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; |
| 387 | default: |
| 388 | PyErr_SetString(PyExc_RuntimeError, |
| 389 | "unknown runtime error"); |
| 390 | return -1; |
| 391 | } |
| 392 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 393 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 394 | if (errors == ERROR_REPLACE) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 395 | if (_PyUnicodeWriter_WriteChar(&buf->writer, |
| 396 | Py_UNICODE_REPLACEMENT_CHARACTER) < 0) |
| 397 | goto errorexit; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 398 | } |
| 399 | if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { |
| 400 | buf->inbuf += esize; |
| 401 | return 0; |
| 402 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 403 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 404 | start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top); |
| 405 | end = start + esize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 406 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 407 | /* use cached exception object if available */ |
| 408 | if (buf->excobj == NULL) { |
| 409 | buf->excobj = PyUnicodeDecodeError_Create(codec->encoding, |
| 410 | (const char *)buf->inbuf_top, |
| 411 | (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top), |
| 412 | start, end, reason); |
| 413 | if (buf->excobj == NULL) |
| 414 | goto errorexit; |
| 415 | } |
| 416 | else |
| 417 | if (PyUnicodeDecodeError_SetStart(buf->excobj, start) || |
| 418 | PyUnicodeDecodeError_SetEnd(buf->excobj, end) || |
| 419 | PyUnicodeDecodeError_SetReason(buf->excobj, reason)) |
| 420 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 421 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 422 | if (errors == ERROR_STRICT) { |
| 423 | PyCodec_StrictErrors(buf->excobj); |
| 424 | goto errorexit; |
| 425 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 426 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 427 | retobj = call_error_callback(errors, buf->excobj); |
| 428 | if (retobj == NULL) |
| 429 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 430 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 431 | if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || |
| 432 | !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) || |
| 433 | !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { |
| 434 | PyErr_SetString(PyExc_TypeError, |
| 435 | "decoding error handler must return " |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 436 | "(str, int) tuple"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 437 | goto errorexit; |
| 438 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 439 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 440 | if (_PyUnicodeWriter_WriteStr(&buf->writer, retuni) < 0) |
Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 441 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 442 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 443 | newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); |
| 444 | if (newpos < 0 && !PyErr_Occurred()) |
| 445 | newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top); |
| 446 | if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) { |
| 447 | PyErr_Clear(); |
| 448 | PyErr_Format(PyExc_IndexError, |
| 449 | "position %zd from error handler out of bounds", |
| 450 | newpos); |
| 451 | goto errorexit; |
| 452 | } |
| 453 | buf->inbuf = buf->inbuf_top + newpos; |
| 454 | Py_DECREF(retobj); |
| 455 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 456 | |
| 457 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 458 | Py_XDECREF(retobj); |
| 459 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 460 | } |
| 461 | |
| 462 | static PyObject * |
| 463 | multibytecodec_encode(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 464 | MultibyteCodec_State *state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 465 | PyObject *text, Py_ssize_t *inpos_t, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 466 | PyObject *errors, int flags) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 467 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 468 | MultibyteEncodeBuffer buf; |
| 469 | Py_ssize_t finalsize, r = 0; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 470 | Py_ssize_t datalen; |
| 471 | int kind; |
| 472 | void *data; |
| 473 | |
| 474 | if (PyUnicode_READY(text) < 0) |
| 475 | return NULL; |
| 476 | datalen = PyUnicode_GET_LENGTH(text); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 477 | |
Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 478 | if (datalen == 0 && !(flags & MBENC_RESET)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 479 | return PyBytes_FromStringAndSize(NULL, 0); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 480 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 481 | buf.excobj = NULL; |
Brett Cannon | b94767f | 2011-02-22 20:15:44 +0000 | [diff] [blame] | 482 | buf.outobj = NULL; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 483 | buf.inobj = text; /* borrowed reference */ |
| 484 | buf.inpos = 0; |
| 485 | buf.inlen = datalen; |
| 486 | kind = PyUnicode_KIND(buf.inobj); |
| 487 | data = PyUnicode_DATA(buf.inobj); |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 488 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 489 | if (datalen > (PY_SSIZE_T_MAX - 16) / 2) { |
| 490 | PyErr_NoMemory(); |
| 491 | goto errorexit; |
| 492 | } |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 493 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 494 | buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16); |
| 495 | if (buf.outobj == NULL) |
| 496 | goto errorexit; |
| 497 | buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj); |
| 498 | buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 499 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 500 | while (buf.inpos < buf.inlen) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 501 | /* we don't reuse inleft and outleft here. |
| 502 | * error callbacks can relocate the cursor anywhere on buffer*/ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 503 | Py_ssize_t outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
| 504 | |
| 505 | r = codec->encode(state, codec->config, |
| 506 | kind, data, |
| 507 | &buf.inpos, buf.inlen, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 508 | &buf.outbuf, outleft, flags); |
| 509 | if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH))) |
| 510 | break; |
| 511 | else if (multibytecodec_encerror(codec, state, &buf, errors,r)) |
| 512 | goto errorexit; |
| 513 | else if (r == MBERR_TOOFEW) |
| 514 | break; |
| 515 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 516 | |
Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 517 | if (codec->encreset != NULL && (flags & MBENC_RESET)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 518 | for (;;) { |
| 519 | Py_ssize_t outleft; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 520 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 521 | outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
| 522 | r = codec->encreset(state, codec->config, &buf.outbuf, |
| 523 | outleft); |
| 524 | if (r == 0) |
| 525 | break; |
| 526 | else if (multibytecodec_encerror(codec, state, |
| 527 | &buf, errors, r)) |
| 528 | goto errorexit; |
| 529 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 530 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 531 | finalsize = (Py_ssize_t)((char *)buf.outbuf - |
| 532 | PyBytes_AS_STRING(buf.outobj)); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 533 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 534 | if (finalsize != PyBytes_GET_SIZE(buf.outobj)) |
| 535 | if (_PyBytes_Resize(&buf.outobj, finalsize) == -1) |
| 536 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 537 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 538 | if (inpos_t) |
| 539 | *inpos_t = buf.inpos; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 540 | Py_XDECREF(buf.excobj); |
| 541 | return buf.outobj; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 542 | |
| 543 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 544 | Py_XDECREF(buf.excobj); |
| 545 | Py_XDECREF(buf.outobj); |
| 546 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 547 | } |
| 548 | |
| 549 | static PyObject * |
| 550 | MultibyteCodec_Encode(MultibyteCodecObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 551 | PyObject *args, PyObject *kwargs) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 552 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 553 | MultibyteCodec_State state; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 554 | PyObject *errorcb, *r, *arg, *ucvt; |
| 555 | const char *errors = NULL; |
| 556 | Py_ssize_t datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 557 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 558 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode", |
| 559 | codeckwarglist, &arg, &errors)) |
| 560 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 561 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 562 | if (PyUnicode_Check(arg)) |
| 563 | ucvt = NULL; |
| 564 | else { |
| 565 | arg = ucvt = PyObject_Str(arg); |
| 566 | if (arg == NULL) |
| 567 | return NULL; |
| 568 | else if (!PyUnicode_Check(arg)) { |
| 569 | PyErr_SetString(PyExc_TypeError, |
| 570 | "couldn't convert the object to unicode."); |
| 571 | Py_DECREF(ucvt); |
| 572 | return NULL; |
| 573 | } |
| 574 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 575 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 576 | if (PyUnicode_READY(arg) < 0) { |
Victor Stinner | 9a80fab | 2011-11-21 02:50:14 +0100 | [diff] [blame] | 577 | Py_XDECREF(ucvt); |
| 578 | return NULL; |
| 579 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 580 | datalen = PyUnicode_GET_LENGTH(arg); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 581 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 582 | errorcb = internal_error_callback(errors); |
| 583 | if (errorcb == NULL) { |
| 584 | Py_XDECREF(ucvt); |
| 585 | return NULL; |
| 586 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 587 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 588 | if (self->codec->encinit != NULL && |
| 589 | self->codec->encinit(&state, self->codec->config) != 0) |
| 590 | goto errorexit; |
| 591 | r = multibytecodec_encode(self->codec, &state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 592 | arg, NULL, errorcb, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 593 | MBENC_FLUSH | MBENC_RESET); |
| 594 | if (r == NULL) |
| 595 | goto errorexit; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 596 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 597 | ERROR_DECREF(errorcb); |
| 598 | Py_XDECREF(ucvt); |
| 599 | return make_tuple(r, datalen); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 600 | |
| 601 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 602 | ERROR_DECREF(errorcb); |
| 603 | Py_XDECREF(ucvt); |
| 604 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 605 | } |
| 606 | |
| 607 | static PyObject * |
| 608 | MultibyteCodec_Decode(MultibyteCodecObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 609 | PyObject *args, PyObject *kwargs) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 610 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 611 | MultibyteCodec_State state; |
| 612 | MultibyteDecodeBuffer buf; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 613 | PyObject *errorcb, *res; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 614 | Py_buffer pdata; |
| 615 | const char *data, *errors = NULL; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 616 | Py_ssize_t datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 617 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 618 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|z:decode", |
| 619 | codeckwarglist, &pdata, &errors)) |
| 620 | return NULL; |
| 621 | data = pdata.buf; |
| 622 | datalen = pdata.len; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 623 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 624 | errorcb = internal_error_callback(errors); |
| 625 | if (errorcb == NULL) { |
| 626 | PyBuffer_Release(&pdata); |
| 627 | return NULL; |
| 628 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 629 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 630 | if (datalen == 0) { |
| 631 | PyBuffer_Release(&pdata); |
| 632 | ERROR_DECREF(errorcb); |
Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 633 | return make_tuple(PyUnicode_New(0, 0), 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 634 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 635 | |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 636 | _PyUnicodeWriter_Init(&buf.writer); |
| 637 | buf.writer.min_length = datalen; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 638 | buf.excobj = NULL; |
| 639 | buf.inbuf = buf.inbuf_top = (unsigned char *)data; |
| 640 | buf.inbuf_end = buf.inbuf_top + datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 641 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 642 | if (self->codec->decinit != NULL && |
| 643 | self->codec->decinit(&state, self->codec->config) != 0) |
| 644 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 645 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 646 | while (buf.inbuf < buf.inbuf_end) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 647 | Py_ssize_t inleft, r; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 648 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 649 | inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf); |
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 | r = self->codec->decode(&state, self->codec->config, |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 652 | &buf.inbuf, inleft, &buf.writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 653 | if (r == 0) |
| 654 | break; |
| 655 | else if (multibytecodec_decerror(self->codec, &state, |
| 656 | &buf, errorcb, r)) |
| 657 | goto errorexit; |
| 658 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 659 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 660 | res = _PyUnicodeWriter_Finish(&buf.writer); |
| 661 | if (res == NULL) |
| 662 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 663 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 664 | PyBuffer_Release(&pdata); |
| 665 | Py_XDECREF(buf.excobj); |
| 666 | ERROR_DECREF(errorcb); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 667 | return make_tuple(res, datalen); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 668 | |
| 669 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 670 | PyBuffer_Release(&pdata); |
| 671 | ERROR_DECREF(errorcb); |
| 672 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 673 | _PyUnicodeWriter_Dealloc(&buf.writer); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 674 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 675 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 676 | } |
| 677 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 678 | static struct PyMethodDef multibytecodec_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 679 | {"encode", (PyCFunction)MultibyteCodec_Encode, |
| 680 | METH_VARARGS | METH_KEYWORDS, |
| 681 | MultibyteCodec_Encode__doc__}, |
| 682 | {"decode", (PyCFunction)MultibyteCodec_Decode, |
| 683 | METH_VARARGS | METH_KEYWORDS, |
| 684 | MultibyteCodec_Decode__doc__}, |
| 685 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 686 | }; |
| 687 | |
| 688 | static void |
| 689 | multibytecodec_dealloc(MultibyteCodecObject *self) |
| 690 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 691 | PyObject_Del(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 692 | } |
| 693 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 694 | static PyTypeObject MultibyteCodec_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 695 | PyVarObject_HEAD_INIT(NULL, 0) |
| 696 | "MultibyteCodec", /* tp_name */ |
| 697 | sizeof(MultibyteCodecObject), /* tp_basicsize */ |
| 698 | 0, /* tp_itemsize */ |
| 699 | /* methods */ |
| 700 | (destructor)multibytecodec_dealloc, /* tp_dealloc */ |
| 701 | 0, /* tp_print */ |
| 702 | 0, /* tp_getattr */ |
| 703 | 0, /* tp_setattr */ |
| 704 | 0, /* tp_reserved */ |
| 705 | 0, /* tp_repr */ |
| 706 | 0, /* tp_as_number */ |
| 707 | 0, /* tp_as_sequence */ |
| 708 | 0, /* tp_as_mapping */ |
| 709 | 0, /* tp_hash */ |
| 710 | 0, /* tp_call */ |
| 711 | 0, /* tp_str */ |
| 712 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 713 | 0, /* tp_setattro */ |
| 714 | 0, /* tp_as_buffer */ |
| 715 | Py_TPFLAGS_DEFAULT, /* tp_flags */ |
| 716 | 0, /* tp_doc */ |
| 717 | 0, /* tp_traverse */ |
| 718 | 0, /* tp_clear */ |
| 719 | 0, /* tp_richcompare */ |
| 720 | 0, /* tp_weaklistoffset */ |
| 721 | 0, /* tp_iter */ |
| 722 | 0, /* tp_iterext */ |
| 723 | multibytecodec_methods, /* tp_methods */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 724 | }; |
| 725 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 726 | |
| 727 | /** |
| 728 | * Utility functions for stateful codec mechanism |
| 729 | */ |
| 730 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 731 | #define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o)) |
| 732 | #define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 733 | |
| 734 | static PyObject * |
| 735 | encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 736 | PyObject *unistr, int final) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 737 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 738 | PyObject *ucvt, *r = NULL; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 739 | PyObject *inbuf = NULL; |
| 740 | Py_ssize_t inpos, datalen; |
| 741 | PyObject *origpending = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 742 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 743 | if (PyUnicode_Check(unistr)) |
| 744 | ucvt = NULL; |
| 745 | else { |
| 746 | unistr = ucvt = PyObject_Str(unistr); |
| 747 | if (unistr == NULL) |
| 748 | return NULL; |
| 749 | else if (!PyUnicode_Check(unistr)) { |
| 750 | PyErr_SetString(PyExc_TypeError, |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 751 | "couldn't convert the object to str."); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 752 | Py_DECREF(ucvt); |
| 753 | return NULL; |
| 754 | } |
| 755 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 756 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 757 | if (ctx->pending) { |
| 758 | PyObject *inbuf_tmp; |
| 759 | |
| 760 | Py_INCREF(ctx->pending); |
| 761 | origpending = ctx->pending; |
| 762 | |
| 763 | Py_INCREF(ctx->pending); |
| 764 | inbuf_tmp = ctx->pending; |
| 765 | PyUnicode_Append(&inbuf_tmp, unistr); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 766 | if (inbuf_tmp == NULL) |
| 767 | goto errorexit; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 768 | Py_CLEAR(ctx->pending); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 769 | inbuf = inbuf_tmp; |
| 770 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 771 | else { |
| 772 | origpending = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 773 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 774 | Py_INCREF(unistr); |
| 775 | inbuf = unistr; |
| 776 | } |
| 777 | if (PyUnicode_READY(inbuf) < 0) |
| 778 | goto errorexit; |
| 779 | inpos = 0; |
| 780 | datalen = PyUnicode_GET_LENGTH(inbuf); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 781 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 782 | r = multibytecodec_encode(ctx->codec, &ctx->state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 783 | inbuf, &inpos, |
| 784 | ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 785 | if (r == NULL) { |
| 786 | /* recover the original pending buffer */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 787 | Py_CLEAR(ctx->pending); |
| 788 | ctx->pending = origpending; |
| 789 | origpending = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 790 | goto errorexit; |
| 791 | } |
Victor Stinner | 322cc74 | 2013-04-14 18:11:41 +0200 | [diff] [blame] | 792 | Py_XDECREF(origpending); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 793 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 794 | if (inpos < datalen) { |
| 795 | if (datalen - inpos > MAXENCPENDING) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 796 | /* normal codecs can't reach here */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 797 | PyErr_SetString(PyExc_UnicodeError, |
| 798 | "pending buffer overflow"); |
| 799 | goto errorexit; |
| 800 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 801 | ctx->pending = PyUnicode_Substring(inbuf, inpos, datalen); |
| 802 | if (ctx->pending == NULL) { |
| 803 | /* normal codecs can't reach here */ |
| 804 | goto errorexit; |
| 805 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 806 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 807 | |
Victor Stinner | 322cc74 | 2013-04-14 18:11:41 +0200 | [diff] [blame] | 808 | Py_DECREF(inbuf); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 809 | Py_XDECREF(ucvt); |
| 810 | return r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 811 | |
| 812 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 813 | Py_XDECREF(r); |
| 814 | Py_XDECREF(ucvt); |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 815 | Py_XDECREF(origpending); |
Victor Stinner | 322cc74 | 2013-04-14 18:11:41 +0200 | [diff] [blame] | 816 | Py_XDECREF(inbuf); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 817 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 818 | } |
| 819 | |
| 820 | static int |
| 821 | decoder_append_pending(MultibyteStatefulDecoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 822 | MultibyteDecodeBuffer *buf) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 823 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 824 | Py_ssize_t npendings; |
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 | npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
| 827 | if (npendings + ctx->pendingsize > MAXDECPENDING || |
| 828 | npendings > PY_SSIZE_T_MAX - ctx->pendingsize) { |
| 829 | PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow"); |
| 830 | return -1; |
| 831 | } |
| 832 | memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings); |
| 833 | ctx->pendingsize += npendings; |
| 834 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 835 | } |
| 836 | |
| 837 | static int |
| 838 | decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 839 | Py_ssize_t size) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 840 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 841 | buf->inbuf = buf->inbuf_top = (const unsigned char *)data; |
| 842 | buf->inbuf_end = buf->inbuf_top + size; |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 843 | buf->writer.min_length += size; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 844 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 845 | } |
| 846 | |
| 847 | static int |
| 848 | decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 849 | MultibyteDecodeBuffer *buf) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 850 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 851 | while (buf->inbuf < buf->inbuf_end) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 852 | Py_ssize_t inleft; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 853 | Py_ssize_t r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 854 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 855 | inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
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 | r = ctx->codec->decode(&ctx->state, ctx->codec->config, |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 858 | &buf->inbuf, inleft, &buf->writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 859 | if (r == 0 || r == MBERR_TOOFEW) |
| 860 | break; |
| 861 | else if (multibytecodec_decerror(ctx->codec, &ctx->state, |
| 862 | buf, ctx->errors, r)) |
| 863 | return -1; |
| 864 | } |
| 865 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 866 | } |
| 867 | |
| 868 | |
| 869 | /** |
| 870 | * MultibyteIncrementalEncoder object |
| 871 | */ |
| 872 | |
| 873 | static PyObject * |
| 874 | mbiencoder_encode(MultibyteIncrementalEncoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 875 | PyObject *args, PyObject *kwargs) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 876 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 877 | PyObject *data; |
| 878 | int final = 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 879 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 880 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode", |
| 881 | incrementalkwarglist, &data, &final)) |
| 882 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 883 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 884 | return encoder_encode_stateful(STATEFUL_ECTX(self), data, final); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 885 | } |
| 886 | |
| 887 | static PyObject * |
| 888 | mbiencoder_reset(MultibyteIncrementalEncoderObject *self) |
| 889 | { |
Victor Stinner | e15dce3 | 2011-05-30 22:56:00 +0200 | [diff] [blame] | 890 | /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */ |
| 891 | unsigned char buffer[4], *outbuf; |
| 892 | Py_ssize_t r; |
| 893 | if (self->codec->encreset != NULL) { |
| 894 | outbuf = buffer; |
| 895 | r = self->codec->encreset(&self->state, self->codec->config, |
| 896 | &outbuf, sizeof(buffer)); |
| 897 | if (r != 0) |
| 898 | return NULL; |
| 899 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 900 | Py_CLEAR(self->pending); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 901 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 902 | } |
| 903 | |
| 904 | static struct PyMethodDef mbiencoder_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 905 | {"encode", (PyCFunction)mbiencoder_encode, |
| 906 | METH_VARARGS | METH_KEYWORDS, NULL}, |
| 907 | {"reset", (PyCFunction)mbiencoder_reset, |
| 908 | METH_NOARGS, NULL}, |
| 909 | {NULL, NULL}, |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 910 | }; |
| 911 | |
| 912 | static PyObject * |
| 913 | mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 914 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 915 | MultibyteIncrementalEncoderObject *self; |
| 916 | PyObject *codec = NULL; |
| 917 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 918 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 919 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder", |
| 920 | incnewkwarglist, &errors)) |
| 921 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 922 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 923 | self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0); |
| 924 | if (self == NULL) |
| 925 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 926 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 927 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 928 | if (codec == NULL) |
| 929 | goto errorexit; |
| 930 | if (!MultibyteCodec_Check(codec)) { |
| 931 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 932 | goto errorexit; |
| 933 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 934 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 935 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 936 | self->pending = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 937 | self->errors = internal_error_callback(errors); |
| 938 | if (self->errors == NULL) |
| 939 | goto errorexit; |
| 940 | if (self->codec->encinit != NULL && |
| 941 | self->codec->encinit(&self->state, self->codec->config) != 0) |
| 942 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 943 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 944 | Py_DECREF(codec); |
| 945 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 946 | |
| 947 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 948 | Py_XDECREF(self); |
| 949 | Py_XDECREF(codec); |
| 950 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 951 | } |
| 952 | |
| 953 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 954 | mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 955 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 956 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 957 | } |
| 958 | |
| 959 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 960 | mbiencoder_traverse(MultibyteIncrementalEncoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 961 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 962 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 963 | if (ERROR_ISCUSTOM(self->errors)) |
| 964 | Py_VISIT(self->errors); |
| 965 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 966 | } |
| 967 | |
| 968 | static void |
| 969 | mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self) |
| 970 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 971 | PyObject_GC_UnTrack(self); |
| 972 | ERROR_DECREF(self->errors); |
| 973 | Py_TYPE(self)->tp_free(self); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 974 | } |
| 975 | |
| 976 | static PyTypeObject MultibyteIncrementalEncoder_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 977 | PyVarObject_HEAD_INIT(NULL, 0) |
| 978 | "MultibyteIncrementalEncoder", /* tp_name */ |
| 979 | sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */ |
| 980 | 0, /* tp_itemsize */ |
| 981 | /* methods */ |
| 982 | (destructor)mbiencoder_dealloc, /* tp_dealloc */ |
| 983 | 0, /* tp_print */ |
| 984 | 0, /* tp_getattr */ |
| 985 | 0, /* tp_setattr */ |
| 986 | 0, /* tp_reserved */ |
| 987 | 0, /* tp_repr */ |
| 988 | 0, /* tp_as_number */ |
| 989 | 0, /* tp_as_sequence */ |
| 990 | 0, /* tp_as_mapping */ |
| 991 | 0, /* tp_hash */ |
| 992 | 0, /* tp_call */ |
| 993 | 0, /* tp_str */ |
| 994 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 995 | 0, /* tp_setattro */ |
| 996 | 0, /* tp_as_buffer */ |
| 997 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 998 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 999 | 0, /* tp_doc */ |
| 1000 | (traverseproc)mbiencoder_traverse, /* tp_traverse */ |
| 1001 | 0, /* tp_clear */ |
| 1002 | 0, /* tp_richcompare */ |
| 1003 | 0, /* tp_weaklistoffset */ |
| 1004 | 0, /* tp_iter */ |
| 1005 | 0, /* tp_iterext */ |
| 1006 | mbiencoder_methods, /* tp_methods */ |
| 1007 | 0, /* tp_members */ |
| 1008 | codecctx_getsets, /* tp_getset */ |
| 1009 | 0, /* tp_base */ |
| 1010 | 0, /* tp_dict */ |
| 1011 | 0, /* tp_descr_get */ |
| 1012 | 0, /* tp_descr_set */ |
| 1013 | 0, /* tp_dictoffset */ |
| 1014 | mbiencoder_init, /* tp_init */ |
| 1015 | 0, /* tp_alloc */ |
| 1016 | mbiencoder_new, /* tp_new */ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1017 | }; |
| 1018 | |
| 1019 | |
| 1020 | /** |
| 1021 | * MultibyteIncrementalDecoder object |
| 1022 | */ |
| 1023 | |
| 1024 | static PyObject * |
| 1025 | mbidecoder_decode(MultibyteIncrementalDecoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1026 | PyObject *args, PyObject *kwargs) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1027 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1028 | MultibyteDecodeBuffer buf; |
| 1029 | char *data, *wdata = NULL; |
| 1030 | Py_buffer pdata; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1031 | Py_ssize_t wsize, size, origpending; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1032 | int final = 0; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1033 | PyObject *res; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1034 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1035 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|i:decode", |
| 1036 | incrementalkwarglist, &pdata, &final)) |
| 1037 | return NULL; |
| 1038 | data = pdata.buf; |
| 1039 | size = pdata.len; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1040 | |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 1041 | _PyUnicodeWriter_Init(&buf.writer); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1042 | buf.excobj = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1043 | origpending = self->pendingsize; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1044 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1045 | if (self->pendingsize == 0) { |
| 1046 | wsize = size; |
| 1047 | wdata = data; |
| 1048 | } |
| 1049 | else { |
| 1050 | if (size > PY_SSIZE_T_MAX - self->pendingsize) { |
| 1051 | PyErr_NoMemory(); |
| 1052 | goto errorexit; |
| 1053 | } |
| 1054 | wsize = size + self->pendingsize; |
| 1055 | wdata = PyMem_Malloc(wsize); |
Victor Stinner | 33283ba | 2013-07-15 17:47:39 +0200 | [diff] [blame] | 1056 | if (wdata == NULL) { |
| 1057 | PyErr_NoMemory(); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1058 | goto errorexit; |
Victor Stinner | 33283ba | 2013-07-15 17:47:39 +0200 | [diff] [blame] | 1059 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1060 | memcpy(wdata, self->pending, self->pendingsize); |
| 1061 | memcpy(wdata + self->pendingsize, data, size); |
| 1062 | self->pendingsize = 0; |
| 1063 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1064 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1065 | if (decoder_prepare_buffer(&buf, wdata, wsize) != 0) |
| 1066 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1067 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1068 | if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf)) |
| 1069 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1070 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1071 | if (final && buf.inbuf < buf.inbuf_end) { |
| 1072 | if (multibytecodec_decerror(self->codec, &self->state, |
| 1073 | &buf, self->errors, MBERR_TOOFEW)) { |
| 1074 | /* recover the original pending buffer */ |
| 1075 | memcpy(self->pending, wdata, origpending); |
| 1076 | self->pendingsize = origpending; |
| 1077 | goto errorexit; |
| 1078 | } |
| 1079 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1080 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1081 | if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */ |
| 1082 | if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0) |
| 1083 | goto errorexit; |
| 1084 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1085 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1086 | res = _PyUnicodeWriter_Finish(&buf.writer); |
| 1087 | if (res == NULL) |
| 1088 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1089 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1090 | PyBuffer_Release(&pdata); |
| 1091 | if (wdata != data) |
| 1092 | PyMem_Del(wdata); |
| 1093 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1094 | return res; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1095 | |
| 1096 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1097 | PyBuffer_Release(&pdata); |
| 1098 | if (wdata != NULL && wdata != data) |
| 1099 | PyMem_Del(wdata); |
| 1100 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1101 | _PyUnicodeWriter_Dealloc(&buf.writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1102 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1103 | } |
| 1104 | |
| 1105 | static PyObject * |
| 1106 | mbidecoder_reset(MultibyteIncrementalDecoderObject *self) |
| 1107 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1108 | if (self->codec->decreset != NULL && |
| 1109 | self->codec->decreset(&self->state, self->codec->config) != 0) |
| 1110 | return NULL; |
| 1111 | self->pendingsize = 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1112 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1113 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1114 | } |
| 1115 | |
| 1116 | static struct PyMethodDef mbidecoder_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1117 | {"decode", (PyCFunction)mbidecoder_decode, |
| 1118 | METH_VARARGS | METH_KEYWORDS, NULL}, |
| 1119 | {"reset", (PyCFunction)mbidecoder_reset, |
| 1120 | METH_NOARGS, NULL}, |
| 1121 | {NULL, NULL}, |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1122 | }; |
| 1123 | |
| 1124 | static PyObject * |
| 1125 | mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1126 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1127 | MultibyteIncrementalDecoderObject *self; |
| 1128 | PyObject *codec = NULL; |
| 1129 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1130 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1131 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder", |
| 1132 | incnewkwarglist, &errors)) |
| 1133 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1134 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1135 | self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0); |
| 1136 | if (self == NULL) |
| 1137 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1138 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1139 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1140 | if (codec == NULL) |
| 1141 | goto errorexit; |
| 1142 | if (!MultibyteCodec_Check(codec)) { |
| 1143 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1144 | goto errorexit; |
| 1145 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1146 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1147 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1148 | self->pendingsize = 0; |
| 1149 | self->errors = internal_error_callback(errors); |
| 1150 | if (self->errors == NULL) |
| 1151 | goto errorexit; |
| 1152 | if (self->codec->decinit != NULL && |
| 1153 | self->codec->decinit(&self->state, self->codec->config) != 0) |
| 1154 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1155 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1156 | Py_DECREF(codec); |
| 1157 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1158 | |
| 1159 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1160 | Py_XDECREF(self); |
| 1161 | Py_XDECREF(codec); |
| 1162 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1163 | } |
| 1164 | |
| 1165 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1166 | mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1167 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1168 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1169 | } |
| 1170 | |
| 1171 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1172 | mbidecoder_traverse(MultibyteIncrementalDecoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1173 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1174 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1175 | if (ERROR_ISCUSTOM(self->errors)) |
| 1176 | Py_VISIT(self->errors); |
| 1177 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1178 | } |
| 1179 | |
| 1180 | static void |
| 1181 | mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self) |
| 1182 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1183 | PyObject_GC_UnTrack(self); |
| 1184 | ERROR_DECREF(self->errors); |
| 1185 | Py_TYPE(self)->tp_free(self); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1186 | } |
| 1187 | |
| 1188 | static PyTypeObject MultibyteIncrementalDecoder_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1189 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1190 | "MultibyteIncrementalDecoder", /* tp_name */ |
| 1191 | sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */ |
| 1192 | 0, /* tp_itemsize */ |
| 1193 | /* methods */ |
| 1194 | (destructor)mbidecoder_dealloc, /* tp_dealloc */ |
| 1195 | 0, /* tp_print */ |
| 1196 | 0, /* tp_getattr */ |
| 1197 | 0, /* tp_setattr */ |
| 1198 | 0, /* tp_reserved */ |
| 1199 | 0, /* tp_repr */ |
| 1200 | 0, /* tp_as_number */ |
| 1201 | 0, /* tp_as_sequence */ |
| 1202 | 0, /* tp_as_mapping */ |
| 1203 | 0, /* tp_hash */ |
| 1204 | 0, /* tp_call */ |
| 1205 | 0, /* tp_str */ |
| 1206 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1207 | 0, /* tp_setattro */ |
| 1208 | 0, /* tp_as_buffer */ |
| 1209 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1210 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1211 | 0, /* tp_doc */ |
| 1212 | (traverseproc)mbidecoder_traverse, /* tp_traverse */ |
| 1213 | 0, /* tp_clear */ |
| 1214 | 0, /* tp_richcompare */ |
| 1215 | 0, /* tp_weaklistoffset */ |
| 1216 | 0, /* tp_iter */ |
| 1217 | 0, /* tp_iterext */ |
| 1218 | mbidecoder_methods, /* tp_methods */ |
| 1219 | 0, /* tp_members */ |
| 1220 | codecctx_getsets, /* tp_getset */ |
| 1221 | 0, /* tp_base */ |
| 1222 | 0, /* tp_dict */ |
| 1223 | 0, /* tp_descr_get */ |
| 1224 | 0, /* tp_descr_set */ |
| 1225 | 0, /* tp_dictoffset */ |
| 1226 | mbidecoder_init, /* tp_init */ |
| 1227 | 0, /* tp_alloc */ |
| 1228 | mbidecoder_new, /* tp_new */ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1229 | }; |
| 1230 | |
| 1231 | |
| 1232 | /** |
| 1233 | * MultibyteStreamReader object |
| 1234 | */ |
| 1235 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1236 | static PyObject * |
| 1237 | mbstreamreader_iread(MultibyteStreamReaderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1238 | const char *method, Py_ssize_t sizehint) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1239 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1240 | MultibyteDecodeBuffer buf; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1241 | PyObject *cres, *res; |
| 1242 | Py_ssize_t rsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1243 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1244 | if (sizehint == 0) |
Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 1245 | return PyUnicode_New(0, 0); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1246 | |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 1247 | _PyUnicodeWriter_Init(&buf.writer); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1248 | buf.excobj = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1249 | cres = NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1250 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1251 | for (;;) { |
| 1252 | int endoffile; |
Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1253 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1254 | if (sizehint < 0) |
| 1255 | cres = PyObject_CallMethod(self->stream, |
| 1256 | (char *)method, NULL); |
| 1257 | else |
| 1258 | cres = PyObject_CallMethod(self->stream, |
| 1259 | (char *)method, "i", sizehint); |
| 1260 | if (cres == NULL) |
| 1261 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1262 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1263 | if (!PyBytes_Check(cres)) { |
| 1264 | PyErr_Format(PyExc_TypeError, |
| 1265 | "stream function returned a " |
| 1266 | "non-bytes object (%.100s)", |
| 1267 | cres->ob_type->tp_name); |
| 1268 | goto errorexit; |
| 1269 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1270 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1271 | endoffile = (PyBytes_GET_SIZE(cres) == 0); |
Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1272 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1273 | if (self->pendingsize > 0) { |
| 1274 | PyObject *ctr; |
| 1275 | char *ctrdata; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1276 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1277 | if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) { |
| 1278 | PyErr_NoMemory(); |
| 1279 | goto errorexit; |
Victor Stinner | 064bbdc | 2013-07-08 22:28:27 +0200 | [diff] [blame] | 1280 | } |
| 1281 | rsize = PyBytes_GET_SIZE(cres) + self->pendingsize; |
| 1282 | ctr = PyBytes_FromStringAndSize(NULL, rsize); |
| 1283 | if (ctr == NULL) |
| 1284 | goto errorexit; |
| 1285 | ctrdata = PyBytes_AS_STRING(ctr); |
| 1286 | memcpy(ctrdata, self->pending, self->pendingsize); |
| 1287 | memcpy(ctrdata + self->pendingsize, |
| 1288 | PyBytes_AS_STRING(cres), |
| 1289 | PyBytes_GET_SIZE(cres)); |
| 1290 | Py_DECREF(cres); |
| 1291 | cres = ctr; |
| 1292 | self->pendingsize = 0; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1293 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1294 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1295 | rsize = PyBytes_GET_SIZE(cres); |
| 1296 | if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres), |
| 1297 | rsize) != 0) |
| 1298 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1299 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1300 | if (rsize > 0 && decoder_feed_buffer( |
| 1301 | (MultibyteStatefulDecoderContext *)self, &buf)) |
| 1302 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1303 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1304 | if (endoffile || sizehint < 0) { |
| 1305 | if (buf.inbuf < buf.inbuf_end && |
| 1306 | multibytecodec_decerror(self->codec, &self->state, |
| 1307 | &buf, self->errors, MBERR_TOOFEW)) |
| 1308 | goto errorexit; |
| 1309 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1310 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1311 | if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */ |
| 1312 | if (decoder_append_pending(STATEFUL_DCTX(self), |
| 1313 | &buf) != 0) |
| 1314 | goto errorexit; |
| 1315 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1316 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1317 | Py_DECREF(cres); |
| 1318 | cres = NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1319 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1320 | if (sizehint < 0 || buf.writer.pos != 0 || rsize == 0) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1321 | break; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1322 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1323 | sizehint = 1; /* read 1 more byte and retry */ |
| 1324 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1325 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1326 | res = _PyUnicodeWriter_Finish(&buf.writer); |
| 1327 | if (res == NULL) |
| 1328 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1329 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1330 | Py_XDECREF(cres); |
| 1331 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1332 | return res; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1333 | |
| 1334 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1335 | Py_XDECREF(cres); |
| 1336 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1337 | _PyUnicodeWriter_Dealloc(&buf.writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1338 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1339 | } |
| 1340 | |
| 1341 | static PyObject * |
| 1342 | mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args) |
| 1343 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1344 | PyObject *sizeobj = NULL; |
| 1345 | Py_ssize_t size; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1346 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1347 | if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj)) |
| 1348 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1349 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1350 | if (sizeobj == Py_None || sizeobj == NULL) |
| 1351 | size = -1; |
| 1352 | else if (PyLong_Check(sizeobj)) |
| 1353 | size = PyLong_AsSsize_t(sizeobj); |
| 1354 | else { |
| 1355 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1356 | return NULL; |
| 1357 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1358 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1359 | if (size == -1 && PyErr_Occurred()) |
| 1360 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1361 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1362 | return mbstreamreader_iread(self, "read", size); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1363 | } |
| 1364 | |
| 1365 | static PyObject * |
| 1366 | mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args) |
| 1367 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1368 | PyObject *sizeobj = NULL; |
| 1369 | Py_ssize_t size; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1370 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1371 | if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj)) |
| 1372 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1373 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1374 | if (sizeobj == Py_None || sizeobj == NULL) |
| 1375 | size = -1; |
| 1376 | else if (PyLong_Check(sizeobj)) |
| 1377 | size = PyLong_AsSsize_t(sizeobj); |
| 1378 | else { |
| 1379 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1380 | return NULL; |
| 1381 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1382 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1383 | if (size == -1 && PyErr_Occurred()) |
| 1384 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1385 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1386 | return mbstreamreader_iread(self, "readline", size); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1387 | } |
| 1388 | |
| 1389 | static PyObject * |
| 1390 | mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args) |
| 1391 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1392 | PyObject *sizehintobj = NULL, *r, *sr; |
| 1393 | Py_ssize_t sizehint; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1394 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1395 | if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj)) |
| 1396 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1397 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1398 | if (sizehintobj == Py_None || sizehintobj == NULL) |
| 1399 | sizehint = -1; |
| 1400 | else if (PyLong_Check(sizehintobj)) |
| 1401 | sizehint = PyLong_AsSsize_t(sizehintobj); |
| 1402 | else { |
| 1403 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1404 | return NULL; |
| 1405 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1406 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1407 | if (sizehint == -1 && PyErr_Occurred()) |
| 1408 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1409 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1410 | r = mbstreamreader_iread(self, "read", sizehint); |
| 1411 | if (r == NULL) |
| 1412 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1413 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1414 | sr = PyUnicode_Splitlines(r, 1); |
| 1415 | Py_DECREF(r); |
| 1416 | return sr; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1417 | } |
| 1418 | |
| 1419 | static PyObject * |
| 1420 | mbstreamreader_reset(MultibyteStreamReaderObject *self) |
| 1421 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1422 | if (self->codec->decreset != NULL && |
| 1423 | self->codec->decreset(&self->state, self->codec->config) != 0) |
| 1424 | return NULL; |
| 1425 | self->pendingsize = 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1426 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1427 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1428 | } |
| 1429 | |
| 1430 | static struct PyMethodDef mbstreamreader_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1431 | {"read", (PyCFunction)mbstreamreader_read, |
| 1432 | METH_VARARGS, NULL}, |
| 1433 | {"readline", (PyCFunction)mbstreamreader_readline, |
| 1434 | METH_VARARGS, NULL}, |
| 1435 | {"readlines", (PyCFunction)mbstreamreader_readlines, |
| 1436 | METH_VARARGS, NULL}, |
| 1437 | {"reset", (PyCFunction)mbstreamreader_reset, |
| 1438 | METH_NOARGS, NULL}, |
| 1439 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1440 | }; |
| 1441 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1442 | static PyMemberDef mbstreamreader_members[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1443 | {"stream", T_OBJECT, |
| 1444 | offsetof(MultibyteStreamReaderObject, stream), |
| 1445 | READONLY, NULL}, |
| 1446 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1447 | }; |
| 1448 | |
| 1449 | static PyObject * |
| 1450 | mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1451 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1452 | MultibyteStreamReaderObject *self; |
| 1453 | PyObject *stream, *codec = NULL; |
| 1454 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1455 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1456 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader", |
| 1457 | streamkwarglist, &stream, &errors)) |
| 1458 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1459 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1460 | self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0); |
| 1461 | if (self == NULL) |
| 1462 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1463 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1464 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1465 | if (codec == NULL) |
| 1466 | goto errorexit; |
| 1467 | if (!MultibyteCodec_Check(codec)) { |
| 1468 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1469 | goto errorexit; |
| 1470 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1471 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1472 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1473 | self->stream = stream; |
| 1474 | Py_INCREF(stream); |
| 1475 | self->pendingsize = 0; |
| 1476 | self->errors = internal_error_callback(errors); |
| 1477 | if (self->errors == NULL) |
| 1478 | goto errorexit; |
| 1479 | if (self->codec->decinit != NULL && |
| 1480 | self->codec->decinit(&self->state, self->codec->config) != 0) |
| 1481 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1482 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1483 | Py_DECREF(codec); |
| 1484 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1485 | |
| 1486 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1487 | Py_XDECREF(self); |
| 1488 | Py_XDECREF(codec); |
| 1489 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1490 | } |
| 1491 | |
| 1492 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1493 | mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1494 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1495 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1496 | } |
| 1497 | |
| 1498 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1499 | mbstreamreader_traverse(MultibyteStreamReaderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1500 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1501 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1502 | if (ERROR_ISCUSTOM(self->errors)) |
| 1503 | Py_VISIT(self->errors); |
| 1504 | Py_VISIT(self->stream); |
| 1505 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1506 | } |
| 1507 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1508 | static void |
| 1509 | mbstreamreader_dealloc(MultibyteStreamReaderObject *self) |
| 1510 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1511 | PyObject_GC_UnTrack(self); |
| 1512 | ERROR_DECREF(self->errors); |
| 1513 | Py_XDECREF(self->stream); |
| 1514 | Py_TYPE(self)->tp_free(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1515 | } |
| 1516 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1517 | static PyTypeObject MultibyteStreamReader_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1518 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1519 | "MultibyteStreamReader", /* tp_name */ |
| 1520 | sizeof(MultibyteStreamReaderObject), /* tp_basicsize */ |
| 1521 | 0, /* tp_itemsize */ |
| 1522 | /* methods */ |
| 1523 | (destructor)mbstreamreader_dealloc, /* tp_dealloc */ |
| 1524 | 0, /* tp_print */ |
| 1525 | 0, /* tp_getattr */ |
| 1526 | 0, /* tp_setattr */ |
| 1527 | 0, /* tp_reserved */ |
| 1528 | 0, /* tp_repr */ |
| 1529 | 0, /* tp_as_number */ |
| 1530 | 0, /* tp_as_sequence */ |
| 1531 | 0, /* tp_as_mapping */ |
| 1532 | 0, /* tp_hash */ |
| 1533 | 0, /* tp_call */ |
| 1534 | 0, /* tp_str */ |
| 1535 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1536 | 0, /* tp_setattro */ |
| 1537 | 0, /* tp_as_buffer */ |
| 1538 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1539 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1540 | 0, /* tp_doc */ |
| 1541 | (traverseproc)mbstreamreader_traverse, /* tp_traverse */ |
| 1542 | 0, /* tp_clear */ |
| 1543 | 0, /* tp_richcompare */ |
| 1544 | 0, /* tp_weaklistoffset */ |
| 1545 | 0, /* tp_iter */ |
| 1546 | 0, /* tp_iterext */ |
| 1547 | mbstreamreader_methods, /* tp_methods */ |
| 1548 | mbstreamreader_members, /* tp_members */ |
| 1549 | codecctx_getsets, /* tp_getset */ |
| 1550 | 0, /* tp_base */ |
| 1551 | 0, /* tp_dict */ |
| 1552 | 0, /* tp_descr_get */ |
| 1553 | 0, /* tp_descr_set */ |
| 1554 | 0, /* tp_dictoffset */ |
| 1555 | mbstreamreader_init, /* tp_init */ |
| 1556 | 0, /* tp_alloc */ |
| 1557 | mbstreamreader_new, /* tp_new */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1558 | }; |
| 1559 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1560 | |
| 1561 | /** |
| 1562 | * MultibyteStreamWriter object |
| 1563 | */ |
| 1564 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1565 | static int |
| 1566 | mbstreamwriter_iwrite(MultibyteStreamWriterObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1567 | PyObject *unistr) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1568 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1569 | PyObject *str, *wr; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1570 | _Py_IDENTIFIER(write); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1571 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1572 | str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0); |
| 1573 | if (str == NULL) |
| 1574 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1575 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1576 | wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1577 | Py_DECREF(str); |
| 1578 | if (wr == NULL) |
| 1579 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1580 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1581 | Py_DECREF(wr); |
| 1582 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1583 | } |
| 1584 | |
| 1585 | static PyObject * |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1586 | mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1587 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1588 | if (mbstreamwriter_iwrite(self, strobj)) |
| 1589 | return NULL; |
| 1590 | else |
| 1591 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1592 | } |
| 1593 | |
| 1594 | static PyObject * |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1595 | mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1596 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1597 | PyObject *strobj; |
| 1598 | int i, r; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1599 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1600 | if (!PySequence_Check(lines)) { |
| 1601 | PyErr_SetString(PyExc_TypeError, |
| 1602 | "arg must be a sequence object"); |
| 1603 | return NULL; |
| 1604 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1605 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1606 | for (i = 0; i < PySequence_Length(lines); i++) { |
| 1607 | /* length can be changed even within this loop */ |
| 1608 | strobj = PySequence_GetItem(lines, i); |
| 1609 | if (strobj == NULL) |
| 1610 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1611 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1612 | r = mbstreamwriter_iwrite(self, strobj); |
| 1613 | Py_DECREF(strobj); |
| 1614 | if (r == -1) |
| 1615 | return NULL; |
| 1616 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1617 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1618 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1619 | } |
| 1620 | |
| 1621 | static PyObject * |
| 1622 | mbstreamwriter_reset(MultibyteStreamWriterObject *self) |
| 1623 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1624 | PyObject *pwrt; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1625 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1626 | pwrt = multibytecodec_encode(self->codec, &self->state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 1627 | self->pending, NULL, self->errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1628 | MBENC_FLUSH | MBENC_RESET); |
| 1629 | /* some pending buffer can be truncated when UnicodeEncodeError is |
| 1630 | * raised on 'strict' mode. but, 'reset' method is designed to |
| 1631 | * reset the pending buffer or states so failed string sequence |
| 1632 | * ought to be missed */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 1633 | Py_CLEAR(self->pending); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1634 | if (pwrt == NULL) |
| 1635 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1636 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1637 | assert(PyBytes_Check(pwrt)); |
| 1638 | if (PyBytes_Size(pwrt) > 0) { |
| 1639 | PyObject *wr; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1640 | _Py_IDENTIFIER(write); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1641 | |
| 1642 | wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1643 | if (wr == NULL) { |
| 1644 | Py_DECREF(pwrt); |
| 1645 | return NULL; |
| 1646 | } |
| 1647 | } |
| 1648 | Py_DECREF(pwrt); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1649 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1650 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1651 | } |
| 1652 | |
| 1653 | static PyObject * |
| 1654 | mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1655 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1656 | MultibyteStreamWriterObject *self; |
| 1657 | PyObject *stream, *codec = NULL; |
| 1658 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1659 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1660 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter", |
| 1661 | streamkwarglist, &stream, &errors)) |
| 1662 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1663 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1664 | self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0); |
| 1665 | if (self == NULL) |
| 1666 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1667 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1668 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1669 | if (codec == NULL) |
| 1670 | goto errorexit; |
| 1671 | if (!MultibyteCodec_Check(codec)) { |
| 1672 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1673 | goto errorexit; |
| 1674 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1675 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1676 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1677 | self->stream = stream; |
| 1678 | Py_INCREF(stream); |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 1679 | self->pending = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1680 | self->errors = internal_error_callback(errors); |
| 1681 | if (self->errors == NULL) |
| 1682 | goto errorexit; |
| 1683 | if (self->codec->encinit != NULL && |
| 1684 | self->codec->encinit(&self->state, self->codec->config) != 0) |
| 1685 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1686 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1687 | Py_DECREF(codec); |
| 1688 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1689 | |
| 1690 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1691 | Py_XDECREF(self); |
| 1692 | Py_XDECREF(codec); |
| 1693 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1694 | } |
| 1695 | |
| 1696 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1697 | mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1698 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1699 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1700 | } |
| 1701 | |
| 1702 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1703 | mbstreamwriter_traverse(MultibyteStreamWriterObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1704 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1705 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1706 | if (ERROR_ISCUSTOM(self->errors)) |
| 1707 | Py_VISIT(self->errors); |
| 1708 | Py_VISIT(self->stream); |
| 1709 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1710 | } |
| 1711 | |
| 1712 | static void |
| 1713 | mbstreamwriter_dealloc(MultibyteStreamWriterObject *self) |
| 1714 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1715 | PyObject_GC_UnTrack(self); |
| 1716 | ERROR_DECREF(self->errors); |
| 1717 | Py_XDECREF(self->stream); |
| 1718 | Py_TYPE(self)->tp_free(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1719 | } |
| 1720 | |
| 1721 | static struct PyMethodDef mbstreamwriter_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1722 | {"write", (PyCFunction)mbstreamwriter_write, |
| 1723 | METH_O, NULL}, |
| 1724 | {"writelines", (PyCFunction)mbstreamwriter_writelines, |
| 1725 | METH_O, NULL}, |
| 1726 | {"reset", (PyCFunction)mbstreamwriter_reset, |
| 1727 | METH_NOARGS, NULL}, |
| 1728 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1729 | }; |
| 1730 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1731 | static PyMemberDef mbstreamwriter_members[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1732 | {"stream", T_OBJECT, |
| 1733 | offsetof(MultibyteStreamWriterObject, stream), |
| 1734 | READONLY, NULL}, |
| 1735 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1736 | }; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1737 | |
| 1738 | static PyTypeObject MultibyteStreamWriter_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1739 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1740 | "MultibyteStreamWriter", /* tp_name */ |
| 1741 | sizeof(MultibyteStreamWriterObject), /* tp_basicsize */ |
| 1742 | 0, /* tp_itemsize */ |
| 1743 | /* methods */ |
| 1744 | (destructor)mbstreamwriter_dealloc, /* tp_dealloc */ |
| 1745 | 0, /* tp_print */ |
| 1746 | 0, /* tp_getattr */ |
| 1747 | 0, /* tp_setattr */ |
| 1748 | 0, /* tp_reserved */ |
| 1749 | 0, /* tp_repr */ |
| 1750 | 0, /* tp_as_number */ |
| 1751 | 0, /* tp_as_sequence */ |
| 1752 | 0, /* tp_as_mapping */ |
| 1753 | 0, /* tp_hash */ |
| 1754 | 0, /* tp_call */ |
| 1755 | 0, /* tp_str */ |
| 1756 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1757 | 0, /* tp_setattro */ |
| 1758 | 0, /* tp_as_buffer */ |
| 1759 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1760 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1761 | 0, /* tp_doc */ |
| 1762 | (traverseproc)mbstreamwriter_traverse, /* tp_traverse */ |
| 1763 | 0, /* tp_clear */ |
| 1764 | 0, /* tp_richcompare */ |
| 1765 | 0, /* tp_weaklistoffset */ |
| 1766 | 0, /* tp_iter */ |
| 1767 | 0, /* tp_iterext */ |
| 1768 | mbstreamwriter_methods, /* tp_methods */ |
| 1769 | mbstreamwriter_members, /* tp_members */ |
| 1770 | codecctx_getsets, /* tp_getset */ |
| 1771 | 0, /* tp_base */ |
| 1772 | 0, /* tp_dict */ |
| 1773 | 0, /* tp_descr_get */ |
| 1774 | 0, /* tp_descr_set */ |
| 1775 | 0, /* tp_dictoffset */ |
| 1776 | mbstreamwriter_init, /* tp_init */ |
| 1777 | 0, /* tp_alloc */ |
| 1778 | mbstreamwriter_new, /* tp_new */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1779 | }; |
| 1780 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1781 | |
| 1782 | /** |
| 1783 | * Exposed factory function |
| 1784 | */ |
| 1785 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1786 | static PyObject * |
| 1787 | __create_codec(PyObject *ignore, PyObject *arg) |
| 1788 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1789 | MultibyteCodecObject *self; |
| 1790 | MultibyteCodec *codec; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1791 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1792 | if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) { |
| 1793 | PyErr_SetString(PyExc_ValueError, "argument type invalid"); |
| 1794 | return NULL; |
| 1795 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1796 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1797 | codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME); |
| 1798 | if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0) |
| 1799 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1800 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1801 | self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type); |
| 1802 | if (self == NULL) |
| 1803 | return NULL; |
| 1804 | self->codec = codec; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1805 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1806 | return (PyObject *)self; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1807 | } |
| 1808 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1809 | static struct PyMethodDef __methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1810 | {"__create_codec", (PyCFunction)__create_codec, METH_O}, |
| 1811 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1812 | }; |
| 1813 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1814 | |
| 1815 | static struct PyModuleDef _multibytecodecmodule = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1816 | PyModuleDef_HEAD_INIT, |
| 1817 | "_multibytecodec", |
| 1818 | NULL, |
| 1819 | -1, |
| 1820 | __methods, |
| 1821 | NULL, |
| 1822 | NULL, |
| 1823 | NULL, |
| 1824 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1825 | }; |
| 1826 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1827 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1828 | PyInit__multibytecodec(void) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1829 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1830 | int i; |
| 1831 | PyObject *m; |
| 1832 | PyTypeObject *typelist[] = { |
| 1833 | &MultibyteIncrementalEncoder_Type, |
| 1834 | &MultibyteIncrementalDecoder_Type, |
| 1835 | &MultibyteStreamReader_Type, |
| 1836 | &MultibyteStreamWriter_Type, |
| 1837 | NULL |
| 1838 | }; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1839 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1840 | if (PyType_Ready(&MultibyteCodec_Type) < 0) |
| 1841 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1842 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1843 | m = PyModule_Create(&_multibytecodecmodule); |
| 1844 | if (m == NULL) |
| 1845 | return NULL; |
Neal Norwitz | 058bde1 | 2005-09-21 06:44:25 +0000 | [diff] [blame] | 1846 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1847 | for (i = 0; typelist[i] != NULL; i++) { |
| 1848 | if (PyType_Ready(typelist[i]) < 0) |
| 1849 | return NULL; |
| 1850 | Py_INCREF(typelist[i]); |
| 1851 | PyModule_AddObject(m, typelist[i]->tp_name, |
| 1852 | (PyObject *)typelist[i]); |
| 1853 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1854 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1855 | if (PyErr_Occurred()) { |
| 1856 | Py_FatalError("can't initialize the _multibytecodec module"); |
| 1857 | Py_DECREF(m); |
| 1858 | m = NULL; |
| 1859 | } |
| 1860 | return m; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1861 | } |