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" |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 11 | #include "clinic/multibytecodec.c.h" |
| 12 | |
| 13 | /*[clinic input] |
| 14 | output preset file |
| 15 | module _multibytecodec |
| 16 | [clinic start generated code]*/ |
| 17 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=e0cf1b7f3c472d17]*/ |
| 18 | |
| 19 | /*[clinic input] |
| 20 | class _multibytecodec.MultibyteCodec "MultibyteCodecObject *" "&MultibyteCodec_Type" |
| 21 | [clinic start generated code]*/ |
| 22 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=d5b1fc1fec8eb003]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 23 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 24 | typedef struct { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 25 | PyObject *inobj; |
| 26 | Py_ssize_t inpos, inlen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 27 | unsigned char *outbuf, *outbuf_end; |
| 28 | PyObject *excobj, *outobj; |
| 29 | } MultibyteEncodeBuffer; |
| 30 | |
| 31 | typedef struct { |
| 32 | const unsigned char *inbuf, *inbuf_top, *inbuf_end; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 33 | PyObject *excobj; |
| 34 | _PyUnicodeWriter writer; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 35 | } MultibyteDecodeBuffer; |
| 36 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 37 | static char *incnewkwarglist[] = {"errors", NULL}; |
Martin v. Löwis | 02cbf4a | 2006-02-27 17:20:04 +0000 | [diff] [blame] | 38 | static char *streamkwarglist[] = {"stream", "errors", NULL}; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 39 | |
| 40 | static PyObject *multibytecodec_encode(MultibyteCodec *, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 41 | MultibyteCodec_State *, PyObject *, Py_ssize_t *, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 42 | PyObject *, int); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 43 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 44 | #define MBENC_RESET MBENC_MAX<<1 /* reset after an encoding session */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 45 | |
Victor Stinner | 3f36a57 | 2013-11-12 21:39:02 +0100 | [diff] [blame] | 46 | _Py_IDENTIFIER(write); |
| 47 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 48 | static PyObject * |
Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 49 | make_tuple(PyObject *object, Py_ssize_t len) |
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 | PyObject *v, *w; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 52 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 53 | if (object == NULL) |
| 54 | return NULL; |
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 | v = PyTuple_New(2); |
| 57 | if (v == NULL) { |
| 58 | Py_DECREF(object); |
| 59 | return NULL; |
| 60 | } |
| 61 | PyTuple_SET_ITEM(v, 0, object); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 62 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 63 | w = PyLong_FromSsize_t(len); |
| 64 | if (w == NULL) { |
| 65 | Py_DECREF(v); |
| 66 | return NULL; |
| 67 | } |
| 68 | PyTuple_SET_ITEM(v, 1, w); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 69 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 70 | return v; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 71 | } |
| 72 | |
| 73 | static PyObject * |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 74 | internal_error_callback(const char *errors) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 75 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 76 | if (errors == NULL || strcmp(errors, "strict") == 0) |
| 77 | return ERROR_STRICT; |
| 78 | else if (strcmp(errors, "ignore") == 0) |
| 79 | return ERROR_IGNORE; |
| 80 | else if (strcmp(errors, "replace") == 0) |
| 81 | return ERROR_REPLACE; |
| 82 | else |
| 83 | return PyUnicode_FromString(errors); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 84 | } |
| 85 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 86 | static PyObject * |
| 87 | call_error_callback(PyObject *errors, PyObject *exc) |
| 88 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 89 | PyObject *args, *cb, *r; |
| 90 | const char *str; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 91 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 92 | assert(PyUnicode_Check(errors)); |
| 93 | str = _PyUnicode_AsString(errors); |
| 94 | if (str == NULL) |
| 95 | return NULL; |
| 96 | cb = PyCodec_LookupError(str); |
| 97 | if (cb == NULL) |
| 98 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 99 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 100 | args = PyTuple_New(1); |
| 101 | if (args == NULL) { |
| 102 | Py_DECREF(cb); |
| 103 | return NULL; |
| 104 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 105 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 106 | PyTuple_SET_ITEM(args, 0, exc); |
| 107 | Py_INCREF(exc); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 108 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 109 | r = PyObject_CallObject(cb, args); |
| 110 | Py_DECREF(args); |
| 111 | Py_DECREF(cb); |
| 112 | return r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 113 | } |
| 114 | |
| 115 | static PyObject * |
| 116 | codecctx_errors_get(MultibyteStatefulCodecContext *self) |
| 117 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 118 | const char *errors; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 119 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 120 | if (self->errors == ERROR_STRICT) |
| 121 | errors = "strict"; |
| 122 | else if (self->errors == ERROR_IGNORE) |
| 123 | errors = "ignore"; |
| 124 | else if (self->errors == ERROR_REPLACE) |
| 125 | errors = "replace"; |
| 126 | else { |
| 127 | Py_INCREF(self->errors); |
| 128 | return self->errors; |
| 129 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 130 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 131 | return PyUnicode_FromString(errors); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 132 | } |
| 133 | |
| 134 | static int |
| 135 | codecctx_errors_set(MultibyteStatefulCodecContext *self, PyObject *value, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 136 | void *closure) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 137 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 138 | PyObject *cb; |
| 139 | const char *str; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 140 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 141 | if (!PyUnicode_Check(value)) { |
| 142 | PyErr_SetString(PyExc_TypeError, "errors must be a string"); |
| 143 | return -1; |
| 144 | } |
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 | str = _PyUnicode_AsString(value); |
| 147 | if (str == NULL) |
| 148 | return -1; |
Neal Norwitz | 6ea45d3 | 2007-08-26 04:19:43 +0000 | [diff] [blame] | 149 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 150 | cb = internal_error_callback(str); |
| 151 | if (cb == NULL) |
| 152 | return -1; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 153 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 154 | ERROR_DECREF(self->errors); |
| 155 | self->errors = cb; |
| 156 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 157 | } |
| 158 | |
| 159 | /* This getset handlers list is used by all the stateful codec objects */ |
| 160 | static PyGetSetDef codecctx_getsets[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 161 | {"errors", (getter)codecctx_errors_get, |
| 162 | (setter)codecctx_errors_set, |
| 163 | PyDoc_STR("how to treat errors")}, |
| 164 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 165 | }; |
| 166 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 167 | static int |
Hye-Shik Chang | 4b96c13 | 2006-03-04 16:08:19 +0000 | [diff] [blame] | 168 | expand_encodebuffer(MultibyteEncodeBuffer *buf, Py_ssize_t esize) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 169 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 170 | Py_ssize_t orgpos, orgsize, incsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 171 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 172 | orgpos = (Py_ssize_t)((char *)buf->outbuf - |
| 173 | PyBytes_AS_STRING(buf->outobj)); |
| 174 | orgsize = PyBytes_GET_SIZE(buf->outobj); |
| 175 | incsize = (esize < (orgsize >> 1) ? (orgsize >> 1) | 1 : esize); |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 176 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 177 | if (orgsize > PY_SSIZE_T_MAX - incsize) |
| 178 | return -1; |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 179 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 180 | if (_PyBytes_Resize(&buf->outobj, orgsize + incsize) == -1) |
| 181 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 182 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 183 | buf->outbuf = (unsigned char *)PyBytes_AS_STRING(buf->outobj) +orgpos; |
| 184 | buf->outbuf_end = (unsigned char *)PyBytes_AS_STRING(buf->outobj) |
| 185 | + PyBytes_GET_SIZE(buf->outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 186 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 187 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 188 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 189 | #define REQUIRE_ENCODEBUFFER(buf, s) { \ |
| 190 | if ((s) < 1 || (buf)->outbuf + (s) > (buf)->outbuf_end) \ |
| 191 | if (expand_encodebuffer(buf, s) == -1) \ |
| 192 | goto errorexit; \ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 193 | } |
| 194 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 195 | |
| 196 | /** |
| 197 | * MultibyteCodec object |
| 198 | */ |
| 199 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 200 | static int |
| 201 | multibytecodec_encerror(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 202 | MultibyteCodec_State *state, |
| 203 | MultibyteEncodeBuffer *buf, |
| 204 | PyObject *errors, Py_ssize_t e) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 205 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 206 | PyObject *retobj = NULL, *retstr = NULL, *tobj; |
| 207 | Py_ssize_t retstrsize, newpos; |
| 208 | Py_ssize_t esize, start, end; |
| 209 | const char *reason; |
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 | if (e > 0) { |
| 212 | reason = "illegal multibyte sequence"; |
| 213 | esize = e; |
| 214 | } |
| 215 | else { |
| 216 | switch (e) { |
| 217 | case MBERR_TOOSMALL: |
| 218 | REQUIRE_ENCODEBUFFER(buf, -1); |
| 219 | return 0; /* retry it */ |
| 220 | case MBERR_TOOFEW: |
| 221 | reason = "incomplete multibyte sequence"; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 222 | esize = (Py_ssize_t)buf->inpos; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 223 | break; |
| 224 | case MBERR_INTERNAL: |
| 225 | PyErr_SetString(PyExc_RuntimeError, |
| 226 | "internal codec error"); |
| 227 | return -1; |
| 228 | default: |
| 229 | PyErr_SetString(PyExc_RuntimeError, |
| 230 | "unknown runtime error"); |
| 231 | return -1; |
| 232 | } |
| 233 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 234 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 235 | if (errors == ERROR_REPLACE) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 236 | PyObject *replchar; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 237 | Py_ssize_t r; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 238 | Py_ssize_t inpos; |
| 239 | int kind; |
| 240 | void *data; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 241 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 242 | replchar = PyUnicode_FromOrdinal('?'); |
| 243 | if (replchar == NULL) |
| 244 | goto errorexit; |
| 245 | kind = PyUnicode_KIND(replchar); |
| 246 | data = PyUnicode_DATA(replchar); |
| 247 | |
| 248 | inpos = 0; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 249 | for (;;) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 250 | Py_ssize_t outleft = (Py_ssize_t)(buf->outbuf_end - buf->outbuf); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 251 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 252 | r = codec->encode(state, codec->config, |
| 253 | kind, data, &inpos, 1, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 254 | &buf->outbuf, outleft, 0); |
| 255 | if (r == MBERR_TOOSMALL) { |
| 256 | REQUIRE_ENCODEBUFFER(buf, -1); |
| 257 | continue; |
| 258 | } |
| 259 | else |
| 260 | break; |
| 261 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 262 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 263 | Py_DECREF(replchar); |
| 264 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 265 | if (r != 0) { |
| 266 | REQUIRE_ENCODEBUFFER(buf, 1); |
| 267 | *buf->outbuf++ = '?'; |
| 268 | } |
| 269 | } |
| 270 | if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 271 | buf->inpos += esize; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 272 | return 0; |
| 273 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 274 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 275 | start = (Py_ssize_t)buf->inpos; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 276 | end = start + esize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 277 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 278 | /* use cached exception object if available */ |
| 279 | if (buf->excobj == NULL) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 280 | buf->excobj = PyObject_CallFunction(PyExc_UnicodeEncodeError, |
| 281 | "sOnns", |
| 282 | codec->encoding, buf->inobj, |
| 283 | start, end, reason); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 284 | if (buf->excobj == NULL) |
| 285 | goto errorexit; |
| 286 | } |
| 287 | else |
| 288 | if (PyUnicodeEncodeError_SetStart(buf->excobj, start) != 0 || |
| 289 | PyUnicodeEncodeError_SetEnd(buf->excobj, end) != 0 || |
| 290 | PyUnicodeEncodeError_SetReason(buf->excobj, reason) != 0) |
| 291 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 292 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 293 | if (errors == ERROR_STRICT) { |
| 294 | PyCodec_StrictErrors(buf->excobj); |
| 295 | goto errorexit; |
| 296 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 297 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 298 | retobj = call_error_callback(errors, buf->excobj); |
| 299 | if (retobj == NULL) |
| 300 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 301 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 302 | if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || |
Benjamin Peterson | 47a00f3 | 2012-12-02 11:20:28 -0500 | [diff] [blame] | 303 | (!PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) && !PyBytes_Check(tobj)) || |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 304 | !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { |
| 305 | PyErr_SetString(PyExc_TypeError, |
| 306 | "encoding error handler must return " |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 307 | "(str, int) tuple"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 308 | goto errorexit; |
| 309 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 310 | |
Benjamin Peterson | 47a00f3 | 2012-12-02 11:20:28 -0500 | [diff] [blame] | 311 | if (PyUnicode_Check(tobj)) { |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 312 | Py_ssize_t inpos; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 313 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 314 | retstr = multibytecodec_encode(codec, state, tobj, |
| 315 | &inpos, ERROR_STRICT, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 316 | MBENC_FLUSH); |
| 317 | if (retstr == NULL) |
| 318 | goto errorexit; |
| 319 | } |
Benjamin Peterson | 47a00f3 | 2012-12-02 11:20:28 -0500 | [diff] [blame] | 320 | else { |
| 321 | Py_INCREF(tobj); |
| 322 | retstr = tobj; |
| 323 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 324 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 325 | assert(PyBytes_Check(retstr)); |
| 326 | retstrsize = PyBytes_GET_SIZE(retstr); |
| 327 | REQUIRE_ENCODEBUFFER(buf, retstrsize); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 328 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 329 | memcpy(buf->outbuf, PyBytes_AS_STRING(retstr), retstrsize); |
| 330 | buf->outbuf += retstrsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 331 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 332 | newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); |
| 333 | if (newpos < 0 && !PyErr_Occurred()) |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 334 | newpos += (Py_ssize_t)buf->inlen; |
| 335 | if (newpos < 0 || newpos > buf->inlen) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 336 | PyErr_Clear(); |
| 337 | PyErr_Format(PyExc_IndexError, |
| 338 | "position %zd from error handler out of bounds", |
| 339 | newpos); |
| 340 | goto errorexit; |
| 341 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 342 | buf->inpos = newpos; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 343 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 344 | Py_DECREF(retobj); |
| 345 | Py_DECREF(retstr); |
| 346 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 347 | |
| 348 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 349 | Py_XDECREF(retobj); |
| 350 | Py_XDECREF(retstr); |
| 351 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 352 | } |
| 353 | |
| 354 | static int |
| 355 | multibytecodec_decerror(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 356 | MultibyteCodec_State *state, |
| 357 | MultibyteDecodeBuffer *buf, |
| 358 | PyObject *errors, Py_ssize_t e) |
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 | PyObject *retobj = NULL, *retuni = NULL; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 361 | Py_ssize_t newpos; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 362 | const char *reason; |
| 363 | Py_ssize_t esize, start, end; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 364 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 365 | if (e > 0) { |
| 366 | reason = "illegal multibyte sequence"; |
| 367 | esize = e; |
| 368 | } |
| 369 | else { |
| 370 | switch (e) { |
| 371 | case MBERR_TOOSMALL: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 372 | return 0; /* retry it */ |
| 373 | case MBERR_TOOFEW: |
| 374 | reason = "incomplete multibyte sequence"; |
| 375 | esize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
| 376 | break; |
| 377 | case MBERR_INTERNAL: |
| 378 | PyErr_SetString(PyExc_RuntimeError, |
| 379 | "internal codec error"); |
| 380 | return -1; |
Victor Stinner | d1f9942 | 2013-07-16 21:41:43 +0200 | [diff] [blame] | 381 | case MBERR_EXCEPTION: |
| 382 | return -1; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 383 | default: |
| 384 | PyErr_SetString(PyExc_RuntimeError, |
| 385 | "unknown runtime error"); |
| 386 | return -1; |
| 387 | } |
| 388 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 389 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 390 | if (errors == ERROR_REPLACE) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 391 | if (_PyUnicodeWriter_WriteChar(&buf->writer, |
| 392 | Py_UNICODE_REPLACEMENT_CHARACTER) < 0) |
| 393 | goto errorexit; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 394 | } |
| 395 | if (errors == ERROR_IGNORE || errors == ERROR_REPLACE) { |
| 396 | buf->inbuf += esize; |
| 397 | return 0; |
| 398 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 399 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 400 | start = (Py_ssize_t)(buf->inbuf - buf->inbuf_top); |
| 401 | end = start + esize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 402 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 403 | /* use cached exception object if available */ |
| 404 | if (buf->excobj == NULL) { |
| 405 | buf->excobj = PyUnicodeDecodeError_Create(codec->encoding, |
| 406 | (const char *)buf->inbuf_top, |
| 407 | (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top), |
| 408 | start, end, reason); |
| 409 | if (buf->excobj == NULL) |
| 410 | goto errorexit; |
| 411 | } |
| 412 | else |
| 413 | if (PyUnicodeDecodeError_SetStart(buf->excobj, start) || |
| 414 | PyUnicodeDecodeError_SetEnd(buf->excobj, end) || |
| 415 | PyUnicodeDecodeError_SetReason(buf->excobj, reason)) |
| 416 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 417 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 418 | if (errors == ERROR_STRICT) { |
| 419 | PyCodec_StrictErrors(buf->excobj); |
| 420 | goto errorexit; |
| 421 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 422 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 423 | retobj = call_error_callback(errors, buf->excobj); |
| 424 | if (retobj == NULL) |
| 425 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 426 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 427 | if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || |
| 428 | !PyUnicode_Check((retuni = PyTuple_GET_ITEM(retobj, 0))) || |
| 429 | !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { |
| 430 | PyErr_SetString(PyExc_TypeError, |
| 431 | "decoding error handler must return " |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 432 | "(str, int) tuple"); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 433 | goto errorexit; |
| 434 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 435 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 436 | if (_PyUnicodeWriter_WriteStr(&buf->writer, retuni) < 0) |
Victor Stinner | 4eea849 | 2011-11-21 03:01:27 +0100 | [diff] [blame] | 437 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 438 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 439 | newpos = PyLong_AsSsize_t(PyTuple_GET_ITEM(retobj, 1)); |
| 440 | if (newpos < 0 && !PyErr_Occurred()) |
| 441 | newpos += (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top); |
| 442 | if (newpos < 0 || buf->inbuf_top + newpos > buf->inbuf_end) { |
| 443 | PyErr_Clear(); |
| 444 | PyErr_Format(PyExc_IndexError, |
| 445 | "position %zd from error handler out of bounds", |
| 446 | newpos); |
| 447 | goto errorexit; |
| 448 | } |
| 449 | buf->inbuf = buf->inbuf_top + newpos; |
| 450 | Py_DECREF(retobj); |
| 451 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 452 | |
| 453 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 454 | Py_XDECREF(retobj); |
| 455 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 456 | } |
| 457 | |
| 458 | static PyObject * |
| 459 | multibytecodec_encode(MultibyteCodec *codec, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 460 | MultibyteCodec_State *state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 461 | PyObject *text, Py_ssize_t *inpos_t, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 462 | PyObject *errors, int flags) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 463 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 464 | MultibyteEncodeBuffer buf; |
| 465 | Py_ssize_t finalsize, r = 0; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 466 | Py_ssize_t datalen; |
| 467 | int kind; |
| 468 | void *data; |
| 469 | |
| 470 | if (PyUnicode_READY(text) < 0) |
| 471 | return NULL; |
| 472 | datalen = PyUnicode_GET_LENGTH(text); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 473 | |
Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 474 | if (datalen == 0 && !(flags & MBENC_RESET)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 475 | return PyBytes_FromStringAndSize(NULL, 0); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 476 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 477 | buf.excobj = NULL; |
Brett Cannon | b94767f | 2011-02-22 20:15:44 +0000 | [diff] [blame] | 478 | buf.outobj = NULL; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 479 | buf.inobj = text; /* borrowed reference */ |
| 480 | buf.inpos = 0; |
| 481 | buf.inlen = datalen; |
| 482 | kind = PyUnicode_KIND(buf.inobj); |
| 483 | data = PyUnicode_DATA(buf.inobj); |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 484 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 485 | if (datalen > (PY_SSIZE_T_MAX - 16) / 2) { |
| 486 | PyErr_NoMemory(); |
| 487 | goto errorexit; |
| 488 | } |
Amaury Forgeot d'Arc | 9c74b14 | 2008-06-18 00:47:36 +0000 | [diff] [blame] | 489 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 490 | buf.outobj = PyBytes_FromStringAndSize(NULL, datalen * 2 + 16); |
| 491 | if (buf.outobj == NULL) |
| 492 | goto errorexit; |
| 493 | buf.outbuf = (unsigned char *)PyBytes_AS_STRING(buf.outobj); |
| 494 | buf.outbuf_end = buf.outbuf + PyBytes_GET_SIZE(buf.outobj); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 495 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 496 | while (buf.inpos < buf.inlen) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 497 | /* we don't reuse inleft and outleft here. |
| 498 | * error callbacks can relocate the cursor anywhere on buffer*/ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 499 | Py_ssize_t outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
| 500 | |
| 501 | r = codec->encode(state, codec->config, |
| 502 | kind, data, |
| 503 | &buf.inpos, buf.inlen, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 504 | &buf.outbuf, outleft, flags); |
| 505 | if ((r == 0) || (r == MBERR_TOOFEW && !(flags & MBENC_FLUSH))) |
| 506 | break; |
| 507 | else if (multibytecodec_encerror(codec, state, &buf, errors,r)) |
| 508 | goto errorexit; |
| 509 | else if (r == MBERR_TOOFEW) |
| 510 | break; |
| 511 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 512 | |
Victor Stinner | 6bcbef7 | 2011-05-24 22:17:55 +0200 | [diff] [blame] | 513 | if (codec->encreset != NULL && (flags & MBENC_RESET)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 514 | for (;;) { |
| 515 | Py_ssize_t outleft; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 516 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 517 | outleft = (Py_ssize_t)(buf.outbuf_end - buf.outbuf); |
| 518 | r = codec->encreset(state, codec->config, &buf.outbuf, |
| 519 | outleft); |
| 520 | if (r == 0) |
| 521 | break; |
| 522 | else if (multibytecodec_encerror(codec, state, |
| 523 | &buf, errors, r)) |
| 524 | goto errorexit; |
| 525 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 526 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 527 | finalsize = (Py_ssize_t)((char *)buf.outbuf - |
| 528 | PyBytes_AS_STRING(buf.outobj)); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 529 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 530 | if (finalsize != PyBytes_GET_SIZE(buf.outobj)) |
| 531 | if (_PyBytes_Resize(&buf.outobj, finalsize) == -1) |
| 532 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 533 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 534 | if (inpos_t) |
| 535 | *inpos_t = buf.inpos; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 536 | Py_XDECREF(buf.excobj); |
| 537 | return buf.outobj; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 538 | |
| 539 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 540 | Py_XDECREF(buf.excobj); |
| 541 | Py_XDECREF(buf.outobj); |
| 542 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 543 | } |
| 544 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 545 | /*[clinic input] |
| 546 | _multibytecodec.MultibyteCodec.encode |
| 547 | |
| 548 | input: object |
| 549 | errors: str(nullable=True) = NULL |
| 550 | |
| 551 | Return an encoded string version of `input'. |
| 552 | |
| 553 | 'errors' may be given to set a different error handling scheme. Default is |
| 554 | 'strict' meaning that encoding errors raise a UnicodeEncodeError. Other possible |
| 555 | values are 'ignore', 'replace' and 'xmlcharrefreplace' as well as any other name |
| 556 | registered with codecs.register_error that can handle UnicodeEncodeErrors. |
| 557 | [clinic start generated code]*/ |
| 558 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 559 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 560 | _multibytecodec_MultibyteCodec_encode_impl(MultibyteCodecObject *self, PyObject *input, const char *errors) |
| 561 | /*[clinic end generated code: output=a36bfa08783a0d0b input=252e7ee695867b2d]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 562 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 563 | MultibyteCodec_State state; |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 564 | PyObject *errorcb, *r, *ucvt; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 565 | Py_ssize_t datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 566 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 567 | if (PyUnicode_Check(input)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 568 | ucvt = NULL; |
| 569 | else { |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 570 | input = ucvt = PyObject_Str(input); |
| 571 | if (input == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 572 | return NULL; |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 573 | else if (!PyUnicode_Check(input)) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 574 | PyErr_SetString(PyExc_TypeError, |
| 575 | "couldn't convert the object to unicode."); |
| 576 | Py_DECREF(ucvt); |
| 577 | return NULL; |
| 578 | } |
| 579 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 580 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 581 | if (PyUnicode_READY(input) < 0) { |
Victor Stinner | 9a80fab | 2011-11-21 02:50:14 +0100 | [diff] [blame] | 582 | Py_XDECREF(ucvt); |
| 583 | return NULL; |
| 584 | } |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 585 | datalen = PyUnicode_GET_LENGTH(input); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 586 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 587 | errorcb = internal_error_callback(errors); |
| 588 | if (errorcb == NULL) { |
| 589 | Py_XDECREF(ucvt); |
| 590 | return NULL; |
| 591 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 592 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 593 | if (self->codec->encinit != NULL && |
| 594 | self->codec->encinit(&state, self->codec->config) != 0) |
| 595 | goto errorexit; |
| 596 | r = multibytecodec_encode(self->codec, &state, |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 597 | input, NULL, errorcb, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 598 | MBENC_FLUSH | MBENC_RESET); |
| 599 | if (r == NULL) |
| 600 | goto errorexit; |
Hye-Shik Chang | 2bb146f | 2004-07-18 03:06:29 +0000 | [diff] [blame] | 601 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 602 | ERROR_DECREF(errorcb); |
| 603 | Py_XDECREF(ucvt); |
| 604 | return make_tuple(r, datalen); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 605 | |
| 606 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 607 | ERROR_DECREF(errorcb); |
| 608 | Py_XDECREF(ucvt); |
| 609 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 610 | } |
| 611 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 612 | /*[clinic input] |
| 613 | _multibytecodec.MultibyteCodec.decode |
| 614 | |
| 615 | input: Py_buffer |
| 616 | errors: str(nullable=True) = NULL |
| 617 | |
| 618 | Decodes 'input'. |
| 619 | |
| 620 | 'errors' may be given to set a different error handling scheme. Default is |
| 621 | 'strict' meaning that encoding errors raise a UnicodeDecodeError. Other possible |
| 622 | values are 'ignore' and 'replace' as well as any other name registered with |
| 623 | codecs.register_error that is able to handle UnicodeDecodeErrors." |
| 624 | [clinic start generated code]*/ |
| 625 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 626 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 627 | _multibytecodec_MultibyteCodec_decode_impl(MultibyteCodecObject *self, Py_buffer *input, const char *errors) |
| 628 | /*[clinic end generated code: output=4c8ee8b2931b014e input=37e1d9236e3ce8f3]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 629 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 630 | MultibyteCodec_State state; |
| 631 | MultibyteDecodeBuffer buf; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 632 | PyObject *errorcb, *res; |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 633 | const char *data; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 634 | Py_ssize_t datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 635 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 636 | data = input->buf; |
| 637 | datalen = input->len; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 638 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 639 | errorcb = internal_error_callback(errors); |
| 640 | if (errorcb == NULL) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 641 | return NULL; |
| 642 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 643 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 644 | if (datalen == 0) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 645 | ERROR_DECREF(errorcb); |
Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 646 | return make_tuple(PyUnicode_New(0, 0), 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 647 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 648 | |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 649 | _PyUnicodeWriter_Init(&buf.writer); |
| 650 | buf.writer.min_length = datalen; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 651 | buf.excobj = NULL; |
| 652 | buf.inbuf = buf.inbuf_top = (unsigned char *)data; |
| 653 | buf.inbuf_end = buf.inbuf_top + datalen; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 654 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 655 | if (self->codec->decinit != NULL && |
| 656 | self->codec->decinit(&state, self->codec->config) != 0) |
| 657 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 658 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 659 | while (buf.inbuf < buf.inbuf_end) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 660 | Py_ssize_t inleft, r; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 661 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 662 | inleft = (Py_ssize_t)(buf.inbuf_end - buf.inbuf); |
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, |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 665 | &buf.inbuf, inleft, &buf.writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 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 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 673 | res = _PyUnicodeWriter_Finish(&buf.writer); |
| 674 | if (res == NULL) |
| 675 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 676 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 677 | Py_XDECREF(buf.excobj); |
| 678 | ERROR_DECREF(errorcb); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 679 | return make_tuple(res, datalen); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 680 | |
| 681 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 682 | ERROR_DECREF(errorcb); |
| 683 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 684 | _PyUnicodeWriter_Dealloc(&buf.writer); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 685 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 686 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 687 | } |
| 688 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 689 | static struct PyMethodDef multibytecodec_methods[] = { |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 690 | _MULTIBYTECODEC_MULTIBYTECODEC_ENCODE_METHODDEF |
| 691 | _MULTIBYTECODEC_MULTIBYTECODEC_DECODE_METHODDEF |
| 692 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 693 | }; |
| 694 | |
| 695 | static void |
| 696 | multibytecodec_dealloc(MultibyteCodecObject *self) |
| 697 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 698 | PyObject_Del(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 699 | } |
| 700 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 701 | static PyTypeObject MultibyteCodec_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 702 | PyVarObject_HEAD_INIT(NULL, 0) |
| 703 | "MultibyteCodec", /* tp_name */ |
| 704 | sizeof(MultibyteCodecObject), /* tp_basicsize */ |
| 705 | 0, /* tp_itemsize */ |
| 706 | /* methods */ |
| 707 | (destructor)multibytecodec_dealloc, /* tp_dealloc */ |
| 708 | 0, /* tp_print */ |
| 709 | 0, /* tp_getattr */ |
| 710 | 0, /* tp_setattr */ |
| 711 | 0, /* tp_reserved */ |
| 712 | 0, /* tp_repr */ |
| 713 | 0, /* tp_as_number */ |
| 714 | 0, /* tp_as_sequence */ |
| 715 | 0, /* tp_as_mapping */ |
| 716 | 0, /* tp_hash */ |
| 717 | 0, /* tp_call */ |
| 718 | 0, /* tp_str */ |
| 719 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 720 | 0, /* tp_setattro */ |
| 721 | 0, /* tp_as_buffer */ |
| 722 | Py_TPFLAGS_DEFAULT, /* tp_flags */ |
| 723 | 0, /* tp_doc */ |
| 724 | 0, /* tp_traverse */ |
| 725 | 0, /* tp_clear */ |
| 726 | 0, /* tp_richcompare */ |
| 727 | 0, /* tp_weaklistoffset */ |
| 728 | 0, /* tp_iter */ |
| 729 | 0, /* tp_iterext */ |
| 730 | multibytecodec_methods, /* tp_methods */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 731 | }; |
| 732 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 733 | |
| 734 | /** |
| 735 | * Utility functions for stateful codec mechanism |
| 736 | */ |
| 737 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 738 | #define STATEFUL_DCTX(o) ((MultibyteStatefulDecoderContext *)(o)) |
| 739 | #define STATEFUL_ECTX(o) ((MultibyteStatefulEncoderContext *)(o)) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 740 | |
| 741 | static PyObject * |
| 742 | encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 743 | PyObject *unistr, int final) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 744 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 745 | PyObject *ucvt, *r = NULL; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 746 | PyObject *inbuf = NULL; |
| 747 | Py_ssize_t inpos, datalen; |
| 748 | PyObject *origpending = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 749 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 750 | if (PyUnicode_Check(unistr)) |
| 751 | ucvt = NULL; |
| 752 | else { |
| 753 | unistr = ucvt = PyObject_Str(unistr); |
| 754 | if (unistr == NULL) |
| 755 | return NULL; |
| 756 | else if (!PyUnicode_Check(unistr)) { |
| 757 | PyErr_SetString(PyExc_TypeError, |
Benjamin Peterson | aff4723 | 2012-12-02 10:53:41 -0500 | [diff] [blame] | 758 | "couldn't convert the object to str."); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 759 | Py_DECREF(ucvt); |
| 760 | return NULL; |
| 761 | } |
| 762 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 763 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 764 | if (ctx->pending) { |
| 765 | PyObject *inbuf_tmp; |
| 766 | |
| 767 | Py_INCREF(ctx->pending); |
| 768 | origpending = ctx->pending; |
| 769 | |
| 770 | Py_INCREF(ctx->pending); |
| 771 | inbuf_tmp = ctx->pending; |
| 772 | PyUnicode_Append(&inbuf_tmp, unistr); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 773 | if (inbuf_tmp == NULL) |
| 774 | goto errorexit; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 775 | Py_CLEAR(ctx->pending); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 776 | inbuf = inbuf_tmp; |
| 777 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 778 | else { |
| 779 | origpending = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 780 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 781 | Py_INCREF(unistr); |
| 782 | inbuf = unistr; |
| 783 | } |
| 784 | if (PyUnicode_READY(inbuf) < 0) |
| 785 | goto errorexit; |
| 786 | inpos = 0; |
| 787 | datalen = PyUnicode_GET_LENGTH(inbuf); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 788 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 789 | r = multibytecodec_encode(ctx->codec, &ctx->state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 790 | inbuf, &inpos, |
| 791 | ctx->errors, final ? MBENC_FLUSH | MBENC_RESET : 0); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 792 | if (r == NULL) { |
| 793 | /* recover the original pending buffer */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 794 | Py_CLEAR(ctx->pending); |
| 795 | ctx->pending = origpending; |
| 796 | origpending = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 797 | goto errorexit; |
| 798 | } |
Victor Stinner | 322cc74 | 2013-04-14 18:11:41 +0200 | [diff] [blame] | 799 | Py_XDECREF(origpending); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 800 | |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 801 | if (inpos < datalen) { |
| 802 | if (datalen - inpos > MAXENCPENDING) { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 803 | /* normal codecs can't reach here */ |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 804 | PyErr_SetString(PyExc_UnicodeError, |
| 805 | "pending buffer overflow"); |
| 806 | goto errorexit; |
| 807 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 808 | ctx->pending = PyUnicode_Substring(inbuf, inpos, datalen); |
| 809 | if (ctx->pending == NULL) { |
| 810 | /* normal codecs can't reach here */ |
| 811 | goto errorexit; |
| 812 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 813 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 814 | |
Victor Stinner | 322cc74 | 2013-04-14 18:11:41 +0200 | [diff] [blame] | 815 | Py_DECREF(inbuf); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 816 | Py_XDECREF(ucvt); |
| 817 | return r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 818 | |
| 819 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 820 | Py_XDECREF(r); |
| 821 | Py_XDECREF(ucvt); |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 822 | Py_XDECREF(origpending); |
Victor Stinner | 322cc74 | 2013-04-14 18:11:41 +0200 | [diff] [blame] | 823 | Py_XDECREF(inbuf); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 824 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 825 | } |
| 826 | |
| 827 | static int |
| 828 | decoder_append_pending(MultibyteStatefulDecoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 829 | MultibyteDecodeBuffer *buf) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 830 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 831 | Py_ssize_t npendings; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 832 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 833 | npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
| 834 | if (npendings + ctx->pendingsize > MAXDECPENDING || |
| 835 | npendings > PY_SSIZE_T_MAX - ctx->pendingsize) { |
| 836 | PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow"); |
| 837 | return -1; |
| 838 | } |
| 839 | memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings); |
| 840 | ctx->pendingsize += npendings; |
| 841 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 842 | } |
| 843 | |
| 844 | static int |
| 845 | decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 846 | Py_ssize_t size) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 847 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 848 | buf->inbuf = buf->inbuf_top = (const unsigned char *)data; |
| 849 | buf->inbuf_end = buf->inbuf_top + size; |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 850 | buf->writer.min_length += size; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 851 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 852 | } |
| 853 | |
| 854 | static int |
| 855 | decoder_feed_buffer(MultibyteStatefulDecoderContext *ctx, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 856 | MultibyteDecodeBuffer *buf) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 857 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 858 | while (buf->inbuf < buf->inbuf_end) { |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 859 | Py_ssize_t inleft; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 860 | Py_ssize_t r; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 861 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 862 | inleft = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 863 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 864 | r = ctx->codec->decode(&ctx->state, ctx->codec->config, |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 865 | &buf->inbuf, inleft, &buf->writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 866 | if (r == 0 || r == MBERR_TOOFEW) |
| 867 | break; |
| 868 | else if (multibytecodec_decerror(ctx->codec, &ctx->state, |
| 869 | buf, ctx->errors, r)) |
| 870 | return -1; |
| 871 | } |
| 872 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 873 | } |
| 874 | |
| 875 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 876 | /*[clinic input] |
| 877 | class _multibytecodec.MultibyteIncrementalEncoder "MultibyteIncrementalEncoderObject *" "&MultibyteIncrementalEncoder_Type" |
| 878 | [clinic start generated code]*/ |
| 879 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=3be82909cd08924d]*/ |
| 880 | |
| 881 | /*[clinic input] |
| 882 | _multibytecodec.MultibyteIncrementalEncoder.encode |
| 883 | |
| 884 | input: object |
| 885 | final: int = 0 |
| 886 | [clinic start generated code]*/ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 887 | |
| 888 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 889 | _multibytecodec_MultibyteIncrementalEncoder_encode_impl(MultibyteIncrementalEncoderObject *self, PyObject *input, int final) |
| 890 | /*[clinic end generated code: output=3cd8780c8a719bbf input=456b76d73e464661]*/ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 891 | { |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 892 | return encoder_encode_stateful(STATEFUL_ECTX(self), input, final); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 893 | } |
| 894 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 895 | /*[clinic input] |
| 896 | _multibytecodec.MultibyteIncrementalEncoder.reset |
| 897 | [clinic start generated code]*/ |
| 898 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 899 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 900 | _multibytecodec_MultibyteIncrementalEncoder_reset_impl(MultibyteIncrementalEncoderObject *self) |
| 901 | /*[clinic end generated code: output=b4125d8f537a253f input=930f06760707b6ea]*/ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 902 | { |
Victor Stinner | e15dce3 | 2011-05-30 22:56:00 +0200 | [diff] [blame] | 903 | /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */ |
| 904 | unsigned char buffer[4], *outbuf; |
| 905 | Py_ssize_t r; |
| 906 | if (self->codec->encreset != NULL) { |
| 907 | outbuf = buffer; |
| 908 | r = self->codec->encreset(&self->state, self->codec->config, |
| 909 | &outbuf, sizeof(buffer)); |
| 910 | if (r != 0) |
| 911 | return NULL; |
| 912 | } |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 913 | Py_CLEAR(self->pending); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 914 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 915 | } |
| 916 | |
| 917 | static struct PyMethodDef mbiencoder_methods[] = { |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 918 | _MULTIBYTECODEC_MULTIBYTEINCREMENTALENCODER_ENCODE_METHODDEF |
| 919 | _MULTIBYTECODEC_MULTIBYTEINCREMENTALENCODER_RESET_METHODDEF |
| 920 | {NULL, NULL}, |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 921 | }; |
| 922 | |
| 923 | static PyObject * |
| 924 | mbiencoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 925 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 926 | MultibyteIncrementalEncoderObject *self; |
| 927 | PyObject *codec = NULL; |
| 928 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 929 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 930 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalEncoder", |
| 931 | incnewkwarglist, &errors)) |
| 932 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 933 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 934 | self = (MultibyteIncrementalEncoderObject *)type->tp_alloc(type, 0); |
| 935 | if (self == NULL) |
| 936 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 937 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 938 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 939 | if (codec == NULL) |
| 940 | goto errorexit; |
| 941 | if (!MultibyteCodec_Check(codec)) { |
| 942 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 943 | goto errorexit; |
| 944 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 945 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 946 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 947 | self->pending = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 948 | self->errors = internal_error_callback(errors); |
| 949 | if (self->errors == NULL) |
| 950 | goto errorexit; |
| 951 | if (self->codec->encinit != NULL && |
| 952 | self->codec->encinit(&self->state, self->codec->config) != 0) |
| 953 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 954 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 955 | Py_DECREF(codec); |
| 956 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 957 | |
| 958 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 959 | Py_XDECREF(self); |
| 960 | Py_XDECREF(codec); |
| 961 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 962 | } |
| 963 | |
| 964 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 965 | mbiencoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 966 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 967 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 968 | } |
| 969 | |
| 970 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 971 | mbiencoder_traverse(MultibyteIncrementalEncoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 972 | visitproc visit, void *arg) |
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 | if (ERROR_ISCUSTOM(self->errors)) |
| 975 | Py_VISIT(self->errors); |
| 976 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 977 | } |
| 978 | |
| 979 | static void |
| 980 | mbiencoder_dealloc(MultibyteIncrementalEncoderObject *self) |
| 981 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 982 | PyObject_GC_UnTrack(self); |
| 983 | ERROR_DECREF(self->errors); |
| 984 | Py_TYPE(self)->tp_free(self); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 985 | } |
| 986 | |
| 987 | static PyTypeObject MultibyteIncrementalEncoder_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 988 | PyVarObject_HEAD_INIT(NULL, 0) |
| 989 | "MultibyteIncrementalEncoder", /* tp_name */ |
| 990 | sizeof(MultibyteIncrementalEncoderObject), /* tp_basicsize */ |
| 991 | 0, /* tp_itemsize */ |
| 992 | /* methods */ |
| 993 | (destructor)mbiencoder_dealloc, /* tp_dealloc */ |
| 994 | 0, /* tp_print */ |
| 995 | 0, /* tp_getattr */ |
| 996 | 0, /* tp_setattr */ |
| 997 | 0, /* tp_reserved */ |
| 998 | 0, /* tp_repr */ |
| 999 | 0, /* tp_as_number */ |
| 1000 | 0, /* tp_as_sequence */ |
| 1001 | 0, /* tp_as_mapping */ |
| 1002 | 0, /* tp_hash */ |
| 1003 | 0, /* tp_call */ |
| 1004 | 0, /* tp_str */ |
| 1005 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1006 | 0, /* tp_setattro */ |
| 1007 | 0, /* tp_as_buffer */ |
| 1008 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1009 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1010 | 0, /* tp_doc */ |
| 1011 | (traverseproc)mbiencoder_traverse, /* tp_traverse */ |
| 1012 | 0, /* tp_clear */ |
| 1013 | 0, /* tp_richcompare */ |
| 1014 | 0, /* tp_weaklistoffset */ |
| 1015 | 0, /* tp_iter */ |
| 1016 | 0, /* tp_iterext */ |
| 1017 | mbiencoder_methods, /* tp_methods */ |
| 1018 | 0, /* tp_members */ |
| 1019 | codecctx_getsets, /* tp_getset */ |
| 1020 | 0, /* tp_base */ |
| 1021 | 0, /* tp_dict */ |
| 1022 | 0, /* tp_descr_get */ |
| 1023 | 0, /* tp_descr_set */ |
| 1024 | 0, /* tp_dictoffset */ |
| 1025 | mbiencoder_init, /* tp_init */ |
| 1026 | 0, /* tp_alloc */ |
| 1027 | mbiencoder_new, /* tp_new */ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1028 | }; |
| 1029 | |
| 1030 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1031 | /*[clinic input] |
| 1032 | class _multibytecodec.MultibyteIncrementalDecoder "MultibyteIncrementalDecoderObject *" "&MultibyteIncrementalDecoder_Type" |
| 1033 | [clinic start generated code]*/ |
| 1034 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=f6003faaf2cea692]*/ |
| 1035 | |
| 1036 | /*[clinic input] |
| 1037 | _multibytecodec.MultibyteIncrementalDecoder.decode |
| 1038 | |
| 1039 | input: Py_buffer |
| 1040 | final: int = 0 |
| 1041 | [clinic start generated code]*/ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1042 | |
| 1043 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1044 | _multibytecodec_MultibyteIncrementalDecoder_decode_impl(MultibyteIncrementalDecoderObject *self, Py_buffer *input, int final) |
| 1045 | /*[clinic end generated code: output=a0f3f92aa7303cf7 input=eb18c2f6e83589e1]*/ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1046 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1047 | MultibyteDecodeBuffer buf; |
| 1048 | char *data, *wdata = NULL; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1049 | Py_ssize_t wsize, size, origpending; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1050 | PyObject *res; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1051 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1052 | data = input->buf; |
| 1053 | size = input->len; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1054 | |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 1055 | _PyUnicodeWriter_Init(&buf.writer); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1056 | buf.excobj = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1057 | origpending = self->pendingsize; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1058 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1059 | if (self->pendingsize == 0) { |
| 1060 | wsize = size; |
| 1061 | wdata = data; |
| 1062 | } |
| 1063 | else { |
| 1064 | if (size > PY_SSIZE_T_MAX - self->pendingsize) { |
| 1065 | PyErr_NoMemory(); |
| 1066 | goto errorexit; |
| 1067 | } |
| 1068 | wsize = size + self->pendingsize; |
| 1069 | wdata = PyMem_Malloc(wsize); |
Victor Stinner | 33283ba | 2013-07-15 17:47:39 +0200 | [diff] [blame] | 1070 | if (wdata == NULL) { |
| 1071 | PyErr_NoMemory(); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1072 | goto errorexit; |
Victor Stinner | 33283ba | 2013-07-15 17:47:39 +0200 | [diff] [blame] | 1073 | } |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1074 | memcpy(wdata, self->pending, self->pendingsize); |
| 1075 | memcpy(wdata + self->pendingsize, data, size); |
| 1076 | self->pendingsize = 0; |
| 1077 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1078 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1079 | if (decoder_prepare_buffer(&buf, wdata, wsize) != 0) |
| 1080 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1081 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1082 | if (decoder_feed_buffer(STATEFUL_DCTX(self), &buf)) |
| 1083 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1084 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1085 | if (final && buf.inbuf < buf.inbuf_end) { |
| 1086 | if (multibytecodec_decerror(self->codec, &self->state, |
| 1087 | &buf, self->errors, MBERR_TOOFEW)) { |
| 1088 | /* recover the original pending buffer */ |
| 1089 | memcpy(self->pending, wdata, origpending); |
| 1090 | self->pendingsize = origpending; |
| 1091 | goto errorexit; |
| 1092 | } |
| 1093 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1094 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1095 | if (buf.inbuf < buf.inbuf_end) { /* pending sequence still exists */ |
| 1096 | if (decoder_append_pending(STATEFUL_DCTX(self), &buf) != 0) |
| 1097 | goto errorexit; |
| 1098 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1099 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1100 | res = _PyUnicodeWriter_Finish(&buf.writer); |
| 1101 | if (res == NULL) |
| 1102 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1103 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1104 | if (wdata != data) |
| 1105 | PyMem_Del(wdata); |
| 1106 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1107 | return res; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1108 | |
| 1109 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1110 | if (wdata != NULL && wdata != data) |
| 1111 | PyMem_Del(wdata); |
| 1112 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1113 | _PyUnicodeWriter_Dealloc(&buf.writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1114 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1115 | } |
| 1116 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1117 | /*[clinic input] |
| 1118 | _multibytecodec.MultibyteIncrementalDecoder.reset |
| 1119 | [clinic start generated code]*/ |
| 1120 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1121 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1122 | _multibytecodec_MultibyteIncrementalDecoder_reset_impl(MultibyteIncrementalDecoderObject *self) |
| 1123 | /*[clinic end generated code: output=da423b1782c23ed1 input=3b63b3be85b2fb45]*/ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1124 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1125 | if (self->codec->decreset != NULL && |
| 1126 | self->codec->decreset(&self->state, self->codec->config) != 0) |
| 1127 | return NULL; |
| 1128 | self->pendingsize = 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1129 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1130 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1131 | } |
| 1132 | |
| 1133 | static struct PyMethodDef mbidecoder_methods[] = { |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1134 | _MULTIBYTECODEC_MULTIBYTEINCREMENTALDECODER_DECODE_METHODDEF |
| 1135 | _MULTIBYTECODEC_MULTIBYTEINCREMENTALDECODER_RESET_METHODDEF |
| 1136 | {NULL, NULL}, |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1137 | }; |
| 1138 | |
| 1139 | static PyObject * |
| 1140 | mbidecoder_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1141 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1142 | MultibyteIncrementalDecoderObject *self; |
| 1143 | PyObject *codec = NULL; |
| 1144 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1145 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1146 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "|s:IncrementalDecoder", |
| 1147 | incnewkwarglist, &errors)) |
| 1148 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1149 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1150 | self = (MultibyteIncrementalDecoderObject *)type->tp_alloc(type, 0); |
| 1151 | if (self == NULL) |
| 1152 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1153 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1154 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1155 | if (codec == NULL) |
| 1156 | goto errorexit; |
| 1157 | if (!MultibyteCodec_Check(codec)) { |
| 1158 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1159 | goto errorexit; |
| 1160 | } |
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->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1163 | self->pendingsize = 0; |
| 1164 | self->errors = internal_error_callback(errors); |
| 1165 | if (self->errors == NULL) |
| 1166 | goto errorexit; |
| 1167 | if (self->codec->decinit != NULL && |
| 1168 | self->codec->decinit(&self->state, self->codec->config) != 0) |
| 1169 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1170 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1171 | Py_DECREF(codec); |
| 1172 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1173 | |
| 1174 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1175 | Py_XDECREF(self); |
| 1176 | Py_XDECREF(codec); |
| 1177 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1178 | } |
| 1179 | |
| 1180 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1181 | mbidecoder_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1182 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1183 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1184 | } |
| 1185 | |
| 1186 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1187 | mbidecoder_traverse(MultibyteIncrementalDecoderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1188 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1189 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1190 | if (ERROR_ISCUSTOM(self->errors)) |
| 1191 | Py_VISIT(self->errors); |
| 1192 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1193 | } |
| 1194 | |
| 1195 | static void |
| 1196 | mbidecoder_dealloc(MultibyteIncrementalDecoderObject *self) |
| 1197 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1198 | PyObject_GC_UnTrack(self); |
| 1199 | ERROR_DECREF(self->errors); |
| 1200 | Py_TYPE(self)->tp_free(self); |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1201 | } |
| 1202 | |
| 1203 | static PyTypeObject MultibyteIncrementalDecoder_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1204 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1205 | "MultibyteIncrementalDecoder", /* tp_name */ |
| 1206 | sizeof(MultibyteIncrementalDecoderObject), /* tp_basicsize */ |
| 1207 | 0, /* tp_itemsize */ |
| 1208 | /* methods */ |
| 1209 | (destructor)mbidecoder_dealloc, /* tp_dealloc */ |
| 1210 | 0, /* tp_print */ |
| 1211 | 0, /* tp_getattr */ |
| 1212 | 0, /* tp_setattr */ |
| 1213 | 0, /* tp_reserved */ |
| 1214 | 0, /* tp_repr */ |
| 1215 | 0, /* tp_as_number */ |
| 1216 | 0, /* tp_as_sequence */ |
| 1217 | 0, /* tp_as_mapping */ |
| 1218 | 0, /* tp_hash */ |
| 1219 | 0, /* tp_call */ |
| 1220 | 0, /* tp_str */ |
| 1221 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1222 | 0, /* tp_setattro */ |
| 1223 | 0, /* tp_as_buffer */ |
| 1224 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1225 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1226 | 0, /* tp_doc */ |
| 1227 | (traverseproc)mbidecoder_traverse, /* tp_traverse */ |
| 1228 | 0, /* tp_clear */ |
| 1229 | 0, /* tp_richcompare */ |
| 1230 | 0, /* tp_weaklistoffset */ |
| 1231 | 0, /* tp_iter */ |
| 1232 | 0, /* tp_iterext */ |
| 1233 | mbidecoder_methods, /* tp_methods */ |
| 1234 | 0, /* tp_members */ |
| 1235 | codecctx_getsets, /* tp_getset */ |
| 1236 | 0, /* tp_base */ |
| 1237 | 0, /* tp_dict */ |
| 1238 | 0, /* tp_descr_get */ |
| 1239 | 0, /* tp_descr_set */ |
| 1240 | 0, /* tp_dictoffset */ |
| 1241 | mbidecoder_init, /* tp_init */ |
| 1242 | 0, /* tp_alloc */ |
| 1243 | mbidecoder_new, /* tp_new */ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1244 | }; |
| 1245 | |
| 1246 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1247 | /*[clinic input] |
| 1248 | class _multibytecodec.MultibyteStreamReader "MultibyteStreamReaderObject *" "MultibyteStreamReader_Type" |
| 1249 | [clinic start generated code]*/ |
| 1250 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=d323634b74976f09]*/ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1251 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1252 | static PyObject * |
| 1253 | mbstreamreader_iread(MultibyteStreamReaderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1254 | const char *method, Py_ssize_t sizehint) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1255 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1256 | MultibyteDecodeBuffer buf; |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1257 | PyObject *cres, *res; |
| 1258 | Py_ssize_t rsize; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1259 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1260 | if (sizehint == 0) |
Victor Stinner | b37b174 | 2011-12-01 03:18:59 +0100 | [diff] [blame] | 1261 | return PyUnicode_New(0, 0); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1262 | |
Victor Stinner | 8f674cc | 2013-04-17 23:02:17 +0200 | [diff] [blame] | 1263 | _PyUnicodeWriter_Init(&buf.writer); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1264 | buf.excobj = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1265 | cres = NULL; |
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 | for (;;) { |
| 1268 | int endoffile; |
Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1269 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1270 | if (sizehint < 0) |
| 1271 | cres = PyObject_CallMethod(self->stream, |
| 1272 | (char *)method, NULL); |
| 1273 | else |
| 1274 | cres = PyObject_CallMethod(self->stream, |
| 1275 | (char *)method, "i", sizehint); |
| 1276 | if (cres == NULL) |
| 1277 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1278 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1279 | if (!PyBytes_Check(cres)) { |
| 1280 | PyErr_Format(PyExc_TypeError, |
| 1281 | "stream function returned a " |
| 1282 | "non-bytes object (%.100s)", |
| 1283 | cres->ob_type->tp_name); |
| 1284 | goto errorexit; |
| 1285 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1286 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1287 | endoffile = (PyBytes_GET_SIZE(cres) == 0); |
Guido van Rossum | cd16bf6 | 2007-06-13 18:07:49 +0000 | [diff] [blame] | 1288 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1289 | if (self->pendingsize > 0) { |
| 1290 | PyObject *ctr; |
| 1291 | char *ctrdata; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1292 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1293 | if (PyBytes_GET_SIZE(cres) > PY_SSIZE_T_MAX - self->pendingsize) { |
| 1294 | PyErr_NoMemory(); |
| 1295 | goto errorexit; |
Victor Stinner | 064bbdc | 2013-07-08 22:28:27 +0200 | [diff] [blame] | 1296 | } |
| 1297 | rsize = PyBytes_GET_SIZE(cres) + self->pendingsize; |
| 1298 | ctr = PyBytes_FromStringAndSize(NULL, rsize); |
| 1299 | if (ctr == NULL) |
| 1300 | goto errorexit; |
| 1301 | ctrdata = PyBytes_AS_STRING(ctr); |
| 1302 | memcpy(ctrdata, self->pending, self->pendingsize); |
| 1303 | memcpy(ctrdata + self->pendingsize, |
| 1304 | PyBytes_AS_STRING(cres), |
| 1305 | PyBytes_GET_SIZE(cres)); |
| 1306 | Py_DECREF(cres); |
| 1307 | cres = ctr; |
| 1308 | self->pendingsize = 0; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1309 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1310 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1311 | rsize = PyBytes_GET_SIZE(cres); |
| 1312 | if (decoder_prepare_buffer(&buf, PyBytes_AS_STRING(cres), |
| 1313 | rsize) != 0) |
| 1314 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1315 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1316 | if (rsize > 0 && decoder_feed_buffer( |
| 1317 | (MultibyteStatefulDecoderContext *)self, &buf)) |
| 1318 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1319 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1320 | if (endoffile || sizehint < 0) { |
| 1321 | if (buf.inbuf < buf.inbuf_end && |
| 1322 | multibytecodec_decerror(self->codec, &self->state, |
| 1323 | &buf, self->errors, MBERR_TOOFEW)) |
| 1324 | goto errorexit; |
| 1325 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1326 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1327 | if (buf.inbuf < buf.inbuf_end) { /* pending sequence exists */ |
| 1328 | if (decoder_append_pending(STATEFUL_DCTX(self), |
| 1329 | &buf) != 0) |
| 1330 | goto errorexit; |
| 1331 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1332 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1333 | Py_DECREF(cres); |
| 1334 | cres = NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1335 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1336 | if (sizehint < 0 || buf.writer.pos != 0 || rsize == 0) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1337 | break; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1338 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1339 | sizehint = 1; /* read 1 more byte and retry */ |
| 1340 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1341 | |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1342 | res = _PyUnicodeWriter_Finish(&buf.writer); |
| 1343 | if (res == NULL) |
| 1344 | goto errorexit; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1345 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1346 | Py_XDECREF(cres); |
| 1347 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1348 | return res; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1349 | |
| 1350 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1351 | Py_XDECREF(cres); |
| 1352 | Py_XDECREF(buf.excobj); |
Victor Stinner | a0dd021 | 2013-04-11 22:09:04 +0200 | [diff] [blame] | 1353 | _PyUnicodeWriter_Dealloc(&buf.writer); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1354 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1355 | } |
| 1356 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1357 | /*[clinic input] |
| 1358 | _multibytecodec.MultibyteStreamReader.read |
| 1359 | |
| 1360 | sizeobj: object = None |
| 1361 | / |
| 1362 | [clinic start generated code]*/ |
| 1363 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1364 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1365 | _multibytecodec_MultibyteStreamReader_read_impl(MultibyteStreamReaderObject *self, PyObject *sizeobj) |
| 1366 | /*[clinic end generated code: output=f298ea6e1bd2083c input=015b0d3ff2fca485]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1367 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1368 | Py_ssize_t size; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1369 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1370 | if (sizeobj == Py_None) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1371 | size = -1; |
| 1372 | else if (PyLong_Check(sizeobj)) |
| 1373 | size = PyLong_AsSsize_t(sizeobj); |
| 1374 | else { |
| 1375 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1376 | return NULL; |
| 1377 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1378 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1379 | if (size == -1 && PyErr_Occurred()) |
| 1380 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1381 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1382 | return mbstreamreader_iread(self, "read", size); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1383 | } |
| 1384 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1385 | /*[clinic input] |
| 1386 | _multibytecodec.MultibyteStreamReader.readline |
| 1387 | |
| 1388 | sizeobj: object = None |
| 1389 | / |
| 1390 | [clinic start generated code]*/ |
| 1391 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1392 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1393 | _multibytecodec_MultibyteStreamReader_readline_impl(MultibyteStreamReaderObject *self, PyObject *sizeobj) |
| 1394 | /*[clinic end generated code: output=e5ac302a6d0999de input=41ccc64f9bb0cec3]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1395 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1396 | Py_ssize_t size; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1397 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1398 | if (sizeobj == Py_None) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1399 | size = -1; |
| 1400 | else if (PyLong_Check(sizeobj)) |
| 1401 | size = PyLong_AsSsize_t(sizeobj); |
| 1402 | else { |
| 1403 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1404 | return NULL; |
| 1405 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1406 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1407 | if (size == -1 && PyErr_Occurred()) |
| 1408 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1409 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1410 | return mbstreamreader_iread(self, "readline", size); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1411 | } |
| 1412 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1413 | /*[clinic input] |
| 1414 | _multibytecodec.MultibyteStreamReader.readlines |
| 1415 | |
| 1416 | sizehintobj: object = None |
| 1417 | / |
| 1418 | [clinic start generated code]*/ |
| 1419 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1420 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1421 | _multibytecodec_MultibyteStreamReader_readlines_impl(MultibyteStreamReaderObject *self, PyObject *sizehintobj) |
| 1422 | /*[clinic end generated code: output=68f024178b77cb0f input=54932f5d4d88e880]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1423 | { |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1424 | PyObject *r, *sr; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1425 | Py_ssize_t sizehint; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1426 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1427 | if (sizehintobj == Py_None) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1428 | sizehint = -1; |
| 1429 | else if (PyLong_Check(sizehintobj)) |
| 1430 | sizehint = PyLong_AsSsize_t(sizehintobj); |
| 1431 | else { |
| 1432 | PyErr_SetString(PyExc_TypeError, "arg 1 must be an integer"); |
| 1433 | return NULL; |
| 1434 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1435 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1436 | if (sizehint == -1 && PyErr_Occurred()) |
| 1437 | return NULL; |
Guido van Rossum | ddefaf3 | 2007-01-14 03:31:43 +0000 | [diff] [blame] | 1438 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1439 | r = mbstreamreader_iread(self, "read", sizehint); |
| 1440 | if (r == NULL) |
| 1441 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1442 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1443 | sr = PyUnicode_Splitlines(r, 1); |
| 1444 | Py_DECREF(r); |
| 1445 | return sr; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1446 | } |
| 1447 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1448 | /*[clinic input] |
| 1449 | _multibytecodec.MultibyteStreamReader.reset |
| 1450 | [clinic start generated code]*/ |
| 1451 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1452 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1453 | _multibytecodec_MultibyteStreamReader_reset_impl(MultibyteStreamReaderObject *self) |
| 1454 | /*[clinic end generated code: output=138490370a680abc input=5d4140db84b5e1e2]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1455 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1456 | if (self->codec->decreset != NULL && |
| 1457 | self->codec->decreset(&self->state, self->codec->config) != 0) |
| 1458 | return NULL; |
| 1459 | self->pendingsize = 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1460 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1461 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1462 | } |
| 1463 | |
| 1464 | static struct PyMethodDef mbstreamreader_methods[] = { |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1465 | _MULTIBYTECODEC_MULTIBYTESTREAMREADER_READ_METHODDEF |
| 1466 | _MULTIBYTECODEC_MULTIBYTESTREAMREADER_READLINE_METHODDEF |
| 1467 | _MULTIBYTECODEC_MULTIBYTESTREAMREADER_READLINES_METHODDEF |
| 1468 | _MULTIBYTECODEC_MULTIBYTESTREAMREADER_RESET_METHODDEF |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1469 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1470 | }; |
| 1471 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1472 | static PyMemberDef mbstreamreader_members[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1473 | {"stream", T_OBJECT, |
| 1474 | offsetof(MultibyteStreamReaderObject, stream), |
| 1475 | READONLY, NULL}, |
| 1476 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1477 | }; |
| 1478 | |
| 1479 | static PyObject * |
| 1480 | mbstreamreader_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1481 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1482 | MultibyteStreamReaderObject *self; |
| 1483 | PyObject *stream, *codec = NULL; |
| 1484 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1485 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1486 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamReader", |
| 1487 | streamkwarglist, &stream, &errors)) |
| 1488 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1489 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1490 | self = (MultibyteStreamReaderObject *)type->tp_alloc(type, 0); |
| 1491 | if (self == NULL) |
| 1492 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1493 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1494 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1495 | if (codec == NULL) |
| 1496 | goto errorexit; |
| 1497 | if (!MultibyteCodec_Check(codec)) { |
| 1498 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1499 | goto errorexit; |
| 1500 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1501 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1502 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1503 | self->stream = stream; |
| 1504 | Py_INCREF(stream); |
| 1505 | self->pendingsize = 0; |
| 1506 | self->errors = internal_error_callback(errors); |
| 1507 | if (self->errors == NULL) |
| 1508 | goto errorexit; |
| 1509 | if (self->codec->decinit != NULL && |
| 1510 | self->codec->decinit(&self->state, self->codec->config) != 0) |
| 1511 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1512 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1513 | Py_DECREF(codec); |
| 1514 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1515 | |
| 1516 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1517 | Py_XDECREF(self); |
| 1518 | Py_XDECREF(codec); |
| 1519 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1520 | } |
| 1521 | |
| 1522 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1523 | mbstreamreader_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1524 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1525 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1526 | } |
| 1527 | |
| 1528 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1529 | mbstreamreader_traverse(MultibyteStreamReaderObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1530 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1531 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1532 | if (ERROR_ISCUSTOM(self->errors)) |
| 1533 | Py_VISIT(self->errors); |
| 1534 | Py_VISIT(self->stream); |
| 1535 | return 0; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1536 | } |
| 1537 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1538 | static void |
| 1539 | mbstreamreader_dealloc(MultibyteStreamReaderObject *self) |
| 1540 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1541 | PyObject_GC_UnTrack(self); |
| 1542 | ERROR_DECREF(self->errors); |
| 1543 | Py_XDECREF(self->stream); |
| 1544 | Py_TYPE(self)->tp_free(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1545 | } |
| 1546 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1547 | static PyTypeObject MultibyteStreamReader_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1548 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1549 | "MultibyteStreamReader", /* tp_name */ |
| 1550 | sizeof(MultibyteStreamReaderObject), /* tp_basicsize */ |
| 1551 | 0, /* tp_itemsize */ |
| 1552 | /* methods */ |
| 1553 | (destructor)mbstreamreader_dealloc, /* tp_dealloc */ |
| 1554 | 0, /* tp_print */ |
| 1555 | 0, /* tp_getattr */ |
| 1556 | 0, /* tp_setattr */ |
| 1557 | 0, /* tp_reserved */ |
| 1558 | 0, /* tp_repr */ |
| 1559 | 0, /* tp_as_number */ |
| 1560 | 0, /* tp_as_sequence */ |
| 1561 | 0, /* tp_as_mapping */ |
| 1562 | 0, /* tp_hash */ |
| 1563 | 0, /* tp_call */ |
| 1564 | 0, /* tp_str */ |
| 1565 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1566 | 0, /* tp_setattro */ |
| 1567 | 0, /* tp_as_buffer */ |
| 1568 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1569 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1570 | 0, /* tp_doc */ |
| 1571 | (traverseproc)mbstreamreader_traverse, /* tp_traverse */ |
| 1572 | 0, /* tp_clear */ |
| 1573 | 0, /* tp_richcompare */ |
| 1574 | 0, /* tp_weaklistoffset */ |
| 1575 | 0, /* tp_iter */ |
| 1576 | 0, /* tp_iterext */ |
| 1577 | mbstreamreader_methods, /* tp_methods */ |
| 1578 | mbstreamreader_members, /* tp_members */ |
| 1579 | codecctx_getsets, /* tp_getset */ |
| 1580 | 0, /* tp_base */ |
| 1581 | 0, /* tp_dict */ |
| 1582 | 0, /* tp_descr_get */ |
| 1583 | 0, /* tp_descr_set */ |
| 1584 | 0, /* tp_dictoffset */ |
| 1585 | mbstreamreader_init, /* tp_init */ |
| 1586 | 0, /* tp_alloc */ |
| 1587 | mbstreamreader_new, /* tp_new */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1588 | }; |
| 1589 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1590 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1591 | /*[clinic input] |
| 1592 | class _multibytecodec.MultibyteStreamWriter "MultibyteStreamWriterObject *" "&MultibyteStreamWriter_Type" |
| 1593 | [clinic start generated code]*/ |
| 1594 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=cde22780a215d6ac]*/ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1595 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1596 | static int |
| 1597 | mbstreamwriter_iwrite(MultibyteStreamWriterObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1598 | PyObject *unistr) |
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 | PyObject *str, *wr; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1601 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1602 | str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0); |
| 1603 | if (str == NULL) |
| 1604 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1605 | |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1606 | wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1607 | Py_DECREF(str); |
| 1608 | if (wr == NULL) |
| 1609 | return -1; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1610 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1611 | Py_DECREF(wr); |
| 1612 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1613 | } |
| 1614 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1615 | /*[clinic input] |
| 1616 | _multibytecodec.MultibyteStreamWriter.write |
| 1617 | |
| 1618 | strobj: object |
| 1619 | / |
| 1620 | [clinic start generated code]*/ |
| 1621 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1622 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1623 | _multibytecodec_MultibyteStreamWriter_write(MultibyteStreamWriterObject *self, PyObject *strobj) |
| 1624 | /*[clinic end generated code: output=44e9eb0db0374cb1 input=551dc4c018c10a2b]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1625 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1626 | if (mbstreamwriter_iwrite(self, strobj)) |
| 1627 | return NULL; |
| 1628 | else |
| 1629 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1630 | } |
| 1631 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1632 | /*[clinic input] |
| 1633 | _multibytecodec.MultibyteStreamWriter.writelines |
| 1634 | |
| 1635 | lines: object |
| 1636 | / |
| 1637 | [clinic start generated code]*/ |
| 1638 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1639 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1640 | _multibytecodec_MultibyteStreamWriter_writelines(MultibyteStreamWriterObject *self, PyObject *lines) |
| 1641 | /*[clinic end generated code: output=4facbb0638dde172 input=57797fe7008d4e96]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1642 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1643 | PyObject *strobj; |
| 1644 | int i, r; |
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 | if (!PySequence_Check(lines)) { |
| 1647 | PyErr_SetString(PyExc_TypeError, |
| 1648 | "arg must be a sequence object"); |
| 1649 | return NULL; |
| 1650 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1651 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1652 | for (i = 0; i < PySequence_Length(lines); i++) { |
| 1653 | /* length can be changed even within this loop */ |
| 1654 | strobj = PySequence_GetItem(lines, i); |
| 1655 | if (strobj == NULL) |
| 1656 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1657 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1658 | r = mbstreamwriter_iwrite(self, strobj); |
| 1659 | Py_DECREF(strobj); |
| 1660 | if (r == -1) |
| 1661 | return NULL; |
| 1662 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1663 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1664 | Py_RETURN_NONE; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1665 | } |
| 1666 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1667 | /*[clinic input] |
| 1668 | _multibytecodec.MultibyteStreamWriter.reset |
| 1669 | [clinic start generated code]*/ |
| 1670 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1671 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1672 | _multibytecodec_MultibyteStreamWriter_reset_impl(MultibyteStreamWriterObject *self) |
| 1673 | /*[clinic end generated code: output=8f54a4d9b03db5ff input=b56dbcbaf35cc10c]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1674 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1675 | PyObject *pwrt; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1676 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1677 | pwrt = multibytecodec_encode(self->codec, &self->state, |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 1678 | self->pending, NULL, self->errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1679 | MBENC_FLUSH | MBENC_RESET); |
| 1680 | /* some pending buffer can be truncated when UnicodeEncodeError is |
| 1681 | * raised on 'strict' mode. but, 'reset' method is designed to |
| 1682 | * reset the pending buffer or states so failed string sequence |
| 1683 | * ought to be missed */ |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 1684 | Py_CLEAR(self->pending); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1685 | if (pwrt == NULL) |
| 1686 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1687 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1688 | assert(PyBytes_Check(pwrt)); |
| 1689 | if (PyBytes_Size(pwrt) > 0) { |
| 1690 | PyObject *wr; |
Martin v. Löwis | afe55bb | 2011-10-09 10:38:36 +0200 | [diff] [blame] | 1691 | |
| 1692 | wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1693 | if (wr == NULL) { |
| 1694 | Py_DECREF(pwrt); |
| 1695 | return NULL; |
| 1696 | } |
| 1697 | } |
| 1698 | Py_DECREF(pwrt); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1699 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1700 | Py_RETURN_NONE; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1701 | } |
| 1702 | |
| 1703 | static PyObject * |
| 1704 | mbstreamwriter_new(PyTypeObject *type, PyObject *args, PyObject *kwds) |
| 1705 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1706 | MultibyteStreamWriterObject *self; |
| 1707 | PyObject *stream, *codec = NULL; |
| 1708 | char *errors = NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1709 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1710 | if (!PyArg_ParseTupleAndKeywords(args, kwds, "O|s:StreamWriter", |
| 1711 | streamkwarglist, &stream, &errors)) |
| 1712 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1713 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1714 | self = (MultibyteStreamWriterObject *)type->tp_alloc(type, 0); |
| 1715 | if (self == NULL) |
| 1716 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1717 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1718 | codec = PyObject_GetAttrString((PyObject *)type, "codec"); |
| 1719 | if (codec == NULL) |
| 1720 | goto errorexit; |
| 1721 | if (!MultibyteCodec_Check(codec)) { |
| 1722 | PyErr_SetString(PyExc_TypeError, "codec is unexpected type"); |
| 1723 | goto errorexit; |
| 1724 | } |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1725 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1726 | self->codec = ((MultibyteCodecObject *)codec)->codec; |
| 1727 | self->stream = stream; |
| 1728 | Py_INCREF(stream); |
Victor Stinner | d949126 | 2013-04-14 02:06:32 +0200 | [diff] [blame] | 1729 | self->pending = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1730 | self->errors = internal_error_callback(errors); |
| 1731 | if (self->errors == NULL) |
| 1732 | goto errorexit; |
| 1733 | if (self->codec->encinit != NULL && |
| 1734 | self->codec->encinit(&self->state, self->codec->config) != 0) |
| 1735 | goto errorexit; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1736 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1737 | Py_DECREF(codec); |
| 1738 | return (PyObject *)self; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1739 | |
| 1740 | errorexit: |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1741 | Py_XDECREF(self); |
| 1742 | Py_XDECREF(codec); |
| 1743 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1744 | } |
| 1745 | |
| 1746 | static int |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1747 | mbstreamwriter_init(PyObject *self, PyObject *args, PyObject *kwds) |
| 1748 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1749 | return 0; |
Hye-Shik Chang | 13247bf | 2006-04-21 16:21:44 +0000 | [diff] [blame] | 1750 | } |
| 1751 | |
| 1752 | static int |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1753 | mbstreamwriter_traverse(MultibyteStreamWriterObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1754 | visitproc visit, void *arg) |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1755 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1756 | if (ERROR_ISCUSTOM(self->errors)) |
| 1757 | Py_VISIT(self->errors); |
| 1758 | Py_VISIT(self->stream); |
| 1759 | return 0; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1760 | } |
| 1761 | |
| 1762 | static void |
| 1763 | mbstreamwriter_dealloc(MultibyteStreamWriterObject *self) |
| 1764 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1765 | PyObject_GC_UnTrack(self); |
| 1766 | ERROR_DECREF(self->errors); |
| 1767 | Py_XDECREF(self->stream); |
| 1768 | Py_TYPE(self)->tp_free(self); |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1769 | } |
| 1770 | |
| 1771 | static struct PyMethodDef mbstreamwriter_methods[] = { |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1772 | _MULTIBYTECODEC_MULTIBYTESTREAMWRITER_WRITE_METHODDEF |
| 1773 | _MULTIBYTECODEC_MULTIBYTESTREAMWRITER_WRITELINES_METHODDEF |
| 1774 | _MULTIBYTECODEC_MULTIBYTESTREAMWRITER_RESET_METHODDEF |
| 1775 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1776 | }; |
| 1777 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1778 | static PyMemberDef mbstreamwriter_members[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1779 | {"stream", T_OBJECT, |
| 1780 | offsetof(MultibyteStreamWriterObject, stream), |
| 1781 | READONLY, NULL}, |
| 1782 | {NULL,} |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1783 | }; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1784 | |
| 1785 | static PyTypeObject MultibyteStreamWriter_Type = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1786 | PyVarObject_HEAD_INIT(NULL, 0) |
| 1787 | "MultibyteStreamWriter", /* tp_name */ |
| 1788 | sizeof(MultibyteStreamWriterObject), /* tp_basicsize */ |
| 1789 | 0, /* tp_itemsize */ |
| 1790 | /* methods */ |
| 1791 | (destructor)mbstreamwriter_dealloc, /* tp_dealloc */ |
| 1792 | 0, /* tp_print */ |
| 1793 | 0, /* tp_getattr */ |
| 1794 | 0, /* tp_setattr */ |
| 1795 | 0, /* tp_reserved */ |
| 1796 | 0, /* tp_repr */ |
| 1797 | 0, /* tp_as_number */ |
| 1798 | 0, /* tp_as_sequence */ |
| 1799 | 0, /* tp_as_mapping */ |
| 1800 | 0, /* tp_hash */ |
| 1801 | 0, /* tp_call */ |
| 1802 | 0, /* tp_str */ |
| 1803 | PyObject_GenericGetAttr, /* tp_getattro */ |
| 1804 | 0, /* tp_setattro */ |
| 1805 | 0, /* tp_as_buffer */ |
| 1806 | Py_TPFLAGS_DEFAULT | Py_TPFLAGS_HAVE_GC |
| 1807 | | Py_TPFLAGS_BASETYPE, /* tp_flags */ |
| 1808 | 0, /* tp_doc */ |
| 1809 | (traverseproc)mbstreamwriter_traverse, /* tp_traverse */ |
| 1810 | 0, /* tp_clear */ |
| 1811 | 0, /* tp_richcompare */ |
| 1812 | 0, /* tp_weaklistoffset */ |
| 1813 | 0, /* tp_iter */ |
| 1814 | 0, /* tp_iterext */ |
| 1815 | mbstreamwriter_methods, /* tp_methods */ |
| 1816 | mbstreamwriter_members, /* tp_members */ |
| 1817 | codecctx_getsets, /* tp_getset */ |
| 1818 | 0, /* tp_base */ |
| 1819 | 0, /* tp_dict */ |
| 1820 | 0, /* tp_descr_get */ |
| 1821 | 0, /* tp_descr_set */ |
| 1822 | 0, /* tp_dictoffset */ |
| 1823 | mbstreamwriter_init, /* tp_init */ |
| 1824 | 0, /* tp_alloc */ |
| 1825 | mbstreamwriter_new, /* tp_new */ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1826 | }; |
| 1827 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1828 | |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1829 | /*[clinic input] |
| 1830 | _multibytecodec.__create_codec |
| 1831 | |
| 1832 | arg: object |
| 1833 | / |
| 1834 | [clinic start generated code]*/ |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1835 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1836 | static PyObject * |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1837 | _multibytecodec___create_codec(PyModuleDef *module, PyObject *arg) |
| 1838 | /*[clinic end generated code: output=fbe74f6510640163 input=6840b2a6b183fcfa]*/ |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1839 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1840 | MultibyteCodecObject *self; |
| 1841 | MultibyteCodec *codec; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1842 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1843 | if (!PyCapsule_IsValid(arg, PyMultibyteCodec_CAPSULE_NAME)) { |
| 1844 | PyErr_SetString(PyExc_ValueError, "argument type invalid"); |
| 1845 | return NULL; |
| 1846 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1847 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1848 | codec = PyCapsule_GetPointer(arg, PyMultibyteCodec_CAPSULE_NAME); |
| 1849 | if (codec->codecinit != NULL && codec->codecinit(codec->config) != 0) |
| 1850 | return NULL; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1851 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1852 | self = PyObject_New(MultibyteCodecObject, &MultibyteCodec_Type); |
| 1853 | if (self == NULL) |
| 1854 | return NULL; |
| 1855 | self->codec = codec; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1856 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1857 | return (PyObject *)self; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1858 | } |
| 1859 | |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1860 | static struct PyMethodDef __methods[] = { |
Brett Cannon | f2de1fc | 2014-08-22 11:45:03 -0400 | [diff] [blame] | 1861 | _MULTIBYTECODEC___CREATE_CODEC_METHODDEF |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1862 | {NULL, NULL}, |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1863 | }; |
| 1864 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1865 | |
| 1866 | static struct PyModuleDef _multibytecodecmodule = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1867 | PyModuleDef_HEAD_INIT, |
| 1868 | "_multibytecodec", |
| 1869 | NULL, |
| 1870 | -1, |
| 1871 | __methods, |
| 1872 | NULL, |
| 1873 | NULL, |
| 1874 | NULL, |
| 1875 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1876 | }; |
| 1877 | |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1878 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1879 | PyInit__multibytecodec(void) |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1880 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1881 | int i; |
| 1882 | PyObject *m; |
| 1883 | PyTypeObject *typelist[] = { |
| 1884 | &MultibyteIncrementalEncoder_Type, |
| 1885 | &MultibyteIncrementalDecoder_Type, |
| 1886 | &MultibyteStreamReader_Type, |
| 1887 | &MultibyteStreamWriter_Type, |
| 1888 | NULL |
| 1889 | }; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1890 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1891 | if (PyType_Ready(&MultibyteCodec_Type) < 0) |
| 1892 | return NULL; |
Thomas Wouters | 49fd7fa | 2006-04-21 10:40:58 +0000 | [diff] [blame] | 1893 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1894 | m = PyModule_Create(&_multibytecodecmodule); |
| 1895 | if (m == NULL) |
| 1896 | return NULL; |
Neal Norwitz | 058bde1 | 2005-09-21 06:44:25 +0000 | [diff] [blame] | 1897 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1898 | for (i = 0; typelist[i] != NULL; i++) { |
| 1899 | if (PyType_Ready(typelist[i]) < 0) |
| 1900 | return NULL; |
| 1901 | Py_INCREF(typelist[i]); |
| 1902 | PyModule_AddObject(m, typelist[i]->tp_name, |
| 1903 | (PyObject *)typelist[i]); |
| 1904 | } |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1905 | |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1906 | if (PyErr_Occurred()) { |
| 1907 | Py_FatalError("can't initialize the _multibytecodec module"); |
| 1908 | Py_DECREF(m); |
| 1909 | m = NULL; |
| 1910 | } |
| 1911 | return m; |
Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame] | 1912 | } |