docutils/io.py

   1 # $Id$
   2 # Author: David Goodger <goodger@python.org>
   3 # Copyright: This module has been placed in the public domain.
   4
   5 """
   6 I/O classes provide a uniform API for low-level input and output.  Subclasses
   7 will exist for a variety of input/output mechanisms.
   8 """
   9
  10 __docformat__ = 'reStructuredText'
  11
  12 import sys
  13 try:
  14     import locale
  15 except:
  16     pass
  17 import re
  18 from types import UnicodeType
  19 from docutils import TransformSpec
  20
  21
  22 class Input(TransformSpec):
  23
  24     """
  25     Abstract base class for input wrappers.
  26     """
  27
  28     component_type = 'input'
  29
  30     default_source_path = None
  31
  32     def __init__(self, source=None, source_path=None, encoding=None,
  33                  error_handler='strict'):
  34         self.encoding = encoding
  35         """Text encoding for the input source."""
  36
  37         self.error_handler = error_handler
  38         """Text decoding error handler."""
  39
  40         self.source = source
  41         """The source of input data."""
  42
  43         self.source_path = source_path
  44         """A text reference to the source."""
  45
  46         if not source_path:
  47             self.source_path = self.default_source_path
  48
  49         self.successful_encoding = None
  50         """The encoding that successfully decoded the source data."""
  51
  52     def __repr__(self):
  53         return '%s: source=%r, source_path=%r' % (self.__class__, self.source,
  54                                                   self.source_path)
  55
  56     def read(self):
  57         raise NotImplementedError
  58
  59     def decode(self, data):
  60         """
  61         Decode a string, `data`, heuristically.
  62         Raise UnicodeError if unsuccessful.
  63
  64         The client application should call ``locale.setlocale`` at the
  65         beginning of processing::
  66
  67             locale.setlocale(locale.LC_ALL, '')
  68         """
  69         if self.encoding and self.encoding.lower() == 'unicode':
  70             assert isinstance(data, UnicodeType), (
  71                 'input encoding is "unicode" '
  72                 'but input is not a unicode object')
  73         if isinstance(data, UnicodeType):
  74             # Accept unicode even if self.encoding != 'unicode'.
  75             return data
  76         if self.encoding:
  77             # We believe the user/application when the encoding is
  78             # explicitly given.
  79             encodings = [self.encoding]
  80         else:
  81             data_encoding = self.determine_encoding_from_data(data)
  82             if data_encoding:
  83                 # If the data declares its encoding (explicitly or via a BOM),
  84                 # we believe it.
  85                 encodings = [data_encoding]
  86             else:
  87                 # Apply heuristics only if no encoding is explicitly given and
  88                 # no BOM found.  Start with UTF-8, because that only matches
  89                 # data that *IS* UTF-8:
  90                 encodings = ['utf-8']
  91                 try:
  92                     # for Python 2.2 compatibility
  93                     encodings.append(locale.nl_langinfo(locale.CODESET))
  94                 except:
  95                     pass
  96                 try:
  97                     encodings.append(locale.getlocale()[1])
  98                 except:
  99                     pass
 100                 try:
 101                     encodings.append(locale.getdefaultlocale()[1])
 102                 except:
 103                     pass
 104                 # fallback encoding:
 105                 encodings.append('latin-1')
 106         error = None
 107         error_details = ''
 108         for enc in encodings:
 109             if not enc:
 110                 continue
 111             try:
 112                 decoded = unicode(data, enc, self.error_handler)
 113                 self.successful_encoding = enc
 114                 # Return decoded, removing BOMs.
 115                 return decoded.replace(u'\ufeff', u'')
 116             except (UnicodeError, LookupError), error:
 117                 pass
 118         if error is not None:
 119             error_details = '\n(%s: %s)' % (error.__class__.__name__, error)
 120         raise UnicodeError(
 121             'Unable to decode input data.  Tried the following encodings: '
 122             '%s.%s'
 123             % (', '.join([repr(enc) for enc in encodings if enc]),
 124                error_details))
 125
 126     coding_slug = re.compile("coding[:=]\s*([-\w.]+)")
 127     """Encoding declaration pattern."""
 128
 129     byte_order_marks = (('\xef\xbb\xbf', 'utf-8'),
 130                         ('\xfe\xff', 'utf-16-be'),
 131                         ('\xff\xfe', 'utf-16-le'),)
 132     """Sequence of (start_bytes, encoding) tuples to for encoding detection.
 133     The first bytes of input data are checked against the start_bytes strings.
 134     A match indicates the given encoding."""
 135
 136     def determine_encoding_from_data(self, data):
 137         """
 138         Try to determine the encoding of `data` by looking *in* `data`.
 139         Check for a byte order mark (BOM) or an encoding declaration.
 140         """
 141         # check for a byte order mark:
 142         for start_bytes, encoding in self.byte_order_marks:
 143             if data.startswith(start_bytes):
 144                 return encoding
 145         # check for an encoding declaration pattern in first 2 lines of file:
 146         for line in data.splitlines()[:2]:
 147             match = self.coding_slug.search(line)
 148             if match:
 149                 return match.group(1)
 150         return None
 151
 152
 153 class Output(TransformSpec):
 154
 155     """
 156     Abstract base class for output wrappers.
 157     """
 158
 159     component_type = 'output'
 160
 161     default_destination_path = None
 162
 163     def __init__(self, destination=None, destination_path=None,
 164                  encoding=None, error_handler='strict'):
 165         self.encoding = encoding
 166         """Text encoding for the output destination."""
 167
 168         self.error_handler = error_handler or 'strict'
 169         """Text encoding error handler."""
 170
 171         self.destination = destination
 172         """The destination for output data."""
 173
 174         self.destination_path = destination_path
 175         """A text reference to the destination."""
 176
 177         if not destination_path:
 178             self.destination_path = self.default_destination_path
 179
 180     def __repr__(self):
 181         return ('%s: destination=%r, destination_path=%r'
 182                 % (self.__class__, self.destination, self.destination_path))
 183
 184     def write(self, data):
 185         """`data` is a Unicode string, to be encoded by `self.encode`."""
 186         raise NotImplementedError
 187
 188     def encode(self, data):
 189         if self.encoding and self.encoding.lower() == 'unicode':
 190             assert isinstance(data, UnicodeType), (
 191                 'the encoding given is "unicode" but the output is not '
 192                 'a Unicode string')
 193             return data
 194         if not isinstance(data, UnicodeType):
 195             # Non-unicode (e.g. binary) output.
 196             return data
 197         else:
 198             try:
 199                 return data.encode(self.encoding, self.error_handler)
 200             except (LookupError, ValueError):
 201                 # LookupError is raised if there are unencodable chars
 202                 # in data and the error_handler isn't found. In old
 203                 # Python versions, ValueError is raised.
 204                 if self.error_handler == 'xmlcharrefreplace':
 205                     # We are using xmlcharrefreplace with a Python
 206                     # version that doesn't support it (2.1, 2.2, or
 207                     # IronPython 1.0) so we emulate its behavior.
 208                     return ''.join([self.xmlcharref_encode(char)
 209                                     for char in data])
 210                 else:
 211                     raise
 212
 213     def xmlcharref_encode(self, char):
 214         """Emulate Python 2.3's 'xmlcharrefreplace' encoding error handler."""
 215         try:
 216             return char.encode(self.encoding, 'strict')
 217         except UnicodeError:
 218             return '&#%i;' % ord(char)
 219
 220
 221 class FileInput(Input):
 222
 223     """
 224     Input for single, simple file-like objects.
 225     """
 226
 227     def __init__(self, source=None, source_path=None,
 228                  encoding=None, error_handler='strict',
 229                  autoclose=1, handle_io_errors=1):
 230         """
 231         :Parameters:
 232             - `source`: either a file-like object (which is read directly), or
 233               `None` (which implies `sys.stdin` if no `source_path` given).
 234             - `source_path`: a path to a file, which is opened and then read.
 235             - `encoding`: the expected text encoding of the input file.
 236             - `error_handler`: the encoding error handler to use.
 237             - `autoclose`: close automatically after read (boolean); always
 238               false if `sys.stdin` is the source.
 239             - `handle_io_errors`: summarize I/O errors here, and exit?
 240         """
 241         Input.__init__(self, source, source_path, encoding, error_handler)
 242         self.autoclose = autoclose
 243         self.handle_io_errors = handle_io_errors
 244         if source is None:
 245             if source_path:
 246                 try:
 247                     self.source = open(source_path)
 248                 except IOError, error:
 249                     if not handle_io_errors:
 250                         raise
 251                     print >>sys.stderr, '%s: %s' % (error.__class__.__name__,
 252                                                     error)
 253                     print >>sys.stderr, (
 254                         'Unable to open source file for reading (%r).  Exiting.'
 255                         % source_path)
 256                     sys.exit(1)
 257             else:
 258                 self.source = sys.stdin
 259                 self.autoclose = None
 260         if not source_path:
 261             try:
 262                 self.source_path = self.source.name
 263             except AttributeError:
 264                 pass
 265
 266     def read(self):
 267         """
 268         Read and decode a single file and return the data (Unicode string).
 269         """
 270         try:
 271             data = self.source.read()
 272         finally:
 273             if self.autoclose:
 274                 self.close()
 275         return self.decode(data)
 276
 277     def close(self):
 278         self.source.close()
 279
 280
 281 class FileOutput(Output):
 282
 283     """
 284     Output for single, simple file-like objects.
 285     """
 286
 287     def __init__(self, destination=None, destination_path=None,
 288                  encoding=None, error_handler='strict', autoclose=1,
 289                  handle_io_errors=1):
 290         """
 291         :Parameters:
 292             - `destination`: either a file-like object (which is written
 293               directly) or `None` (which implies `sys.stdout` if no
 294               `destination_path` given).
 295             - `destination_path`: a path to a file, which is opened and then
 296               written.
 297             - `autoclose`: close automatically after write (boolean); always
 298               false if `sys.stdout` is the destination.
 299         """
 300         Output.__init__(self, destination, destination_path,
 301                         encoding, error_handler)
 302         self.opened = 1
 303         self.autoclose = autoclose
 304         self.handle_io_errors = handle_io_errors
 305         if destination is None:
 306             if destination_path:
 307                 self.opened = None
 308             else:
 309                 self.destination = sys.stdout
 310                 self.autoclose = None
 311         if not destination_path:
 312             try:
 313                 self.destination_path = self.destination.name
 314             except AttributeError:
 315                 pass
 316
 317     def open(self):
 318         try:
 319             self.destination = open(self.destination_path, 'w')
 320         except IOError, error:
 321             if not self.handle_io_errors:
 322                 raise
 323             print >>sys.stderr, '%s: %s' % (error.__class__.__name__,
 324                                             error)
 325             print >>sys.stderr, ('Unable to open destination file for writing '
 326                                  '(%r).  Exiting.' % self.destination_path)
 327             sys.exit(1)
 328         self.opened = 1
 329
 330     def write(self, data):
 331         """Encode `data`, write it to a single file, and return it."""
 332         output = self.encode(data)
 333         if not self.opened:
 334             self.open()
 335         try:
 336             self.destination.write(output)
 337         finally:
 338             if self.autoclose:
 339                 self.close()
 340         return output
 341
 342     def close(self):
 343         self.destination.close()
 344         self.opened = None
 345
 346
 347 class StringInput(Input):
 348
 349     """
 350     Direct string input.
 351     """
 352
 353     default_source_path = '<string>'
 354
 355     def read(self):
 356         """Decode and return the source string."""
 357         return self.decode(self.source)
 358
 359
 360 class StringOutput(Output):
 361
 362     """
 363     Direct string output.
 364     """
 365
 366     default_destination_path = '<string>'
 367
 368     def write(self, data):
 369         """Encode `data`, store it in `self.destination`, and return it."""
 370         self.destination = self.encode(data)
 371         return self.destination
 372
 373
 374 class NullInput(Input):
 375
 376     """
 377     Degenerate input: read nothing.
 378     """
 379
 380     default_source_path = 'null input'
 381
 382     def read(self):
 383         """Return a null string."""
 384         return u''
 385
 386
 387 class NullOutput(Output):
 388
 389     """
 390     Degenerate output: write nothing.
 391     """
 392
 393     default_destination_path = 'null output'
 394
 395     def write(self, data):
 396         """Do nothing ([don't even] send data to the bit bucket)."""
 397         pass
 398
 399
 400 class DocTreeInput(Input):
 401
 402     """
 403     Adapter for document tree input.
 404
 405     The document tree must be passed in the ``source`` parameter.
 406     """
 407
 408     default_source_path = 'doctree input'
 409
 410     def read(self):
 411         """Return the document tree."""
 412         return self.source