1 | n/a | /* |
---|
2 | n/a | * multibytecodec.c: Common Multibyte Codec Implementation |
---|
3 | n/a | * |
---|
4 | n/a | * Written by Hye-Shik Chang <perky@FreeBSD.org> |
---|
5 | n/a | */ |
---|
6 | n/a | |
---|
7 | n/a | #define PY_SSIZE_T_CLEAN |
---|
8 | n/a | #include "Python.h" |
---|
9 | n/a | #include "structmember.h" |
---|
10 | n/a | #include "multibytecodec.h" |
---|
11 | n/a | #include "clinic/multibytecodec.c.h" |
---|
12 | n/a | |
---|
13 | n/a | /*[clinic input] |
---|
14 | n/a | module _multibytecodec |
---|
15 | n/a | class _multibytecodec.MultibyteCodec "MultibyteCodecObject *" "&MultibyteCodec_Type" |
---|
16 | n/a | [clinic start generated code]*/ |
---|
17 | n/a | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=6ad689546cbb5450]*/ |
---|
18 | n/a | |
---|
19 | n/a | typedef struct { |
---|
20 | n/a | PyObject *inobj; |
---|
21 | n/a | Py_ssize_t inpos, inlen; |
---|
22 | n/a | unsigned char *outbuf, *outbuf_end; |
---|
23 | n/a | PyObject *excobj, *outobj; |
---|
24 | n/a | } MultibyteEncodeBuffer; |
---|
25 | n/a | |
---|
26 | n/a | typedef struct { |
---|
27 | n/a | const unsigned char *inbuf, *inbuf_top, *inbuf_end; |
---|
28 | n/a | PyObject *excobj; |
---|
29 | n/a | _PyUnicodeWriter writer; |
---|
30 | n/a | } MultibyteDecodeBuffer; |
---|
31 | n/a | |
---|
32 | n/a | static char *incnewkwarglist[] = {"errors", NULL}; |
---|
33 | n/a | static char *streamkwarglist[] = {"stream", "errors", NULL}; |
---|
34 | n/a | |
---|
35 | n/a | static PyObject *multibytecodec_encode(MultibyteCodec *, |
---|
36 | n/a | MultibyteCodec_State *, PyObject *, Py_ssize_t *, |
---|
37 | n/a | PyObject *, int); |
---|
38 | n/a | |
---|
39 | n/a | #define MBENC_RESET MBENC_MAX<<1 /* reset after an encoding session */ |
---|
40 | n/a | |
---|
41 | n/a | _Py_IDENTIFIER(write); |
---|
42 | n/a | |
---|
43 | n/a | static PyObject * |
---|
44 | n/a | make_tuple(PyObject *object, Py_ssize_t len) |
---|
45 | n/a | { |
---|
46 | n/a | PyObject *v, *w; |
---|
47 | n/a | |
---|
48 | n/a | if (object == NULL) |
---|
49 | n/a | return NULL; |
---|
50 | n/a | |
---|
51 | n/a | v = PyTuple_New(2); |
---|
52 | n/a | if (v == NULL) { |
---|
53 | n/a | Py_DECREF(object); |
---|
54 | n/a | return NULL; |
---|
55 | n/a | } |
---|
56 | n/a | PyTuple_SET_ITEM(v, 0, object); |
---|
57 | n/a | |
---|
58 | n/a | w = PyLong_FromSsize_t(len); |
---|
59 | n/a | if (w == NULL) { |
---|
60 | n/a | Py_DECREF(v); |
---|
61 | n/a | return NULL; |
---|
62 | n/a | } |
---|
63 | n/a | PyTuple_SET_ITEM(v, 1, w); |
---|
64 | n/a | |
---|
65 | n/a | return v; |
---|
66 | n/a | } |
---|
67 | n/a | |
---|
68 | n/a | static PyObject * |
---|
69 | n/a | internal_error_callback(const char *errors) |
---|
70 | n/a | { |
---|
71 | n/a | if (errors == NULL || strcmp(errors, "strict") == 0) |
---|
72 | n/a | return ERROR_STRICT; |
---|
73 | n/a | else if (strcmp(errors, "ignore") == 0) |
---|
74 | n/a | return ERROR_IGNORE; |
---|
75 | n/a | else if (strcmp(errors, "replace") == 0) |
---|
76 | n/a | return ERROR_REPLACE; |
---|
77 | n/a | else |
---|
78 | n/a | return PyUnicode_FromString(errors); |
---|
79 | n/a | } |
---|
80 | n/a | |
---|
81 | n/a | static PyObject * |
---|
82 | n/a | call_error_callback(PyObject *errors, PyObject *exc) |
---|
83 | n/a | { |
---|
84 | n/a | PyObject *args, *cb, *r; |
---|
85 | n/a | const char *str; |
---|
86 | n/a | |
---|
87 | n/a | assert(PyUnicode_Check(errors)); |
---|
88 | n/a | str = PyUnicode_AsUTF8(errors); |
---|
89 | n/a | if (str == NULL) |
---|
90 | n/a | return NULL; |
---|
91 | n/a | cb = PyCodec_LookupError(str); |
---|
92 | n/a | if (cb == NULL) |
---|
93 | n/a | return NULL; |
---|
94 | n/a | |
---|
95 | n/a | args = PyTuple_New(1); |
---|
96 | n/a | if (args == NULL) { |
---|
97 | n/a | Py_DECREF(cb); |
---|
98 | n/a | return NULL; |
---|
99 | n/a | } |
---|
100 | n/a | |
---|
101 | n/a | PyTuple_SET_ITEM(args, 0, exc); |
---|
102 | n/a | Py_INCREF(exc); |
---|
103 | n/a | |
---|
104 | n/a | r = PyObject_CallObject(cb, args); |
---|
105 | n/a | Py_DECREF(args); |
---|
106 | n/a | Py_DECREF(cb); |
---|
107 | n/a | return r; |
---|
108 | n/a | } |
---|
109 | n/a | |
---|
110 | n/a | static PyObject * |
---|
111 | n/a | codecctx_errors_get(MultibyteStatefulCodecContext *self) |
---|
112 | n/a | { |
---|
113 | n/a | const char *errors; |
---|
114 | n/a | |
---|
115 | n/a | if (self->errors == ERROR_STRICT) |
---|
116 | n/a | errors = "strict"; |
---|
117 | n/a | else if (self->errors == ERROR_IGNORE) |
---|
118 | n/a | errors = "ignore"; |
---|
119 | n/a | else if (self->errors == ERROR_REPLACE) |
---|
120 | n/a | errors = "replace"; |
---|
121 | n/a | else { |
---|
122 | n/a | Py_INCREF(self->errors); |
---|
123 | n/a | return self->errors; |
---|
124 | n/a | } |
---|
125 | n/a | |
---|
126 | n/a | return PyUnicode_FromString(errors); |
---|
127 | n/a | } |
---|
128 | n/a | |
---|
129 | n/a | static int |
---|
130 | n/a | codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value, |
---|
131 | n/a | void *closure) |
---|
132 | n/a | { |
---|
133 | n/a | PyObject *cb; |
---|
134 | n/a | const char *str; |
---|
135 | n/a | |
---|
136 | n/a | if (!PyUnicode_Check(value)) { |
---|
137 | n/a | PyErr_SetString(PyExc_TypeError, "errors must be a string"); |
---|
138 | n/a | return -1; |
---|
139 | n/a | } |
---|
140 | n/a | |
---|
141 | n/a | str = PyUnicode_AsUTF8(value); |
---|
142 | n/a | if (str == NULL) |
---|
143 | n/a | return -1; |
---|
144 | n/a | |
---|
145 | n/a | cb = internal_error_callback(str); |
---|
146 | n/a | if (cb == NULL) |
---|
147 | n/a | return -1; |
---|
148 | n/a | |
---|
149 | n/a | ERROR_DECREF(self->errors); |
---|
150 | n/a | self->errors = cb; |
---|
151 | n/a | return 0; |
---|
152 | n/a | } |
---|
153 | n/a | |
---|
154 | n/a | /* This getset handlers list is used by all the stateful codec objects */ |
---|
155 | n/a | static PyGetSetDef codecctx_getsets[] = { |
---|
156 | n/a | {"errors", (getter)codecctx_errors_get, |
---|
157 | n/a | (setter)codecctx_errors_set, |
---|
158 | n/a | PyDoc_STR("how to treat errors")}, |
---|
159 | n/a | {NULL,} |
---|
160 | n/a | }; |
---|
161 | n/a | |
---|
162 | n/a | static int |
---|
163 | n/a | expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize) |
---|
164 | n/a | { |
---|
165 | n/a | Py_ssize_t orgpos, orgsize, incsize; |
---|
166 | n/a | |
---|
167 | n/a | orgpos = (Py_ssize_t)((char *)buf->outbuf - |
---|
168 | n/a | PyBytes_AS_STRING(buf->outobj)); |
---|
169 | n/a | orgsize = PyBytes_GET_SIZE(buf->outobj); |
---|
170 | n/a | incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize); |
---|
171 | n/a | |
---|
172 | n/a | if (orgsize > PY_SSIZE_T_MAX - incsize) { |
---|
173 | n/a | PyErr_NoMemory(); |
---|
174 | n/a | return -1; |
---|
175 | n/a | } |
---|
176 | n/a | |
---|
177 | n/a | if (_PyBytes_Resize(&buf->outobj, orgsize + incsize) == -1) |
---|
178 | n/a | return -1; |
---|
179 | n/a | |
---|
180 | n/a | buf->outbuf = (unsigned char *)PyBytes_AS_STRING(buf->outobj) +orgpos; |
---|
181 | n/a | buf->outbuf_end = (unsigned char *)PyBytes_AS_STRING(buf->outobj) |
---|
182 | n/a | + PyBytes_GET_SIZE(buf->outobj); |
---|
183 | n/a | |
---|
184 | n/a | return 0; |
---|
185 | n/a | } |
---|
186 | n/a | #define REQUIRE_ENCODEBUFFER(buf, s) do { \ |
---|
187 | n/a | if ((s) < 0 || (s) > (buf)->outbuf_end - (buf)->outbuf) \ |
---|
188 | n/a | if (expand_encodebuffer(buf, s) == -1) \ |
---|
189 | n/a | goto errorexit; \ |
---|
190 | n/a | } while(0) |
---|
191 | n/a | |
---|
192 | n/a | |
---|
193 | n/a | /** |
---|
194 | n/a | * MultibyteCodec object |
---|
195 | n/a | */ |
---|
196 | n/a | |
---|
197 | n/a | static int |
---|
198 | n/a | multibytecodec_encerror(MultibyteCodec *codec, |
---|
199 | n/a | MultibyteCodec_State *state, |
---|
200 | n/a | MultibyteEncodeBuffer *buf, |
---|
201 | n/a | PyObject *errors, Py_ssize_t e) |
---|
202 | n/a | { |
---|
203 | n/a | PyObject *retobj = NULL, *retstr = NULL, *tobj; |
---|
204 | n/a | Py_ssize_t retstrsize, newpos; |
---|
205 | n/a | Py_ssize_t esize, start, end; |
---|
206 | n/a | const char *reason; |
---|
207 | n/a | |
---|
208 | n/a | if (e > 0) { |
---|
209 | n/a | reason = "illegal multibyte sequence"; |
---|
210 | n/a | esize = e; |
---|
211 | n/a | } |
---|
212 | n/a | else { |
---|
213 | n/a | switch (e) { |
---|
214 | n/a | case MBERR_TOOSMALL: |
---|
215 | n/a | REQUIRE_ENCODEBUFFER(buf, -1); |
---|
216 | n/a | return 0; /* retry it */ |
---|
217 | n/a | case MBERR_TOOFEW: |
---|
218 | n/a | reason = "incomplete multibyte sequence"; |
---|
219 | n/a | esize = (Py_ssize_t)buf->inpos; |
---|
220 | n/a | break; |
---|
221 | n/a | case MBERR_INTERNAL: |
---|
222 | n/a | PyErr_SetString(PyExc_RuntimeError, |
---|
223 | n/a | "internal codec error"); |
---|
224 | n/a | return -1; |
---|
225 | n/a | default: |
---|
226 | n/a | PyErr_SetString(PyExc_RuntimeError, |
---|
227 | n/a | "unknown runtime error"); |
---|
228 | n/a | return -1; |
---|
229 | n/a | } |
---|
230 | n/a | } |
---|
231 | n/a | |
---|
232 | n/a | if (errors == ERROR_REPLACE) { |
---|
233 | n/a | PyObject *replchar; |
---|
234 | n/a | Py_ssize_t r; |
---|
235 | n/a | Py_ssize_t inpos; |
---|
236 | n/a | int kind; |
---|
237 | n/a | void *data; |
---|
238 | n/a | |
---|
239 | n/a | replchar = PyUnicode_FromOrdinal('?'); |
---|
240 | n/a | if (replchar == NULL) |
---|
241 | n/a | goto errorexit; |
---|
242 | n/a | kind = PyUnicode_KIND(replchar); |
---|
243 | n/a | data = PyUnicode_DATA(replchar); |
---|
244 | n/a | |
---|
245 | n/a | inpos = 0; |
---|
246 | n/a | for (;;) { |
---|
247 | n/a | Py_ssize_t outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf); |
---|
248 | n/a | |
---|
249 | n/a | r = codec->encode(state, codec->config, |
---|
250 | n/a | kind, data, &inpos, 1, |
---|
251 | n/a | &buf->outbuf, outleft, 0); |
---|
252 | n/a | if (r == MBERR_TOOSMALL) { |
---|
253 | n/a | REQUIRE_ENCODEBUFFER(buf, -1); |
---|
254 | n/a | continue; |
---|
255 | n/a | } |
---|
256 | n/a | else |
---|
257 | n/a | break; |
---|
258 | n/a | } |
---|
259 | n/a | |
---|
260 | n/a | Py_DECREF(replchar); |
---|
261 | n/a | |
---|
262 | n/a | if (r != 0) { |
---|
263 | n/a | REQUIRE_ENCODEBUFFER(buf, 1); |
---|
264 | n/a | *buf->outbuf++ = '?'; |
---|
265 | n/a | } |
---|
266 | n/a | } |
---|
267 | n/a | if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { |
---|
268 | n/a | buf->inpos += esize; |
---|
269 | n/a | return 0; |
---|
270 | n/a | } |
---|
271 | n/a | |
---|
272 | n/a | start = (Py_ssize_t)buf->inpos; |
---|
273 | n/a | end = start + esize; |
---|
274 | n/a | |
---|
275 | n/a | /* use cached exception object if available */ |
---|
276 | n/a | if (buf->excobj == NULL) { |
---|
277 | n/a | buf->excobj = PyObject_CallFunction(PyExc_UnicodeEncodeError, |
---|
278 | n/a | "sOnns", |
---|
279 | n/a | codec->encoding, buf->inobj, |
---|
280 | n/a | start, end, reason); |
---|
281 | n/a | if (buf->excobj == NULL) |
---|
282 | n/a | goto errorexit; |
---|
283 | n/a | } |
---|
284 | n/a | else |
---|
285 | n/a | if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 || |
---|
286 | n/a | PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 || |
---|
287 | n/a | PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0) |
---|
288 | n/a | goto errorexit; |
---|
289 | n/a | |
---|
290 | n/a | if (errors == ERROR_STRICT) { |
---|
291 | n/a | PyCodec_StrictErrors(buf->excobj); |
---|
292 | n/a | goto errorexit; |
---|
293 | n/a | } |
---|
294 | n/a | |
---|
295 | n/a | retobj = call_error_callback(errors, buf->excobj); |
---|
296 | n/a | if (retobj == NULL) |
---|
297 | n/a | goto errorexit; |
---|
298 | n/a | |
---|
299 | n/a | if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || |
---|
300 | n/a | (!PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) && !PyBytes_Check(tobj)) || |
---|
301 | n/a | !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { |
---|
302 | n/a | PyErr_SetString(PyExc_TypeError, |
---|
303 | n/a | "encoding error handler must return " |
---|
304 | n/a | "(str, int) tuple"); |
---|
305 | n/a | goto errorexit; |
---|
306 | n/a | } |
---|
307 | n/a | |
---|
308 | n/a | if (PyUnicode_Check(tobj)) { |
---|
309 | n/a | Py_ssize_t inpos; |
---|
310 | n/a | |
---|
311 | n/a | retstr = multibytecodec_encode(codec, state, tobj, |
---|
312 | n/a | &inpos, ERROR_STRICT, |
---|
313 | n/a | MBENC_FLUSH); |
---|
314 | n/a | if (retstr == NULL) |
---|
315 | n/a | goto errorexit; |
---|
316 | n/a | } |
---|
317 | n/a | else { |
---|
318 | n/a | Py_INCREF(tobj); |
---|
319 | n/a | retstr = tobj; |
---|
320 | n/a | } |
---|
321 | n/a | |
---|
322 | n/a | assert(PyBytes_Check(retstr)); |
---|
323 | n/a | retstrsize = PyBytes_GET_SIZE(retstr); |
---|
324 | n/a | if (retstrsize > 0) { |
---|
325 | n/a | REQUIRE_ENCODEBUFFER(buf, retstrsize); |
---|
326 | n/a | memcpy(buf->outbuf, PyBytes_AS_STRING(retstr), retstrsize); |
---|
327 | n/a | buf->outbuf += retstrsize; |
---|
328 | n/a | } |
---|
329 | n/a | |
---|
330 | n/a | newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); |
---|
331 | n/a | if (newpos < 0 && !PyErr_Occurred()) |
---|
332 | n/a | newpos += (Py_ssize_t)buf->inlen; |
---|
333 | n/a | if (newpos < 0 || newpos > buf->inlen) { |
---|
334 | n/a | PyErr_Clear(); |
---|
335 | n/a | PyErr_Format(PyExc_IndexError, |
---|
336 | n/a | "position %zd from error handler out of bounds", |
---|
337 | n/a | newpos); |
---|
338 | n/a | goto errorexit; |
---|
339 | n/a | } |
---|
340 | n/a | buf->inpos = newpos; |
---|
341 | n/a | |
---|
342 | n/a | Py_DECREF(retobj); |
---|
343 | n/a | Py_DECREF(retstr); |
---|
344 | n/a | return 0; |
---|
345 | n/a | |
---|
346 | n/a | errorexit: |
---|
347 | n/a | Py_XDECREF(retobj); |
---|
348 | n/a | Py_XDECREF(retstr); |
---|
349 | n/a | return -1; |
---|
350 | n/a | } |
---|
351 | n/a | |
---|
352 | n/a | static int |
---|
353 | n/a | multibytecodec_decerror(MultibyteCodec *codec, |
---|
354 | n/a | MultibyteCodec_State *state, |
---|
355 | n/a | MultibyteDecodeBuffer *buf, |
---|
356 | n/a | PyObject *errors, Py_ssize_t e) |
---|
357 | n/a | { |
---|
358 | n/a | PyObject *retobj = NULL, *retuni = NULL; |
---|
359 | n/a | Py_ssize_t newpos; |
---|
360 | n/a | const char *reason; |
---|
361 | n/a | Py_ssize_t esize, start, end; |
---|
362 | n/a | |
---|
363 | n/a | if (e > 0) { |
---|
364 | n/a | reason = "illegal multibyte sequence"; |
---|
365 | n/a | esize = e; |
---|
366 | n/a | } |
---|
367 | n/a | else { |
---|
368 | n/a | switch (e) { |
---|
369 | n/a | case MBERR_TOOSMALL: |
---|
370 | n/a | return 0; /* retry it */ |
---|
371 | n/a | case MBERR_TOOFEW: |
---|
372 | n/a | reason = "incomplete multibyte sequence"; |
---|
373 | n/a | esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
---|
374 | n/a | break; |
---|
375 | n/a | case MBERR_INTERNAL: |
---|
376 | n/a | PyErr_SetString(PyExc_RuntimeError, |
---|
377 | n/a | "internal codec error"); |
---|
378 | n/a | return -1; |
---|
379 | n/a | case MBERR_EXCEPTION: |
---|
380 | n/a | return -1; |
---|
381 | n/a | default: |
---|
382 | n/a | PyErr_SetString(PyExc_RuntimeError, |
---|
383 | n/a | "unknown runtime error"); |
---|
384 | n/a | return -1; |
---|
385 | n/a | } |
---|
386 | n/a | } |
---|
387 | n/a | |
---|
388 | n/a | if (errors == ERROR_REPLACE) { |
---|
389 | n/a | if (_PyUnicodeWriter_WriteChar(&buf->writer, |
---|
390 | n/a | Py_UNICODE_REPLACEMENT_CHARACTER) < 0) |
---|
391 | n/a | goto errorexit; |
---|
392 | n/a | } |
---|
393 | n/a | if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { |
---|
394 | n/a | buf->inbuf += esize; |
---|
395 | n/a | return 0; |
---|
396 | n/a | } |
---|
397 | n/a | |
---|
398 | n/a | start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top); |
---|
399 | n/a | end = start + esize; |
---|
400 | n/a | |
---|
401 | n/a | /* use cached exception object if available */ |
---|
402 | n/a | if (buf->excobj == NULL) { |
---|
403 | n/a | buf->excobj = PyUnicodeDecodeError_Create(codec->encoding, |
---|
404 | n/a | (const char *)buf->inbuf_top, |
---|
405 | n/a | (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top), |
---|
406 | n/a | start, end, reason); |
---|
407 | n/a | if (buf->excobj == NULL) |
---|
408 | n/a | goto errorexit; |
---|
409 | n/a | } |
---|
410 | n/a | else |
---|
411 | n/a | if (PyUnicodeDecodeError_SetStart(buf->excobj, start) || |
---|
412 | n/a | PyUnicodeDecodeError_SetEnd(buf->excobj, end) || |
---|
413 | n/a | PyUnicodeDecodeError_SetReason(buf->excobj, reason)) |
---|
414 | n/a | goto errorexit; |
---|
415 | n/a | |
---|
416 | n/a | if (errors == ERROR_STRICT) { |
---|
417 | n/a | PyCodec_StrictErrors(buf->excobj); |
---|
418 | n/a | goto errorexit; |
---|
419 | n/a | } |
---|
420 | n/a | |
---|
421 | n/a | retobj = call_error_callback(errors, buf->excobj); |
---|
422 | n/a | if (retobj == NULL) |
---|
423 | n/a | goto errorexit; |
---|
424 | n/a | |
---|
425 | n/a | if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || |
---|
426 | n/a | !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) || |
---|
427 | n/a | !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { |
---|
428 | n/a | PyErr_SetString(PyExc_TypeError, |
---|
429 | n/a | "decoding error handler must return " |
---|
430 | n/a | "(str, int) tuple"); |
---|
431 | n/a | goto errorexit; |
---|
432 | n/a | } |
---|
433 | n/a | |
---|
434 | n/a | if (_PyUnicodeWriter_WriteStr(&buf->writer, retuni) < 0) |
---|
435 | n/a | goto errorexit; |
---|
436 | n/a | |
---|
437 | n/a | newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); |
---|
438 | n/a | if (newpos < 0 && !PyErr_Occurred()) |
---|
439 | n/a | newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top); |
---|
440 | n/a | if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) { |
---|
441 | n/a | PyErr_Clear(); |
---|
442 | n/a | PyErr_Format(PyExc_IndexError, |
---|
443 | n/a | "position %zd from error handler out of bounds", |
---|
444 | n/a | newpos); |
---|
445 | n/a | goto errorexit; |
---|
446 | n/a | } |
---|
447 | n/a | buf->inbuf = buf->inbuf_top + newpos; |
---|
448 | n/a | Py_DECREF(retobj); |
---|
449 | n/a | return 0; |
---|
450 | n/a | |
---|
451 | n/a | errorexit: |
---|
452 | n/a | Py_XDECREF(retobj); |
---|
453 | n/a | return -1; |
---|
454 | n/a | } |
---|
455 | n/a | |
---|
456 | n/a | static PyObject * |
---|
457 | n/a | multibytecodec_encode(MultibyteCodec *codec, |
---|
458 | n/a | MultibyteCodec_State *state, |
---|
459 | n/a | PyObject *text, Py_ssize_t *inpos_t, |
---|
460 | n/a | PyObject *errors, int flags) |
---|
461 | n/a | { |
---|
462 | n/a | MultibyteEncodeBuffer buf; |
---|
463 | n/a | Py_ssize_t finalsize, r = 0; |
---|
464 | n/a | Py_ssize_t datalen; |
---|
465 | n/a | int kind; |
---|
466 | n/a | void *data; |
---|
467 | n/a | |
---|
468 | n/a | if (PyUnicode_READY(text) < 0) |
---|
469 | n/a | return NULL; |
---|
470 | n/a | datalen = PyUnicode_GET_LENGTH(text); |
---|
471 | n/a | |
---|
472 | n/a | if (datalen == 0 && !(flags & MBENC_RESET)) |
---|
473 | n/a | return PyBytes_FromStringAndSize(NULL, 0); |
---|
474 | n/a | |
---|
475 | n/a | buf.excobj = NULL; |
---|
476 | n/a | buf.outobj = NULL; |
---|
477 | n/a | buf.inobj = text; /* borrowed reference */ |
---|
478 | n/a | buf.inpos = 0; |
---|
479 | n/a | buf.inlen = datalen; |
---|
480 | n/a | kind = PyUnicode_KIND(buf.inobj); |
---|
481 | n/a | data = PyUnicode_DATA(buf.inobj); |
---|
482 | n/a | |
---|
483 | n/a | if (datalen > (PY_SSIZE_T_MAX - 16) / 2) { |
---|
484 | n/a | PyErr_NoMemory(); |
---|
485 | n/a | goto errorexit; |
---|
486 | n/a | } |
---|
487 | n/a | |
---|
488 | n/a | buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16); |
---|
489 | n/a | if (buf.outobj == NULL) |
---|
490 | n/a | goto errorexit; |
---|
491 | n/a | buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj); |
---|
492 | n/a | buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj); |
---|
493 | n/a | |
---|
494 | n/a | while (buf.inpos < buf.inlen) { |
---|
495 | n/a | /* we don't reuse inleft and outleft here. |
---|
496 | n/a | * error callbacks can relocate the cursor anywhere on buffer*/ |
---|
497 | n/a | Py_ssize_t outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
---|
498 | n/a | |
---|
499 | n/a | r = codec->encode(state, codec->config, |
---|
500 | n/a | kind, data, |
---|
501 | n/a | &buf.inpos, buf.inlen, |
---|
502 | n/a | &buf.outbuf, outleft, flags); |
---|
503 | n/a | if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH))) |
---|
504 | n/a | break; |
---|
505 | n/a | else if (multibytecodec_encerror(codec, state, &buf, errors,r)) |
---|
506 | n/a | goto errorexit; |
---|
507 | n/a | else if (r == MBERR_TOOFEW) |
---|
508 | n/a | break; |
---|
509 | n/a | } |
---|
510 | n/a | |
---|
511 | n/a | if (codec->encreset != NULL && (flags & MBENC_RESET)) |
---|
512 | n/a | for (;;) { |
---|
513 | n/a | Py_ssize_t outleft; |
---|
514 | n/a | |
---|
515 | n/a | outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
---|
516 | n/a | r = codec->encreset(state, codec->config, &buf.outbuf, |
---|
517 | n/a | outleft); |
---|
518 | n/a | if (r == 0) |
---|
519 | n/a | break; |
---|
520 | n/a | else if (multibytecodec_encerror(codec, state, |
---|
521 | n/a | &buf, errors, r)) |
---|
522 | n/a | goto errorexit; |
---|
523 | n/a | } |
---|
524 | n/a | |
---|
525 | n/a | finalsize = (Py_ssize_t)((char *)buf.outbuf - |
---|
526 | n/a | PyBytes_AS_STRING(buf.outobj)); |
---|
527 | n/a | |
---|
528 | n/a | if (finalsize != PyBytes_GET_SIZE(buf.outobj)) |
---|
529 | n/a | if (_PyBytes_Resize(&buf.outobj, finalsize) == -1) |
---|
530 | n/a | goto errorexit; |
---|
531 | n/a | |
---|
532 | n/a | if (inpos_t) |
---|
533 | n/a | *inpos_t = buf.inpos; |
---|
534 | n/a | Py_XDECREF(buf.excobj); |
---|
535 | n/a | return buf.outobj; |
---|
536 | n/a | |
---|
537 | n/a | errorexit: |
---|
538 | n/a | Py_XDECREF(buf.excobj); |
---|
539 | n/a | Py_XDECREF(buf.outobj); |
---|
540 | n/a | return NULL; |
---|
541 | n/a | } |
---|
542 | n/a | |
---|
543 | n/a | /*[clinic input] |
---|
544 | n/a | _multibytecodec.MultibyteCodec.encode |
---|
545 | n/a | |
---|
546 | n/a | input: object |
---|
547 | n/a | errors: str(accept={str, NoneType}) = NULL |
---|
548 | n/a | |
---|
549 | n/a | Return an encoded string version of `input'. |
---|
550 | n/a | |
---|
551 | n/a | 'errors' may be given to set a different error handling scheme. Default is |
---|
552 | n/a | 'strict' meaning that encoding errors raise a UnicodeEncodeError. Other possible |
---|
553 | n/a | values are 'ignore', 'replace' and 'xmlcharrefreplace' as well as any other name |
---|
554 | n/a | registered with codecs.register_error that can handle UnicodeEncodeErrors. |
---|
555 | n/a | [clinic start generated code]*/ |
---|
556 | n/a | |
---|
557 | n/a | static PyObject * |
---|
558 | n/a | _multibytecodec_MultibyteCodec_encode_impl(MultibyteCodecObject *self, |
---|
559 | n/a | PyObject *input, |
---|
560 | n/a | const char *errors) |
---|
561 | n/a | /*[clinic end generated code: output=7b26652045ba56a9 input=05f6ced3c8dd0582]*/ |
---|
562 | n/a | { |
---|
563 | n/a | MultibyteCodec_State state; |
---|
564 | n/a | PyObject *errorcb, *r, *ucvt; |
---|
565 | n/a | Py_ssize_t datalen; |
---|
566 | n/a | |
---|
567 | n/a | if (PyUnicode_Check(input)) |
---|
568 | n/a | ucvt = NULL; |
---|
569 | n/a | else { |
---|
570 | n/a | input = ucvt = PyObject_Str(input); |
---|
571 | n/a | if (input == NULL) |
---|
572 | n/a | return NULL; |
---|
573 | n/a | else if (!PyUnicode_Check(input)) { |
---|
574 | n/a | PyErr_SetString(PyExc_TypeError, |
---|
575 | n/a | "couldn't convert the object to unicode."); |
---|
576 | n/a | Py_DECREF(ucvt); |
---|
577 | n/a | return NULL; |
---|
578 | n/a | } |
---|
579 | n/a | } |
---|
580 | n/a | |
---|
581 | n/a | if (PyUnicode_READY(input) < 0) { |
---|
582 | n/a | Py_XDECREF(ucvt); |
---|
583 | n/a | return NULL; |
---|
584 | n/a | } |
---|
585 | n/a | datalen = PyUnicode_GET_LENGTH(input); |
---|
586 | n/a | |
---|
587 | n/a | errorcb = internal_error_callback(errors); |
---|
588 | n/a | if (errorcb == NULL) { |
---|
589 | n/a | Py_XDECREF(ucvt); |
---|
590 | n/a | return NULL; |
---|
591 | n/a | } |
---|
592 | n/a | |
---|
593 | n/a | if (self->codec->encinit != NULL && |
---|
594 | n/a | self->codec->encinit(&state, self->codec->config) != 0) |
---|
595 | n/a | goto errorexit; |
---|
596 | n/a | r = multibytecodec_encode(self->codec, &state, |
---|
597 | n/a | input, NULL, errorcb, |
---|
598 | n/a | MBENC_FLUSH | MBENC_RESET); |
---|
599 | n/a | if (r == NULL) |
---|
600 | n/a | goto errorexit; |
---|
601 | n/a | |
---|
602 | n/a | ERROR_DECREF(errorcb); |
---|
603 | n/a | Py_XDECREF(ucvt); |
---|
604 | n/a | return make_tuple(r, datalen); |
---|
605 | n/a | |
---|
606 | n/a | errorexit: |
---|
607 | n/a | ERROR_DECREF(errorcb); |
---|
608 | n/a | Py_XDECREF(ucvt); |
---|
609 | n/a | return NULL; |
---|
610 | n/a | } |
---|
611 | n/a | |
---|
612 | n/a | /*[clinic input] |
---|
613 | n/a | _multibytecodec.MultibyteCodec.decode |
---|
614 | n/a | |
---|
615 | n/a | input: Py_buffer |
---|
616 | n/a | errors: str(accept={str, NoneType}) = NULL |
---|
617 | n/a | |
---|
618 | n/a | Decodes 'input'. |
---|
619 | n/a | |
---|
620 | n/a | 'errors' may be given to set a different error handling scheme. Default is |
---|
621 | n/a | 'strict' meaning that encoding errors raise a UnicodeDecodeError. Other possible |
---|
622 | n/a | values are 'ignore' and 'replace' as well as any other name registered with |
---|
623 | n/a | codecs.register_error that is able to handle UnicodeDecodeErrors." |
---|
624 | n/a | [clinic start generated code]*/ |
---|
625 | n/a | |
---|
626 | n/a | static PyObject * |
---|
627 | n/a | _multibytecodec_MultibyteCodec_decode_impl(MultibyteCodecObject *self, |
---|
628 | n/a | Py_buffer *input, |
---|
629 | n/a | const char *errors) |
---|
630 | n/a | /*[clinic end generated code: output=ff419f65bad6cc77 input=a7d45f87f75e5e02]*/ |
---|
631 | n/a | { |
---|
632 | n/a | MultibyteCodec_State state; |
---|
633 | n/a | MultibyteDecodeBuffer buf; |
---|
634 | n/a | PyObject *errorcb, *res; |
---|
635 | n/a | const char *data; |
---|
636 | n/a | Py_ssize_t datalen; |
---|
637 | n/a | |
---|
638 | n/a | data = input->buf; |
---|
639 | n/a | datalen = input->len; |
---|
640 | n/a | |
---|
641 | n/a | errorcb = internal_error_callback(errors); |
---|
642 | n/a | if (errorcb == NULL) { |
---|
643 | n/a | return NULL; |
---|
644 | n/a | } |
---|
645 | n/a | |
---|
646 | n/a | if (datalen == 0) { |
---|
647 | n/a | ERROR_DECREF(errorcb); |
---|
648 | n/a | return make_tuple(PyUnicode_New(0, 0), 0); |
---|
649 | n/a | } |
---|
650 | n/a | |
---|
651 | n/a | _PyUnicodeWriter_Init(&buf.writer); |
---|
652 | n/a | buf.writer.min_length = datalen; |
---|
653 | n/a | buf.excobj = NULL; |
---|
654 | n/a | buf.inbuf = buf.inbuf_top = (unsigned char *)data; |
---|
655 | n/a | buf.inbuf_end = buf.inbuf_top + datalen; |
---|
656 | n/a | |
---|
657 | n/a | if (self->codec->decinit != NULL && |
---|
658 | n/a | self->codec->decinit(&state, self->codec->config) != 0) |
---|
659 | n/a | goto errorexit; |
---|
660 | n/a | |
---|
661 | n/a | while (buf.inbuf < buf.inbuf_end) { |
---|
662 | n/a | Py_ssize_t inleft, r; |
---|
663 | n/a | |
---|
664 | n/a | inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf); |
---|
665 | n/a | |
---|
666 | n/a | r = self->codec->decode(&state, self->codec->config, |
---|
667 | n/a | &buf.inbuf, inleft, &buf.writer); |
---|
668 | n/a | if (r == 0) |
---|
669 | n/a | break; |
---|
670 | n/a | else if (multibytecodec_decerror(self->codec, &state, |
---|
671 | n/a | &buf, errorcb, r)) |
---|
672 | n/a | goto errorexit; |
---|
673 | n/a | } |
---|
674 | n/a | |
---|
675 | n/a | res = _PyUnicodeWriter_Finish(&buf.writer); |
---|
676 | n/a | if (res == NULL) |
---|
677 | n/a | goto errorexit; |
---|
678 | n/a | |
---|
679 | n/a | Py_XDECREF(buf.excobj); |
---|
680 | n/a | ERROR_DECREF(errorcb); |
---|
681 | n/a | return make_tuple(res, datalen); |
---|
682 | n/a | |
---|
683 | n/a | errorexit: |
---|
684 | n/a | ERROR_DECREF(errorcb); |
---|
685 | n/a | Py_XDECREF(buf.excobj); |
---|
686 | n/a | _PyUnicodeWriter_Dealloc(&buf.writer); |
---|
687 | n/a | |
---|
688 | n/a | return NULL; |
---|
689 | n/a | } |
---|
690 | n/a | |
---|
691 | n/a | static struct PyMethodDef multibytecodec_methods[] = { |
---|
692 | n/a | _MULTIBYTECODEC_MULTIBYTECODEC_ENCODE_METHODDEF |
---|
693 | n/a | _MULTIBYTECODEC_MULTIBYTECODEC_DECODE_METHODDEF |
---|
694 | n/a | {NULL, NULL}, |
---|
695 | n/a | }; |
---|
696 | n/a | |
---|
697 | n/a | static void |
---|
698 | n/a | multibytecodec_dealloc(MultibyteCodecObject *self) |
---|
699 | n/a | { |
---|
700 | n/a | PyObject_Del(self); |
---|
701 | n/a | } |
---|
702 | n/a | |
---|
703 | n/a | static PyTypeObject MultibyteCodec_Type = { |
---|
704 | n/a | PyVarObject_HEAD_INIT(NULL, 0) |
---|
705 | n/a | "MultibyteCodec", /* tp_name */ |
---|
706 | n/a | sizeof(MultibyteCodecObject), /* tp_basicsize */ |
---|
707 | n/a | 0, /* tp_itemsize */ |
---|
708 | n/a | /* methods */ |
---|
709 | n/a | (destructor)multibytecodec_dealloc, /* tp_dealloc */ |
---|
710 | n/a | 0, /* tp_print */ |
---|
711 | n/a | 0, /* tp_getattr */ |
---|
712 | n/a | 0, /* tp_setattr */ |
---|
713 | n/a | 0, /* tp_reserved */ |
---|
714 | n/a | 0, /* tp_repr */ |
---|
715 | n/a | 0, /* tp_as_number */ |
---|
716 | n/a | 0, /* tp_as_sequence */ |
---|
717 | n/a | 0, /* tp_as_mapping */ |
---|
718 | n/a | 0, /* tp_hash */ |
---|
719 | n/a | 0, /* tp_call */ |
---|
720 | n/a | 0, /* tp_str */ |
---|
721 | n/a | PyObject_GenericGetAttr, /* tp_getattro */ |
---|
722 | n/a | 0, /* tp_setattro */ |
---|
723 | n/a | 0, /* tp_as_buffer */ |
---|
724 | n/a | Py_TPFLAGS_DEFAULT, /* tp_flags */ |
---|
725 | n/a | 0, /* tp_doc */ |
---|
726 | n/a | 0, /* tp_traverse */ |
---|
727 | n/a | 0, /* tp_clear */ |
---|
728 | n/a | 0, /* tp_richcompare */ |
---|
729 | n/a | 0, /* tp_weaklistoffset */ |
---|
730 | n/a | 0, /* tp_iter */ |
---|
731 | n/a | 0, /* tp_iterext */ |
---|
732 | n/a | multibytecodec_methods, /* tp_methods */ |
---|
733 | n/a | }; |
---|
734 | n/a | |
---|
735 | n/a | |
---|
736 | n/a | /** |
---|
737 | n/a | * Utility functions for stateful codec mechanism |
---|
738 | n/a | */ |
---|
739 | n/a | |
---|
740 | n/a | #define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o)) |
---|
741 | n/a | #define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o)) |
---|
742 | n/a | |
---|
743 | n/a | static PyObject * |
---|
744 | n/a | encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx, |
---|
745 | n/a | PyObject *unistr, int final) |
---|
746 | n/a | { |
---|
747 | n/a | PyObject *ucvt, *r = NULL; |
---|
748 | n/a | PyObject *inbuf = NULL; |
---|
749 | n/a | Py_ssize_t inpos, datalen; |
---|
750 | n/a | PyObject *origpending = NULL; |
---|
751 | n/a | |
---|
752 | n/a | if (PyUnicode_Check(unistr)) |
---|
753 | n/a | ucvt = NULL; |
---|
754 | n/a | else { |
---|
755 | n/a | unistr = ucvt = PyObject_Str(unistr); |
---|
756 | n/a | if (unistr == NULL) |
---|
757 | n/a | return NULL; |
---|
758 | n/a | else if (!PyUnicode_Check(unistr)) { |
---|
759 | n/a | PyErr_SetString(PyExc_TypeError, |
---|
760 | n/a | "couldn't convert the object to str."); |
---|
761 | n/a | Py_DECREF(ucvt); |
---|
762 | n/a | return NULL; |
---|
763 | n/a | } |
---|
764 | n/a | } |
---|
765 | n/a | |
---|
766 | n/a | if (ctx->pending) { |
---|
767 | n/a | PyObject *inbuf_tmp; |
---|
768 | n/a | |
---|
769 | n/a | Py_INCREF(ctx->pending); |
---|
770 | n/a | origpending = ctx->pending; |
---|
771 | n/a | |
---|
772 | n/a | Py_INCREF(ctx->pending); |
---|
773 | n/a | inbuf_tmp = ctx->pending; |
---|
774 | n/a | PyUnicode_Append(&inbuf_tmp, unistr); |
---|
775 | n/a | if (inbuf_tmp == NULL) |
---|
776 | n/a | goto errorexit; |
---|
777 | n/a | Py_CLEAR(ctx->pending); |
---|
778 | n/a | inbuf = inbuf_tmp; |
---|
779 | n/a | } |
---|
780 | n/a | else { |
---|
781 | n/a | origpending = NULL; |
---|
782 | n/a | |
---|
783 | n/a | Py_INCREF(unistr); |
---|
784 | n/a | inbuf = unistr; |
---|
785 | n/a | } |
---|
786 | n/a | if (PyUnicode_READY(inbuf) < 0) |
---|
787 | n/a | goto errorexit; |
---|
788 | n/a | inpos = 0; |
---|
789 | n/a | datalen = PyUnicode_GET_LENGTH(inbuf); |
---|
790 | n/a | |
---|
791 | n/a | r = multibytecodec_encode(ctx->codec, &ctx->state, |
---|
792 | n/a | inbuf, &inpos, |
---|
793 | n/a | ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0); |
---|
794 | n/a | if (r == NULL) { |
---|
795 | n/a | /* recover the original pending buffer */ |
---|
796 | n/a | Py_XSETREF(ctx->pending, origpending); |
---|
797 | n/a | origpending = NULL; |
---|
798 | n/a | goto errorexit; |
---|
799 | n/a | } |
---|
800 | n/a | Py_XDECREF(origpending); |
---|
801 | n/a | |
---|
802 | n/a | if (inpos < datalen) { |
---|
803 | n/a | if (datalen - inpos > MAXENCPENDING) { |
---|
804 | n/a | /* normal codecs can't reach here */ |
---|
805 | n/a | PyErr_SetString(PyExc_UnicodeError, |
---|
806 | n/a | "pending buffer overflow"); |
---|
807 | n/a | goto errorexit; |
---|
808 | n/a | } |
---|
809 | n/a | ctx->pending = PyUnicode_Substring(inbuf, inpos, datalen); |
---|
810 | n/a | if (ctx->pending == NULL) { |
---|
811 | n/a | /* normal codecs can't reach here */ |
---|
812 | n/a | goto errorexit; |
---|
813 | n/a | } |
---|
814 | n/a | } |
---|
815 | n/a | |
---|
816 | n/a | Py_DECREF(inbuf); |
---|
817 | n/a | Py_XDECREF(ucvt); |
---|
818 | n/a | return r; |
---|
819 | n/a | |
---|
820 | n/a | errorexit: |
---|
821 | n/a | Py_XDECREF(r); |
---|
822 | n/a | Py_XDECREF(ucvt); |
---|
823 | n/a | Py_XDECREF(origpending); |
---|
824 | n/a | Py_XDECREF(inbuf); |
---|
825 | n/a | return NULL; |
---|
826 | n/a | } |
---|
827 | n/a | |
---|
828 | n/a | static int |
---|
829 | n/a | decoder_append_pending(MultibyteStatefulDecoderContext *ctx, |
---|
830 | n/a | MultibyteDecodeBuffer *buf) |
---|
831 | n/a | { |
---|
832 | n/a | Py_ssize_t npendings; |
---|
833 | n/a | |
---|
834 | n/a | npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
---|
835 | n/a | if (npendings + ctx->pendingsize > MAXDECPENDING || |
---|
836 | n/a | npendings > PY_SSIZE_T_MAX - ctx->pendingsize) { |
---|
837 | n/a | PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow"); |
---|
838 | n/a | return -1; |
---|
839 | n/a | } |
---|
840 | n/a | memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings); |
---|
841 | n/a | ctx->pendingsize += npendings; |
---|
842 | n/a | return 0; |
---|
843 | n/a | } |
---|
844 | n/a | |
---|
845 | n/a | static int |
---|
846 | n/a | decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data, |
---|
847 | n/a | Py_ssize_t size) |
---|
848 | n/a | { |
---|
849 | n/a | buf->inbuf = buf->inbuf_top = (const unsigned char *)data; |
---|
850 | n/a | buf->inbuf_end = buf->inbuf_top + size; |
---|
851 | n/a | buf->writer.min_length += size; |
---|
852 | n/a | return 0; |
---|
853 | n/a | } |
---|
854 | n/a | |
---|
855 | n/a | static int |
---|
856 | n/a | decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx, |
---|
857 | n/a | MultibyteDecodeBuffer *buf) |
---|
858 | n/a | { |
---|
859 | n/a | while (buf->inbuf < buf->inbuf_end) { |
---|
860 | n/a | Py_ssize_t inleft; |
---|
861 | n/a | Py_ssize_t r; |
---|
862 | n/a | |
---|
863 | n/a | inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
---|
864 | n/a | |
---|
865 | n/a | r = ctx->codec->decode(&ctx->state, ctx->codec->config, |
---|
866 | n/a | &buf->inbuf, inleft, &buf->writer); |
---|
867 | n/a | if (r == 0 || r == MBERR_TOOFEW) |
---|
868 | n/a | break; |
---|
869 | n/a | else if (multibytecodec_decerror(ctx->codec, &ctx->state, |
---|
870 | n/a | buf, ctx->errors, r)) |
---|
871 | n/a | return -1; |
---|
872 | n/a | } |
---|
873 | n/a | return 0; |
---|
874 | n/a | } |
---|
875 | n/a | |
---|
876 | n/a | |
---|
877 | n/a | /*[clinic input] |
---|
878 | n/a | class _multibytecodec.MultibyteIncrementalEncoder "MultibyteIncrementalEncoderObject *" "&MultibyteIncrementalEncoder_Type" |
---|
879 | n/a | [clinic start generated code]*/ |
---|
880 | n/a | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=3be82909cd08924d]*/ |
---|
881 | n/a | |
---|
882 | n/a | /*[clinic input] |
---|
883 | n/a | _multibytecodec.MultibyteIncrementalEncoder.encode |
---|
884 | n/a | |
---|
885 | n/a | input: object |
---|
886 | n/a | final: int(c_default="0") = False |
---|
887 | n/a | [clinic start generated code]*/ |
---|
888 | n/a | |
---|
889 | n/a | static PyObject * |
---|
890 | n/a | _multibytecodec_MultibyteIncrementalEncoder_encode_impl(MultibyteIncrementalEncoderObject *self, |
---|
891 | n/a | PyObject *input, |
---|
892 | n/a | int final) |
---|
893 | n/a | /*[clinic end generated code: output=123361b6c505e2c1 input=a345c688fa664f92]*/ |
---|
894 | n/a | { |
---|
895 | n/a | return encoder_encode_stateful(STATEFUL_ECTX(self), input, final); |
---|
896 | n/a | } |
---|
897 | n/a | |
---|
898 | n/a | /*[clinic input] |
---|
899 | n/a | _multibytecodec.MultibyteIncrementalEncoder.reset |
---|
900 | n/a | [clinic start generated code]*/ |
---|
901 | n/a | |
---|
902 | n/a | static PyObject * |
---|
903 | n/a | _multibytecodec_MultibyteIncrementalEncoder_reset_impl(MultibyteIncrementalEncoderObject *self) |
---|
904 | n/a | /*[clinic end generated code: output=b4125d8f537a253f input=930f06760707b6ea]*/ |
---|
905 | n/a | { |
---|
906 | n/a | /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */ |
---|
907 | n/a | unsigned char buffer[4], *outbuf; |
---|
908 | n/a | Py_ssize_t r; |
---|
909 | n/a | if (self->codec->encreset != NULL) { |
---|
910 | n/a | outbuf = buffer; |
---|
911 | n/a | r = self->codec->encreset(&self->state, self->codec->config, |
---|
912 | n/a | &outbuf, sizeof(buffer)); |
---|
913 | n/a | if (r != 0) |
---|
914 | n/a | return NULL; |
---|
915 | n/a | } |
---|
916 | n/a | Py_CLEAR(self->pending); |
---|
917 | n/a | Py_RETURN_NONE; |
---|
918 | n/a | } |
---|
919 | n/a | |
---|
920 | n/a | static struct PyMethodDef mbiencoder_methods[] = { |
---|
921 | n/a | _MULTIBYTECODEC_MULTIBYTEINCREMENTALENCODER_ENCODE_METHODDEF |
---|
922 | n/a | _MULTIBYTECODEC_MULTIBYTEINCREMENTALENCODER_RESET_METHODDEF |
---|
923 | n/a | {NULL, NULL}, |
---|
924 | n/a | }; |
---|
925 | n/a | |
---|
926 | n/a | static PyObject * |
---|
927 | n/a | mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
---|
928 | n/a | { |
---|
929 | n/a | MultibyteIncrementalEncoderObject *self; |
---|
930 | n/a | PyObject *codec = NULL; |
---|
931 | n/a | char *errors = NULL; |
---|
932 | n/a | |
---|
933 | n/a | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder", |
---|
934 | n/a | incnewkwarglist, &errors)) |
---|
935 | n/a | return NULL; |
---|
936 | n/a | |
---|
937 | n/a | self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0); |
---|
938 | n/a | if (self == NULL) |
---|
939 | n/a | return NULL; |
---|
940 | n/a | |
---|
941 | n/a | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
---|
942 | n/a | if (codec == NULL) |
---|
943 | n/a | goto errorexit; |
---|
944 | n/a | if (!MultibyteCodec_Check(codec)) { |
---|
945 | n/a | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
---|
946 | n/a | goto errorexit; |
---|
947 | n/a | } |
---|
948 | n/a | |
---|
949 | n/a | self->codec = ((MultibyteCodecObject *)codec)->codec; |
---|
950 | n/a | self->pending = NULL; |
---|
951 | n/a | self->errors = internal_error_callback(errors); |
---|
952 | n/a | if (self->errors == NULL) |
---|
953 | n/a | goto errorexit; |
---|
954 | n/a | if (self->codec->encinit != NULL && |
---|
955 | n/a | self->codec->encinit(&self->state, self->codec->config) != 0) |
---|
956 | n/a | goto errorexit; |
---|
957 | n/a | |
---|
958 | n/a | Py_DECREF(codec); |
---|
959 | n/a | return (PyObject *)self; |
---|
960 | n/a | |
---|
961 | n/a | errorexit: |
---|
962 | n/a | Py_XDECREF(self); |
---|
963 | n/a | Py_XDECREF(codec); |
---|
964 | n/a | return NULL; |
---|
965 | n/a | } |
---|
966 | n/a | |
---|
967 | n/a | static int |
---|
968 | n/a | mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
---|
969 | n/a | { |
---|
970 | n/a | return 0; |
---|
971 | n/a | } |
---|
972 | n/a | |
---|
973 | n/a | static int |
---|
974 | n/a | mbiencoder_traverse(MultibyteIncrementalEncoderObject *self, |
---|
975 | n/a | visitproc visit, void *arg) |
---|
976 | n/a | { |
---|
977 | n/a | if (ERROR_ISCUSTOM(self->errors)) |
---|
978 | n/a | Py_VISIT(self->errors); |
---|
979 | n/a | return 0; |
---|
980 | n/a | } |
---|
981 | n/a | |
---|
982 | n/a | static void |
---|
983 | n/a | mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self) |
---|
984 | n/a | { |
---|
985 | n/a | PyObject_GC_UnTrack(self); |
---|
986 | n/a | ERROR_DECREF(self->errors); |
---|
987 | n/a | Py_TYPE(self)->tp_free(self); |
---|
988 | n/a | } |
---|
989 | n/a | |
---|
990 | n/a | static PyTypeObject MultibyteIncrementalEncoder_Type = { |
---|
991 | n/a | PyVarObject_HEAD_INIT(NULL, 0) |
---|
992 | n/a | "MultibyteIncrementalEncoder", /* tp_name */ |
---|
993 | n/a | sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */ |
---|
994 | n/a | 0, /* tp_itemsize */ |
---|
995 | n/a | /* methods */ |
---|
996 | n/a | (destructor)mbiencoder_dealloc, /* tp_dealloc */ |
---|
997 | n/a | 0, /* tp_print */ |
---|
998 | n/a | 0, /* tp_getattr */ |
---|
999 | n/a | 0, /* tp_setattr */ |
---|
1000 | n/a | 0, /* tp_reserved */ |
---|
1001 | n/a | 0, /* tp_repr */ |
---|
1002 | n/a | 0, /* tp_as_number */ |
---|
1003 | n/a | 0, /* tp_as_sequence */ |
---|
1004 | n/a | 0, /* tp_as_mapping */ |
---|
1005 | n/a | 0, /* tp_hash */ |
---|
1006 | n/a | 0, /* tp_call */ |
---|
1007 | n/a | 0, /* tp_str */ |
---|
1008 | n/a | PyObject_GenericGetAttr, /* tp_getattro */ |
---|
1009 | n/a | 0, /* tp_setattro */ |
---|
1010 | n/a | 0, /* tp_as_buffer */ |
---|
1011 | n/a | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
---|
1012 | n/a | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
---|
1013 | n/a | 0, /* tp_doc */ |
---|
1014 | n/a | (traverseproc)mbiencoder_traverse, /* tp_traverse */ |
---|
1015 | n/a | 0, /* tp_clear */ |
---|
1016 | n/a | 0, /* tp_richcompare */ |
---|
1017 | n/a | 0, /* tp_weaklistoffset */ |
---|
1018 | n/a | 0, /* tp_iter */ |
---|
1019 | n/a | 0, /* tp_iterext */ |
---|
1020 | n/a | mbiencoder_methods, /* tp_methods */ |
---|
1021 | n/a | 0, /* tp_members */ |
---|
1022 | n/a | codecctx_getsets, /* tp_getset */ |
---|
1023 | n/a | 0, /* tp_base */ |
---|
1024 | n/a | 0, /* tp_dict */ |
---|
1025 | n/a | 0, /* tp_descr_get */ |
---|
1026 | n/a | 0, /* tp_descr_set */ |
---|
1027 | n/a | 0, /* tp_dictoffset */ |
---|
1028 | n/a | mbiencoder_init, /* tp_init */ |
---|
1029 | n/a | 0, /* tp_alloc */ |
---|
1030 | n/a | mbiencoder_new, /* tp_new */ |
---|
1031 | n/a | }; |
---|
1032 | n/a | |
---|
1033 | n/a | |
---|
1034 | n/a | /*[clinic input] |
---|
1035 | n/a | class _multibytecodec.MultibyteIncrementalDecoder "MultibyteIncrementalDecoderObject *" "&MultibyteIncrementalDecoder_Type" |
---|
1036 | n/a | [clinic start generated code]*/ |
---|
1037 | n/a | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=f6003faaf2cea692]*/ |
---|
1038 | n/a | |
---|
1039 | n/a | /*[clinic input] |
---|
1040 | n/a | _multibytecodec.MultibyteIncrementalDecoder.decode |
---|
1041 | n/a | |
---|
1042 | n/a | input: Py_buffer |
---|
1043 | n/a | final: int(c_default="0") = False |
---|
1044 | n/a | [clinic start generated code]*/ |
---|
1045 | n/a | |
---|
1046 | n/a | static PyObject * |
---|
1047 | n/a | _multibytecodec_MultibyteIncrementalDecoder_decode_impl(MultibyteIncrementalDecoderObject *self, |
---|
1048 | n/a | Py_buffer *input, |
---|
1049 | n/a | int final) |
---|
1050 | n/a | /*[clinic end generated code: output=b9b9090e8a9ce2ba input=576631c61906d39d]*/ |
---|
1051 | n/a | { |
---|
1052 | n/a | MultibyteDecodeBuffer buf; |
---|
1053 | n/a | char *data, *wdata = NULL; |
---|
1054 | n/a | Py_ssize_t wsize, size, origpending; |
---|
1055 | n/a | PyObject *res; |
---|
1056 | n/a | |
---|
1057 | n/a | data = input->buf; |
---|
1058 | n/a | size = input->len; |
---|
1059 | n/a | |
---|
1060 | n/a | _PyUnicodeWriter_Init(&buf.writer); |
---|
1061 | n/a | buf.excobj = NULL; |
---|
1062 | n/a | origpending = self->pendingsize; |
---|
1063 | n/a | |
---|
1064 | n/a | if (self->pendingsize == 0) { |
---|
1065 | n/a | wsize = size; |
---|
1066 | n/a | wdata = data; |
---|
1067 | n/a | } |
---|
1068 | n/a | else { |
---|
1069 | n/a | if (size > PY_SSIZE_T_MAX - self->pendingsize) { |
---|
1070 | n/a | PyErr_NoMemory(); |
---|
1071 | n/a | goto errorexit; |
---|
1072 | n/a | } |
---|
1073 | n/a | wsize = size + self->pendingsize; |
---|
1074 | n/a | wdata = PyMem_Malloc(wsize); |
---|
1075 | n/a | if (wdata == NULL) { |
---|
1076 | n/a | PyErr_NoMemory(); |
---|
1077 | n/a | goto errorexit; |
---|
1078 | n/a | } |
---|
1079 | n/a | memcpy(wdata, self->pending, self->pendingsize); |
---|
1080 | n/a | memcpy(wdata + self->pendingsize, data, size); |
---|
1081 | n/a | self->pendingsize = 0; |
---|
1082 | n/a | } |
---|
1083 | n/a | |
---|
1084 | n/a | if (decoder_prepare_buffer(&buf, wdata, wsize) != 0) |
---|
1085 | n/a | goto errorexit; |
---|
1086 | n/a | |
---|
1087 | n/a | if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf)) |
---|
1088 | n/a | goto errorexit; |
---|
1089 | n/a | |
---|
1090 | n/a | if (final && buf.inbuf < buf.inbuf_end) { |
---|
1091 | n/a | if (multibytecodec_decerror(self->codec, &self->state, |
---|
1092 | n/a | &buf, self->errors, MBERR_TOOFEW)) { |
---|
1093 | n/a | /* recover the original pending buffer */ |
---|
1094 | n/a | memcpy(self->pending, wdata, origpending); |
---|
1095 | n/a | self->pendingsize = origpending; |
---|
1096 | n/a | goto errorexit; |
---|
1097 | n/a | } |
---|
1098 | n/a | } |
---|
1099 | n/a | |
---|
1100 | n/a | if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */ |
---|
1101 | n/a | if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0) |
---|
1102 | n/a | goto errorexit; |
---|
1103 | n/a | } |
---|
1104 | n/a | |
---|
1105 | n/a | res = _PyUnicodeWriter_Finish(&buf.writer); |
---|
1106 | n/a | if (res == NULL) |
---|
1107 | n/a | goto errorexit; |
---|
1108 | n/a | |
---|
1109 | n/a | if (wdata != data) |
---|
1110 | n/a | PyMem_Del(wdata); |
---|
1111 | n/a | Py_XDECREF(buf.excobj); |
---|
1112 | n/a | return res; |
---|
1113 | n/a | |
---|
1114 | n/a | errorexit: |
---|
1115 | n/a | if (wdata != NULL && wdata != data) |
---|
1116 | n/a | PyMem_Del(wdata); |
---|
1117 | n/a | Py_XDECREF(buf.excobj); |
---|
1118 | n/a | _PyUnicodeWriter_Dealloc(&buf.writer); |
---|
1119 | n/a | return NULL; |
---|
1120 | n/a | } |
---|
1121 | n/a | |
---|
1122 | n/a | /*[clinic input] |
---|
1123 | n/a | _multibytecodec.MultibyteIncrementalDecoder.reset |
---|
1124 | n/a | [clinic start generated code]*/ |
---|
1125 | n/a | |
---|
1126 | n/a | static PyObject * |
---|
1127 | n/a | _multibytecodec_MultibyteIncrementalDecoder_reset_impl(MultibyteIncrementalDecoderObject *self) |
---|
1128 | n/a | /*[clinic end generated code: output=da423b1782c23ed1 input=3b63b3be85b2fb45]*/ |
---|
1129 | n/a | { |
---|
1130 | n/a | if (self->codec->decreset != NULL && |
---|
1131 | n/a | self->codec->decreset(&self->state, self->codec->config) != 0) |
---|
1132 | n/a | return NULL; |
---|
1133 | n/a | self->pendingsize = 0; |
---|
1134 | n/a | |
---|
1135 | n/a | Py_RETURN_NONE; |
---|
1136 | n/a | } |
---|
1137 | n/a | |
---|
1138 | n/a | static struct PyMethodDef mbidecoder_methods[] = { |
---|
1139 | n/a | _MULTIBYTECODEC_MULTIBYTEINCREMENTALDECODER_DECODE_METHODDEF |
---|
1140 | n/a | _MULTIBYTECODEC_MULTIBYTEINCREMENTALDECODER_RESET_METHODDEF |
---|
1141 | n/a | {NULL, NULL}, |
---|
1142 | n/a | }; |
---|
1143 | n/a | |
---|
1144 | n/a | static PyObject * |
---|
1145 | n/a | mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
---|
1146 | n/a | { |
---|
1147 | n/a | MultibyteIncrementalDecoderObject *self; |
---|
1148 | n/a | PyObject *codec = NULL; |
---|
1149 | n/a | char *errors = NULL; |
---|
1150 | n/a | |
---|
1151 | n/a | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder", |
---|
1152 | n/a | incnewkwarglist, &errors)) |
---|
1153 | n/a | return NULL; |
---|
1154 | n/a | |
---|
1155 | n/a | self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0); |
---|
1156 | n/a | if (self == NULL) |
---|
1157 | n/a | return NULL; |
---|
1158 | n/a | |
---|
1159 | n/a | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
---|
1160 | n/a | if (codec == NULL) |
---|
1161 | n/a | goto errorexit; |
---|
1162 | n/a | if (!MultibyteCodec_Check(codec)) { |
---|
1163 | n/a | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
---|
1164 | n/a | goto errorexit; |
---|
1165 | n/a | } |
---|
1166 | n/a | |
---|
1167 | n/a | self->codec = ((MultibyteCodecObject *)codec)->codec; |
---|
1168 | n/a | self->pendingsize = 0; |
---|
1169 | n/a | self->errors = internal_error_callback(errors); |
---|
1170 | n/a | if (self->errors == NULL) |
---|
1171 | n/a | goto errorexit; |
---|
1172 | n/a | if (self->codec->decinit != NULL && |
---|
1173 | n/a | self->codec->decinit(&self->state, self->codec->config) != 0) |
---|
1174 | n/a | goto errorexit; |
---|
1175 | n/a | |
---|
1176 | n/a | Py_DECREF(codec); |
---|
1177 | n/a | return (PyObject *)self; |
---|
1178 | n/a | |
---|
1179 | n/a | errorexit: |
---|
1180 | n/a | Py_XDECREF(self); |
---|
1181 | n/a | Py_XDECREF(codec); |
---|
1182 | n/a | return NULL; |
---|
1183 | n/a | } |
---|
1184 | n/a | |
---|
1185 | n/a | static int |
---|
1186 | n/a | mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
---|
1187 | n/a | { |
---|
1188 | n/a | return 0; |
---|
1189 | n/a | } |
---|
1190 | n/a | |
---|
1191 | n/a | static int |
---|
1192 | n/a | mbidecoder_traverse(MultibyteIncrementalDecoderObject *self, |
---|
1193 | n/a | visitproc visit, void *arg) |
---|
1194 | n/a | { |
---|
1195 | n/a | if (ERROR_ISCUSTOM(self->errors)) |
---|
1196 | n/a | Py_VISIT(self->errors); |
---|
1197 | n/a | return 0; |
---|
1198 | n/a | } |
---|
1199 | n/a | |
---|
1200 | n/a | static void |
---|
1201 | n/a | mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self) |
---|
1202 | n/a | { |
---|
1203 | n/a | PyObject_GC_UnTrack(self); |
---|
1204 | n/a | ERROR_DECREF(self->errors); |
---|
1205 | n/a | Py_TYPE(self)->tp_free(self); |
---|
1206 | n/a | } |
---|
1207 | n/a | |
---|
1208 | n/a | static PyTypeObject MultibyteIncrementalDecoder_Type = { |
---|
1209 | n/a | PyVarObject_HEAD_INIT(NULL, 0) |
---|
1210 | n/a | "MultibyteIncrementalDecoder", /* tp_name */ |
---|
1211 | n/a | sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */ |
---|
1212 | n/a | 0, /* tp_itemsize */ |
---|
1213 | n/a | /* methods */ |
---|
1214 | n/a | (destructor)mbidecoder_dealloc, /* tp_dealloc */ |
---|
1215 | n/a | 0, /* tp_print */ |
---|
1216 | n/a | 0, /* tp_getattr */ |
---|
1217 | n/a | 0, /* tp_setattr */ |
---|
1218 | n/a | 0, /* tp_reserved */ |
---|
1219 | n/a | 0, /* tp_repr */ |
---|
1220 | n/a | 0, /* tp_as_number */ |
---|
1221 | n/a | 0, /* tp_as_sequence */ |
---|
1222 | n/a | 0, /* tp_as_mapping */ |
---|
1223 | n/a | 0, /* tp_hash */ |
---|
1224 | n/a | 0, /* tp_call */ |
---|
1225 | n/a | 0, /* tp_str */ |
---|
1226 | n/a | PyObject_GenericGetAttr, /* tp_getattro */ |
---|
1227 | n/a | 0, /* tp_setattro */ |
---|
1228 | n/a | 0, /* tp_as_buffer */ |
---|
1229 | n/a | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
---|
1230 | n/a | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
---|
1231 | n/a | 0, /* tp_doc */ |
---|
1232 | n/a | (traverseproc)mbidecoder_traverse, /* tp_traverse */ |
---|
1233 | n/a | 0, /* tp_clear */ |
---|
1234 | n/a | 0, /* tp_richcompare */ |
---|
1235 | n/a | 0, /* tp_weaklistoffset */ |
---|
1236 | n/a | 0, /* tp_iter */ |
---|
1237 | n/a | 0, /* tp_iterext */ |
---|
1238 | n/a | mbidecoder_methods, /* tp_methods */ |
---|
1239 | n/a | 0, /* tp_members */ |
---|
1240 | n/a | codecctx_getsets, /* tp_getset */ |
---|
1241 | n/a | 0, /* tp_base */ |
---|
1242 | n/a | 0, /* tp_dict */ |
---|
1243 | n/a | 0, /* tp_descr_get */ |
---|
1244 | n/a | 0, /* tp_descr_set */ |
---|
1245 | n/a | 0, /* tp_dictoffset */ |
---|
1246 | n/a | mbidecoder_init, /* tp_init */ |
---|
1247 | n/a | 0, /* tp_alloc */ |
---|
1248 | n/a | mbidecoder_new, /* tp_new */ |
---|
1249 | n/a | }; |
---|
1250 | n/a | |
---|
1251 | n/a | |
---|
1252 | n/a | /*[clinic input] |
---|
1253 | n/a | class _multibytecodec.MultibyteStreamReader "MultibyteStreamReaderObject *" "MultibyteStreamReader_Type" |
---|
1254 | n/a | [clinic start generated code]*/ |
---|
1255 | n/a | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=d323634b74976f09]*/ |
---|
1256 | n/a | |
---|
1257 | n/a | static PyObject * |
---|
1258 | n/a | mbstreamreader_iread(MultibyteStreamReaderObject *self, |
---|
1259 | n/a | const char *method, Py_ssize_t sizehint) |
---|
1260 | n/a | { |
---|
1261 | n/a | MultibyteDecodeBuffer buf; |
---|
1262 | n/a | PyObject *cres, *res; |
---|
1263 | n/a | Py_ssize_t rsize; |
---|
1264 | n/a | |
---|
1265 | n/a | if (sizehint == 0) |
---|
1266 | n/a | return PyUnicode_New(0, 0); |
---|
1267 | n/a | |
---|
1268 | n/a | _PyUnicodeWriter_Init(&buf.writer); |
---|
1269 | n/a | buf.excobj = NULL; |
---|
1270 | n/a | cres = NULL; |
---|
1271 | n/a | |
---|
1272 | n/a | for (;;) { |
---|
1273 | n/a | int endoffile; |
---|
1274 | n/a | |
---|
1275 | n/a | if (sizehint < 0) |
---|
1276 | n/a | cres = PyObject_CallMethod(self->stream, |
---|
1277 | n/a | method, NULL); |
---|
1278 | n/a | else |
---|
1279 | n/a | cres = PyObject_CallMethod(self->stream, |
---|
1280 | n/a | method, "i", sizehint); |
---|
1281 | n/a | if (cres == NULL) |
---|
1282 | n/a | goto errorexit; |
---|
1283 | n/a | |
---|
1284 | n/a | if (!PyBytes_Check(cres)) { |
---|
1285 | n/a | PyErr_Format(PyExc_TypeError, |
---|
1286 | n/a | "stream function returned a " |
---|
1287 | n/a | "non-bytes object (%.100s)", |
---|
1288 | n/a | cres->ob_type->tp_name); |
---|
1289 | n/a | goto errorexit; |
---|
1290 | n/a | } |
---|
1291 | n/a | |
---|
1292 | n/a | endoffile = (PyBytes_GET_SIZE(cres) == 0); |
---|
1293 | n/a | |
---|
1294 | n/a | if (self->pendingsize > 0) { |
---|
1295 | n/a | PyObject *ctr; |
---|
1296 | n/a | char *ctrdata; |
---|
1297 | n/a | |
---|
1298 | n/a | if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) { |
---|
1299 | n/a | PyErr_NoMemory(); |
---|
1300 | n/a | goto errorexit; |
---|
1301 | n/a | } |
---|
1302 | n/a | rsize = PyBytes_GET_SIZE(cres) + self->pendingsize; |
---|
1303 | n/a | ctr = PyBytes_FromStringAndSize(NULL, rsize); |
---|
1304 | n/a | if (ctr == NULL) |
---|
1305 | n/a | goto errorexit; |
---|
1306 | n/a | ctrdata = PyBytes_AS_STRING(ctr); |
---|
1307 | n/a | memcpy(ctrdata, self->pending, self->pendingsize); |
---|
1308 | n/a | memcpy(ctrdata + self->pendingsize, |
---|
1309 | n/a | PyBytes_AS_STRING(cres), |
---|
1310 | n/a | PyBytes_GET_SIZE(cres)); |
---|
1311 | n/a | Py_DECREF(cres); |
---|
1312 | n/a | cres = ctr; |
---|
1313 | n/a | self->pendingsize = 0; |
---|
1314 | n/a | } |
---|
1315 | n/a | |
---|
1316 | n/a | rsize = PyBytes_GET_SIZE(cres); |
---|
1317 | n/a | if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres), |
---|
1318 | n/a | rsize) != 0) |
---|
1319 | n/a | goto errorexit; |
---|
1320 | n/a | |
---|
1321 | n/a | if (rsize > 0 && decoder_feed_buffer( |
---|
1322 | n/a | (MultibyteStatefulDecoderContext *)self, &buf)) |
---|
1323 | n/a | goto errorexit; |
---|
1324 | n/a | |
---|
1325 | n/a | if (endoffile || sizehint < 0) { |
---|
1326 | n/a | if (buf.inbuf < buf.inbuf_end && |
---|
1327 | n/a | multibytecodec_decerror(self->codec, &self->state, |
---|
1328 | n/a | &buf, self->errors, MBERR_TOOFEW)) |
---|
1329 | n/a | goto errorexit; |
---|
1330 | n/a | } |
---|
1331 | n/a | |
---|
1332 | n/a | if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */ |
---|
1333 | n/a | if (decoder_append_pending(STATEFUL_DCTX(self), |
---|
1334 | n/a | &buf) != 0) |
---|
1335 | n/a | goto errorexit; |
---|
1336 | n/a | } |
---|
1337 | n/a | |
---|
1338 | n/a | Py_DECREF(cres); |
---|
1339 | n/a | cres = NULL; |
---|
1340 | n/a | |
---|
1341 | n/a | if (sizehint < 0 || buf.writer.pos != 0 || rsize == 0) |
---|
1342 | n/a | break; |
---|
1343 | n/a | |
---|
1344 | n/a | sizehint = 1; /* read 1 more byte and retry */ |
---|
1345 | n/a | } |
---|
1346 | n/a | |
---|
1347 | n/a | res = _PyUnicodeWriter_Finish(&buf.writer); |
---|
1348 | n/a | if (res == NULL) |
---|
1349 | n/a | goto errorexit; |
---|
1350 | n/a | |
---|
1351 | n/a | Py_XDECREF(cres); |
---|
1352 | n/a | Py_XDECREF(buf.excobj); |
---|
1353 | n/a | return res; |
---|
1354 | n/a | |
---|
1355 | n/a | errorexit: |
---|
1356 | n/a | Py_XDECREF(cres); |
---|
1357 | n/a | Py_XDECREF(buf.excobj); |
---|
1358 | n/a | _PyUnicodeWriter_Dealloc(&buf.writer); |
---|
1359 | n/a | return NULL; |
---|
1360 | n/a | } |
---|
1361 | n/a | |
---|
1362 | n/a | /*[clinic input] |
---|
1363 | n/a | _multibytecodec.MultibyteStreamReader.read |
---|
1364 | n/a | |
---|
1365 | n/a | sizeobj: object = None |
---|
1366 | n/a | / |
---|
1367 | n/a | [clinic start generated code]*/ |
---|
1368 | n/a | |
---|
1369 | n/a | static PyObject * |
---|
1370 | n/a | _multibytecodec_MultibyteStreamReader_read_impl(MultibyteStreamReaderObject *self, |
---|
1371 | n/a | PyObject *sizeobj) |
---|
1372 | n/a | /*[clinic end generated code: output=35621eb75355d5b8 input=015b0d3ff2fca485]*/ |
---|
1373 | n/a | { |
---|
1374 | n/a | Py_ssize_t size; |
---|
1375 | n/a | |
---|
1376 | n/a | if (sizeobj == Py_None) |
---|
1377 | n/a | size = -1; |
---|
1378 | n/a | else if (PyLong_Check(sizeobj)) |
---|
1379 | n/a | size = PyLong_AsSsize_t(sizeobj); |
---|
1380 | n/a | else { |
---|
1381 | n/a | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
---|
1382 | n/a | return NULL; |
---|
1383 | n/a | } |
---|
1384 | n/a | |
---|
1385 | n/a | if (size == -1 && PyErr_Occurred()) |
---|
1386 | n/a | return NULL; |
---|
1387 | n/a | |
---|
1388 | n/a | return mbstreamreader_iread(self, "read", size); |
---|
1389 | n/a | } |
---|
1390 | n/a | |
---|
1391 | n/a | /*[clinic input] |
---|
1392 | n/a | _multibytecodec.MultibyteStreamReader.readline |
---|
1393 | n/a | |
---|
1394 | n/a | sizeobj: object = None |
---|
1395 | n/a | / |
---|
1396 | n/a | [clinic start generated code]*/ |
---|
1397 | n/a | |
---|
1398 | n/a | static PyObject * |
---|
1399 | n/a | _multibytecodec_MultibyteStreamReader_readline_impl(MultibyteStreamReaderObject *self, |
---|
1400 | n/a | PyObject *sizeobj) |
---|
1401 | n/a | /*[clinic end generated code: output=4fbfaae1ed457a11 input=41ccc64f9bb0cec3]*/ |
---|
1402 | n/a | { |
---|
1403 | n/a | Py_ssize_t size; |
---|
1404 | n/a | |
---|
1405 | n/a | if (sizeobj == Py_None) |
---|
1406 | n/a | size = -1; |
---|
1407 | n/a | else if (PyLong_Check(sizeobj)) |
---|
1408 | n/a | size = PyLong_AsSsize_t(sizeobj); |
---|
1409 | n/a | else { |
---|
1410 | n/a | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
---|
1411 | n/a | return NULL; |
---|
1412 | n/a | } |
---|
1413 | n/a | |
---|
1414 | n/a | if (size == -1 && PyErr_Occurred()) |
---|
1415 | n/a | return NULL; |
---|
1416 | n/a | |
---|
1417 | n/a | return mbstreamreader_iread(self, "readline", size); |
---|
1418 | n/a | } |
---|
1419 | n/a | |
---|
1420 | n/a | /*[clinic input] |
---|
1421 | n/a | _multibytecodec.MultibyteStreamReader.readlines |
---|
1422 | n/a | |
---|
1423 | n/a | sizehintobj: object = None |
---|
1424 | n/a | / |
---|
1425 | n/a | [clinic start generated code]*/ |
---|
1426 | n/a | |
---|
1427 | n/a | static PyObject * |
---|
1428 | n/a | _multibytecodec_MultibyteStreamReader_readlines_impl(MultibyteStreamReaderObject *self, |
---|
1429 | n/a | PyObject *sizehintobj) |
---|
1430 | n/a | /*[clinic end generated code: output=e7c4310768ed2ad4 input=54932f5d4d88e880]*/ |
---|
1431 | n/a | { |
---|
1432 | n/a | PyObject *r, *sr; |
---|
1433 | n/a | Py_ssize_t sizehint; |
---|
1434 | n/a | |
---|
1435 | n/a | if (sizehintobj == Py_None) |
---|
1436 | n/a | sizehint = -1; |
---|
1437 | n/a | else if (PyLong_Check(sizehintobj)) |
---|
1438 | n/a | sizehint = PyLong_AsSsize_t(sizehintobj); |
---|
1439 | n/a | else { |
---|
1440 | n/a | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
---|
1441 | n/a | return NULL; |
---|
1442 | n/a | } |
---|
1443 | n/a | |
---|
1444 | n/a | if (sizehint == -1 && PyErr_Occurred()) |
---|
1445 | n/a | return NULL; |
---|
1446 | n/a | |
---|
1447 | n/a | r = mbstreamreader_iread(self, "read", sizehint); |
---|
1448 | n/a | if (r == NULL) |
---|
1449 | n/a | return NULL; |
---|
1450 | n/a | |
---|
1451 | n/a | sr = PyUnicode_Splitlines(r, 1); |
---|
1452 | n/a | Py_DECREF(r); |
---|
1453 | n/a | return sr; |
---|
1454 | n/a | } |
---|
1455 | n/a | |
---|
1456 | n/a | /*[clinic input] |
---|
1457 | n/a | _multibytecodec.MultibyteStreamReader.reset |
---|
1458 | n/a | [clinic start generated code]*/ |
---|
1459 | n/a | |
---|
1460 | n/a | static PyObject * |
---|
1461 | n/a | _multibytecodec_MultibyteStreamReader_reset_impl(MultibyteStreamReaderObject *self) |
---|
1462 | n/a | /*[clinic end generated code: output=138490370a680abc input=5d4140db84b5e1e2]*/ |
---|
1463 | n/a | { |
---|
1464 | n/a | if (self->codec->decreset != NULL && |
---|
1465 | n/a | self->codec->decreset(&self->state, self->codec->config) != 0) |
---|
1466 | n/a | return NULL; |
---|
1467 | n/a | self->pendingsize = 0; |
---|
1468 | n/a | |
---|
1469 | n/a | Py_RETURN_NONE; |
---|
1470 | n/a | } |
---|
1471 | n/a | |
---|
1472 | n/a | static struct PyMethodDef mbstreamreader_methods[] = { |
---|
1473 | n/a | _MULTIBYTECODEC_MULTIBYTESTREAMREADER_READ_METHODDEF |
---|
1474 | n/a | _MULTIBYTECODEC_MULTIBYTESTREAMREADER_READLINE_METHODDEF |
---|
1475 | n/a | _MULTIBYTECODEC_MULTIBYTESTREAMREADER_READLINES_METHODDEF |
---|
1476 | n/a | _MULTIBYTECODEC_MULTIBYTESTREAMREADER_RESET_METHODDEF |
---|
1477 | n/a | {NULL, NULL}, |
---|
1478 | n/a | }; |
---|
1479 | n/a | |
---|
1480 | n/a | static PyMemberDef mbstreamreader_members[] = { |
---|
1481 | n/a | {"stream", T_OBJECT, |
---|
1482 | n/a | offsetof(MultibyteStreamReaderObject, stream), |
---|
1483 | n/a | READONLY, NULL}, |
---|
1484 | n/a | {NULL,} |
---|
1485 | n/a | }; |
---|
1486 | n/a | |
---|
1487 | n/a | static PyObject * |
---|
1488 | n/a | mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
---|
1489 | n/a | { |
---|
1490 | n/a | MultibyteStreamReaderObject *self; |
---|
1491 | n/a | PyObject *stream, *codec = NULL; |
---|
1492 | n/a | char *errors = NULL; |
---|
1493 | n/a | |
---|
1494 | n/a | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader", |
---|
1495 | n/a | streamkwarglist, &stream, &errors)) |
---|
1496 | n/a | return NULL; |
---|
1497 | n/a | |
---|
1498 | n/a | self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0); |
---|
1499 | n/a | if (self == NULL) |
---|
1500 | n/a | return NULL; |
---|
1501 | n/a | |
---|
1502 | n/a | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
---|
1503 | n/a | if (codec == NULL) |
---|
1504 | n/a | goto errorexit; |
---|
1505 | n/a | if (!MultibyteCodec_Check(codec)) { |
---|
1506 | n/a | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
---|
1507 | n/a | goto errorexit; |
---|
1508 | n/a | } |
---|
1509 | n/a | |
---|
1510 | n/a | self->codec = ((MultibyteCodecObject *)codec)->codec; |
---|
1511 | n/a | self->stream = stream; |
---|
1512 | n/a | Py_INCREF(stream); |
---|
1513 | n/a | self->pendingsize = 0; |
---|
1514 | n/a | self->errors = internal_error_callback(errors); |
---|
1515 | n/a | if (self->errors == NULL) |
---|
1516 | n/a | goto errorexit; |
---|
1517 | n/a | if (self->codec->decinit != NULL && |
---|
1518 | n/a | self->codec->decinit(&self->state, self->codec->config) != 0) |
---|
1519 | n/a | goto errorexit; |
---|
1520 | n/a | |
---|
1521 | n/a | Py_DECREF(codec); |
---|
1522 | n/a | return (PyObject *)self; |
---|
1523 | n/a | |
---|
1524 | n/a | errorexit: |
---|
1525 | n/a | Py_XDECREF(self); |
---|
1526 | n/a | Py_XDECREF(codec); |
---|
1527 | n/a | return NULL; |
---|
1528 | n/a | } |
---|
1529 | n/a | |
---|
1530 | n/a | static int |
---|
1531 | n/a | mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds) |
---|
1532 | n/a | { |
---|
1533 | n/a | return 0; |
---|
1534 | n/a | } |
---|
1535 | n/a | |
---|
1536 | n/a | static int |
---|
1537 | n/a | mbstreamreader_traverse(MultibyteStreamReaderObject *self, |
---|
1538 | n/a | visitproc visit, void *arg) |
---|
1539 | n/a | { |
---|
1540 | n/a | if (ERROR_ISCUSTOM(self->errors)) |
---|
1541 | n/a | Py_VISIT(self->errors); |
---|
1542 | n/a | Py_VISIT(self->stream); |
---|
1543 | n/a | return 0; |
---|
1544 | n/a | } |
---|
1545 | n/a | |
---|
1546 | n/a | static void |
---|
1547 | n/a | mbstreamreader_dealloc(MultibyteStreamReaderObject *self) |
---|
1548 | n/a | { |
---|
1549 | n/a | PyObject_GC_UnTrack(self); |
---|
1550 | n/a | ERROR_DECREF(self->errors); |
---|
1551 | n/a | Py_XDECREF(self->stream); |
---|
1552 | n/a | Py_TYPE(self)->tp_free(self); |
---|
1553 | n/a | } |
---|
1554 | n/a | |
---|
1555 | n/a | static PyTypeObject MultibyteStreamReader_Type = { |
---|
1556 | n/a | PyVarObject_HEAD_INIT(NULL, 0) |
---|
1557 | n/a | "MultibyteStreamReader", /* tp_name */ |
---|
1558 | n/a | sizeof(MultibyteStreamReaderObject), /* tp_basicsize */ |
---|
1559 | n/a | 0, /* tp_itemsize */ |
---|
1560 | n/a | /* methods */ |
---|
1561 | n/a | (destructor)mbstreamreader_dealloc, /* tp_dealloc */ |
---|
1562 | n/a | 0, /* tp_print */ |
---|
1563 | n/a | 0, /* tp_getattr */ |
---|
1564 | n/a | 0, /* tp_setattr */ |
---|
1565 | n/a | 0, /* tp_reserved */ |
---|
1566 | n/a | 0, /* tp_repr */ |
---|
1567 | n/a | 0, /* tp_as_number */ |
---|
1568 | n/a | 0, /* tp_as_sequence */ |
---|
1569 | n/a | 0, /* tp_as_mapping */ |
---|
1570 | n/a | 0, /* tp_hash */ |
---|
1571 | n/a | 0, /* tp_call */ |
---|
1572 | n/a | 0, /* tp_str */ |
---|
1573 | n/a | PyObject_GenericGetAttr, /* tp_getattro */ |
---|
1574 | n/a | 0, /* tp_setattro */ |
---|
1575 | n/a | 0, /* tp_as_buffer */ |
---|
1576 | n/a | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
---|
1577 | n/a | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
---|
1578 | n/a | 0, /* tp_doc */ |
---|
1579 | n/a | (traverseproc)mbstreamreader_traverse, /* tp_traverse */ |
---|
1580 | n/a | 0, /* tp_clear */ |
---|
1581 | n/a | 0, /* tp_richcompare */ |
---|
1582 | n/a | 0, /* tp_weaklistoffset */ |
---|
1583 | n/a | 0, /* tp_iter */ |
---|
1584 | n/a | 0, /* tp_iterext */ |
---|
1585 | n/a | mbstreamreader_methods, /* tp_methods */ |
---|
1586 | n/a | mbstreamreader_members, /* tp_members */ |
---|
1587 | n/a | codecctx_getsets, /* tp_getset */ |
---|
1588 | n/a | 0, /* tp_base */ |
---|
1589 | n/a | 0, /* tp_dict */ |
---|
1590 | n/a | 0, /* tp_descr_get */ |
---|
1591 | n/a | 0, /* tp_descr_set */ |
---|
1592 | n/a | 0, /* tp_dictoffset */ |
---|
1593 | n/a | mbstreamreader_init, /* tp_init */ |
---|
1594 | n/a | 0, /* tp_alloc */ |
---|
1595 | n/a | mbstreamreader_new, /* tp_new */ |
---|
1596 | n/a | }; |
---|
1597 | n/a | |
---|
1598 | n/a | |
---|
1599 | n/a | /*[clinic input] |
---|
1600 | n/a | class _multibytecodec.MultibyteStreamWriter "MultibyteStreamWriterObject *" "&MultibyteStreamWriter_Type" |
---|
1601 | n/a | [clinic start generated code]*/ |
---|
1602 | n/a | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=cde22780a215d6ac]*/ |
---|
1603 | n/a | |
---|
1604 | n/a | static int |
---|
1605 | n/a | mbstreamwriter_iwrite(MultibyteStreamWriterObject *self, |
---|
1606 | n/a | PyObject *unistr) |
---|
1607 | n/a | { |
---|
1608 | n/a | PyObject *str, *wr; |
---|
1609 | n/a | |
---|
1610 | n/a | str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0); |
---|
1611 | n/a | if (str == NULL) |
---|
1612 | n/a | return -1; |
---|
1613 | n/a | |
---|
1614 | n/a | wr = _PyObject_CallMethodIdObjArgs(self->stream, &PyId_write, str, NULL); |
---|
1615 | n/a | Py_DECREF(str); |
---|
1616 | n/a | if (wr == NULL) |
---|
1617 | n/a | return -1; |
---|
1618 | n/a | |
---|
1619 | n/a | Py_DECREF(wr); |
---|
1620 | n/a | return 0; |
---|
1621 | n/a | } |
---|
1622 | n/a | |
---|
1623 | n/a | /*[clinic input] |
---|
1624 | n/a | _multibytecodec.MultibyteStreamWriter.write |
---|
1625 | n/a | |
---|
1626 | n/a | strobj: object |
---|
1627 | n/a | / |
---|
1628 | n/a | [clinic start generated code]*/ |
---|
1629 | n/a | |
---|
1630 | n/a | static PyObject * |
---|
1631 | n/a | _multibytecodec_MultibyteStreamWriter_write(MultibyteStreamWriterObject *self, |
---|
1632 | n/a | PyObject *strobj) |
---|
1633 | n/a | /*[clinic end generated code: output=e13ae841c895251e input=551dc4c018c10a2b]*/ |
---|
1634 | n/a | { |
---|
1635 | n/a | if (mbstreamwriter_iwrite(self, strobj)) |
---|
1636 | n/a | return NULL; |
---|
1637 | n/a | else |
---|
1638 | n/a | Py_RETURN_NONE; |
---|
1639 | n/a | } |
---|
1640 | n/a | |
---|
1641 | n/a | /*[clinic input] |
---|
1642 | n/a | _multibytecodec.MultibyteStreamWriter.writelines |
---|
1643 | n/a | |
---|
1644 | n/a | lines: object |
---|
1645 | n/a | / |
---|
1646 | n/a | [clinic start generated code]*/ |
---|
1647 | n/a | |
---|
1648 | n/a | static PyObject * |
---|
1649 | n/a | _multibytecodec_MultibyteStreamWriter_writelines(MultibyteStreamWriterObject *self, |
---|
1650 | n/a | PyObject *lines) |
---|
1651 | n/a | /*[clinic end generated code: output=e5c4285ac8e7d522 input=57797fe7008d4e96]*/ |
---|
1652 | n/a | { |
---|
1653 | n/a | PyObject *strobj; |
---|
1654 | n/a | int i, r; |
---|
1655 | n/a | |
---|
1656 | n/a | if (!PySequence_Check(lines)) { |
---|
1657 | n/a | PyErr_SetString(PyExc_TypeError, |
---|
1658 | n/a | "arg must be a sequence object"); |
---|
1659 | n/a | return NULL; |
---|
1660 | n/a | } |
---|
1661 | n/a | |
---|
1662 | n/a | for (i = 0; i < PySequence_Length(lines); i++) { |
---|
1663 | n/a | /* length can be changed even within this loop */ |
---|
1664 | n/a | strobj = PySequence_GetItem(lines, i); |
---|
1665 | n/a | if (strobj == NULL) |
---|
1666 | n/a | return NULL; |
---|
1667 | n/a | |
---|
1668 | n/a | r = mbstreamwriter_iwrite(self, strobj); |
---|
1669 | n/a | Py_DECREF(strobj); |
---|
1670 | n/a | if (r == -1) |
---|
1671 | n/a | return NULL; |
---|
1672 | n/a | } |
---|
1673 | n/a | |
---|
1674 | n/a | Py_RETURN_NONE; |
---|
1675 | n/a | } |
---|
1676 | n/a | |
---|
1677 | n/a | /*[clinic input] |
---|
1678 | n/a | _multibytecodec.MultibyteStreamWriter.reset |
---|
1679 | n/a | [clinic start generated code]*/ |
---|
1680 | n/a | |
---|
1681 | n/a | static PyObject * |
---|
1682 | n/a | _multibytecodec_MultibyteStreamWriter_reset_impl(MultibyteStreamWriterObject *self) |
---|
1683 | n/a | /*[clinic end generated code: output=8f54a4d9b03db5ff input=b56dbcbaf35cc10c]*/ |
---|
1684 | n/a | { |
---|
1685 | n/a | PyObject *pwrt; |
---|
1686 | n/a | |
---|
1687 | n/a | if (!self->pending) |
---|
1688 | n/a | Py_RETURN_NONE; |
---|
1689 | n/a | |
---|
1690 | n/a | pwrt = multibytecodec_encode(self->codec, &self->state, |
---|
1691 | n/a | self->pending, NULL, self->errors, |
---|
1692 | n/a | MBENC_FLUSH | MBENC_RESET); |
---|
1693 | n/a | /* some pending buffer can be truncated when UnicodeEncodeError is |
---|
1694 | n/a | * raised on 'strict' mode. but, 'reset' method is designed to |
---|
1695 | n/a | * reset the pending buffer or states so failed string sequence |
---|
1696 | n/a | * ought to be missed */ |
---|
1697 | n/a | Py_CLEAR(self->pending); |
---|
1698 | n/a | if (pwrt == NULL) |
---|
1699 | n/a | return NULL; |
---|
1700 | n/a | |
---|
1701 | n/a | assert(PyBytes_Check(pwrt)); |
---|
1702 | n/a | if (PyBytes_Size(pwrt) > 0) { |
---|
1703 | n/a | PyObject *wr; |
---|
1704 | n/a | |
---|
1705 | n/a | wr = _PyObject_CallMethodIdObjArgs(self->stream, &PyId_write, pwrt); |
---|
1706 | n/a | if (wr == NULL) { |
---|
1707 | n/a | Py_DECREF(pwrt); |
---|
1708 | n/a | return NULL; |
---|
1709 | n/a | } |
---|
1710 | n/a | } |
---|
1711 | n/a | Py_DECREF(pwrt); |
---|
1712 | n/a | |
---|
1713 | n/a | Py_RETURN_NONE; |
---|
1714 | n/a | } |
---|
1715 | n/a | |
---|
1716 | n/a | static PyObject * |
---|
1717 | n/a | mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
---|
1718 | n/a | { |
---|
1719 | n/a | MultibyteStreamWriterObject *self; |
---|
1720 | n/a | PyObject *stream, *codec = NULL; |
---|
1721 | n/a | char *errors = NULL; |
---|
1722 | n/a | |
---|
1723 | n/a | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter", |
---|
1724 | n/a | streamkwarglist, &stream, &errors)) |
---|
1725 | n/a | return NULL; |
---|
1726 | n/a | |
---|
1727 | n/a | self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0); |
---|
1728 | n/a | if (self == NULL) |
---|
1729 | n/a | return NULL; |
---|
1730 | n/a | |
---|
1731 | n/a | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
---|
1732 | n/a | if (codec == NULL) |
---|
1733 | n/a | goto errorexit; |
---|
1734 | n/a | if (!MultibyteCodec_Check(codec)) { |
---|
1735 | n/a | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
---|
1736 | n/a | goto errorexit; |
---|
1737 | n/a | } |
---|
1738 | n/a | |
---|
1739 | n/a | self->codec = ((MultibyteCodecObject *)codec)->codec; |
---|
1740 | n/a | self->stream = stream; |
---|
1741 | n/a | Py_INCREF(stream); |
---|
1742 | n/a | self->pending = NULL; |
---|
1743 | n/a | self->errors = internal_error_callback(errors); |
---|
1744 | n/a | if (self->errors == NULL) |
---|
1745 | n/a | goto errorexit; |
---|
1746 | n/a | if (self->codec->encinit != NULL && |
---|
1747 | n/a | self->codec->encinit(&self->state, self->codec->config) != 0) |
---|
1748 | n/a | goto errorexit; |
---|
1749 | n/a | |
---|
1750 | n/a | Py_DECREF(codec); |
---|
1751 | n/a | return (PyObject *)self; |
---|
1752 | n/a | |
---|
1753 | n/a | errorexit: |
---|
1754 | n/a | Py_XDECREF(self); |
---|
1755 | n/a | Py_XDECREF(codec); |
---|
1756 | n/a | return NULL; |
---|
1757 | n/a | } |
---|
1758 | n/a | |
---|
1759 | n/a | static int |
---|
1760 | n/a | mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds) |
---|
1761 | n/a | { |
---|
1762 | n/a | return 0; |
---|
1763 | n/a | } |
---|
1764 | n/a | |
---|
1765 | n/a | static int |
---|
1766 | n/a | mbstreamwriter_traverse(MultibyteStreamWriterObject *self, |
---|
1767 | n/a | visitproc visit, void *arg) |
---|
1768 | n/a | { |
---|
1769 | n/a | if (ERROR_ISCUSTOM(self->errors)) |
---|
1770 | n/a | Py_VISIT(self->errors); |
---|
1771 | n/a | Py_VISIT(self->stream); |
---|
1772 | n/a | return 0; |
---|
1773 | n/a | } |
---|
1774 | n/a | |
---|
1775 | n/a | static void |
---|
1776 | n/a | mbstreamwriter_dealloc(MultibyteStreamWriterObject *self) |
---|
1777 | n/a | { |
---|
1778 | n/a | PyObject_GC_UnTrack(self); |
---|
1779 | n/a | ERROR_DECREF(self->errors); |
---|
1780 | n/a | Py_XDECREF(self->stream); |
---|
1781 | n/a | Py_TYPE(self)->tp_free(self); |
---|
1782 | n/a | } |
---|
1783 | n/a | |
---|
1784 | n/a | static struct PyMethodDef mbstreamwriter_methods[] = { |
---|
1785 | n/a | _MULTIBYTECODEC_MULTIBYTESTREAMWRITER_WRITE_METHODDEF |
---|
1786 | n/a | _MULTIBYTECODEC_MULTIBYTESTREAMWRITER_WRITELINES_METHODDEF |
---|
1787 | n/a | _MULTIBYTECODEC_MULTIBYTESTREAMWRITER_RESET_METHODDEF |
---|
1788 | n/a | {NULL, NULL}, |
---|
1789 | n/a | }; |
---|
1790 | n/a | |
---|
1791 | n/a | static PyMemberDef mbstreamwriter_members[] = { |
---|
1792 | n/a | {"stream", T_OBJECT, |
---|
1793 | n/a | offsetof(MultibyteStreamWriterObject, stream), |
---|
1794 | n/a | READONLY, NULL}, |
---|
1795 | n/a | {NULL,} |
---|
1796 | n/a | }; |
---|
1797 | n/a | |
---|
1798 | n/a | static PyTypeObject MultibyteStreamWriter_Type = { |
---|
1799 | n/a | PyVarObject_HEAD_INIT(NULL, 0) |
---|
1800 | n/a | "MultibyteStreamWriter", /* tp_name */ |
---|
1801 | n/a | sizeof(MultibyteStreamWriterObject), /* tp_basicsize */ |
---|
1802 | n/a | 0, /* tp_itemsize */ |
---|
1803 | n/a | /* methods */ |
---|
1804 | n/a | (destructor)mbstreamwriter_dealloc, /* tp_dealloc */ |
---|
1805 | n/a | 0, /* tp_print */ |
---|
1806 | n/a | 0, /* tp_getattr */ |
---|
1807 | n/a | 0, /* tp_setattr */ |
---|
1808 | n/a | 0, /* tp_reserved */ |
---|
1809 | n/a | 0, /* tp_repr */ |
---|
1810 | n/a | 0, /* tp_as_number */ |
---|
1811 | n/a | 0, /* tp_as_sequence */ |
---|
1812 | n/a | 0, /* tp_as_mapping */ |
---|
1813 | n/a | 0, /* tp_hash */ |
---|
1814 | n/a | 0, /* tp_call */ |
---|
1815 | n/a | 0, /* tp_str */ |
---|
1816 | n/a | PyObject_GenericGetAttr, /* tp_getattro */ |
---|
1817 | n/a | 0, /* tp_setattro */ |
---|
1818 | n/a | 0, /* tp_as_buffer */ |
---|
1819 | n/a | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
---|
1820 | n/a | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
---|
1821 | n/a | 0, /* tp_doc */ |
---|
1822 | n/a | (traverseproc)mbstreamwriter_traverse, /* tp_traverse */ |
---|
1823 | n/a | 0, /* tp_clear */ |
---|
1824 | n/a | 0, /* tp_richcompare */ |
---|
1825 | n/a | 0, /* tp_weaklistoffset */ |
---|
1826 | n/a | 0, /* tp_iter */ |
---|
1827 | n/a | 0, /* tp_iterext */ |
---|
1828 | n/a | mbstreamwriter_methods, /* tp_methods */ |
---|
1829 | n/a | mbstreamwriter_members, /* tp_members */ |
---|
1830 | n/a | codecctx_getsets, /* tp_getset */ |
---|
1831 | n/a | 0, /* tp_base */ |
---|
1832 | n/a | 0, /* tp_dict */ |
---|
1833 | n/a | 0, /* tp_descr_get */ |
---|
1834 | n/a | 0, /* tp_descr_set */ |
---|
1835 | n/a | 0, /* tp_dictoffset */ |
---|
1836 | n/a | mbstreamwriter_init, /* tp_init */ |
---|
1837 | n/a | 0, /* tp_alloc */ |
---|
1838 | n/a | mbstreamwriter_new, /* tp_new */ |
---|
1839 | n/a | }; |
---|
1840 | n/a | |
---|
1841 | n/a | |
---|
1842 | n/a | /*[clinic input] |
---|
1843 | n/a | _multibytecodec.__create_codec |
---|
1844 | n/a | |
---|
1845 | n/a | arg: object |
---|
1846 | n/a | / |
---|
1847 | n/a | [clinic start generated code]*/ |
---|
1848 | n/a | |
---|
1849 | n/a | static PyObject * |
---|
1850 | n/a | _multibytecodec___create_codec(PyObject *module, PyObject *arg) |
---|
1851 | n/a | /*[clinic end generated code: output=cfa3dce8260e809d input=6840b2a6b183fcfa]*/ |
---|
1852 | n/a | { |
---|
1853 | n/a | MultibyteCodecObject *self; |
---|
1854 | n/a | MultibyteCodec *codec; |
---|
1855 | n/a | |
---|
1856 | n/a | if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) { |
---|
1857 | n/a | PyErr_SetString(PyExc_ValueError, "argument type invalid"); |
---|
1858 | n/a | return NULL; |
---|
1859 | n/a | } |
---|
1860 | n/a | |
---|
1861 | n/a | codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME); |
---|
1862 | n/a | if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0) |
---|
1863 | n/a | return NULL; |
---|
1864 | n/a | |
---|
1865 | n/a | self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type); |
---|
1866 | n/a | if (self == NULL) |
---|
1867 | n/a | return NULL; |
---|
1868 | n/a | self->codec = codec; |
---|
1869 | n/a | |
---|
1870 | n/a | return (PyObject *)self; |
---|
1871 | n/a | } |
---|
1872 | n/a | |
---|
1873 | n/a | static struct PyMethodDef __methods[] = { |
---|
1874 | n/a | _MULTIBYTECODEC___CREATE_CODEC_METHODDEF |
---|
1875 | n/a | {NULL, NULL}, |
---|
1876 | n/a | }; |
---|
1877 | n/a | |
---|
1878 | n/a | |
---|
1879 | n/a | static struct PyModuleDef _multibytecodecmodule = { |
---|
1880 | n/a | PyModuleDef_HEAD_INIT, |
---|
1881 | n/a | "_multibytecodec", |
---|
1882 | n/a | NULL, |
---|
1883 | n/a | -1, |
---|
1884 | n/a | __methods, |
---|
1885 | n/a | NULL, |
---|
1886 | n/a | NULL, |
---|
1887 | n/a | NULL, |
---|
1888 | n/a | NULL |
---|
1889 | n/a | }; |
---|
1890 | n/a | |
---|
1891 | n/a | PyMODINIT_FUNC |
---|
1892 | n/a | PyInit__multibytecodec(void) |
---|
1893 | n/a | { |
---|
1894 | n/a | int i; |
---|
1895 | n/a | PyObject *m; |
---|
1896 | n/a | PyTypeObject *typelist[] = { |
---|
1897 | n/a | &MultibyteIncrementalEncoder_Type, |
---|
1898 | n/a | &MultibyteIncrementalDecoder_Type, |
---|
1899 | n/a | &MultibyteStreamReader_Type, |
---|
1900 | n/a | &MultibyteStreamWriter_Type, |
---|
1901 | n/a | NULL |
---|
1902 | n/a | }; |
---|
1903 | n/a | |
---|
1904 | n/a | if (PyType_Ready(&MultibyteCodec_Type) < 0) |
---|
1905 | n/a | return NULL; |
---|
1906 | n/a | |
---|
1907 | n/a | m = PyModule_Create(&_multibytecodecmodule); |
---|
1908 | n/a | if (m == NULL) |
---|
1909 | n/a | return NULL; |
---|
1910 | n/a | |
---|
1911 | n/a | for (i = 0; typelist[i] != NULL; i++) { |
---|
1912 | n/a | if (PyType_Ready(typelist[i]) < 0) |
---|
1913 | n/a | return NULL; |
---|
1914 | n/a | Py_INCREF(typelist[i]); |
---|
1915 | n/a | PyModule_AddObject(m, typelist[i]->tp_name, |
---|
1916 | n/a | (PyObject *)typelist[i]); |
---|
1917 | n/a | } |
---|
1918 | n/a | |
---|
1919 | n/a | if (PyErr_Occurred()) { |
---|
1920 | n/a | Py_FatalError("can't initialize the _multibytecodec module"); |
---|
1921 | n/a | Py_DECREF(m); |
---|
1922 | n/a | m = NULL; |
---|
1923 | n/a | } |
---|
1924 | n/a | return m; |
---|
1925 | n/a | } |
---|