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 { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 13 | PyObject *inobj; |
| 14 | Py_ssize_t inpos, inlen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 15 | unsigned char *outbuf, *outbuf_end; |
| 16 | PyObject *excobj, *outobj; |
| 17 | } MultibyteEncodeBuffer; |
| 18 | |
| 19 | typedef struct { |
| 20 | const unsigned char *inbuf, *inbuf_top, *inbuf_end; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 21 | PyObject *excobj; |
| 22 | _PyUnicodeWriter writer; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 23 | } MultibyteDecodeBuffer; |
| 24 | |
| 25 | PyDoc_STRVAR(MultibyteCodec_Encode__doc__, |
| 26 | "I.encode(unicode[, errors]) -> (string, length consumed)\n\ |
| 27 | \n\ |
| 28 | Return an encoded string version of `unicode'. errors may be given to\n\ |
| 29 | set a different error handling scheme. Default is 'strict' meaning that\n\ |
| 30 | encoding errors raise a UnicodeEncodeError. Other possible values are\n\ |
| 31 | 'ignore', 'replace' and 'xmlcharrefreplace' as well as any other name\n\ |
| 32 | registered with codecs.register_error that can handle UnicodeEncodeErrors."); |
| 33 | |
| 34 | PyDoc_STRVAR(MultibyteCodec_Decode__doc__, |
| 35 | "I.decode(string[, errors]) -> (unicodeobject, length consumed)\n\ |
| 36 | \n\ |
| 37 | Decodes `string' using I, an MultibyteCodec instance. errors may be given\n\ |
| 38 | to set a different error handling scheme. Default is 'strict' meaning\n\ |
| 39 | that encoding errors raise a UnicodeDecodeError. Other possible values\n\ |
Benjamin Peterson | f10a79a | 2008-10-11 00:49:57 +0000 | [diff] [blame] | 40 | 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] | 41 | codecs.register_error that is able to handle UnicodeDecodeErrors."); |
| 42 | |
Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 43 | static char *codeckwarglist[] = {"input", "errors", NULL}; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 44 | static char *incnewkwarglist[] = {"errors", NULL}; |
| 45 | static char *incrementalkwarglist[] = {"input", "final", NULL}; |
Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 46 | static char *streamkwarglist[] = {"stream", "errors", NULL}; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 47 | |
| 48 | static PyObject *multibytecodec_encode(MultibyteCodec *, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 49 | MultibyteCodec_State *, PyObject *, Py_ssize_t *, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 50 | PyObject *, int); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 51 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 52 | #define MBENC_RESET MBENC_MAX<<1 /* reset after an encoding session */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 53 | |
| 54 | static PyObject * |
Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 55 | make_tuple(PyObject *object, Py_ssize_t len) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 56 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 57 | PyObject *v, *w; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 58 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 59 | if (object == NULL) |
| 60 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 61 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 62 | v = PyTuple_New(2); |
| 63 | if (v == NULL) { |
| 64 | Py_DECREF(object); |
| 65 | return NULL; |
| 66 | } |
| 67 | PyTuple_SET_ITEM(v, 0, object); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 68 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 69 | w = PyLong_FromSsize_t(len); |
| 70 | if (w == NULL) { |
| 71 | Py_DECREF(v); |
| 72 | return NULL; |
| 73 | } |
| 74 | PyTuple_SET_ITEM(v, 1, w); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 75 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 76 | return v; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 77 | } |
| 78 | |
| 79 | static PyObject * |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 80 | internal_error_callback(const char *errors) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 81 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 82 | if (errors == NULL || strcmp(errors, "strict") == 0) |
| 83 | return ERROR_STRICT; |
| 84 | else if (strcmp(errors, "ignore") == 0) |
| 85 | return ERROR_IGNORE; |
| 86 | else if (strcmp(errors, "replace") == 0) |
| 87 | return ERROR_REPLACE; |
| 88 | else |
| 89 | return PyUnicode_FromString(errors); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 90 | } |
| 91 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 92 | static PyObject * |
| 93 | call_error_callback(PyObject *errors, PyObject *exc) |
| 94 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 95 | PyObject *args, *cb, *r; |
| 96 | const char *str; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 97 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 98 | assert(PyUnicode_Check(errors)); |
| 99 | str = _PyUnicode_AsString(errors); |
| 100 | if (str == NULL) |
| 101 | return NULL; |
| 102 | cb = PyCodec_LookupError(str); |
| 103 | if (cb == NULL) |
| 104 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 105 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 106 | args = PyTuple_New(1); |
| 107 | if (args == NULL) { |
| 108 | Py_DECREF(cb); |
| 109 | return NULL; |
| 110 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 111 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 112 | PyTuple_SET_ITEM(args, 0, exc); |
| 113 | Py_INCREF(exc); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 114 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 115 | r = PyObject_CallObject(cb, args); |
| 116 | Py_DECREF(args); |
| 117 | Py_DECREF(cb); |
| 118 | return r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 119 | } |
| 120 | |
| 121 | static PyObject * |
| 122 | codecctx_errors_get(MultibyteStatefulCodecContext *self) |
| 123 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 124 | const char *errors; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 125 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 126 | if (self->errors == ERROR_STRICT) |
| 127 | errors = "strict"; |
| 128 | else if (self->errors == ERROR_IGNORE) |
| 129 | errors = "ignore"; |
| 130 | else if (self->errors == ERROR_REPLACE) |
| 131 | errors = "replace"; |
| 132 | else { |
| 133 | Py_INCREF(self->errors); |
| 134 | return self->errors; |
| 135 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 136 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 137 | return PyUnicode_FromString(errors); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 138 | } |
| 139 | |
| 140 | static int |
| 141 | codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 142 | void *closure) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 143 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 144 | PyObject *cb; |
| 145 | const char *str; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 146 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 147 | if (!PyUnicode_Check(value)) { |
| 148 | PyErr_SetString(PyExc_TypeError, "errors must be a string"); |
| 149 | return -1; |
| 150 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 151 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 152 | str = _PyUnicode_AsString(value); |
| 153 | if (str == NULL) |
| 154 | return -1; |
Neal Norwitz | 6ea45d3 | 2007-08-26 04:19:43 +0000 | [diff] [blame] | 155 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 156 | cb = internal_error_callback(str); |
| 157 | if (cb == NULL) |
| 158 | return -1; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 159 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 160 | ERROR_DECREF(self->errors); |
| 161 | self->errors = cb; |
| 162 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 163 | } |
| 164 | |
| 165 | /* This getset handlers list is used by all the stateful codec objects */ |
| 166 | static PyGetSetDef codecctx_getsets[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 167 | {"errors", (getter)codecctx_errors_get, |
| 168 | (setter)codecctx_errors_set, |
| 169 | PyDoc_STR("how to treat errors")}, |
| 170 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 171 | }; |
| 172 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 173 | static int |
Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 174 | expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 175 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 176 | Py_ssize_t orgpos, orgsize, incsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 177 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 178 | orgpos = (Py_ssize_t)((char *)buf->outbuf - |
| 179 | PyBytes_AS_STRING(buf->outobj)); |
| 180 | orgsize = PyBytes_GET_SIZE(buf->outobj); |
| 181 | incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize); |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 182 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 183 | if (orgsize > PY_SSIZE_T_MAX - incsize) |
| 184 | return -1; |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 185 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 186 | if (_PyBytes_Resize(&buf->outobj, orgsize + incsize) == -1) |
| 187 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 188 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 189 | buf->outbuf = (unsigned char *)PyBytes_AS_STRING(buf->outobj) +orgpos; |
| 190 | buf->outbuf_end = (unsigned char *)PyBytes_AS_STRING(buf->outobj) |
| 191 | + PyBytes_GET_SIZE(buf->outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 192 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 193 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 194 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 195 | #define REQUIRE_ENCODEBUFFER(buf, s) { \ |
| 196 | if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end) \ |
| 197 | if (expand_encodebuffer(buf, s) == -1) \ |
| 198 | goto errorexit; \ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 199 | } |
| 200 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 201 | |
| 202 | /** |
| 203 | * MultibyteCodec object |
| 204 | */ |
| 205 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 206 | static int |
| 207 | multibytecodec_encerror(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 208 | MultibyteCodec_State *state, |
| 209 | MultibyteEncodeBuffer *buf, |
| 210 | PyObject *errors, Py_ssize_t e) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 211 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 212 | PyObject *retobj = NULL, *retstr = NULL, *tobj; |
| 213 | Py_ssize_t retstrsize, newpos; |
| 214 | Py_ssize_t esize, start, end; |
| 215 | const char *reason; |
Hye-Shik 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 | if (e > 0) { |
| 218 | reason = "illegal multibyte sequence"; |
| 219 | esize = e; |
| 220 | } |
| 221 | else { |
| 222 | switch (e) { |
| 223 | case MBERR_TOOSMALL: |
| 224 | REQUIRE_ENCODEBUFFER(buf, -1); |
| 225 | return 0; /* retry it */ |
| 226 | case MBERR_TOOFEW: |
| 227 | reason = "incomplete multibyte sequence"; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 228 | esize = (Py_ssize_t)buf->inpos; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 229 | break; |
| 230 | case MBERR_INTERNAL: |
| 231 | PyErr_SetString(PyExc_RuntimeError, |
| 232 | "internal codec error"); |
| 233 | return -1; |
| 234 | default: |
| 235 | PyErr_SetString(PyExc_RuntimeError, |
| 236 | "unknown runtime error"); |
| 237 | return -1; |
| 238 | } |
| 239 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 240 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 241 | if (errors == ERROR_REPLACE) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 242 | PyObject *replchar; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 243 | Py_ssize_t r; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 244 | Py_ssize_t inpos; |
| 245 | int kind; |
| 246 | void *data; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 247 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 248 | replchar = PyUnicode_FromOrdinal('?'); |
| 249 | if (replchar == NULL) |
| 250 | goto errorexit; |
| 251 | kind = PyUnicode_KIND(replchar); |
| 252 | data = PyUnicode_DATA(replchar); |
| 253 | |
| 254 | inpos = 0; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 255 | for (;;) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 256 | Py_ssize_t outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 257 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 258 | r = codec->encode(state, codec->config, |
| 259 | kind, data, &inpos, 1, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 260 | &buf->outbuf, outleft, 0); |
| 261 | if (r == MBERR_TOOSMALL) { |
| 262 | REQUIRE_ENCODEBUFFER(buf, -1); |
| 263 | continue; |
| 264 | } |
| 265 | else |
| 266 | break; |
| 267 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 268 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 269 | Py_DECREF(replchar); |
| 270 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 271 | if (r != 0) { |
| 272 | REQUIRE_ENCODEBUFFER(buf, 1); |
| 273 | *buf->outbuf++ = '?'; |
| 274 | } |
| 275 | } |
| 276 | if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 277 | buf->inpos += esize; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 278 | return 0; |
| 279 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 280 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 281 | start = (Py_ssize_t)buf->inpos; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 282 | end = start + esize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 283 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 284 | /* use cached exception object if available */ |
| 285 | if (buf->excobj == NULL) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 286 | buf->excobj = PyObject_CallFunction(PyExc_UnicodeEncodeError, |
| 287 | "sOnns", |
| 288 | codec->encoding, buf->inobj, |
| 289 | start, end, reason); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 290 | if (buf->excobj == NULL) |
| 291 | goto errorexit; |
| 292 | } |
| 293 | else |
| 294 | if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 || |
| 295 | PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 || |
| 296 | PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0) |
| 297 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 298 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 299 | if (errors == ERROR_STRICT) { |
| 300 | PyCodec_StrictErrors(buf->excobj); |
| 301 | goto errorexit; |
| 302 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 303 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 304 | retobj = call_error_callback(errors, buf->excobj); |
| 305 | if (retobj == NULL) |
| 306 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 307 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 308 | if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || |
Benjamin Peterson | 47a00f3 | 2012-12-02 11:20:28 -0500 | [diff] [blame] | 309 | (!PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) && !PyBytes_Check(tobj)) || |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 310 | !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { |
| 311 | PyErr_SetString(PyExc_TypeError, |
| 312 | "encoding error handler must return " |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 313 | "(str, int) tuple"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 314 | goto errorexit; |
| 315 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 316 | |
Benjamin Peterson | 47a00f3 | 2012-12-02 11:20:28 -0500 | [diff] [blame] | 317 | if (PyUnicode_Check(tobj)) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 318 | Py_ssize_t inpos; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 319 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 320 | retstr = multibytecodec_encode(codec, state, tobj, |
| 321 | &inpos, ERROR_STRICT, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 322 | MBENC_FLUSH); |
| 323 | if (retstr == NULL) |
| 324 | goto errorexit; |
| 325 | } |
Benjamin Peterson | 47a00f3 | 2012-12-02 11:20:28 -0500 | [diff] [blame] | 326 | else { |
| 327 | Py_INCREF(tobj); |
| 328 | retstr = tobj; |
| 329 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 330 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 331 | assert(PyBytes_Check(retstr)); |
| 332 | retstrsize = PyBytes_GET_SIZE(retstr); |
| 333 | REQUIRE_ENCODEBUFFER(buf, retstrsize); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 334 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 335 | memcpy(buf->outbuf, PyBytes_AS_STRING(retstr), retstrsize); |
| 336 | buf->outbuf += retstrsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 337 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 338 | newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); |
| 339 | if (newpos < 0 && !PyErr_Occurred()) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 340 | newpos += (Py_ssize_t)buf->inlen; |
| 341 | if (newpos < 0 || newpos > buf->inlen) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 342 | PyErr_Clear(); |
| 343 | PyErr_Format(PyExc_IndexError, |
| 344 | "position %zd from error handler out of bounds", |
| 345 | newpos); |
| 346 | goto errorexit; |
| 347 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 348 | buf->inpos = newpos; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 349 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 350 | Py_DECREF(retobj); |
| 351 | Py_DECREF(retstr); |
| 352 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 353 | |
| 354 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 355 | Py_XDECREF(retobj); |
| 356 | Py_XDECREF(retstr); |
| 357 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 358 | } |
| 359 | |
| 360 | static int |
| 361 | multibytecodec_decerror(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 362 | MultibyteCodec_State *state, |
| 363 | MultibyteDecodeBuffer *buf, |
| 364 | PyObject *errors, Py_ssize_t e) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 365 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 366 | PyObject *retobj = NULL, *retuni = NULL; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 367 | Py_ssize_t newpos; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 368 | const char *reason; |
| 369 | Py_ssize_t esize, start, end; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 370 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 371 | if (e > 0) { |
| 372 | reason = "illegal multibyte sequence"; |
| 373 | esize = e; |
| 374 | } |
| 375 | else { |
| 376 | switch (e) { |
| 377 | case MBERR_TOOSMALL: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 378 | return 0; /* retry it */ |
| 379 | case MBERR_TOOFEW: |
| 380 | reason = "incomplete multibyte sequence"; |
| 381 | esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
| 382 | break; |
| 383 | case MBERR_INTERNAL: |
| 384 | PyErr_SetString(PyExc_RuntimeError, |
| 385 | "internal codec error"); |
| 386 | return -1; |
Victor Stinner | d1f9942 | 2013-07-16 21:41:43 +0200 | [diff] [blame] | 387 | case MBERR_EXCEPTION: |
| 388 | return -1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 389 | default: |
| 390 | PyErr_SetString(PyExc_RuntimeError, |
| 391 | "unknown runtime error"); |
| 392 | return -1; |
| 393 | } |
| 394 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 395 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 396 | if (errors == ERROR_REPLACE) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 397 | if (_PyUnicodeWriter_WriteChar(&buf->writer, |
| 398 | Py_UNICODE_REPLACEMENT_CHARACTER) < 0) |
| 399 | goto errorexit; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 400 | } |
| 401 | if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { |
| 402 | buf->inbuf += esize; |
| 403 | return 0; |
| 404 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 405 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 406 | start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top); |
| 407 | end = start + esize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 408 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 409 | /* use cached exception object if available */ |
| 410 | if (buf->excobj == NULL) { |
| 411 | buf->excobj = PyUnicodeDecodeError_Create(codec->encoding, |
| 412 | (const char *)buf->inbuf_top, |
| 413 | (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top), |
| 414 | start, end, reason); |
| 415 | if (buf->excobj == NULL) |
| 416 | goto errorexit; |
| 417 | } |
| 418 | else |
| 419 | if (PyUnicodeDecodeError_SetStart(buf->excobj, start) || |
| 420 | PyUnicodeDecodeError_SetEnd(buf->excobj, end) || |
| 421 | PyUnicodeDecodeError_SetReason(buf->excobj, reason)) |
| 422 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 423 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 424 | if (errors == ERROR_STRICT) { |
| 425 | PyCodec_StrictErrors(buf->excobj); |
| 426 | goto errorexit; |
| 427 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 428 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 429 | retobj = call_error_callback(errors, buf->excobj); |
| 430 | if (retobj == NULL) |
| 431 | goto errorexit; |
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 | if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || |
| 434 | !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) || |
| 435 | !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { |
| 436 | PyErr_SetString(PyExc_TypeError, |
| 437 | "decoding error handler must return " |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 438 | "(str, int) tuple"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 439 | goto errorexit; |
| 440 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 441 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 442 | if (_PyUnicodeWriter_WriteStr(&buf->writer, retuni) < 0) |
Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 443 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 444 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 445 | newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); |
| 446 | if (newpos < 0 && !PyErr_Occurred()) |
| 447 | newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top); |
| 448 | if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) { |
| 449 | PyErr_Clear(); |
| 450 | PyErr_Format(PyExc_IndexError, |
| 451 | "position %zd from error handler out of bounds", |
| 452 | newpos); |
| 453 | goto errorexit; |
| 454 | } |
| 455 | buf->inbuf = buf->inbuf_top + newpos; |
| 456 | Py_DECREF(retobj); |
| 457 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 458 | |
| 459 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 460 | Py_XDECREF(retobj); |
| 461 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 462 | } |
| 463 | |
| 464 | static PyObject * |
| 465 | multibytecodec_encode(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 466 | MultibyteCodec_State *state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 467 | PyObject *text, Py_ssize_t *inpos_t, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 468 | PyObject *errors, int flags) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 469 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 470 | MultibyteEncodeBuffer buf; |
| 471 | Py_ssize_t finalsize, r = 0; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 472 | Py_ssize_t datalen; |
| 473 | int kind; |
| 474 | void *data; |
| 475 | |
| 476 | if (PyUnicode_READY(text) < 0) |
| 477 | return NULL; |
| 478 | datalen = PyUnicode_GET_LENGTH(text); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 479 | |
Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 480 | if (datalen == 0 && !(flags & MBENC_RESET)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 481 | return PyBytes_FromStringAndSize(NULL, 0); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 482 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 483 | buf.excobj = NULL; |
Brett Cannon | b94767f | 2011-02-22 20:15:44 +0000 | [diff] [blame] | 484 | buf.outobj = NULL; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 485 | buf.inobj = text; /* borrowed reference */ |
| 486 | buf.inpos = 0; |
| 487 | buf.inlen = datalen; |
| 488 | kind = PyUnicode_KIND(buf.inobj); |
| 489 | data = PyUnicode_DATA(buf.inobj); |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 490 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 491 | if (datalen > (PY_SSIZE_T_MAX - 16) / 2) { |
| 492 | PyErr_NoMemory(); |
| 493 | goto errorexit; |
| 494 | } |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 495 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 496 | buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16); |
| 497 | if (buf.outobj == NULL) |
| 498 | goto errorexit; |
| 499 | buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj); |
| 500 | buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 501 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 502 | while (buf.inpos < buf.inlen) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 503 | /* we don't reuse inleft and outleft here. |
| 504 | * error callbacks can relocate the cursor anywhere on buffer*/ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 505 | Py_ssize_t outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
| 506 | |
| 507 | r = codec->encode(state, codec->config, |
| 508 | kind, data, |
| 509 | &buf.inpos, buf.inlen, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 510 | &buf.outbuf, outleft, flags); |
| 511 | if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH))) |
| 512 | break; |
| 513 | else if (multibytecodec_encerror(codec, state, &buf, errors,r)) |
| 514 | goto errorexit; |
| 515 | else if (r == MBERR_TOOFEW) |
| 516 | break; |
| 517 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 518 | |
Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 519 | if (codec->encreset != NULL && (flags & MBENC_RESET)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 520 | for (;;) { |
| 521 | Py_ssize_t outleft; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 522 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 523 | outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
| 524 | r = codec->encreset(state, codec->config, &buf.outbuf, |
| 525 | outleft); |
| 526 | if (r == 0) |
| 527 | break; |
| 528 | else if (multibytecodec_encerror(codec, state, |
| 529 | &buf, errors, r)) |
| 530 | goto errorexit; |
| 531 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 532 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 533 | finalsize = (Py_ssize_t)((char *)buf.outbuf - |
| 534 | PyBytes_AS_STRING(buf.outobj)); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 535 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 536 | if (finalsize != PyBytes_GET_SIZE(buf.outobj)) |
| 537 | if (_PyBytes_Resize(&buf.outobj, finalsize) == -1) |
| 538 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 539 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 540 | if (inpos_t) |
| 541 | *inpos_t = buf.inpos; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 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; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 556 | PyObject *errorcb, *r, *arg, *ucvt; |
| 557 | const char *errors = NULL; |
| 558 | Py_ssize_t datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 559 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 560 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode", |
| 561 | codeckwarglist, &arg, &errors)) |
| 562 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 563 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 564 | if (PyUnicode_Check(arg)) |
| 565 | ucvt = NULL; |
| 566 | else { |
| 567 | arg = ucvt = PyObject_Str(arg); |
| 568 | if (arg == NULL) |
| 569 | return NULL; |
| 570 | else if (!PyUnicode_Check(arg)) { |
| 571 | PyErr_SetString(PyExc_TypeError, |
| 572 | "couldn't convert the object to unicode."); |
| 573 | Py_DECREF(ucvt); |
| 574 | return NULL; |
| 575 | } |
| 576 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 577 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 578 | if (PyUnicode_READY(arg) < 0) { |
Victor Stinner | 9a80fab | 2011-11-21 02:50:14 +0100 | [diff] [blame] | 579 | Py_XDECREF(ucvt); |
| 580 | return NULL; |
| 581 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 582 | datalen = PyUnicode_GET_LENGTH(arg); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 583 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 584 | errorcb = internal_error_callback(errors); |
| 585 | if (errorcb == NULL) { |
| 586 | Py_XDECREF(ucvt); |
| 587 | return NULL; |
| 588 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 589 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 590 | if (self->codec->encinit != NULL && |
| 591 | self->codec->encinit(&state, self->codec->config) != 0) |
| 592 | goto errorexit; |
| 593 | r = multibytecodec_encode(self->codec, &state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 594 | arg, NULL, errorcb, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 595 | MBENC_FLUSH | MBENC_RESET); |
| 596 | if (r == NULL) |
| 597 | goto errorexit; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 598 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 599 | ERROR_DECREF(errorcb); |
| 600 | Py_XDECREF(ucvt); |
| 601 | return make_tuple(r, datalen); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 602 | |
| 603 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 604 | ERROR_DECREF(errorcb); |
| 605 | Py_XDECREF(ucvt); |
| 606 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 607 | } |
| 608 | |
| 609 | static PyObject * |
| 610 | MultibyteCodec_Decode(MultibyteCodecObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 611 | PyObject *args, PyObject *kwargs) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 612 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 613 | MultibyteCodec_State state; |
| 614 | MultibyteDecodeBuffer buf; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 615 | PyObject *errorcb, *res; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 616 | Py_buffer pdata; |
| 617 | const char *data, *errors = NULL; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 618 | Py_ssize_t datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 619 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 620 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|z:decode", |
| 621 | codeckwarglist, &pdata, &errors)) |
| 622 | return NULL; |
| 623 | data = pdata.buf; |
| 624 | datalen = pdata.len; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 625 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 626 | errorcb = internal_error_callback(errors); |
| 627 | if (errorcb == NULL) { |
| 628 | PyBuffer_Release(&pdata); |
| 629 | return NULL; |
| 630 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 631 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 632 | if (datalen == 0) { |
| 633 | PyBuffer_Release(&pdata); |
| 634 | ERROR_DECREF(errorcb); |
Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 635 | return make_tuple(PyUnicode_New(0, 0), 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 636 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 637 | |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 638 | _PyUnicodeWriter_Init(&buf.writer); |
| 639 | buf.writer.min_length = datalen; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 640 | buf.excobj = NULL; |
| 641 | buf.inbuf = buf.inbuf_top = (unsigned char *)data; |
| 642 | buf.inbuf_end = buf.inbuf_top + datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 643 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 644 | if (self->codec->decinit != NULL && |
| 645 | self->codec->decinit(&state, self->codec->config) != 0) |
| 646 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 647 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 648 | while (buf.inbuf < buf.inbuf_end) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 649 | Py_ssize_t inleft, r; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 650 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 651 | inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 652 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 653 | r = self->codec->decode(&state, self->codec->config, |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 654 | &buf.inbuf, inleft, &buf.writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 655 | if (r == 0) |
| 656 | break; |
| 657 | else if (multibytecodec_decerror(self->codec, &state, |
| 658 | &buf, errorcb, r)) |
| 659 | goto errorexit; |
| 660 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 661 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 662 | res = _PyUnicodeWriter_Finish(&buf.writer); |
| 663 | if (res == NULL) |
| 664 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 665 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 666 | PyBuffer_Release(&pdata); |
| 667 | Py_XDECREF(buf.excobj); |
| 668 | ERROR_DECREF(errorcb); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 669 | return make_tuple(res, datalen); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 670 | |
| 671 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 672 | PyBuffer_Release(&pdata); |
| 673 | ERROR_DECREF(errorcb); |
| 674 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 675 | _PyUnicodeWriter_Dealloc(&buf.writer); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 676 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 677 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 678 | } |
| 679 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 680 | static struct PyMethodDef multibytecodec_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 681 | {"encode", (PyCFunction)MultibyteCodec_Encode, |
| 682 | METH_VARARGS | METH_KEYWORDS, |
| 683 | MultibyteCodec_Encode__doc__}, |
| 684 | {"decode", (PyCFunction)MultibyteCodec_Decode, |
| 685 | METH_VARARGS | METH_KEYWORDS, |
| 686 | MultibyteCodec_Decode__doc__}, |
| 687 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 688 | }; |
| 689 | |
| 690 | static void |
| 691 | multibytecodec_dealloc(MultibyteCodecObject *self) |
| 692 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 693 | PyObject_Del(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 694 | } |
| 695 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 696 | static PyTypeObject MultibyteCodec_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 697 | PyVarObject_HEAD_INIT(NULL, 0) |
| 698 | "MultibyteCodec", /* tp_name */ |
| 699 | sizeof(MultibyteCodecObject), /* tp_basicsize */ |
| 700 | 0, /* tp_itemsize */ |
| 701 | /* methods */ |
| 702 | (destructor)multibytecodec_dealloc, /* tp_dealloc */ |
| 703 | 0, /* tp_print */ |
| 704 | 0, /* tp_getattr */ |
| 705 | 0, /* tp_setattr */ |
| 706 | 0, /* tp_reserved */ |
| 707 | 0, /* tp_repr */ |
| 708 | 0, /* tp_as_number */ |
| 709 | 0, /* tp_as_sequence */ |
| 710 | 0, /* tp_as_mapping */ |
| 711 | 0, /* tp_hash */ |
| 712 | 0, /* tp_call */ |
| 713 | 0, /* tp_str */ |
| 714 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 715 | 0, /* tp_setattro */ |
| 716 | 0, /* tp_as_buffer */ |
| 717 | Py_TPFLAGS_DEFAULT, /* tp_flags */ |
| 718 | 0, /* tp_doc */ |
| 719 | 0, /* tp_traverse */ |
| 720 | 0, /* tp_clear */ |
| 721 | 0, /* tp_richcompare */ |
| 722 | 0, /* tp_weaklistoffset */ |
| 723 | 0, /* tp_iter */ |
| 724 | 0, /* tp_iterext */ |
| 725 | multibytecodec_methods, /* tp_methods */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 726 | }; |
| 727 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 728 | |
| 729 | /** |
| 730 | * Utility functions for stateful codec mechanism |
| 731 | */ |
| 732 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 733 | #define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o)) |
| 734 | #define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 735 | |
| 736 | static PyObject * |
| 737 | encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 738 | PyObject *unistr, int final) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 739 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 740 | PyObject *ucvt, *r = NULL; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 741 | PyObject *inbuf = NULL; |
| 742 | Py_ssize_t inpos, datalen; |
| 743 | PyObject *origpending = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 744 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 745 | if (PyUnicode_Check(unistr)) |
| 746 | ucvt = NULL; |
| 747 | else { |
| 748 | unistr = ucvt = PyObject_Str(unistr); |
| 749 | if (unistr == NULL) |
| 750 | return NULL; |
| 751 | else if (!PyUnicode_Check(unistr)) { |
| 752 | PyErr_SetString(PyExc_TypeError, |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 753 | "couldn't convert the object to str."); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 754 | Py_DECREF(ucvt); |
| 755 | return NULL; |
| 756 | } |
| 757 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 758 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 759 | if (ctx->pending) { |
| 760 | PyObject *inbuf_tmp; |
| 761 | |
| 762 | Py_INCREF(ctx->pending); |
| 763 | origpending = ctx->pending; |
| 764 | |
| 765 | Py_INCREF(ctx->pending); |
| 766 | inbuf_tmp = ctx->pending; |
| 767 | PyUnicode_Append(&inbuf_tmp, unistr); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 768 | if (inbuf_tmp == NULL) |
| 769 | goto errorexit; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 770 | Py_CLEAR(ctx->pending); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 771 | inbuf = inbuf_tmp; |
| 772 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 773 | else { |
| 774 | origpending = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 775 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 776 | Py_INCREF(unistr); |
| 777 | inbuf = unistr; |
| 778 | } |
| 779 | if (PyUnicode_READY(inbuf) < 0) |
| 780 | goto errorexit; |
| 781 | inpos = 0; |
| 782 | datalen = PyUnicode_GET_LENGTH(inbuf); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 783 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 784 | r = multibytecodec_encode(ctx->codec, &ctx->state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 785 | inbuf, &inpos, |
| 786 | ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 787 | if (r == NULL) { |
| 788 | /* recover the original pending buffer */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 789 | Py_CLEAR(ctx->pending); |
| 790 | ctx->pending = origpending; |
| 791 | origpending = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 792 | goto errorexit; |
| 793 | } |
Victor Stinner | 322cc74 | 2013-04-14 18:11:41 +0200 | [diff] [blame] | 794 | Py_XDECREF(origpending); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 795 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 796 | if (inpos < datalen) { |
| 797 | if (datalen - inpos > MAXENCPENDING) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 798 | /* normal codecs can't reach here */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 799 | PyErr_SetString(PyExc_UnicodeError, |
| 800 | "pending buffer overflow"); |
| 801 | goto errorexit; |
| 802 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 803 | ctx->pending = PyUnicode_Substring(inbuf, inpos, datalen); |
| 804 | if (ctx->pending == NULL) { |
| 805 | /* normal codecs can't reach here */ |
| 806 | goto errorexit; |
| 807 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 808 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 809 | |
Victor Stinner | 322cc74 | 2013-04-14 18:11:41 +0200 | [diff] [blame] | 810 | Py_DECREF(inbuf); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 811 | Py_XDECREF(ucvt); |
| 812 | return r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 813 | |
| 814 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 815 | Py_XDECREF(r); |
| 816 | Py_XDECREF(ucvt); |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 817 | Py_XDECREF(origpending); |
Victor Stinner | 322cc74 | 2013-04-14 18:11:41 +0200 | [diff] [blame] | 818 | Py_XDECREF(inbuf); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 819 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 820 | } |
| 821 | |
| 822 | static int |
| 823 | decoder_append_pending(MultibyteStatefulDecoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 824 | MultibyteDecodeBuffer *buf) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 825 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 826 | Py_ssize_t npendings; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 827 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 828 | npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
| 829 | if (npendings + ctx->pendingsize > MAXDECPENDING || |
| 830 | npendings > PY_SSIZE_T_MAX - ctx->pendingsize) { |
| 831 | PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow"); |
| 832 | return -1; |
| 833 | } |
| 834 | memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings); |
| 835 | ctx->pendingsize += npendings; |
| 836 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 837 | } |
| 838 | |
| 839 | static int |
| 840 | decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 841 | Py_ssize_t size) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 842 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 843 | buf->inbuf = buf->inbuf_top = (const unsigned char *)data; |
| 844 | buf->inbuf_end = buf->inbuf_top + size; |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 845 | buf->writer.min_length += size; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 846 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 847 | } |
| 848 | |
| 849 | static int |
| 850 | decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 851 | MultibyteDecodeBuffer *buf) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 852 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 853 | while (buf->inbuf < buf->inbuf_end) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 854 | Py_ssize_t inleft; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 855 | Py_ssize_t r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 856 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 857 | inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 858 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 859 | r = ctx->codec->decode(&ctx->state, ctx->codec->config, |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 860 | &buf->inbuf, inleft, &buf->writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 861 | if (r == 0 || r == MBERR_TOOFEW) |
| 862 | break; |
| 863 | else if (multibytecodec_decerror(ctx->codec, &ctx->state, |
| 864 | buf, ctx->errors, r)) |
| 865 | return -1; |
| 866 | } |
| 867 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 868 | } |
| 869 | |
| 870 | |
| 871 | /** |
| 872 | * MultibyteIncrementalEncoder object |
| 873 | */ |
| 874 | |
| 875 | static PyObject * |
| 876 | mbiencoder_encode(MultibyteIncrementalEncoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 877 | PyObject *args, PyObject *kwargs) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 878 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 879 | PyObject *data; |
| 880 | int final = 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 881 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 882 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode", |
| 883 | incrementalkwarglist, &data, &final)) |
| 884 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 885 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 886 | return encoder_encode_stateful(STATEFUL_ECTX(self), data, final); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 887 | } |
| 888 | |
| 889 | static PyObject * |
| 890 | mbiencoder_reset(MultibyteIncrementalEncoderObject *self) |
| 891 | { |
Victor Stinner | e15dce3 | 2011-05-30 22:56:00 +0200 | [diff] [blame] | 892 | /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */ |
| 893 | unsigned char buffer[4], *outbuf; |
| 894 | Py_ssize_t r; |
| 895 | if (self->codec->encreset != NULL) { |
| 896 | outbuf = buffer; |
| 897 | r = self->codec->encreset(&self->state, self->codec->config, |
| 898 | &outbuf, sizeof(buffer)); |
| 899 | if (r != 0) |
| 900 | return NULL; |
| 901 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 902 | Py_CLEAR(self->pending); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 903 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 904 | } |
| 905 | |
| 906 | static struct PyMethodDef mbiencoder_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 907 | {"encode", (PyCFunction)mbiencoder_encode, |
| 908 | METH_VARARGS | METH_KEYWORDS, NULL}, |
| 909 | {"reset", (PyCFunction)mbiencoder_reset, |
| 910 | METH_NOARGS, NULL}, |
| 911 | {NULL, NULL}, |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 912 | }; |
| 913 | |
| 914 | static PyObject * |
| 915 | mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 916 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 917 | MultibyteIncrementalEncoderObject *self; |
| 918 | PyObject *codec = NULL; |
| 919 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 920 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 921 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder", |
| 922 | incnewkwarglist, &errors)) |
| 923 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 924 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 925 | self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0); |
| 926 | if (self == NULL) |
| 927 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 928 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 929 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 930 | if (codec == NULL) |
| 931 | goto errorexit; |
| 932 | if (!MultibyteCodec_Check(codec)) { |
| 933 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 934 | goto errorexit; |
| 935 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 936 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 937 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 938 | self->pending = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 939 | self->errors = internal_error_callback(errors); |
| 940 | if (self->errors == NULL) |
| 941 | goto errorexit; |
| 942 | if (self->codec->encinit != NULL && |
| 943 | self->codec->encinit(&self->state, self->codec->config) != 0) |
| 944 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 945 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 946 | Py_DECREF(codec); |
| 947 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 948 | |
| 949 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 950 | Py_XDECREF(self); |
| 951 | Py_XDECREF(codec); |
| 952 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 953 | } |
| 954 | |
| 955 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 956 | mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 957 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 958 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 959 | } |
| 960 | |
| 961 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 962 | mbiencoder_traverse(MultibyteIncrementalEncoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 963 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 964 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 965 | if (ERROR_ISCUSTOM(self->errors)) |
| 966 | Py_VISIT(self->errors); |
| 967 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 968 | } |
| 969 | |
| 970 | static void |
| 971 | mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self) |
| 972 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 973 | PyObject_GC_UnTrack(self); |
| 974 | ERROR_DECREF(self->errors); |
| 975 | Py_TYPE(self)->tp_free(self); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 976 | } |
| 977 | |
| 978 | static PyTypeObject MultibyteIncrementalEncoder_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 979 | PyVarObject_HEAD_INIT(NULL, 0) |
| 980 | "MultibyteIncrementalEncoder", /* tp_name */ |
| 981 | sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */ |
| 982 | 0, /* tp_itemsize */ |
| 983 | /* methods */ |
| 984 | (destructor)mbiencoder_dealloc, /* tp_dealloc */ |
| 985 | 0, /* tp_print */ |
| 986 | 0, /* tp_getattr */ |
| 987 | 0, /* tp_setattr */ |
| 988 | 0, /* tp_reserved */ |
| 989 | 0, /* tp_repr */ |
| 990 | 0, /* tp_as_number */ |
| 991 | 0, /* tp_as_sequence */ |
| 992 | 0, /* tp_as_mapping */ |
| 993 | 0, /* tp_hash */ |
| 994 | 0, /* tp_call */ |
| 995 | 0, /* tp_str */ |
| 996 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 997 | 0, /* tp_setattro */ |
| 998 | 0, /* tp_as_buffer */ |
| 999 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1000 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1001 | 0, /* tp_doc */ |
| 1002 | (traverseproc)mbiencoder_traverse, /* tp_traverse */ |
| 1003 | 0, /* tp_clear */ |
| 1004 | 0, /* tp_richcompare */ |
| 1005 | 0, /* tp_weaklistoffset */ |
| 1006 | 0, /* tp_iter */ |
| 1007 | 0, /* tp_iterext */ |
| 1008 | mbiencoder_methods, /* tp_methods */ |
| 1009 | 0, /* tp_members */ |
| 1010 | codecctx_getsets, /* tp_getset */ |
| 1011 | 0, /* tp_base */ |
| 1012 | 0, /* tp_dict */ |
| 1013 | 0, /* tp_descr_get */ |
| 1014 | 0, /* tp_descr_set */ |
| 1015 | 0, /* tp_dictoffset */ |
| 1016 | mbiencoder_init, /* tp_init */ |
| 1017 | 0, /* tp_alloc */ |
| 1018 | mbiencoder_new, /* tp_new */ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1019 | }; |
| 1020 | |
| 1021 | |
| 1022 | /** |
| 1023 | * MultibyteIncrementalDecoder object |
| 1024 | */ |
| 1025 | |
| 1026 | static PyObject * |
| 1027 | mbidecoder_decode(MultibyteIncrementalDecoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1028 | PyObject *args, PyObject *kwargs) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1029 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1030 | MultibyteDecodeBuffer buf; |
| 1031 | char *data, *wdata = NULL; |
| 1032 | Py_buffer pdata; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1033 | Py_ssize_t wsize, size, origpending; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1034 | int final = 0; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1035 | PyObject *res; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1036 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1037 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|i:decode", |
| 1038 | incrementalkwarglist, &pdata, &final)) |
| 1039 | return NULL; |
| 1040 | data = pdata.buf; |
| 1041 | size = pdata.len; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1042 | |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 1043 | _PyUnicodeWriter_Init(&buf.writer); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1044 | buf.excobj = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1045 | origpending = self->pendingsize; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1046 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1047 | if (self->pendingsize == 0) { |
| 1048 | wsize = size; |
| 1049 | wdata = data; |
| 1050 | } |
| 1051 | else { |
| 1052 | if (size > PY_SSIZE_T_MAX - self->pendingsize) { |
| 1053 | PyErr_NoMemory(); |
| 1054 | goto errorexit; |
| 1055 | } |
| 1056 | wsize = size + self->pendingsize; |
| 1057 | wdata = PyMem_Malloc(wsize); |
Victor Stinner | 33283ba | 2013-07-15 17:47:39 +0200 | [diff] [blame] | 1058 | if (wdata == NULL) { |
| 1059 | PyErr_NoMemory(); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1060 | goto errorexit; |
Victor Stinner | 33283ba | 2013-07-15 17:47:39 +0200 | [diff] [blame] | 1061 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1062 | memcpy(wdata, self->pending, self->pendingsize); |
| 1063 | memcpy(wdata + self->pendingsize, data, size); |
| 1064 | self->pendingsize = 0; |
| 1065 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1066 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1067 | if (decoder_prepare_buffer(&buf, wdata, wsize) != 0) |
| 1068 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1069 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1070 | if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf)) |
| 1071 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1072 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1073 | if (final && buf.inbuf < buf.inbuf_end) { |
| 1074 | if (multibytecodec_decerror(self->codec, &self->state, |
| 1075 | &buf, self->errors, MBERR_TOOFEW)) { |
| 1076 | /* recover the original pending buffer */ |
| 1077 | memcpy(self->pending, wdata, origpending); |
| 1078 | self->pendingsize = origpending; |
| 1079 | goto errorexit; |
| 1080 | } |
| 1081 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1082 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1083 | if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */ |
| 1084 | if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0) |
| 1085 | goto errorexit; |
| 1086 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1087 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1088 | res = _PyUnicodeWriter_Finish(&buf.writer); |
| 1089 | if (res == NULL) |
| 1090 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1091 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1092 | PyBuffer_Release(&pdata); |
| 1093 | if (wdata != data) |
| 1094 | PyMem_Del(wdata); |
| 1095 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1096 | return res; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1097 | |
| 1098 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1099 | PyBuffer_Release(&pdata); |
| 1100 | if (wdata != NULL && wdata != data) |
| 1101 | PyMem_Del(wdata); |
| 1102 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1103 | _PyUnicodeWriter_Dealloc(&buf.writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1104 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1105 | } |
| 1106 | |
| 1107 | static PyObject * |
| 1108 | mbidecoder_reset(MultibyteIncrementalDecoderObject *self) |
| 1109 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1110 | if (self->codec->decreset != NULL && |
| 1111 | self->codec->decreset(&self->state, self->codec->config) != 0) |
| 1112 | return NULL; |
| 1113 | self->pendingsize = 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1114 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1115 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1116 | } |
| 1117 | |
| 1118 | static struct PyMethodDef mbidecoder_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1119 | {"decode", (PyCFunction)mbidecoder_decode, |
| 1120 | METH_VARARGS | METH_KEYWORDS, NULL}, |
| 1121 | {"reset", (PyCFunction)mbidecoder_reset, |
| 1122 | METH_NOARGS, NULL}, |
| 1123 | {NULL, NULL}, |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1124 | }; |
| 1125 | |
| 1126 | static PyObject * |
| 1127 | mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1128 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1129 | MultibyteIncrementalDecoderObject *self; |
| 1130 | PyObject *codec = NULL; |
| 1131 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1132 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1133 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder", |
| 1134 | incnewkwarglist, &errors)) |
| 1135 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1136 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1137 | self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0); |
| 1138 | if (self == NULL) |
| 1139 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1140 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1141 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1142 | if (codec == NULL) |
| 1143 | goto errorexit; |
| 1144 | if (!MultibyteCodec_Check(codec)) { |
| 1145 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1146 | goto errorexit; |
| 1147 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1148 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1149 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1150 | self->pendingsize = 0; |
| 1151 | self->errors = internal_error_callback(errors); |
| 1152 | if (self->errors == NULL) |
| 1153 | goto errorexit; |
| 1154 | if (self->codec->decinit != NULL && |
| 1155 | self->codec->decinit(&self->state, self->codec->config) != 0) |
| 1156 | goto errorexit; |
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 | Py_DECREF(codec); |
| 1159 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1160 | |
| 1161 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1162 | Py_XDECREF(self); |
| 1163 | Py_XDECREF(codec); |
| 1164 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1165 | } |
| 1166 | |
| 1167 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1168 | mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1169 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1170 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1171 | } |
| 1172 | |
| 1173 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1174 | mbidecoder_traverse(MultibyteIncrementalDecoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1175 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1176 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1177 | if (ERROR_ISCUSTOM(self->errors)) |
| 1178 | Py_VISIT(self->errors); |
| 1179 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1180 | } |
| 1181 | |
| 1182 | static void |
| 1183 | mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self) |
| 1184 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1185 | PyObject_GC_UnTrack(self); |
| 1186 | ERROR_DECREF(self->errors); |
| 1187 | Py_TYPE(self)->tp_free(self); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1188 | } |
| 1189 | |
| 1190 | static PyTypeObject MultibyteIncrementalDecoder_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1191 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1192 | "MultibyteIncrementalDecoder", /* tp_name */ |
| 1193 | sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */ |
| 1194 | 0, /* tp_itemsize */ |
| 1195 | /* methods */ |
| 1196 | (destructor)mbidecoder_dealloc, /* tp_dealloc */ |
| 1197 | 0, /* tp_print */ |
| 1198 | 0, /* tp_getattr */ |
| 1199 | 0, /* tp_setattr */ |
| 1200 | 0, /* tp_reserved */ |
| 1201 | 0, /* tp_repr */ |
| 1202 | 0, /* tp_as_number */ |
| 1203 | 0, /* tp_as_sequence */ |
| 1204 | 0, /* tp_as_mapping */ |
| 1205 | 0, /* tp_hash */ |
| 1206 | 0, /* tp_call */ |
| 1207 | 0, /* tp_str */ |
| 1208 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1209 | 0, /* tp_setattro */ |
| 1210 | 0, /* tp_as_buffer */ |
| 1211 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1212 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1213 | 0, /* tp_doc */ |
| 1214 | (traverseproc)mbidecoder_traverse, /* tp_traverse */ |
| 1215 | 0, /* tp_clear */ |
| 1216 | 0, /* tp_richcompare */ |
| 1217 | 0, /* tp_weaklistoffset */ |
| 1218 | 0, /* tp_iter */ |
| 1219 | 0, /* tp_iterext */ |
| 1220 | mbidecoder_methods, /* tp_methods */ |
| 1221 | 0, /* tp_members */ |
| 1222 | codecctx_getsets, /* tp_getset */ |
| 1223 | 0, /* tp_base */ |
| 1224 | 0, /* tp_dict */ |
| 1225 | 0, /* tp_descr_get */ |
| 1226 | 0, /* tp_descr_set */ |
| 1227 | 0, /* tp_dictoffset */ |
| 1228 | mbidecoder_init, /* tp_init */ |
| 1229 | 0, /* tp_alloc */ |
| 1230 | mbidecoder_new, /* tp_new */ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1231 | }; |
| 1232 | |
| 1233 | |
| 1234 | /** |
| 1235 | * MultibyteStreamReader object |
| 1236 | */ |
| 1237 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1238 | static PyObject * |
| 1239 | mbstreamreader_iread(MultibyteStreamReaderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1240 | const char *method, Py_ssize_t sizehint) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1241 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1242 | MultibyteDecodeBuffer buf; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1243 | PyObject *cres, *res; |
| 1244 | Py_ssize_t rsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1245 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1246 | if (sizehint == 0) |
Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 1247 | return PyUnicode_New(0, 0); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1248 | |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 1249 | _PyUnicodeWriter_Init(&buf.writer); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1250 | buf.excobj = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1251 | cres = NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1252 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1253 | for (;;) { |
| 1254 | int endoffile; |
Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1255 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1256 | if (sizehint < 0) |
| 1257 | cres = PyObject_CallMethod(self->stream, |
| 1258 | (char *)method, NULL); |
| 1259 | else |
| 1260 | cres = PyObject_CallMethod(self->stream, |
| 1261 | (char *)method, "i", sizehint); |
| 1262 | if (cres == NULL) |
| 1263 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1264 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1265 | if (!PyBytes_Check(cres)) { |
| 1266 | PyErr_Format(PyExc_TypeError, |
| 1267 | "stream function returned a " |
| 1268 | "non-bytes object (%.100s)", |
| 1269 | cres->ob_type->tp_name); |
| 1270 | goto errorexit; |
| 1271 | } |
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 | endoffile = (PyBytes_GET_SIZE(cres) == 0); |
Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1274 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1275 | if (self->pendingsize > 0) { |
| 1276 | PyObject *ctr; |
| 1277 | char *ctrdata; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1278 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1279 | if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) { |
| 1280 | PyErr_NoMemory(); |
| 1281 | goto errorexit; |
Victor Stinner | 064bbdc | 2013-07-08 22:28:27 +0200 | [diff] [blame] | 1282 | } |
| 1283 | rsize = PyBytes_GET_SIZE(cres) + self->pendingsize; |
| 1284 | ctr = PyBytes_FromStringAndSize(NULL, rsize); |
| 1285 | if (ctr == NULL) |
| 1286 | goto errorexit; |
| 1287 | ctrdata = PyBytes_AS_STRING(ctr); |
| 1288 | memcpy(ctrdata, self->pending, self->pendingsize); |
| 1289 | memcpy(ctrdata + self->pendingsize, |
| 1290 | PyBytes_AS_STRING(cres), |
| 1291 | PyBytes_GET_SIZE(cres)); |
| 1292 | Py_DECREF(cres); |
| 1293 | cres = ctr; |
| 1294 | self->pendingsize = 0; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1295 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1296 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1297 | rsize = PyBytes_GET_SIZE(cres); |
| 1298 | if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres), |
| 1299 | rsize) != 0) |
| 1300 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1301 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1302 | if (rsize > 0 && decoder_feed_buffer( |
| 1303 | (MultibyteStatefulDecoderContext *)self, &buf)) |
| 1304 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1305 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1306 | if (endoffile || sizehint < 0) { |
| 1307 | if (buf.inbuf < buf.inbuf_end && |
| 1308 | multibytecodec_decerror(self->codec, &self->state, |
| 1309 | &buf, self->errors, MBERR_TOOFEW)) |
| 1310 | goto errorexit; |
| 1311 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1312 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1313 | if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */ |
| 1314 | if (decoder_append_pending(STATEFUL_DCTX(self), |
| 1315 | &buf) != 0) |
| 1316 | goto errorexit; |
| 1317 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1318 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1319 | Py_DECREF(cres); |
| 1320 | cres = NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1321 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1322 | if (sizehint < 0 || buf.writer.pos != 0 || rsize == 0) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1323 | break; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1324 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1325 | sizehint = 1; /* read 1 more byte and retry */ |
| 1326 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1327 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1328 | res = _PyUnicodeWriter_Finish(&buf.writer); |
| 1329 | if (res == NULL) |
| 1330 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1331 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1332 | Py_XDECREF(cres); |
| 1333 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1334 | return res; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1335 | |
| 1336 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1337 | Py_XDECREF(cres); |
| 1338 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1339 | _PyUnicodeWriter_Dealloc(&buf.writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1340 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1341 | } |
| 1342 | |
| 1343 | static PyObject * |
| 1344 | mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args) |
| 1345 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1346 | PyObject *sizeobj = NULL; |
| 1347 | Py_ssize_t size; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1348 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1349 | if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj)) |
| 1350 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1351 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1352 | if (sizeobj == Py_None || sizeobj == NULL) |
| 1353 | size = -1; |
| 1354 | else if (PyLong_Check(sizeobj)) |
| 1355 | size = PyLong_AsSsize_t(sizeobj); |
| 1356 | else { |
| 1357 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1358 | return NULL; |
| 1359 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1360 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1361 | if (size == -1 && PyErr_Occurred()) |
| 1362 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1363 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1364 | return mbstreamreader_iread(self, "read", size); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1365 | } |
| 1366 | |
| 1367 | static PyObject * |
| 1368 | mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args) |
| 1369 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1370 | PyObject *sizeobj = NULL; |
| 1371 | Py_ssize_t size; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1372 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1373 | if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj)) |
| 1374 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1375 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1376 | if (sizeobj == Py_None || sizeobj == NULL) |
| 1377 | size = -1; |
| 1378 | else if (PyLong_Check(sizeobj)) |
| 1379 | size = PyLong_AsSsize_t(sizeobj); |
| 1380 | else { |
| 1381 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1382 | return NULL; |
| 1383 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1384 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1385 | if (size == -1 && PyErr_Occurred()) |
| 1386 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1387 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1388 | return mbstreamreader_iread(self, "readline", size); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1389 | } |
| 1390 | |
| 1391 | static PyObject * |
| 1392 | mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args) |
| 1393 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1394 | PyObject *sizehintobj = NULL, *r, *sr; |
| 1395 | Py_ssize_t sizehint; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1396 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1397 | if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj)) |
| 1398 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1399 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1400 | if (sizehintobj == Py_None || sizehintobj == NULL) |
| 1401 | sizehint = -1; |
| 1402 | else if (PyLong_Check(sizehintobj)) |
| 1403 | sizehint = PyLong_AsSsize_t(sizehintobj); |
| 1404 | else { |
| 1405 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1406 | return NULL; |
| 1407 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1408 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1409 | if (sizehint == -1 && PyErr_Occurred()) |
| 1410 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1411 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1412 | r = mbstreamreader_iread(self, "read", sizehint); |
| 1413 | if (r == NULL) |
| 1414 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1415 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1416 | sr = PyUnicode_Splitlines(r, 1); |
| 1417 | Py_DECREF(r); |
| 1418 | return sr; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1419 | } |
| 1420 | |
| 1421 | static PyObject * |
| 1422 | mbstreamreader_reset(MultibyteStreamReaderObject *self) |
| 1423 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1424 | if (self->codec->decreset != NULL && |
| 1425 | self->codec->decreset(&self->state, self->codec->config) != 0) |
| 1426 | return NULL; |
| 1427 | self->pendingsize = 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1428 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1429 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1430 | } |
| 1431 | |
| 1432 | static struct PyMethodDef mbstreamreader_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1433 | {"read", (PyCFunction)mbstreamreader_read, |
| 1434 | METH_VARARGS, NULL}, |
| 1435 | {"readline", (PyCFunction)mbstreamreader_readline, |
| 1436 | METH_VARARGS, NULL}, |
| 1437 | {"readlines", (PyCFunction)mbstreamreader_readlines, |
| 1438 | METH_VARARGS, NULL}, |
| 1439 | {"reset", (PyCFunction)mbstreamreader_reset, |
| 1440 | METH_NOARGS, NULL}, |
| 1441 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1442 | }; |
| 1443 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1444 | static PyMemberDef mbstreamreader_members[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1445 | {"stream", T_OBJECT, |
| 1446 | offsetof(MultibyteStreamReaderObject, stream), |
| 1447 | READONLY, NULL}, |
| 1448 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1449 | }; |
| 1450 | |
| 1451 | static PyObject * |
| 1452 | mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1453 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1454 | MultibyteStreamReaderObject *self; |
| 1455 | PyObject *stream, *codec = NULL; |
| 1456 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1457 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1458 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader", |
| 1459 | streamkwarglist, &stream, &errors)) |
| 1460 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1461 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1462 | self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0); |
| 1463 | if (self == NULL) |
| 1464 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1465 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1466 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1467 | if (codec == NULL) |
| 1468 | goto errorexit; |
| 1469 | if (!MultibyteCodec_Check(codec)) { |
| 1470 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1471 | goto errorexit; |
| 1472 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1473 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1474 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1475 | self->stream = stream; |
| 1476 | Py_INCREF(stream); |
| 1477 | self->pendingsize = 0; |
| 1478 | self->errors = internal_error_callback(errors); |
| 1479 | if (self->errors == NULL) |
| 1480 | goto errorexit; |
| 1481 | if (self->codec->decinit != NULL && |
| 1482 | self->codec->decinit(&self->state, self->codec->config) != 0) |
| 1483 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1484 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1485 | Py_DECREF(codec); |
| 1486 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1487 | |
| 1488 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1489 | Py_XDECREF(self); |
| 1490 | Py_XDECREF(codec); |
| 1491 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1492 | } |
| 1493 | |
| 1494 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1495 | mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1496 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1497 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1498 | } |
| 1499 | |
| 1500 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1501 | mbstreamreader_traverse(MultibyteStreamReaderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1502 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1503 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1504 | if (ERROR_ISCUSTOM(self->errors)) |
| 1505 | Py_VISIT(self->errors); |
| 1506 | Py_VISIT(self->stream); |
| 1507 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1508 | } |
| 1509 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1510 | static void |
| 1511 | mbstreamreader_dealloc(MultibyteStreamReaderObject *self) |
| 1512 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1513 | PyObject_GC_UnTrack(self); |
| 1514 | ERROR_DECREF(self->errors); |
| 1515 | Py_XDECREF(self->stream); |
| 1516 | Py_TYPE(self)->tp_free(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1517 | } |
| 1518 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1519 | static PyTypeObject MultibyteStreamReader_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1520 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1521 | "MultibyteStreamReader", /* tp_name */ |
| 1522 | sizeof(MultibyteStreamReaderObject), /* tp_basicsize */ |
| 1523 | 0, /* tp_itemsize */ |
| 1524 | /* methods */ |
| 1525 | (destructor)mbstreamreader_dealloc, /* tp_dealloc */ |
| 1526 | 0, /* tp_print */ |
| 1527 | 0, /* tp_getattr */ |
| 1528 | 0, /* tp_setattr */ |
| 1529 | 0, /* tp_reserved */ |
| 1530 | 0, /* tp_repr */ |
| 1531 | 0, /* tp_as_number */ |
| 1532 | 0, /* tp_as_sequence */ |
| 1533 | 0, /* tp_as_mapping */ |
| 1534 | 0, /* tp_hash */ |
| 1535 | 0, /* tp_call */ |
| 1536 | 0, /* tp_str */ |
| 1537 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1538 | 0, /* tp_setattro */ |
| 1539 | 0, /* tp_as_buffer */ |
| 1540 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1541 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1542 | 0, /* tp_doc */ |
| 1543 | (traverseproc)mbstreamreader_traverse, /* tp_traverse */ |
| 1544 | 0, /* tp_clear */ |
| 1545 | 0, /* tp_richcompare */ |
| 1546 | 0, /* tp_weaklistoffset */ |
| 1547 | 0, /* tp_iter */ |
| 1548 | 0, /* tp_iterext */ |
| 1549 | mbstreamreader_methods, /* tp_methods */ |
| 1550 | mbstreamreader_members, /* tp_members */ |
| 1551 | codecctx_getsets, /* tp_getset */ |
| 1552 | 0, /* tp_base */ |
| 1553 | 0, /* tp_dict */ |
| 1554 | 0, /* tp_descr_get */ |
| 1555 | 0, /* tp_descr_set */ |
| 1556 | 0, /* tp_dictoffset */ |
| 1557 | mbstreamreader_init, /* tp_init */ |
| 1558 | 0, /* tp_alloc */ |
| 1559 | mbstreamreader_new, /* tp_new */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1560 | }; |
| 1561 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1562 | |
| 1563 | /** |
| 1564 | * MultibyteStreamWriter object |
| 1565 | */ |
| 1566 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1567 | static int |
| 1568 | mbstreamwriter_iwrite(MultibyteStreamWriterObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1569 | PyObject *unistr) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1570 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1571 | PyObject *str, *wr; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1572 | _Py_IDENTIFIER(write); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1573 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1574 | str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0); |
| 1575 | if (str == NULL) |
| 1576 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1577 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1578 | wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1579 | Py_DECREF(str); |
| 1580 | if (wr == NULL) |
| 1581 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1582 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1583 | Py_DECREF(wr); |
| 1584 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1585 | } |
| 1586 | |
| 1587 | static PyObject * |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1588 | mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1589 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1590 | if (mbstreamwriter_iwrite(self, strobj)) |
| 1591 | return NULL; |
| 1592 | else |
| 1593 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1594 | } |
| 1595 | |
| 1596 | static PyObject * |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1597 | mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1598 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1599 | PyObject *strobj; |
| 1600 | int i, r; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1601 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1602 | if (!PySequence_Check(lines)) { |
| 1603 | PyErr_SetString(PyExc_TypeError, |
| 1604 | "arg must be a sequence object"); |
| 1605 | return NULL; |
| 1606 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1607 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1608 | for (i = 0; i < PySequence_Length(lines); i++) { |
| 1609 | /* length can be changed even within this loop */ |
| 1610 | strobj = PySequence_GetItem(lines, i); |
| 1611 | if (strobj == NULL) |
| 1612 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1613 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1614 | r = mbstreamwriter_iwrite(self, strobj); |
| 1615 | Py_DECREF(strobj); |
| 1616 | if (r == -1) |
| 1617 | return NULL; |
| 1618 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1619 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1620 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1621 | } |
| 1622 | |
| 1623 | static PyObject * |
| 1624 | mbstreamwriter_reset(MultibyteStreamWriterObject *self) |
| 1625 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1626 | PyObject *pwrt; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1627 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1628 | pwrt = multibytecodec_encode(self->codec, &self->state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 1629 | self->pending, NULL, self->errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1630 | MBENC_FLUSH | MBENC_RESET); |
| 1631 | /* some pending buffer can be truncated when UnicodeEncodeError is |
| 1632 | * raised on 'strict' mode. but, 'reset' method is designed to |
| 1633 | * reset the pending buffer or states so failed string sequence |
| 1634 | * ought to be missed */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 1635 | Py_CLEAR(self->pending); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1636 | if (pwrt == NULL) |
| 1637 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1638 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1639 | assert(PyBytes_Check(pwrt)); |
| 1640 | if (PyBytes_Size(pwrt) > 0) { |
| 1641 | PyObject *wr; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1642 | _Py_IDENTIFIER(write); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1643 | |
| 1644 | wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1645 | if (wr == NULL) { |
| 1646 | Py_DECREF(pwrt); |
| 1647 | return NULL; |
| 1648 | } |
| 1649 | } |
| 1650 | Py_DECREF(pwrt); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1651 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1652 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1653 | } |
| 1654 | |
| 1655 | static PyObject * |
| 1656 | mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1657 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1658 | MultibyteStreamWriterObject *self; |
| 1659 | PyObject *stream, *codec = NULL; |
| 1660 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1661 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1662 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter", |
| 1663 | streamkwarglist, &stream, &errors)) |
| 1664 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1665 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1666 | self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0); |
| 1667 | if (self == NULL) |
| 1668 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1669 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1670 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1671 | if (codec == NULL) |
| 1672 | goto errorexit; |
| 1673 | if (!MultibyteCodec_Check(codec)) { |
| 1674 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1675 | goto errorexit; |
| 1676 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1677 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1678 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1679 | self->stream = stream; |
| 1680 | Py_INCREF(stream); |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 1681 | self->pending = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1682 | self->errors = internal_error_callback(errors); |
| 1683 | if (self->errors == NULL) |
| 1684 | goto errorexit; |
| 1685 | if (self->codec->encinit != NULL && |
| 1686 | self->codec->encinit(&self->state, self->codec->config) != 0) |
| 1687 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1688 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1689 | Py_DECREF(codec); |
| 1690 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1691 | |
| 1692 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1693 | Py_XDECREF(self); |
| 1694 | Py_XDECREF(codec); |
| 1695 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1696 | } |
| 1697 | |
| 1698 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1699 | mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1700 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1701 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1702 | } |
| 1703 | |
| 1704 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1705 | mbstreamwriter_traverse(MultibyteStreamWriterObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1706 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1707 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1708 | if (ERROR_ISCUSTOM(self->errors)) |
| 1709 | Py_VISIT(self->errors); |
| 1710 | Py_VISIT(self->stream); |
| 1711 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1712 | } |
| 1713 | |
| 1714 | static void |
| 1715 | mbstreamwriter_dealloc(MultibyteStreamWriterObject *self) |
| 1716 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1717 | PyObject_GC_UnTrack(self); |
| 1718 | ERROR_DECREF(self->errors); |
| 1719 | Py_XDECREF(self->stream); |
| 1720 | Py_TYPE(self)->tp_free(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1721 | } |
| 1722 | |
| 1723 | static struct PyMethodDef mbstreamwriter_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1724 | {"write", (PyCFunction)mbstreamwriter_write, |
| 1725 | METH_O, NULL}, |
| 1726 | {"writelines", (PyCFunction)mbstreamwriter_writelines, |
| 1727 | METH_O, NULL}, |
| 1728 | {"reset", (PyCFunction)mbstreamwriter_reset, |
| 1729 | METH_NOARGS, NULL}, |
| 1730 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1731 | }; |
| 1732 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1733 | static PyMemberDef mbstreamwriter_members[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1734 | {"stream", T_OBJECT, |
| 1735 | offsetof(MultibyteStreamWriterObject, stream), |
| 1736 | READONLY, NULL}, |
| 1737 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1738 | }; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1739 | |
| 1740 | static PyTypeObject MultibyteStreamWriter_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1741 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1742 | "MultibyteStreamWriter", /* tp_name */ |
| 1743 | sizeof(MultibyteStreamWriterObject), /* tp_basicsize */ |
| 1744 | 0, /* tp_itemsize */ |
| 1745 | /* methods */ |
| 1746 | (destructor)mbstreamwriter_dealloc, /* tp_dealloc */ |
| 1747 | 0, /* tp_print */ |
| 1748 | 0, /* tp_getattr */ |
| 1749 | 0, /* tp_setattr */ |
| 1750 | 0, /* tp_reserved */ |
| 1751 | 0, /* tp_repr */ |
| 1752 | 0, /* tp_as_number */ |
| 1753 | 0, /* tp_as_sequence */ |
| 1754 | 0, /* tp_as_mapping */ |
| 1755 | 0, /* tp_hash */ |
| 1756 | 0, /* tp_call */ |
| 1757 | 0, /* tp_str */ |
| 1758 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1759 | 0, /* tp_setattro */ |
| 1760 | 0, /* tp_as_buffer */ |
| 1761 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1762 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1763 | 0, /* tp_doc */ |
| 1764 | (traverseproc)mbstreamwriter_traverse, /* tp_traverse */ |
| 1765 | 0, /* tp_clear */ |
| 1766 | 0, /* tp_richcompare */ |
| 1767 | 0, /* tp_weaklistoffset */ |
| 1768 | 0, /* tp_iter */ |
| 1769 | 0, /* tp_iterext */ |
| 1770 | mbstreamwriter_methods, /* tp_methods */ |
| 1771 | mbstreamwriter_members, /* tp_members */ |
| 1772 | codecctx_getsets, /* tp_getset */ |
| 1773 | 0, /* tp_base */ |
| 1774 | 0, /* tp_dict */ |
| 1775 | 0, /* tp_descr_get */ |
| 1776 | 0, /* tp_descr_set */ |
| 1777 | 0, /* tp_dictoffset */ |
| 1778 | mbstreamwriter_init, /* tp_init */ |
| 1779 | 0, /* tp_alloc */ |
| 1780 | mbstreamwriter_new, /* tp_new */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1781 | }; |
| 1782 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1783 | |
| 1784 | /** |
| 1785 | * Exposed factory function |
| 1786 | */ |
| 1787 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1788 | static PyObject * |
| 1789 | __create_codec(PyObject *ignore, PyObject *arg) |
| 1790 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1791 | MultibyteCodecObject *self; |
| 1792 | MultibyteCodec *codec; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1793 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1794 | if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) { |
| 1795 | PyErr_SetString(PyExc_ValueError, "argument type invalid"); |
| 1796 | return NULL; |
| 1797 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1798 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1799 | codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME); |
| 1800 | if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0) |
| 1801 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1802 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1803 | self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type); |
| 1804 | if (self == NULL) |
| 1805 | return NULL; |
| 1806 | self->codec = codec; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1807 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1808 | return (PyObject *)self; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1809 | } |
| 1810 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1811 | static struct PyMethodDef __methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1812 | {"__create_codec", (PyCFunction)__create_codec, METH_O}, |
| 1813 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1814 | }; |
| 1815 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1816 | |
| 1817 | static struct PyModuleDef _multibytecodecmodule = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1818 | PyModuleDef_HEAD_INIT, |
| 1819 | "_multibytecodec", |
| 1820 | NULL, |
| 1821 | -1, |
| 1822 | __methods, |
| 1823 | NULL, |
| 1824 | NULL, |
| 1825 | NULL, |
| 1826 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1827 | }; |
| 1828 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1829 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1830 | PyInit__multibytecodec(void) |
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 | int i; |
| 1833 | PyObject *m; |
| 1834 | PyTypeObject *typelist[] = { |
| 1835 | &MultibyteIncrementalEncoder_Type, |
| 1836 | &MultibyteIncrementalDecoder_Type, |
| 1837 | &MultibyteStreamReader_Type, |
| 1838 | &MultibyteStreamWriter_Type, |
| 1839 | NULL |
| 1840 | }; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1841 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1842 | if (PyType_Ready(&MultibyteCodec_Type) < 0) |
| 1843 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1844 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1845 | m = PyModule_Create(&_multibytecodecmodule); |
| 1846 | if (m == NULL) |
| 1847 | return NULL; |
Neal Norwitz | 058bde1 | 2005-09-21 06:44:25 +0000 | [diff] [blame] | 1848 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1849 | for (i = 0; typelist[i] != NULL; i++) { |
| 1850 | if (PyType_Ready(typelist[i]) < 0) |
| 1851 | return NULL; |
| 1852 | Py_INCREF(typelist[i]); |
| 1853 | PyModule_AddObject(m, typelist[i]->tp_name, |
| 1854 | (PyObject *)typelist[i]); |
| 1855 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1856 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1857 | if (PyErr_Occurred()) { |
| 1858 | Py_FatalError("can't initialize the _multibytecodec module"); |
| 1859 | Py_DECREF(m); |
| 1860 | m = NULL; |
| 1861 | } |
| 1862 | return m; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1863 | } |