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 | |
| 23 | These <encoding>s are available: utf_8, unicode_escape, |
Marc-André Lemburg | 5f0e29e | 2000-07-05 11:24:13 +0000 | [diff] [blame] | 24 | raw_unicode_escape, unicode_internal, latin_1, ascii (7-bit), |
| 25 | mbcs (on win32). |
| 26 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 27 | |
| 28 | Written by Marc-Andre Lemburg (mal@lemburg.com). |
| 29 | |
Guido van Rossum | 16b1ad9 | 2000-08-03 16:24:25 +0000 | [diff] [blame] | 30 | Copyright (c) Corporation for National Research Initiatives. |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 31 | |
| 32 | ------------------------------------------------------------------------ */ |
| 33 | |
Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 34 | #define PY_SSIZE_T_CLEAN |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 35 | #include "Python.h" |
| 36 | |
Victor Stinner | 87a7c82 | 2011-11-10 20:05:55 +0100 | [diff] [blame] | 37 | #ifdef MS_WINDOWS |
| 38 | #include <windows.h> |
| 39 | #endif |
| 40 | |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 41 | /*[clinic input] |
| 42 | module _codecs |
| 43 | [clinic start generated code]*/ |
| 44 | /*[clinic end generated code: output=da39a3ee5e6b4b0d input=e1390e3da3cb9deb]*/ |
| 45 | |
Serhiy Storchaka | 1009bf1 | 2015-04-03 23:53:51 +0300 | [diff] [blame] | 46 | #include "clinic/_codecsmodule.c.h" |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 47 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 48 | /* --- Registry ----------------------------------------------------------- */ |
| 49 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 50 | /*[clinic input] |
| 51 | _codecs.register |
| 52 | search_function: object |
| 53 | / |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 54 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 55 | Register a codec search function. |
| 56 | |
| 57 | Search functions are expected to take one argument, the encoding name in |
| 58 | all lower case letters, and either return None, or a tuple of functions |
| 59 | (encoder, decoder, stream_reader, stream_writer) (or a CodecInfo object). |
| 60 | [clinic start generated code]*/ |
| 61 | |
| 62 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 63 | _codecs_register(PyObject *module, PyObject *search_function) |
| 64 | /*[clinic end generated code: output=d1bf21e99db7d6d3 input=369578467955cae4]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 65 | { |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 66 | if (PyCodec_Register(search_function)) |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 67 | return NULL; |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 68 | |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 69 | Py_RETURN_NONE; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 70 | } |
| 71 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 72 | /*[clinic input] |
| 73 | _codecs.lookup |
| 74 | encoding: str |
| 75 | / |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 76 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 77 | Looks up a codec tuple in the Python codec registry and returns a CodecInfo object. |
| 78 | [clinic start generated code]*/ |
| 79 | |
| 80 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 81 | _codecs_lookup_impl(PyObject *module, const char *encoding) |
| 82 | /*[clinic end generated code: output=9f0afa572080c36d input=3c572c0db3febe9c]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 83 | { |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 84 | return _PyCodec_Lookup(encoding); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 85 | } |
| 86 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 87 | /*[clinic input] |
| 88 | _codecs.encode |
| 89 | obj: object |
Serhiy Storchaka | c97a962 | 2015-08-09 12:23:08 +0300 | [diff] [blame] | 90 | encoding: str(c_default="NULL") = "utf-8" |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 91 | errors: str(c_default="NULL") = "strict" |
| 92 | |
| 93 | Encodes obj using the codec registered for encoding. |
| 94 | |
Serhiy Storchaka | c97a962 | 2015-08-09 12:23:08 +0300 | [diff] [blame] | 95 | The default encoding is 'utf-8'. errors may be given to set a |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 96 | different error handling scheme. Default is 'strict' meaning that encoding |
| 97 | errors raise a ValueError. Other possible values are 'ignore', 'replace' |
| 98 | and 'backslashreplace' as well as any other name registered with |
| 99 | codecs.register_error that can handle ValueErrors. |
| 100 | [clinic start generated code]*/ |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 101 | |
| 102 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 103 | _codecs_encode_impl(PyObject *module, PyObject *obj, const char *encoding, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 104 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 105 | /*[clinic end generated code: output=385148eb9a067c86 input=cd5b685040ff61f0]*/ |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 106 | { |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 107 | if (encoding == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 108 | encoding = PyUnicode_GetDefaultEncoding(); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 109 | |
| 110 | /* Encode via the codec registry */ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 111 | return PyCodec_Encode(obj, encoding, errors); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 112 | } |
| 113 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 114 | /*[clinic input] |
| 115 | _codecs.decode |
| 116 | obj: object |
Serhiy Storchaka | c97a962 | 2015-08-09 12:23:08 +0300 | [diff] [blame] | 117 | encoding: str(c_default="NULL") = "utf-8" |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 118 | errors: str(c_default="NULL") = "strict" |
| 119 | |
| 120 | Decodes obj using the codec registered for encoding. |
| 121 | |
Serhiy Storchaka | c97a962 | 2015-08-09 12:23:08 +0300 | [diff] [blame] | 122 | Default encoding is 'utf-8'. errors may be given to set a |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 123 | different error handling scheme. Default is 'strict' meaning that encoding |
| 124 | errors raise a ValueError. Other possible values are 'ignore', 'replace' |
| 125 | and 'backslashreplace' as well as any other name registered with |
| 126 | codecs.register_error that can handle ValueErrors. |
| 127 | [clinic start generated code]*/ |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 128 | |
| 129 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 130 | _codecs_decode_impl(PyObject *module, PyObject *obj, const char *encoding, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 131 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 132 | /*[clinic end generated code: output=679882417dc3a0bd input=7702c0cc2fa1add6]*/ |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 133 | { |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 134 | if (encoding == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 135 | encoding = PyUnicode_GetDefaultEncoding(); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 136 | |
| 137 | /* Decode via the codec registry */ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 138 | return PyCodec_Decode(obj, encoding, errors); |
Marc-André Lemburg | 3f41974 | 2004-07-10 12:06:10 +0000 | [diff] [blame] | 139 | } |
| 140 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 141 | /* --- Helpers ------------------------------------------------------------ */ |
| 142 | |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 143 | /*[clinic input] |
| 144 | _codecs._forget_codec |
| 145 | |
| 146 | encoding: str |
| 147 | / |
| 148 | |
| 149 | Purge the named codec from the internal codec lookup cache |
| 150 | [clinic start generated code]*/ |
| 151 | |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 152 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 153 | _codecs__forget_codec_impl(PyObject *module, const char *encoding) |
| 154 | /*[clinic end generated code: output=0bde9f0a5b084aa2 input=18d5d92d0e386c38]*/ |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 155 | { |
| 156 | if (_PyCodec_Forget(encoding) < 0) { |
| 157 | return NULL; |
| 158 | }; |
| 159 | Py_RETURN_NONE; |
| 160 | } |
| 161 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 162 | static |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 163 | PyObject *codec_tuple(PyObject *decoded, |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 164 | Py_ssize_t len) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 165 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 166 | if (decoded == NULL) |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 167 | return NULL; |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 168 | return Py_BuildValue("Nn", decoded, len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 169 | } |
| 170 | |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 171 | /* --- String codecs ------------------------------------------------------ */ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 172 | /*[clinic input] |
| 173 | _codecs.escape_decode |
| 174 | data: Py_buffer(accept={str, buffer}) |
| 175 | errors: str(accept={str, NoneType}) = NULL |
| 176 | / |
| 177 | [clinic start generated code]*/ |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 178 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 179 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 180 | _codecs_escape_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 181 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 182 | /*[clinic end generated code: output=505200ba8056979a input=0018edfd99db714d]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 183 | { |
| 184 | PyObject *decoded = PyBytes_DecodeEscape(data->buf, data->len, |
| 185 | errors, 0, NULL); |
| 186 | return codec_tuple(decoded, data->len); |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 187 | } |
| 188 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 189 | /*[clinic input] |
| 190 | _codecs.escape_encode |
| 191 | data: object(subclass_of='&PyBytes_Type') |
| 192 | errors: str(accept={str, NoneType}) = NULL |
| 193 | / |
| 194 | [clinic start generated code]*/ |
| 195 | |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 196 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 197 | _codecs_escape_encode_impl(PyObject *module, PyObject *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 198 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 199 | /*[clinic end generated code: output=4af1d477834bab34 input=da9ded00992f32f2]*/ |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 200 | { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 201 | Py_ssize_t size; |
| 202 | Py_ssize_t newsize; |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 203 | PyObject *v; |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 204 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 205 | size = PyBytes_GET_SIZE(data); |
Mark Dickinson | c04ddff | 2012-10-06 18:04:49 +0100 | [diff] [blame] | 206 | if (size > PY_SSIZE_T_MAX / 4) { |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 207 | PyErr_SetString(PyExc_OverflowError, |
| 208 | "string is too large to encode"); |
| 209 | return NULL; |
| 210 | } |
Mark Dickinson | c04ddff | 2012-10-06 18:04:49 +0100 | [diff] [blame] | 211 | newsize = 4*size; |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 212 | v = PyBytes_FromStringAndSize(NULL, newsize); |
| 213 | |
| 214 | if (v == NULL) { |
| 215 | return NULL; |
| 216 | } |
| 217 | else { |
Antoine Pitrou | 9ed5f27 | 2013-08-13 20:18:52 +0200 | [diff] [blame] | 218 | Py_ssize_t i; |
| 219 | char c; |
| 220 | char *p = PyBytes_AS_STRING(v); |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 221 | |
| 222 | for (i = 0; i < size; i++) { |
| 223 | /* There's at least enough room for a hex escape */ |
| 224 | assert(newsize - (p - PyBytes_AS_STRING(v)) >= 4); |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 225 | c = PyBytes_AS_STRING(data)[i]; |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 226 | if (c == '\'' || c == '\\') |
| 227 | *p++ = '\\', *p++ = c; |
| 228 | else if (c == '\t') |
| 229 | *p++ = '\\', *p++ = 't'; |
| 230 | else if (c == '\n') |
| 231 | *p++ = '\\', *p++ = 'n'; |
| 232 | else if (c == '\r') |
| 233 | *p++ = '\\', *p++ = 'r'; |
| 234 | else if (c < ' ' || c >= 0x7f) { |
| 235 | *p++ = '\\'; |
| 236 | *p++ = 'x'; |
Victor Stinner | f5cff56 | 2011-10-14 02:13:11 +0200 | [diff] [blame] | 237 | *p++ = Py_hexdigits[(c & 0xf0) >> 4]; |
| 238 | *p++ = Py_hexdigits[c & 0xf]; |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 239 | } |
| 240 | else |
| 241 | *p++ = c; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 242 | } |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 243 | *p = '\0'; |
| 244 | if (_PyBytes_Resize(&v, (p - PyBytes_AS_STRING(v)))) { |
| 245 | return NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 246 | } |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 247 | } |
Walter Dörwald | 1ab8330 | 2007-05-18 17:15:44 +0000 | [diff] [blame] | 248 | |
Antoine Pitrou | d118856 | 2010-06-09 16:38:55 +0000 | [diff] [blame] | 249 | return codec_tuple(v, size); |
Martin v. Löwis | 8a8da79 | 2002-08-14 07:46:28 +0000 | [diff] [blame] | 250 | } |
| 251 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 252 | /* --- Decoder ------------------------------------------------------------ */ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 253 | /*[clinic input] |
| 254 | _codecs.unicode_internal_decode |
| 255 | obj: object |
| 256 | errors: str(accept={str, NoneType}) = NULL |
| 257 | / |
| 258 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 259 | |
| 260 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 261 | _codecs_unicode_internal_decode_impl(PyObject *module, PyObject *obj, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 262 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 263 | /*[clinic end generated code: output=edbfe175e09eff9a input=8d57930aeda170c6]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 264 | { |
Marc-André Lemburg | 29273c8 | 2003-02-04 19:35:03 +0000 | [diff] [blame] | 265 | if (PyUnicode_Check(obj)) { |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 266 | if (PyUnicode_READY(obj) < 0) |
| 267 | return NULL; |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 268 | Py_INCREF(obj); |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 269 | return codec_tuple(obj, PyUnicode_GET_LENGTH(obj)); |
Marc-André Lemburg | 29273c8 | 2003-02-04 19:35:03 +0000 | [diff] [blame] | 270 | } |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 271 | else { |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 272 | Py_buffer view; |
| 273 | PyObject *result; |
| 274 | if (PyObject_GetBuffer(obj, &view, PyBUF_SIMPLE) != 0) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 275 | return NULL; |
Walter Dörwald | a47d1c0 | 2005-08-30 10:23:14 +0000 | [diff] [blame] | 276 | |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 277 | result = codec_tuple( |
| 278 | _PyUnicode_DecodeUnicodeInternal(view.buf, view.len, errors), |
| 279 | view.len); |
| 280 | PyBuffer_Release(&view); |
| 281 | return result; |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 282 | } |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 283 | } |
| 284 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 285 | /*[clinic input] |
| 286 | _codecs.utf_7_decode |
| 287 | data: Py_buffer |
| 288 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 289 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 290 | / |
| 291 | [clinic start generated code]*/ |
| 292 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 293 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 294 | _codecs_utf_7_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 295 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 296 | /*[clinic end generated code: output=0cd3a944a32a4089 input=2d94a5a1f170c8ae]*/ |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 297 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 298 | Py_ssize_t consumed = data->len; |
| 299 | PyObject *decoded = PyUnicode_DecodeUTF7Stateful(data->buf, data->len, |
| 300 | errors, |
| 301 | final ? NULL : &consumed); |
Christian Heimes | 5d14c2b | 2007-11-20 23:38:09 +0000 | [diff] [blame] | 302 | return codec_tuple(decoded, consumed); |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 303 | } |
| 304 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 305 | /*[clinic input] |
| 306 | _codecs.utf_8_decode |
| 307 | data: Py_buffer |
| 308 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 309 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 310 | / |
| 311 | [clinic start generated code]*/ |
| 312 | |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 313 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 314 | _codecs_utf_8_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 315 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 316 | /*[clinic end generated code: output=10f74dec8d9bb8bf input=1ea6c21492e8bcbe]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 317 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 318 | Py_ssize_t consumed = data->len; |
| 319 | PyObject *decoded = PyUnicode_DecodeUTF8Stateful(data->buf, data->len, |
| 320 | errors, |
| 321 | final ? NULL : &consumed); |
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 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 325 | /*[clinic input] |
| 326 | _codecs.utf_16_decode |
| 327 | data: Py_buffer |
| 328 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 329 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 330 | / |
| 331 | [clinic start generated code]*/ |
| 332 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 333 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 334 | _codecs_utf_16_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 335 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 336 | /*[clinic end generated code: output=783b442abcbcc2d0 input=2ba128c28ea0bb40]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 337 | { |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 338 | int byteorder = 0; |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 339 | /* This is overwritten unless final is true. */ |
| 340 | Py_ssize_t consumed = data->len; |
| 341 | PyObject *decoded = PyUnicode_DecodeUTF16Stateful(data->buf, data->len, |
| 342 | errors, &byteorder, |
| 343 | final ? NULL : &consumed); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 344 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 345 | } |
| 346 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 347 | /*[clinic input] |
| 348 | _codecs.utf_16_le_decode |
| 349 | data: Py_buffer |
| 350 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 351 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 352 | / |
| 353 | [clinic start generated code]*/ |
| 354 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 355 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 356 | _codecs_utf_16_le_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 357 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 358 | /*[clinic end generated code: output=899b9e6364379dcd input=43aeb8b0461cace5]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 359 | { |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 360 | int byteorder = -1; |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 361 | /* This is overwritten unless final is true. */ |
| 362 | Py_ssize_t consumed = data->len; |
| 363 | PyObject *decoded = PyUnicode_DecodeUTF16Stateful(data->buf, data->len, |
| 364 | errors, &byteorder, |
| 365 | final ? NULL : &consumed); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 366 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 367 | } |
| 368 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 369 | /*[clinic input] |
| 370 | _codecs.utf_16_be_decode |
| 371 | data: Py_buffer |
| 372 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 373 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 374 | / |
| 375 | [clinic start generated code]*/ |
| 376 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 377 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 378 | _codecs_utf_16_be_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 379 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 380 | /*[clinic end generated code: output=49f6465ea07669c8 input=339e554c804f34b2]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 381 | { |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 382 | int byteorder = 1; |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 383 | /* This is overwritten unless final is true. */ |
| 384 | Py_ssize_t consumed = data->len; |
| 385 | PyObject *decoded = PyUnicode_DecodeUTF16Stateful(data->buf, data->len, |
| 386 | errors, &byteorder, |
| 387 | final ? NULL : &consumed); |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 388 | return codec_tuple(decoded, consumed); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 389 | } |
| 390 | |
| 391 | /* This non-standard version also provides access to the byteorder |
| 392 | parameter of the builtin UTF-16 codec. |
| 393 | |
| 394 | It returns a tuple (unicode, bytesread, byteorder) with byteorder |
| 395 | being the value in effect at the end of data. |
| 396 | |
| 397 | */ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 398 | /*[clinic input] |
| 399 | _codecs.utf_16_ex_decode |
| 400 | data: Py_buffer |
| 401 | errors: str(accept={str, NoneType}) = NULL |
| 402 | byteorder: int = 0 |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 403 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 404 | / |
| 405 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 406 | |
| 407 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 408 | _codecs_utf_16_ex_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 409 | const char *errors, int byteorder, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 410 | /*[clinic end generated code: output=0f385f251ecc1988 input=3201aeddb9636889]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 411 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 412 | /* This is overwritten unless final is true. */ |
| 413 | Py_ssize_t consumed = data->len; |
Walter Dörwald | 6965203 | 2004-09-07 20:24:22 +0000 | [diff] [blame] | 414 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 415 | PyObject *decoded = PyUnicode_DecodeUTF16Stateful(data->buf, data->len, |
| 416 | errors, &byteorder, |
| 417 | final ? NULL : &consumed); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 418 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 419 | return NULL; |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 420 | return Py_BuildValue("Nni", decoded, consumed, byteorder); |
| 421 | } |
| 422 | |
| 423 | /*[clinic input] |
| 424 | _codecs.utf_32_decode |
| 425 | data: Py_buffer |
| 426 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 427 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 428 | / |
| 429 | [clinic start generated code]*/ |
| 430 | |
| 431 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 432 | _codecs_utf_32_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 433 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 434 | /*[clinic end generated code: output=2fc961807f7b145f input=155a5c673a4e2514]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 435 | { |
| 436 | int byteorder = 0; |
| 437 | /* This is overwritten unless final is true. */ |
| 438 | Py_ssize_t consumed = data->len; |
| 439 | PyObject *decoded = PyUnicode_DecodeUTF32Stateful(data->buf, data->len, |
| 440 | errors, &byteorder, |
| 441 | final ? NULL : &consumed); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 442 | return codec_tuple(decoded, consumed); |
| 443 | } |
| 444 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 445 | /*[clinic input] |
| 446 | _codecs.utf_32_le_decode |
| 447 | data: Py_buffer |
| 448 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 449 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 450 | / |
| 451 | [clinic start generated code]*/ |
| 452 | |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 453 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 454 | _codecs_utf_32_le_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 455 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 456 | /*[clinic end generated code: output=ec8f46b67a94f3e6 input=7baf061069e92d3b]*/ |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 457 | { |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 458 | int byteorder = -1; |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 459 | /* This is overwritten unless final is true. */ |
| 460 | Py_ssize_t consumed = data->len; |
| 461 | PyObject *decoded = PyUnicode_DecodeUTF32Stateful(data->buf, data->len, |
| 462 | errors, &byteorder, |
| 463 | final ? NULL : &consumed); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 464 | return codec_tuple(decoded, consumed); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 465 | } |
| 466 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 467 | /*[clinic input] |
| 468 | _codecs.utf_32_be_decode |
| 469 | data: Py_buffer |
| 470 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 471 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 472 | / |
| 473 | [clinic start generated code]*/ |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 474 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 475 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 476 | _codecs_utf_32_be_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 477 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 478 | /*[clinic end generated code: output=ff82bae862c92c4e input=b182026300dae595]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 479 | { |
| 480 | int byteorder = 1; |
| 481 | /* This is overwritten unless final is true. */ |
| 482 | Py_ssize_t consumed = data->len; |
| 483 | PyObject *decoded = PyUnicode_DecodeUTF32Stateful(data->buf, data->len, |
| 484 | errors, &byteorder, |
| 485 | final ? NULL : &consumed); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 486 | return codec_tuple(decoded, consumed); |
| 487 | } |
| 488 | |
| 489 | /* This non-standard version also provides access to the byteorder |
| 490 | parameter of the builtin UTF-32 codec. |
| 491 | |
| 492 | It returns a tuple (unicode, bytesread, byteorder) with byteorder |
| 493 | being the value in effect at the end of data. |
| 494 | |
| 495 | */ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 496 | /*[clinic input] |
| 497 | _codecs.utf_32_ex_decode |
| 498 | data: Py_buffer |
| 499 | errors: str(accept={str, NoneType}) = NULL |
| 500 | byteorder: int = 0 |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 501 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 502 | / |
| 503 | [clinic start generated code]*/ |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 504 | |
| 505 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 506 | _codecs_utf_32_ex_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 507 | const char *errors, int byteorder, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 508 | /*[clinic end generated code: output=6bfb177dceaf4848 input=7b9c2cb819fb237a]*/ |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 509 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 510 | Py_ssize_t consumed = data->len; |
| 511 | PyObject *decoded = PyUnicode_DecodeUTF32Stateful(data->buf, data->len, |
| 512 | errors, &byteorder, |
| 513 | final ? NULL : &consumed); |
| 514 | if (decoded == NULL) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 515 | return NULL; |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 516 | return Py_BuildValue("Nni", decoded, consumed, byteorder); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 517 | } |
| 518 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 519 | /*[clinic input] |
| 520 | _codecs.unicode_escape_decode |
| 521 | data: Py_buffer(accept={str, buffer}) |
| 522 | errors: str(accept={str, NoneType}) = NULL |
| 523 | / |
| 524 | [clinic start generated code]*/ |
| 525 | |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 526 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 527 | _codecs_unicode_escape_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 528 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 529 | /*[clinic end generated code: output=3ca3c917176b82ab input=49fd27d06813a7f5]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 530 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 531 | PyObject *decoded = PyUnicode_DecodeUnicodeEscape(data->buf, data->len, |
| 532 | errors); |
| 533 | return codec_tuple(decoded, data->len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 534 | } |
| 535 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 536 | /*[clinic input] |
| 537 | _codecs.raw_unicode_escape_decode |
| 538 | data: Py_buffer(accept={str, buffer}) |
| 539 | errors: str(accept={str, NoneType}) = NULL |
| 540 | / |
| 541 | [clinic start generated code]*/ |
| 542 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 543 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 544 | _codecs_raw_unicode_escape_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 545 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 546 | /*[clinic end generated code: output=c98eeb56028070a6 input=770903a211434ebc]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 547 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 548 | PyObject *decoded = PyUnicode_DecodeRawUnicodeEscape(data->buf, data->len, |
| 549 | errors); |
| 550 | return codec_tuple(decoded, data->len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 551 | } |
| 552 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 553 | /*[clinic input] |
| 554 | _codecs.latin_1_decode |
| 555 | data: Py_buffer |
| 556 | errors: str(accept={str, NoneType}) = NULL |
| 557 | / |
| 558 | [clinic start generated code]*/ |
| 559 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 560 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 561 | _codecs_latin_1_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 562 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 563 | /*[clinic end generated code: output=07f3dfa3f72c7d8f input=5cad0f1759c618ec]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 564 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 565 | PyObject *decoded = PyUnicode_DecodeLatin1(data->buf, data->len, errors); |
| 566 | return codec_tuple(decoded, data->len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 567 | } |
| 568 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 569 | /*[clinic input] |
| 570 | _codecs.ascii_decode |
| 571 | data: Py_buffer |
| 572 | errors: str(accept={str, NoneType}) = NULL |
| 573 | / |
| 574 | [clinic start generated code]*/ |
| 575 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 576 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 577 | _codecs_ascii_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 578 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 579 | /*[clinic end generated code: output=2627d72058d42429 input=ad1106f64037bd16]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 580 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 581 | PyObject *decoded = PyUnicode_DecodeASCII(data->buf, data->len, errors); |
| 582 | return codec_tuple(decoded, data->len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 583 | } |
| 584 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 585 | /*[clinic input] |
| 586 | _codecs.charmap_decode |
| 587 | data: Py_buffer |
| 588 | errors: str(accept={str, NoneType}) = NULL |
| 589 | mapping: object = NULL |
| 590 | / |
| 591 | [clinic start generated code]*/ |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 592 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 593 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 594 | _codecs_charmap_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 595 | const char *errors, PyObject *mapping) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 596 | /*[clinic end generated code: output=2c335b09778cf895 input=19712ca35c5a80e2]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 597 | { |
| 598 | PyObject *decoded; |
| 599 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 600 | if (mapping == Py_None) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 601 | mapping = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 602 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 603 | decoded = PyUnicode_DecodeCharmap(data->buf, data->len, mapping, errors); |
| 604 | return codec_tuple(decoded, data->len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 605 | } |
| 606 | |
Steve Dower | cc16be8 | 2016-09-08 10:35:16 -0700 | [diff] [blame] | 607 | #ifdef MS_WINDOWS |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 608 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 609 | /*[clinic input] |
| 610 | _codecs.mbcs_decode |
| 611 | data: Py_buffer |
| 612 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 613 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 614 | / |
| 615 | [clinic start generated code]*/ |
| 616 | |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 617 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 618 | _codecs_mbcs_decode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 619 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 620 | /*[clinic end generated code: output=39b65b8598938c4b input=b5f2fe568f311297]*/ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 621 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 622 | Py_ssize_t consumed = data->len; |
| 623 | PyObject *decoded = PyUnicode_DecodeMBCSStateful(data->buf, data->len, |
| 624 | errors, final ? NULL : &consumed); |
Martin v. Löwis | 423be95 | 2008-08-13 15:53:07 +0000 | [diff] [blame] | 625 | return codec_tuple(decoded, consumed); |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 626 | } |
| 627 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 628 | /*[clinic input] |
Steve Dower | f5aba58 | 2016-09-06 19:42:27 -0700 | [diff] [blame] | 629 | _codecs.oem_decode |
| 630 | data: Py_buffer |
| 631 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 632 | final: bool(accept={int}) = False |
Steve Dower | f5aba58 | 2016-09-06 19:42:27 -0700 | [diff] [blame] | 633 | / |
| 634 | [clinic start generated code]*/ |
| 635 | |
| 636 | static PyObject * |
| 637 | _codecs_oem_decode_impl(PyObject *module, Py_buffer *data, |
| 638 | const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 639 | /*[clinic end generated code: output=da1617612f3fcad8 input=278709bcfd374a9c]*/ |
Steve Dower | f5aba58 | 2016-09-06 19:42:27 -0700 | [diff] [blame] | 640 | { |
| 641 | Py_ssize_t consumed = data->len; |
| 642 | PyObject *decoded = PyUnicode_DecodeCodePageStateful(CP_OEMCP, |
| 643 | data->buf, data->len, errors, final ? NULL : &consumed); |
| 644 | return codec_tuple(decoded, consumed); |
| 645 | } |
| 646 | |
| 647 | /*[clinic input] |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 648 | _codecs.code_page_decode |
| 649 | codepage: int |
| 650 | data: Py_buffer |
| 651 | errors: str(accept={str, NoneType}) = NULL |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 652 | final: bool(accept={int}) = False |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 653 | / |
| 654 | [clinic start generated code]*/ |
| 655 | |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 656 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 657 | _codecs_code_page_decode_impl(PyObject *module, int codepage, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 658 | Py_buffer *data, const char *errors, int final) |
Serhiy Storchaka | 202fda5 | 2017-03-12 10:10:47 +0200 | [diff] [blame] | 659 | /*[clinic end generated code: output=53008ea967da3fff input=51f6169021c68dd5]*/ |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 660 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 661 | Py_ssize_t consumed = data->len; |
Serhiy Storchaka | 1aa5e1d | 2015-05-12 14:00:22 +0300 | [diff] [blame] | 662 | PyObject *decoded = PyUnicode_DecodeCodePageStateful(codepage, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 663 | data->buf, data->len, |
| 664 | errors, |
| 665 | final ? NULL : &consumed); |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 666 | return codec_tuple(decoded, consumed); |
| 667 | } |
| 668 | |
Steve Dower | cc16be8 | 2016-09-08 10:35:16 -0700 | [diff] [blame] | 669 | #endif /* MS_WINDOWS */ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 670 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 671 | /* --- Encoder ------------------------------------------------------------ */ |
| 672 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 673 | /*[clinic input] |
| 674 | _codecs.readbuffer_encode |
| 675 | data: Py_buffer(accept={str, buffer}) |
| 676 | errors: str(accept={str, NoneType}) = NULL |
| 677 | / |
| 678 | [clinic start generated code]*/ |
| 679 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 680 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 681 | _codecs_readbuffer_encode_impl(PyObject *module, Py_buffer *data, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 682 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 683 | /*[clinic end generated code: output=c645ea7cdb3d6e86 input=b7c322b89d4ab923]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 684 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 685 | PyObject *result = PyBytes_FromStringAndSize(data->buf, data->len); |
| 686 | return codec_tuple(result, data->len); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 687 | } |
| 688 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 689 | /*[clinic input] |
| 690 | _codecs.unicode_internal_encode |
| 691 | obj: object |
| 692 | errors: str(accept={str, NoneType}) = NULL |
| 693 | / |
| 694 | [clinic start generated code]*/ |
Walter Dörwald | 9fd115c | 2005-11-02 08:30:08 +0000 | [diff] [blame] | 695 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 696 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 697 | _codecs_unicode_internal_encode_impl(PyObject *module, PyObject *obj, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 698 | const char *errors) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 699 | /*[clinic end generated code: output=a72507dde4ea558f input=8628f0280cf5ba61]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 700 | { |
Ezio Melotti | adc417c | 2011-11-17 12:23:34 +0200 | [diff] [blame] | 701 | if (PyErr_WarnEx(PyExc_DeprecationWarning, |
| 702 | "unicode_internal codec has been deprecated", |
| 703 | 1)) |
| 704 | return NULL; |
| 705 | |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 706 | if (PyUnicode_Check(obj)) { |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 707 | Py_UNICODE *u; |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 708 | Py_ssize_t len, size; |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 709 | |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 710 | if (PyUnicode_READY(obj) < 0) |
| 711 | return NULL; |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 712 | |
Victor Stinner | 9f4b1e9 | 2011-11-10 20:56:30 +0100 | [diff] [blame] | 713 | u = PyUnicode_AsUnicodeAndSize(obj, &len); |
| 714 | if (u == NULL) |
| 715 | return NULL; |
Victor Stinner | 049e509 | 2014-08-17 22:20:00 +0200 | [diff] [blame] | 716 | 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] | 717 | return PyErr_NoMemory(); |
| 718 | size = len * sizeof(Py_UNICODE); |
| 719 | return codec_tuple(PyBytes_FromStringAndSize((const char*)u, size), |
Martin v. Löwis | 1db7c13 | 2011-11-10 18:24:32 +0100 | [diff] [blame] | 720 | PyUnicode_GET_LENGTH(obj)); |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 721 | } |
| 722 | else { |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 723 | Py_buffer view; |
| 724 | PyObject *result; |
| 725 | if (PyObject_GetBuffer(obj, &view, PyBUF_SIMPLE) != 0) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 726 | return NULL; |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 727 | result = codec_tuple(PyBytes_FromStringAndSize(view.buf, view.len), |
| 728 | view.len); |
Serhiy Storchaka | 4fdb684 | 2015-02-03 01:21:08 +0200 | [diff] [blame] | 729 | PyBuffer_Release(&view); |
| 730 | return result; |
Marc-André Lemburg | b425f5e | 2000-09-21 21:09:45 +0000 | [diff] [blame] | 731 | } |
| 732 | } |
| 733 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 734 | /*[clinic input] |
| 735 | _codecs.utf_7_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 736 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 737 | errors: str(accept={str, NoneType}) = NULL |
| 738 | / |
| 739 | [clinic start generated code]*/ |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 740 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 741 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 742 | _codecs_utf_7_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 743 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 744 | /*[clinic end generated code: output=0feda21ffc921bc8 input=d1a47579e79cbe15]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 745 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 746 | return codec_tuple(_PyUnicode_EncodeUTF7(str, 0, 0, errors), |
| 747 | PyUnicode_GET_LENGTH(str)); |
Marc-André Lemburg | c60e6f7 | 2001-09-20 10:35:46 +0000 | [diff] [blame] | 748 | } |
| 749 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 750 | /*[clinic input] |
| 751 | _codecs.utf_8_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 752 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 753 | errors: str(accept={str, NoneType}) = NULL |
| 754 | / |
| 755 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 756 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 757 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 758 | _codecs_utf_8_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 759 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 760 | /*[clinic end generated code: output=02bf47332b9c796c input=42e3ba73c4392eef]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 761 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 762 | return codec_tuple(_PyUnicode_AsUTF8String(str, errors), |
| 763 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 764 | } |
| 765 | |
| 766 | /* This version provides access to the byteorder parameter of the |
| 767 | builtin UTF-16 codecs as optional third argument. It defaults to 0 |
| 768 | 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] | 769 | BOM mark. |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 770 | |
| 771 | */ |
| 772 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 773 | /*[clinic input] |
| 774 | _codecs.utf_16_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 775 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 776 | errors: str(accept={str, NoneType}) = NULL |
| 777 | byteorder: int = 0 |
| 778 | / |
| 779 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 780 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 781 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 782 | _codecs_utf_16_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 783 | const char *errors, int byteorder) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 784 | /*[clinic end generated code: output=c654e13efa2e64e4 input=ff46416b04edb944]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 785 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 786 | return codec_tuple(_PyUnicode_EncodeUTF16(str, errors, byteorder), |
| 787 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 788 | } |
| 789 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 790 | /*[clinic input] |
| 791 | _codecs.utf_16_le_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 792 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 793 | errors: str(accept={str, NoneType}) = NULL |
| 794 | / |
| 795 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 796 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 797 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 798 | _codecs_utf_16_le_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 799 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 800 | /*[clinic end generated code: output=431b01e55f2d4995 input=cb385455ea8f2fe0]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 801 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 802 | return codec_tuple(_PyUnicode_EncodeUTF16(str, errors, -1), |
| 803 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 804 | } |
| 805 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 806 | /*[clinic input] |
| 807 | _codecs.utf_16_be_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 808 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 809 | errors: str(accept={str, NoneType}) = NULL |
| 810 | / |
| 811 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 812 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 813 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 814 | _codecs_utf_16_be_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 815 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 816 | /*[clinic end generated code: output=96886a6fd54dcae3 input=9119997066bdaefd]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 817 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 818 | return codec_tuple(_PyUnicode_EncodeUTF16(str, errors, +1), |
| 819 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 820 | } |
| 821 | |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 822 | /* This version provides access to the byteorder parameter of the |
| 823 | builtin UTF-32 codecs as optional third argument. It defaults to 0 |
| 824 | which means: use the native byte order and prepend the data with a |
| 825 | BOM mark. |
| 826 | |
| 827 | */ |
| 828 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 829 | /*[clinic input] |
| 830 | _codecs.utf_32_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 831 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 832 | errors: str(accept={str, NoneType}) = NULL |
| 833 | byteorder: int = 0 |
| 834 | / |
| 835 | [clinic start generated code]*/ |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 836 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 837 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 838 | _codecs_utf_32_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 839 | const char *errors, int byteorder) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 840 | /*[clinic end generated code: output=5c760da0c09a8b83 input=c5e77da82fbe5c2a]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 841 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 842 | return codec_tuple(_PyUnicode_EncodeUTF32(str, errors, byteorder), |
| 843 | PyUnicode_GET_LENGTH(str)); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 844 | } |
| 845 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 846 | /*[clinic input] |
| 847 | _codecs.utf_32_le_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 848 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 849 | errors: str(accept={str, NoneType}) = NULL |
| 850 | / |
| 851 | [clinic start generated code]*/ |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 852 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 853 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 854 | _codecs_utf_32_le_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 855 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 856 | /*[clinic end generated code: output=b65cd176de8e36d6 input=9993b25fe0877848]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 857 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 858 | return codec_tuple(_PyUnicode_EncodeUTF32(str, errors, -1), |
| 859 | PyUnicode_GET_LENGTH(str)); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 860 | } |
| 861 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 862 | /*[clinic input] |
| 863 | _codecs.utf_32_be_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 864 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 865 | errors: str(accept={str, NoneType}) = NULL |
| 866 | / |
| 867 | [clinic start generated code]*/ |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 868 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 869 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 870 | _codecs_utf_32_be_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 871 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 872 | /*[clinic end generated code: output=1d9e71a9358709e9 input=d3e0ccaa02920431]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 873 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 874 | return codec_tuple(_PyUnicode_EncodeUTF32(str, errors, +1), |
| 875 | PyUnicode_GET_LENGTH(str)); |
Walter Dörwald | 41980ca | 2007-08-16 21:55:45 +0000 | [diff] [blame] | 876 | } |
| 877 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 878 | /*[clinic input] |
| 879 | _codecs.unicode_escape_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 880 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 881 | errors: str(accept={str, NoneType}) = NULL |
| 882 | / |
| 883 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 884 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 885 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 886 | _codecs_unicode_escape_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 887 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 888 | /*[clinic end generated code: output=66271b30bc4f7a3c input=65d9eefca65b455a]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 889 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 890 | return codec_tuple(PyUnicode_AsUnicodeEscapeString(str), |
| 891 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 892 | } |
| 893 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 894 | /*[clinic input] |
| 895 | _codecs.raw_unicode_escape_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 896 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 897 | errors: str(accept={str, NoneType}) = NULL |
| 898 | / |
| 899 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 900 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 901 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 902 | _codecs_raw_unicode_escape_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 903 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 904 | /*[clinic end generated code: output=a66a806ed01c830a input=5aa33e4a133391ab]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 905 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 906 | return codec_tuple(PyUnicode_AsRawUnicodeEscapeString(str), |
| 907 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 908 | } |
| 909 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 910 | /*[clinic input] |
| 911 | _codecs.latin_1_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 912 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 913 | errors: str(accept={str, NoneType}) = NULL |
| 914 | / |
| 915 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 916 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 917 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 918 | _codecs_latin_1_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 919 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 920 | /*[clinic end generated code: output=2c28c83a27884e08 input=30b11c9e49a65150]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 921 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 922 | return codec_tuple(_PyUnicode_AsLatin1String(str, errors), |
| 923 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 924 | } |
| 925 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 926 | /*[clinic input] |
| 927 | _codecs.ascii_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 928 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 929 | errors: str(accept={str, NoneType}) = NULL |
| 930 | / |
| 931 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 932 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 933 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 934 | _codecs_ascii_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 935 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 936 | /*[clinic end generated code: output=b5e035182d33befc input=843a1d268e6dfa8e]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 937 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 938 | return codec_tuple(_PyUnicode_AsASCIIString(str, errors), |
| 939 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 940 | } |
| 941 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 942 | /*[clinic input] |
| 943 | _codecs.charmap_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 944 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 945 | errors: str(accept={str, NoneType}) = NULL |
| 946 | mapping: object = NULL |
| 947 | / |
| 948 | [clinic start generated code]*/ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 949 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 950 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 951 | _codecs_charmap_encode_impl(PyObject *module, PyObject *str, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 952 | const char *errors, PyObject *mapping) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 953 | /*[clinic end generated code: output=047476f48495a9e9 input=0752cde07a6d6d00]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 954 | { |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 955 | if (mapping == Py_None) |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 956 | mapping = NULL; |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 957 | |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 958 | return codec_tuple(_PyUnicode_EncodeCharmap(str, mapping, errors), |
| 959 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 960 | } |
| 961 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 962 | /*[clinic input] |
| 963 | _codecs.charmap_build |
| 964 | map: unicode |
| 965 | / |
| 966 | [clinic start generated code]*/ |
| 967 | |
| 968 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 969 | _codecs_charmap_build_impl(PyObject *module, PyObject *map) |
| 970 | /*[clinic end generated code: output=bb073c27031db9ac input=d91a91d1717dbc6d]*/ |
Thomas Wouters | 73e5a5b | 2006-06-08 15:35:45 +0000 | [diff] [blame] | 971 | { |
Thomas Wouters | 73e5a5b | 2006-06-08 15:35:45 +0000 | [diff] [blame] | 972 | return PyUnicode_BuildEncodingMap(map); |
| 973 | } |
| 974 | |
Steve Dower | cc16be8 | 2016-09-08 10:35:16 -0700 | [diff] [blame] | 975 | #ifdef MS_WINDOWS |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 976 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 977 | /*[clinic input] |
| 978 | _codecs.mbcs_encode |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 979 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 980 | errors: str(accept={str, NoneType}) = NULL |
| 981 | / |
| 982 | [clinic start generated code]*/ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 983 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 984 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 985 | _codecs_mbcs_encode_impl(PyObject *module, PyObject *str, const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 986 | /*[clinic end generated code: output=76e2e170c966c080 input=de471e0815947553]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 987 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 988 | return codec_tuple(PyUnicode_EncodeCodePage(CP_ACP, str, errors), |
| 989 | PyUnicode_GET_LENGTH(str)); |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 990 | } |
| 991 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 992 | /*[clinic input] |
Steve Dower | f5aba58 | 2016-09-06 19:42:27 -0700 | [diff] [blame] | 993 | _codecs.oem_encode |
| 994 | str: unicode |
| 995 | errors: str(accept={str, NoneType}) = NULL |
| 996 | / |
| 997 | [clinic start generated code]*/ |
| 998 | |
| 999 | static PyObject * |
| 1000 | _codecs_oem_encode_impl(PyObject *module, PyObject *str, const char *errors) |
| 1001 | /*[clinic end generated code: output=65d5982c737de649 input=3fc5f0028aad3cda]*/ |
| 1002 | { |
| 1003 | return codec_tuple(PyUnicode_EncodeCodePage(CP_OEMCP, str, errors), |
| 1004 | PyUnicode_GET_LENGTH(str)); |
| 1005 | } |
| 1006 | |
| 1007 | /*[clinic input] |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1008 | _codecs.code_page_encode |
| 1009 | code_page: int |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 1010 | str: unicode |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1011 | errors: str(accept={str, NoneType}) = NULL |
| 1012 | / |
| 1013 | [clinic start generated code]*/ |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 1014 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1015 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 1016 | _codecs_code_page_encode_impl(PyObject *module, int code_page, PyObject *str, |
| 1017 | const char *errors) |
Serhiy Storchaka | 2954f83 | 2016-07-07 18:20:03 +0300 | [diff] [blame] | 1018 | /*[clinic end generated code: output=45673f6085657a9e input=786421ae617d680b]*/ |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1019 | { |
Serhiy Storchaka | 21a663e | 2016-04-13 15:37:23 +0300 | [diff] [blame] | 1020 | return codec_tuple(PyUnicode_EncodeCodePage(code_page, str, errors), |
| 1021 | PyUnicode_GET_LENGTH(str)); |
Victor Stinner | 3a50e70 | 2011-10-18 21:21:00 +0200 | [diff] [blame] | 1022 | } |
| 1023 | |
Steve Dower | cc16be8 | 2016-09-08 10:35:16 -0700 | [diff] [blame] | 1024 | #endif /* MS_WINDOWS */ |
Guido van Rossum | 24bdb04 | 2000-03-28 20:29:59 +0000 | [diff] [blame] | 1025 | |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1026 | /* --- Error handler registry --------------------------------------------- */ |
| 1027 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1028 | /*[clinic input] |
| 1029 | _codecs.register_error |
| 1030 | errors: str |
| 1031 | handler: object |
| 1032 | / |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1033 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1034 | Register the specified error handler under the name errors. |
| 1035 | |
| 1036 | handler must be a callable object, that will be called with an exception |
| 1037 | instance containing information about the location of the encoding/decoding |
| 1038 | error and must return a (replacement, new position) tuple. |
| 1039 | [clinic start generated code]*/ |
| 1040 | |
| 1041 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 1042 | _codecs_register_error_impl(PyObject *module, const char *errors, |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1043 | PyObject *handler) |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 1044 | /*[clinic end generated code: output=fa2f7d1879b3067d input=5e6709203c2e33fe]*/ |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1045 | { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1046 | if (PyCodec_RegisterError(errors, handler)) |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1047 | return NULL; |
Thomas Wouters | 4d70c3d | 2006-06-08 14:42:34 +0000 | [diff] [blame] | 1048 | Py_RETURN_NONE; |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1049 | } |
| 1050 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1051 | /*[clinic input] |
| 1052 | _codecs.lookup_error |
| 1053 | name: str |
| 1054 | / |
Walter Dörwald | 0ae2981 | 2002-10-31 13:36:29 +0000 | [diff] [blame] | 1055 | |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1056 | lookup_error(errors) -> handler |
| 1057 | |
| 1058 | Return the error handler for the specified error handling name or raise a |
| 1059 | LookupError, if no handler exists under this name. |
| 1060 | [clinic start generated code]*/ |
| 1061 | |
| 1062 | static PyObject * |
Serhiy Storchaka | 1a2b24f | 2016-07-07 17:35:15 +0300 | [diff] [blame] | 1063 | _codecs_lookup_error_impl(PyObject *module, const char *name) |
| 1064 | /*[clinic end generated code: output=087f05dc0c9a98cc input=4775dd65e6235aba]*/ |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1065 | { |
Walter Dörwald | 3aeb632 | 2002-09-02 13:14:32 +0000 | [diff] [blame] | 1066 | return PyCodec_LookupError(name); |
| 1067 | } |
| 1068 | |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1069 | /* --- Module API --------------------------------------------------------- */ |
| 1070 | |
| 1071 | static PyMethodDef _codecs_functions[] = { |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1072 | _CODECS_REGISTER_METHODDEF |
| 1073 | _CODECS_LOOKUP_METHODDEF |
| 1074 | _CODECS_ENCODE_METHODDEF |
| 1075 | _CODECS_DECODE_METHODDEF |
| 1076 | _CODECS_ESCAPE_ENCODE_METHODDEF |
| 1077 | _CODECS_ESCAPE_DECODE_METHODDEF |
| 1078 | _CODECS_UTF_8_ENCODE_METHODDEF |
| 1079 | _CODECS_UTF_8_DECODE_METHODDEF |
| 1080 | _CODECS_UTF_7_ENCODE_METHODDEF |
| 1081 | _CODECS_UTF_7_DECODE_METHODDEF |
| 1082 | _CODECS_UTF_16_ENCODE_METHODDEF |
| 1083 | _CODECS_UTF_16_LE_ENCODE_METHODDEF |
| 1084 | _CODECS_UTF_16_BE_ENCODE_METHODDEF |
| 1085 | _CODECS_UTF_16_DECODE_METHODDEF |
| 1086 | _CODECS_UTF_16_LE_DECODE_METHODDEF |
| 1087 | _CODECS_UTF_16_BE_DECODE_METHODDEF |
| 1088 | _CODECS_UTF_16_EX_DECODE_METHODDEF |
| 1089 | _CODECS_UTF_32_ENCODE_METHODDEF |
| 1090 | _CODECS_UTF_32_LE_ENCODE_METHODDEF |
| 1091 | _CODECS_UTF_32_BE_ENCODE_METHODDEF |
| 1092 | _CODECS_UTF_32_DECODE_METHODDEF |
| 1093 | _CODECS_UTF_32_LE_DECODE_METHODDEF |
| 1094 | _CODECS_UTF_32_BE_DECODE_METHODDEF |
| 1095 | _CODECS_UTF_32_EX_DECODE_METHODDEF |
| 1096 | _CODECS_UNICODE_ESCAPE_ENCODE_METHODDEF |
| 1097 | _CODECS_UNICODE_ESCAPE_DECODE_METHODDEF |
| 1098 | _CODECS_UNICODE_INTERNAL_ENCODE_METHODDEF |
| 1099 | _CODECS_UNICODE_INTERNAL_DECODE_METHODDEF |
| 1100 | _CODECS_RAW_UNICODE_ESCAPE_ENCODE_METHODDEF |
| 1101 | _CODECS_RAW_UNICODE_ESCAPE_DECODE_METHODDEF |
| 1102 | _CODECS_LATIN_1_ENCODE_METHODDEF |
| 1103 | _CODECS_LATIN_1_DECODE_METHODDEF |
| 1104 | _CODECS_ASCII_ENCODE_METHODDEF |
| 1105 | _CODECS_ASCII_DECODE_METHODDEF |
| 1106 | _CODECS_CHARMAP_ENCODE_METHODDEF |
| 1107 | _CODECS_CHARMAP_DECODE_METHODDEF |
| 1108 | _CODECS_CHARMAP_BUILD_METHODDEF |
| 1109 | _CODECS_READBUFFER_ENCODE_METHODDEF |
| 1110 | _CODECS_MBCS_ENCODE_METHODDEF |
| 1111 | _CODECS_MBCS_DECODE_METHODDEF |
Steve Dower | f5aba58 | 2016-09-06 19:42:27 -0700 | [diff] [blame] | 1112 | _CODECS_OEM_ENCODE_METHODDEF |
| 1113 | _CODECS_OEM_DECODE_METHODDEF |
Serhiy Storchaka | 0c59ff6 | 2015-05-12 13:15:57 +0300 | [diff] [blame] | 1114 | _CODECS_CODE_PAGE_ENCODE_METHODDEF |
| 1115 | _CODECS_CODE_PAGE_DECODE_METHODDEF |
| 1116 | _CODECS_REGISTER_ERROR_METHODDEF |
| 1117 | _CODECS_LOOKUP_ERROR_METHODDEF |
Nick Coghlan | 8fad167 | 2014-09-15 23:50:44 +1200 | [diff] [blame] | 1118 | _CODECS__FORGET_CODEC_METHODDEF |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1119 | {NULL, NULL} /* sentinel */ |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1120 | }; |
| 1121 | |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1122 | static struct PyModuleDef codecsmodule = { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1123 | PyModuleDef_HEAD_INIT, |
| 1124 | "_codecs", |
| 1125 | NULL, |
| 1126 | -1, |
| 1127 | _codecs_functions, |
| 1128 | NULL, |
| 1129 | NULL, |
| 1130 | NULL, |
| 1131 | NULL |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1132 | }; |
| 1133 | |
Mark Hammond | fe51c6d | 2002-08-02 02:27:13 +0000 | [diff] [blame] | 1134 | PyMODINIT_FUNC |
Martin v. Löwis | 1a21451 | 2008-06-11 05:26:20 +0000 | [diff] [blame] | 1135 | PyInit__codecs(void) |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1136 | { |
Antoine Pitrou | f95a1b3 | 2010-05-09 15:52:27 +0000 | [diff] [blame] | 1137 | return PyModule_Create(&codecsmodule); |
Guido van Rossum | e2d67f9 | 2000-03-10 23:09:23 +0000 | [diff] [blame] | 1138 | } |