Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 1 | """ codecs -- Python Codec Registry, API and helpers. |
| 2 | |
| 3 | |
| 4 | Written by Marc-Andre Lemburg (mal@lemburg.com). |
| 5 | |
| 6 | (c) Copyright CNRI, All Rights Reserved. NO WARRANTY. |
| 7 | |
| 8 | """#" |
| 9 | |
| 10 | import struct,types,__builtin__ |
| 11 | |
| 12 | ### Registry and builtin stateless codec functions |
| 13 | |
Guido van Rossum | b95de4f | 2000-03-31 17:25:23 +0000 | [diff] [blame] | 14 | try: |
| 15 | from _codecs import * |
| 16 | except ImportError,why: |
| 17 | raise SystemError,\ |
| 18 | 'Failed to load the builtin codecs: %s' % why |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 19 | |
| 20 | ### Constants |
| 21 | |
| 22 | # |
| 23 | # Byte Order Mark (BOM) and its possible values (BOM_BE, BOM_LE) |
| 24 | # |
| 25 | BOM = struct.pack('=H',0xFEFF) |
| 26 | # |
| 27 | BOM_BE = BOM32_BE = '\376\377' |
| 28 | # corresponds to Unicode U+FEFF in UTF-16 on big endian |
| 29 | # platforms == ZERO WIDTH NO-BREAK SPACE |
| 30 | BOM_LE = BOM32_LE = '\377\376' |
| 31 | # corresponds to Unicode U+FFFE in UTF-16 on little endian |
| 32 | # platforms == defined as being an illegal Unicode character |
| 33 | |
| 34 | # |
| 35 | # 64-bit Byte Order Marks |
| 36 | # |
| 37 | BOM64_BE = '\000\000\376\377' |
| 38 | # corresponds to Unicode U+0000FEFF in UCS-4 |
| 39 | BOM64_LE = '\377\376\000\000' |
| 40 | # corresponds to Unicode U+0000FFFE in UCS-4 |
| 41 | |
| 42 | |
| 43 | ### Codec base classes (defining the API) |
| 44 | |
| 45 | class Codec: |
| 46 | |
| 47 | """ Defines the interface for stateless encoders/decoders. |
| 48 | |
| 49 | The .encode()/.decode() methods may implement different error |
| 50 | handling schemes by providing the errors argument. These |
| 51 | string values are defined: |
| 52 | |
Guido van Rossum | d8855fd | 2000-03-24 22:14:19 +0000 | [diff] [blame] | 53 | 'strict' - raise a ValueError error (or a subclass) |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 54 | 'ignore' - ignore the character and continue with the next |
| 55 | 'replace' - replace with a suitable replacement character; |
| 56 | Python will use the official U+FFFD REPLACEMENT |
| 57 | CHARACTER for the builtin Unicode codecs. |
| 58 | |
| 59 | """ |
| 60 | def encode(self,input,errors='strict'): |
| 61 | |
Fred Drake | 3e74c0d | 2000-03-17 15:40:35 +0000 | [diff] [blame] | 62 | """ Encodes the object input and returns a tuple (output |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 63 | object, length consumed). |
| 64 | |
| 65 | errors defines the error handling to apply. It defaults to |
| 66 | 'strict' handling. |
| 67 | |
| 68 | The method may not store state in the Codec instance. Use |
| 69 | StreamCodec for codecs which have to keep state in order to |
| 70 | make encoding/decoding efficient. |
| 71 | |
| 72 | The encoder must be able to handle zero length input and |
| 73 | return an empty object of the output object type in this |
| 74 | situation. |
| 75 | |
| 76 | """ |
| 77 | raise NotImplementedError |
| 78 | |
| 79 | def decode(self,input,errors='strict'): |
| 80 | |
| 81 | """ Decodes the object input and returns a tuple (output |
| 82 | object, length consumed). |
| 83 | |
| 84 | input must be an object which provides the bf_getreadbuf |
| 85 | buffer slot. Python strings, buffer objects and memory |
| 86 | mapped files are examples of objects providing this slot. |
| 87 | |
| 88 | errors defines the error handling to apply. It defaults to |
| 89 | 'strict' handling. |
| 90 | |
| 91 | The method may not store state in the Codec instance. Use |
| 92 | StreamCodec for codecs which have to keep state in order to |
| 93 | make encoding/decoding efficient. |
| 94 | |
| 95 | The decoder must be able to handle zero length input and |
| 96 | return an empty object of the output object type in this |
| 97 | situation. |
| 98 | |
| 99 | """ |
| 100 | raise NotImplementedError |
| 101 | |
| 102 | # |
| 103 | # The StreamWriter and StreamReader class provide generic working |
| 104 | # interfaces which can be used to implement new encodings submodules |
| 105 | # very easily. See encodings/utf_8.py for an example on how this is |
| 106 | # done. |
| 107 | # |
| 108 | |
| 109 | class StreamWriter(Codec): |
| 110 | |
| 111 | def __init__(self,stream,errors='strict'): |
| 112 | |
| 113 | """ Creates a StreamWriter instance. |
| 114 | |
| 115 | stream must be a file-like object open for writing |
| 116 | (binary) data. |
| 117 | |
| 118 | The StreamWriter may implement different error handling |
| 119 | schemes by providing the errors keyword argument. These |
| 120 | parameters are defined: |
| 121 | |
| 122 | 'strict' - raise a ValueError (or a subclass) |
| 123 | 'ignore' - ignore the character and continue with the next |
| 124 | 'replace'- replace with a suitable replacement character |
| 125 | |
| 126 | """ |
| 127 | self.stream = stream |
| 128 | self.errors = errors |
| 129 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 130 | def write(self, object): |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 131 | |
| 132 | """ Writes the object's contents encoded to self.stream. |
| 133 | """ |
| 134 | data, consumed = self.encode(object,self.errors) |
| 135 | self.stream.write(data) |
| 136 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 137 | def writelines(self, list): |
| 138 | |
| 139 | """ Writes the concatenated list of strings to the stream |
| 140 | using .write(). |
| 141 | """ |
| 142 | self.write(''.join(list)) |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 143 | |
| 144 | def reset(self): |
| 145 | |
| 146 | """ Flushes and resets the codec buffers used for keeping state. |
| 147 | |
| 148 | Calling this method should ensure that the data on the |
| 149 | output is put into a clean state, that allows appending |
| 150 | of new fresh data without having to rescan the whole |
| 151 | stream to recover state. |
| 152 | |
| 153 | """ |
| 154 | pass |
| 155 | |
| 156 | def __getattr__(self,name, |
| 157 | |
| 158 | getattr=getattr): |
| 159 | |
| 160 | """ Inherit all other methods from the underlying stream. |
| 161 | """ |
| 162 | return getattr(self.stream,name) |
| 163 | |
| 164 | ### |
| 165 | |
| 166 | class StreamReader(Codec): |
| 167 | |
| 168 | def __init__(self,stream,errors='strict'): |
| 169 | |
| 170 | """ Creates a StreamReader instance. |
| 171 | |
| 172 | stream must be a file-like object open for reading |
| 173 | (binary) data. |
| 174 | |
| 175 | The StreamReader may implement different error handling |
| 176 | schemes by providing the errors keyword argument. These |
| 177 | parameters are defined: |
| 178 | |
| 179 | 'strict' - raise a ValueError (or a subclass) |
| 180 | 'ignore' - ignore the character and continue with the next |
| 181 | 'replace'- replace with a suitable replacement character; |
| 182 | |
| 183 | """ |
| 184 | self.stream = stream |
| 185 | self.errors = errors |
| 186 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 187 | def read(self, size=-1): |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 188 | |
| 189 | """ Decodes data from the stream self.stream and returns the |
| 190 | resulting object. |
| 191 | |
| 192 | size indicates the approximate maximum number of bytes to |
| 193 | read from the stream for decoding purposes. The decoder |
| 194 | can modify this setting as appropriate. The default value |
| 195 | -1 indicates to read and decode as much as possible. size |
| 196 | is intended to prevent having to decode huge files in one |
| 197 | step. |
| 198 | |
| 199 | The method should use a greedy read strategy meaning that |
| 200 | it should read as much data as is allowed within the |
| 201 | definition of the encoding and the given size, e.g. if |
| 202 | optional encoding endings or state markers are available |
| 203 | on the stream, these should be read too. |
| 204 | |
| 205 | """ |
| 206 | # Unsliced reading: |
| 207 | if size < 0: |
| 208 | return self.decode(self.stream.read())[0] |
| 209 | |
| 210 | # Sliced reading: |
| 211 | read = self.stream.read |
| 212 | decode = self.decode |
| 213 | data = read(size) |
| 214 | i = 0 |
| 215 | while 1: |
| 216 | try: |
| 217 | object, decodedbytes = decode(data) |
| 218 | except ValueError,why: |
| 219 | # This method is slow but should work under pretty much |
| 220 | # all conditions; at most 10 tries are made |
| 221 | i = i + 1 |
| 222 | newdata = read(1) |
| 223 | if not newdata or i > 10: |
| 224 | raise |
| 225 | data = data + newdata |
| 226 | else: |
| 227 | return object |
| 228 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 229 | def readline(self, size=None): |
| 230 | |
| 231 | """ Read one line from the input stream and return the |
| 232 | decoded data. |
| 233 | |
| 234 | Note: Unlike the .readlines() method, line breaking must |
| 235 | be implemented by the underlying stream's .readline() |
| 236 | method -- there is currently no support for line breaking |
| 237 | using the codec decoder due to lack of line buffering. |
| 238 | |
| 239 | size, if given, is passed as size argument to the stream's |
| 240 | .readline() method. |
| 241 | |
| 242 | """ |
| 243 | if size is None: |
| 244 | line = self.stream.readline() |
| 245 | else: |
| 246 | line = self.stream.readline(size) |
| 247 | return self.decode(line)[0] |
| 248 | |
| 249 | |
| 250 | def readlines(self, sizehint=0): |
| 251 | |
| 252 | """ Read all lines available on the input stream |
| 253 | and return them as list of lines. |
| 254 | |
| 255 | Line breaks are implemented using the codec's decoder |
| 256 | method and are included in the list entries. |
| 257 | |
| 258 | sizehint, if given, is passed as size argument to the |
| 259 | stream's .read() method. |
| 260 | |
| 261 | """ |
| 262 | if sizehint is None: |
| 263 | data = self.stream.read() |
| 264 | else: |
| 265 | data = self.stream.read(sizehint) |
| 266 | return self.decode(data)[0].splitlines(1) |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 267 | |
| 268 | def reset(self): |
| 269 | |
| 270 | """ Resets the codec buffers used for keeping state. |
| 271 | |
| 272 | Note that no stream repositioning should take place. |
| 273 | This method is primarely intended to be able to recover |
| 274 | from decoding errors. |
| 275 | |
| 276 | """ |
| 277 | pass |
| 278 | |
| 279 | def __getattr__(self,name, |
| 280 | |
| 281 | getattr=getattr): |
| 282 | |
| 283 | """ Inherit all other methods from the underlying stream. |
| 284 | """ |
| 285 | return getattr(self.stream,name) |
| 286 | |
| 287 | ### |
| 288 | |
| 289 | class StreamReaderWriter: |
| 290 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 291 | # Optional attributes set by the file wrappers below |
| 292 | encoding = 'unknown' |
| 293 | |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 294 | def __init__(self,stream,Reader,Writer,errors='strict'): |
| 295 | |
| 296 | """ Creates a StreamReaderWriter instance. |
| 297 | |
| 298 | stream must be a Stream-like object. |
| 299 | |
| 300 | Reader, Writer must be factory functions or classes |
| 301 | providing the StreamReader, StreamWriter interface resp. |
| 302 | |
| 303 | Error handling is done in the same way as defined for the |
| 304 | StreamWriter/Readers. |
| 305 | |
| 306 | """ |
| 307 | self.stream = stream |
| 308 | self.reader = Reader(stream, errors) |
| 309 | self.writer = Writer(stream, errors) |
| 310 | self.errors = errors |
| 311 | |
| 312 | def read(self,size=-1): |
| 313 | |
| 314 | return self.reader.read(size) |
| 315 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 316 | def readline(size=None): |
| 317 | |
| 318 | return self.reader.readline(size) |
| 319 | |
| 320 | def readlines(sizehint=None): |
| 321 | |
| 322 | return self.reader.readlines(sizehint) |
| 323 | |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 324 | def write(self,data): |
| 325 | |
| 326 | return self.writer.write(data) |
| 327 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 328 | def writelines(self,list): |
| 329 | |
| 330 | return self.writer.writelines(list) |
| 331 | |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 332 | def reset(self): |
| 333 | |
| 334 | self.reader.reset() |
| 335 | self.writer.reset() |
| 336 | |
| 337 | def __getattr__(self,name, |
| 338 | |
| 339 | getattr=getattr): |
| 340 | |
| 341 | """ Inherit all other methods from the underlying stream. |
| 342 | """ |
| 343 | return getattr(self.stream,name) |
| 344 | |
| 345 | ### |
| 346 | |
| 347 | class StreamRecoder: |
| 348 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 349 | # Optional attributes set by the file wrappers below |
| 350 | data_encoding = 'unknown' |
| 351 | file_encoding = 'unknown' |
| 352 | |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 353 | def __init__(self,stream,encode,decode,Reader,Writer,errors='strict'): |
| 354 | |
| 355 | """ Creates a StreamRecoder instance which implements a two-way |
| 356 | conversion: encode and decode work on the frontend (the |
| 357 | input to .read() and output of .write()) while |
| 358 | Reader and Writer work on the backend (reading and |
Fred Drake | 908670c | 2000-03-17 15:42:11 +0000 | [diff] [blame] | 359 | writing to the stream). |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 360 | |
| 361 | You can use these objects to do transparent direct |
| 362 | recodings from e.g. latin-1 to utf-8 and back. |
| 363 | |
| 364 | stream must be a file-like object. |
| 365 | |
| 366 | encode, decode must adhere to the Codec interface, Reader, |
| 367 | Writer must be factory functions or classes providing the |
| 368 | StreamReader, StreamWriter interface resp. |
| 369 | |
| 370 | encode and decode are needed for the frontend translation, |
| 371 | Reader and Writer for the backend translation. Unicode is |
| 372 | used as intermediate encoding. |
| 373 | |
| 374 | Error handling is done in the same way as defined for the |
| 375 | StreamWriter/Readers. |
| 376 | |
| 377 | """ |
| 378 | self.stream = stream |
| 379 | self.encode = encode |
| 380 | self.decode = decode |
| 381 | self.reader = Reader(stream, errors) |
| 382 | self.writer = Writer(stream, errors) |
| 383 | self.errors = errors |
| 384 | |
| 385 | def read(self,size=-1): |
| 386 | |
| 387 | data = self.reader.read(size) |
| 388 | data, bytesencoded = self.encode(data, self.errors) |
| 389 | return data |
| 390 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 391 | def readline(self,size=None): |
| 392 | |
| 393 | if size is None: |
| 394 | data = self.reader.readline() |
| 395 | else: |
| 396 | data = self.reader.readline(size) |
| 397 | data, bytesencoded = self.encode(data, self.errors) |
| 398 | return data |
| 399 | |
| 400 | def readlines(self,sizehint=None): |
| 401 | |
| 402 | if sizehint is None: |
| 403 | data = self.reader.read() |
| 404 | else: |
| 405 | data = self.reader.read(sizehint) |
| 406 | data, bytesencoded = self.encode(data, self.errors) |
| 407 | return data.splitlines(1) |
| 408 | |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 409 | def write(self,data): |
| 410 | |
| 411 | data, bytesdecoded = self.decode(data, self.errors) |
| 412 | return self.writer.write(data) |
| 413 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 414 | def writelines(self,list): |
| 415 | |
| 416 | data = ''.join(list) |
| 417 | data, bytesdecoded = self.decode(data, self.errors) |
| 418 | return self.writer.write(data) |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 419 | |
| 420 | def reset(self): |
| 421 | |
| 422 | self.reader.reset() |
| 423 | self.writer.reset() |
| 424 | |
| 425 | def __getattr__(self,name, |
| 426 | |
| 427 | getattr=getattr): |
| 428 | |
| 429 | """ Inherit all other methods from the underlying stream. |
| 430 | """ |
| 431 | return getattr(self.stream,name) |
| 432 | |
| 433 | ### Shortcuts |
| 434 | |
| 435 | def open(filename, mode, encoding=None, errors='strict', buffering=1): |
| 436 | |
| 437 | """ Open an encoded file using the given mode and return |
| 438 | a wrapped version providing transparent encoding/decoding. |
| 439 | |
| 440 | Note: The wrapped version will only accept the object format |
| 441 | defined by the codecs, i.e. Unicode objects for most builtin |
| 442 | codecs. Output is also codec dependent and will usually by |
| 443 | Unicode as well. |
| 444 | |
| 445 | encoding specifies the encoding which is to be used for the |
| 446 | the file. |
| 447 | |
| 448 | errors may be given to define the error handling. It defaults |
| 449 | to 'strict' which causes ValueErrors to be raised in case an |
| 450 | encoding error occurs. |
| 451 | |
| 452 | buffering has the same meaning as for the builtin open() API. |
| 453 | It defaults to line buffered. |
| 454 | |
| 455 | """ |
| 456 | if encoding is not None and \ |
| 457 | 'b' not in mode: |
| 458 | # Force opening of the file in binary mode |
| 459 | mode = mode + 'b' |
| 460 | file = __builtin__.open(filename, mode, buffering) |
| 461 | if encoding is None: |
| 462 | return file |
| 463 | (e,d,sr,sw) = lookup(encoding) |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 464 | srw = StreamReaderWriter(file, sr, sw, errors) |
| 465 | # Add attributes to simplify introspection |
| 466 | srw.encoding = encoding |
| 467 | return srw |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 468 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 469 | def EncodedFile(file, data_encoding, file_encoding=None, errors='strict'): |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 470 | |
| 471 | """ Return a wrapped version of file which provides transparent |
| 472 | encoding translation. |
| 473 | |
| 474 | Strings written to the wrapped file are interpreted according |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 475 | to the given data_encoding and then written to the original |
| 476 | file as string using file_encoding. The intermediate encoding |
| 477 | will usually be Unicode but depends on the specified codecs. |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 478 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 479 | Strings are read from the file using file_encoding and then |
| 480 | passed back to the caller as string using data_encoding. |
| 481 | |
| 482 | If file_encoding is not given, it defaults to data_encoding. |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 483 | |
| 484 | errors may be given to define the error handling. It defaults |
| 485 | to 'strict' which causes ValueErrors to be raised in case an |
| 486 | encoding error occurs. |
| 487 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 488 | data_encoding and file_encoding are added to the wrapped file |
| 489 | object as attributes .data_encoding and .file_encoding resp. |
| 490 | |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 491 | """ |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 492 | if file_encoding is None: |
| 493 | file_encoding = data_encoding |
| 494 | encode, decode = lookup(data_encoding)[:2] |
| 495 | Reader, Writer = lookup(file_encoding)[2:] |
| 496 | sr = StreamRecoder(file, |
| 497 | encode,decode,Reader,Writer, |
| 498 | errors) |
| 499 | # Add attributes to simplify introspection |
| 500 | sr.data_encoding = data_encoding |
| 501 | sr.file_encoding = file_encoding |
| 502 | return sr |
Guido van Rossum | 0612d84 | 2000-03-10 23:20:43 +0000 | [diff] [blame] | 503 | |
| 504 | ### Tests |
| 505 | |
| 506 | if __name__ == '__main__': |
| 507 | |
| 508 | import sys |
| 509 | |
Guido van Rossum | a327713 | 2000-04-11 15:37:43 +0000 | [diff] [blame^] | 510 | # Make stdout translate Latin-1 output into UTF-8 output |
| 511 | sys.stdout = EncodedFile(sys.stdout, 'latin-1', 'utf-8') |
| 512 | |
| 513 | # Have stdin translate Latin-1 input into UTF-8 input |
| 514 | sys.stdin = EncodedFile(sys.stdin, 'utf-8', 'latin-1') |