Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1 | /* ------------------------------------------------------------------------ |
| 2 | |
| 3 | _codecs -- Provides access to the codec registry and the builtin |
| 4 | codecs. |
| 5 | |
| 6 | This module should never be imported directly. The standard library |
| 7 | module "codecs" wraps this builtin module for use within Python. |
| 8 | |
| 9 | The codec registry is accessible via: |
| 10 | |
| 11 | register(search_function) -> None |
| 12 | |
Walter Dörwald | 219336a | 2007-07-19 13:04:38 +0000 | [diff] [blame] | 13 | lookup(encoding) -> CodecInfo object |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 14 | |
| 15 | The builtin Unicode codecs use the following interface: |
| 16 | |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 17 | <encoding>_encode(Unicode_object[,errors='strict']) -> |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 18 | (string object, bytes consumed) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 19 | |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 20 | <encoding>_decode(char_buffer_obj[,errors='strict']) -> |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 21 | (Unicode object, bytes consumed) |
| 22 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 23 | <encoding>_encode() interfaces also accept non-Unicode object as |
| 24 | input. The objects are then converted to Unicode using |
| 25 | PyUnicode_FromObject() prior to applying the conversion. |
| 26 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 27 | These <encoding>s are available: utf_8, unicode_escape, |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 28 | raw_unicode_escape, unicode_internal, latin_1, ascii (7-bit), |
| 29 | mbcs (on win32). |
| 30 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 31 | |
| 32 | Written by Marc-Andre Lemburg (mal@lemburg.com). |
| 33 | |
Guido van Rossum | 16b1ad9 | 2000-08-03 16:24:25 +0000 | [diff] [blame] | 34 | Copyright (c) Corporation for National Research Initiatives. |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 35 | |
| 36 | ------------------------------------------------------------------------ */ |
| 37 | |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 38 | #define PY_SSIZE_T_CLEAN |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 39 | #include "Python.h" |
| 40 | |
| 41 | /* --- Registry ----------------------------------------------------------- */ |
| 42 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 43 | PyDoc_STRVAR(register__doc__, |
| 44 | "register(search_function)\n\ |
| 45 | \n\ |
| 46 | Register a codec search function. Search functions are expected to take\n\ |
| 47 | one argument, the encoding name in all lower case letters, and return\n\ |
Walter Dörwald | 219336a | 2007-07-19 13:04:38 +0000 | [diff] [blame] | 48 | a tuple of functions (encoder, decoder, stream_reader, stream_writer)\n\ |
| 49 | (or a CodecInfo object)."); |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 50 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 51 | static |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 52 | PyObject *codec_register(PyObject *self, PyObject *search_function) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 53 | { |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 54 | if (PyCodec_Register(search_function)) |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 55 | return NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 56 | |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 57 | Py_RETURN_NONE; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 58 | } |
| 59 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 60 | PyDoc_STRVAR(lookup__doc__, |
Walter Dörwald | 219336a | 2007-07-19 13:04:38 +0000 | [diff] [blame] | 61 | "lookup(encoding) -> CodecInfo\n\ |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 62 | \n\ |
| 63 | Looks up a codec tuple in the Python codec registry and returns\n\ |
Georg Brandl | 4c95e09 | 2009-03-15 21:32:06 +0000 | [diff] [blame] | 64 | a CodecInfo object."); |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 65 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 66 | static |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 67 | PyObject *codec_lookup(PyObject *self, PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 68 | { |
| 69 | char *encoding; |
| 70 | |
| 71 | if (!PyArg_ParseTuple(args, "s:lookup", &encoding)) |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 72 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 73 | |
| 74 | return _PyCodec_Lookup(encoding); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 75 | } |
| 76 | |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 77 | PyDoc_STRVAR(encode__doc__, |
| 78 | "encode(obj, [encoding[,errors]]) -> object\n\ |
| 79 | \n\ |
| 80 | Encodes obj using the codec registered for encoding. encoding defaults\n\ |
| 81 | to the default encoding. errors may be given to set a different error\n\ |
| 82 | handling scheme. Default is 'strict' meaning that encoding errors raise\n\ |
| 83 | a ValueError. Other possible values are 'ignore', 'replace' and\n\ |
| 84 | 'xmlcharrefreplace' as well as any other name registered with\n\ |
| 85 | codecs.register_error that can handle ValueErrors."); |
| 86 | |
| 87 | static PyObject * |
| 88 | codec_encode(PyObject *self, PyObject *args) |
| 89 | { |
Brett Cannon | 3e377de | 2004-07-10 21:41:14 +0000 | [diff] [blame] | 90 | const char *encoding = NULL; |
| 91 | const char *errors = NULL; |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 92 | PyObject *v; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 93 | |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 94 | if (!PyArg_ParseTuple(args, "O|ss:encode", &v, &encoding, &errors)) |
| 95 | return NULL; |
| 96 | |
Martin v. Löwis | e2713be | 2005-03-08 15:03:08 +0000 | [diff] [blame] | 97 | #ifdef Py_USING_UNICODE |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 98 | if (encoding == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 99 | encoding = PyUnicode_GetDefaultEncoding(); |
Martin v. Löwis | e2713be | 2005-03-08 15:03:08 +0000 | [diff] [blame] | 100 | #else |
| 101 | if (encoding == NULL) { |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 102 | PyErr_SetString(PyExc_ValueError, "no encoding specified"); |
| 103 | return NULL; |
Martin v. Löwis | e2713be | 2005-03-08 15:03:08 +0000 | [diff] [blame] | 104 | } |
| 105 | #endif |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 106 | |
| 107 | /* Encode via the codec registry */ |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 108 | return PyCodec_Encode(v, encoding, errors); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 109 | } |
| 110 | |
| 111 | PyDoc_STRVAR(decode__doc__, |
| 112 | "decode(obj, [encoding[,errors]]) -> object\n\ |
| 113 | \n\ |
| 114 | Decodes obj using the codec registered for encoding. encoding defaults\n\ |
| 115 | to the default encoding. errors may be given to set a different error\n\ |
| 116 | handling scheme. Default is 'strict' meaning that encoding errors raise\n\ |
| 117 | a ValueError. Other possible values are 'ignore' and 'replace'\n\ |
Andrew M. Kuchling | 658c45f | 2008-10-03 12:26:42 +0000 | [diff] [blame] | 118 | as well as any other name registered with codecs.register_error that is\n\ |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 119 | able to handle ValueErrors."); |
| 120 | |
| 121 | static PyObject * |
| 122 | codec_decode(PyObject *self, PyObject *args) |
| 123 | { |
Brett Cannon | 3e377de | 2004-07-10 21:41:14 +0000 | [diff] [blame] | 124 | const char *encoding = NULL; |
| 125 | const char *errors = NULL; |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 126 | PyObject *v; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 127 | |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 128 | if (!PyArg_ParseTuple(args, "O|ss:decode", &v, &encoding, &errors)) |
| 129 | return NULL; |
| 130 | |
Martin v. Löwis | e2713be | 2005-03-08 15:03:08 +0000 | [diff] [blame] | 131 | #ifdef Py_USING_UNICODE |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 132 | if (encoding == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 133 | encoding = PyUnicode_GetDefaultEncoding(); |
Martin v. Löwis | e2713be | 2005-03-08 15:03:08 +0000 | [diff] [blame] | 134 | #else |
| 135 | if (encoding == NULL) { |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 136 | PyErr_SetString(PyExc_ValueError, "no encoding specified"); |
| 137 | return NULL; |
Martin v. Löwis | e2713be | 2005-03-08 15:03:08 +0000 | [diff] [blame] | 138 | } |
| 139 | #endif |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 140 | |
| 141 | /* Decode via the codec registry */ |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 142 | return PyCodec_Decode(v, encoding, errors); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 143 | } |
| 144 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 145 | /* --- Helpers ------------------------------------------------------------ */ |
| 146 | |
| 147 | static |
| 148 | PyObject *codec_tuple(PyObject *unicode, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 149 | Py_ssize_t len) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 150 | { |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 151 | PyObject *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 152 | if (unicode == NULL) |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 153 | return NULL; |
| 154 | v = Py_BuildValue("On", unicode, len); |
| 155 | Py_DECREF(unicode); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 156 | return v; |
| 157 | } |
| 158 | |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 159 | /* --- String codecs ------------------------------------------------------ */ |
| 160 | static PyObject * |
| 161 | escape_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 162 | PyObject *args) |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 163 | { |
| 164 | const char *errors = NULL; |
| 165 | const char *data; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 166 | Py_ssize_t size; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 167 | |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 168 | if (!PyArg_ParseTuple(args, "s#|z:escape_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 169 | &data, &size, &errors)) |
| 170 | return NULL; |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 171 | return codec_tuple(PyString_DecodeEscape(data, size, errors, 0, NULL), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 172 | size); |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 173 | } |
| 174 | |
| 175 | static PyObject * |
| 176 | escape_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 177 | PyObject *args) |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 178 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 179 | PyObject *str; |
| 180 | const char *errors = NULL; |
| 181 | char *buf; |
| 182 | Py_ssize_t consumed, len; |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 183 | |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 184 | if (!PyArg_ParseTuple(args, "S|z:escape_encode", |
| 185 | &str, &errors)) |
| 186 | return NULL; |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 187 | |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 188 | consumed = PyString_GET_SIZE(str); |
| 189 | str = PyString_Repr(str, 0); |
| 190 | if (!str) |
| 191 | return NULL; |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 192 | |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 193 | /* The string will be quoted. Unquote, similar to unicode-escape. */ |
| 194 | buf = PyString_AS_STRING (str); |
| 195 | len = PyString_GET_SIZE (str); |
| 196 | memmove(buf, buf+1, len-2); |
| 197 | if (_PyString_Resize(&str, len-2) < 0) |
| 198 | return NULL; |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 199 | |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 200 | return codec_tuple(str, consumed); |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 201 | } |
| 202 | |
| 203 | #ifdef Py_USING_UNICODE |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 204 | /* --- Decoder ------------------------------------------------------------ */ |
| 205 | |
| 206 | static PyObject * |
| 207 | unicode_internal_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 208 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 209 | { |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 210 | PyObject *obj; |
| 211 | const char *errors = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 212 | const char *data; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 213 | Py_ssize_t size; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 214 | |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 215 | if (!PyArg_ParseTuple(args, "O|z:unicode_internal_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 216 | &obj, &errors)) |
| 217 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 218 | |
Marc-André Lemburg | 29273c8 | 2003-02-04 19:35:03 +0000 | [diff] [blame] | 219 | if (PyUnicode_Check(obj)) { |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 220 | Py_INCREF(obj); |
| 221 | return codec_tuple(obj, PyUnicode_GET_SIZE(obj)); |
Marc-André Lemburg | 29273c8 | 2003-02-04 19:35:03 +0000 | [diff] [blame] | 222 | } |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 223 | else { |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 224 | if (PyObject_AsReadBuffer(obj, (const void **)&data, &size)) |
| 225 | return NULL; |
Walter Dörwald | a47d1c0 | 2005-08-30 10:23:14 +0000 | [diff] [blame] | 226 | |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 227 | return codec_tuple(_PyUnicode_DecodeUnicodeInternal(data, size, errors), |
| 228 | size); |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 229 | } |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 230 | } |
| 231 | |
| 232 | static PyObject * |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 233 | utf_7_decode(PyObject *self, |
Amaury Forgeot d'Arc | 5087980 | 2007-11-20 23:31:27 +0000 | [diff] [blame] | 234 | PyObject *args) |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 235 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 236 | Py_buffer pbuf; |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 237 | const char *errors = NULL; |
Amaury Forgeot d'Arc | 5087980 | 2007-11-20 23:31:27 +0000 | [diff] [blame] | 238 | int final = 0; |
| 239 | Py_ssize_t consumed; |
| 240 | PyObject *decoded = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 241 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 242 | if (!PyArg_ParseTuple(args, "s*|zi:utf_7_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 243 | &pbuf, &errors, &final)) |
| 244 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 245 | consumed = pbuf.len; |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 246 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 247 | decoded = PyUnicode_DecodeUTF7Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 248 | final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 249 | PyBuffer_Release(&pbuf); |
Amaury Forgeot d'Arc | 5087980 | 2007-11-20 23:31:27 +0000 | [diff] [blame] | 250 | if (decoded == NULL) |
| 251 | return NULL; |
| 252 | return codec_tuple(decoded, consumed); |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 253 | } |
| 254 | |
| 255 | static PyObject * |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 256 | utf_8_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 257 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 258 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 259 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 260 | const char *errors = NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 261 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 262 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 263 | PyObject *decoded = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 264 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 265 | if (!PyArg_ParseTuple(args, "s*|zi:utf_8_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 266 | &pbuf, &errors, &final)) |
| 267 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 268 | consumed = pbuf.len; |
| 269 | |
| 270 | decoded = PyUnicode_DecodeUTF8Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 271 | final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 272 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 273 | if (decoded == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 274 | return NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 275 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 276 | } |
| 277 | |
| 278 | static PyObject * |
| 279 | utf_16_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 280 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 281 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 282 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 283 | const char *errors = NULL; |
| 284 | int byteorder = 0; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 285 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 286 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 287 | PyObject *decoded; |
| 288 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 289 | if (!PyArg_ParseTuple(args, "s*|zi:utf_16_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 290 | &pbuf, &errors, &final)) |
| 291 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 292 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 293 | decoded = PyUnicode_DecodeUTF16Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 294 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 295 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 296 | if (decoded == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 297 | return NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 298 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 299 | } |
| 300 | |
| 301 | static PyObject * |
| 302 | utf_16_le_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 303 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 304 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 305 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 306 | const char *errors = NULL; |
| 307 | int byteorder = -1; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 308 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 309 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 310 | PyObject *decoded = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 311 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 312 | if (!PyArg_ParseTuple(args, "s*|zi:utf_16_le_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 313 | &pbuf, &errors, &final)) |
| 314 | return NULL; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 315 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 316 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 317 | decoded = PyUnicode_DecodeUTF16Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 318 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 319 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 320 | if (decoded == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 321 | return NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 322 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 323 | } |
| 324 | |
| 325 | static PyObject * |
| 326 | utf_16_be_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 327 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 328 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 329 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 330 | const char *errors = NULL; |
| 331 | int byteorder = 1; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 332 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 333 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 334 | PyObject *decoded = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 335 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 336 | if (!PyArg_ParseTuple(args, "s*|zi:utf_16_be_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 337 | &pbuf, &errors, &final)) |
| 338 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 339 | |
| 340 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 341 | decoded = PyUnicode_DecodeUTF16Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 342 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 343 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 344 | if (decoded == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 345 | return NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 346 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 347 | } |
| 348 | |
| 349 | /* This non-standard version also provides access to the byteorder |
| 350 | parameter of the builtin UTF-16 codec. |
| 351 | |
| 352 | It returns a tuple (unicode, bytesread, byteorder) with byteorder |
| 353 | being the value in effect at the end of data. |
| 354 | |
| 355 | */ |
| 356 | |
| 357 | static PyObject * |
| 358 | utf_16_ex_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 359 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 360 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 361 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 362 | const char *errors = NULL; |
| 363 | int byteorder = 0; |
| 364 | PyObject *unicode, *tuple; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 365 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 366 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 367 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 368 | if (!PyArg_ParseTuple(args, "s*|zii:utf_16_ex_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 369 | &pbuf, &errors, &byteorder, &final)) |
| 370 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 371 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 372 | unicode = PyUnicode_DecodeUTF16Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 373 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 374 | PyBuffer_Release(&pbuf); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 375 | if (unicode == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 376 | return NULL; |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 377 | tuple = Py_BuildValue("Oni", unicode, consumed, byteorder); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 378 | Py_DECREF(unicode); |
| 379 | return tuple; |
| 380 | } |
| 381 | |
| 382 | static PyObject * |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 383 | utf_32_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 384 | PyObject *args) |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 385 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 386 | Py_buffer pbuf; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 387 | const char *errors = NULL; |
| 388 | int byteorder = 0; |
| 389 | int final = 0; |
| 390 | Py_ssize_t consumed; |
| 391 | PyObject *decoded; |
| 392 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 393 | if (!PyArg_ParseTuple(args, "s*|zi:utf_32_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 394 | &pbuf, &errors, &final)) |
| 395 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 396 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 397 | decoded = PyUnicode_DecodeUTF32Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 398 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 399 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 400 | if (decoded == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 401 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 402 | return codec_tuple(decoded, consumed); |
| 403 | } |
| 404 | |
| 405 | static PyObject * |
| 406 | utf_32_le_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 407 | PyObject *args) |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 408 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 409 | Py_buffer pbuf; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 410 | const char *errors = NULL; |
| 411 | int byteorder = -1; |
| 412 | int final = 0; |
| 413 | Py_ssize_t consumed; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 414 | PyObject *decoded; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 415 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 416 | if (!PyArg_ParseTuple(args, "s*|zi:utf_32_le_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 417 | &pbuf, &errors, &final)) |
| 418 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 419 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 420 | decoded = PyUnicode_DecodeUTF32Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 421 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 422 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 423 | if (decoded == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 424 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 425 | return codec_tuple(decoded, consumed); |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 426 | } |
| 427 | |
| 428 | static PyObject * |
| 429 | utf_32_be_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 430 | PyObject *args) |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 431 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 432 | Py_buffer pbuf; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 433 | const char *errors = NULL; |
| 434 | int byteorder = 1; |
| 435 | int final = 0; |
| 436 | Py_ssize_t consumed; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 437 | PyObject *decoded; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 438 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 439 | if (!PyArg_ParseTuple(args, "s*|zi:utf_32_be_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 440 | &pbuf, &errors, &final)) |
| 441 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 442 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 443 | decoded = PyUnicode_DecodeUTF32Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 444 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 445 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 446 | if (decoded == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 447 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 448 | return codec_tuple(decoded, consumed); |
| 449 | } |
| 450 | |
| 451 | /* This non-standard version also provides access to the byteorder |
| 452 | parameter of the builtin UTF-32 codec. |
| 453 | |
| 454 | It returns a tuple (unicode, bytesread, byteorder) with byteorder |
| 455 | being the value in effect at the end of data. |
| 456 | |
| 457 | */ |
| 458 | |
| 459 | static PyObject * |
| 460 | utf_32_ex_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 461 | PyObject *args) |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 462 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 463 | Py_buffer pbuf; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 464 | const char *errors = NULL; |
| 465 | int byteorder = 0; |
| 466 | PyObject *unicode, *tuple; |
| 467 | int final = 0; |
| 468 | Py_ssize_t consumed; |
| 469 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 470 | if (!PyArg_ParseTuple(args, "s*|zii:utf_32_ex_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 471 | &pbuf, &errors, &byteorder, &final)) |
| 472 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 473 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 474 | unicode = PyUnicode_DecodeUTF32Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 475 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 476 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 477 | if (unicode == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 478 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 479 | tuple = Py_BuildValue("Oni", unicode, consumed, byteorder); |
| 480 | Py_DECREF(unicode); |
| 481 | return tuple; |
| 482 | } |
| 483 | |
| 484 | static PyObject * |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 485 | unicode_escape_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 486 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 487 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 488 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 489 | const char *errors = NULL; |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 490 | PyObject *unicode; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 491 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 492 | if (!PyArg_ParseTuple(args, "s*|z:unicode_escape_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 493 | &pbuf, &errors)) |
| 494 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 495 | |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 496 | unicode = PyUnicode_DecodeUnicodeEscape(pbuf.buf, pbuf.len, errors); |
| 497 | PyBuffer_Release(&pbuf); |
| 498 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 499 | } |
| 500 | |
| 501 | static PyObject * |
| 502 | raw_unicode_escape_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 503 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 504 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 505 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 506 | const char *errors = NULL; |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 507 | PyObject *unicode; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 508 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 509 | if (!PyArg_ParseTuple(args, "s*|z:raw_unicode_escape_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 510 | &pbuf, &errors)) |
| 511 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 512 | |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 513 | unicode = PyUnicode_DecodeRawUnicodeEscape(pbuf.buf, pbuf.len, errors); |
| 514 | PyBuffer_Release(&pbuf); |
| 515 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 516 | } |
| 517 | |
| 518 | static PyObject * |
| 519 | latin_1_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 520 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 521 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 522 | Py_buffer pbuf; |
| 523 | PyObject *unicode; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 524 | const char *errors = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 525 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 526 | if (!PyArg_ParseTuple(args, "s*|z:latin_1_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 527 | &pbuf, &errors)) |
| 528 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 529 | |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 530 | unicode = PyUnicode_DecodeLatin1(pbuf.buf, pbuf.len, errors); |
| 531 | PyBuffer_Release(&pbuf); |
| 532 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 533 | } |
| 534 | |
| 535 | static PyObject * |
| 536 | ascii_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 537 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 538 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 539 | Py_buffer pbuf; |
| 540 | PyObject *unicode; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 541 | const char *errors = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 542 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 543 | if (!PyArg_ParseTuple(args, "s*|z:ascii_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 544 | &pbuf, &errors)) |
| 545 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 546 | |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 547 | unicode = PyUnicode_DecodeASCII(pbuf.buf, pbuf.len, errors); |
| 548 | PyBuffer_Release(&pbuf); |
| 549 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 550 | } |
| 551 | |
| 552 | static PyObject * |
| 553 | charmap_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 554 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 555 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 556 | Py_buffer pbuf; |
| 557 | PyObject *unicode; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 558 | const char *errors = NULL; |
| 559 | PyObject *mapping = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 560 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 561 | if (!PyArg_ParseTuple(args, "s*|zO:charmap_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 562 | &pbuf, &errors, &mapping)) |
| 563 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 564 | if (mapping == Py_None) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 565 | mapping = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 566 | |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 567 | unicode = PyUnicode_DecodeCharmap(pbuf.buf, pbuf.len, mapping, errors); |
| 568 | PyBuffer_Release(&pbuf); |
| 569 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 570 | } |
| 571 | |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 572 | #if defined(MS_WINDOWS) && defined(HAVE_USABLE_WCHAR_T) |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 573 | |
| 574 | static PyObject * |
| 575 | mbcs_decode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 576 | PyObject *args) |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 577 | { |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 578 | Py_buffer pbuf; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 579 | const char *errors = NULL; |
Martin v. Löwis | 961b91b | 2006-08-02 13:53:55 +0000 | [diff] [blame] | 580 | int final = 0; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 581 | Py_ssize_t consumed; |
| 582 | PyObject *decoded = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 583 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 584 | if (!PyArg_ParseTuple(args, "s*|zi:mbcs_decode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 585 | &pbuf, &errors, &final)) |
| 586 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 587 | consumed = pbuf.len; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 588 | |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 589 | decoded = PyUnicode_DecodeMBCSStateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 590 | final ? NULL : &consumed); |
Antoine Pitrou | 619f16e | 2010-06-09 16:24:00 +0000 | [diff] [blame] | 591 | PyBuffer_Release(&pbuf); |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 592 | if (decoded == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 593 | return NULL; |
Martin v. Löwis | f91d46a | 2008-08-12 14:49:50 +0000 | [diff] [blame] | 594 | return codec_tuple(decoded, consumed); |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 595 | } |
| 596 | |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 597 | #endif /* MS_WINDOWS */ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 598 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 599 | /* --- Encoder ------------------------------------------------------------ */ |
| 600 | |
| 601 | static PyObject * |
| 602 | readbuffer_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 603 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 604 | { |
| 605 | const char *data; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 606 | Py_ssize_t size; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 607 | const char *errors = NULL; |
| 608 | |
| 609 | if (!PyArg_ParseTuple(args, "s#|z:readbuffer_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 610 | &data, &size, &errors)) |
| 611 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 612 | |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 613 | return codec_tuple(PyString_FromStringAndSize(data, size), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 614 | size); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 615 | } |
| 616 | |
| 617 | static PyObject * |
| 618 | charbuffer_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 619 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 620 | { |
| 621 | const char *data; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 622 | Py_ssize_t size; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 623 | const char *errors = NULL; |
| 624 | |
| 625 | if (!PyArg_ParseTuple(args, "t#|z:charbuffer_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 626 | &data, &size, &errors)) |
| 627 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 628 | |
Gregory P. Smith | dd96db6 | 2008-06-09 04:58:54 +0000 | [diff] [blame] | 629 | return codec_tuple(PyString_FromStringAndSize(data, size), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 630 | size); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 631 | } |
| 632 | |
| 633 | static PyObject * |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 634 | unicode_internal_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 635 | PyObject *args) |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 636 | { |
| 637 | PyObject *obj; |
| 638 | const char *errors = NULL; |
| 639 | const char *data; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 640 | Py_ssize_t size; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 641 | |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 642 | if (!PyArg_ParseTuple(args, "O|z:unicode_internal_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 643 | &obj, &errors)) |
| 644 | return NULL; |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 645 | |
| 646 | if (PyUnicode_Check(obj)) { |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 647 | data = PyUnicode_AS_DATA(obj); |
| 648 | size = PyUnicode_GET_DATA_SIZE(obj); |
| 649 | return codec_tuple(PyString_FromStringAndSize(data, size), |
| 650 | PyUnicode_GET_SIZE(obj)); |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 651 | } |
| 652 | else { |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 653 | if (PyObject_AsReadBuffer(obj, (const void **)&data, &size)) |
| 654 | return NULL; |
| 655 | return codec_tuple(PyString_FromStringAndSize(data, size), |
| 656 | size); |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 657 | } |
| 658 | } |
| 659 | |
| 660 | static PyObject * |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 661 | utf_7_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 662 | PyObject *args) |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 663 | { |
| 664 | PyObject *str, *v; |
| 665 | const char *errors = NULL; |
| 666 | |
| 667 | if (!PyArg_ParseTuple(args, "O|z:utf_7_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 668 | &str, &errors)) |
| 669 | return NULL; |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 670 | |
| 671 | str = PyUnicode_FromObject(str); |
| 672 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 673 | return NULL; |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 674 | v = codec_tuple(PyUnicode_EncodeUTF7(PyUnicode_AS_UNICODE(str), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 675 | PyUnicode_GET_SIZE(str), |
| 676 | 0, |
| 677 | 0, |
| 678 | errors), |
| 679 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 680 | Py_DECREF(str); |
| 681 | return v; |
| 682 | } |
| 683 | |
| 684 | static PyObject * |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 685 | utf_8_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 686 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 687 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 688 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 689 | const char *errors = NULL; |
| 690 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 691 | if (!PyArg_ParseTuple(args, "O|z:utf_8_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 692 | &str, &errors)) |
| 693 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 694 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 695 | str = PyUnicode_FromObject(str); |
| 696 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 697 | return NULL; |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 698 | v = codec_tuple(PyUnicode_EncodeUTF8(PyUnicode_AS_UNICODE(str), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 699 | PyUnicode_GET_SIZE(str), |
| 700 | errors), |
| 701 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 702 | Py_DECREF(str); |
| 703 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 704 | } |
| 705 | |
| 706 | /* This version provides access to the byteorder parameter of the |
| 707 | builtin UTF-16 codecs as optional third argument. It defaults to 0 |
| 708 | which means: use the native byte order and prepend the data with a |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 709 | BOM mark. |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 710 | |
| 711 | */ |
| 712 | |
| 713 | static PyObject * |
| 714 | utf_16_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 715 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 716 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 717 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 718 | const char *errors = NULL; |
| 719 | int byteorder = 0; |
| 720 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 721 | if (!PyArg_ParseTuple(args, "O|zi:utf_16_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 722 | &str, &errors, &byteorder)) |
| 723 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 724 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 725 | str = PyUnicode_FromObject(str); |
| 726 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 727 | return NULL; |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 728 | v = codec_tuple(PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(str), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 729 | PyUnicode_GET_SIZE(str), |
| 730 | errors, |
| 731 | byteorder), |
| 732 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 733 | Py_DECREF(str); |
| 734 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 735 | } |
| 736 | |
| 737 | static PyObject * |
| 738 | utf_16_le_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 739 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 740 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 741 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 742 | const char *errors = NULL; |
| 743 | |
Marc-André Lemburg | 4157dd5 | 2001-06-17 18:32:36 +0000 | [diff] [blame] | 744 | if (!PyArg_ParseTuple(args, "O|z:utf_16_le_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 745 | &str, &errors)) |
| 746 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 747 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 748 | str = PyUnicode_FromObject(str); |
| 749 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 750 | return NULL; |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 751 | v = codec_tuple(PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(str), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 752 | PyUnicode_GET_SIZE(str), |
| 753 | errors, |
| 754 | -1), |
| 755 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 756 | Py_DECREF(str); |
| 757 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 758 | } |
| 759 | |
| 760 | static PyObject * |
| 761 | utf_16_be_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 762 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 763 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 764 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 765 | const char *errors = NULL; |
| 766 | |
Marc-André Lemburg | 4157dd5 | 2001-06-17 18:32:36 +0000 | [diff] [blame] | 767 | if (!PyArg_ParseTuple(args, "O|z:utf_16_be_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 768 | &str, &errors)) |
| 769 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 770 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 771 | str = PyUnicode_FromObject(str); |
| 772 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 773 | return NULL; |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 774 | v = codec_tuple(PyUnicode_EncodeUTF16(PyUnicode_AS_UNICODE(str), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 775 | PyUnicode_GET_SIZE(str), |
| 776 | errors, |
| 777 | +1), |
| 778 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 779 | Py_DECREF(str); |
| 780 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 781 | } |
| 782 | |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 783 | /* This version provides access to the byteorder parameter of the |
| 784 | builtin UTF-32 codecs as optional third argument. It defaults to 0 |
| 785 | which means: use the native byte order and prepend the data with a |
| 786 | BOM mark. |
| 787 | |
| 788 | */ |
| 789 | |
| 790 | static PyObject * |
| 791 | utf_32_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 792 | PyObject *args) |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 793 | { |
| 794 | PyObject *str, *v; |
| 795 | const char *errors = NULL; |
| 796 | int byteorder = 0; |
| 797 | |
| 798 | if (!PyArg_ParseTuple(args, "O|zi:utf_32_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 799 | &str, &errors, &byteorder)) |
| 800 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 801 | |
| 802 | str = PyUnicode_FromObject(str); |
| 803 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 804 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 805 | v = codec_tuple(PyUnicode_EncodeUTF32(PyUnicode_AS_UNICODE(str), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 806 | PyUnicode_GET_SIZE(str), |
| 807 | errors, |
| 808 | byteorder), |
| 809 | PyUnicode_GET_SIZE(str)); |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 810 | Py_DECREF(str); |
| 811 | return v; |
| 812 | } |
| 813 | |
| 814 | static PyObject * |
| 815 | utf_32_le_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 816 | PyObject *args) |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 817 | { |
| 818 | PyObject *str, *v; |
| 819 | const char *errors = NULL; |
| 820 | |
| 821 | if (!PyArg_ParseTuple(args, "O|z:utf_32_le_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 822 | &str, &errors)) |
| 823 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 824 | |
| 825 | str = PyUnicode_FromObject(str); |
| 826 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 827 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 828 | v = codec_tuple(PyUnicode_EncodeUTF32(PyUnicode_AS_UNICODE(str), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 829 | PyUnicode_GET_SIZE(str), |
| 830 | errors, |
| 831 | -1), |
| 832 | PyUnicode_GET_SIZE(str)); |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 833 | Py_DECREF(str); |
| 834 | return v; |
| 835 | } |
| 836 | |
| 837 | static PyObject * |
| 838 | utf_32_be_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 839 | PyObject *args) |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 840 | { |
| 841 | PyObject *str, *v; |
| 842 | const char *errors = NULL; |
| 843 | |
| 844 | if (!PyArg_ParseTuple(args, "O|z:utf_32_be_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 845 | &str, &errors)) |
| 846 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 847 | |
| 848 | str = PyUnicode_FromObject(str); |
| 849 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 850 | return NULL; |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 851 | v = codec_tuple(PyUnicode_EncodeUTF32(PyUnicode_AS_UNICODE(str), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 852 | PyUnicode_GET_SIZE(str), |
| 853 | errors, |
| 854 | +1), |
| 855 | PyUnicode_GET_SIZE(str)); |
Walter Dörwald | 6e39080 | 2007-08-17 16:41:28 +0000 | [diff] [blame] | 856 | Py_DECREF(str); |
| 857 | return v; |
| 858 | } |
| 859 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 860 | static PyObject * |
| 861 | unicode_escape_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 862 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 863 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 864 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 865 | const char *errors = NULL; |
| 866 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 867 | if (!PyArg_ParseTuple(args, "O|z:unicode_escape_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 868 | &str, &errors)) |
| 869 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 870 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 871 | str = PyUnicode_FromObject(str); |
| 872 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 873 | return NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 874 | v = codec_tuple(PyUnicode_EncodeUnicodeEscape(PyUnicode_AS_UNICODE(str), |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 875 | PyUnicode_GET_SIZE(str)), |
| 876 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 877 | Py_DECREF(str); |
| 878 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 879 | } |
| 880 | |
| 881 | static PyObject * |
| 882 | raw_unicode_escape_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 883 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 884 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 885 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 886 | const char *errors = NULL; |
| 887 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 888 | if (!PyArg_ParseTuple(args, "O|z:raw_unicode_escape_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 889 | &str, &errors)) |
| 890 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 891 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 892 | str = PyUnicode_FromObject(str); |
| 893 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 894 | return NULL; |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 895 | v = codec_tuple(PyUnicode_EncodeRawUnicodeEscape( |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 896 | PyUnicode_AS_UNICODE(str), |
| 897 | PyUnicode_GET_SIZE(str)), |
| 898 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 899 | Py_DECREF(str); |
| 900 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 901 | } |
| 902 | |
| 903 | static PyObject * |
| 904 | latin_1_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 905 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 906 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 907 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 908 | const char *errors = NULL; |
| 909 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 910 | if (!PyArg_ParseTuple(args, "O|z:latin_1_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 911 | &str, &errors)) |
| 912 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 913 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 914 | str = PyUnicode_FromObject(str); |
| 915 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 916 | return NULL; |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 917 | v = codec_tuple(PyUnicode_EncodeLatin1( |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 918 | PyUnicode_AS_UNICODE(str), |
| 919 | PyUnicode_GET_SIZE(str), |
| 920 | errors), |
| 921 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 922 | Py_DECREF(str); |
| 923 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 924 | } |
| 925 | |
| 926 | static PyObject * |
| 927 | ascii_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 928 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 929 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 930 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 931 | const char *errors = NULL; |
| 932 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 933 | if (!PyArg_ParseTuple(args, "O|z:ascii_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 934 | &str, &errors)) |
| 935 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 936 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 937 | str = PyUnicode_FromObject(str); |
| 938 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 939 | return NULL; |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 940 | v = codec_tuple(PyUnicode_EncodeASCII( |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 941 | PyUnicode_AS_UNICODE(str), |
| 942 | PyUnicode_GET_SIZE(str), |
| 943 | errors), |
| 944 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 945 | Py_DECREF(str); |
| 946 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 947 | } |
| 948 | |
| 949 | static PyObject * |
| 950 | charmap_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 951 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 952 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 953 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 954 | const char *errors = NULL; |
| 955 | PyObject *mapping = NULL; |
| 956 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 957 | if (!PyArg_ParseTuple(args, "O|zO:charmap_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 958 | &str, &errors, &mapping)) |
| 959 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 960 | if (mapping == Py_None) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 961 | mapping = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 962 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 963 | str = PyUnicode_FromObject(str); |
| 964 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 965 | return NULL; |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 966 | v = codec_tuple(PyUnicode_EncodeCharmap( |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 967 | PyUnicode_AS_UNICODE(str), |
| 968 | PyUnicode_GET_SIZE(str), |
| 969 | mapping, |
| 970 | errors), |
| 971 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 972 | Py_DECREF(str); |
| 973 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 974 | } |
| 975 | |
Martin v. Löwis | 3f76779 | 2006-06-04 19:36:28 +0000 | [diff] [blame] | 976 | static PyObject* |
| 977 | charmap_build(PyObject *self, PyObject *args) |
| 978 | { |
| 979 | PyObject *map; |
| 980 | if (!PyArg_ParseTuple(args, "U:charmap_build", &map)) |
| 981 | return NULL; |
| 982 | return PyUnicode_BuildEncodingMap(map); |
| 983 | } |
| 984 | |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 985 | #if defined(MS_WINDOWS) && defined(HAVE_USABLE_WCHAR_T) |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 986 | |
| 987 | static PyObject * |
| 988 | mbcs_encode(PyObject *self, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 989 | PyObject *args) |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 990 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 991 | PyObject *str, *v; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 992 | const char *errors = NULL; |
| 993 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 994 | if (!PyArg_ParseTuple(args, "O|z:mbcs_encode", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 995 | &str, &errors)) |
| 996 | return NULL; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 997 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 998 | str = PyUnicode_FromObject(str); |
| 999 | if (str == NULL) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1000 | return NULL; |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1001 | v = codec_tuple(PyUnicode_EncodeMBCS( |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1002 | PyUnicode_AS_UNICODE(str), |
| 1003 | PyUnicode_GET_SIZE(str), |
| 1004 | errors), |
| 1005 | PyUnicode_GET_SIZE(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1006 | Py_DECREF(str); |
| 1007 | return v; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1008 | } |
| 1009 | |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 1010 | #endif /* MS_WINDOWS */ |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 1011 | #endif /* Py_USING_UNICODE */ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1012 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1013 | /* --- Error handler registry --------------------------------------------- */ |
| 1014 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1015 | PyDoc_STRVAR(register_error__doc__, |
| 1016 | "register_error(errors, handler)\n\ |
| 1017 | \n\ |
| 1018 | Register the specified error handler under the name\n\ |
| 1019 | errors. handler must be a callable object, that\n\ |
| 1020 | will be called with an exception instance containing\n\ |
| 1021 | information about the location of the encoding/decoding\n\ |
| 1022 | error and must return a (replacement, new position) tuple."); |
| 1023 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1024 | static PyObject *register_error(PyObject *self, PyObject *args) |
| 1025 | { |
| 1026 | const char *name; |
| 1027 | PyObject *handler; |
| 1028 | |
| 1029 | if (!PyArg_ParseTuple(args, "sO:register_error", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1030 | &name, &handler)) |
| 1031 | return NULL; |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1032 | if (PyCodec_RegisterError(name, handler)) |
| 1033 | return NULL; |
Georg Brandl | 96a8c39 | 2006-05-29 21:04:52 +0000 | [diff] [blame] | 1034 | Py_RETURN_NONE; |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1035 | } |
| 1036 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1037 | PyDoc_STRVAR(lookup_error__doc__, |
| 1038 | "lookup_error(errors) -> handler\n\ |
| 1039 | \n\ |
| 1040 | Return the error handler for the specified error handling name\n\ |
| 1041 | or raise a LookupError, if no handler exists under this name."); |
| 1042 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1043 | static PyObject *lookup_error(PyObject *self, PyObject *args) |
| 1044 | { |
| 1045 | const char *name; |
| 1046 | |
| 1047 | if (!PyArg_ParseTuple(args, "s:lookup_error", |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1048 | &name)) |
| 1049 | return NULL; |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1050 | return PyCodec_LookupError(name); |
| 1051 | } |
| 1052 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1053 | /* --- Module API --------------------------------------------------------- */ |
| 1054 | |
| 1055 | static PyMethodDef _codecs_functions[] = { |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1056 | {"register", codec_register, METH_O, |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1057 | register__doc__}, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1058 | {"lookup", codec_lookup, METH_VARARGS, |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1059 | lookup__doc__}, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1060 | {"encode", codec_encode, METH_VARARGS, |
| 1061 | encode__doc__}, |
| 1062 | {"decode", codec_decode, METH_VARARGS, |
| 1063 | decode__doc__}, |
| 1064 | {"escape_encode", escape_encode, METH_VARARGS}, |
| 1065 | {"escape_decode", escape_decode, METH_VARARGS}, |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 1066 | #ifdef Py_USING_UNICODE |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1067 | {"utf_8_encode", utf_8_encode, METH_VARARGS}, |
| 1068 | {"utf_8_decode", utf_8_decode, METH_VARARGS}, |
| 1069 | {"utf_7_encode", utf_7_encode, METH_VARARGS}, |
| 1070 | {"utf_7_decode", utf_7_decode, METH_VARARGS}, |
| 1071 | {"utf_16_encode", utf_16_encode, METH_VARARGS}, |
| 1072 | {"utf_16_le_encode", utf_16_le_encode, METH_VARARGS}, |
| 1073 | {"utf_16_be_encode", utf_16_be_encode, METH_VARARGS}, |
| 1074 | {"utf_16_decode", utf_16_decode, METH_VARARGS}, |
| 1075 | {"utf_16_le_decode", utf_16_le_decode, METH_VARARGS}, |
| 1076 | {"utf_16_be_decode", utf_16_be_decode, METH_VARARGS}, |
| 1077 | {"utf_16_ex_decode", utf_16_ex_decode, METH_VARARGS}, |
| 1078 | {"utf_32_encode", utf_32_encode, METH_VARARGS}, |
| 1079 | {"utf_32_le_encode", utf_32_le_encode, METH_VARARGS}, |
| 1080 | {"utf_32_be_encode", utf_32_be_encode, METH_VARARGS}, |
| 1081 | {"utf_32_decode", utf_32_decode, METH_VARARGS}, |
| 1082 | {"utf_32_le_decode", utf_32_le_decode, METH_VARARGS}, |
| 1083 | {"utf_32_be_decode", utf_32_be_decode, METH_VARARGS}, |
| 1084 | {"utf_32_ex_decode", utf_32_ex_decode, METH_VARARGS}, |
| 1085 | {"unicode_escape_encode", unicode_escape_encode, METH_VARARGS}, |
| 1086 | {"unicode_escape_decode", unicode_escape_decode, METH_VARARGS}, |
| 1087 | {"unicode_internal_encode", unicode_internal_encode, METH_VARARGS}, |
| 1088 | {"unicode_internal_decode", unicode_internal_decode, METH_VARARGS}, |
| 1089 | {"raw_unicode_escape_encode", raw_unicode_escape_encode, METH_VARARGS}, |
| 1090 | {"raw_unicode_escape_decode", raw_unicode_escape_decode, METH_VARARGS}, |
| 1091 | {"latin_1_encode", latin_1_encode, METH_VARARGS}, |
| 1092 | {"latin_1_decode", latin_1_decode, METH_VARARGS}, |
| 1093 | {"ascii_encode", ascii_encode, METH_VARARGS}, |
| 1094 | {"ascii_decode", ascii_decode, METH_VARARGS}, |
| 1095 | {"charmap_encode", charmap_encode, METH_VARARGS}, |
| 1096 | {"charmap_decode", charmap_decode, METH_VARARGS}, |
| 1097 | {"charmap_build", charmap_build, METH_VARARGS}, |
| 1098 | {"readbuffer_encode", readbuffer_encode, METH_VARARGS}, |
| 1099 | {"charbuffer_encode", charbuffer_encode, METH_VARARGS}, |
Martin v. Löwis | 6238d2b | 2002-06-30 15:26:10 +0000 | [diff] [blame] | 1100 | #if defined(MS_WINDOWS) && defined(HAVE_USABLE_WCHAR_T) |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1101 | {"mbcs_encode", mbcs_encode, METH_VARARGS}, |
| 1102 | {"mbcs_decode", mbcs_decode, METH_VARARGS}, |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1103 | #endif |
Martin v. Löwis | 339d0f7 | 2001-08-17 18:39:25 +0000 | [diff] [blame] | 1104 | #endif /* Py_USING_UNICODE */ |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1105 | {"register_error", register_error, METH_VARARGS, |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1106 | register_error__doc__}, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1107 | {"lookup_error", lookup_error, METH_VARARGS, |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1108 | lookup_error__doc__}, |
Antoine Pitrou | c83ea13 | 2010-05-09 14:46:46 +0000 | [diff] [blame] | 1109 | {NULL, NULL} /* sentinel */ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1110 | }; |
| 1111 | |
Mark Hammond | fe51c6d | 2002-08-02 02:27:13 +0000 | [diff] [blame] | 1112 | PyMODINIT_FUNC |
Thomas Wouters | f3f33dc | 2000-07-21 06:00:07 +0000 | [diff] [blame] | 1113 | init_codecs(void) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1114 | { |
| 1115 | Py_InitModule("_codecs", _codecs_functions); |
| 1116 | } |