| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1 | /* | 
 | 2 |  * multibytecodec.c: Common Multibyte Codec Implementation | 
 | 3 |  * | 
 | 4 |  * Written by Hye-Shik Chang <perky@FreeBSD.org> | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 5 |  */ | 
 | 6 |  | 
| Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 7 | #define PY_SSIZE_T_CLEAN | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 8 | #include "Python.h" | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 9 | #include "structmember.h" | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 10 | #include "multibytecodec.h" | 
 | 11 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 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\ | 
| Benjamin Peterson | f10a79a | 2008-10-11 00:49:57 +0000 | [diff] [blame] | 39 | are 'ignore' and 'replace' as well as any other name registered with\n\ | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 40 | codecs.register_error that is able to handle UnicodeDecodeErrors."); | 
 | 41 |  | 
| Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 42 | static char *codeckwarglist[] = {"input", "errors", NULL}; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 43 | static char *incnewkwarglist[] = {"errors", NULL}; | 
 | 44 | static char *incrementalkwarglist[] = {"input", "final", NULL}; | 
| Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 45 | static char *streamkwarglist[] = {"stream", "errors", NULL}; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 46 |  | 
 | 47 | static PyObject *multibytecodec_encode(MultibyteCodec *, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 48 |                 MultibyteCodec_State *, const Py_UNICODE **, Py_ssize_t, | 
 | 49 |                 PyObject *, int); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 50 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 51 | #define MBENC_RESET     MBENC_MAX<<1 /* reset after an encoding session */ | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 52 |  | 
 | 53 | static PyObject * | 
| Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 54 | make_tuple(PyObject *object, Py_ssize_t len) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 55 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 56 |     PyObject *v, *w; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 57 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 58 |     if (object == NULL) | 
 | 59 |         return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 60 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 61 |     v = PyTuple_New(2); | 
 | 62 |     if (v == NULL) { | 
 | 63 |         Py_DECREF(object); | 
 | 64 |         return NULL; | 
 | 65 |     } | 
 | 66 |     PyTuple_SET_ITEM(v, 0, object); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 67 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 68 |     w = PyLong_FromSsize_t(len); | 
 | 69 |     if (w == NULL) { | 
 | 70 |         Py_DECREF(v); | 
 | 71 |         return NULL; | 
 | 72 |     } | 
 | 73 |     PyTuple_SET_ITEM(v, 1, w); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 74 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 75 |     return v; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 76 | } | 
 | 77 |  | 
 | 78 | static PyObject * | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 79 | internal_error_callback(const char *errors) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 80 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 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 PyUnicode_FromString(errors); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 89 | } | 
 | 90 |  | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 91 | static PyObject * | 
 | 92 | call_error_callback(PyObject *errors, PyObject *exc) | 
 | 93 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 94 |     PyObject *args, *cb, *r; | 
 | 95 |     const char *str; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 96 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 97 |     assert(PyUnicode_Check(errors)); | 
 | 98 |     str = _PyUnicode_AsString(errors); | 
 | 99 |     if (str == NULL) | 
 | 100 |         return NULL; | 
 | 101 |     cb = PyCodec_LookupError(str); | 
 | 102 |     if (cb == NULL) | 
 | 103 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 104 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 105 |     args = PyTuple_New(1); | 
 | 106 |     if (args == NULL) { | 
 | 107 |         Py_DECREF(cb); | 
 | 108 |         return NULL; | 
 | 109 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 110 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 111 |     PyTuple_SET_ITEM(args, 0, exc); | 
 | 112 |     Py_INCREF(exc); | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 113 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 114 |     r = PyObject_CallObject(cb, args); | 
 | 115 |     Py_DECREF(args); | 
 | 116 |     Py_DECREF(cb); | 
 | 117 |     return r; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 118 | } | 
 | 119 |  | 
 | 120 | static PyObject * | 
 | 121 | codecctx_errors_get(MultibyteStatefulCodecContext *self) | 
 | 122 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 123 |     const char *errors; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 124 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 125 |     if (self->errors == ERROR_STRICT) | 
 | 126 |         errors = "strict"; | 
 | 127 |     else if (self->errors == ERROR_IGNORE) | 
 | 128 |         errors = "ignore"; | 
 | 129 |     else if (self->errors == ERROR_REPLACE) | 
 | 130 |         errors = "replace"; | 
 | 131 |     else { | 
 | 132 |         Py_INCREF(self->errors); | 
 | 133 |         return self->errors; | 
 | 134 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 135 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 136 |     return PyUnicode_FromString(errors); | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 137 | } | 
 | 138 |  | 
 | 139 | static int | 
 | 140 | codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 141 |                     void *closure) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 142 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 143 |     PyObject *cb; | 
 | 144 |     const char *str; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 145 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 146 |     if (!PyUnicode_Check(value)) { | 
 | 147 |         PyErr_SetString(PyExc_TypeError, "errors must be a string"); | 
 | 148 |         return -1; | 
 | 149 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 150 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 151 |     str = _PyUnicode_AsString(value); | 
 | 152 |     if (str == NULL) | 
 | 153 |         return -1; | 
| Neal Norwitz | 6ea45d3 | 2007-08-26 04:19:43 +0000 | [diff] [blame] | 154 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 155 |     cb = internal_error_callback(str); | 
 | 156 |     if (cb == NULL) | 
 | 157 |         return -1; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 158 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 159 |     ERROR_DECREF(self->errors); | 
 | 160 |     self->errors = cb; | 
 | 161 |     return 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 162 | } | 
 | 163 |  | 
 | 164 | /* This getset handlers list is used by all the stateful codec objects */ | 
 | 165 | static PyGetSetDef codecctx_getsets[] = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 166 |     {"errors",          (getter)codecctx_errors_get, | 
 | 167 |                     (setter)codecctx_errors_set, | 
 | 168 |                     PyDoc_STR("how to treat errors")}, | 
 | 169 |     {NULL,} | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 170 | }; | 
 | 171 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 172 | static int | 
| Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 173 | expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 174 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 175 |     Py_ssize_t orgpos, orgsize, incsize; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 176 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 177 |     orgpos = (Py_ssize_t)((char *)buf->outbuf - | 
 | 178 |                             PyBytes_AS_STRING(buf->outobj)); | 
 | 179 |     orgsize = PyBytes_GET_SIZE(buf->outobj); | 
 | 180 |     incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize); | 
| Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 181 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 182 |     if (orgsize > PY_SSIZE_T_MAX - incsize) | 
 | 183 |         return -1; | 
| Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 184 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 185 |     if (_PyBytes_Resize(&buf->outobj, orgsize + incsize) == -1) | 
 | 186 |         return -1; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 187 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 188 |     buf->outbuf = (unsigned char *)PyBytes_AS_STRING(buf->outobj) +orgpos; | 
 | 189 |     buf->outbuf_end = (unsigned char *)PyBytes_AS_STRING(buf->outobj) | 
 | 190 |         + PyBytes_GET_SIZE(buf->outobj); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 191 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 192 |     return 0; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 193 | } | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 194 | #define REQUIRE_ENCODEBUFFER(buf, s) {                                  \ | 
 | 195 |     if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end)             \ | 
 | 196 |         if (expand_encodebuffer(buf, s) == -1)                          \ | 
 | 197 |             goto errorexit;                                             \ | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 198 | } | 
 | 199 |  | 
 | 200 | static int | 
| Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 201 | expand_decodebuffer(MultibyteDecodeBuffer *buf, Py_ssize_t esize) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 202 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 203 |     Py_ssize_t orgpos, orgsize; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 204 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 205 |     orgpos = (Py_ssize_t)(buf->outbuf - PyUnicode_AS_UNICODE(buf->outobj)); | 
 | 206 |     orgsize = PyUnicode_GET_SIZE(buf->outobj); | 
 | 207 |     if (PyUnicode_Resize(&buf->outobj, orgsize + ( | 
 | 208 |         esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize)) == -1) | 
 | 209 |         return -1; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 210 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 211 |     buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj) + orgpos; | 
 | 212 |     buf->outbuf_end = PyUnicode_AS_UNICODE(buf->outobj) | 
 | 213 |                       + PyUnicode_GET_SIZE(buf->outobj); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 214 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 215 |     return 0; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 216 | } | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 217 | #define REQUIRE_DECODEBUFFER(buf, s) {                                  \ | 
 | 218 |     if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end)             \ | 
 | 219 |         if (expand_decodebuffer(buf, s) == -1)                          \ | 
 | 220 |             goto errorexit;                                             \ | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 221 | } | 
 | 222 |  | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 223 |  | 
 | 224 | /** | 
 | 225 |  * MultibyteCodec object | 
 | 226 |  */ | 
 | 227 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 228 | static int | 
 | 229 | multibytecodec_encerror(MultibyteCodec *codec, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 230 |                         MultibyteCodec_State *state, | 
 | 231 |                         MultibyteEncodeBuffer *buf, | 
 | 232 |                         PyObject *errors, Py_ssize_t e) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 233 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 234 |     PyObject *retobj = NULL, *retstr = NULL, *tobj; | 
 | 235 |     Py_ssize_t retstrsize, newpos; | 
 | 236 |     Py_ssize_t esize, start, end; | 
 | 237 |     const char *reason; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 238 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 239 |     if (e > 0) { | 
 | 240 |         reason = "illegal multibyte sequence"; | 
 | 241 |         esize = e; | 
 | 242 |     } | 
 | 243 |     else { | 
 | 244 |         switch (e) { | 
 | 245 |         case MBERR_TOOSMALL: | 
 | 246 |             REQUIRE_ENCODEBUFFER(buf, -1); | 
 | 247 |             return 0; /* retry it */ | 
 | 248 |         case MBERR_TOOFEW: | 
 | 249 |             reason = "incomplete multibyte sequence"; | 
 | 250 |             esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); | 
 | 251 |             break; | 
 | 252 |         case MBERR_INTERNAL: | 
 | 253 |             PyErr_SetString(PyExc_RuntimeError, | 
 | 254 |                             "internal codec error"); | 
 | 255 |             return -1; | 
 | 256 |         default: | 
 | 257 |             PyErr_SetString(PyExc_RuntimeError, | 
 | 258 |                             "unknown runtime error"); | 
 | 259 |             return -1; | 
 | 260 |         } | 
 | 261 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 262 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 263 |     if (errors == ERROR_REPLACE) { | 
 | 264 |         const Py_UNICODE replchar = '?', *inbuf = &replchar; | 
 | 265 |         Py_ssize_t r; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 266 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 267 |         for (;;) { | 
 | 268 |             Py_ssize_t outleft; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 269 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 270 |             outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf); | 
 | 271 |             r = codec->encode(state, codec->config, &inbuf, 1, | 
 | 272 |                               &buf->outbuf, outleft, 0); | 
 | 273 |             if (r == MBERR_TOOSMALL) { | 
 | 274 |                 REQUIRE_ENCODEBUFFER(buf, -1); | 
 | 275 |                 continue; | 
 | 276 |             } | 
 | 277 |             else | 
 | 278 |                 break; | 
 | 279 |         } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 280 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 281 |         if (r != 0) { | 
 | 282 |             REQUIRE_ENCODEBUFFER(buf, 1); | 
 | 283 |             *buf->outbuf++ = '?'; | 
 | 284 |         } | 
 | 285 |     } | 
 | 286 |     if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { | 
 | 287 |         buf->inbuf += esize; | 
 | 288 |         return 0; | 
 | 289 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 290 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 291 |     start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top); | 
 | 292 |     end = start + esize; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 293 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 294 |     /* use cached exception object if available */ | 
 | 295 |     if (buf->excobj == NULL) { | 
 | 296 |         buf->excobj = PyUnicodeEncodeError_Create(codec->encoding, | 
 | 297 |                         buf->inbuf_top, | 
 | 298 |                         buf->inbuf_end - buf->inbuf_top, | 
 | 299 |                         start, end, reason); | 
 | 300 |         if (buf->excobj == NULL) | 
 | 301 |             goto errorexit; | 
 | 302 |     } | 
 | 303 |     else | 
 | 304 |         if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 || | 
 | 305 |             PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 || | 
 | 306 |             PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0) | 
 | 307 |             goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 308 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 309 |     if (errors == ERROR_STRICT) { | 
 | 310 |         PyCodec_StrictErrors(buf->excobj); | 
 | 311 |         goto errorexit; | 
 | 312 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 313 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 314 |     retobj = call_error_callback(errors, buf->excobj); | 
 | 315 |     if (retobj == NULL) | 
 | 316 |         goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 317 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 318 |     if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || | 
 | 319 |         !PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) || | 
 | 320 |         !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { | 
 | 321 |         PyErr_SetString(PyExc_TypeError, | 
 | 322 |                         "encoding error handler must return " | 
 | 323 |                         "(unicode, int) tuple"); | 
 | 324 |         goto errorexit; | 
 | 325 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 326 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 327 |     { | 
 | 328 |         const Py_UNICODE *uraw = PyUnicode_AS_UNICODE(tobj); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 329 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 330 |         retstr = multibytecodec_encode(codec, state, &uraw, | 
 | 331 |                         PyUnicode_GET_SIZE(tobj), ERROR_STRICT, | 
 | 332 |                         MBENC_FLUSH); | 
 | 333 |         if (retstr == NULL) | 
 | 334 |             goto errorexit; | 
 | 335 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 336 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 337 |     assert(PyBytes_Check(retstr)); | 
 | 338 |     retstrsize = PyBytes_GET_SIZE(retstr); | 
 | 339 |     REQUIRE_ENCODEBUFFER(buf, retstrsize); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 340 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 341 |     memcpy(buf->outbuf, PyBytes_AS_STRING(retstr), retstrsize); | 
 | 342 |     buf->outbuf += retstrsize; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 343 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 344 |     newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); | 
 | 345 |     if (newpos < 0 && !PyErr_Occurred()) | 
 | 346 |         newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top); | 
 | 347 |     if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) { | 
 | 348 |         PyErr_Clear(); | 
 | 349 |         PyErr_Format(PyExc_IndexError, | 
 | 350 |                      "position %zd from error handler out of bounds", | 
 | 351 |                      newpos); | 
 | 352 |         goto errorexit; | 
 | 353 |     } | 
 | 354 |     buf->inbuf = buf->inbuf_top + newpos; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 355 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 356 |     Py_DECREF(retobj); | 
 | 357 |     Py_DECREF(retstr); | 
 | 358 |     return 0; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 359 |  | 
 | 360 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 361 |     Py_XDECREF(retobj); | 
 | 362 |     Py_XDECREF(retstr); | 
 | 363 |     return -1; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 364 | } | 
 | 365 |  | 
 | 366 | static int | 
 | 367 | multibytecodec_decerror(MultibyteCodec *codec, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 368 |                         MultibyteCodec_State *state, | 
 | 369 |                         MultibyteDecodeBuffer *buf, | 
 | 370 |                         PyObject *errors, Py_ssize_t e) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 371 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 372 |     PyObject *retobj = NULL, *retuni = NULL; | 
 | 373 |     Py_ssize_t retunisize, newpos; | 
 | 374 |     const char *reason; | 
 | 375 |     Py_ssize_t esize, start, end; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 376 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 377 |     if (e > 0) { | 
 | 378 |         reason = "illegal multibyte sequence"; | 
 | 379 |         esize = e; | 
 | 380 |     } | 
 | 381 |     else { | 
 | 382 |         switch (e) { | 
 | 383 |         case MBERR_TOOSMALL: | 
 | 384 |             REQUIRE_DECODEBUFFER(buf, -1); | 
 | 385 |             return 0; /* retry it */ | 
 | 386 |         case MBERR_TOOFEW: | 
 | 387 |             reason = "incomplete multibyte sequence"; | 
 | 388 |             esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); | 
 | 389 |             break; | 
 | 390 |         case MBERR_INTERNAL: | 
 | 391 |             PyErr_SetString(PyExc_RuntimeError, | 
 | 392 |                             "internal codec error"); | 
 | 393 |             return -1; | 
 | 394 |         default: | 
 | 395 |             PyErr_SetString(PyExc_RuntimeError, | 
 | 396 |                             "unknown runtime error"); | 
 | 397 |             return -1; | 
 | 398 |         } | 
 | 399 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 400 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 401 |     if (errors == ERROR_REPLACE) { | 
 | 402 |         REQUIRE_DECODEBUFFER(buf, 1); | 
 | 403 |         *buf->outbuf++ = Py_UNICODE_REPLACEMENT_CHARACTER; | 
 | 404 |     } | 
 | 405 |     if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { | 
 | 406 |         buf->inbuf += esize; | 
 | 407 |         return 0; | 
 | 408 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 409 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 410 |     start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top); | 
 | 411 |     end = start + esize; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 412 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 413 |     /* use cached exception object if available */ | 
 | 414 |     if (buf->excobj == NULL) { | 
 | 415 |         buf->excobj = PyUnicodeDecodeError_Create(codec->encoding, | 
 | 416 |                         (const char *)buf->inbuf_top, | 
 | 417 |                         (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top), | 
 | 418 |                         start, end, reason); | 
 | 419 |         if (buf->excobj == NULL) | 
 | 420 |             goto errorexit; | 
 | 421 |     } | 
 | 422 |     else | 
 | 423 |         if (PyUnicodeDecodeError_SetStart(buf->excobj, start) || | 
 | 424 |             PyUnicodeDecodeError_SetEnd(buf->excobj, end) || | 
 | 425 |             PyUnicodeDecodeError_SetReason(buf->excobj, reason)) | 
 | 426 |             goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 427 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 428 |     if (errors == ERROR_STRICT) { | 
 | 429 |         PyCodec_StrictErrors(buf->excobj); | 
 | 430 |         goto errorexit; | 
 | 431 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 432 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 433 |     retobj = call_error_callback(errors, buf->excobj); | 
 | 434 |     if (retobj == NULL) | 
 | 435 |         goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 436 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 437 |     if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || | 
 | 438 |         !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) || | 
 | 439 |         !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { | 
 | 440 |         PyErr_SetString(PyExc_TypeError, | 
 | 441 |                         "decoding error handler must return " | 
 | 442 |                         "(unicode, int) tuple"); | 
 | 443 |         goto errorexit; | 
 | 444 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 445 |  | 
| Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 446 |     if (PyUnicode_AsUnicode(retuni) == NULL) | 
 | 447 |         goto errorexit; | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 448 |     retunisize = PyUnicode_GET_SIZE(retuni); | 
 | 449 |     if (retunisize > 0) { | 
 | 450 |         REQUIRE_DECODEBUFFER(buf, retunisize); | 
| Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 451 |         memcpy((char *)buf->outbuf, PyUnicode_AS_UNICODE(retuni), | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 452 |                         retunisize * Py_UNICODE_SIZE); | 
 | 453 |         buf->outbuf += retunisize; | 
 | 454 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 455 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 456 |     newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); | 
 | 457 |     if (newpos < 0 && !PyErr_Occurred()) | 
 | 458 |         newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top); | 
 | 459 |     if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) { | 
 | 460 |         PyErr_Clear(); | 
 | 461 |         PyErr_Format(PyExc_IndexError, | 
 | 462 |                      "position %zd from error handler out of bounds", | 
 | 463 |                      newpos); | 
 | 464 |         goto errorexit; | 
 | 465 |     } | 
 | 466 |     buf->inbuf = buf->inbuf_top + newpos; | 
 | 467 |     Py_DECREF(retobj); | 
 | 468 |     return 0; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 469 |  | 
 | 470 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 471 |     Py_XDECREF(retobj); | 
 | 472 |     return -1; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 473 | } | 
 | 474 |  | 
 | 475 | static PyObject * | 
 | 476 | multibytecodec_encode(MultibyteCodec *codec, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 477 |                       MultibyteCodec_State *state, | 
 | 478 |                       const Py_UNICODE **data, Py_ssize_t datalen, | 
 | 479 |                       PyObject *errors, int flags) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 480 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 481 |     MultibyteEncodeBuffer buf; | 
 | 482 |     Py_ssize_t finalsize, r = 0; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 483 |  | 
| Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 484 |     if (datalen == 0 && !(flags & MBENC_RESET)) | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 485 |         return PyBytes_FromStringAndSize(NULL, 0); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 486 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 487 |     buf.excobj = NULL; | 
| Brett Cannon | b94767f | 2011-02-22 20:15:44 +0000 | [diff] [blame] | 488 |     buf.outobj = NULL; | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 489 |     buf.inbuf = buf.inbuf_top = *data; | 
 | 490 |     buf.inbuf_end = buf.inbuf_top + datalen; | 
| Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 491 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 492 |     if (datalen > (PY_SSIZE_T_MAX - 16) / 2) { | 
 | 493 |         PyErr_NoMemory(); | 
 | 494 |         goto errorexit; | 
 | 495 |     } | 
| Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 496 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 497 |     buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16); | 
 | 498 |     if (buf.outobj == NULL) | 
 | 499 |         goto errorexit; | 
 | 500 |     buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj); | 
 | 501 |     buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 502 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 503 |     while (buf.inbuf < buf.inbuf_end) { | 
 | 504 |         Py_ssize_t inleft, outleft; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 505 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 506 |         /* we don't reuse inleft and outleft here. | 
 | 507 |          * error callbacks can relocate the cursor anywhere on buffer*/ | 
 | 508 |         inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf); | 
 | 509 |         outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); | 
 | 510 |         r = codec->encode(state, codec->config, &buf.inbuf, inleft, | 
 | 511 |                           &buf.outbuf, outleft, flags); | 
 | 512 |         if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH))) | 
 | 513 |             break; | 
 | 514 |         else if (multibytecodec_encerror(codec, state, &buf, errors,r)) | 
 | 515 |             goto errorexit; | 
 | 516 |         else if (r == MBERR_TOOFEW) | 
 | 517 |             break; | 
 | 518 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 519 |  | 
| Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 520 |     if (codec->encreset != NULL && (flags & MBENC_RESET)) | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 521 |         for (;;) { | 
 | 522 |             Py_ssize_t outleft; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 523 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 524 |             outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); | 
 | 525 |             r = codec->encreset(state, codec->config, &buf.outbuf, | 
 | 526 |                                 outleft); | 
 | 527 |             if (r == 0) | 
 | 528 |                 break; | 
 | 529 |             else if (multibytecodec_encerror(codec, state, | 
 | 530 |                                              &buf, errors, r)) | 
 | 531 |                 goto errorexit; | 
 | 532 |         } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 533 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 534 |     finalsize = (Py_ssize_t)((char *)buf.outbuf - | 
 | 535 |                              PyBytes_AS_STRING(buf.outobj)); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 536 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 537 |     if (finalsize != PyBytes_GET_SIZE(buf.outobj)) | 
 | 538 |         if (_PyBytes_Resize(&buf.outobj, finalsize) == -1) | 
 | 539 |             goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 540 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 541 |     *data = buf.inbuf; | 
 | 542 |     Py_XDECREF(buf.excobj); | 
 | 543 |     return buf.outobj; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 544 |  | 
 | 545 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 546 |     Py_XDECREF(buf.excobj); | 
 | 547 |     Py_XDECREF(buf.outobj); | 
 | 548 |     return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 549 | } | 
 | 550 |  | 
 | 551 | static PyObject * | 
 | 552 | MultibyteCodec_Encode(MultibyteCodecObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 553 |                       PyObject *args, PyObject *kwargs) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 554 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 555 |     MultibyteCodec_State state; | 
 | 556 |     Py_UNICODE *data; | 
 | 557 |     PyObject *errorcb, *r, *arg, *ucvt; | 
 | 558 |     const char *errors = NULL; | 
 | 559 |     Py_ssize_t datalen; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 560 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 561 |     if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode", | 
 | 562 |                             codeckwarglist, &arg, &errors)) | 
 | 563 |         return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 564 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 565 |     if (PyUnicode_Check(arg)) | 
 | 566 |         ucvt = NULL; | 
 | 567 |     else { | 
 | 568 |         arg = ucvt = PyObject_Str(arg); | 
 | 569 |         if (arg == NULL) | 
 | 570 |             return NULL; | 
 | 571 |         else if (!PyUnicode_Check(arg)) { | 
 | 572 |             PyErr_SetString(PyExc_TypeError, | 
 | 573 |                 "couldn't convert the object to unicode."); | 
 | 574 |             Py_DECREF(ucvt); | 
 | 575 |             return NULL; | 
 | 576 |         } | 
 | 577 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 578 |  | 
| Victor Stinner | 9a80fab | 2011-11-21 02:50:14 +0100 | [diff] [blame] | 579 |     data = PyUnicode_AsUnicodeAndSize(arg, &datalen); | 
 | 580 |     if (data == NULL) { | 
 | 581 |         Py_XDECREF(ucvt); | 
 | 582 |         return NULL; | 
 | 583 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 584 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 585 |     errorcb = internal_error_callback(errors); | 
 | 586 |     if (errorcb == NULL) { | 
 | 587 |         Py_XDECREF(ucvt); | 
 | 588 |         return NULL; | 
 | 589 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 590 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 591 |     if (self->codec->encinit != NULL && | 
 | 592 |         self->codec->encinit(&state, self->codec->config) != 0) | 
 | 593 |         goto errorexit; | 
 | 594 |     r = multibytecodec_encode(self->codec, &state, | 
 | 595 |                     (const Py_UNICODE **)&data, datalen, errorcb, | 
 | 596 |                     MBENC_FLUSH | MBENC_RESET); | 
 | 597 |     if (r == NULL) | 
 | 598 |         goto errorexit; | 
| Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 599 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 600 |     ERROR_DECREF(errorcb); | 
 | 601 |     Py_XDECREF(ucvt); | 
 | 602 |     return make_tuple(r, datalen); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 603 |  | 
 | 604 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 605 |     ERROR_DECREF(errorcb); | 
 | 606 |     Py_XDECREF(ucvt); | 
 | 607 |     return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 608 | } | 
 | 609 |  | 
 | 610 | static PyObject * | 
 | 611 | MultibyteCodec_Decode(MultibyteCodecObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 612 |                       PyObject *args, PyObject *kwargs) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 613 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 614 |     MultibyteCodec_State state; | 
 | 615 |     MultibyteDecodeBuffer buf; | 
 | 616 |     PyObject *errorcb; | 
 | 617 |     Py_buffer pdata; | 
 | 618 |     const char *data, *errors = NULL; | 
 | 619 |     Py_ssize_t datalen, finalsize; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 620 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 621 |     if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|z:decode", | 
 | 622 |                             codeckwarglist, &pdata, &errors)) | 
 | 623 |         return NULL; | 
 | 624 |     data = pdata.buf; | 
 | 625 |     datalen = pdata.len; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 626 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 627 |     errorcb = internal_error_callback(errors); | 
 | 628 |     if (errorcb == NULL) { | 
 | 629 |         PyBuffer_Release(&pdata); | 
 | 630 |         return NULL; | 
 | 631 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 632 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 633 |     if (datalen == 0) { | 
 | 634 |         PyBuffer_Release(&pdata); | 
 | 635 |         ERROR_DECREF(errorcb); | 
| Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 636 |         return make_tuple(PyUnicode_New(0, 0), 0); | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 637 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 638 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 639 |     buf.excobj = NULL; | 
 | 640 |     buf.inbuf = buf.inbuf_top = (unsigned char *)data; | 
 | 641 |     buf.inbuf_end = buf.inbuf_top + datalen; | 
 | 642 |     buf.outobj = PyUnicode_FromUnicode(NULL, datalen); | 
 | 643 |     if (buf.outobj == NULL) | 
 | 644 |         goto errorexit; | 
 | 645 |     buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj); | 
| Victor Stinner | 08b523a | 2011-12-01 03:18:30 +0100 | [diff] [blame] | 646 |     if (buf.outbuf == NULL) | 
 | 647 |         goto errorexit; | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 648 |     buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 649 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 650 |     if (self->codec->decinit != NULL && | 
 | 651 |         self->codec->decinit(&state, self->codec->config) != 0) | 
 | 652 |         goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 653 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 654 |     while (buf.inbuf < buf.inbuf_end) { | 
 | 655 |         Py_ssize_t inleft, outleft, r; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 656 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 657 |         inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf); | 
 | 658 |         outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 659 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 660 |         r = self->codec->decode(&state, self->codec->config, | 
 | 661 |                         &buf.inbuf, inleft, &buf.outbuf, outleft); | 
 | 662 |         if (r == 0) | 
 | 663 |             break; | 
 | 664 |         else if (multibytecodec_decerror(self->codec, &state, | 
 | 665 |                                          &buf, errorcb, r)) | 
 | 666 |             goto errorexit; | 
 | 667 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 668 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 669 |     finalsize = (Py_ssize_t)(buf.outbuf - | 
 | 670 |                              PyUnicode_AS_UNICODE(buf.outobj)); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 671 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 672 |     if (finalsize != PyUnicode_GET_SIZE(buf.outobj)) | 
 | 673 |         if (PyUnicode_Resize(&buf.outobj, finalsize) == -1) | 
 | 674 |             goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 675 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 676 |     PyBuffer_Release(&pdata); | 
 | 677 |     Py_XDECREF(buf.excobj); | 
 | 678 |     ERROR_DECREF(errorcb); | 
 | 679 |     return make_tuple(buf.outobj, datalen); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 680 |  | 
 | 681 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 682 |     PyBuffer_Release(&pdata); | 
 | 683 |     ERROR_DECREF(errorcb); | 
 | 684 |     Py_XDECREF(buf.excobj); | 
 | 685 |     Py_XDECREF(buf.outobj); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 686 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 687 |     return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 688 | } | 
 | 689 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 690 | static struct PyMethodDef multibytecodec_methods[] = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 691 |     {"encode",          (PyCFunction)MultibyteCodec_Encode, | 
 | 692 |                     METH_VARARGS | METH_KEYWORDS, | 
 | 693 |                     MultibyteCodec_Encode__doc__}, | 
 | 694 |     {"decode",          (PyCFunction)MultibyteCodec_Decode, | 
 | 695 |                     METH_VARARGS | METH_KEYWORDS, | 
 | 696 |                     MultibyteCodec_Decode__doc__}, | 
 | 697 |     {NULL,              NULL}, | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 698 | }; | 
 | 699 |  | 
 | 700 | static void | 
 | 701 | multibytecodec_dealloc(MultibyteCodecObject *self) | 
 | 702 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 703 |     PyObject_Del(self); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 704 | } | 
 | 705 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 706 | static PyTypeObject MultibyteCodec_Type = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 707 |     PyVarObject_HEAD_INIT(NULL, 0) | 
 | 708 |     "MultibyteCodec",                   /* tp_name */ | 
 | 709 |     sizeof(MultibyteCodecObject),       /* tp_basicsize */ | 
 | 710 |     0,                                  /* tp_itemsize */ | 
 | 711 |     /* methods */ | 
 | 712 |     (destructor)multibytecodec_dealloc, /* tp_dealloc */ | 
 | 713 |     0,                                  /* tp_print */ | 
 | 714 |     0,                                  /* tp_getattr */ | 
 | 715 |     0,                                  /* tp_setattr */ | 
 | 716 |     0,                                  /* tp_reserved */ | 
 | 717 |     0,                                  /* tp_repr */ | 
 | 718 |     0,                                  /* tp_as_number */ | 
 | 719 |     0,                                  /* tp_as_sequence */ | 
 | 720 |     0,                                  /* tp_as_mapping */ | 
 | 721 |     0,                                  /* tp_hash */ | 
 | 722 |     0,                                  /* tp_call */ | 
 | 723 |     0,                                  /* tp_str */ | 
 | 724 |     PyObject_GenericGetAttr,            /* tp_getattro */ | 
 | 725 |     0,                                  /* tp_setattro */ | 
 | 726 |     0,                                  /* tp_as_buffer */ | 
 | 727 |     Py_TPFLAGS_DEFAULT,                 /* tp_flags */ | 
 | 728 |     0,                                  /* tp_doc */ | 
 | 729 |     0,                                  /* tp_traverse */ | 
 | 730 |     0,                                  /* tp_clear */ | 
 | 731 |     0,                                  /* tp_richcompare */ | 
 | 732 |     0,                                  /* tp_weaklistoffset */ | 
 | 733 |     0,                                  /* tp_iter */ | 
 | 734 |     0,                                  /* tp_iterext */ | 
 | 735 |     multibytecodec_methods,             /* tp_methods */ | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 736 | }; | 
 | 737 |  | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 738 |  | 
 | 739 | /** | 
 | 740 |  * Utility functions for stateful codec mechanism | 
 | 741 |  */ | 
 | 742 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 743 | #define STATEFUL_DCTX(o)        ((MultibyteStatefulDecoderContext *)(o)) | 
 | 744 | #define STATEFUL_ECTX(o)        ((MultibyteStatefulEncoderContext *)(o)) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 745 |  | 
 | 746 | static PyObject * | 
 | 747 | encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 748 |                         PyObject *unistr, int final) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 749 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 750 |     PyObject *ucvt, *r = NULL; | 
 | 751 |     Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL; | 
 | 752 |     Py_ssize_t datalen, origpending; | 
| Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 753 |     wchar_t *data; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 754 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 755 |     if (PyUnicode_Check(unistr)) | 
 | 756 |         ucvt = NULL; | 
 | 757 |     else { | 
 | 758 |         unistr = ucvt = PyObject_Str(unistr); | 
 | 759 |         if (unistr == NULL) | 
 | 760 |             return NULL; | 
 | 761 |         else if (!PyUnicode_Check(unistr)) { | 
 | 762 |             PyErr_SetString(PyExc_TypeError, | 
 | 763 |                 "couldn't convert the object to unicode."); | 
 | 764 |             Py_DECREF(ucvt); | 
 | 765 |             return NULL; | 
 | 766 |         } | 
 | 767 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 768 |  | 
| Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 769 |     data = PyUnicode_AsUnicodeAndSize(unistr, &datalen); | 
 | 770 |     if (data == NULL) | 
 | 771 |         goto errorexit; | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 772 |     origpending = ctx->pendingsize; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 773 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 774 |     if (origpending > 0) { | 
 | 775 |         if (datalen > PY_SSIZE_T_MAX - ctx->pendingsize) { | 
 | 776 |             PyErr_NoMemory(); | 
 | 777 |             /* inbuf_tmp == NULL */ | 
 | 778 |             goto errorexit; | 
 | 779 |         } | 
 | 780 |         inbuf_tmp = PyMem_New(Py_UNICODE, datalen + ctx->pendingsize); | 
 | 781 |         if (inbuf_tmp == NULL) | 
 | 782 |             goto errorexit; | 
 | 783 |         memcpy(inbuf_tmp, ctx->pending, | 
 | 784 |             Py_UNICODE_SIZE * ctx->pendingsize); | 
 | 785 |         memcpy(inbuf_tmp + ctx->pendingsize, | 
 | 786 |             PyUnicode_AS_UNICODE(unistr), | 
 | 787 |             Py_UNICODE_SIZE * datalen); | 
 | 788 |         datalen += ctx->pendingsize; | 
 | 789 |         ctx->pendingsize = 0; | 
 | 790 |         inbuf = inbuf_tmp; | 
 | 791 |     } | 
 | 792 |     else | 
 | 793 |         inbuf = (Py_UNICODE *)PyUnicode_AS_UNICODE(unistr); | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 794 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 795 |     inbuf_end = inbuf + datalen; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 796 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 797 |     r = multibytecodec_encode(ctx->codec, &ctx->state, | 
| Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 798 |                     (const Py_UNICODE **)&inbuf, datalen, | 
 | 799 |                     ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0); | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 800 |     if (r == NULL) { | 
 | 801 |         /* recover the original pending buffer */ | 
 | 802 |         if (origpending > 0) | 
 | 803 |             memcpy(ctx->pending, inbuf_tmp, | 
 | 804 |                 Py_UNICODE_SIZE * origpending); | 
 | 805 |         ctx->pendingsize = origpending; | 
 | 806 |         goto errorexit; | 
 | 807 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 808 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 809 |     if (inbuf < inbuf_end) { | 
 | 810 |         ctx->pendingsize = (Py_ssize_t)(inbuf_end - inbuf); | 
 | 811 |         if (ctx->pendingsize > MAXENCPENDING) { | 
 | 812 |             /* normal codecs can't reach here */ | 
 | 813 |             ctx->pendingsize = 0; | 
 | 814 |             PyErr_SetString(PyExc_UnicodeError, | 
 | 815 |                             "pending buffer overflow"); | 
 | 816 |             goto errorexit; | 
 | 817 |         } | 
 | 818 |         memcpy(ctx->pending, inbuf, | 
 | 819 |             ctx->pendingsize * Py_UNICODE_SIZE); | 
 | 820 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 821 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 822 |     if (inbuf_tmp != NULL) | 
 | 823 |         PyMem_Del(inbuf_tmp); | 
 | 824 |     Py_XDECREF(ucvt); | 
 | 825 |     return r; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 826 |  | 
 | 827 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 828 |     if (inbuf_tmp != NULL) | 
 | 829 |         PyMem_Del(inbuf_tmp); | 
 | 830 |     Py_XDECREF(r); | 
 | 831 |     Py_XDECREF(ucvt); | 
 | 832 |     return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 833 | } | 
 | 834 |  | 
 | 835 | static int | 
 | 836 | decoder_append_pending(MultibyteStatefulDecoderContext *ctx, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 837 |                        MultibyteDecodeBuffer *buf) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 838 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 839 |     Py_ssize_t npendings; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 840 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 841 |     npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); | 
 | 842 |     if (npendings + ctx->pendingsize > MAXDECPENDING || | 
 | 843 |         npendings > PY_SSIZE_T_MAX - ctx->pendingsize) { | 
 | 844 |             PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow"); | 
 | 845 |             return -1; | 
 | 846 |     } | 
 | 847 |     memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings); | 
 | 848 |     ctx->pendingsize += npendings; | 
 | 849 |     return 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 850 | } | 
 | 851 |  | 
 | 852 | static int | 
 | 853 | decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 854 |                        Py_ssize_t size) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 855 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 856 |     buf->inbuf = buf->inbuf_top = (const unsigned char *)data; | 
 | 857 |     buf->inbuf_end = buf->inbuf_top + size; | 
 | 858 |     if (buf->outobj == NULL) { /* only if outobj is not allocated yet */ | 
 | 859 |         buf->outobj = PyUnicode_FromUnicode(NULL, size); | 
 | 860 |         if (buf->outobj == NULL) | 
 | 861 |             return -1; | 
| Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 862 |         buf->outbuf = PyUnicode_AsUnicode(buf->outobj); | 
 | 863 |         if (buf->outbuf == NULL) | 
 | 864 |             return -1; | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 865 |         buf->outbuf_end = buf->outbuf + | 
 | 866 |                           PyUnicode_GET_SIZE(buf->outobj); | 
 | 867 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 868 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 869 |     return 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 870 | } | 
 | 871 |  | 
 | 872 | static int | 
 | 873 | decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 874 |                     MultibyteDecodeBuffer *buf) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 875 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 876 |     while (buf->inbuf < buf->inbuf_end) { | 
 | 877 |         Py_ssize_t inleft, outleft; | 
 | 878 |         Py_ssize_t r; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 879 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 880 |         inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); | 
 | 881 |         outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf); | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 882 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 883 |         r = ctx->codec->decode(&ctx->state, ctx->codec->config, | 
 | 884 |             &buf->inbuf, inleft, &buf->outbuf, outleft); | 
 | 885 |         if (r == 0 || r == MBERR_TOOFEW) | 
 | 886 |             break; | 
 | 887 |         else if (multibytecodec_decerror(ctx->codec, &ctx->state, | 
 | 888 |                                          buf, ctx->errors, r)) | 
 | 889 |             return -1; | 
 | 890 |     } | 
 | 891 |     return 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 892 | } | 
 | 893 |  | 
 | 894 |  | 
 | 895 | /** | 
 | 896 |  * MultibyteIncrementalEncoder object | 
 | 897 |  */ | 
 | 898 |  | 
 | 899 | static PyObject * | 
 | 900 | mbiencoder_encode(MultibyteIncrementalEncoderObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 901 |                   PyObject *args, PyObject *kwargs) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 902 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 903 |     PyObject *data; | 
 | 904 |     int final = 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 905 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 906 |     if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode", | 
 | 907 |                     incrementalkwarglist, &data, &final)) | 
 | 908 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 909 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 910 |     return encoder_encode_stateful(STATEFUL_ECTX(self), data, final); | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 911 | } | 
 | 912 |  | 
 | 913 | static PyObject * | 
 | 914 | mbiencoder_reset(MultibyteIncrementalEncoderObject *self) | 
 | 915 | { | 
| Victor Stinner | e15dce3 | 2011-05-30 22:56:00 +0200 | [diff] [blame] | 916 |     /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */ | 
 | 917 |     unsigned char buffer[4], *outbuf; | 
 | 918 |     Py_ssize_t r; | 
 | 919 |     if (self->codec->encreset != NULL) { | 
 | 920 |         outbuf = buffer; | 
 | 921 |         r = self->codec->encreset(&self->state, self->codec->config, | 
 | 922 |                                   &outbuf, sizeof(buffer)); | 
 | 923 |         if (r != 0) | 
 | 924 |             return NULL; | 
 | 925 |     } | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 926 |     self->pendingsize = 0; | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 927 |     Py_RETURN_NONE; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 928 | } | 
 | 929 |  | 
 | 930 | static struct PyMethodDef mbiencoder_methods[] = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 931 |     {"encode",          (PyCFunction)mbiencoder_encode, | 
 | 932 |                     METH_VARARGS | METH_KEYWORDS, NULL}, | 
 | 933 |     {"reset",           (PyCFunction)mbiencoder_reset, | 
 | 934 |                     METH_NOARGS, NULL}, | 
 | 935 |     {NULL,              NULL}, | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 936 | }; | 
 | 937 |  | 
 | 938 | static PyObject * | 
 | 939 | mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) | 
 | 940 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 941 |     MultibyteIncrementalEncoderObject *self; | 
 | 942 |     PyObject *codec = NULL; | 
 | 943 |     char *errors = NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 944 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 945 |     if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder", | 
 | 946 |                                      incnewkwarglist, &errors)) | 
 | 947 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 948 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 949 |     self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0); | 
 | 950 |     if (self == NULL) | 
 | 951 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 952 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 953 |     codec = PyObject_GetAttrString((PyObject *)type, "codec"); | 
 | 954 |     if (codec == NULL) | 
 | 955 |         goto errorexit; | 
 | 956 |     if (!MultibyteCodec_Check(codec)) { | 
 | 957 |         PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); | 
 | 958 |         goto errorexit; | 
 | 959 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 960 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 961 |     self->codec = ((MultibyteCodecObject *)codec)->codec; | 
 | 962 |     self->pendingsize = 0; | 
 | 963 |     self->errors = internal_error_callback(errors); | 
 | 964 |     if (self->errors == NULL) | 
 | 965 |         goto errorexit; | 
 | 966 |     if (self->codec->encinit != NULL && | 
 | 967 |         self->codec->encinit(&self->state, self->codec->config) != 0) | 
 | 968 |         goto errorexit; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 969 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 970 |     Py_DECREF(codec); | 
 | 971 |     return (PyObject *)self; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 972 |  | 
 | 973 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 974 |     Py_XDECREF(self); | 
 | 975 |     Py_XDECREF(codec); | 
 | 976 |     return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 977 | } | 
 | 978 |  | 
 | 979 | static int | 
| Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 980 | mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds) | 
 | 981 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 982 |     return 0; | 
| Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 983 | } | 
 | 984 |  | 
 | 985 | static int | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 986 | mbiencoder_traverse(MultibyteIncrementalEncoderObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 987 |                     visitproc visit, void *arg) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 988 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 989 |     if (ERROR_ISCUSTOM(self->errors)) | 
 | 990 |         Py_VISIT(self->errors); | 
 | 991 |     return 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 992 | } | 
 | 993 |  | 
 | 994 | static void | 
 | 995 | mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self) | 
 | 996 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 997 |     PyObject_GC_UnTrack(self); | 
 | 998 |     ERROR_DECREF(self->errors); | 
 | 999 |     Py_TYPE(self)->tp_free(self); | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1000 | } | 
 | 1001 |  | 
 | 1002 | static PyTypeObject MultibyteIncrementalEncoder_Type = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1003 |     PyVarObject_HEAD_INIT(NULL, 0) | 
 | 1004 |     "MultibyteIncrementalEncoder",      /* tp_name */ | 
 | 1005 |     sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */ | 
 | 1006 |     0,                                  /* tp_itemsize */ | 
 | 1007 |     /*  methods  */ | 
 | 1008 |     (destructor)mbiencoder_dealloc, /* tp_dealloc */ | 
 | 1009 |     0,                                  /* tp_print */ | 
 | 1010 |     0,                                  /* tp_getattr */ | 
 | 1011 |     0,                                  /* tp_setattr */ | 
 | 1012 |     0,                                  /* tp_reserved */ | 
 | 1013 |     0,                                  /* tp_repr */ | 
 | 1014 |     0,                                  /* tp_as_number */ | 
 | 1015 |     0,                                  /* tp_as_sequence */ | 
 | 1016 |     0,                                  /* tp_as_mapping */ | 
 | 1017 |     0,                                  /* tp_hash */ | 
 | 1018 |     0,                                  /* tp_call */ | 
 | 1019 |     0,                                  /* tp_str */ | 
 | 1020 |     PyObject_GenericGetAttr,            /* tp_getattro */ | 
 | 1021 |     0,                                  /* tp_setattro */ | 
 | 1022 |     0,                                  /* tp_as_buffer */ | 
 | 1023 |     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC | 
 | 1024 |         | Py_TPFLAGS_BASETYPE,          /* tp_flags */ | 
 | 1025 |     0,                                  /* tp_doc */ | 
 | 1026 |     (traverseproc)mbiencoder_traverse,          /* tp_traverse */ | 
 | 1027 |     0,                                  /* tp_clear */ | 
 | 1028 |     0,                                  /* tp_richcompare */ | 
 | 1029 |     0,                                  /* tp_weaklistoffset */ | 
 | 1030 |     0,                                  /* tp_iter */ | 
 | 1031 |     0,                                  /* tp_iterext */ | 
 | 1032 |     mbiencoder_methods,                 /* tp_methods */ | 
 | 1033 |     0,                                  /* tp_members */ | 
 | 1034 |     codecctx_getsets,                   /* tp_getset */ | 
 | 1035 |     0,                                  /* tp_base */ | 
 | 1036 |     0,                                  /* tp_dict */ | 
 | 1037 |     0,                                  /* tp_descr_get */ | 
 | 1038 |     0,                                  /* tp_descr_set */ | 
 | 1039 |     0,                                  /* tp_dictoffset */ | 
 | 1040 |     mbiencoder_init,                    /* tp_init */ | 
 | 1041 |     0,                                  /* tp_alloc */ | 
 | 1042 |     mbiencoder_new,                     /* tp_new */ | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1043 | }; | 
 | 1044 |  | 
 | 1045 |  | 
 | 1046 | /** | 
 | 1047 |  * MultibyteIncrementalDecoder object | 
 | 1048 |  */ | 
 | 1049 |  | 
 | 1050 | static PyObject * | 
 | 1051 | mbidecoder_decode(MultibyteIncrementalDecoderObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1052 |                   PyObject *args, PyObject *kwargs) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1053 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1054 |     MultibyteDecodeBuffer buf; | 
 | 1055 |     char *data, *wdata = NULL; | 
 | 1056 |     Py_buffer pdata; | 
 | 1057 |     Py_ssize_t wsize, finalsize = 0, size, origpending; | 
 | 1058 |     int final = 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1059 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1060 |     if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|i:decode", | 
 | 1061 |                     incrementalkwarglist, &pdata, &final)) | 
 | 1062 |         return NULL; | 
 | 1063 |     data = pdata.buf; | 
 | 1064 |     size = pdata.len; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1065 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1066 |     buf.outobj = buf.excobj = NULL; | 
 | 1067 |     origpending = self->pendingsize; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1068 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1069 |     if (self->pendingsize == 0) { | 
 | 1070 |         wsize = size; | 
 | 1071 |         wdata = data; | 
 | 1072 |     } | 
 | 1073 |     else { | 
 | 1074 |         if (size > PY_SSIZE_T_MAX - self->pendingsize) { | 
 | 1075 |             PyErr_NoMemory(); | 
 | 1076 |             goto errorexit; | 
 | 1077 |         } | 
 | 1078 |         wsize = size + self->pendingsize; | 
 | 1079 |         wdata = PyMem_Malloc(wsize); | 
 | 1080 |         if (wdata == NULL) | 
 | 1081 |             goto errorexit; | 
 | 1082 |         memcpy(wdata, self->pending, self->pendingsize); | 
 | 1083 |         memcpy(wdata + self->pendingsize, data, size); | 
 | 1084 |         self->pendingsize = 0; | 
 | 1085 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1086 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1087 |     if (decoder_prepare_buffer(&buf, wdata, wsize) != 0) | 
 | 1088 |         goto errorexit; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1089 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1090 |     if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf)) | 
 | 1091 |         goto errorexit; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1092 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1093 |     if (final && buf.inbuf < buf.inbuf_end) { | 
 | 1094 |         if (multibytecodec_decerror(self->codec, &self->state, | 
 | 1095 |                         &buf, self->errors, MBERR_TOOFEW)) { | 
 | 1096 |             /* recover the original pending buffer */ | 
 | 1097 |             memcpy(self->pending, wdata, origpending); | 
 | 1098 |             self->pendingsize = origpending; | 
 | 1099 |             goto errorexit; | 
 | 1100 |         } | 
 | 1101 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1102 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1103 |     if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */ | 
 | 1104 |         if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0) | 
 | 1105 |             goto errorexit; | 
 | 1106 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1107 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1108 |     finalsize = (Py_ssize_t)(buf.outbuf - PyUnicode_AS_UNICODE(buf.outobj)); | 
 | 1109 |     if (finalsize != PyUnicode_GET_SIZE(buf.outobj)) | 
 | 1110 |         if (PyUnicode_Resize(&buf.outobj, finalsize) == -1) | 
 | 1111 |             goto errorexit; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1112 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1113 |     PyBuffer_Release(&pdata); | 
 | 1114 |     if (wdata != data) | 
 | 1115 |         PyMem_Del(wdata); | 
 | 1116 |     Py_XDECREF(buf.excobj); | 
 | 1117 |     return buf.outobj; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1118 |  | 
 | 1119 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1120 |     PyBuffer_Release(&pdata); | 
 | 1121 |     if (wdata != NULL && wdata != data) | 
 | 1122 |         PyMem_Del(wdata); | 
 | 1123 |     Py_XDECREF(buf.excobj); | 
 | 1124 |     Py_XDECREF(buf.outobj); | 
 | 1125 |     return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1126 | } | 
 | 1127 |  | 
 | 1128 | static PyObject * | 
 | 1129 | mbidecoder_reset(MultibyteIncrementalDecoderObject *self) | 
 | 1130 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1131 |     if (self->codec->decreset != NULL && | 
 | 1132 |         self->codec->decreset(&self->state, self->codec->config) != 0) | 
 | 1133 |         return NULL; | 
 | 1134 |     self->pendingsize = 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1135 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1136 |     Py_RETURN_NONE; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1137 | } | 
 | 1138 |  | 
 | 1139 | static struct PyMethodDef mbidecoder_methods[] = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1140 |     {"decode",          (PyCFunction)mbidecoder_decode, | 
 | 1141 |                     METH_VARARGS | METH_KEYWORDS, NULL}, | 
 | 1142 |     {"reset",           (PyCFunction)mbidecoder_reset, | 
 | 1143 |                     METH_NOARGS, NULL}, | 
 | 1144 |     {NULL,              NULL}, | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1145 | }; | 
 | 1146 |  | 
 | 1147 | static PyObject * | 
 | 1148 | mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) | 
 | 1149 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1150 |     MultibyteIncrementalDecoderObject *self; | 
 | 1151 |     PyObject *codec = NULL; | 
 | 1152 |     char *errors = NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1153 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1154 |     if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder", | 
 | 1155 |                                      incnewkwarglist, &errors)) | 
 | 1156 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1157 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1158 |     self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0); | 
 | 1159 |     if (self == NULL) | 
 | 1160 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1161 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1162 |     codec = PyObject_GetAttrString((PyObject *)type, "codec"); | 
 | 1163 |     if (codec == NULL) | 
 | 1164 |         goto errorexit; | 
 | 1165 |     if (!MultibyteCodec_Check(codec)) { | 
 | 1166 |         PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); | 
 | 1167 |         goto errorexit; | 
 | 1168 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1169 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1170 |     self->codec = ((MultibyteCodecObject *)codec)->codec; | 
 | 1171 |     self->pendingsize = 0; | 
 | 1172 |     self->errors = internal_error_callback(errors); | 
 | 1173 |     if (self->errors == NULL) | 
 | 1174 |         goto errorexit; | 
 | 1175 |     if (self->codec->decinit != NULL && | 
 | 1176 |         self->codec->decinit(&self->state, self->codec->config) != 0) | 
 | 1177 |         goto errorexit; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1178 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1179 |     Py_DECREF(codec); | 
 | 1180 |     return (PyObject *)self; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1181 |  | 
 | 1182 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1183 |     Py_XDECREF(self); | 
 | 1184 |     Py_XDECREF(codec); | 
 | 1185 |     return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1186 | } | 
 | 1187 |  | 
 | 1188 | static int | 
| Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1189 | mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds) | 
 | 1190 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1191 |     return 0; | 
| Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1192 | } | 
 | 1193 |  | 
 | 1194 | static int | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1195 | mbidecoder_traverse(MultibyteIncrementalDecoderObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1196 |                     visitproc visit, void *arg) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1197 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1198 |     if (ERROR_ISCUSTOM(self->errors)) | 
 | 1199 |         Py_VISIT(self->errors); | 
 | 1200 |     return 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1201 | } | 
 | 1202 |  | 
 | 1203 | static void | 
 | 1204 | mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self) | 
 | 1205 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1206 |     PyObject_GC_UnTrack(self); | 
 | 1207 |     ERROR_DECREF(self->errors); | 
 | 1208 |     Py_TYPE(self)->tp_free(self); | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1209 | } | 
 | 1210 |  | 
 | 1211 | static PyTypeObject MultibyteIncrementalDecoder_Type = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1212 |     PyVarObject_HEAD_INIT(NULL, 0) | 
 | 1213 |     "MultibyteIncrementalDecoder",      /* tp_name */ | 
 | 1214 |     sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */ | 
 | 1215 |     0,                                  /* tp_itemsize */ | 
 | 1216 |     /*  methods  */ | 
 | 1217 |     (destructor)mbidecoder_dealloc, /* tp_dealloc */ | 
 | 1218 |     0,                                  /* tp_print */ | 
 | 1219 |     0,                                  /* tp_getattr */ | 
 | 1220 |     0,                                  /* tp_setattr */ | 
 | 1221 |     0,                                  /* tp_reserved */ | 
 | 1222 |     0,                                  /* tp_repr */ | 
 | 1223 |     0,                                  /* tp_as_number */ | 
 | 1224 |     0,                                  /* tp_as_sequence */ | 
 | 1225 |     0,                                  /* tp_as_mapping */ | 
 | 1226 |     0,                                  /* tp_hash */ | 
 | 1227 |     0,                                  /* tp_call */ | 
 | 1228 |     0,                                  /* tp_str */ | 
 | 1229 |     PyObject_GenericGetAttr,            /* tp_getattro */ | 
 | 1230 |     0,                                  /* tp_setattro */ | 
 | 1231 |     0,                                  /* tp_as_buffer */ | 
 | 1232 |     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC | 
 | 1233 |         | Py_TPFLAGS_BASETYPE,          /* tp_flags */ | 
 | 1234 |     0,                                  /* tp_doc */ | 
 | 1235 |     (traverseproc)mbidecoder_traverse,          /* tp_traverse */ | 
 | 1236 |     0,                                  /* tp_clear */ | 
 | 1237 |     0,                                  /* tp_richcompare */ | 
 | 1238 |     0,                                  /* tp_weaklistoffset */ | 
 | 1239 |     0,                                  /* tp_iter */ | 
 | 1240 |     0,                                  /* tp_iterext */ | 
 | 1241 |     mbidecoder_methods,                 /* tp_methods */ | 
 | 1242 |     0,                                  /* tp_members */ | 
 | 1243 |     codecctx_getsets,                   /* tp_getset */ | 
 | 1244 |     0,                                  /* tp_base */ | 
 | 1245 |     0,                                  /* tp_dict */ | 
 | 1246 |     0,                                  /* tp_descr_get */ | 
 | 1247 |     0,                                  /* tp_descr_set */ | 
 | 1248 |     0,                                  /* tp_dictoffset */ | 
 | 1249 |     mbidecoder_init,                    /* tp_init */ | 
 | 1250 |     0,                                  /* tp_alloc */ | 
 | 1251 |     mbidecoder_new,                     /* tp_new */ | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1252 | }; | 
 | 1253 |  | 
 | 1254 |  | 
 | 1255 | /** | 
 | 1256 |  * MultibyteStreamReader object | 
 | 1257 |  */ | 
 | 1258 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1259 | static PyObject * | 
 | 1260 | mbstreamreader_iread(MultibyteStreamReaderObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1261 |                      const char *method, Py_ssize_t sizehint) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1262 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1263 |     MultibyteDecodeBuffer buf; | 
 | 1264 |     PyObject *cres; | 
 | 1265 |     Py_ssize_t rsize, finalsize = 0; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1266 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1267 |     if (sizehint == 0) | 
| Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 1268 |         return PyUnicode_New(0, 0); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1269 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1270 |     buf.outobj = buf.excobj = NULL; | 
 | 1271 |     cres = NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1272 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1273 |     for (;;) { | 
 | 1274 |         int endoffile; | 
| Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1275 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1276 |         if (sizehint < 0) | 
 | 1277 |             cres = PyObject_CallMethod(self->stream, | 
 | 1278 |                             (char *)method, NULL); | 
 | 1279 |         else | 
 | 1280 |             cres = PyObject_CallMethod(self->stream, | 
 | 1281 |                             (char *)method, "i", sizehint); | 
 | 1282 |         if (cres == NULL) | 
 | 1283 |             goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1284 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1285 |         if (!PyBytes_Check(cres)) { | 
 | 1286 |             PyErr_Format(PyExc_TypeError, | 
 | 1287 |                          "stream function returned a " | 
 | 1288 |                          "non-bytes object (%.100s)", | 
 | 1289 |                          cres->ob_type->tp_name); | 
 | 1290 |             goto errorexit; | 
 | 1291 |         } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1292 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1293 |         endoffile = (PyBytes_GET_SIZE(cres) == 0); | 
| Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1294 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1295 |         if (self->pendingsize > 0) { | 
 | 1296 |             PyObject *ctr; | 
 | 1297 |             char *ctrdata; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1298 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1299 |             if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) { | 
 | 1300 |                 PyErr_NoMemory(); | 
 | 1301 |                 goto errorexit; | 
 | 1302 |         } | 
 | 1303 |                     rsize = PyBytes_GET_SIZE(cres) + self->pendingsize; | 
 | 1304 |                     ctr = PyBytes_FromStringAndSize(NULL, rsize); | 
 | 1305 |                     if (ctr == NULL) | 
 | 1306 |                             goto errorexit; | 
 | 1307 |                     ctrdata = PyBytes_AS_STRING(ctr); | 
 | 1308 |                     memcpy(ctrdata, self->pending, self->pendingsize); | 
 | 1309 |                     memcpy(ctrdata + self->pendingsize, | 
 | 1310 |                             PyBytes_AS_STRING(cres), | 
 | 1311 |                             PyBytes_GET_SIZE(cres)); | 
 | 1312 |                     Py_DECREF(cres); | 
 | 1313 |                     cres = ctr; | 
 | 1314 |                     self->pendingsize = 0; | 
 | 1315 |         } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1316 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1317 |         rsize = PyBytes_GET_SIZE(cres); | 
 | 1318 |         if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres), | 
 | 1319 |                                    rsize) != 0) | 
 | 1320 |             goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1321 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1322 |         if (rsize > 0 && decoder_feed_buffer( | 
 | 1323 |                         (MultibyteStatefulDecoderContext *)self, &buf)) | 
 | 1324 |             goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1325 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1326 |         if (endoffile || sizehint < 0) { | 
 | 1327 |             if (buf.inbuf < buf.inbuf_end && | 
 | 1328 |                 multibytecodec_decerror(self->codec, &self->state, | 
 | 1329 |                             &buf, self->errors, MBERR_TOOFEW)) | 
 | 1330 |                 goto errorexit; | 
 | 1331 |         } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1332 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1333 |         if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */ | 
 | 1334 |             if (decoder_append_pending(STATEFUL_DCTX(self), | 
 | 1335 |                                        &buf) != 0) | 
 | 1336 |                 goto errorexit; | 
 | 1337 |         } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1338 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1339 |         finalsize = (Py_ssize_t)(buf.outbuf - | 
 | 1340 |                         PyUnicode_AS_UNICODE(buf.outobj)); | 
 | 1341 |         Py_DECREF(cres); | 
 | 1342 |         cres = NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1343 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1344 |         if (sizehint < 0 || finalsize != 0 || rsize == 0) | 
 | 1345 |             break; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1346 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1347 |         sizehint = 1; /* read 1 more byte and retry */ | 
 | 1348 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1349 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1350 |     if (finalsize != PyUnicode_GET_SIZE(buf.outobj)) | 
 | 1351 |         if (PyUnicode_Resize(&buf.outobj, finalsize) == -1) | 
 | 1352 |             goto errorexit; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1353 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1354 |     Py_XDECREF(cres); | 
 | 1355 |     Py_XDECREF(buf.excobj); | 
 | 1356 |     return buf.outobj; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1357 |  | 
 | 1358 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1359 |     Py_XDECREF(cres); | 
 | 1360 |     Py_XDECREF(buf.excobj); | 
 | 1361 |     Py_XDECREF(buf.outobj); | 
 | 1362 |     return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1363 | } | 
 | 1364 |  | 
 | 1365 | static PyObject * | 
 | 1366 | mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args) | 
 | 1367 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1368 |     PyObject *sizeobj = NULL; | 
 | 1369 |     Py_ssize_t size; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1370 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1371 |     if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj)) | 
 | 1372 |         return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1373 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1374 |     if (sizeobj == Py_None || sizeobj == NULL) | 
 | 1375 |         size = -1; | 
 | 1376 |     else if (PyLong_Check(sizeobj)) | 
 | 1377 |         size = PyLong_AsSsize_t(sizeobj); | 
 | 1378 |     else { | 
 | 1379 |         PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); | 
 | 1380 |         return NULL; | 
 | 1381 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1382 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1383 |     if (size == -1 && PyErr_Occurred()) | 
 | 1384 |         return NULL; | 
| Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1385 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1386 |     return mbstreamreader_iread(self, "read", size); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1387 | } | 
 | 1388 |  | 
 | 1389 | static PyObject * | 
 | 1390 | mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args) | 
 | 1391 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1392 |     PyObject *sizeobj = NULL; | 
 | 1393 |     Py_ssize_t size; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1394 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1395 |     if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj)) | 
 | 1396 |         return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1397 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1398 |     if (sizeobj == Py_None || sizeobj == NULL) | 
 | 1399 |         size = -1; | 
 | 1400 |     else if (PyLong_Check(sizeobj)) | 
 | 1401 |         size = PyLong_AsSsize_t(sizeobj); | 
 | 1402 |     else { | 
 | 1403 |         PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); | 
 | 1404 |         return NULL; | 
 | 1405 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1406 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1407 |     if (size == -1 && PyErr_Occurred()) | 
 | 1408 |         return NULL; | 
| Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1409 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1410 |     return mbstreamreader_iread(self, "readline", size); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1411 | } | 
 | 1412 |  | 
 | 1413 | static PyObject * | 
 | 1414 | mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args) | 
 | 1415 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1416 |     PyObject *sizehintobj = NULL, *r, *sr; | 
 | 1417 |     Py_ssize_t sizehint; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1418 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1419 |     if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj)) | 
 | 1420 |         return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1421 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1422 |     if (sizehintobj == Py_None || sizehintobj == NULL) | 
 | 1423 |         sizehint = -1; | 
 | 1424 |     else if (PyLong_Check(sizehintobj)) | 
 | 1425 |         sizehint = PyLong_AsSsize_t(sizehintobj); | 
 | 1426 |     else { | 
 | 1427 |         PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); | 
 | 1428 |         return NULL; | 
 | 1429 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1430 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1431 |     if (sizehint == -1 && PyErr_Occurred()) | 
 | 1432 |         return NULL; | 
| Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1433 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1434 |     r = mbstreamreader_iread(self, "read", sizehint); | 
 | 1435 |     if (r == NULL) | 
 | 1436 |         return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1437 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1438 |     sr = PyUnicode_Splitlines(r, 1); | 
 | 1439 |     Py_DECREF(r); | 
 | 1440 |     return sr; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1441 | } | 
 | 1442 |  | 
 | 1443 | static PyObject * | 
 | 1444 | mbstreamreader_reset(MultibyteStreamReaderObject *self) | 
 | 1445 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1446 |     if (self->codec->decreset != NULL && | 
 | 1447 |         self->codec->decreset(&self->state, self->codec->config) != 0) | 
 | 1448 |         return NULL; | 
 | 1449 |     self->pendingsize = 0; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1450 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1451 |     Py_RETURN_NONE; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1452 | } | 
 | 1453 |  | 
 | 1454 | static struct PyMethodDef mbstreamreader_methods[] = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1455 |     {"read",            (PyCFunction)mbstreamreader_read, | 
 | 1456 |                     METH_VARARGS, NULL}, | 
 | 1457 |     {"readline",        (PyCFunction)mbstreamreader_readline, | 
 | 1458 |                     METH_VARARGS, NULL}, | 
 | 1459 |     {"readlines",       (PyCFunction)mbstreamreader_readlines, | 
 | 1460 |                     METH_VARARGS, NULL}, | 
 | 1461 |     {"reset",           (PyCFunction)mbstreamreader_reset, | 
 | 1462 |                     METH_NOARGS, NULL}, | 
 | 1463 |     {NULL,              NULL}, | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1464 | }; | 
 | 1465 |  | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1466 | static PyMemberDef mbstreamreader_members[] = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1467 |     {"stream",          T_OBJECT, | 
 | 1468 |                     offsetof(MultibyteStreamReaderObject, stream), | 
 | 1469 |                     READONLY, NULL}, | 
 | 1470 |     {NULL,} | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1471 | }; | 
 | 1472 |  | 
 | 1473 | static PyObject * | 
 | 1474 | mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds) | 
 | 1475 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1476 |     MultibyteStreamReaderObject *self; | 
 | 1477 |     PyObject *stream, *codec = NULL; | 
 | 1478 |     char *errors = NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1479 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1480 |     if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader", | 
 | 1481 |                             streamkwarglist, &stream, &errors)) | 
 | 1482 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1483 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1484 |     self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0); | 
 | 1485 |     if (self == NULL) | 
 | 1486 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1487 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1488 |     codec = PyObject_GetAttrString((PyObject *)type, "codec"); | 
 | 1489 |     if (codec == NULL) | 
 | 1490 |         goto errorexit; | 
 | 1491 |     if (!MultibyteCodec_Check(codec)) { | 
 | 1492 |         PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); | 
 | 1493 |         goto errorexit; | 
 | 1494 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1495 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1496 |     self->codec = ((MultibyteCodecObject *)codec)->codec; | 
 | 1497 |     self->stream = stream; | 
 | 1498 |     Py_INCREF(stream); | 
 | 1499 |     self->pendingsize = 0; | 
 | 1500 |     self->errors = internal_error_callback(errors); | 
 | 1501 |     if (self->errors == NULL) | 
 | 1502 |         goto errorexit; | 
 | 1503 |     if (self->codec->decinit != NULL && | 
 | 1504 |         self->codec->decinit(&self->state, self->codec->config) != 0) | 
 | 1505 |         goto errorexit; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1506 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1507 |     Py_DECREF(codec); | 
 | 1508 |     return (PyObject *)self; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1509 |  | 
 | 1510 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1511 |     Py_XDECREF(self); | 
 | 1512 |     Py_XDECREF(codec); | 
 | 1513 |     return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1514 | } | 
 | 1515 |  | 
 | 1516 | static int | 
| Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1517 | mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds) | 
 | 1518 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1519 |     return 0; | 
| Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1520 | } | 
 | 1521 |  | 
 | 1522 | static int | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1523 | mbstreamreader_traverse(MultibyteStreamReaderObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1524 |                         visitproc visit, void *arg) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1525 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1526 |     if (ERROR_ISCUSTOM(self->errors)) | 
 | 1527 |         Py_VISIT(self->errors); | 
 | 1528 |     Py_VISIT(self->stream); | 
 | 1529 |     return 0; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1530 | } | 
 | 1531 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1532 | static void | 
 | 1533 | mbstreamreader_dealloc(MultibyteStreamReaderObject *self) | 
 | 1534 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1535 |     PyObject_GC_UnTrack(self); | 
 | 1536 |     ERROR_DECREF(self->errors); | 
 | 1537 |     Py_XDECREF(self->stream); | 
 | 1538 |     Py_TYPE(self)->tp_free(self); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1539 | } | 
 | 1540 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1541 | static PyTypeObject MultibyteStreamReader_Type = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1542 |     PyVarObject_HEAD_INIT(NULL, 0) | 
 | 1543 |     "MultibyteStreamReader",            /* tp_name */ | 
 | 1544 |     sizeof(MultibyteStreamReaderObject), /* tp_basicsize */ | 
 | 1545 |     0,                                  /* tp_itemsize */ | 
 | 1546 |     /*  methods  */ | 
 | 1547 |     (destructor)mbstreamreader_dealloc, /* tp_dealloc */ | 
 | 1548 |     0,                                  /* tp_print */ | 
 | 1549 |     0,                                  /* tp_getattr */ | 
 | 1550 |     0,                                  /* tp_setattr */ | 
 | 1551 |     0,                                  /* tp_reserved */ | 
 | 1552 |     0,                                  /* tp_repr */ | 
 | 1553 |     0,                                  /* tp_as_number */ | 
 | 1554 |     0,                                  /* tp_as_sequence */ | 
 | 1555 |     0,                                  /* tp_as_mapping */ | 
 | 1556 |     0,                                  /* tp_hash */ | 
 | 1557 |     0,                                  /* tp_call */ | 
 | 1558 |     0,                                  /* tp_str */ | 
 | 1559 |     PyObject_GenericGetAttr,            /* tp_getattro */ | 
 | 1560 |     0,                                  /* tp_setattro */ | 
 | 1561 |     0,                                  /* tp_as_buffer */ | 
 | 1562 |     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC | 
 | 1563 |         | Py_TPFLAGS_BASETYPE,          /* tp_flags */ | 
 | 1564 |     0,                                  /* tp_doc */ | 
 | 1565 |     (traverseproc)mbstreamreader_traverse,      /* tp_traverse */ | 
 | 1566 |     0,                                  /* tp_clear */ | 
 | 1567 |     0,                                  /* tp_richcompare */ | 
 | 1568 |     0,                                  /* tp_weaklistoffset */ | 
 | 1569 |     0,                                  /* tp_iter */ | 
 | 1570 |     0,                                  /* tp_iterext */ | 
 | 1571 |     mbstreamreader_methods,             /* tp_methods */ | 
 | 1572 |     mbstreamreader_members,             /* tp_members */ | 
 | 1573 |     codecctx_getsets,                   /* tp_getset */ | 
 | 1574 |     0,                                  /* tp_base */ | 
 | 1575 |     0,                                  /* tp_dict */ | 
 | 1576 |     0,                                  /* tp_descr_get */ | 
 | 1577 |     0,                                  /* tp_descr_set */ | 
 | 1578 |     0,                                  /* tp_dictoffset */ | 
 | 1579 |     mbstreamreader_init,                /* tp_init */ | 
 | 1580 |     0,                                  /* tp_alloc */ | 
 | 1581 |     mbstreamreader_new,                 /* tp_new */ | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1582 | }; | 
 | 1583 |  | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1584 |  | 
 | 1585 | /** | 
 | 1586 |  * MultibyteStreamWriter object | 
 | 1587 |  */ | 
 | 1588 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1589 | static int | 
 | 1590 | mbstreamwriter_iwrite(MultibyteStreamWriterObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1591 |                       PyObject *unistr) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1592 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1593 |     PyObject *str, *wr; | 
| Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1594 |     _Py_IDENTIFIER(write); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1595 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1596 |     str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0); | 
 | 1597 |     if (str == NULL) | 
 | 1598 |         return -1; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1599 |  | 
| Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1600 |     wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str); | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1601 |     Py_DECREF(str); | 
 | 1602 |     if (wr == NULL) | 
 | 1603 |         return -1; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1604 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1605 |     Py_DECREF(wr); | 
 | 1606 |     return 0; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1607 | } | 
 | 1608 |  | 
 | 1609 | static PyObject * | 
| Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1610 | mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1611 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1612 |     if (mbstreamwriter_iwrite(self, strobj)) | 
 | 1613 |         return NULL; | 
 | 1614 |     else | 
 | 1615 |         Py_RETURN_NONE; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1616 | } | 
 | 1617 |  | 
 | 1618 | static PyObject * | 
| Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1619 | mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1620 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1621 |     PyObject *strobj; | 
 | 1622 |     int i, r; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1623 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1624 |     if (!PySequence_Check(lines)) { | 
 | 1625 |         PyErr_SetString(PyExc_TypeError, | 
 | 1626 |                         "arg must be a sequence object"); | 
 | 1627 |         return NULL; | 
 | 1628 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1629 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1630 |     for (i = 0; i < PySequence_Length(lines); i++) { | 
 | 1631 |         /* length can be changed even within this loop */ | 
 | 1632 |         strobj = PySequence_GetItem(lines, i); | 
 | 1633 |         if (strobj == NULL) | 
 | 1634 |             return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1635 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1636 |         r = mbstreamwriter_iwrite(self, strobj); | 
 | 1637 |         Py_DECREF(strobj); | 
 | 1638 |         if (r == -1) | 
 | 1639 |             return NULL; | 
 | 1640 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1641 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1642 |     Py_RETURN_NONE; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1643 | } | 
 | 1644 |  | 
 | 1645 | static PyObject * | 
 | 1646 | mbstreamwriter_reset(MultibyteStreamWriterObject *self) | 
 | 1647 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1648 |     const Py_UNICODE *pending; | 
 | 1649 |     PyObject *pwrt; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1650 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1651 |     pending = self->pending; | 
 | 1652 |     pwrt = multibytecodec_encode(self->codec, &self->state, | 
 | 1653 |                     &pending, self->pendingsize, self->errors, | 
 | 1654 |                     MBENC_FLUSH | MBENC_RESET); | 
 | 1655 |     /* some pending buffer can be truncated when UnicodeEncodeError is | 
 | 1656 |      * raised on 'strict' mode. but, 'reset' method is designed to | 
 | 1657 |      * reset the pending buffer or states so failed string sequence | 
 | 1658 |      * ought to be missed */ | 
 | 1659 |     self->pendingsize = 0; | 
 | 1660 |     if (pwrt == NULL) | 
 | 1661 |         return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1662 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1663 |     assert(PyBytes_Check(pwrt)); | 
 | 1664 |     if (PyBytes_Size(pwrt) > 0) { | 
 | 1665 |         PyObject *wr; | 
| Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1666 |         _Py_IDENTIFIER(write); | 
| Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1667 |  | 
 | 1668 |         wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt); | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1669 |         if (wr == NULL) { | 
 | 1670 |             Py_DECREF(pwrt); | 
 | 1671 |             return NULL; | 
 | 1672 |         } | 
 | 1673 |     } | 
 | 1674 |     Py_DECREF(pwrt); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1675 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1676 |     Py_RETURN_NONE; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1677 | } | 
 | 1678 |  | 
 | 1679 | static PyObject * | 
 | 1680 | mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds) | 
 | 1681 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1682 |     MultibyteStreamWriterObject *self; | 
 | 1683 |     PyObject *stream, *codec = NULL; | 
 | 1684 |     char *errors = NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1685 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1686 |     if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter", | 
 | 1687 |                             streamkwarglist, &stream, &errors)) | 
 | 1688 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1689 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1690 |     self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0); | 
 | 1691 |     if (self == NULL) | 
 | 1692 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1693 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1694 |     codec = PyObject_GetAttrString((PyObject *)type, "codec"); | 
 | 1695 |     if (codec == NULL) | 
 | 1696 |         goto errorexit; | 
 | 1697 |     if (!MultibyteCodec_Check(codec)) { | 
 | 1698 |         PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); | 
 | 1699 |         goto errorexit; | 
 | 1700 |     } | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1701 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1702 |     self->codec = ((MultibyteCodecObject *)codec)->codec; | 
 | 1703 |     self->stream = stream; | 
 | 1704 |     Py_INCREF(stream); | 
 | 1705 |     self->pendingsize = 0; | 
 | 1706 |     self->errors = internal_error_callback(errors); | 
 | 1707 |     if (self->errors == NULL) | 
 | 1708 |         goto errorexit; | 
 | 1709 |     if (self->codec->encinit != NULL && | 
 | 1710 |         self->codec->encinit(&self->state, self->codec->config) != 0) | 
 | 1711 |         goto errorexit; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1712 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1713 |     Py_DECREF(codec); | 
 | 1714 |     return (PyObject *)self; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1715 |  | 
 | 1716 | errorexit: | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1717 |     Py_XDECREF(self); | 
 | 1718 |     Py_XDECREF(codec); | 
 | 1719 |     return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1720 | } | 
 | 1721 |  | 
 | 1722 | static int | 
| Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1723 | mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds) | 
 | 1724 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1725 |     return 0; | 
| Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1726 | } | 
 | 1727 |  | 
 | 1728 | static int | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1729 | mbstreamwriter_traverse(MultibyteStreamWriterObject *self, | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1730 |                         visitproc visit, void *arg) | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1731 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1732 |     if (ERROR_ISCUSTOM(self->errors)) | 
 | 1733 |         Py_VISIT(self->errors); | 
 | 1734 |     Py_VISIT(self->stream); | 
 | 1735 |     return 0; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1736 | } | 
 | 1737 |  | 
 | 1738 | static void | 
 | 1739 | mbstreamwriter_dealloc(MultibyteStreamWriterObject *self) | 
 | 1740 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1741 |     PyObject_GC_UnTrack(self); | 
 | 1742 |     ERROR_DECREF(self->errors); | 
 | 1743 |     Py_XDECREF(self->stream); | 
 | 1744 |     Py_TYPE(self)->tp_free(self); | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1745 | } | 
 | 1746 |  | 
 | 1747 | static struct PyMethodDef mbstreamwriter_methods[] = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1748 |     {"write",           (PyCFunction)mbstreamwriter_write, | 
 | 1749 |                     METH_O, NULL}, | 
 | 1750 |     {"writelines",      (PyCFunction)mbstreamwriter_writelines, | 
 | 1751 |                     METH_O, NULL}, | 
 | 1752 |     {"reset",           (PyCFunction)mbstreamwriter_reset, | 
 | 1753 |                     METH_NOARGS, NULL}, | 
 | 1754 |     {NULL,              NULL}, | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1755 | }; | 
 | 1756 |  | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1757 | static PyMemberDef mbstreamwriter_members[] = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1758 |     {"stream",          T_OBJECT, | 
 | 1759 |                     offsetof(MultibyteStreamWriterObject, stream), | 
 | 1760 |                     READONLY, NULL}, | 
 | 1761 |     {NULL,} | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1762 | }; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1763 |  | 
 | 1764 | static PyTypeObject MultibyteStreamWriter_Type = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1765 |     PyVarObject_HEAD_INIT(NULL, 0) | 
 | 1766 |     "MultibyteStreamWriter",            /* tp_name */ | 
 | 1767 |     sizeof(MultibyteStreamWriterObject), /* tp_basicsize */ | 
 | 1768 |     0,                                  /* tp_itemsize */ | 
 | 1769 |     /*  methods  */ | 
 | 1770 |     (destructor)mbstreamwriter_dealloc, /* tp_dealloc */ | 
 | 1771 |     0,                                  /* tp_print */ | 
 | 1772 |     0,                                  /* tp_getattr */ | 
 | 1773 |     0,                                  /* tp_setattr */ | 
 | 1774 |     0,                                  /* tp_reserved */ | 
 | 1775 |     0,                                  /* tp_repr */ | 
 | 1776 |     0,                                  /* tp_as_number */ | 
 | 1777 |     0,                                  /* tp_as_sequence */ | 
 | 1778 |     0,                                  /* tp_as_mapping */ | 
 | 1779 |     0,                                  /* tp_hash */ | 
 | 1780 |     0,                                  /* tp_call */ | 
 | 1781 |     0,                                  /* tp_str */ | 
 | 1782 |     PyObject_GenericGetAttr,            /* tp_getattro */ | 
 | 1783 |     0,                                  /* tp_setattro */ | 
 | 1784 |     0,                                  /* tp_as_buffer */ | 
 | 1785 |     Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC | 
 | 1786 |         | Py_TPFLAGS_BASETYPE,          /* tp_flags */ | 
 | 1787 |     0,                                  /* tp_doc */ | 
 | 1788 |     (traverseproc)mbstreamwriter_traverse,      /* tp_traverse */ | 
 | 1789 |     0,                                  /* tp_clear */ | 
 | 1790 |     0,                                  /* tp_richcompare */ | 
 | 1791 |     0,                                  /* tp_weaklistoffset */ | 
 | 1792 |     0,                                  /* tp_iter */ | 
 | 1793 |     0,                                  /* tp_iterext */ | 
 | 1794 |     mbstreamwriter_methods,             /* tp_methods */ | 
 | 1795 |     mbstreamwriter_members,             /* tp_members */ | 
 | 1796 |     codecctx_getsets,                   /* tp_getset */ | 
 | 1797 |     0,                                  /* tp_base */ | 
 | 1798 |     0,                                  /* tp_dict */ | 
 | 1799 |     0,                                  /* tp_descr_get */ | 
 | 1800 |     0,                                  /* tp_descr_set */ | 
 | 1801 |     0,                                  /* tp_dictoffset */ | 
 | 1802 |     mbstreamwriter_init,                /* tp_init */ | 
 | 1803 |     0,                                  /* tp_alloc */ | 
 | 1804 |     mbstreamwriter_new,                 /* tp_new */ | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1805 | }; | 
 | 1806 |  | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1807 |  | 
 | 1808 | /** | 
 | 1809 |  * Exposed factory function | 
 | 1810 |  */ | 
 | 1811 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1812 | static PyObject * | 
 | 1813 | __create_codec(PyObject *ignore, PyObject *arg) | 
 | 1814 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1815 |     MultibyteCodecObject *self; | 
 | 1816 |     MultibyteCodec *codec; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1817 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1818 |     if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) { | 
 | 1819 |         PyErr_SetString(PyExc_ValueError, "argument type invalid"); | 
 | 1820 |         return NULL; | 
 | 1821 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1822 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1823 |     codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME); | 
 | 1824 |     if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0) | 
 | 1825 |         return NULL; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1826 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1827 |     self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type); | 
 | 1828 |     if (self == NULL) | 
 | 1829 |         return NULL; | 
 | 1830 |     self->codec = codec; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1831 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1832 |     return (PyObject *)self; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1833 | } | 
 | 1834 |  | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1835 | static struct PyMethodDef __methods[] = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1836 |     {"__create_codec", (PyCFunction)__create_codec, METH_O}, | 
 | 1837 |     {NULL, NULL}, | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1838 | }; | 
 | 1839 |  | 
| Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1840 |  | 
 | 1841 | static struct PyModuleDef _multibytecodecmodule = { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1842 |     PyModuleDef_HEAD_INIT, | 
 | 1843 |     "_multibytecodec", | 
 | 1844 |     NULL, | 
 | 1845 |     -1, | 
 | 1846 |     __methods, | 
 | 1847 |     NULL, | 
 | 1848 |     NULL, | 
 | 1849 |     NULL, | 
 | 1850 |     NULL | 
| Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1851 | }; | 
 | 1852 |  | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1853 | PyMODINIT_FUNC | 
| Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1854 | PyInit__multibytecodec(void) | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1855 | { | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1856 |     int i; | 
 | 1857 |     PyObject *m; | 
 | 1858 |     PyTypeObject *typelist[] = { | 
 | 1859 |         &MultibyteIncrementalEncoder_Type, | 
 | 1860 |         &MultibyteIncrementalDecoder_Type, | 
 | 1861 |         &MultibyteStreamReader_Type, | 
 | 1862 |         &MultibyteStreamWriter_Type, | 
 | 1863 |         NULL | 
 | 1864 |     }; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1865 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1866 |     if (PyType_Ready(&MultibyteCodec_Type) < 0) | 
 | 1867 |         return NULL; | 
| Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1868 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1869 |     m = PyModule_Create(&_multibytecodecmodule); | 
 | 1870 |     if (m == NULL) | 
 | 1871 |         return NULL; | 
| Neal Norwitz | 058bde1 | 2005-09-21 06:44:25 +0000 | [diff] [blame] | 1872 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1873 |     for (i = 0; typelist[i] != NULL; i++) { | 
 | 1874 |         if (PyType_Ready(typelist[i]) < 0) | 
 | 1875 |             return NULL; | 
 | 1876 |         Py_INCREF(typelist[i]); | 
 | 1877 |         PyModule_AddObject(m, typelist[i]->tp_name, | 
 | 1878 |                            (PyObject *)typelist[i]); | 
 | 1879 |     } | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1880 |  | 
| Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1881 |     if (PyErr_Occurred()) { | 
 | 1882 |         Py_FatalError("can't initialize the _multibytecodec module"); | 
 | 1883 |         Py_DECREF(m); | 
 | 1884 |         m = NULL; | 
 | 1885 |     } | 
 | 1886 |     return m; | 
| Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1887 | } |