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 | |
Guido van Rossum | 36e0a92 | 2007-07-20 04:05:57 +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 | f95a1b3 | 2010-05-09 15:52:27 +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 | |
Victor Stinner | 87a7c82 | 2011-11-10 20:05:55 +0100 | [diff] [blame] | 41 | #ifdef MS_WINDOWS |
| 42 | #include <windows.h> |
| 43 | #endif |
| 44 | |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 45 | /*[clinic input] |
| 46 | module _codecs |
| 47 | [clinic start generated code]*/ |
| 48 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=e1390e3da3cb9deb]*/ |
| 49 | |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 50 | #include "clinic/_codecsmodule.c.h" |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 51 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 52 | /* --- Registry ----------------------------------------------------------- */ |
| 53 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 54 | PyDoc_STRVAR(register__doc__, |
| 55 | "register(search_function)\n\ |
| 56 | \n\ |
| 57 | Register a codec search function. Search functions are expected to take\n\ |
Nick Coghlan | b9fdb7a | 2015-01-07 00:22:00 +1000 | [diff] [blame] | 58 | one argument, the encoding name in all lower case letters, and either\n\ |
| 59 | return None, or a tuple of functions (encoder, decoder, stream_reader,\n\ |
| 60 | stream_writer) (or a CodecInfo object)."); |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 61 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 62 | static |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 63 | PyObject *codec_register(PyObject *self, PyObject *search_function) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 64 | { |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 65 | if (PyCodec_Register(search_function)) |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 66 | return NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 67 | |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 68 | Py_RETURN_NONE; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 69 | } |
| 70 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 71 | PyDoc_STRVAR(lookup__doc__, |
Guido van Rossum | 36e0a92 | 2007-07-20 04:05:57 +0000 | [diff] [blame] | 72 | "lookup(encoding) -> CodecInfo\n\ |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 73 | \n\ |
| 74 | Looks up a codec tuple in the Python codec registry and returns\n\ |
Benjamin Peterson | f07d002 | 2009-03-21 17:31:58 +0000 | [diff] [blame] | 75 | a CodecInfo object."); |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 76 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 77 | static |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 78 | PyObject *codec_lookup(PyObject *self, PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 79 | { |
| 80 | char *encoding; |
| 81 | |
| 82 | if (!PyArg_ParseTuple(args, "s:lookup", &encoding)) |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 83 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 84 | |
| 85 | return _PyCodec_Lookup(encoding); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 86 | } |
| 87 | |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 88 | PyDoc_STRVAR(encode__doc__, |
| 89 | "encode(obj, [encoding[,errors]]) -> object\n\ |
| 90 | \n\ |
| 91 | Encodes obj using the codec registered for encoding. encoding defaults\n\ |
| 92 | to the default encoding. errors may be given to set a different error\n\ |
| 93 | handling scheme. Default is 'strict' meaning that encoding errors raise\n\ |
| 94 | a ValueError. Other possible values are 'ignore', 'replace' and\n\ |
| 95 | 'xmlcharrefreplace' as well as any other name registered with\n\ |
| 96 | codecs.register_error that can handle ValueErrors."); |
| 97 | |
| 98 | static PyObject * |
Victor Stinner | a57dfd0 | 2014-05-14 17:13:14 +0200 | [diff] [blame] | 99 | codec_encode(PyObject *self, PyObject *args, PyObject *kwargs) |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 100 | { |
Victor Stinner | a57dfd0 | 2014-05-14 17:13:14 +0200 | [diff] [blame] | 101 | static char *kwlist[] = {"obj", "encoding", "errors", NULL}; |
Brett Cannon | 3e377de | 2004-07-10 21:41:14 +0000 | [diff] [blame] | 102 | const char *encoding = NULL; |
| 103 | const char *errors = NULL; |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 104 | PyObject *v; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 105 | |
Victor Stinner | a57dfd0 | 2014-05-14 17:13:14 +0200 | [diff] [blame] | 106 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|ss:encode", kwlist, |
| 107 | &v, &encoding, &errors)) |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 108 | return NULL; |
| 109 | |
| 110 | if (encoding == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 111 | encoding = PyUnicode_GetDefaultEncoding(); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 112 | |
| 113 | /* Encode via the codec registry */ |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 114 | return PyCodec_Encode(v, encoding, errors); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 115 | } |
| 116 | |
| 117 | PyDoc_STRVAR(decode__doc__, |
| 118 | "decode(obj, [encoding[,errors]]) -> object\n\ |
| 119 | \n\ |
| 120 | Decodes obj using the codec registered for encoding. encoding defaults\n\ |
| 121 | to the default encoding. errors may be given to set a different error\n\ |
| 122 | handling scheme. Default is 'strict' meaning that encoding errors raise\n\ |
| 123 | a ValueError. Other possible values are 'ignore' and 'replace'\n\ |
Benjamin Peterson | f10a79a | 2008-10-11 00:49:57 +0000 | [diff] [blame] | 124 | 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] | 125 | able to handle ValueErrors."); |
| 126 | |
| 127 | static PyObject * |
Victor Stinner | a57dfd0 | 2014-05-14 17:13:14 +0200 | [diff] [blame] | 128 | codec_decode(PyObject *self, PyObject *args, PyObject *kwargs) |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 129 | { |
Victor Stinner | a57dfd0 | 2014-05-14 17:13:14 +0200 | [diff] [blame] | 130 | static char *kwlist[] = {"obj", "encoding", "errors", NULL}; |
Brett Cannon | 3e377de | 2004-07-10 21:41:14 +0000 | [diff] [blame] | 131 | const char *encoding = NULL; |
| 132 | const char *errors = NULL; |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 133 | PyObject *v; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 134 | |
Victor Stinner | a57dfd0 | 2014-05-14 17:13:14 +0200 | [diff] [blame] | 135 | if (!PyArg_ParseTupleAndKeywords(args, kwargs, "O|ss:decode", kwlist, |
| 136 | &v, &encoding, &errors)) |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 137 | return NULL; |
| 138 | |
| 139 | if (encoding == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 140 | encoding = PyUnicode_GetDefaultEncoding(); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 141 | |
| 142 | /* Decode via the codec registry */ |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 143 | return PyCodec_Decode(v, encoding, errors); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 144 | } |
| 145 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 146 | /* --- Helpers ------------------------------------------------------------ */ |
| 147 | |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 148 | /*[clinic input] |
| 149 | _codecs._forget_codec |
| 150 | |
| 151 | encoding: str |
| 152 | / |
| 153 | |
| 154 | Purge the named codec from the internal codec lookup cache |
| 155 | [clinic start generated code]*/ |
| 156 | |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 157 | static PyObject * |
| 158 | _codecs__forget_codec_impl(PyModuleDef *module, const char *encoding) |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 159 | /*[clinic end generated code: output=b56a9b99d2d28080 input=18d5d92d0e386c38]*/ |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 160 | { |
| 161 | if (_PyCodec_Forget(encoding) < 0) { |
| 162 | return NULL; |
| 163 | }; |
| 164 | Py_RETURN_NONE; |
| 165 | } |
| 166 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 167 | static |
| 168 | PyObject *codec_tuple(PyObject *unicode, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 169 | Py_ssize_t len) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 170 | { |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 171 | PyObject *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 172 | if (unicode == NULL) |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 173 | return NULL; |
| 174 | v = Py_BuildValue("On", unicode, len); |
| 175 | Py_DECREF(unicode); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 176 | return v; |
| 177 | } |
| 178 | |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 179 | /* --- String codecs ------------------------------------------------------ */ |
| 180 | static PyObject * |
| 181 | escape_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 182 | PyObject *args) |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 183 | { |
Serhiy Storchaka | 8490f5a | 2015-03-20 09:00:36 +0200 | [diff] [blame] | 184 | Py_buffer pbuf; |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 185 | const char *errors = NULL; |
Serhiy Storchaka | 8490f5a | 2015-03-20 09:00:36 +0200 | [diff] [blame] | 186 | PyObject *result; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 187 | |
Serhiy Storchaka | 8490f5a | 2015-03-20 09:00:36 +0200 | [diff] [blame] | 188 | if (!PyArg_ParseTuple(args, "s*|z:escape_decode", |
| 189 | &pbuf, &errors)) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 190 | return NULL; |
Serhiy Storchaka | 8490f5a | 2015-03-20 09:00:36 +0200 | [diff] [blame] | 191 | result = codec_tuple( |
| 192 | PyBytes_DecodeEscape(pbuf.buf, pbuf.len, errors, 0, NULL), |
| 193 | pbuf.len); |
| 194 | PyBuffer_Release(&pbuf); |
| 195 | return result; |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 196 | } |
| 197 | |
| 198 | static PyObject * |
| 199 | escape_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 200 | PyObject *args) |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 201 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 202 | PyObject *str; |
| 203 | Py_ssize_t size; |
| 204 | Py_ssize_t newsize; |
| 205 | const char *errors = NULL; |
| 206 | PyObject *v; |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 207 | |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 208 | if (!PyArg_ParseTuple(args, "O!|z:escape_encode", |
| 209 | &PyBytes_Type, &str, &errors)) |
| 210 | return NULL; |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 211 | |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 212 | size = PyBytes_GET_SIZE(str); |
Mark Dickinson | c04ddff | 2012-10-06 18:04:49 +0100 | [diff] [blame] | 213 | if (size > PY_SSIZE_T_MAX / 4) { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 214 | PyErr_SetString(PyExc_OverflowError, |
| 215 | "string is too large to encode"); |
| 216 | return NULL; |
| 217 | } |
Mark Dickinson | c04ddff | 2012-10-06 18:04:49 +0100 | [diff] [blame] | 218 | newsize = 4*size; |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 219 | v = PyBytes_FromStringAndSize(NULL, newsize); |
| 220 | |
| 221 | if (v == NULL) { |
| 222 | return NULL; |
| 223 | } |
| 224 | else { |
Antoine Pitrou | 9ed5f27 | 2013-08-13 20:18:52 +0200 | [diff] [blame] | 225 | Py_ssize_t i; |
| 226 | char c; |
| 227 | char *p = PyBytes_AS_STRING(v); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 228 | |
| 229 | for (i = 0; i < size; i++) { |
| 230 | /* There's at least enough room for a hex escape */ |
| 231 | assert(newsize - (p - PyBytes_AS_STRING(v)) >= 4); |
| 232 | c = PyBytes_AS_STRING(str)[i]; |
| 233 | if (c == '\'' || c == '\\') |
| 234 | *p++ = '\\', *p++ = c; |
| 235 | else if (c == '\t') |
| 236 | *p++ = '\\', *p++ = 't'; |
| 237 | else if (c == '\n') |
| 238 | *p++ = '\\', *p++ = 'n'; |
| 239 | else if (c == '\r') |
| 240 | *p++ = '\\', *p++ = 'r'; |
| 241 | else if (c < ' ' || c >= 0x7f) { |
| 242 | *p++ = '\\'; |
| 243 | *p++ = 'x'; |
Victor Stinner | f5cff56 | 2011-10-14 02:13:11 +0200 | [diff] [blame] | 244 | *p++ = Py_hexdigits[(c & 0xf0) >> 4]; |
| 245 | *p++ = Py_hexdigits[c & 0xf]; |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 246 | } |
| 247 | else |
| 248 | *p++ = c; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 249 | } |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 250 | *p = '\0'; |
| 251 | if (_PyBytes_Resize(&v, (p - PyBytes_AS_STRING(v)))) { |
| 252 | return NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 253 | } |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 254 | } |
Walter Dörwald | 1ab8330 | 2007-05-18 17:15:44 +0000 | [diff] [blame] | 255 | |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 256 | return codec_tuple(v, size); |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 257 | } |
| 258 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 259 | /* --- Decoder ------------------------------------------------------------ */ |
| 260 | |
| 261 | static PyObject * |
| 262 | unicode_internal_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 263 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 264 | { |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 265 | PyObject *obj; |
| 266 | const char *errors = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 267 | |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 268 | if (!PyArg_ParseTuple(args, "O|z:unicode_internal_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 269 | &obj, &errors)) |
| 270 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 271 | |
Marc-André Lemburg | 29273c8 | 2003-02-04 19:35:03 +0000 | [diff] [blame] | 272 | if (PyUnicode_Check(obj)) { |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 273 | if (PyUnicode_READY(obj) < 0) |
| 274 | return NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 275 | Py_INCREF(obj); |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 276 | return codec_tuple(obj, PyUnicode_GET_LENGTH(obj)); |
Marc-André Lemburg | 29273c8 | 2003-02-04 19:35:03 +0000 | [diff] [blame] | 277 | } |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 278 | else { |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 279 | Py_buffer view; |
| 280 | PyObject *result; |
| 281 | if (PyObject_GetBuffer(obj, &view, PyBUF_SIMPLE) != 0) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 282 | return NULL; |
Walter Dörwald | a47d1c0 | 2005-08-30 10:23:14 +0000 | [diff] [blame] | 283 | |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 284 | result = codec_tuple( |
| 285 | _PyUnicode_DecodeUnicodeInternal(view.buf, view.len, errors), |
| 286 | view.len); |
| 287 | PyBuffer_Release(&view); |
| 288 | return result; |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 289 | } |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 290 | } |
| 291 | |
| 292 | static PyObject * |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 293 | utf_7_decode(PyObject *self, |
Christian Heimes | 5d14c2b | 2007-11-20 23:38:09 +0000 | [diff] [blame] | 294 | PyObject *args) |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 295 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 296 | Py_buffer pbuf; |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 297 | const char *errors = NULL; |
Christian Heimes | 5d14c2b | 2007-11-20 23:38:09 +0000 | [diff] [blame] | 298 | int final = 0; |
| 299 | Py_ssize_t consumed; |
| 300 | PyObject *decoded = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 301 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 302 | if (!PyArg_ParseTuple(args, "y*|zi:utf_7_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 303 | &pbuf, &errors, &final)) |
| 304 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 305 | consumed = pbuf.len; |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 306 | |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 307 | decoded = PyUnicode_DecodeUTF7Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 308 | final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 309 | PyBuffer_Release(&pbuf); |
Christian Heimes | 5d14c2b | 2007-11-20 23:38:09 +0000 | [diff] [blame] | 310 | if (decoded == NULL) |
| 311 | return NULL; |
| 312 | return codec_tuple(decoded, consumed); |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 313 | } |
| 314 | |
| 315 | static PyObject * |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 316 | utf_8_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 317 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 318 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 319 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 320 | const char *errors = NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 321 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 322 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 323 | PyObject *decoded = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 324 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 325 | if (!PyArg_ParseTuple(args, "y*|zi:utf_8_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 326 | &pbuf, &errors, &final)) |
| 327 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 328 | consumed = pbuf.len; |
| 329 | |
| 330 | decoded = PyUnicode_DecodeUTF8Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 331 | final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 332 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 333 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 334 | return NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 335 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 336 | } |
| 337 | |
| 338 | static PyObject * |
| 339 | utf_16_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 340 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 341 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 342 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 343 | const char *errors = NULL; |
| 344 | int byteorder = 0; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 345 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 346 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 347 | PyObject *decoded; |
| 348 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 349 | if (!PyArg_ParseTuple(args, "y*|zi:utf_16_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 350 | &pbuf, &errors, &final)) |
| 351 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 352 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 353 | decoded = PyUnicode_DecodeUTF16Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 354 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 355 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 356 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 357 | return NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 358 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 359 | } |
| 360 | |
| 361 | static PyObject * |
| 362 | utf_16_le_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 363 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 364 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 365 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 366 | const char *errors = NULL; |
| 367 | int byteorder = -1; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 368 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 369 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 370 | PyObject *decoded = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 371 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 372 | if (!PyArg_ParseTuple(args, "y*|zi:utf_16_le_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 373 | &pbuf, &errors, &final)) |
| 374 | return NULL; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 375 | |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 376 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 377 | decoded = PyUnicode_DecodeUTF16Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 378 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 379 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 380 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 381 | return NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 382 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 383 | } |
| 384 | |
| 385 | static PyObject * |
| 386 | utf_16_be_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 387 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 388 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 389 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 390 | const char *errors = NULL; |
| 391 | int byteorder = 1; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 392 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 393 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 394 | PyObject *decoded = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 395 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 396 | if (!PyArg_ParseTuple(args, "y*|zi:utf_16_be_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 397 | &pbuf, &errors, &final)) |
| 398 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 399 | |
| 400 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 401 | decoded = PyUnicode_DecodeUTF16Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 402 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 403 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 404 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 405 | return NULL; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 406 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 407 | } |
| 408 | |
| 409 | /* This non-standard version also provides access to the byteorder |
| 410 | parameter of the builtin UTF-16 codec. |
| 411 | |
| 412 | It returns a tuple (unicode, bytesread, byteorder) with byteorder |
| 413 | being the value in effect at the end of data. |
| 414 | |
| 415 | */ |
| 416 | |
| 417 | static PyObject * |
| 418 | utf_16_ex_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 419 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 420 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 421 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 422 | const char *errors = NULL; |
| 423 | int byteorder = 0; |
| 424 | PyObject *unicode, *tuple; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 425 | int final = 0; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 426 | Py_ssize_t consumed; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 427 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 428 | if (!PyArg_ParseTuple(args, "y*|zii:utf_16_ex_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 429 | &pbuf, &errors, &byteorder, &final)) |
| 430 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 431 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 432 | unicode = PyUnicode_DecodeUTF16Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 433 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 434 | PyBuffer_Release(&pbuf); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 435 | if (unicode == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 436 | return NULL; |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 437 | tuple = Py_BuildValue("Oni", unicode, consumed, byteorder); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 438 | Py_DECREF(unicode); |
| 439 | return tuple; |
| 440 | } |
| 441 | |
| 442 | static PyObject * |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 443 | utf_32_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 444 | PyObject *args) |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 445 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 446 | Py_buffer pbuf; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 447 | const char *errors = NULL; |
| 448 | int byteorder = 0; |
| 449 | int final = 0; |
| 450 | Py_ssize_t consumed; |
| 451 | PyObject *decoded; |
| 452 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 453 | if (!PyArg_ParseTuple(args, "y*|zi:utf_32_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 454 | &pbuf, &errors, &final)) |
| 455 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 456 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 457 | decoded = PyUnicode_DecodeUTF32Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 458 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 459 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 460 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 461 | return NULL; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 462 | return codec_tuple(decoded, consumed); |
| 463 | } |
| 464 | |
| 465 | static PyObject * |
| 466 | utf_32_le_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 467 | PyObject *args) |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 468 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 469 | Py_buffer pbuf; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 470 | const char *errors = NULL; |
| 471 | int byteorder = -1; |
| 472 | int final = 0; |
| 473 | Py_ssize_t consumed; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 474 | PyObject *decoded; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 475 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 476 | if (!PyArg_ParseTuple(args, "y*|zi:utf_32_le_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 477 | &pbuf, &errors, &final)) |
| 478 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 479 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 480 | decoded = PyUnicode_DecodeUTF32Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 481 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 482 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 483 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 484 | return NULL; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 485 | return codec_tuple(decoded, consumed); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 486 | } |
| 487 | |
| 488 | static PyObject * |
| 489 | utf_32_be_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 490 | PyObject *args) |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 491 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 492 | Py_buffer pbuf; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 493 | const char *errors = NULL; |
| 494 | int byteorder = 1; |
| 495 | int final = 0; |
| 496 | Py_ssize_t consumed; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 497 | PyObject *decoded; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 498 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 499 | if (!PyArg_ParseTuple(args, "y*|zi:utf_32_be_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 500 | &pbuf, &errors, &final)) |
| 501 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 502 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 503 | decoded = PyUnicode_DecodeUTF32Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 504 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 505 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 506 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 507 | return NULL; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 508 | return codec_tuple(decoded, consumed); |
| 509 | } |
| 510 | |
| 511 | /* This non-standard version also provides access to the byteorder |
| 512 | parameter of the builtin UTF-32 codec. |
| 513 | |
| 514 | It returns a tuple (unicode, bytesread, byteorder) with byteorder |
| 515 | being the value in effect at the end of data. |
| 516 | |
| 517 | */ |
| 518 | |
| 519 | static PyObject * |
| 520 | utf_32_ex_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 521 | PyObject *args) |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 522 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 523 | Py_buffer pbuf; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 524 | const char *errors = NULL; |
| 525 | int byteorder = 0; |
| 526 | PyObject *unicode, *tuple; |
| 527 | int final = 0; |
| 528 | Py_ssize_t consumed; |
| 529 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 530 | if (!PyArg_ParseTuple(args, "y*|zii:utf_32_ex_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 531 | &pbuf, &errors, &byteorder, &final)) |
| 532 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 533 | consumed = pbuf.len; /* This is overwritten unless final is true. */ |
| 534 | unicode = PyUnicode_DecodeUTF32Stateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 535 | &byteorder, final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 536 | PyBuffer_Release(&pbuf); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 537 | if (unicode == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 538 | return NULL; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 539 | tuple = Py_BuildValue("Oni", unicode, consumed, byteorder); |
| 540 | Py_DECREF(unicode); |
| 541 | return tuple; |
| 542 | } |
| 543 | |
| 544 | static PyObject * |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 545 | unicode_escape_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 546 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 547 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 548 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 549 | const char *errors = NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 550 | PyObject *unicode; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 551 | |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 552 | if (!PyArg_ParseTuple(args, "s*|z:unicode_escape_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 553 | &pbuf, &errors)) |
| 554 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 555 | |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 556 | unicode = PyUnicode_DecodeUnicodeEscape(pbuf.buf, pbuf.len, errors); |
| 557 | PyBuffer_Release(&pbuf); |
| 558 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 559 | } |
| 560 | |
| 561 | static PyObject * |
| 562 | raw_unicode_escape_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 563 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 564 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 565 | Py_buffer pbuf; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 566 | const char *errors = NULL; |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 567 | PyObject *unicode; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 568 | |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 569 | if (!PyArg_ParseTuple(args, "s*|z:raw_unicode_escape_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 570 | &pbuf, &errors)) |
| 571 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 572 | |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 573 | unicode = PyUnicode_DecodeRawUnicodeEscape(pbuf.buf, pbuf.len, errors); |
| 574 | PyBuffer_Release(&pbuf); |
| 575 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 576 | } |
| 577 | |
| 578 | static PyObject * |
| 579 | latin_1_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 580 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 581 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 582 | Py_buffer pbuf; |
| 583 | PyObject *unicode; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 584 | const char *errors = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 585 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 586 | if (!PyArg_ParseTuple(args, "y*|z:latin_1_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 587 | &pbuf, &errors)) |
| 588 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 589 | |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 590 | unicode = PyUnicode_DecodeLatin1(pbuf.buf, pbuf.len, errors); |
| 591 | PyBuffer_Release(&pbuf); |
| 592 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 593 | } |
| 594 | |
| 595 | static PyObject * |
| 596 | ascii_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 597 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 598 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 599 | Py_buffer pbuf; |
| 600 | PyObject *unicode; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 601 | const char *errors = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 602 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 603 | if (!PyArg_ParseTuple(args, "y*|z:ascii_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 604 | &pbuf, &errors)) |
| 605 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 606 | |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 607 | unicode = PyUnicode_DecodeASCII(pbuf.buf, pbuf.len, errors); |
| 608 | PyBuffer_Release(&pbuf); |
| 609 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 610 | } |
| 611 | |
| 612 | static PyObject * |
| 613 | charmap_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 614 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 615 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 616 | Py_buffer pbuf; |
| 617 | PyObject *unicode; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 618 | const char *errors = NULL; |
| 619 | PyObject *mapping = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 620 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 621 | if (!PyArg_ParseTuple(args, "y*|zO:charmap_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 622 | &pbuf, &errors, &mapping)) |
| 623 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 624 | if (mapping == Py_None) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 625 | mapping = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 626 | |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 627 | unicode = PyUnicode_DecodeCharmap(pbuf.buf, pbuf.len, mapping, errors); |
| 628 | PyBuffer_Release(&pbuf); |
| 629 | return codec_tuple(unicode, pbuf.len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 630 | } |
| 631 | |
Victor Stinner | 99b9538 | 2011-07-04 14:23:54 +0200 | [diff] [blame] | 632 | #ifdef HAVE_MBCS |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 633 | |
| 634 | static PyObject * |
| 635 | mbcs_decode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 636 | PyObject *args) |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 637 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 638 | Py_buffer pbuf; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 639 | const char *errors = NULL; |
Thomas Wouters | 0e3f591 | 2006-08-11 14:57:12 +0000 | [diff] [blame] | 640 | int final = 0; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 641 | Py_ssize_t consumed; |
| 642 | PyObject *decoded = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 643 | |
Antoine Pitrou | 81fabdb | 2009-01-22 10:11:36 +0000 | [diff] [blame] | 644 | if (!PyArg_ParseTuple(args, "y*|zi:mbcs_decode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 645 | &pbuf, &errors, &final)) |
| 646 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 647 | consumed = pbuf.len; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 648 | |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 649 | decoded = PyUnicode_DecodeMBCSStateful(pbuf.buf, pbuf.len, errors, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 650 | final ? NULL : &consumed); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 651 | PyBuffer_Release(&pbuf); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 652 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 653 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 654 | return codec_tuple(decoded, consumed); |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 655 | } |
| 656 | |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 657 | static PyObject * |
| 658 | code_page_decode(PyObject *self, |
| 659 | PyObject *args) |
| 660 | { |
| 661 | Py_buffer pbuf; |
| 662 | const char *errors = NULL; |
| 663 | int final = 0; |
| 664 | Py_ssize_t consumed; |
| 665 | PyObject *decoded = NULL; |
| 666 | int code_page; |
| 667 | |
| 668 | if (!PyArg_ParseTuple(args, "iy*|zi:code_page_decode", |
| 669 | &code_page, &pbuf, &errors, &final)) |
| 670 | return NULL; |
| 671 | consumed = pbuf.len; |
| 672 | |
| 673 | decoded = PyUnicode_DecodeCodePageStateful(code_page, |
| 674 | pbuf.buf, pbuf.len, errors, |
| 675 | final ? NULL : &consumed); |
| 676 | PyBuffer_Release(&pbuf); |
| 677 | if (decoded == NULL) |
| 678 | return NULL; |
| 679 | return codec_tuple(decoded, consumed); |
| 680 | } |
| 681 | |
Victor Stinner | 99b9538 | 2011-07-04 14:23:54 +0200 | [diff] [blame] | 682 | #endif /* HAVE_MBCS */ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 683 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 684 | /* --- Encoder ------------------------------------------------------------ */ |
| 685 | |
| 686 | static PyObject * |
| 687 | readbuffer_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 688 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 689 | { |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 690 | Py_buffer pdata; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 691 | const char *data; |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 692 | Py_ssize_t size; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 693 | const char *errors = NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 694 | PyObject *result; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 695 | |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 696 | if (!PyArg_ParseTuple(args, "s*|z:readbuffer_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 697 | &pdata, &errors)) |
| 698 | return NULL; |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 699 | data = pdata.buf; |
| 700 | size = pdata.len; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 701 | |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 702 | result = PyBytes_FromStringAndSize(data, size); |
| 703 | PyBuffer_Release(&pdata); |
| 704 | return codec_tuple(result, size); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 705 | } |
| 706 | |
| 707 | static PyObject * |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 708 | unicode_internal_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 709 | PyObject *args) |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 710 | { |
| 711 | PyObject *obj; |
| 712 | const char *errors = NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 713 | |
Ezio Melotti | adc417c | 2011-11-17 12:23:34 +0200 | [diff] [blame] | 714 | if (PyErr_WarnEx(PyExc_DeprecationWarning, |
| 715 | "unicode_internal codec has been deprecated", |
| 716 | 1)) |
| 717 | return NULL; |
| 718 | |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 719 | if (!PyArg_ParseTuple(args, "O|z:unicode_internal_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 720 | &obj, &errors)) |
| 721 | return NULL; |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 722 | |
| 723 | if (PyUnicode_Check(obj)) { |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 724 | Py_UNICODE *u; |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 725 | Py_ssize_t len, size; |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 726 | |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 727 | if (PyUnicode_READY(obj) < 0) |
| 728 | return NULL; |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 729 | |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 730 | u = PyUnicode_AsUnicodeAndSize(obj, &len); |
| 731 | if (u == NULL) |
| 732 | return NULL; |
Victor Stinner | 049e509 | 2014-08-17 22:20:00 +0200 | [diff] [blame] | 733 | if ((size_t)len > (size_t)PY_SSIZE_T_MAX / sizeof(Py_UNICODE)) |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 734 | return PyErr_NoMemory(); |
| 735 | size = len * sizeof(Py_UNICODE); |
| 736 | return codec_tuple(PyBytes_FromStringAndSize((const char*)u, size), |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 737 | PyUnicode_GET_LENGTH(obj)); |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 738 | } |
| 739 | else { |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 740 | Py_buffer view; |
| 741 | PyObject *result; |
| 742 | if (PyObject_GetBuffer(obj, &view, PyBUF_SIMPLE) != 0) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 743 | return NULL; |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 744 | result = codec_tuple(PyBytes_FromStringAndSize(view.buf, view.len), view.len); |
| 745 | PyBuffer_Release(&view); |
| 746 | return result; |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 747 | } |
| 748 | } |
| 749 | |
| 750 | static PyObject * |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 751 | utf_7_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 752 | PyObject *args) |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 753 | { |
| 754 | PyObject *str, *v; |
| 755 | const char *errors = NULL; |
| 756 | |
| 757 | if (!PyArg_ParseTuple(args, "O|z:utf_7_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 758 | &str, &errors)) |
| 759 | return NULL; |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 760 | |
| 761 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 762 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 763 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 764 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 765 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 766 | v = codec_tuple(_PyUnicode_EncodeUTF7(str, 0, 0, errors), |
| 767 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 768 | Py_DECREF(str); |
| 769 | return v; |
| 770 | } |
| 771 | |
| 772 | static PyObject * |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 773 | utf_8_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 774 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 775 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 776 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 777 | const char *errors = NULL; |
| 778 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 779 | if (!PyArg_ParseTuple(args, "O|z:utf_8_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 780 | &str, &errors)) |
| 781 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 782 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 783 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 784 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 785 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 786 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 787 | } |
Martin v. Löwis | d63a3b8 | 2011-09-28 07:41:54 +0200 | [diff] [blame] | 788 | v = codec_tuple(PyUnicode_AsEncodedString(str, "utf-8", errors), |
| 789 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 790 | Py_DECREF(str); |
| 791 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 792 | } |
| 793 | |
| 794 | /* This version provides access to the byteorder parameter of the |
| 795 | builtin UTF-16 codecs as optional third argument. It defaults to 0 |
| 796 | 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] | 797 | BOM mark. |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 798 | |
| 799 | */ |
| 800 | |
| 801 | static PyObject * |
| 802 | utf_16_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 803 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 804 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 805 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 806 | const char *errors = NULL; |
| 807 | int byteorder = 0; |
| 808 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 809 | if (!PyArg_ParseTuple(args, "O|zi:utf_16_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 810 | &str, &errors, &byteorder)) |
| 811 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 812 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 813 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 814 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 815 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 816 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 817 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 818 | v = codec_tuple(_PyUnicode_EncodeUTF16(str, errors, byteorder), |
| 819 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 820 | Py_DECREF(str); |
| 821 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 822 | } |
| 823 | |
| 824 | static PyObject * |
| 825 | utf_16_le_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 826 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 827 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 828 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 829 | const char *errors = NULL; |
| 830 | |
Marc-André Lemburg | 4157dd5 | 2001-06-17 18:32:36 +0000 | [diff] [blame] | 831 | if (!PyArg_ParseTuple(args, "O|z:utf_16_le_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 832 | &str, &errors)) |
| 833 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 834 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 835 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 836 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 837 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 838 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 839 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 840 | v = codec_tuple(_PyUnicode_EncodeUTF16(str, errors, -1), |
| 841 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 842 | Py_DECREF(str); |
| 843 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 844 | } |
| 845 | |
| 846 | static PyObject * |
| 847 | utf_16_be_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 848 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 849 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 850 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 851 | const char *errors = NULL; |
| 852 | |
Marc-André Lemburg | 4157dd5 | 2001-06-17 18:32:36 +0000 | [diff] [blame] | 853 | if (!PyArg_ParseTuple(args, "O|z:utf_16_be_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 854 | &str, &errors)) |
| 855 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 856 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 857 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 858 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 859 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 860 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 861 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 862 | v = codec_tuple(_PyUnicode_EncodeUTF16(str, errors, +1), |
| 863 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 864 | Py_DECREF(str); |
| 865 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 866 | } |
| 867 | |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 868 | /* This version provides access to the byteorder parameter of the |
| 869 | builtin UTF-32 codecs as optional third argument. It defaults to 0 |
| 870 | which means: use the native byte order and prepend the data with a |
| 871 | BOM mark. |
| 872 | |
| 873 | */ |
| 874 | |
| 875 | static PyObject * |
| 876 | utf_32_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 877 | PyObject *args) |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 878 | { |
| 879 | PyObject *str, *v; |
| 880 | const char *errors = NULL; |
| 881 | int byteorder = 0; |
| 882 | |
| 883 | if (!PyArg_ParseTuple(args, "O|zi:utf_32_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 884 | &str, &errors, &byteorder)) |
| 885 | return NULL; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 886 | |
| 887 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 888 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 889 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 890 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 891 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 892 | v = codec_tuple(_PyUnicode_EncodeUTF32(str, errors, byteorder), |
| 893 | PyUnicode_GET_LENGTH(str)); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 894 | Py_DECREF(str); |
| 895 | return v; |
| 896 | } |
| 897 | |
| 898 | static PyObject * |
| 899 | utf_32_le_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 900 | PyObject *args) |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 901 | { |
| 902 | PyObject *str, *v; |
| 903 | const char *errors = NULL; |
| 904 | |
| 905 | if (!PyArg_ParseTuple(args, "O|z:utf_32_le_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 906 | &str, &errors)) |
| 907 | return NULL; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 908 | |
| 909 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 910 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 911 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 912 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 913 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 914 | v = codec_tuple(_PyUnicode_EncodeUTF32(str, errors, -1), |
| 915 | PyUnicode_GET_LENGTH(str)); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 916 | Py_DECREF(str); |
| 917 | return v; |
| 918 | } |
| 919 | |
| 920 | static PyObject * |
| 921 | utf_32_be_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 922 | PyObject *args) |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 923 | { |
| 924 | PyObject *str, *v; |
| 925 | const char *errors = NULL; |
| 926 | |
| 927 | if (!PyArg_ParseTuple(args, "O|z:utf_32_be_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 928 | &str, &errors)) |
| 929 | return NULL; |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 930 | |
| 931 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 932 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 933 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 934 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 935 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 936 | v = codec_tuple(_PyUnicode_EncodeUTF32(str, errors, +1), |
| 937 | PyUnicode_GET_LENGTH(str)); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 938 | Py_DECREF(str); |
| 939 | return v; |
| 940 | } |
| 941 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 942 | static PyObject * |
| 943 | unicode_escape_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 944 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 945 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 946 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 947 | const char *errors = NULL; |
| 948 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 949 | if (!PyArg_ParseTuple(args, "O|z:unicode_escape_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 950 | &str, &errors)) |
| 951 | return NULL; |
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 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 954 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 955 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 956 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 957 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 958 | v = codec_tuple(PyUnicode_AsUnicodeEscapeString(str), |
| 959 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 960 | Py_DECREF(str); |
| 961 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 962 | } |
| 963 | |
| 964 | static PyObject * |
| 965 | raw_unicode_escape_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 966 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 967 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 968 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 969 | const char *errors = NULL; |
| 970 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 971 | if (!PyArg_ParseTuple(args, "O|z:raw_unicode_escape_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 972 | &str, &errors)) |
| 973 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 974 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 975 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 976 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 977 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 978 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 979 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 980 | v = codec_tuple(PyUnicode_AsRawUnicodeEscapeString(str), |
| 981 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 982 | Py_DECREF(str); |
| 983 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 984 | } |
| 985 | |
| 986 | static PyObject * |
| 987 | latin_1_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 988 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 989 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 990 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 991 | const char *errors = NULL; |
| 992 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 993 | if (!PyArg_ParseTuple(args, "O|z:latin_1_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 994 | &str, &errors)) |
| 995 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 996 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 997 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 998 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 999 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1000 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 1001 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 1002 | v = codec_tuple(_PyUnicode_AsLatin1String(str, errors), |
| 1003 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1004 | Py_DECREF(str); |
| 1005 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1006 | } |
| 1007 | |
| 1008 | static PyObject * |
| 1009 | ascii_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1010 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1011 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1012 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1013 | const char *errors = NULL; |
| 1014 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1015 | if (!PyArg_ParseTuple(args, "O|z:ascii_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1016 | &str, &errors)) |
| 1017 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1018 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1019 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 1020 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 1021 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1022 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 1023 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 1024 | v = codec_tuple(_PyUnicode_AsASCIIString(str, errors), |
| 1025 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1026 | Py_DECREF(str); |
| 1027 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1028 | } |
| 1029 | |
| 1030 | static PyObject * |
| 1031 | charmap_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1032 | PyObject *args) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1033 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1034 | PyObject *str, *v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1035 | const char *errors = NULL; |
| 1036 | PyObject *mapping = NULL; |
| 1037 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1038 | if (!PyArg_ParseTuple(args, "O|zO:charmap_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1039 | &str, &errors, &mapping)) |
| 1040 | return NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1041 | if (mapping == Py_None) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1042 | mapping = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1043 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1044 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 1045 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 1046 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1047 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 1048 | } |
Martin v. Löwis | 23e275b | 2011-11-02 18:02:51 +0100 | [diff] [blame] | 1049 | v = codec_tuple(_PyUnicode_EncodeCharmap(str, mapping, errors), |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 1050 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1051 | Py_DECREF(str); |
| 1052 | return v; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1053 | } |
| 1054 | |
Thomas Wouters | 73e5a5b | 2006-06-08 15:35:45 +0000 | [diff] [blame] | 1055 | static PyObject* |
| 1056 | charmap_build(PyObject *self, PyObject *args) |
| 1057 | { |
| 1058 | PyObject *map; |
| 1059 | if (!PyArg_ParseTuple(args, "U:charmap_build", &map)) |
| 1060 | return NULL; |
| 1061 | return PyUnicode_BuildEncodingMap(map); |
| 1062 | } |
| 1063 | |
Victor Stinner | 99b9538 | 2011-07-04 14:23:54 +0200 | [diff] [blame] | 1064 | #ifdef HAVE_MBCS |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1065 | |
| 1066 | static PyObject * |
| 1067 | mbcs_encode(PyObject *self, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1068 | PyObject *args) |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1069 | { |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1070 | PyObject *str, *v; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1071 | const char *errors = NULL; |
| 1072 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1073 | if (!PyArg_ParseTuple(args, "O|z:mbcs_encode", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1074 | &str, &errors)) |
| 1075 | return NULL; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1076 | |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1077 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 1078 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 1079 | Py_XDECREF(str); |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1080 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 1081 | } |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 1082 | v = codec_tuple(PyUnicode_EncodeCodePage(CP_ACP, str, errors), |
| 1083 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 1084 | Py_DECREF(str); |
| 1085 | return v; |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1086 | } |
| 1087 | |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 1088 | static PyObject * |
| 1089 | code_page_encode(PyObject *self, |
| 1090 | PyObject *args) |
| 1091 | { |
| 1092 | PyObject *str, *v; |
| 1093 | const char *errors = NULL; |
| 1094 | int code_page; |
| 1095 | |
| 1096 | if (!PyArg_ParseTuple(args, "iO|z:code_page_encode", |
| 1097 | &code_page, &str, &errors)) |
| 1098 | return NULL; |
| 1099 | |
| 1100 | str = PyUnicode_FromObject(str); |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 1101 | if (str == NULL || PyUnicode_READY(str) < 0) { |
| 1102 | Py_XDECREF(str); |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 1103 | return NULL; |
Antoine Pitrou | f72d4ef | 2011-11-12 18:05:15 +0100 | [diff] [blame] | 1104 | } |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 1105 | v = codec_tuple(PyUnicode_EncodeCodePage(code_page, |
| 1106 | str, |
| 1107 | errors), |
| 1108 | PyUnicode_GET_LENGTH(str)); |
| 1109 | Py_DECREF(str); |
| 1110 | return v; |
| 1111 | } |
| 1112 | |
Victor Stinner | 99b9538 | 2011-07-04 14:23:54 +0200 | [diff] [blame] | 1113 | #endif /* HAVE_MBCS */ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1114 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1115 | /* --- Error handler registry --------------------------------------------- */ |
| 1116 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1117 | PyDoc_STRVAR(register_error__doc__, |
| 1118 | "register_error(errors, handler)\n\ |
| 1119 | \n\ |
| 1120 | Register the specified error handler under the name\n\ |
| 1121 | errors. handler must be a callable object, that\n\ |
| 1122 | will be called with an exception instance containing\n\ |
| 1123 | information about the location of the encoding/decoding\n\ |
| 1124 | error and must return a (replacement, new position) tuple."); |
| 1125 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1126 | static PyObject *register_error(PyObject *self, PyObject *args) |
| 1127 | { |
| 1128 | const char *name; |
| 1129 | PyObject *handler; |
| 1130 | |
| 1131 | if (!PyArg_ParseTuple(args, "sO:register_error", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1132 | &name, &handler)) |
| 1133 | return NULL; |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1134 | if (PyCodec_RegisterError(name, handler)) |
| 1135 | return NULL; |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1136 | Py_RETURN_NONE; |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1137 | } |
| 1138 | |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1139 | PyDoc_STRVAR(lookup_error__doc__, |
| 1140 | "lookup_error(errors) -> handler\n\ |
| 1141 | \n\ |
| 1142 | Return the error handler for the specified error handling name\n\ |
| 1143 | or raise a LookupError, if no handler exists under this name."); |
| 1144 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1145 | static PyObject *lookup_error(PyObject *self, PyObject *args) |
| 1146 | { |
| 1147 | const char *name; |
| 1148 | |
| 1149 | if (!PyArg_ParseTuple(args, "s:lookup_error", |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1150 | &name)) |
| 1151 | return NULL; |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1152 | return PyCodec_LookupError(name); |
| 1153 | } |
| 1154 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1155 | /* --- Module API --------------------------------------------------------- */ |
| 1156 | |
| 1157 | static PyMethodDef _codecs_functions[] = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1158 | {"register", codec_register, METH_O, |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1159 | register__doc__}, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1160 | {"lookup", codec_lookup, METH_VARARGS, |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1161 | lookup__doc__}, |
Victor Stinner | a57dfd0 | 2014-05-14 17:13:14 +0200 | [diff] [blame] | 1162 | {"encode", (PyCFunction)codec_encode, METH_VARARGS|METH_KEYWORDS, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1163 | encode__doc__}, |
Victor Stinner | a57dfd0 | 2014-05-14 17:13:14 +0200 | [diff] [blame] | 1164 | {"decode", (PyCFunction)codec_decode, METH_VARARGS|METH_KEYWORDS, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1165 | decode__doc__}, |
| 1166 | {"escape_encode", escape_encode, METH_VARARGS}, |
| 1167 | {"escape_decode", escape_decode, METH_VARARGS}, |
| 1168 | {"utf_8_encode", utf_8_encode, METH_VARARGS}, |
| 1169 | {"utf_8_decode", utf_8_decode, METH_VARARGS}, |
| 1170 | {"utf_7_encode", utf_7_encode, METH_VARARGS}, |
| 1171 | {"utf_7_decode", utf_7_decode, METH_VARARGS}, |
| 1172 | {"utf_16_encode", utf_16_encode, METH_VARARGS}, |
| 1173 | {"utf_16_le_encode", utf_16_le_encode, METH_VARARGS}, |
| 1174 | {"utf_16_be_encode", utf_16_be_encode, METH_VARARGS}, |
| 1175 | {"utf_16_decode", utf_16_decode, METH_VARARGS}, |
| 1176 | {"utf_16_le_decode", utf_16_le_decode, METH_VARARGS}, |
| 1177 | {"utf_16_be_decode", utf_16_be_decode, METH_VARARGS}, |
| 1178 | {"utf_16_ex_decode", utf_16_ex_decode, METH_VARARGS}, |
| 1179 | {"utf_32_encode", utf_32_encode, METH_VARARGS}, |
| 1180 | {"utf_32_le_encode", utf_32_le_encode, METH_VARARGS}, |
| 1181 | {"utf_32_be_encode", utf_32_be_encode, METH_VARARGS}, |
| 1182 | {"utf_32_decode", utf_32_decode, METH_VARARGS}, |
| 1183 | {"utf_32_le_decode", utf_32_le_decode, METH_VARARGS}, |
| 1184 | {"utf_32_be_decode", utf_32_be_decode, METH_VARARGS}, |
| 1185 | {"utf_32_ex_decode", utf_32_ex_decode, METH_VARARGS}, |
| 1186 | {"unicode_escape_encode", unicode_escape_encode, METH_VARARGS}, |
| 1187 | {"unicode_escape_decode", unicode_escape_decode, METH_VARARGS}, |
| 1188 | {"unicode_internal_encode", unicode_internal_encode, METH_VARARGS}, |
| 1189 | {"unicode_internal_decode", unicode_internal_decode, METH_VARARGS}, |
| 1190 | {"raw_unicode_escape_encode", raw_unicode_escape_encode, METH_VARARGS}, |
| 1191 | {"raw_unicode_escape_decode", raw_unicode_escape_decode, METH_VARARGS}, |
| 1192 | {"latin_1_encode", latin_1_encode, METH_VARARGS}, |
| 1193 | {"latin_1_decode", latin_1_decode, METH_VARARGS}, |
| 1194 | {"ascii_encode", ascii_encode, METH_VARARGS}, |
| 1195 | {"ascii_decode", ascii_decode, METH_VARARGS}, |
| 1196 | {"charmap_encode", charmap_encode, METH_VARARGS}, |
| 1197 | {"charmap_decode", charmap_decode, METH_VARARGS}, |
| 1198 | {"charmap_build", charmap_build, METH_VARARGS}, |
| 1199 | {"readbuffer_encode", readbuffer_encode, METH_VARARGS}, |
Victor Stinner | 99b9538 | 2011-07-04 14:23:54 +0200 | [diff] [blame] | 1200 | #ifdef HAVE_MBCS |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1201 | {"mbcs_encode", mbcs_encode, METH_VARARGS}, |
| 1202 | {"mbcs_decode", mbcs_decode, METH_VARARGS}, |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 1203 | {"code_page_encode", code_page_encode, METH_VARARGS}, |
| 1204 | {"code_page_decode", code_page_decode, METH_VARARGS}, |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1205 | #endif |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1206 | {"register_error", register_error, METH_VARARGS, |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1207 | register_error__doc__}, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1208 | {"lookup_error", lookup_error, METH_VARARGS, |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1209 | lookup_error__doc__}, |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 1210 | _CODECS__FORGET_CODEC_METHODDEF |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1211 | {NULL, NULL} /* sentinel */ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1212 | }; |
| 1213 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1214 | static struct PyModuleDef codecsmodule = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1215 | PyModuleDef_HEAD_INIT, |
| 1216 | "_codecs", |
| 1217 | NULL, |
| 1218 | -1, |
| 1219 | _codecs_functions, |
| 1220 | NULL, |
| 1221 | NULL, |
| 1222 | NULL, |
| 1223 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1224 | }; |
| 1225 | |
Mark Hammond | fe51c6d | 2002-08-02 02:27:13 +0000 | [diff] [blame] | 1226 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1227 | PyInit__codecs(void) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1228 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1229 | return PyModule_Create(&codecsmodule); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1230 | } |