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         return -1;
175 
176     if (_PyString_Resize(&buf->outobj, orgsize + incsize) == -1)
177         return -1;
178 
179     buf->outbuf = (unsigned char *)PyString_AS_STRING(buf->outobj) +orgpos;
180     buf->outbuf_end = (unsigned char *)PyString_AS_STRING(buf->outobj)
181         + PyString_GET_SIZE(buf->outobj);
182 
183     return 0;
184 }
185 #define REQUIRE_ENCODEBUFFER(buf, s) {                                  \
186     if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end)             \
187         if (expand_encodebuffer(buf, s) == -1)                          \
188             goto errorexit;                                             \
189 }
190 
191 static int
expand_decodebuffer(MultibyteDecodeBuffer * buf,Py_ssize_t esize)192 expand_decodebuffer(MultibyteDecodeBuffer *buf, Py_ssize_t esize)
193 {
194     Py_ssize_t orgpos, orgsize;
195 
196     orgpos = (Py_ssize_t)(buf->outbuf - PyUnicode_AS_UNICODE(buf->outobj));
197     orgsize = PyUnicode_GET_SIZE(buf->outobj);
198     if (PyUnicode_Resize(&buf->outobj, orgsize + (
199         esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize)) == -1)
200         return -1;
201 
202     buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj) + orgpos;
203     buf->outbuf_end = PyUnicode_AS_UNICODE(buf->outobj)
204                       + PyUnicode_GET_SIZE(buf->outobj);
205 
206     return 0;
207 }
208 #define REQUIRE_DECODEBUFFER(buf, s) {                                  \
209     if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end)             \
210         if (expand_decodebuffer(buf, s) == -1)                          \
211             goto errorexit;                                             \
212 }
213 
214 
215 /**
216  * MultibyteCodec object
217  */
218 
219 static int
multibytecodec_encerror(MultibyteCodec * codec,MultibyteCodec_State * state,MultibyteEncodeBuffer * buf,PyObject * errors,Py_ssize_t e)220 multibytecodec_encerror(MultibyteCodec *codec,
221                         MultibyteCodec_State *state,
222                         MultibyteEncodeBuffer *buf,
223                         PyObject *errors, Py_ssize_t e)
224 {
225     PyObject *retobj = NULL, *retstr = NULL, *tobj;
226     Py_ssize_t retstrsize, newpos;
227     Py_ssize_t esize, start, end;
228     const char *reason;
229 
230     if (e > 0) {
231         reason = "illegal multibyte sequence";
232         esize = e;
233     }
234     else {
235         switch (e) {
236         case MBERR_TOOSMALL:
237             REQUIRE_ENCODEBUFFER(buf, -1);
238             return 0; /* retry it */
239         case MBERR_TOOFEW:
240             reason = "incomplete multibyte sequence";
241             esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
242             break;
243         case MBERR_INTERNAL:
244             PyErr_SetString(PyExc_RuntimeError,
245                             "internal codec error");
246             return -1;
247         default:
248             PyErr_SetString(PyExc_RuntimeError,
249                             "unknown runtime error");
250             return -1;
251         }
252     }
253 
254     if (errors == ERROR_REPLACE) {
255         const Py_UNICODE replchar = '?', *inbuf = &replchar;
256         Py_ssize_t r;
257 
258         for (;;) {
259             Py_ssize_t outleft;
260 
261             outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
262             r = codec->encode(state, codec->config, &inbuf, 1,
263                               &buf->outbuf, outleft, 0);
264             if (r == MBERR_TOOSMALL) {
265                 REQUIRE_ENCODEBUFFER(buf, -1);
266                 continue;
267             }
268             else
269                 break;
270         }
271 
272         if (r != 0) {
273             REQUIRE_ENCODEBUFFER(buf, 1);
274             *buf->outbuf++ = '?';
275         }
276     }
277     if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
278         buf->inbuf += esize;
279         return 0;
280     }
281 
282     start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
283     end = start + esize;
284 
285     /* use cached exception object if available */
286     if (buf->excobj == NULL) {
287         buf->excobj = PyUnicodeEncodeError_Create(codec->encoding,
288                         buf->inbuf_top,
289                         buf->inbuf_end - buf->inbuf_top,
290                         start, end, reason);
291         if (buf->excobj == NULL)
292             goto errorexit;
293     }
294     else
295         if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 ||
296             PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 ||
297             PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0)
298             goto errorexit;
299 
300     if (errors == ERROR_STRICT) {
301         PyCodec_StrictErrors(buf->excobj);
302         goto errorexit;
303     }
304 
305     retobj = call_error_callback(errors, buf->excobj);
306     if (retobj == NULL)
307         goto errorexit;
308 
309     if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
310         !PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) ||
311         !(PyInt_Check(PyTuple_GET_ITEM(retobj, 1)) ||
312           PyLong_Check(PyTuple_GET_ITEM(retobj, 1)))) {
313         PyErr_SetString(PyExc_TypeError,
314                         "encoding error handler must return "
315                         "(unicode, int) tuple");
316         goto errorexit;
317     }
318 
319     {
320         const Py_UNICODE *uraw = PyUnicode_AS_UNICODE(tobj);
321 
322         retstr = multibytecodec_encode(codec, state, &uraw,
323                         PyUnicode_GET_SIZE(tobj), ERROR_STRICT,
324                         MBENC_FLUSH);
325         if (retstr == NULL)
326             goto errorexit;
327     }
328 
329     retstrsize = PyString_GET_SIZE(retstr);
330     REQUIRE_ENCODEBUFFER(buf, retstrsize);
331 
332     memcpy(buf->outbuf, PyString_AS_STRING(retstr), retstrsize);
333     buf->outbuf += retstrsize;
334 
335     newpos = PyInt_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
336     if (newpos < 0 && !PyErr_Occurred())
337         newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
338     if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
339         PyErr_Clear();
340         PyErr_Format(PyExc_IndexError,
341                      "position %zd from error handler out of bounds",
342                      newpos);
343         goto errorexit;
344     }
345     buf->inbuf = buf->inbuf_top + newpos;
346 
347     Py_DECREF(retobj);
348     Py_DECREF(retstr);
349     return 0;
350 
351 errorexit:
352     Py_XDECREF(retobj);
353     Py_XDECREF(retstr);
354     return -1;
355 }
356 
357 static int
multibytecodec_decerror(MultibyteCodec * codec,MultibyteCodec_State * state,MultibyteDecodeBuffer * buf,PyObject * errors,Py_ssize_t e)358 multibytecodec_decerror(MultibyteCodec *codec,
359                         MultibyteCodec_State *state,
360                         MultibyteDecodeBuffer *buf,
361                         PyObject *errors, Py_ssize_t e)
362 {
363     PyObject *retobj = NULL, *retuni = NULL;
364     Py_ssize_t retunisize, newpos;
365     const char *reason;
366     Py_ssize_t esize, start, end;
367 
368     if (e > 0) {
369         reason = "illegal multibyte sequence";
370         esize = e;
371     }
372     else {
373         switch (e) {
374         case MBERR_TOOSMALL:
375             REQUIRE_DECODEBUFFER(buf, -1);
376             return 0; /* retry it */
377         case MBERR_TOOFEW:
378             reason = "incomplete multibyte sequence";
379             esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
380             break;
381         case MBERR_INTERNAL:
382             PyErr_SetString(PyExc_RuntimeError,
383                             "internal codec error");
384             return -1;
385         default:
386             PyErr_SetString(PyExc_RuntimeError,
387                             "unknown runtime error");
388             return -1;
389         }
390     }
391 
392     if (errors == ERROR_REPLACE) {
393         REQUIRE_DECODEBUFFER(buf, 1);
394         *buf->outbuf++ = Py_UNICODE_REPLACEMENT_CHARACTER;
395     }
396     if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) {
397         buf->inbuf += esize;
398         return 0;
399     }
400 
401     start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top);
402     end = start + esize;
403 
404     /* use cached exception object if available */
405     if (buf->excobj == NULL) {
406         buf->excobj = PyUnicodeDecodeError_Create(codec->encoding,
407                         (const char *)buf->inbuf_top,
408                         (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top),
409                         start, end, reason);
410         if (buf->excobj == NULL)
411             goto errorexit;
412     }
413     else
414         if (PyUnicodeDecodeError_SetStart(buf->excobj, start) ||
415             PyUnicodeDecodeError_SetEnd(buf->excobj, end) ||
416             PyUnicodeDecodeError_SetReason(buf->excobj, reason))
417             goto errorexit;
418 
419     if (errors == ERROR_STRICT) {
420         PyCodec_StrictErrors(buf->excobj);
421         goto errorexit;
422     }
423 
424     retobj = call_error_callback(errors, buf->excobj);
425     if (retobj == NULL)
426         goto errorexit;
427 
428     if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 ||
429         !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) ||
430         !(PyInt_Check(PyTuple_GET_ITEM(retobj, 1)) ||
431           PyLong_Check(PyTuple_GET_ITEM(retobj, 1)))) {
432         PyErr_SetString(PyExc_TypeError,
433                         "decoding error handler must return "
434                         "(unicode, int) tuple");
435         goto errorexit;
436     }
437 
438     retunisize = PyUnicode_GET_SIZE(retuni);
439     if (retunisize > 0) {
440         REQUIRE_DECODEBUFFER(buf, retunisize);
441         memcpy((char *)buf->outbuf, PyUnicode_AS_DATA(retuni),
442                         retunisize * Py_UNICODE_SIZE);
443         buf->outbuf += retunisize;
444     }
445 
446     newpos = PyInt_AsSsize_t(PyTuple_GET_ITEM(retobj, 1));
447     if (newpos < 0 && !PyErr_Occurred())
448         newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top);
449     if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) {
450         PyErr_Clear();
451         PyErr_Format(PyExc_IndexError,
452                      "position %zd from error handler out of bounds",
453                      newpos);
454         goto errorexit;
455     }
456     buf->inbuf = buf->inbuf_top + newpos;
457     Py_DECREF(retobj);
458     return 0;
459 
460 errorexit:
461     Py_XDECREF(retobj);
462     return -1;
463 }
464 
465 static PyObject *
multibytecodec_encode(MultibyteCodec * codec,MultibyteCodec_State * state,const Py_UNICODE ** data,Py_ssize_t datalen,PyObject * errors,int flags)466 multibytecodec_encode(MultibyteCodec *codec,
467                       MultibyteCodec_State *state,
468                       const Py_UNICODE **data, Py_ssize_t datalen,
469                       PyObject *errors, int flags)
470 {
471     MultibyteEncodeBuffer buf;
472     Py_ssize_t finalsize, r = 0;
473 
474     if (datalen == 0 && !(flags & MBENC_RESET))
475         return PyString_FromString("");
476 
477     buf.excobj = NULL;
478     buf.inbuf = buf.inbuf_top = *data;
479     buf.inbuf_end = buf.inbuf_top + datalen;
480 
481     if (datalen > (PY_SSIZE_T_MAX - 16) / 2) {
482         PyErr_NoMemory();
483         goto errorexit;
484     }
485 
486     buf.outobj = PyString_FromStringAndSize(NULL, datalen * 2 + 16);
487     if (buf.outobj == NULL)
488         goto errorexit;
489     buf.outbuf = (unsigned char *)PyString_AS_STRING(buf.outobj);
490     buf.outbuf_end = buf.outbuf + PyString_GET_SIZE(buf.outobj);
491 
492     while (buf.inbuf < buf.inbuf_end) {
493         Py_ssize_t inleft, outleft;
494 
495         /* we don't reuse inleft and outleft here.
496          * error callbacks can relocate the cursor anywhere on buffer*/
497         inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
498         outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
499         r = codec->encode(state, codec->config, &buf.inbuf, inleft,
500                           &buf.outbuf, outleft, flags);
501         if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH)))
502             break;
503         else if (multibytecodec_encerror(codec, state, &buf, errors,r))
504             goto errorexit;
505         else if (r == MBERR_TOOFEW)
506             break;
507     }
508 
509     if (codec->encreset != NULL && (flags & MBENC_RESET))
510         for (;;) {
511             Py_ssize_t outleft;
512 
513             outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
514             r = codec->encreset(state, codec->config, &buf.outbuf,
515                                 outleft);
516             if (r == 0)
517                 break;
518             else if (multibytecodec_encerror(codec, state,
519                                              &buf, errors, r))
520                 goto errorexit;
521         }
522 
523     finalsize = (Py_ssize_t)((char *)buf.outbuf -
524                              PyString_AS_STRING(buf.outobj));
525 
526     if (finalsize != PyString_GET_SIZE(buf.outobj))
527         if (_PyString_Resize(&buf.outobj, finalsize) == -1)
528             goto errorexit;
529 
530 	*data = buf.inbuf;
531     Py_XDECREF(buf.excobj);
532     return buf.outobj;
533 
534 errorexit:
535     Py_XDECREF(buf.excobj);
536     Py_XDECREF(buf.outobj);
537     return NULL;
538 }
539 
540 static PyObject *
MultibyteCodec_Encode(MultibyteCodecObject * self,PyObject * args,PyObject * kwargs)541 MultibyteCodec_Encode(MultibyteCodecObject *self,
542                       PyObject *args, PyObject *kwargs)
543 {
544     MultibyteCodec_State state;
545     Py_UNICODE *data;
546     PyObject *errorcb, *r, *arg, *ucvt;
547     const char *errors = NULL;
548     Py_ssize_t datalen;
549 
550     if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode",
551                             codeckwarglist, &arg, &errors))
552         return NULL;
553 
554     if (PyUnicode_Check(arg))
555         ucvt = NULL;
556     else {
557         arg = ucvt = PyObject_Unicode(arg);
558         if (arg == NULL)
559             return NULL;
560         else if (!PyUnicode_Check(arg)) {
561             PyErr_SetString(PyExc_TypeError,
562                 "couldn't convert the object to unicode.");
563             Py_DECREF(ucvt);
564             return NULL;
565         }
566     }
567 
568     data = PyUnicode_AS_UNICODE(arg);
569     datalen = PyUnicode_GET_SIZE(arg);
570 
571     errorcb = internal_error_callback(errors);
572     if (errorcb == NULL) {
573         Py_XDECREF(ucvt);
574         return NULL;
575     }
576 
577     if (self->codec->encinit != NULL &&
578         self->codec->encinit(&state, self->codec->config) != 0)
579         goto errorexit;
580     r = multibytecodec_encode(self->codec, &state,
581                     (const Py_UNICODE **)&data, datalen, errorcb,
582                     MBENC_FLUSH | MBENC_RESET);
583     if (r == NULL)
584         goto errorexit;
585 
586     ERROR_DECREF(errorcb);
587     Py_XDECREF(ucvt);
588     return make_tuple(r, datalen);
589 
590 errorexit:
591     ERROR_DECREF(errorcb);
592     Py_XDECREF(ucvt);
593     return NULL;
594 }
595 
596 static PyObject *
MultibyteCodec_Decode(MultibyteCodecObject * self,PyObject * args,PyObject * kwargs)597 MultibyteCodec_Decode(MultibyteCodecObject *self,
598                       PyObject *args, PyObject *kwargs)
599 {
600     MultibyteCodec_State state;
601     MultibyteDecodeBuffer buf;
602     PyObject *errorcb;
603     Py_buffer pdata;
604     const char *data, *errors = NULL;
605     Py_ssize_t datalen, finalsize;
606 
607     if (!PyArg_ParseTupleAndKeywords(args, kwargs, "s*|z:decode",
608                             codeckwarglist, &pdata, &errors))
609         return NULL;
610     data = pdata.buf;
611     datalen = pdata.len;
612 
613     errorcb = internal_error_callback(errors);
614     if (errorcb == NULL) {
615         PyBuffer_Release(&pdata);
616         return NULL;
617     }
618 
619     if (datalen == 0) {
620         PyBuffer_Release(&pdata);
621         ERROR_DECREF(errorcb);
622         return make_tuple(PyUnicode_FromUnicode(NULL, 0), 0);
623     }
624 
625     buf.excobj = NULL;
626     buf.inbuf = buf.inbuf_top = (unsigned char *)data;
627     buf.inbuf_end = buf.inbuf_top + datalen;
628     buf.outobj = PyUnicode_FromUnicode(NULL, datalen);
629     if (buf.outobj == NULL)
630         goto errorexit;
631     buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj);
632     buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj);
633 
634     if (self->codec->decinit != NULL &&
635         self->codec->decinit(&state, self->codec->config) != 0)
636         goto errorexit;
637 
638     while (buf.inbuf < buf.inbuf_end) {
639         Py_ssize_t inleft, outleft, r;
640 
641         inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf);
642         outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf);
643 
644         r = self->codec->decode(&state, self->codec->config,
645                         &buf.inbuf, inleft, &buf.outbuf, outleft);
646         if (r == 0)
647             break;
648         else if (multibytecodec_decerror(self->codec, &state,
649                                          &buf, errorcb, r))
650             goto errorexit;
651     }
652 
653     finalsize = (Py_ssize_t)(buf.outbuf -
654                              PyUnicode_AS_UNICODE(buf.outobj));
655 
656     if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
657         if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
658             goto errorexit;
659 
660     PyBuffer_Release(&pdata);
661     Py_XDECREF(buf.excobj);
662     ERROR_DECREF(errorcb);
663     return make_tuple(buf.outobj, datalen);
664 
665 errorexit:
666     PyBuffer_Release(&pdata);
667     ERROR_DECREF(errorcb);
668     Py_XDECREF(buf.excobj);
669     Py_XDECREF(buf.outobj);
670 
671     return NULL;
672 }
673 
674 static struct PyMethodDef multibytecodec_methods[] = {
675     {"encode",          (PyCFunction)MultibyteCodec_Encode,
676                     METH_VARARGS | METH_KEYWORDS,
677                     MultibyteCodec_Encode__doc__},
678     {"decode",          (PyCFunction)MultibyteCodec_Decode,
679                     METH_VARARGS | METH_KEYWORDS,
680                     MultibyteCodec_Decode__doc__},
681     {NULL,              NULL},
682 };
683 
684 static void
multibytecodec_dealloc(MultibyteCodecObject * self)685 multibytecodec_dealloc(MultibyteCodecObject *self)
686 {
687     PyObject_Del(self);
688 }
689 
690 static PyTypeObject MultibyteCodec_Type = {
691     PyVarObject_HEAD_INIT(NULL, 0)
692     "MultibyteCodec",                   /* tp_name */
693     sizeof(MultibyteCodecObject),       /* tp_basicsize */
694     0,                                  /* tp_itemsize */
695     /* methods */
696     (destructor)multibytecodec_dealloc, /* tp_dealloc */
697     0,                                  /* tp_print */
698     0,                                  /* tp_getattr */
699     0,                                  /* tp_setattr */
700     0,                                  /* tp_compare */
701     0,                                  /* tp_repr */
702     0,                                  /* tp_as_number */
703     0,                                  /* tp_as_sequence */
704     0,                                  /* tp_as_mapping */
705     0,                                  /* tp_hash */
706     0,                                  /* tp_call */
707     0,                                  /* tp_str */
708     PyObject_GenericGetAttr,            /* tp_getattro */
709     0,                                  /* tp_setattro */
710     0,                                  /* tp_as_buffer */
711     Py_TPFLAGS_DEFAULT,                 /* tp_flags */
712     0,                                  /* tp_doc */
713     0,                                  /* tp_traverse */
714     0,                                  /* tp_clear */
715     0,                                  /* tp_richcompare */
716     0,                                  /* tp_weaklistoffset */
717     0,                                  /* tp_iter */
718     0,                                  /* tp_iterext */
719     multibytecodec_methods,             /* tp_methods */
720 };
721 
722 
723 /**
724  * Utility functions for stateful codec mechanism
725  */
726 
727 #define STATEFUL_DCTX(o)        ((MultibyteStatefulDecoderContext *)(o))
728 #define STATEFUL_ECTX(o)        ((MultibyteStatefulEncoderContext *)(o))
729 
730 static PyObject *
encoder_encode_stateful(MultibyteStatefulEncoderContext * ctx,PyObject * unistr,int final)731 encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx,
732                         PyObject *unistr, int final)
733 {
734     PyObject *ucvt, *r = NULL;
735     Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL;
736     Py_ssize_t datalen, origpending;
737 
738     if (PyUnicode_Check(unistr))
739         ucvt = NULL;
740     else {
741         unistr = ucvt = PyObject_Unicode(unistr);
742         if (unistr == NULL)
743             return NULL;
744         else if (!PyUnicode_Check(unistr)) {
745             PyErr_SetString(PyExc_TypeError,
746                 "couldn't convert the object to unicode.");
747             Py_DECREF(ucvt);
748             return NULL;
749         }
750     }
751 
752     datalen = PyUnicode_GET_SIZE(unistr);
753     origpending = ctx->pendingsize;
754 
755     if (origpending > 0) {
756         if (datalen > PY_SSIZE_T_MAX - ctx->pendingsize) {
757             PyErr_NoMemory();
758             /* inbuf_tmp == NULL */
759             goto errorexit;
760         }
761         inbuf_tmp = PyMem_New(Py_UNICODE, datalen + ctx->pendingsize);
762         if (inbuf_tmp == NULL)
763             goto errorexit;
764         memcpy(inbuf_tmp, ctx->pending,
765             Py_UNICODE_SIZE * ctx->pendingsize);
766         memcpy(inbuf_tmp + ctx->pendingsize,
767             PyUnicode_AS_UNICODE(unistr),
768             Py_UNICODE_SIZE * datalen);
769         datalen += ctx->pendingsize;
770         ctx->pendingsize = 0;
771         inbuf = inbuf_tmp;
772     }
773     else
774         inbuf = (Py_UNICODE *)PyUnicode_AS_UNICODE(unistr);
775 
776     inbuf_end = inbuf + datalen;
777 
778     r = multibytecodec_encode(ctx->codec, &ctx->state,
779                     (const Py_UNICODE **)&inbuf, datalen,
780                     ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0);
781     if (r == NULL) {
782         /* recover the original pending buffer */
783         if (origpending > 0)
784             memcpy(ctx->pending, inbuf_tmp,
785                 Py_UNICODE_SIZE * origpending);
786         ctx->pendingsize = origpending;
787         goto errorexit;
788     }
789 
790     if (inbuf < inbuf_end) {
791         ctx->pendingsize = (Py_ssize_t)(inbuf_end - inbuf);
792         if (ctx->pendingsize > MAXENCPENDING) {
793             /* normal codecs can't reach here */
794             ctx->pendingsize = 0;
795             PyErr_SetString(PyExc_UnicodeError,
796                             "pending buffer overflow");
797             goto errorexit;
798         }
799         memcpy(ctx->pending, inbuf,
800             ctx->pendingsize * Py_UNICODE_SIZE);
801     }
802 
803     if (inbuf_tmp != NULL)
804         PyMem_Del(inbuf_tmp);
805     Py_XDECREF(ucvt);
806     return r;
807 
808 errorexit:
809     if (inbuf_tmp != NULL)
810         PyMem_Del(inbuf_tmp);
811     Py_XDECREF(r);
812     Py_XDECREF(ucvt);
813     return NULL;
814 }
815 
816 static int
decoder_append_pending(MultibyteStatefulDecoderContext * ctx,MultibyteDecodeBuffer * buf)817 decoder_append_pending(MultibyteStatefulDecoderContext *ctx,
818                        MultibyteDecodeBuffer *buf)
819 {
820     Py_ssize_t npendings;
821 
822     npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
823     if (npendings + ctx->pendingsize > MAXDECPENDING ||
824         npendings > PY_SSIZE_T_MAX - ctx->pendingsize) {
825             PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow");
826             return -1;
827     }
828     memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings);
829     ctx->pendingsize += npendings;
830     return 0;
831 }
832 
833 static int
decoder_prepare_buffer(MultibyteDecodeBuffer * buf,const char * data,Py_ssize_t size)834 decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data,
835                        Py_ssize_t size)
836 {
837     buf->inbuf = buf->inbuf_top = (const unsigned char *)data;
838     buf->inbuf_end = buf->inbuf_top + size;
839     if (buf->outobj == NULL) { /* only if outobj is not allocated yet */
840         buf->outobj = PyUnicode_FromUnicode(NULL, size);
841         if (buf->outobj == NULL)
842             return -1;
843         buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj);
844         buf->outbuf_end = buf->outbuf +
845                           PyUnicode_GET_SIZE(buf->outobj);
846     }
847 
848     return 0;
849 }
850 
851 static int
decoder_feed_buffer(MultibyteStatefulDecoderContext * ctx,MultibyteDecodeBuffer * buf)852 decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx,
853                     MultibyteDecodeBuffer *buf)
854 {
855     while (buf->inbuf < buf->inbuf_end) {
856         Py_ssize_t inleft, outleft;
857         Py_ssize_t r;
858 
859         inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf);
860         outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf);
861 
862         r = ctx->codec->decode(&ctx->state, ctx->codec->config,
863             &buf->inbuf, inleft, &buf->outbuf, outleft);
864         if (r == 0 || r == MBERR_TOOFEW)
865             break;
866         else if (multibytecodec_decerror(ctx->codec, &ctx->state,
867                                          buf, ctx->errors, r))
868             return -1;
869     }
870     return 0;
871 }
872 
873 
874 /**
875  * MultibyteIncrementalEncoder object
876  */
877 
878 static PyObject *
mbiencoder_encode(MultibyteIncrementalEncoderObject * self,PyObject * args,PyObject * kwargs)879 mbiencoder_encode(MultibyteIncrementalEncoderObject *self,
880                   PyObject *args, PyObject *kwargs)
881 {
882     PyObject *data;
883     int final = 0;
884 
885     if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode",
886                     incrementalkwarglist, &data, &final))
887         return NULL;
888 
889     return encoder_encode_stateful(STATEFUL_ECTX(self), data, final);
890 }
891 
892 static PyObject *
mbiencoder_reset(MultibyteIncrementalEncoderObject * self)893 mbiencoder_reset(MultibyteIncrementalEncoderObject *self)
894 {
895     if (self->codec->decreset != NULL &&
896         self->codec->decreset(&self->state, self->codec->config) != 0)
897         return NULL;
898     self->pendingsize = 0;
899 
900     Py_RETURN_NONE;
901 }
902 
903 static struct PyMethodDef mbiencoder_methods[] = {
904     {"encode",          (PyCFunction)mbiencoder_encode,
905                     METH_VARARGS | METH_KEYWORDS, NULL},
906     {"reset",           (PyCFunction)mbiencoder_reset,
907                     METH_NOARGS, NULL},
908     {NULL,              NULL},
909 };
910 
911 static PyObject *
mbiencoder_new(PyTypeObject * type,PyObject * args,PyObject * kwds)912 mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
913 {
914     MultibyteIncrementalEncoderObject *self;
915     PyObject *codec = NULL;
916     char *errors = NULL;
917 
918     if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder",
919                                      incnewkwarglist, &errors))
920         return NULL;
921 
922     self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0);
923     if (self == NULL)
924         return NULL;
925 
926     codec = PyObject_GetAttrString((PyObject *)type, "codec");
927     if (codec == NULL)
928         goto errorexit;
929     if (!MultibyteCodec_Check(codec)) {
930         PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
931         goto errorexit;
932     }
933 
934     self->codec = ((MultibyteCodecObject *)codec)->codec;
935     self->pendingsize = 0;
936     self->errors = internal_error_callback(errors);
937     if (self->errors == NULL)
938         goto errorexit;
939     if (self->codec->encinit != NULL &&
940         self->codec->encinit(&self->state, self->codec->config) != 0)
941         goto errorexit;
942 
943     Py_DECREF(codec);
944     return (PyObject *)self;
945 
946 errorexit:
947     Py_XDECREF(self);
948     Py_XDECREF(codec);
949     return NULL;
950 }
951 
952 static int
mbiencoder_init(PyObject * self,PyObject * args,PyObject * kwds)953 mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds)
954 {
955     return 0;
956 }
957 
958 static int
mbiencoder_traverse(MultibyteIncrementalEncoderObject * self,visitproc visit,void * arg)959 mbiencoder_traverse(MultibyteIncrementalEncoderObject *self,
960                     visitproc visit, void *arg)
961 {
962     if (ERROR_ISCUSTOM(self->errors))
963         Py_VISIT(self->errors);
964     return 0;
965 }
966 
967 static void
mbiencoder_dealloc(MultibyteIncrementalEncoderObject * self)968 mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self)
969 {
970     PyObject_GC_UnTrack(self);
971     ERROR_DECREF(self->errors);
972     Py_TYPE(self)->tp_free(self);
973 }
974 
975 static PyTypeObject MultibyteIncrementalEncoder_Type = {
976     PyVarObject_HEAD_INIT(NULL, 0)
977     "MultibyteIncrementalEncoder",      /* tp_name */
978     sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */
979     0,                                  /* tp_itemsize */
980     /*  methods  */
981     (destructor)mbiencoder_dealloc, /* tp_dealloc */
982     0,                                  /* tp_print */
983     0,                                  /* tp_getattr */
984     0,                                  /* tp_setattr */
985     0,                                  /* tp_compare */
986     0,                                  /* tp_repr */
987     0,                                  /* tp_as_number */
988     0,                                  /* tp_as_sequence */
989     0,                                  /* tp_as_mapping */
990     0,                                  /* tp_hash */
991     0,                                  /* tp_call */
992     0,                                  /* tp_str */
993     PyObject_GenericGetAttr,            /* tp_getattro */
994     0,                                  /* tp_setattro */
995     0,                                  /* tp_as_buffer */
996     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
997         | Py_TPFLAGS_BASETYPE,          /* tp_flags */
998     0,                                  /* tp_doc */
999     (traverseproc)mbiencoder_traverse,          /* tp_traverse */
1000     0,                                  /* tp_clear */
1001     0,                                  /* tp_richcompare */
1002     0,                                  /* tp_weaklistoffset */
1003     0,                                  /* tp_iter */
1004     0,                                  /* tp_iterext */
1005     mbiencoder_methods,                 /* tp_methods */
1006     0,                                  /* tp_members */
1007     codecctx_getsets,                   /* tp_getset */
1008     0,                                  /* tp_base */
1009     0,                                  /* tp_dict */
1010     0,                                  /* tp_descr_get */
1011     0,                                  /* tp_descr_set */
1012     0,                                  /* tp_dictoffset */
1013     mbiencoder_init,                    /* tp_init */
1014     0,                                  /* tp_alloc */
1015     mbiencoder_new,                     /* tp_new */
1016 };
1017 
1018 
1019 /**
1020  * MultibyteIncrementalDecoder object
1021  */
1022 
1023 static PyObject *
mbidecoder_decode(MultibyteIncrementalDecoderObject * self,PyObject * args,PyObject * kwargs)1024 mbidecoder_decode(MultibyteIncrementalDecoderObject *self,
1025                   PyObject *args, PyObject *kwargs)
1026 {
1027     MultibyteDecodeBuffer buf;
1028     char *data, *wdata = NULL;
1029     Py_buffer pdata;
1030     Py_ssize_t wsize, finalsize = 0, size, origpending;
1031     int final = 0;
1032 
1033     if (!PyArg_ParseTupleAndKeywords(args, kwargs, "s*|i:decode",
1034                     incrementalkwarglist, &pdata, &final))
1035         return NULL;
1036     data = pdata.buf;
1037     size = pdata.len;
1038 
1039     buf.outobj = buf.excobj = NULL;
1040     origpending = self->pendingsize;
1041 
1042     if (self->pendingsize == 0) {
1043         wsize = size;
1044         wdata = data;
1045     }
1046     else {
1047         if (size > PY_SSIZE_T_MAX - self->pendingsize) {
1048             PyErr_NoMemory();
1049             goto errorexit;
1050         }
1051         wsize = size + self->pendingsize;
1052         wdata = PyMem_Malloc(wsize);
1053         if (wdata == NULL)
1054             goto errorexit;
1055         memcpy(wdata, self->pending, self->pendingsize);
1056         memcpy(wdata + self->pendingsize, data, size);
1057         self->pendingsize = 0;
1058     }
1059 
1060     if (decoder_prepare_buffer(&buf, wdata, wsize) != 0)
1061         goto errorexit;
1062 
1063     if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf))
1064         goto errorexit;
1065 
1066     if (final && buf.inbuf < buf.inbuf_end) {
1067         if (multibytecodec_decerror(self->codec, &self->state,
1068                         &buf, self->errors, MBERR_TOOFEW)) {
1069             /* recover the original pending buffer */
1070             memcpy(self->pending, wdata, origpending);
1071             self->pendingsize = origpending;
1072             goto errorexit;
1073         }
1074     }
1075 
1076     if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */
1077         if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0)
1078             goto errorexit;
1079     }
1080 
1081     finalsize = (Py_ssize_t)(buf.outbuf - PyUnicode_AS_UNICODE(buf.outobj));
1082     if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1083         if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1084             goto errorexit;
1085 
1086     PyBuffer_Release(&pdata);
1087     if (wdata != data)
1088         PyMem_Del(wdata);
1089     Py_XDECREF(buf.excobj);
1090     return buf.outobj;
1091 
1092 errorexit:
1093     PyBuffer_Release(&pdata);
1094     if (wdata != NULL && wdata != data)
1095         PyMem_Del(wdata);
1096     Py_XDECREF(buf.excobj);
1097     Py_XDECREF(buf.outobj);
1098     return NULL;
1099 }
1100 
1101 static PyObject *
mbidecoder_reset(MultibyteIncrementalDecoderObject * self)1102 mbidecoder_reset(MultibyteIncrementalDecoderObject *self)
1103 {
1104     if (self->codec->decreset != NULL &&
1105         self->codec->decreset(&self->state, self->codec->config) != 0)
1106         return NULL;
1107     self->pendingsize = 0;
1108 
1109     Py_RETURN_NONE;
1110 }
1111 
1112 static struct PyMethodDef mbidecoder_methods[] = {
1113     {"decode",          (PyCFunction)mbidecoder_decode,
1114                     METH_VARARGS | METH_KEYWORDS, NULL},
1115     {"reset",           (PyCFunction)mbidecoder_reset,
1116                     METH_NOARGS, NULL},
1117     {NULL,              NULL},
1118 };
1119 
1120 static PyObject *
mbidecoder_new(PyTypeObject * type,PyObject * args,PyObject * kwds)1121 mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1122 {
1123     MultibyteIncrementalDecoderObject *self;
1124     PyObject *codec = NULL;
1125     char *errors = NULL;
1126 
1127     if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder",
1128                                      incnewkwarglist, &errors))
1129         return NULL;
1130 
1131     self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0);
1132     if (self == NULL)
1133         return NULL;
1134 
1135     codec = PyObject_GetAttrString((PyObject *)type, "codec");
1136     if (codec == NULL)
1137         goto errorexit;
1138     if (!MultibyteCodec_Check(codec)) {
1139         PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1140         goto errorexit;
1141     }
1142 
1143     self->codec = ((MultibyteCodecObject *)codec)->codec;
1144     self->pendingsize = 0;
1145     self->errors = internal_error_callback(errors);
1146     if (self->errors == NULL)
1147         goto errorexit;
1148     if (self->codec->decinit != NULL &&
1149         self->codec->decinit(&self->state, self->codec->config) != 0)
1150         goto errorexit;
1151 
1152     Py_DECREF(codec);
1153     return (PyObject *)self;
1154 
1155 errorexit:
1156     Py_XDECREF(self);
1157     Py_XDECREF(codec);
1158     return NULL;
1159 }
1160 
1161 static int
mbidecoder_init(PyObject * self,PyObject * args,PyObject * kwds)1162 mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds)
1163 {
1164     return 0;
1165 }
1166 
1167 static int
mbidecoder_traverse(MultibyteIncrementalDecoderObject * self,visitproc visit,void * arg)1168 mbidecoder_traverse(MultibyteIncrementalDecoderObject *self,
1169                     visitproc visit, void *arg)
1170 {
1171     if (ERROR_ISCUSTOM(self->errors))
1172         Py_VISIT(self->errors);
1173     return 0;
1174 }
1175 
1176 static void
mbidecoder_dealloc(MultibyteIncrementalDecoderObject * self)1177 mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self)
1178 {
1179     PyObject_GC_UnTrack(self);
1180     ERROR_DECREF(self->errors);
1181     Py_TYPE(self)->tp_free(self);
1182 }
1183 
1184 static PyTypeObject MultibyteIncrementalDecoder_Type = {
1185     PyVarObject_HEAD_INIT(NULL, 0)
1186     "MultibyteIncrementalDecoder",      /* tp_name */
1187     sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */
1188     0,                                  /* tp_itemsize */
1189     /*  methods  */
1190     (destructor)mbidecoder_dealloc, /* tp_dealloc */
1191     0,                                  /* tp_print */
1192     0,                                  /* tp_getattr */
1193     0,                                  /* tp_setattr */
1194     0,                                  /* tp_compare */
1195     0,                                  /* tp_repr */
1196     0,                                  /* tp_as_number */
1197     0,                                  /* tp_as_sequence */
1198     0,                                  /* tp_as_mapping */
1199     0,                                  /* tp_hash */
1200     0,                                  /* tp_call */
1201     0,                                  /* tp_str */
1202     PyObject_GenericGetAttr,            /* tp_getattro */
1203     0,                                  /* tp_setattro */
1204     0,                                  /* tp_as_buffer */
1205     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1206         | Py_TPFLAGS_BASETYPE,          /* tp_flags */
1207     0,                                  /* tp_doc */
1208     (traverseproc)mbidecoder_traverse,          /* tp_traverse */
1209     0,                                  /* tp_clear */
1210     0,                                  /* tp_richcompare */
1211     0,                                  /* tp_weaklistoffset */
1212     0,                                  /* tp_iter */
1213     0,                                  /* tp_iterext */
1214     mbidecoder_methods,                 /* tp_methods */
1215     0,                                  /* tp_members */
1216     codecctx_getsets,                   /* tp_getset */
1217     0,                                  /* tp_base */
1218     0,                                  /* tp_dict */
1219     0,                                  /* tp_descr_get */
1220     0,                                  /* tp_descr_set */
1221     0,                                  /* tp_dictoffset */
1222     mbidecoder_init,                    /* tp_init */
1223     0,                                  /* tp_alloc */
1224     mbidecoder_new,                     /* tp_new */
1225 };
1226 
1227 
1228 /**
1229  * MultibyteStreamReader object
1230  */
1231 
1232 static PyObject *
mbstreamreader_iread(MultibyteStreamReaderObject * self,const char * method,Py_ssize_t sizehint)1233 mbstreamreader_iread(MultibyteStreamReaderObject *self,
1234                      const char *method, Py_ssize_t sizehint)
1235 {
1236     MultibyteDecodeBuffer buf;
1237     PyObject *cres;
1238     Py_ssize_t rsize, finalsize = 0;
1239 
1240     if (sizehint == 0)
1241         return PyUnicode_FromUnicode(NULL, 0);
1242 
1243     buf.outobj = buf.excobj = NULL;
1244     cres = NULL;
1245 
1246     for (;;) {
1247         int endoffile;
1248 
1249         if (sizehint < 0)
1250             cres = PyObject_CallMethod(self->stream,
1251                             (char *)method, NULL);
1252         else
1253             cres = PyObject_CallMethod(self->stream,
1254                             (char *)method, "i", sizehint);
1255         if (cres == NULL)
1256             goto errorexit;
1257 
1258         if (!PyString_Check(cres)) {
1259             PyErr_SetString(PyExc_TypeError,
1260                             "stream function returned a "
1261                             "non-string object");
1262             goto errorexit;
1263         }
1264 
1265         endoffile = (PyString_GET_SIZE(cres) == 0);
1266 
1267         if (self->pendingsize > 0) {
1268             PyObject *ctr;
1269             char *ctrdata;
1270 
1271             if (PyString_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) {
1272                 PyErr_NoMemory();
1273                 goto errorexit;
1274         }
1275                     rsize = PyString_GET_SIZE(cres) + self->pendingsize;
1276                     ctr = PyString_FromStringAndSize(NULL, rsize);
1277                     if (ctr == NULL)
1278                             goto errorexit;
1279                     ctrdata = PyString_AS_STRING(ctr);
1280                     memcpy(ctrdata, self->pending, self->pendingsize);
1281                     memcpy(ctrdata + self->pendingsize,
1282                             PyString_AS_STRING(cres),
1283                             PyString_GET_SIZE(cres));
1284                     Py_DECREF(cres);
1285                     cres = ctr;
1286                     self->pendingsize = 0;
1287         }
1288 
1289         rsize = PyString_GET_SIZE(cres);
1290         if (decoder_prepare_buffer(&buf, PyString_AS_STRING(cres),
1291                                    rsize) != 0)
1292             goto errorexit;
1293 
1294         if (rsize > 0 && decoder_feed_buffer(
1295                         (MultibyteStatefulDecoderContext *)self, &buf))
1296             goto errorexit;
1297 
1298         if (endoffile || sizehint < 0) {
1299             if (buf.inbuf < buf.inbuf_end &&
1300                 multibytecodec_decerror(self->codec, &self->state,
1301                             &buf, self->errors, MBERR_TOOFEW))
1302                 goto errorexit;
1303         }
1304 
1305         if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */
1306             if (decoder_append_pending(STATEFUL_DCTX(self),
1307                                        &buf) != 0)
1308                 goto errorexit;
1309         }
1310 
1311         finalsize = (Py_ssize_t)(buf.outbuf -
1312                         PyUnicode_AS_UNICODE(buf.outobj));
1313         Py_DECREF(cres);
1314         cres = NULL;
1315 
1316         if (sizehint < 0 || finalsize != 0 || rsize == 0)
1317             break;
1318 
1319         sizehint = 1; /* read 1 more byte and retry */
1320     }
1321 
1322     if (finalsize != PyUnicode_GET_SIZE(buf.outobj))
1323         if (PyUnicode_Resize(&buf.outobj, finalsize) == -1)
1324             goto errorexit;
1325 
1326     Py_XDECREF(cres);
1327     Py_XDECREF(buf.excobj);
1328     return buf.outobj;
1329 
1330 errorexit:
1331     Py_XDECREF(cres);
1332     Py_XDECREF(buf.excobj);
1333     Py_XDECREF(buf.outobj);
1334     return NULL;
1335 }
1336 
1337 static PyObject *
mbstreamreader_read(MultibyteStreamReaderObject * self,PyObject * args)1338 mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args)
1339 {
1340     PyObject *sizeobj = NULL;
1341     Py_ssize_t size;
1342 
1343     if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj))
1344         return NULL;
1345 
1346     if (sizeobj == Py_None || sizeobj == NULL)
1347         size = -1;
1348     else if (PyInt_Check(sizeobj))
1349         size = PyInt_AsSsize_t(sizeobj);
1350     else {
1351         PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1352         return NULL;
1353     }
1354 
1355     return mbstreamreader_iread(self, "read", size);
1356 }
1357 
1358 static PyObject *
mbstreamreader_readline(MultibyteStreamReaderObject * self,PyObject * args)1359 mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args)
1360 {
1361     PyObject *sizeobj = NULL;
1362     Py_ssize_t size;
1363 
1364     if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj))
1365         return NULL;
1366 
1367     if (sizeobj == Py_None || sizeobj == NULL)
1368         size = -1;
1369     else if (PyInt_Check(sizeobj))
1370         size = PyInt_AsSsize_t(sizeobj);
1371     else {
1372         PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1373         return NULL;
1374     }
1375 
1376     return mbstreamreader_iread(self, "readline", size);
1377 }
1378 
1379 static PyObject *
mbstreamreader_readlines(MultibyteStreamReaderObject * self,PyObject * args)1380 mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args)
1381 {
1382     PyObject *sizehintobj = NULL, *r, *sr;
1383     Py_ssize_t sizehint;
1384 
1385     if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj))
1386         return NULL;
1387 
1388     if (sizehintobj == Py_None || sizehintobj == NULL)
1389         sizehint = -1;
1390     else if (PyInt_Check(sizehintobj))
1391         sizehint = PyInt_AsSsize_t(sizehintobj);
1392     else {
1393         PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer");
1394         return NULL;
1395     }
1396 
1397     r = mbstreamreader_iread(self, "read", sizehint);
1398     if (r == NULL)
1399         return NULL;
1400 
1401     sr = PyUnicode_Splitlines(r, 1);
1402     Py_DECREF(r);
1403     return sr;
1404 }
1405 
1406 static PyObject *
mbstreamreader_reset(MultibyteStreamReaderObject * self)1407 mbstreamreader_reset(MultibyteStreamReaderObject *self)
1408 {
1409     if (self->codec->decreset != NULL &&
1410         self->codec->decreset(&self->state, self->codec->config) != 0)
1411         return NULL;
1412     self->pendingsize = 0;
1413 
1414     Py_RETURN_NONE;
1415 }
1416 
1417 static struct PyMethodDef mbstreamreader_methods[] = {
1418     {"read",            (PyCFunction)mbstreamreader_read,
1419                     METH_VARARGS, NULL},
1420     {"readline",        (PyCFunction)mbstreamreader_readline,
1421                     METH_VARARGS, NULL},
1422     {"readlines",       (PyCFunction)mbstreamreader_readlines,
1423                     METH_VARARGS, NULL},
1424     {"reset",           (PyCFunction)mbstreamreader_reset,
1425                     METH_NOARGS, NULL},
1426     {NULL,              NULL},
1427 };
1428 
1429 static PyMemberDef mbstreamreader_members[] = {
1430     {"stream",          T_OBJECT,
1431                     offsetof(MultibyteStreamReaderObject, stream),
1432                     READONLY, NULL},
1433     {NULL,}
1434 };
1435 
1436 static PyObject *
mbstreamreader_new(PyTypeObject * type,PyObject * args,PyObject * kwds)1437 mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1438 {
1439     MultibyteStreamReaderObject *self;
1440     PyObject *stream, *codec = NULL;
1441     char *errors = NULL;
1442 
1443     if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader",
1444                             streamkwarglist, &stream, &errors))
1445         return NULL;
1446 
1447     self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0);
1448     if (self == NULL)
1449         return NULL;
1450 
1451     codec = PyObject_GetAttrString((PyObject *)type, "codec");
1452     if (codec == NULL)
1453         goto errorexit;
1454     if (!MultibyteCodec_Check(codec)) {
1455         PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1456         goto errorexit;
1457     }
1458 
1459     self->codec = ((MultibyteCodecObject *)codec)->codec;
1460     self->stream = stream;
1461     Py_INCREF(stream);
1462     self->pendingsize = 0;
1463     self->errors = internal_error_callback(errors);
1464     if (self->errors == NULL)
1465         goto errorexit;
1466     if (self->codec->decinit != NULL &&
1467         self->codec->decinit(&self->state, self->codec->config) != 0)
1468         goto errorexit;
1469 
1470     Py_DECREF(codec);
1471     return (PyObject *)self;
1472 
1473 errorexit:
1474     Py_XDECREF(self);
1475     Py_XDECREF(codec);
1476     return NULL;
1477 }
1478 
1479 static int
mbstreamreader_init(PyObject * self,PyObject * args,PyObject * kwds)1480 mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds)
1481 {
1482     return 0;
1483 }
1484 
1485 static int
mbstreamreader_traverse(MultibyteStreamReaderObject * self,visitproc visit,void * arg)1486 mbstreamreader_traverse(MultibyteStreamReaderObject *self,
1487                         visitproc visit, void *arg)
1488 {
1489     if (ERROR_ISCUSTOM(self->errors))
1490         Py_VISIT(self->errors);
1491     Py_VISIT(self->stream);
1492     return 0;
1493 }
1494 
1495 static void
mbstreamreader_dealloc(MultibyteStreamReaderObject * self)1496 mbstreamreader_dealloc(MultibyteStreamReaderObject *self)
1497 {
1498     PyObject_GC_UnTrack(self);
1499     ERROR_DECREF(self->errors);
1500     Py_XDECREF(self->stream);
1501     Py_TYPE(self)->tp_free(self);
1502 }
1503 
1504 static PyTypeObject MultibyteStreamReader_Type = {
1505     PyVarObject_HEAD_INIT(NULL, 0)
1506     "MultibyteStreamReader",            /* tp_name */
1507     sizeof(MultibyteStreamReaderObject), /* tp_basicsize */
1508     0,                                  /* tp_itemsize */
1509     /*  methods  */
1510     (destructor)mbstreamreader_dealloc, /* tp_dealloc */
1511     0,                                  /* tp_print */
1512     0,                                  /* tp_getattr */
1513     0,                                  /* tp_setattr */
1514     0,                                  /* tp_compare */
1515     0,                                  /* tp_repr */
1516     0,                                  /* tp_as_number */
1517     0,                                  /* tp_as_sequence */
1518     0,                                  /* tp_as_mapping */
1519     0,                                  /* tp_hash */
1520     0,                                  /* tp_call */
1521     0,                                  /* tp_str */
1522     PyObject_GenericGetAttr,            /* tp_getattro */
1523     0,                                  /* tp_setattro */
1524     0,                                  /* tp_as_buffer */
1525     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1526         | Py_TPFLAGS_BASETYPE,          /* tp_flags */
1527     0,                                  /* tp_doc */
1528     (traverseproc)mbstreamreader_traverse,      /* tp_traverse */
1529     0,                                  /* tp_clear */
1530     0,                                  /* tp_richcompare */
1531     0,                                  /* tp_weaklistoffset */
1532     0,                                  /* tp_iter */
1533     0,                                  /* tp_iterext */
1534     mbstreamreader_methods,             /* tp_methods */
1535     mbstreamreader_members,             /* tp_members */
1536     codecctx_getsets,                   /* tp_getset */
1537     0,                                  /* tp_base */
1538     0,                                  /* tp_dict */
1539     0,                                  /* tp_descr_get */
1540     0,                                  /* tp_descr_set */
1541     0,                                  /* tp_dictoffset */
1542     mbstreamreader_init,                /* tp_init */
1543     0,                                  /* tp_alloc */
1544     mbstreamreader_new,                 /* tp_new */
1545 };
1546 
1547 
1548 /**
1549  * MultibyteStreamWriter object
1550  */
1551 
1552 static int
mbstreamwriter_iwrite(MultibyteStreamWriterObject * self,PyObject * unistr)1553 mbstreamwriter_iwrite(MultibyteStreamWriterObject *self,
1554                       PyObject *unistr)
1555 {
1556     PyObject *str, *wr;
1557 
1558     str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0);
1559     if (str == NULL)
1560         return -1;
1561 
1562     wr = PyObject_CallMethod(self->stream, "write", "O", str);
1563     Py_DECREF(str);
1564     if (wr == NULL)
1565         return -1;
1566 
1567     Py_DECREF(wr);
1568     return 0;
1569 }
1570 
1571 static PyObject *
mbstreamwriter_write(MultibyteStreamWriterObject * self,PyObject * strobj)1572 mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj)
1573 {
1574     if (mbstreamwriter_iwrite(self, strobj))
1575         return NULL;
1576     else
1577         Py_RETURN_NONE;
1578 }
1579 
1580 static PyObject *
mbstreamwriter_writelines(MultibyteStreamWriterObject * self,PyObject * lines)1581 mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines)
1582 {
1583     PyObject *strobj;
1584     int i, r;
1585 
1586     if (!PySequence_Check(lines)) {
1587         PyErr_SetString(PyExc_TypeError,
1588                         "arg must be a sequence object");
1589         return NULL;
1590     }
1591 
1592     for (i = 0; i < PySequence_Length(lines); i++) {
1593         /* length can be changed even within this loop */
1594         strobj = PySequence_GetItem(lines, i);
1595         if (strobj == NULL)
1596             return NULL;
1597 
1598         r = mbstreamwriter_iwrite(self, strobj);
1599         Py_DECREF(strobj);
1600         if (r == -1)
1601             return NULL;
1602     }
1603 
1604     Py_RETURN_NONE;
1605 }
1606 
1607 static PyObject *
mbstreamwriter_reset(MultibyteStreamWriterObject * self)1608 mbstreamwriter_reset(MultibyteStreamWriterObject *self)
1609 {
1610     const Py_UNICODE *pending;
1611     PyObject *pwrt;
1612 
1613     pending = self->pending;
1614     pwrt = multibytecodec_encode(self->codec, &self->state,
1615                     &pending, self->pendingsize, self->errors,
1616                     MBENC_FLUSH | MBENC_RESET);
1617     /* some pending buffer can be truncated when UnicodeEncodeError is
1618      * raised on 'strict' mode. but, 'reset' method is designed to
1619      * reset the pending buffer or states so failed string sequence
1620      * ought to be missed */
1621     self->pendingsize = 0;
1622     if (pwrt == NULL)
1623         return NULL;
1624 
1625     if (PyString_Size(pwrt) > 0) {
1626         PyObject *wr;
1627         wr = PyObject_CallMethod(self->stream, "write", "O", pwrt);
1628         if (wr == NULL) {
1629             Py_DECREF(pwrt);
1630             return NULL;
1631         }
1632     }
1633     Py_DECREF(pwrt);
1634 
1635     Py_RETURN_NONE;
1636 }
1637 
1638 static PyObject *
mbstreamwriter_new(PyTypeObject * type,PyObject * args,PyObject * kwds)1639 mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
1640 {
1641     MultibyteStreamWriterObject *self;
1642     PyObject *stream, *codec = NULL;
1643     char *errors = NULL;
1644 
1645     if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter",
1646                             streamkwarglist, &stream, &errors))
1647         return NULL;
1648 
1649     self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0);
1650     if (self == NULL)
1651         return NULL;
1652 
1653     codec = PyObject_GetAttrString((PyObject *)type, "codec");
1654     if (codec == NULL)
1655         goto errorexit;
1656     if (!MultibyteCodec_Check(codec)) {
1657         PyErr_SetString(PyExc_TypeError, "codec is unexpected type");
1658         goto errorexit;
1659     }
1660 
1661     self->codec = ((MultibyteCodecObject *)codec)->codec;
1662     self->stream = stream;
1663     Py_INCREF(stream);
1664     self->pendingsize = 0;
1665     self->errors = internal_error_callback(errors);
1666     if (self->errors == NULL)
1667         goto errorexit;
1668     if (self->codec->encinit != NULL &&
1669         self->codec->encinit(&self->state, self->codec->config) != 0)
1670         goto errorexit;
1671 
1672     Py_DECREF(codec);
1673     return (PyObject *)self;
1674 
1675 errorexit:
1676     Py_XDECREF(self);
1677     Py_XDECREF(codec);
1678     return NULL;
1679 }
1680 
1681 static int
mbstreamwriter_init(PyObject * self,PyObject * args,PyObject * kwds)1682 mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds)
1683 {
1684     return 0;
1685 }
1686 
1687 static int
mbstreamwriter_traverse(MultibyteStreamWriterObject * self,visitproc visit,void * arg)1688 mbstreamwriter_traverse(MultibyteStreamWriterObject *self,
1689                         visitproc visit, void *arg)
1690 {
1691     if (ERROR_ISCUSTOM(self->errors))
1692         Py_VISIT(self->errors);
1693     Py_VISIT(self->stream);
1694     return 0;
1695 }
1696 
1697 static void
mbstreamwriter_dealloc(MultibyteStreamWriterObject * self)1698 mbstreamwriter_dealloc(MultibyteStreamWriterObject *self)
1699 {
1700     PyObject_GC_UnTrack(self);
1701     ERROR_DECREF(self->errors);
1702     Py_XDECREF(self->stream);
1703     Py_TYPE(self)->tp_free(self);
1704 }
1705 
1706 static struct PyMethodDef mbstreamwriter_methods[] = {
1707     {"write",           (PyCFunction)mbstreamwriter_write,
1708                     METH_O, NULL},
1709     {"writelines",      (PyCFunction)mbstreamwriter_writelines,
1710                     METH_O, NULL},
1711     {"reset",           (PyCFunction)mbstreamwriter_reset,
1712                     METH_NOARGS, NULL},
1713     {NULL,              NULL},
1714 };
1715 
1716 static PyMemberDef mbstreamwriter_members[] = {
1717     {"stream",          T_OBJECT,
1718                     offsetof(MultibyteStreamWriterObject, stream),
1719                     READONLY, NULL},
1720     {NULL,}
1721 };
1722 
1723 static PyTypeObject MultibyteStreamWriter_Type = {
1724     PyVarObject_HEAD_INIT(NULL, 0)
1725     "MultibyteStreamWriter",            /* tp_name */
1726     sizeof(MultibyteStreamWriterObject), /* tp_basicsize */
1727     0,                                  /* tp_itemsize */
1728     /*  methods  */
1729     (destructor)mbstreamwriter_dealloc, /* tp_dealloc */
1730     0,                                  /* tp_print */
1731     0,                                  /* tp_getattr */
1732     0,                                  /* tp_setattr */
1733     0,                                  /* tp_compare */
1734     0,                                  /* tp_repr */
1735     0,                                  /* tp_as_number */
1736     0,                                  /* tp_as_sequence */
1737     0,                                  /* tp_as_mapping */
1738     0,                                  /* tp_hash */
1739     0,                                  /* tp_call */
1740     0,                                  /* tp_str */
1741     PyObject_GenericGetAttr,            /* tp_getattro */
1742     0,                                  /* tp_setattro */
1743     0,                                  /* tp_as_buffer */
1744     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC
1745         | Py_TPFLAGS_BASETYPE,          /* tp_flags */
1746     0,                                  /* tp_doc */
1747     (traverseproc)mbstreamwriter_traverse,      /* tp_traverse */
1748     0,                                  /* tp_clear */
1749     0,                                  /* tp_richcompare */
1750     0,                                  /* tp_weaklistoffset */
1751     0,                                  /* tp_iter */
1752     0,                                  /* tp_iterext */
1753     mbstreamwriter_methods,             /* tp_methods */
1754     mbstreamwriter_members,             /* tp_members */
1755     codecctx_getsets,                   /* tp_getset */
1756     0,                                  /* tp_base */
1757     0,                                  /* tp_dict */
1758     0,                                  /* tp_descr_get */
1759     0,                                  /* tp_descr_set */
1760     0,                                  /* tp_dictoffset */
1761     mbstreamwriter_init,                /* tp_init */
1762     0,                                  /* tp_alloc */
1763     mbstreamwriter_new,                 /* tp_new */
1764 };
1765 
1766 
1767 /**
1768  * Exposed factory function
1769  */
1770 
1771 static PyObject *
__create_codec(PyObject * ignore,PyObject * arg)1772 __create_codec(PyObject *ignore, PyObject *arg)
1773 {
1774     MultibyteCodecObject *self;
1775     MultibyteCodec *codec;
1776 
1777     if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) {
1778         PyErr_SetString(PyExc_ValueError, "argument type invalid");
1779         return NULL;
1780     }
1781 
1782     codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME);
1783     if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0)
1784         return NULL;
1785 
1786     self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type);
1787     if (self == NULL)
1788         return NULL;
1789     self->codec = codec;
1790 
1791     return (PyObject *)self;
1792 }
1793 
1794 static struct PyMethodDef __methods[] = {
1795     {"__create_codec", (PyCFunction)__create_codec, METH_O},
1796     {NULL, NULL},
1797 };
1798 
1799 PyMODINIT_FUNC
init_multibytecodec(void)1800 init_multibytecodec(void)
1801 {
1802     int i;
1803     PyObject *m;
1804     PyTypeObject *typelist[] = {
1805         &MultibyteIncrementalEncoder_Type,
1806         &MultibyteIncrementalDecoder_Type,
1807         &MultibyteStreamReader_Type,
1808         &MultibyteStreamWriter_Type,
1809         NULL
1810     };
1811 
1812     if (PyType_Ready(&MultibyteCodec_Type) < 0)
1813         return;
1814 
1815     m = Py_InitModule("_multibytecodec", __methods);
1816     if (m == NULL)
1817         return;
1818 
1819     for (i = 0; typelist[i] != NULL; i++) {
1820         if (PyType_Ready(typelist[i]) < 0)
1821             return;
1822         Py_INCREF(typelist[i]);
1823         PyModule_AddObject(m, typelist[i]->tp_name,
1824                            (PyObject *)typelist[i]);
1825     }
1826 
1827     if (PyErr_Occurred())
1828         Py_FatalError("can't initialize the _multibytecodec module");
1829 }
1830