds)1165 mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
1166 {
1167 return 0;
1168 }
1169
1170 static int
mbidecoder_traverse(MultibyteIncrementalDecoderObject * self,visitproc visit,void * arg)1171 mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
1172 visitproc visit, void *arg)
1173 {
1174 if (ERROR_ISCUSTOM(self->errors))
1175 Py_VISIT(self->errors);
1176 return 0;
1177 }
1178
1179 static void
mbidecoder_dealloc(MultibyteIncrementalDecoderObject * self)1180 mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
1181 {
1182 PyObject_GC_UnTrack(self);
1183 ERROR_DECREF(self->errors);
1184 Py_TYPE(self)->tp_free(self);
1185 }
1186
1187 static PyTypeObject MultibyteIncrementalDecoder_Type = {
1188 PyVarObject_HEAD_INIT(NULL, 0)
1189 "MultibyteIncrementalDecoder", /* tp_name */
1190 sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */
1191 0, /* tp_itemsize */
1192 /* methods */
1193 (destructor)mbidecoder_dealloc, /* tp_dealloc */
1194 0, /* tp_print */
1195 0, /* tp_getattr */
1196 0, /* tp_setattr */
1197 0, /* tp_compare */
1198 0, /* tp_repr */
1199 0, /* tp_as_number */
1200 0, /* tp_as_sequence */
1201 0, /* tp_as_mapping */
1202 0, /* tp_hash */
1203 0, /* tp_call */
1204 0, /* tp_str */
1205 PyObject_GenericGetAttr, /* tp_getattro */
1206 0, /* tp_setattro */
1207 0, /* tp_as_buffer */
1208 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1209 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1210 0, /* tp_doc */
1211 (traverseproc)mbidecoder_traverse, /* tp_traverse */
1212 0, /* tp_clear */
1213 0, /* tp_richcompare */
1214 0, /* tp_weaklistoffset */
1215 0, /* tp_iter */
1216 0, /* tp_iterext */
1217 mbidecoder_methods, /* tp_methods */
1218 0, /* tp_members */
1219 codecctx_getsets, /* tp_getset */
1220 0, /* tp_base */
1221 0, /* tp_dict */
1222 0, /* tp_descr_get */
1223 0, /* tp_descr_set */
1224 0, /* tp_dictoffset */
1225 mbidecoder_init, /* tp_init */
1226 0, /* tp_alloc */
1227 mbidecoder_new, /* tp_new */
1228 };
1229
1230
1231 /**
1232 * MultibyteStreamReader object
1233 */
1234
1235 static PyObject *
mbstreamreader_iread(MultibyteStreamReaderObject * self,const char * method,Py_ssize_t sizehint)1236 mbstreamreader_iread(MultibyteStreamReaderObject *self,
1237 const char *method, Py_ssize_t sizehint)
1238 {
1239 MultibyteDecodeBuffer buf;
1240 PyObject *cres;
1241 Py_ssize_t rsize, finalsize = 0;
1242
1243 if (sizehint == 0)
1244 return PyUnicode_FromUnicode(NULL, 0);
1245
1246 buf.outobj = buf.excobj = NULL;
1247 cres = NULL;
1248
1249 for (;;) {
1250 int endoffile;
1251
1252 if (sizehint < 0)
1253 cres = PyObject_CallMethod(self->stream,
1254 (char *)method, NULL);
1255 else
1256 cres = PyObject_CallMethod(self->stream,
1257 (char *)method, "i", sizehint);
1258 if (cres == NULL)
1259 goto errorexit;
1260
1261 if (!PyString_Check(cres)) {
1262 PyErr_SetString(PyExc_TypeError,
1263 "stream function returned a "
1264 "non-string object");
1265 goto errorexit;
1266 }
1267
1268 endoffile = (PyString_GET_SIZE(cres) == 0);
1269
1270 if (self->pendingsize > 0) {
1271 PyObject *ctr;
1272 char *ctrdata;
1273
1274 if (PyString_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
1275 PyErr_NoMemory();
1276 goto errorexit;
1277 }
1278 rsize = PyString_GET_SIZE(cres) + self->pendingsize;
1279 ctr = PyString_FromStringAndSize(NULL, rsize);
1280 if (ctr == NULL)
1281 goto errorexit;
1282 ctrdata = PyString_AS_STRING(ctr);
1283 memcpy(ctrdata, self->pending, self->pendingsize);
1284 memcpy(ctrdata + self->pendingsize,
1285 PyString_AS_STRING(cres),
1286 PyString_GET_SIZE(cres));
1287 Py_DECREF(cres);
1288 cres = ctr;
1289 self->pendingsize = 0;
1290 }
1291
1292 rsize = PyString_GET_SIZE(cres);
1293 if (decoder_prepare_buffer(&buf, PyString_AS_STRING(cres),
1294 rsize) != 0)
1295 goto errorexit;
1296
1297 if (rsize > 0 && decoder_feed_buffer(
1298 (MultibyteStatefulDecoderContext *)self, &buf))
1299 goto errorexit;
1300
1301 if (endoffile || sizehint < 0) {
1302 if (buf.inbuf < buf.inbuf_end &&
1303 multibytecodec_decerror(self->codec, &self->state,
1304 &buf, self->errors, MBERR_TOOFEW))
1305 goto errorexit;
1306 }
1307
1308 if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
1309 if (decoder_append_pending(STATEFUL_DCTX(self),
1310 &buf) != 0)
1311 goto errorexit;
1312 }
1313
1314 finalsize = (Py_ssize_t)(buf.outbuf -
1315 PyUnicode_AS_UNICODE(buf.outobj));
1316 Py_DECREF(cres);
1317 cres = NULL;
1318
1319 if (sizehint < 0 || finalsize != 0 || rsize == 0)
1320 break;
1321
1322 sizehint = 1; /* read 1 more byte and retry */
1323 }
1324
1325 if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1326 if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1327 goto errorexit;
1328
1329 Py_XDECREF(cres);
1330 Py_XDECREF(buf.excobj);
1331 return buf.outobj;
1332
1333 errorexit:
1334 Py_XDECREF(cres);
1335 Py_XDECREF(buf.excobj);
1336 Py_XDECREF(buf.outobj);
1337 return NULL;
1338 }
1339
1340 static PyObject *
mbstreamreader_read(MultibyteStreamReaderObject * self,PyObject * args)1341 mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args)
1342 {
1343 PyObject *sizeobj = NULL;
1344 Py_ssize_t size;
1345
1346 if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj))
1347 return NULL;
1348
1349 if (sizeobj == Py_None || sizeobj == NULL)
1350 size = -1;
1351 else if (PyInt_Check(sizeobj))
1352 size = PyInt_AsSsize_t(sizeobj);
1353 else {
1354 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1355 return NULL;
1356 }
1357
1358 return mbstreamreader_iread(self, "read", size);
1359 }
1360
1361 static PyObject *
mbstreamreader_readline(MultibyteStreamReaderObject * self,PyObject * args)1362 mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args)
1363 {
1364 PyObject *sizeobj = NULL;
1365 Py_ssize_t size;
1366
1367 if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj))
1368 return NULL;
1369
1370 if (sizeobj == Py_None || sizeobj == NULL)
1371 size = -1;
1372 else if (PyInt_Check(sizeobj))
1373 size = PyInt_AsSsize_t(sizeobj);
1374 else {
1375 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1376 return NULL;
1377 }
1378
1379 return mbstreamreader_iread(self, "readline", size);
1380 }
1381
1382 static PyObject *
mbstreamreader_readlines(MultibyteStreamReaderObject * self,PyObject * args)1383 mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args)
1384 {
1385 PyObject *sizehintobj = NULL, *r, *sr;
1386 Py_ssize_t sizehint;
1387
1388 if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj))
1389 return NULL;
1390
1391 if (sizehintobj == Py_None || sizehintobj == NULL)
1392 sizehint = -1;
1393 else if (PyInt_Check(sizehintobj))
1394 sizehint = PyInt_AsSsize_t(sizehintobj);
1395 else {
1396 PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1397 return NULL;
1398 }
1399
1400 r = mbstreamreader_iread(self, "read", sizehint);
1401 if (r == NULL)
1402 return NULL;
1403
1404 sr = PyUnicode_Splitlines(r, 1);
1405 Py_DECREF(r);
1406 return sr;
1407 }
1408
1409 static PyObject *
mbstreamreader_reset(MultibyteStreamReaderObject * self)1410 mbstreamreader_reset(MultibyteStreamReaderObject *self)
1411 {
1412 if (self->codec->decreset != NULL &&
1413 self->codec->decreset(&self->state, self->codec->config) != 0)
1414 return NULL;
1415 self->pendingsize = 0;
1416
1417 Py_RETURN_NONE;
1418 }
1419
1420 static struct PyMethodDef mbstreamreader_methods[] = {
1421 {"read", (PyCFunction)mbstreamreader_read,
1422 METH_VARARGS, NULL},
1423 {"readline", (PyCFunction)mbstreamreader_readline,
1424 METH_VARARGS, NULL},
1425 {"readlines", (PyCFunction)mbstreamreader_readlines,
1426 METH_VARARGS, NULL},
1427 {"reset", (PyCFunction)mbstreamreader_reset,
1428 METH_NOARGS, NULL},
1429 {NULL, NULL},
1430 };
1431
1432 static PyMemberDef mbstreamreader_members[] = {
1433 {"stream", T_OBJECT,
1434 offsetof(MultibyteStreamReaderObject, stream),
1435 READONLY, NULL},
1436 {NULL,}
1437 };
1438
1439 static PyObject *
mbstreamreader_new(PyTypeObject * type,PyObject * args,PyObject * kwds)1440 mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1441 {
1442 MultibyteStreamReaderObject *self;
1443 PyObject *stream, *codec = NULL;
1444 char *errors = NULL;
1445
1446 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
1447 streamkwarglist, &stream, &errors))
1448 return NULL;
1449
1450 self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
1451 if (self == NULL)
1452 return NULL;
1453
1454 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1455 if (codec == NULL)
1456 goto errorexit;
1457 if (!MultibyteCodec_Check(codec)) {
1458 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1459 goto errorexit;
1460 }
1461
1462 self->codec = ((MultibyteCodecObject *)codec)->codec;
1463 self->stream = stream;
1464 Py_INCREF(stream);
1465 self->pendingsize = 0;
1466 self->errors = internal_error_callback(errors);
1467 if (self->errors == NULL)
1468 goto errorexit;
1469 if (self->codec->decinit != NULL &&
1470 self->codec->decinit(&self->state, self->codec->config) != 0)
1471 goto errorexit;
1472
1473 Py_DECREF(codec);
1474 return (PyObject *)self;
1475
1476 errorexit:
1477 Py_XDECREF(self);
1478 Py_XDECREF(codec);
1479 return NULL;
1480 }
1481
1482 static int
mbstreamreader_init(PyObject * self,PyObject * args,PyObject * kwds)1483 mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
1484 {
1485 return 0;
1486 }
1487
1488 static int
mbstreamreader_traverse(MultibyteStreamReaderObject * self,visitproc visit,void * arg)1489 mbstreamreader_traverse(MultibyteStreamReaderObject *self,
1490 visitproc visit, void *arg)
1491 {
1492 if (ERROR_ISCUSTOM(self->errors))
1493 Py_VISIT(self->errors);
1494 Py_VISIT(self->stream);
1495 return 0;
1496 }
1497
1498 static void
mbstreamreader_dealloc(MultibyteStreamReaderObject * self)1499 mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
1500 {
1501 PyObject_GC_UnTrack(self);
1502 ERROR_DECREF(self->errors);
1503 Py_XDECREF(self->stream);
1504 Py_TYPE(self)->tp_free(self);
1505 }
1506
1507 static PyTypeObject MultibyteStreamReader_Type = {
1508 PyVarObject_HEAD_INIT(NULL, 0)
1509 "MultibyteStreamReader", /* tp_name */
1510 sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
1511 0, /* tp_itemsize */
1512 /* methods */
1513 (destructor)mbstreamreader_dealloc, /* tp_dealloc */
1514 0, /* tp_print */
1515 0, /* tp_getattr */
1516 0, /* tp_setattr */
1517 0, /* tp_compare */
1518 0, /* tp_repr */
1519 0, /* tp_as_number */
1520 0, /* tp_as_sequence */
1521 0, /* tp_as_mapping */
1522 0, /* tp_hash */
1523 0, /* tp_call */
1524 0, /* tp_str */
1525 PyObject_GenericGetAttr, /* tp_getattro */
1526 0, /* tp_setattro */
1527 0, /* tp_as_buffer */
1528 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1529 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1530 0, /* tp_doc */
1531 (traverseproc)mbstreamreader_traverse, /* tp_traverse */
1532 0, /* tp_clear */
1533 0, /* tp_richcompare */
1534 0, /* tp_weaklistoffset */
1535 0, /* tp_iter */
1536 0, /* tp_iterext */
1537 mbstreamreader_methods, /* tp_methods */
1538 mbstreamreader_members, /* tp_members */
1539 codecctx_getsets, /* tp_getset */
1540 0, /* tp_base */
1541 0, /* tp_dict */
1542 0, /* tp_descr_get */
1543 0, /* tp_descr_set */
1544 0, /* tp_dictoffset */
1545 mbstreamreader_init, /* tp_init */
1546 0, /* tp_alloc */
1547 mbstreamreader_new, /* tp_new */
1548 };
1549
1550
1551 /**
1552 * MultibyteStreamWriter object
1553 */
1554
1555 static int
mbstreamwriter_iwrite(MultibyteStreamWriterObject * self,PyObject * unistr)1556 mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
1557 PyObject *unistr)
1558 {
1559 PyObject *str, *wr;
1560
1561 str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
1562 if (str == NULL)
1563 return -1;
1564
1565 wr = PyObject_CallMethod(self->stream, "write", "O", str);
1566 Py_DECREF(str);
1567 if (wr == NULL)
1568 return -1;
1569
1570 Py_DECREF(wr);
1571 return 0;
1572 }
1573
1574 static PyObject *
mbstreamwriter_write(MultibyteStreamWriterObject * self,PyObject * strobj)1575 mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj)
1576 {
1577 if (mbstreamwriter_iwrite(self, strobj))
1578 return NULL;
1579 else
1580 Py_RETURN_NONE;
1581 }
1582
1583 static PyObject *
mbstreamwriter_writelines(MultibyteStreamWriterObject * self,PyObject * lines)1584 mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines)
1585 {
1586 PyObject *strobj;
1587 int i, r;
1588
1589 if (!PySequence_Check(lines)) {
1590 PyErr_SetString(PyExc_TypeError,
1591 "arg must be a sequence object");
1592 return NULL;
1593 }
1594
1595 for (i = 0; i < PySequence_Length(lines); i++) {
1596 /* length can be changed even within this loop */
1597 strobj = PySequence_GetItem(lines, i);
1598 if (strobj == NULL)
1599 return NULL;
1600
1601 r = mbstreamwriter_iwrite(self, strobj);
1602 Py_DECREF(strobj);
1603 if (r == -1)
1604 return NULL;
1605 }
1606
1607 Py_RETURN_NONE;
1608 }
1609
1610 static PyObject *
mbstreamwriter_reset(MultibyteStreamWriterObject * self)1611 mbstreamwriter_reset(MultibyteStreamWriterObject *self)
1612 {
1613 const Py_UNICODE *pending;
1614 PyObject *pwrt;
1615
1616 pending = self->pending;
1617 pwrt = multibytecodec_encode(self->codec, &self->state,
1618 &pending, self->pendingsize, self->errors,
1619 MBENC_FLUSH | MBENC_RESET);
1620 /* some pending buffer can be truncated when UnicodeEncodeError is
1621 * raised on 'strict' mode. but, 'reset' method is designed to
1622 * reset the pending buffer or states so failed string sequence
1623 * ought to be missed */
1624 self->pendingsize = 0;
1625 if (pwrt == NULL)
1626 return NULL;
1627
1628 if (PyString_Size(pwrt) > 0) {
1629 PyObject *wr;
1630 wr = PyObject_CallMethod(self->stream, "write", "O", pwrt);
1631 if (wr == NULL) {
1632 Py_DECREF(pwrt);
1633 return NULL;
1634 }
1635 }
1636 Py_DECREF(pwrt);
1637
1638 Py_RETURN_NONE;
1639 }
1640
1641 static PyObject *
mbstreamwriter_new(PyTypeObject * type,PyObject * args,PyObject * kwds)1642 mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1643 {
1644 MultibyteStreamWriterObject *self;
1645 PyObject *stream, *codec = NULL;
1646 char *errors = NULL;
1647
1648 if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
1649 streamkwarglist, &stream, &errors))
1650 return NULL;
1651
1652 self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
1653 if (self == NULL)
1654 return NULL;
1655
1656 codec = PyObject_GetAttrString((PyObject *)type, "codec");
1657 if (codec == NULL)
1658 goto errorexit;
1659 if (!MultibyteCodec_Check(codec)) {
1660 PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1661 goto errorexit;
1662 }
1663
1664 self->codec = ((MultibyteCodecObject *)codec)->codec;
1665 self->stream = stream;
1666 Py_INCREF(stream);
1667 self->pendingsize = 0;
1668 self->errors = internal_error_callback(errors);
1669 if (self->errors == NULL)
1670 goto errorexit;
1671 if (self->codec->encinit != NULL &&
1672 self->codec->encinit(&self->state, self->codec->config) != 0)
1673 goto errorexit;
1674
1675 Py_DECREF(codec);
1676 return (PyObject *)self;
1677
1678 errorexit:
1679 Py_XDECREF(self);
1680 Py_XDECREF(codec);
1681 return NULL;
1682 }
1683
1684 static int
mbstreamwriter_init(PyObject * self,PyObject * args,PyObject * kwds)1685 mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
1686 {
1687 return 0;
1688 }
1689
1690 static int
mbstreamwriter_traverse(MultibyteStreamWriterObject * self,visitproc visit,void * arg)1691 mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
1692 visitproc visit, void *arg)
1693 {
1694 if (ERROR_ISCUSTOM(self->errors))
1695 Py_VISIT(self->errors);
1696 Py_VISIT(self->stream);
1697 return 0;
1698 }
1699
1700 static void
mbstreamwriter_dealloc(MultibyteStreamWriterObject * self)1701 mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
1702 {
1703 PyObject_GC_UnTrack(self);
1704 ERROR_DECREF(self->errors);
1705 Py_XDECREF(self->stream);
1706 Py_TYPE(self)->tp_free(self);
1707 }
1708
1709 static struct PyMethodDef mbstreamwriter_methods[] = {
1710 {"write", (PyCFunction)mbstreamwriter_write,
1711 METH_O, NULL},
1712 {"writelines", (PyCFunction)mbstreamwriter_writelines,
1713 METH_O, NULL},
1714 {"reset", (PyCFunction)mbstreamwriter_reset,
1715 METH_NOARGS, NULL},
1716 {NULL, NULL},
1717 };
1718
1719 static PyMemberDef mbstreamwriter_members[] = {
1720 {"stream", T_OBJECT,
1721 offsetof(MultibyteStreamWriterObject, stream),
1722 READONLY, NULL},
1723 {NULL,}
1724 };
1725
1726 static PyTypeObject MultibyteStreamWriter_Type = {
1727 PyVarObject_HEAD_INIT(NULL, 0)
1728 "MultibyteStreamWriter", /* tp_name */
1729 sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
1730 0, /* tp_itemsize */
1731 /* methods */
1732 (destructor)mbstreamwriter_dealloc, /* tp_dealloc */
1733 0, /* tp_print */
1734 0, /* tp_getattr */
1735 0, /* tp_setattr */
1736 0, /* tp_compare */
1737 0, /* tp_repr */
1738 0, /* tp_as_number */
1739 0, /* tp_as_sequence */
1740 0, /* tp_as_mapping */
1741 0, /* tp_hash */
1742 0, /* tp_call */
1743 0, /* tp_str */
1744 PyObject_GenericGetAttr, /* tp_getattro */
1745 0, /* tp_setattro */
1746 0, /* tp_as_buffer */
1747 Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1748 | Py_TPFLAGS_BASETYPE, /* tp_flags */
1749 0, /* tp_doc */
1750 (traverseproc)mbstreamwriter_traverse, /* tp_traverse */
1751 0, /* tp_clear */
1752 0, /* tp_richcompare */
1753 0, /* tp_weaklistoffset */
1754 0, /* tp_iter */
1755 0, /* tp_iterext */
1756 mbstreamwriter_methods, /* tp_methods */
1757 mbstreamwriter_members, /* tp_members */
1758 codecctx_getsets, /* tp_getset */
1759 0, /* tp_base */
1760 0, /* tp_dict */
1761 0, /* tp_descr_get */
1762 0, /* tp_descr_set */
1763 0, /* tp_dictoffset */
1764 mbstreamwriter_init, /* tp_init */
1765 0, /* tp_alloc */
1766 mbstreamwriter_new, /* tp_new */
1767 };
1768
1769
1770 /**
1771 * Exposed factory function
1772 */
1773
1774 static PyObject *
__create_codec(PyObject * ignore,PyObject * arg)1775 __create_codec(PyObject *ignore, PyObject *arg)
1776 {
1777 MultibyteCodecObject *self;
1778 MultibyteCodec *codec;
1779
1780 if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
1781 PyErr_SetString(PyExc_ValueError, "argument type invalid");
1782 return NULL;
1783 }
1784
1785 codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
1786 if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
1787 return NULL;
1788
1789 self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
1790 if (self == NULL)
1791 return NULL;
1792 self->codec = codec;
1793
1794 return (PyObject *)self;
1795 }
1796
1797 static struct PyMethodDef __methods[] = {
1798 {"__create_codec", (PyCFunction)__create_codec, METH_O},
1799 {NULL, NULL},
1800 };
1801
1802 PyMODINIT_FUNC
init_multibytecodec(void)1803 init_multibytecodec(void)
1804 {
1805 int i;
1806 PyObject *m;
1807 PyTypeObject *typelist[] = {
1808 &MultibyteIncrementalEncoder_Type,
1809 &MultibyteIncrementalDecoder_Type,
1810 &MultibyteStreamReader_Type,
1811 &MultibyteStreamWriter_Type,
1812 NULL
1813 };
1814
1815 if (PyType_Ready(&MultibyteCodec_Type) < 0)
1816 return;
1817
1818 m = Py_InitModule("_multibytecodec", __methods);
1819 if (m == NULL)
1820 return;
1821
1822 for (i = 0; typelist[i] != NULL; i++) {
1823 if (PyType_Ready(typelist[i]) < 0)
1824 return;
1825 Py_INCREF(typelist[i]);
1826 PyModule_AddObject(m, typelist[i]->tp_name,
1827 (PyObject *)typelist[i]);
1828 }
1829
1830 if (PyErr_Occurred())
1831 Py_FatalError("can't initialize the _multibytecodec module");
1832 }
1833