Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1 | /* |
| 2 | * multibytecodec.c: Common Multibyte Codec Implementation |
| 3 | * |
| 4 | * Written by Hye-Shik Chang <perky@FreeBSD.org> |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 5 | */ |
| 6 | |
Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 7 | #define PY_SSIZE_T_CLEAN |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 8 | #include "Python.h" |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 9 | #include "structmember.h" |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 10 | #include "multibytecodec.h" |
| 11 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 12 | typedef struct { |
| 13 | const Py_UNICODE *inbuf, *inbuf_top, *inbuf_end; |
| 14 | unsigned char *outbuf, *outbuf_end; |
| 15 | PyObject *excobj, *outobj; |
| 16 | } MultibyteEncodeBuffer; |
| 17 | |
| 18 | typedef struct { |
| 19 | const unsigned char *inbuf, *inbuf_top, *inbuf_end; |
| 20 | Py_UNICODE *outbuf, *outbuf_end; |
| 21 | PyObject *excobj, *outobj; |
| 22 | } MultibyteDecodeBuffer; |
| 23 | |
| 24 | PyDoc_STRVAR(MultibyteCodec_Encode__doc__, |
| 25 | "I.encode(unicode[, errors]) -> (string, length consumed)\n\ |
| 26 | \n\ |
| 27 | Return an encoded string version of `unicode'. errors may be given to\n\ |
| 28 | set a different error handling scheme. Default is 'strict' meaning that\n\ |
| 29 | encoding errors raise a UnicodeEncodeError. Other possible values are\n\ |
| 30 | 'ignore', 'replace' and 'xmlcharrefreplace' as well as any other name\n\ |
| 31 | registered with codecs.register_error that can handle UnicodeEncodeErrors."); |
| 32 | |
| 33 | PyDoc_STRVAR(MultibyteCodec_Decode__doc__, |
| 34 | "I.decode(string[, errors]) -> (unicodeobject, length consumed)\n\ |
| 35 | \n\ |
| 36 | Decodes `string' using I, an MultibyteCodec instance. errors may be given\n\ |
| 37 | to set a different error handling scheme. Default is 'strict' meaning\n\ |
| 38 | that encoding errors raise a UnicodeDecodeError. Other possible values\n\ |
Benjamin Peterson | f10a79a | 2008-10-11 00:49:57 +0000 | [diff] [blame] | 39 | are 'ignore' and 'replace' as well as any other name registered with\n\ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 40 | codecs.register_error that is able to handle UnicodeDecodeErrors."); |
| 41 | |
Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 42 | static char *codeckwarglist[] = {"input", "errors", NULL}; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 43 | static char *incnewkwarglist[] = {"errors", NULL}; |
| 44 | static char *incrementalkwarglist[] = {"input", "final", NULL}; |
Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 45 | static char *streamkwarglist[] = {"stream", "errors", NULL}; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 46 | |
| 47 | static PyObject *multibytecodec_encode(MultibyteCodec *, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 48 | MultibyteCodec_State *, const Py_UNICODE **, Py_ssize_t, |
| 49 | PyObject *, int); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 50 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 51 | #define MBENC_RESET MBENC_MAX<<1 /* reset after an encoding session */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 52 | |
| 53 | static PyObject * |
Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 54 | make_tuple(PyObject *object, Py_ssize_t len) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 55 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 56 | PyObject *v, *w; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 57 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 58 | if (object == NULL) |
| 59 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 60 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 61 | v = PyTuple_New(2); |
| 62 | if (v == NULL) { |
| 63 | Py_DECREF(object); |
| 64 | return NULL; |
| 65 | } |
| 66 | PyTuple_SET_ITEM(v, 0, object); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 67 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 68 | w = PyLong_FromSsize_t(len); |
| 69 | if (w == NULL) { |
| 70 | Py_DECREF(v); |
| 71 | return NULL; |
| 72 | } |
| 73 | PyTuple_SET_ITEM(v, 1, w); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 74 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 75 | return v; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 76 | } |
| 77 | |
| 78 | static PyObject * |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 79 | internal_error_callback(const char *errors) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 80 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 81 | if (errors == NULL || strcmp(errors, "strict") == 0) |
| 82 | return ERROR_STRICT; |
| 83 | else if (strcmp(errors, "ignore") == 0) |
| 84 | return ERROR_IGNORE; |
| 85 | else if (strcmp(errors, "replace") == 0) |
| 86 | return ERROR_REPLACE; |
| 87 | else |
| 88 | return PyUnicode_FromString(errors); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 89 | } |
| 90 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 91 | static PyObject * |
| 92 | call_error_callback(PyObject *errors, PyObject *exc) |
| 93 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 94 | PyObject *args, *cb, *r; |
| 95 | const char *str; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 96 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 97 | assert(PyUnicode_Check(errors)); |
| 98 | str = _PyUnicode_AsString(errors); |
| 99 | if (str == NULL) |
| 100 | return NULL; |
| 101 | cb = PyCodec_LookupError(str); |
| 102 | if (cb == NULL) |
| 103 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 104 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 105 | args = PyTuple_New(1); |
| 106 | if (args == NULL) { |
| 107 | Py_DECREF(cb); |
| 108 | return NULL; |
| 109 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 110 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 111 | PyTuple_SET_ITEM(args, 0, exc); |
| 112 | Py_INCREF(exc); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 113 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 114 | r = PyObject_CallObject(cb, args); |
| 115 | Py_DECREF(args); |
| 116 | Py_DECREF(cb); |
| 117 | return r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 118 | } |
| 119 | |
| 120 | static PyObject * |
| 121 | codecctx_errors_get(MultibyteStatefulCodecContext *self) |
| 122 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 123 | const char *errors; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 124 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 125 | if (self->errors == ERROR_STRICT) |
| 126 | errors = "strict"; |
| 127 | else if (self->errors == ERROR_IGNORE) |
| 128 | errors = "ignore"; |
| 129 | else if (self->errors == ERROR_REPLACE) |
| 130 | errors = "replace"; |
| 131 | else { |
| 132 | Py_INCREF(self->errors); |
| 133 | return self->errors; |
| 134 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 135 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 136 | return PyUnicode_FromString(errors); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 137 | } |
| 138 | |
| 139 | static int |
| 140 | codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 141 | void *closure) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 142 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 143 | PyObject *cb; |
| 144 | const char *str; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 145 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 146 | if (!PyUnicode_Check(value)) { |
| 147 | PyErr_SetString(PyExc_TypeError, "errors must be a string"); |
| 148 | return -1; |
| 149 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 150 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 151 | str = _PyUnicode_AsString(value); |
| 152 | if (str == NULL) |
| 153 | return -1; |
Neal Norwitz | 6ea45d3 | 2007-08-26 04:19:43 +0000 | [diff] [blame] | 154 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 155 | cb = internal_error_callback(str); |
| 156 | if (cb == NULL) |
| 157 | return -1; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 158 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 159 | ERROR_DECREF(self->errors); |
| 160 | self->errors = cb; |
| 161 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 162 | } |
| 163 | |
| 164 | /* This getset handlers list is used by all the stateful codec objects */ |
| 165 | static PyGetSetDef codecctx_getsets[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 166 | {"errors", (getter)codecctx_errors_get, |
| 167 | (setter)codecctx_errors_set, |
| 168 | PyDoc_STR("how to treat errors")}, |
| 169 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 170 | }; |
| 171 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 172 | static int |
Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 173 | expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 174 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 175 | Py_ssize_t orgpos, orgsize, incsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 176 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 177 | orgpos = (Py_ssize_t)((char *)buf->outbuf - |
| 178 | PyBytes_AS_STRING(buf->outobj)); |
| 179 | orgsize = PyBytes_GET_SIZE(buf->outobj); |
| 180 | incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize); |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 181 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 182 | if (orgsize > PY_SSIZE_T_MAX - incsize) |
| 183 | return -1; |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 184 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 185 | if (_PyBytes_Resize(&buf->outobj, orgsize + incsize) == -1) |
| 186 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 187 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 188 | buf->outbuf = (unsigned char *)PyBytes_AS_STRING(buf->outobj) +orgpos; |
| 189 | buf->outbuf_end = (unsigned char *)PyBytes_AS_STRING(buf->outobj) |
| 190 | + PyBytes_GET_SIZE(buf->outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 191 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 192 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 193 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 194 | #define REQUIRE_ENCODEBUFFER(buf, s) { \ |
| 195 | if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end) \ |
| 196 | if (expand_encodebuffer(buf, s) == -1) \ |
| 197 | goto errorexit; \ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 198 | } |
| 199 | |
| 200 | static int |
Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 201 | expand_decodebuffer(MultibyteDecodeBuffer *buf, Py_ssize_t esize) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 202 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 203 | Py_ssize_t orgpos, orgsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 204 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 205 | orgpos = (Py_ssize_t)(buf->outbuf - PyUnicode_AS_UNICODE(buf->outobj)); |
| 206 | orgsize = PyUnicode_GET_SIZE(buf->outobj); |
| 207 | if (PyUnicode_Resize(&buf->outobj, orgsize + ( |
| 208 | esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize)) == -1) |
| 209 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 210 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 211 | buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj) + orgpos; |
| 212 | buf->outbuf_end = PyUnicode_AS_UNICODE(buf->outobj) |
| 213 | + PyUnicode_GET_SIZE(buf->outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 214 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 215 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 216 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 217 | #define REQUIRE_DECODEBUFFER(buf, s) { \ |
| 218 | if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end) \ |
| 219 | if (expand_decodebuffer(buf, s) == -1) \ |
| 220 | goto errorexit; \ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 221 | } |
| 222 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 223 | |
| 224 | /** |
| 225 | * MultibyteCodec object |
| 226 | */ |
| 227 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 228 | static int |
| 229 | multibytecodec_encerror(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 230 | MultibyteCodec_State *state, |
| 231 | MultibyteEncodeBuffer *buf, |
| 232 | PyObject *errors, Py_ssize_t e) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 233 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 234 | PyObject *retobj = NULL, *retstr = NULL, *tobj; |
| 235 | Py_ssize_t retstrsize, newpos; |
| 236 | Py_ssize_t esize, start, end; |
| 237 | const char *reason; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 238 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 239 | if (e > 0) { |
| 240 | reason = "illegal multibyte sequence"; |
| 241 | esize = e; |
| 242 | } |
| 243 | else { |
| 244 | switch (e) { |
| 245 | case MBERR_TOOSMALL: |
| 246 | REQUIRE_ENCODEBUFFER(buf, -1); |
| 247 | return 0; /* retry it */ |
| 248 | case MBERR_TOOFEW: |
| 249 | reason = "incomplete multibyte sequence"; |
| 250 | esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
| 251 | break; |
| 252 | case MBERR_INTERNAL: |
| 253 | PyErr_SetString(PyExc_RuntimeError, |
| 254 | "internal codec error"); |
| 255 | return -1; |
| 256 | default: |
| 257 | PyErr_SetString(PyExc_RuntimeError, |
| 258 | "unknown runtime error"); |
| 259 | return -1; |
| 260 | } |
| 261 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 262 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 263 | if (errors == ERROR_REPLACE) { |
| 264 | const Py_UNICODE replchar = '?', *inbuf = &replchar; |
| 265 | Py_ssize_t r; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 266 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 267 | for (;;) { |
| 268 | Py_ssize_t outleft; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 269 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 270 | outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf); |
| 271 | r = codec->encode(state, codec->config, &inbuf, 1, |
| 272 | &buf->outbuf, outleft, 0); |
| 273 | if (r == MBERR_TOOSMALL) { |
| 274 | REQUIRE_ENCODEBUFFER(buf, -1); |
| 275 | continue; |
| 276 | } |
| 277 | else |
| 278 | break; |
| 279 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 280 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 281 | if (r != 0) { |
| 282 | REQUIRE_ENCODEBUFFER(buf, 1); |
| 283 | *buf->outbuf++ = '?'; |
| 284 | } |
| 285 | } |
| 286 | if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { |
| 287 | buf->inbuf += esize; |
| 288 | return 0; |
| 289 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 290 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 291 | start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top); |
| 292 | end = start + esize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 293 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 294 | /* use cached exception object if available */ |
| 295 | if (buf->excobj == NULL) { |
| 296 | buf->excobj = PyUnicodeEncodeError_Create(codec->encoding, |
| 297 | buf->inbuf_top, |
| 298 | buf->inbuf_end - buf->inbuf_top, |
| 299 | start, end, reason); |
| 300 | if (buf->excobj == NULL) |
| 301 | goto errorexit; |
| 302 | } |
| 303 | else |
| 304 | if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 || |
| 305 | PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 || |
| 306 | PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0) |
| 307 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 308 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 309 | if (errors == ERROR_STRICT) { |
| 310 | PyCodec_StrictErrors(buf->excobj); |
| 311 | goto errorexit; |
| 312 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 313 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 314 | retobj = call_error_callback(errors, buf->excobj); |
| 315 | if (retobj == NULL) |
| 316 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 317 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 318 | if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || |
Benjamin Peterson | 47a00f3 | 2012-12-02 11:20:28 -0500 | [diff] [blame] | 319 | (!PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) && !PyBytes_Check(tobj)) || |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 320 | !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { |
| 321 | PyErr_SetString(PyExc_TypeError, |
| 322 | "encoding error handler must return " |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 323 | "(str, int) tuple"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 324 | goto errorexit; |
| 325 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 326 | |
Benjamin Peterson | 47a00f3 | 2012-12-02 11:20:28 -0500 | [diff] [blame] | 327 | if (PyUnicode_Check(tobj)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 328 | const Py_UNICODE *uraw = PyUnicode_AS_UNICODE(tobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 329 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 330 | retstr = multibytecodec_encode(codec, state, &uraw, |
| 331 | PyUnicode_GET_SIZE(tobj), ERROR_STRICT, |
| 332 | MBENC_FLUSH); |
| 333 | if (retstr == NULL) |
| 334 | goto errorexit; |
| 335 | } |
Benjamin Peterson | 47a00f3 | 2012-12-02 11:20:28 -0500 | [diff] [blame] | 336 | else { |
| 337 | Py_INCREF(tobj); |
| 338 | retstr = tobj; |
| 339 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 340 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 341 | assert(PyBytes_Check(retstr)); |
| 342 | retstrsize = PyBytes_GET_SIZE(retstr); |
| 343 | REQUIRE_ENCODEBUFFER(buf, retstrsize); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 344 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 345 | memcpy(buf->outbuf, PyBytes_AS_STRING(retstr), retstrsize); |
| 346 | buf->outbuf += retstrsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 347 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 348 | newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); |
| 349 | if (newpos < 0 && !PyErr_Occurred()) |
| 350 | newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top); |
| 351 | if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) { |
| 352 | PyErr_Clear(); |
| 353 | PyErr_Format(PyExc_IndexError, |
| 354 | "position %zd from error handler out of bounds", |
| 355 | newpos); |
| 356 | goto errorexit; |
| 357 | } |
| 358 | buf->inbuf = buf->inbuf_top + newpos; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 359 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 360 | Py_DECREF(retobj); |
| 361 | Py_DECREF(retstr); |
| 362 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 363 | |
| 364 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 365 | Py_XDECREF(retobj); |
| 366 | Py_XDECREF(retstr); |
| 367 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 368 | } |
| 369 | |
| 370 | static int |
| 371 | multibytecodec_decerror(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 372 | MultibyteCodec_State *state, |
| 373 | MultibyteDecodeBuffer *buf, |
| 374 | PyObject *errors, Py_ssize_t e) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 375 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 376 | PyObject *retobj = NULL, *retuni = NULL; |
| 377 | Py_ssize_t retunisize, newpos; |
| 378 | const char *reason; |
| 379 | Py_ssize_t esize, start, end; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 380 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 381 | if (e > 0) { |
| 382 | reason = "illegal multibyte sequence"; |
| 383 | esize = e; |
| 384 | } |
| 385 | else { |
| 386 | switch (e) { |
| 387 | case MBERR_TOOSMALL: |
| 388 | REQUIRE_DECODEBUFFER(buf, -1); |
| 389 | return 0; /* retry it */ |
| 390 | case MBERR_TOOFEW: |
| 391 | reason = "incomplete multibyte sequence"; |
| 392 | esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
| 393 | break; |
| 394 | case MBERR_INTERNAL: |
| 395 | PyErr_SetString(PyExc_RuntimeError, |
| 396 | "internal codec error"); |
| 397 | return -1; |
| 398 | default: |
| 399 | PyErr_SetString(PyExc_RuntimeError, |
| 400 | "unknown runtime error"); |
| 401 | return -1; |
| 402 | } |
| 403 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 404 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 405 | if (errors == ERROR_REPLACE) { |
| 406 | REQUIRE_DECODEBUFFER(buf, 1); |
| 407 | *buf->outbuf++ = Py_UNICODE_REPLACEMENT_CHARACTER; |
| 408 | } |
| 409 | if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { |
| 410 | buf->inbuf += esize; |
| 411 | return 0; |
| 412 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 413 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 414 | start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top); |
| 415 | end = start + esize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 416 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 417 | /* use cached exception object if available */ |
| 418 | if (buf->excobj == NULL) { |
| 419 | buf->excobj = PyUnicodeDecodeError_Create(codec->encoding, |
| 420 | (const char *)buf->inbuf_top, |
| 421 | (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top), |
| 422 | start, end, reason); |
| 423 | if (buf->excobj == NULL) |
| 424 | goto errorexit; |
| 425 | } |
| 426 | else |
| 427 | if (PyUnicodeDecodeError_SetStart(buf->excobj, start) || |
| 428 | PyUnicodeDecodeError_SetEnd(buf->excobj, end) || |
| 429 | PyUnicodeDecodeError_SetReason(buf->excobj, reason)) |
| 430 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 431 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 432 | if (errors == ERROR_STRICT) { |
| 433 | PyCodec_StrictErrors(buf->excobj); |
| 434 | goto errorexit; |
| 435 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 436 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 437 | retobj = call_error_callback(errors, buf->excobj); |
| 438 | if (retobj == NULL) |
| 439 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 440 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 441 | if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || |
| 442 | !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) || |
| 443 | !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { |
| 444 | PyErr_SetString(PyExc_TypeError, |
| 445 | "decoding error handler must return " |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 446 | "(str, int) tuple"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 447 | goto errorexit; |
| 448 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 449 | |
Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 450 | if (PyUnicode_AsUnicode(retuni) == NULL) |
| 451 | goto errorexit; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 452 | retunisize = PyUnicode_GET_SIZE(retuni); |
| 453 | if (retunisize > 0) { |
| 454 | REQUIRE_DECODEBUFFER(buf, retunisize); |
Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 455 | memcpy((char *)buf->outbuf, PyUnicode_AS_UNICODE(retuni), |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 456 | retunisize * Py_UNICODE_SIZE); |
| 457 | buf->outbuf += retunisize; |
| 458 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 459 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 460 | newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); |
| 461 | if (newpos < 0 && !PyErr_Occurred()) |
| 462 | newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top); |
| 463 | if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) { |
| 464 | PyErr_Clear(); |
| 465 | PyErr_Format(PyExc_IndexError, |
| 466 | "position %zd from error handler out of bounds", |
| 467 | newpos); |
| 468 | goto errorexit; |
| 469 | } |
| 470 | buf->inbuf = buf->inbuf_top + newpos; |
| 471 | Py_DECREF(retobj); |
| 472 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 473 | |
| 474 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 475 | Py_XDECREF(retobj); |
| 476 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 477 | } |
| 478 | |
| 479 | static PyObject * |
| 480 | multibytecodec_encode(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 481 | MultibyteCodec_State *state, |
| 482 | const Py_UNICODE **data, Py_ssize_t datalen, |
| 483 | PyObject *errors, int flags) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 484 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 485 | MultibyteEncodeBuffer buf; |
| 486 | Py_ssize_t finalsize, r = 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 487 | |
Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 488 | if (datalen == 0 && !(flags & MBENC_RESET)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 489 | return PyBytes_FromStringAndSize(NULL, 0); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 490 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 491 | buf.excobj = NULL; |
Brett Cannon | b94767f | 2011-02-22 20:15:44 +0000 | [diff] [blame] | 492 | buf.outobj = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 493 | buf.inbuf = buf.inbuf_top = *data; |
| 494 | buf.inbuf_end = buf.inbuf_top + datalen; |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 495 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 496 | if (datalen > (PY_SSIZE_T_MAX - 16) / 2) { |
| 497 | PyErr_NoMemory(); |
| 498 | goto errorexit; |
| 499 | } |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 500 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 501 | buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16); |
| 502 | if (buf.outobj == NULL) |
| 503 | goto errorexit; |
| 504 | buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj); |
| 505 | buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 506 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 507 | while (buf.inbuf < buf.inbuf_end) { |
| 508 | Py_ssize_t inleft, outleft; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 509 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 510 | /* we don't reuse inleft and outleft here. |
| 511 | * error callbacks can relocate the cursor anywhere on buffer*/ |
| 512 | inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf); |
| 513 | outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
| 514 | r = codec->encode(state, codec->config, &buf.inbuf, inleft, |
| 515 | &buf.outbuf, outleft, flags); |
| 516 | if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH))) |
| 517 | break; |
| 518 | else if (multibytecodec_encerror(codec, state, &buf, errors,r)) |
| 519 | goto errorexit; |
| 520 | else if (r == MBERR_TOOFEW) |
| 521 | break; |
| 522 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 523 | |
Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 524 | if (codec->encreset != NULL && (flags & MBENC_RESET)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 525 | for (;;) { |
| 526 | Py_ssize_t outleft; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 527 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 528 | outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
| 529 | r = codec->encreset(state, codec->config, &buf.outbuf, |
| 530 | outleft); |
| 531 | if (r == 0) |
| 532 | break; |
| 533 | else if (multibytecodec_encerror(codec, state, |
| 534 | &buf, errors, r)) |
| 535 | goto errorexit; |
| 536 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 537 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 538 | finalsize = (Py_ssize_t)((char *)buf.outbuf - |
| 539 | PyBytes_AS_STRING(buf.outobj)); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 540 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 541 | if (finalsize != PyBytes_GET_SIZE(buf.outobj)) |
| 542 | if (_PyBytes_Resize(&buf.outobj, finalsize) == -1) |
| 543 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 544 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 545 | *data = buf.inbuf; |
| 546 | Py_XDECREF(buf.excobj); |
| 547 | return buf.outobj; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 548 | |
| 549 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 550 | Py_XDECREF(buf.excobj); |
| 551 | Py_XDECREF(buf.outobj); |
| 552 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 553 | } |
| 554 | |
| 555 | static PyObject * |
| 556 | MultibyteCodec_Encode(MultibyteCodecObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 557 | PyObject *args, PyObject *kwargs) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 558 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 559 | MultibyteCodec_State state; |
| 560 | Py_UNICODE *data; |
| 561 | PyObject *errorcb, *r, *arg, *ucvt; |
| 562 | const char *errors = NULL; |
| 563 | Py_ssize_t datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 564 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 565 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|z:encode", |
| 566 | codeckwarglist, &arg, &errors)) |
| 567 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 568 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 569 | if (PyUnicode_Check(arg)) |
| 570 | ucvt = NULL; |
| 571 | else { |
| 572 | arg = ucvt = PyObject_Str(arg); |
| 573 | if (arg == NULL) |
| 574 | return NULL; |
| 575 | else if (!PyUnicode_Check(arg)) { |
| 576 | PyErr_SetString(PyExc_TypeError, |
| 577 | "couldn't convert the object to unicode."); |
| 578 | Py_DECREF(ucvt); |
| 579 | return NULL; |
| 580 | } |
| 581 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 582 | |
Victor Stinner | 9a80fab | 2011-11-21 02:50:14 +0100 | [diff] [blame] | 583 | data = PyUnicode_AsUnicodeAndSize(arg, &datalen); |
| 584 | if (data == NULL) { |
| 585 | Py_XDECREF(ucvt); |
| 586 | return NULL; |
| 587 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 588 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 589 | errorcb = internal_error_callback(errors); |
| 590 | if (errorcb == NULL) { |
| 591 | Py_XDECREF(ucvt); |
| 592 | return NULL; |
| 593 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 594 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 595 | if (self->codec->encinit != NULL && |
| 596 | self->codec->encinit(&state, self->codec->config) != 0) |
| 597 | goto errorexit; |
| 598 | r = multibytecodec_encode(self->codec, &state, |
| 599 | (const Py_UNICODE **)&data, datalen, errorcb, |
| 600 | MBENC_FLUSH | MBENC_RESET); |
| 601 | if (r == NULL) |
| 602 | goto errorexit; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 603 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 604 | ERROR_DECREF(errorcb); |
| 605 | Py_XDECREF(ucvt); |
| 606 | return make_tuple(r, datalen); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 607 | |
| 608 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 609 | ERROR_DECREF(errorcb); |
| 610 | Py_XDECREF(ucvt); |
| 611 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 612 | } |
| 613 | |
| 614 | static PyObject * |
| 615 | MultibyteCodec_Decode(MultibyteCodecObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 616 | PyObject *args, PyObject *kwargs) |
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 | MultibyteCodec_State state; |
| 619 | MultibyteDecodeBuffer buf; |
| 620 | PyObject *errorcb; |
| 621 | Py_buffer pdata; |
| 622 | const char *data, *errors = NULL; |
| 623 | Py_ssize_t datalen, finalsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 624 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 625 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|z:decode", |
| 626 | codeckwarglist, &pdata, &errors)) |
| 627 | return NULL; |
| 628 | data = pdata.buf; |
| 629 | datalen = pdata.len; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 630 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 631 | errorcb = internal_error_callback(errors); |
| 632 | if (errorcb == NULL) { |
| 633 | PyBuffer_Release(&pdata); |
| 634 | return NULL; |
| 635 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 636 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 637 | if (datalen == 0) { |
| 638 | PyBuffer_Release(&pdata); |
| 639 | ERROR_DECREF(errorcb); |
Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 640 | return make_tuple(PyUnicode_New(0, 0), 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 641 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 642 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 643 | buf.excobj = NULL; |
| 644 | buf.inbuf = buf.inbuf_top = (unsigned char *)data; |
| 645 | buf.inbuf_end = buf.inbuf_top + datalen; |
| 646 | buf.outobj = PyUnicode_FromUnicode(NULL, datalen); |
| 647 | if (buf.outobj == NULL) |
| 648 | goto errorexit; |
| 649 | buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj); |
Victor Stinner | 08b523a | 2011-12-01 03:18:30 +0100 | [diff] [blame] | 650 | if (buf.outbuf == NULL) |
| 651 | goto errorexit; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 652 | buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 653 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 654 | if (self->codec->decinit != NULL && |
| 655 | self->codec->decinit(&state, self->codec->config) != 0) |
| 656 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 657 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 658 | while (buf.inbuf < buf.inbuf_end) { |
| 659 | Py_ssize_t inleft, outleft, r; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 660 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 661 | inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf); |
| 662 | outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
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 | r = self->codec->decode(&state, self->codec->config, |
| 665 | &buf.inbuf, inleft, &buf.outbuf, outleft); |
| 666 | if (r == 0) |
| 667 | break; |
| 668 | else if (multibytecodec_decerror(self->codec, &state, |
| 669 | &buf, errorcb, r)) |
| 670 | goto errorexit; |
| 671 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 672 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 673 | finalsize = (Py_ssize_t)(buf.outbuf - |
| 674 | PyUnicode_AS_UNICODE(buf.outobj)); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 675 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 676 | if (finalsize != PyUnicode_GET_SIZE(buf.outobj)) |
| 677 | if (PyUnicode_Resize(&buf.outobj, finalsize) == -1) |
| 678 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 679 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 680 | PyBuffer_Release(&pdata); |
| 681 | Py_XDECREF(buf.excobj); |
| 682 | ERROR_DECREF(errorcb); |
| 683 | return make_tuple(buf.outobj, datalen); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 684 | |
| 685 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 686 | PyBuffer_Release(&pdata); |
| 687 | ERROR_DECREF(errorcb); |
| 688 | Py_XDECREF(buf.excobj); |
| 689 | Py_XDECREF(buf.outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 690 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 691 | return NULL; |
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 struct PyMethodDef multibytecodec_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 695 | {"encode", (PyCFunction)MultibyteCodec_Encode, |
| 696 | METH_VARARGS | METH_KEYWORDS, |
| 697 | MultibyteCodec_Encode__doc__}, |
| 698 | {"decode", (PyCFunction)MultibyteCodec_Decode, |
| 699 | METH_VARARGS | METH_KEYWORDS, |
| 700 | MultibyteCodec_Decode__doc__}, |
| 701 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 702 | }; |
| 703 | |
| 704 | static void |
| 705 | multibytecodec_dealloc(MultibyteCodecObject *self) |
| 706 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 707 | PyObject_Del(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 708 | } |
| 709 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 710 | static PyTypeObject MultibyteCodec_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 711 | PyVarObject_HEAD_INIT(NULL, 0) |
| 712 | "MultibyteCodec", /* tp_name */ |
| 713 | sizeof(MultibyteCodecObject), /* tp_basicsize */ |
| 714 | 0, /* tp_itemsize */ |
| 715 | /* methods */ |
| 716 | (destructor)multibytecodec_dealloc, /* tp_dealloc */ |
| 717 | 0, /* tp_print */ |
| 718 | 0, /* tp_getattr */ |
| 719 | 0, /* tp_setattr */ |
| 720 | 0, /* tp_reserved */ |
| 721 | 0, /* tp_repr */ |
| 722 | 0, /* tp_as_number */ |
| 723 | 0, /* tp_as_sequence */ |
| 724 | 0, /* tp_as_mapping */ |
| 725 | 0, /* tp_hash */ |
| 726 | 0, /* tp_call */ |
| 727 | 0, /* tp_str */ |
| 728 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 729 | 0, /* tp_setattro */ |
| 730 | 0, /* tp_as_buffer */ |
| 731 | Py_TPFLAGS_DEFAULT, /* tp_flags */ |
| 732 | 0, /* tp_doc */ |
| 733 | 0, /* tp_traverse */ |
| 734 | 0, /* tp_clear */ |
| 735 | 0, /* tp_richcompare */ |
| 736 | 0, /* tp_weaklistoffset */ |
| 737 | 0, /* tp_iter */ |
| 738 | 0, /* tp_iterext */ |
| 739 | multibytecodec_methods, /* tp_methods */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 740 | }; |
| 741 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 742 | |
| 743 | /** |
| 744 | * Utility functions for stateful codec mechanism |
| 745 | */ |
| 746 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 747 | #define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o)) |
| 748 | #define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 749 | |
| 750 | static PyObject * |
| 751 | encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 752 | PyObject *unistr, int final) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 753 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 754 | PyObject *ucvt, *r = NULL; |
| 755 | Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL; |
| 756 | Py_ssize_t datalen, origpending; |
Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 757 | wchar_t *data; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 758 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 759 | if (PyUnicode_Check(unistr)) |
| 760 | ucvt = NULL; |
| 761 | else { |
| 762 | unistr = ucvt = PyObject_Str(unistr); |
| 763 | if (unistr == NULL) |
| 764 | return NULL; |
| 765 | else if (!PyUnicode_Check(unistr)) { |
| 766 | PyErr_SetString(PyExc_TypeError, |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 767 | "couldn't convert the object to str."); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 768 | Py_DECREF(ucvt); |
| 769 | return NULL; |
| 770 | } |
| 771 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 772 | |
Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 773 | data = PyUnicode_AsUnicodeAndSize(unistr, &datalen); |
| 774 | if (data == NULL) |
| 775 | goto errorexit; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 776 | origpending = ctx->pendingsize; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 777 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 778 | if (origpending > 0) { |
| 779 | if (datalen > PY_SSIZE_T_MAX - ctx->pendingsize) { |
| 780 | PyErr_NoMemory(); |
| 781 | /* inbuf_tmp == NULL */ |
| 782 | goto errorexit; |
| 783 | } |
| 784 | inbuf_tmp = PyMem_New(Py_UNICODE, datalen + ctx->pendingsize); |
| 785 | if (inbuf_tmp == NULL) |
| 786 | goto errorexit; |
| 787 | memcpy(inbuf_tmp, ctx->pending, |
| 788 | Py_UNICODE_SIZE * ctx->pendingsize); |
| 789 | memcpy(inbuf_tmp + ctx->pendingsize, |
| 790 | PyUnicode_AS_UNICODE(unistr), |
| 791 | Py_UNICODE_SIZE * datalen); |
| 792 | datalen += ctx->pendingsize; |
| 793 | ctx->pendingsize = 0; |
| 794 | inbuf = inbuf_tmp; |
| 795 | } |
| 796 | else |
| 797 | inbuf = (Py_UNICODE *)PyUnicode_AS_UNICODE(unistr); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 798 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 799 | inbuf_end = inbuf + datalen; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 800 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 801 | r = multibytecodec_encode(ctx->codec, &ctx->state, |
Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 802 | (const Py_UNICODE **)&inbuf, datalen, |
| 803 | ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 804 | if (r == NULL) { |
| 805 | /* recover the original pending buffer */ |
| 806 | if (origpending > 0) |
| 807 | memcpy(ctx->pending, inbuf_tmp, |
| 808 | Py_UNICODE_SIZE * origpending); |
| 809 | ctx->pendingsize = origpending; |
| 810 | goto errorexit; |
| 811 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 812 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 813 | if (inbuf < inbuf_end) { |
| 814 | ctx->pendingsize = (Py_ssize_t)(inbuf_end - inbuf); |
| 815 | if (ctx->pendingsize > MAXENCPENDING) { |
| 816 | /* normal codecs can't reach here */ |
| 817 | ctx->pendingsize = 0; |
| 818 | PyErr_SetString(PyExc_UnicodeError, |
| 819 | "pending buffer overflow"); |
| 820 | goto errorexit; |
| 821 | } |
| 822 | memcpy(ctx->pending, inbuf, |
| 823 | ctx->pendingsize * Py_UNICODE_SIZE); |
| 824 | } |
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 | if (inbuf_tmp != NULL) |
| 827 | PyMem_Del(inbuf_tmp); |
| 828 | Py_XDECREF(ucvt); |
| 829 | return r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 830 | |
| 831 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 832 | if (inbuf_tmp != NULL) |
| 833 | PyMem_Del(inbuf_tmp); |
| 834 | Py_XDECREF(r); |
| 835 | Py_XDECREF(ucvt); |
| 836 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 837 | } |
| 838 | |
| 839 | static int |
| 840 | decoder_append_pending(MultibyteStatefulDecoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 841 | MultibyteDecodeBuffer *buf) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 842 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 843 | Py_ssize_t npendings; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 844 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 845 | npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
| 846 | if (npendings + ctx->pendingsize > MAXDECPENDING || |
| 847 | npendings > PY_SSIZE_T_MAX - ctx->pendingsize) { |
| 848 | PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow"); |
| 849 | return -1; |
| 850 | } |
| 851 | memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings); |
| 852 | ctx->pendingsize += npendings; |
| 853 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 854 | } |
| 855 | |
| 856 | static int |
| 857 | decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 858 | Py_ssize_t size) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 859 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 860 | buf->inbuf = buf->inbuf_top = (const unsigned char *)data; |
| 861 | buf->inbuf_end = buf->inbuf_top + size; |
| 862 | if (buf->outobj == NULL) { /* only if outobj is not allocated yet */ |
| 863 | buf->outobj = PyUnicode_FromUnicode(NULL, size); |
| 864 | if (buf->outobj == NULL) |
| 865 | return -1; |
Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 866 | buf->outbuf = PyUnicode_AsUnicode(buf->outobj); |
| 867 | if (buf->outbuf == NULL) |
| 868 | return -1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 869 | buf->outbuf_end = buf->outbuf + |
| 870 | PyUnicode_GET_SIZE(buf->outobj); |
| 871 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 872 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 873 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 874 | } |
| 875 | |
| 876 | static int |
| 877 | decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 878 | MultibyteDecodeBuffer *buf) |
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 | while (buf->inbuf < buf->inbuf_end) { |
| 881 | Py_ssize_t inleft, outleft; |
| 882 | Py_ssize_t r; |
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 | inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
| 885 | outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 886 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 887 | r = ctx->codec->decode(&ctx->state, ctx->codec->config, |
| 888 | &buf->inbuf, inleft, &buf->outbuf, outleft); |
| 889 | if (r == 0 || r == MBERR_TOOFEW) |
| 890 | break; |
| 891 | else if (multibytecodec_decerror(ctx->codec, &ctx->state, |
| 892 | buf, ctx->errors, r)) |
| 893 | return -1; |
| 894 | } |
| 895 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 896 | } |
| 897 | |
| 898 | |
| 899 | /** |
| 900 | * MultibyteIncrementalEncoder object |
| 901 | */ |
| 902 | |
| 903 | static PyObject * |
| 904 | mbiencoder_encode(MultibyteIncrementalEncoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 905 | PyObject *args, PyObject *kwargs) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 906 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 907 | PyObject *data; |
| 908 | int final = 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 909 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 910 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|i:encode", |
| 911 | incrementalkwarglist, &data, &final)) |
| 912 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 913 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 914 | return encoder_encode_stateful(STATEFUL_ECTX(self), data, final); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 915 | } |
| 916 | |
| 917 | static PyObject * |
| 918 | mbiencoder_reset(MultibyteIncrementalEncoderObject *self) |
| 919 | { |
Victor Stinner | e15dce3 | 2011-05-30 22:56:00 +0200 | [diff] [blame] | 920 | /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */ |
| 921 | unsigned char buffer[4], *outbuf; |
| 922 | Py_ssize_t r; |
| 923 | if (self->codec->encreset != NULL) { |
| 924 | outbuf = buffer; |
| 925 | r = self->codec->encreset(&self->state, self->codec->config, |
| 926 | &outbuf, sizeof(buffer)); |
| 927 | if (r != 0) |
| 928 | return NULL; |
| 929 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 930 | self->pendingsize = 0; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 931 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 932 | } |
| 933 | |
| 934 | static struct PyMethodDef mbiencoder_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 935 | {"encode", (PyCFunction)mbiencoder_encode, |
| 936 | METH_VARARGS | METH_KEYWORDS, NULL}, |
| 937 | {"reset", (PyCFunction)mbiencoder_reset, |
| 938 | METH_NOARGS, NULL}, |
| 939 | {NULL, NULL}, |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 940 | }; |
| 941 | |
| 942 | static PyObject * |
| 943 | mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 944 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 945 | MultibyteIncrementalEncoderObject *self; |
| 946 | PyObject *codec = NULL; |
| 947 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 948 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 949 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder", |
| 950 | incnewkwarglist, &errors)) |
| 951 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 952 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 953 | self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0); |
| 954 | if (self == NULL) |
| 955 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 956 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 957 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 958 | if (codec == NULL) |
| 959 | goto errorexit; |
| 960 | if (!MultibyteCodec_Check(codec)) { |
| 961 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 962 | goto errorexit; |
| 963 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 964 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 965 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 966 | self->pendingsize = 0; |
| 967 | self->errors = internal_error_callback(errors); |
| 968 | if (self->errors == NULL) |
| 969 | goto errorexit; |
| 970 | if (self->codec->encinit != NULL && |
| 971 | self->codec->encinit(&self->state, self->codec->config) != 0) |
| 972 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 973 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 974 | Py_DECREF(codec); |
| 975 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 976 | |
| 977 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 978 | Py_XDECREF(self); |
| 979 | Py_XDECREF(codec); |
| 980 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 981 | } |
| 982 | |
| 983 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 984 | mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 985 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 986 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 987 | } |
| 988 | |
| 989 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 990 | mbiencoder_traverse(MultibyteIncrementalEncoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 991 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 992 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 993 | if (ERROR_ISCUSTOM(self->errors)) |
| 994 | Py_VISIT(self->errors); |
| 995 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 996 | } |
| 997 | |
| 998 | static void |
| 999 | mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self) |
| 1000 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1001 | PyObject_GC_UnTrack(self); |
| 1002 | ERROR_DECREF(self->errors); |
| 1003 | Py_TYPE(self)->tp_free(self); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1004 | } |
| 1005 | |
| 1006 | static PyTypeObject MultibyteIncrementalEncoder_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1007 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1008 | "MultibyteIncrementalEncoder", /* tp_name */ |
| 1009 | sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */ |
| 1010 | 0, /* tp_itemsize */ |
| 1011 | /* methods */ |
| 1012 | (destructor)mbiencoder_dealloc, /* tp_dealloc */ |
| 1013 | 0, /* tp_print */ |
| 1014 | 0, /* tp_getattr */ |
| 1015 | 0, /* tp_setattr */ |
| 1016 | 0, /* tp_reserved */ |
| 1017 | 0, /* tp_repr */ |
| 1018 | 0, /* tp_as_number */ |
| 1019 | 0, /* tp_as_sequence */ |
| 1020 | 0, /* tp_as_mapping */ |
| 1021 | 0, /* tp_hash */ |
| 1022 | 0, /* tp_call */ |
| 1023 | 0, /* tp_str */ |
| 1024 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1025 | 0, /* tp_setattro */ |
| 1026 | 0, /* tp_as_buffer */ |
| 1027 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1028 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1029 | 0, /* tp_doc */ |
| 1030 | (traverseproc)mbiencoder_traverse, /* tp_traverse */ |
| 1031 | 0, /* tp_clear */ |
| 1032 | 0, /* tp_richcompare */ |
| 1033 | 0, /* tp_weaklistoffset */ |
| 1034 | 0, /* tp_iter */ |
| 1035 | 0, /* tp_iterext */ |
| 1036 | mbiencoder_methods, /* tp_methods */ |
| 1037 | 0, /* tp_members */ |
| 1038 | codecctx_getsets, /* tp_getset */ |
| 1039 | 0, /* tp_base */ |
| 1040 | 0, /* tp_dict */ |
| 1041 | 0, /* tp_descr_get */ |
| 1042 | 0, /* tp_descr_set */ |
| 1043 | 0, /* tp_dictoffset */ |
| 1044 | mbiencoder_init, /* tp_init */ |
| 1045 | 0, /* tp_alloc */ |
| 1046 | mbiencoder_new, /* tp_new */ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1047 | }; |
| 1048 | |
| 1049 | |
| 1050 | /** |
| 1051 | * MultibyteIncrementalDecoder object |
| 1052 | */ |
| 1053 | |
| 1054 | static PyObject * |
| 1055 | mbidecoder_decode(MultibyteIncrementalDecoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1056 | PyObject *args, PyObject *kwargs) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1057 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1058 | MultibyteDecodeBuffer buf; |
| 1059 | char *data, *wdata = NULL; |
| 1060 | Py_buffer pdata; |
| 1061 | Py_ssize_t wsize, finalsize = 0, size, origpending; |
| 1062 | int final = 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1063 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1064 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "y*|i:decode", |
| 1065 | incrementalkwarglist, &pdata, &final)) |
| 1066 | return NULL; |
| 1067 | data = pdata.buf; |
| 1068 | size = pdata.len; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1069 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1070 | buf.outobj = buf.excobj = NULL; |
| 1071 | origpending = self->pendingsize; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1072 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1073 | if (self->pendingsize == 0) { |
| 1074 | wsize = size; |
| 1075 | wdata = data; |
| 1076 | } |
| 1077 | else { |
| 1078 | if (size > PY_SSIZE_T_MAX - self->pendingsize) { |
| 1079 | PyErr_NoMemory(); |
| 1080 | goto errorexit; |
| 1081 | } |
| 1082 | wsize = size + self->pendingsize; |
| 1083 | wdata = PyMem_Malloc(wsize); |
| 1084 | if (wdata == NULL) |
| 1085 | goto errorexit; |
| 1086 | memcpy(wdata, self->pending, self->pendingsize); |
| 1087 | memcpy(wdata + self->pendingsize, data, size); |
| 1088 | self->pendingsize = 0; |
| 1089 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1090 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1091 | if (decoder_prepare_buffer(&buf, wdata, wsize) != 0) |
| 1092 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1093 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1094 | if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf)) |
| 1095 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1096 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1097 | if (final && buf.inbuf < buf.inbuf_end) { |
| 1098 | if (multibytecodec_decerror(self->codec, &self->state, |
| 1099 | &buf, self->errors, MBERR_TOOFEW)) { |
| 1100 | /* recover the original pending buffer */ |
| 1101 | memcpy(self->pending, wdata, origpending); |
| 1102 | self->pendingsize = origpending; |
| 1103 | goto errorexit; |
| 1104 | } |
| 1105 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1106 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1107 | if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */ |
| 1108 | if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0) |
| 1109 | goto errorexit; |
| 1110 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1111 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1112 | finalsize = (Py_ssize_t)(buf.outbuf - PyUnicode_AS_UNICODE(buf.outobj)); |
| 1113 | if (finalsize != PyUnicode_GET_SIZE(buf.outobj)) |
| 1114 | if (PyUnicode_Resize(&buf.outobj, finalsize) == -1) |
| 1115 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1116 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1117 | PyBuffer_Release(&pdata); |
| 1118 | if (wdata != data) |
| 1119 | PyMem_Del(wdata); |
| 1120 | Py_XDECREF(buf.excobj); |
| 1121 | return buf.outobj; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1122 | |
| 1123 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1124 | PyBuffer_Release(&pdata); |
| 1125 | if (wdata != NULL && wdata != data) |
| 1126 | PyMem_Del(wdata); |
| 1127 | Py_XDECREF(buf.excobj); |
| 1128 | Py_XDECREF(buf.outobj); |
| 1129 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1130 | } |
| 1131 | |
| 1132 | static PyObject * |
| 1133 | mbidecoder_reset(MultibyteIncrementalDecoderObject *self) |
| 1134 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1135 | if (self->codec->decreset != NULL && |
| 1136 | self->codec->decreset(&self->state, self->codec->config) != 0) |
| 1137 | return NULL; |
| 1138 | self->pendingsize = 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1139 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1140 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1141 | } |
| 1142 | |
| 1143 | static struct PyMethodDef mbidecoder_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1144 | {"decode", (PyCFunction)mbidecoder_decode, |
| 1145 | METH_VARARGS | METH_KEYWORDS, NULL}, |
| 1146 | {"reset", (PyCFunction)mbidecoder_reset, |
| 1147 | METH_NOARGS, NULL}, |
| 1148 | {NULL, NULL}, |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1149 | }; |
| 1150 | |
| 1151 | static PyObject * |
| 1152 | mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1153 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1154 | MultibyteIncrementalDecoderObject *self; |
| 1155 | PyObject *codec = NULL; |
| 1156 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1157 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1158 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder", |
| 1159 | incnewkwarglist, &errors)) |
| 1160 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1161 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1162 | self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0); |
| 1163 | if (self == NULL) |
| 1164 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1165 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1166 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1167 | if (codec == NULL) |
| 1168 | goto errorexit; |
| 1169 | if (!MultibyteCodec_Check(codec)) { |
| 1170 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1171 | goto errorexit; |
| 1172 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1173 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1174 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1175 | self->pendingsize = 0; |
| 1176 | self->errors = internal_error_callback(errors); |
| 1177 | if (self->errors == NULL) |
| 1178 | goto errorexit; |
| 1179 | if (self->codec->decinit != NULL && |
| 1180 | self->codec->decinit(&self->state, self->codec->config) != 0) |
| 1181 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1182 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1183 | Py_DECREF(codec); |
| 1184 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1185 | |
| 1186 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1187 | Py_XDECREF(self); |
| 1188 | Py_XDECREF(codec); |
| 1189 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1190 | } |
| 1191 | |
| 1192 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1193 | mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1194 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1195 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1196 | } |
| 1197 | |
| 1198 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1199 | mbidecoder_traverse(MultibyteIncrementalDecoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1200 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1201 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1202 | if (ERROR_ISCUSTOM(self->errors)) |
| 1203 | Py_VISIT(self->errors); |
| 1204 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1205 | } |
| 1206 | |
| 1207 | static void |
| 1208 | mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self) |
| 1209 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1210 | PyObject_GC_UnTrack(self); |
| 1211 | ERROR_DECREF(self->errors); |
| 1212 | Py_TYPE(self)->tp_free(self); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1213 | } |
| 1214 | |
| 1215 | static PyTypeObject MultibyteIncrementalDecoder_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1216 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1217 | "MultibyteIncrementalDecoder", /* tp_name */ |
| 1218 | sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */ |
| 1219 | 0, /* tp_itemsize */ |
| 1220 | /* methods */ |
| 1221 | (destructor)mbidecoder_dealloc, /* tp_dealloc */ |
| 1222 | 0, /* tp_print */ |
| 1223 | 0, /* tp_getattr */ |
| 1224 | 0, /* tp_setattr */ |
| 1225 | 0, /* tp_reserved */ |
| 1226 | 0, /* tp_repr */ |
| 1227 | 0, /* tp_as_number */ |
| 1228 | 0, /* tp_as_sequence */ |
| 1229 | 0, /* tp_as_mapping */ |
| 1230 | 0, /* tp_hash */ |
| 1231 | 0, /* tp_call */ |
| 1232 | 0, /* tp_str */ |
| 1233 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1234 | 0, /* tp_setattro */ |
| 1235 | 0, /* tp_as_buffer */ |
| 1236 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1237 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1238 | 0, /* tp_doc */ |
| 1239 | (traverseproc)mbidecoder_traverse, /* tp_traverse */ |
| 1240 | 0, /* tp_clear */ |
| 1241 | 0, /* tp_richcompare */ |
| 1242 | 0, /* tp_weaklistoffset */ |
| 1243 | 0, /* tp_iter */ |
| 1244 | 0, /* tp_iterext */ |
| 1245 | mbidecoder_methods, /* tp_methods */ |
| 1246 | 0, /* tp_members */ |
| 1247 | codecctx_getsets, /* tp_getset */ |
| 1248 | 0, /* tp_base */ |
| 1249 | 0, /* tp_dict */ |
| 1250 | 0, /* tp_descr_get */ |
| 1251 | 0, /* tp_descr_set */ |
| 1252 | 0, /* tp_dictoffset */ |
| 1253 | mbidecoder_init, /* tp_init */ |
| 1254 | 0, /* tp_alloc */ |
| 1255 | mbidecoder_new, /* tp_new */ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1256 | }; |
| 1257 | |
| 1258 | |
| 1259 | /** |
| 1260 | * MultibyteStreamReader object |
| 1261 | */ |
| 1262 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1263 | static PyObject * |
| 1264 | mbstreamreader_iread(MultibyteStreamReaderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1265 | const char *method, Py_ssize_t sizehint) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1266 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1267 | MultibyteDecodeBuffer buf; |
| 1268 | PyObject *cres; |
| 1269 | Py_ssize_t rsize, finalsize = 0; |
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 | if (sizehint == 0) |
Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 1272 | return PyUnicode_New(0, 0); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1273 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1274 | buf.outobj = buf.excobj = NULL; |
| 1275 | cres = NULL; |
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 | for (;;) { |
| 1278 | int endoffile; |
Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1279 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1280 | if (sizehint < 0) |
| 1281 | cres = PyObject_CallMethod(self->stream, |
| 1282 | (char *)method, NULL); |
| 1283 | else |
| 1284 | cres = PyObject_CallMethod(self->stream, |
| 1285 | (char *)method, "i", sizehint); |
| 1286 | if (cres == NULL) |
| 1287 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1288 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1289 | if (!PyBytes_Check(cres)) { |
| 1290 | PyErr_Format(PyExc_TypeError, |
| 1291 | "stream function returned a " |
| 1292 | "non-bytes object (%.100s)", |
| 1293 | cres->ob_type->tp_name); |
| 1294 | goto errorexit; |
| 1295 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1296 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1297 | endoffile = (PyBytes_GET_SIZE(cres) == 0); |
Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1298 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1299 | if (self->pendingsize > 0) { |
| 1300 | PyObject *ctr; |
| 1301 | char *ctrdata; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1302 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1303 | if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) { |
| 1304 | PyErr_NoMemory(); |
| 1305 | goto errorexit; |
| 1306 | } |
| 1307 | rsize = PyBytes_GET_SIZE(cres) + self->pendingsize; |
| 1308 | ctr = PyBytes_FromStringAndSize(NULL, rsize); |
| 1309 | if (ctr == NULL) |
| 1310 | goto errorexit; |
| 1311 | ctrdata = PyBytes_AS_STRING(ctr); |
| 1312 | memcpy(ctrdata, self->pending, self->pendingsize); |
| 1313 | memcpy(ctrdata + self->pendingsize, |
| 1314 | PyBytes_AS_STRING(cres), |
| 1315 | PyBytes_GET_SIZE(cres)); |
| 1316 | Py_DECREF(cres); |
| 1317 | cres = ctr; |
| 1318 | self->pendingsize = 0; |
| 1319 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1320 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1321 | rsize = PyBytes_GET_SIZE(cres); |
| 1322 | if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres), |
| 1323 | rsize) != 0) |
| 1324 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1325 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1326 | if (rsize > 0 && decoder_feed_buffer( |
| 1327 | (MultibyteStatefulDecoderContext *)self, &buf)) |
| 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 | if (endoffile || sizehint < 0) { |
| 1331 | if (buf.inbuf < buf.inbuf_end && |
| 1332 | multibytecodec_decerror(self->codec, &self->state, |
| 1333 | &buf, self->errors, MBERR_TOOFEW)) |
| 1334 | goto errorexit; |
| 1335 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1336 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1337 | if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */ |
| 1338 | if (decoder_append_pending(STATEFUL_DCTX(self), |
| 1339 | &buf) != 0) |
| 1340 | goto errorexit; |
| 1341 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1342 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1343 | finalsize = (Py_ssize_t)(buf.outbuf - |
| 1344 | PyUnicode_AS_UNICODE(buf.outobj)); |
| 1345 | Py_DECREF(cres); |
| 1346 | cres = NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1347 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1348 | if (sizehint < 0 || finalsize != 0 || rsize == 0) |
| 1349 | break; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1350 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1351 | sizehint = 1; /* read 1 more byte and retry */ |
| 1352 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1353 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1354 | if (finalsize != PyUnicode_GET_SIZE(buf.outobj)) |
| 1355 | if (PyUnicode_Resize(&buf.outobj, finalsize) == -1) |
| 1356 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1357 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1358 | Py_XDECREF(cres); |
| 1359 | Py_XDECREF(buf.excobj); |
| 1360 | return buf.outobj; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1361 | |
| 1362 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1363 | Py_XDECREF(cres); |
| 1364 | Py_XDECREF(buf.excobj); |
| 1365 | Py_XDECREF(buf.outobj); |
| 1366 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1367 | } |
| 1368 | |
| 1369 | static PyObject * |
| 1370 | mbstreamreader_read(MultibyteStreamReaderObject *self, PyObject *args) |
| 1371 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1372 | PyObject *sizeobj = NULL; |
| 1373 | Py_ssize_t size; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1374 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1375 | if (!PyArg_UnpackTuple(args, "read", 0, 1, &sizeobj)) |
| 1376 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1377 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1378 | if (sizeobj == Py_None || sizeobj == NULL) |
| 1379 | size = -1; |
| 1380 | else if (PyLong_Check(sizeobj)) |
| 1381 | size = PyLong_AsSsize_t(sizeobj); |
| 1382 | else { |
| 1383 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1384 | return NULL; |
| 1385 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1386 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1387 | if (size == -1 && PyErr_Occurred()) |
| 1388 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1389 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1390 | return mbstreamreader_iread(self, "read", size); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1391 | } |
| 1392 | |
| 1393 | static PyObject * |
| 1394 | mbstreamreader_readline(MultibyteStreamReaderObject *self, PyObject *args) |
| 1395 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1396 | PyObject *sizeobj = NULL; |
| 1397 | Py_ssize_t size; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1398 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1399 | if (!PyArg_UnpackTuple(args, "readline", 0, 1, &sizeobj)) |
| 1400 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1401 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1402 | if (sizeobj == Py_None || sizeobj == NULL) |
| 1403 | size = -1; |
| 1404 | else if (PyLong_Check(sizeobj)) |
| 1405 | size = PyLong_AsSsize_t(sizeobj); |
| 1406 | else { |
| 1407 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1408 | return NULL; |
| 1409 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1410 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1411 | if (size == -1 && PyErr_Occurred()) |
| 1412 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1413 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1414 | return mbstreamreader_iread(self, "readline", size); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1415 | } |
| 1416 | |
| 1417 | static PyObject * |
| 1418 | mbstreamreader_readlines(MultibyteStreamReaderObject *self, PyObject *args) |
| 1419 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1420 | PyObject *sizehintobj = NULL, *r, *sr; |
| 1421 | Py_ssize_t sizehint; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1422 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1423 | if (!PyArg_UnpackTuple(args, "readlines", 0, 1, &sizehintobj)) |
| 1424 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1425 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1426 | if (sizehintobj == Py_None || sizehintobj == NULL) |
| 1427 | sizehint = -1; |
| 1428 | else if (PyLong_Check(sizehintobj)) |
| 1429 | sizehint = PyLong_AsSsize_t(sizehintobj); |
| 1430 | else { |
| 1431 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1432 | return NULL; |
| 1433 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1434 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1435 | if (sizehint == -1 && PyErr_Occurred()) |
| 1436 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1437 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1438 | r = mbstreamreader_iread(self, "read", sizehint); |
| 1439 | if (r == NULL) |
| 1440 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1441 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1442 | sr = PyUnicode_Splitlines(r, 1); |
| 1443 | Py_DECREF(r); |
| 1444 | return sr; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1445 | } |
| 1446 | |
| 1447 | static PyObject * |
| 1448 | mbstreamreader_reset(MultibyteStreamReaderObject *self) |
| 1449 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1450 | if (self->codec->decreset != NULL && |
| 1451 | self->codec->decreset(&self->state, self->codec->config) != 0) |
| 1452 | return NULL; |
| 1453 | self->pendingsize = 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1454 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1455 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1456 | } |
| 1457 | |
| 1458 | static struct PyMethodDef mbstreamreader_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1459 | {"read", (PyCFunction)mbstreamreader_read, |
| 1460 | METH_VARARGS, NULL}, |
| 1461 | {"readline", (PyCFunction)mbstreamreader_readline, |
| 1462 | METH_VARARGS, NULL}, |
| 1463 | {"readlines", (PyCFunction)mbstreamreader_readlines, |
| 1464 | METH_VARARGS, NULL}, |
| 1465 | {"reset", (PyCFunction)mbstreamreader_reset, |
| 1466 | METH_NOARGS, NULL}, |
| 1467 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1468 | }; |
| 1469 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1470 | static PyMemberDef mbstreamreader_members[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1471 | {"stream", T_OBJECT, |
| 1472 | offsetof(MultibyteStreamReaderObject, stream), |
| 1473 | READONLY, NULL}, |
| 1474 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1475 | }; |
| 1476 | |
| 1477 | static PyObject * |
| 1478 | mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1479 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1480 | MultibyteStreamReaderObject *self; |
| 1481 | PyObject *stream, *codec = NULL; |
| 1482 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1483 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1484 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader", |
| 1485 | streamkwarglist, &stream, &errors)) |
| 1486 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1487 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1488 | self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0); |
| 1489 | if (self == NULL) |
| 1490 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1491 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1492 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1493 | if (codec == NULL) |
| 1494 | goto errorexit; |
| 1495 | if (!MultibyteCodec_Check(codec)) { |
| 1496 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1497 | goto errorexit; |
| 1498 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1499 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1500 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1501 | self->stream = stream; |
| 1502 | Py_INCREF(stream); |
| 1503 | self->pendingsize = 0; |
| 1504 | self->errors = internal_error_callback(errors); |
| 1505 | if (self->errors == NULL) |
| 1506 | goto errorexit; |
| 1507 | if (self->codec->decinit != NULL && |
| 1508 | self->codec->decinit(&self->state, self->codec->config) != 0) |
| 1509 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1510 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1511 | Py_DECREF(codec); |
| 1512 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1513 | |
| 1514 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1515 | Py_XDECREF(self); |
| 1516 | Py_XDECREF(codec); |
| 1517 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1518 | } |
| 1519 | |
| 1520 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1521 | mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1522 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1523 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1524 | } |
| 1525 | |
| 1526 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1527 | mbstreamreader_traverse(MultibyteStreamReaderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1528 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1529 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1530 | if (ERROR_ISCUSTOM(self->errors)) |
| 1531 | Py_VISIT(self->errors); |
| 1532 | Py_VISIT(self->stream); |
| 1533 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1534 | } |
| 1535 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1536 | static void |
| 1537 | mbstreamreader_dealloc(MultibyteStreamReaderObject *self) |
| 1538 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1539 | PyObject_GC_UnTrack(self); |
| 1540 | ERROR_DECREF(self->errors); |
| 1541 | Py_XDECREF(self->stream); |
| 1542 | Py_TYPE(self)->tp_free(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1543 | } |
| 1544 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1545 | static PyTypeObject MultibyteStreamReader_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1546 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1547 | "MultibyteStreamReader", /* tp_name */ |
| 1548 | sizeof(MultibyteStreamReaderObject), /* tp_basicsize */ |
| 1549 | 0, /* tp_itemsize */ |
| 1550 | /* methods */ |
| 1551 | (destructor)mbstreamreader_dealloc, /* tp_dealloc */ |
| 1552 | 0, /* tp_print */ |
| 1553 | 0, /* tp_getattr */ |
| 1554 | 0, /* tp_setattr */ |
| 1555 | 0, /* tp_reserved */ |
| 1556 | 0, /* tp_repr */ |
| 1557 | 0, /* tp_as_number */ |
| 1558 | 0, /* tp_as_sequence */ |
| 1559 | 0, /* tp_as_mapping */ |
| 1560 | 0, /* tp_hash */ |
| 1561 | 0, /* tp_call */ |
| 1562 | 0, /* tp_str */ |
| 1563 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1564 | 0, /* tp_setattro */ |
| 1565 | 0, /* tp_as_buffer */ |
| 1566 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1567 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1568 | 0, /* tp_doc */ |
| 1569 | (traverseproc)mbstreamreader_traverse, /* tp_traverse */ |
| 1570 | 0, /* tp_clear */ |
| 1571 | 0, /* tp_richcompare */ |
| 1572 | 0, /* tp_weaklistoffset */ |
| 1573 | 0, /* tp_iter */ |
| 1574 | 0, /* tp_iterext */ |
| 1575 | mbstreamreader_methods, /* tp_methods */ |
| 1576 | mbstreamreader_members, /* tp_members */ |
| 1577 | codecctx_getsets, /* tp_getset */ |
| 1578 | 0, /* tp_base */ |
| 1579 | 0, /* tp_dict */ |
| 1580 | 0, /* tp_descr_get */ |
| 1581 | 0, /* tp_descr_set */ |
| 1582 | 0, /* tp_dictoffset */ |
| 1583 | mbstreamreader_init, /* tp_init */ |
| 1584 | 0, /* tp_alloc */ |
| 1585 | mbstreamreader_new, /* tp_new */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1586 | }; |
| 1587 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1588 | |
| 1589 | /** |
| 1590 | * MultibyteStreamWriter object |
| 1591 | */ |
| 1592 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1593 | static int |
| 1594 | mbstreamwriter_iwrite(MultibyteStreamWriterObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1595 | PyObject *unistr) |
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 *str, *wr; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1598 | _Py_IDENTIFIER(write); |
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 | str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0); |
| 1601 | if (str == NULL) |
| 1602 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1603 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1604 | wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1605 | Py_DECREF(str); |
| 1606 | if (wr == NULL) |
| 1607 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1608 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1609 | Py_DECREF(wr); |
| 1610 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1611 | } |
| 1612 | |
| 1613 | static PyObject * |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1614 | mbstreamwriter_write(MultibyteStreamWriterObject *self, PyObject *strobj) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1615 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1616 | if (mbstreamwriter_iwrite(self, strobj)) |
| 1617 | return NULL; |
| 1618 | else |
| 1619 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1620 | } |
| 1621 | |
| 1622 | static PyObject * |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1623 | mbstreamwriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1624 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1625 | PyObject *strobj; |
| 1626 | int i, r; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1627 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1628 | if (!PySequence_Check(lines)) { |
| 1629 | PyErr_SetString(PyExc_TypeError, |
| 1630 | "arg must be a sequence object"); |
| 1631 | return NULL; |
| 1632 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1633 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1634 | for (i = 0; i < PySequence_Length(lines); i++) { |
| 1635 | /* length can be changed even within this loop */ |
| 1636 | strobj = PySequence_GetItem(lines, i); |
| 1637 | if (strobj == NULL) |
| 1638 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1639 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1640 | r = mbstreamwriter_iwrite(self, strobj); |
| 1641 | Py_DECREF(strobj); |
| 1642 | if (r == -1) |
| 1643 | return NULL; |
| 1644 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1645 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1646 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1647 | } |
| 1648 | |
| 1649 | static PyObject * |
| 1650 | mbstreamwriter_reset(MultibyteStreamWriterObject *self) |
| 1651 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1652 | const Py_UNICODE *pending; |
| 1653 | PyObject *pwrt; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1654 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1655 | pending = self->pending; |
| 1656 | pwrt = multibytecodec_encode(self->codec, &self->state, |
| 1657 | &pending, self->pendingsize, self->errors, |
| 1658 | MBENC_FLUSH | MBENC_RESET); |
| 1659 | /* some pending buffer can be truncated when UnicodeEncodeError is |
| 1660 | * raised on 'strict' mode. but, 'reset' method is designed to |
| 1661 | * reset the pending buffer or states so failed string sequence |
| 1662 | * ought to be missed */ |
| 1663 | self->pendingsize = 0; |
| 1664 | if (pwrt == NULL) |
| 1665 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1666 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1667 | assert(PyBytes_Check(pwrt)); |
| 1668 | if (PyBytes_Size(pwrt) > 0) { |
| 1669 | PyObject *wr; |
Martin v. Löwis | bd928fe | 2011-10-14 10:20:37 +0200 | [diff] [blame] | 1670 | _Py_IDENTIFIER(write); |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1671 | |
| 1672 | wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1673 | if (wr == NULL) { |
| 1674 | Py_DECREF(pwrt); |
| 1675 | return NULL; |
| 1676 | } |
| 1677 | } |
| 1678 | Py_DECREF(pwrt); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1679 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1680 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1681 | } |
| 1682 | |
| 1683 | static PyObject * |
| 1684 | mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1685 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1686 | MultibyteStreamWriterObject *self; |
| 1687 | PyObject *stream, *codec = NULL; |
| 1688 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1689 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1690 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter", |
| 1691 | streamkwarglist, &stream, &errors)) |
| 1692 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1693 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1694 | self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0); |
| 1695 | if (self == NULL) |
| 1696 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1697 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1698 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1699 | if (codec == NULL) |
| 1700 | goto errorexit; |
| 1701 | if (!MultibyteCodec_Check(codec)) { |
| 1702 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1703 | goto errorexit; |
| 1704 | } |
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 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1707 | self->stream = stream; |
| 1708 | Py_INCREF(stream); |
| 1709 | self->pendingsize = 0; |
| 1710 | self->errors = internal_error_callback(errors); |
| 1711 | if (self->errors == NULL) |
| 1712 | goto errorexit; |
| 1713 | if (self->codec->encinit != NULL && |
| 1714 | self->codec->encinit(&self->state, self->codec->config) != 0) |
| 1715 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1716 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1717 | Py_DECREF(codec); |
| 1718 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1719 | |
| 1720 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1721 | Py_XDECREF(self); |
| 1722 | Py_XDECREF(codec); |
| 1723 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1724 | } |
| 1725 | |
| 1726 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1727 | mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1728 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1729 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1730 | } |
| 1731 | |
| 1732 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1733 | mbstreamwriter_traverse(MultibyteStreamWriterObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1734 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1735 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1736 | if (ERROR_ISCUSTOM(self->errors)) |
| 1737 | Py_VISIT(self->errors); |
| 1738 | Py_VISIT(self->stream); |
| 1739 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1740 | } |
| 1741 | |
| 1742 | static void |
| 1743 | mbstreamwriter_dealloc(MultibyteStreamWriterObject *self) |
| 1744 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1745 | PyObject_GC_UnTrack(self); |
| 1746 | ERROR_DECREF(self->errors); |
| 1747 | Py_XDECREF(self->stream); |
| 1748 | Py_TYPE(self)->tp_free(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1749 | } |
| 1750 | |
| 1751 | static struct PyMethodDef mbstreamwriter_methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1752 | {"write", (PyCFunction)mbstreamwriter_write, |
| 1753 | METH_O, NULL}, |
| 1754 | {"writelines", (PyCFunction)mbstreamwriter_writelines, |
| 1755 | METH_O, NULL}, |
| 1756 | {"reset", (PyCFunction)mbstreamwriter_reset, |
| 1757 | METH_NOARGS, NULL}, |
| 1758 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1759 | }; |
| 1760 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1761 | static PyMemberDef mbstreamwriter_members[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1762 | {"stream", T_OBJECT, |
| 1763 | offsetof(MultibyteStreamWriterObject, stream), |
| 1764 | READONLY, NULL}, |
| 1765 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1766 | }; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1767 | |
| 1768 | static PyTypeObject MultibyteStreamWriter_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1769 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1770 | "MultibyteStreamWriter", /* tp_name */ |
| 1771 | sizeof(MultibyteStreamWriterObject), /* tp_basicsize */ |
| 1772 | 0, /* tp_itemsize */ |
| 1773 | /* methods */ |
| 1774 | (destructor)mbstreamwriter_dealloc, /* tp_dealloc */ |
| 1775 | 0, /* tp_print */ |
| 1776 | 0, /* tp_getattr */ |
| 1777 | 0, /* tp_setattr */ |
| 1778 | 0, /* tp_reserved */ |
| 1779 | 0, /* tp_repr */ |
| 1780 | 0, /* tp_as_number */ |
| 1781 | 0, /* tp_as_sequence */ |
| 1782 | 0, /* tp_as_mapping */ |
| 1783 | 0, /* tp_hash */ |
| 1784 | 0, /* tp_call */ |
| 1785 | 0, /* tp_str */ |
| 1786 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1787 | 0, /* tp_setattro */ |
| 1788 | 0, /* tp_as_buffer */ |
| 1789 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1790 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1791 | 0, /* tp_doc */ |
| 1792 | (traverseproc)mbstreamwriter_traverse, /* tp_traverse */ |
| 1793 | 0, /* tp_clear */ |
| 1794 | 0, /* tp_richcompare */ |
| 1795 | 0, /* tp_weaklistoffset */ |
| 1796 | 0, /* tp_iter */ |
| 1797 | 0, /* tp_iterext */ |
| 1798 | mbstreamwriter_methods, /* tp_methods */ |
| 1799 | mbstreamwriter_members, /* tp_members */ |
| 1800 | codecctx_getsets, /* tp_getset */ |
| 1801 | 0, /* tp_base */ |
| 1802 | 0, /* tp_dict */ |
| 1803 | 0, /* tp_descr_get */ |
| 1804 | 0, /* tp_descr_set */ |
| 1805 | 0, /* tp_dictoffset */ |
| 1806 | mbstreamwriter_init, /* tp_init */ |
| 1807 | 0, /* tp_alloc */ |
| 1808 | mbstreamwriter_new, /* tp_new */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1809 | }; |
| 1810 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1811 | |
| 1812 | /** |
| 1813 | * Exposed factory function |
| 1814 | */ |
| 1815 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1816 | static PyObject * |
| 1817 | __create_codec(PyObject *ignore, PyObject *arg) |
| 1818 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1819 | MultibyteCodecObject *self; |
| 1820 | MultibyteCodec *codec; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1821 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1822 | if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) { |
| 1823 | PyErr_SetString(PyExc_ValueError, "argument type invalid"); |
| 1824 | return NULL; |
| 1825 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1826 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1827 | codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME); |
| 1828 | if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0) |
| 1829 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1830 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1831 | self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type); |
| 1832 | if (self == NULL) |
| 1833 | return NULL; |
| 1834 | self->codec = codec; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1835 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1836 | return (PyObject *)self; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1837 | } |
| 1838 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1839 | static struct PyMethodDef __methods[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1840 | {"__create_codec", (PyCFunction)__create_codec, METH_O}, |
| 1841 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1842 | }; |
| 1843 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1844 | |
| 1845 | static struct PyModuleDef _multibytecodecmodule = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1846 | PyModuleDef_HEAD_INIT, |
| 1847 | "_multibytecodec", |
| 1848 | NULL, |
| 1849 | -1, |
| 1850 | __methods, |
| 1851 | NULL, |
| 1852 | NULL, |
| 1853 | NULL, |
| 1854 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1855 | }; |
| 1856 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1857 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1858 | PyInit__multibytecodec(void) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1859 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1860 | int i; |
| 1861 | PyObject *m; |
| 1862 | PyTypeObject *typelist[] = { |
| 1863 | &MultibyteIncrementalEncoder_Type, |
| 1864 | &MultibyteIncrementalDecoder_Type, |
| 1865 | &MultibyteStreamReader_Type, |
| 1866 | &MultibyteStreamWriter_Type, |
| 1867 | NULL |
| 1868 | }; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1869 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1870 | if (PyType_Ready(&MultibyteCodec_Type) < 0) |
| 1871 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1872 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1873 | m = PyModule_Create(&_multibytecodecmodule); |
| 1874 | if (m == NULL) |
| 1875 | return NULL; |
Neal Norwitz | 058bde1 | 2005-09-21 06:44:25 +0000 | [diff] [blame] | 1876 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1877 | for (i = 0; typelist[i] != NULL; i++) { |
| 1878 | if (PyType_Ready(typelist[i]) < 0) |
| 1879 | return NULL; |
| 1880 | Py_INCREF(typelist[i]); |
| 1881 | PyModule_AddObject(m, typelist[i]->tp_name, |
| 1882 | (PyObject *)typelist[i]); |
| 1883 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1884 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1885 | if (PyErr_Occurred()) { |
| 1886 | Py_FatalError("can't initialize the _multibytecodec module"); |
| 1887 | Py_DECREF(m); |
| 1888 | m = NULL; |
| 1889 | } |
| 1890 | return m; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1891 | } |