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 import re
  14 import codecs
  15 from docutils import TransformSpec
  16 from docutils._compat import b
  17 from docutils.error_reporting import locale_encoding, ErrorString, ErrorOutput
  18
  19 class Input(TransformSpec):
  20
  21     """
  22     Abstract base class for input wrappers.
  23     """
  24
  25     component_type = 'input'
  26
  27     default_source_path = None
  28
  29     def __init__(self, source=None, source_path=None, encoding=None,
  30                  error_handler='strict'):
  31         self.encoding = encoding
  32         """Text encoding for the input source."""
  33
  34         self.error_handler = error_handler
  35         """Text decoding error handler."""
  36
  37         self.source = source
  38         """The source of input data."""
  39
  40         self.source_path = source_path
  41         """A text reference to the source."""
  42
  43         if not source_path:
  44             self.source_path = self.default_source_path
  45
  46         self.successful_encoding = None
  47         """The encoding that successfully decoded the source data."""
  48
  49     def __repr__(self):
  50         return '%s: source=%r, source_path=%r' % (self.__class__, self.source,
  51                                                   self.source_path)
  52
  53     def read(self):
  54         raise NotImplementedError
  55
  56     def decode(self, data):
  57         """
  58         Decode a string, `data`, heuristically.
  59         Raise UnicodeError if unsuccessful.
  60
  61         The client application should call ``locale.setlocale`` at the
  62         beginning of processing::
  63
  64             locale.setlocale(locale.LC_ALL, '')
  65         """
  66         if self.encoding and self.encoding.lower() == 'unicode':
  67             assert isinstance(data, unicode), (
  68                 'input encoding is "unicode" '
  69                 'but input is not a unicode object')
  70         if isinstance(data, unicode):
  71             # Accept unicode even if self.encoding != 'unicode'.
  72             return data
  73         if self.encoding:
  74             # We believe the user/application when the encoding is
  75             # explicitly given.
  76             encodings = [self.encoding]
  77         else:
  78             data_encoding = self.determine_encoding_from_data(data)
  79             if data_encoding:
  80                 # If the data declares its encoding (explicitly or via a BOM),
  81                 # we believe it.
  82                 encodings = [data_encoding]
  83             else:
  84                 # Apply heuristics only if no encoding is explicitly given and
  85                 # no BOM found.  Start with UTF-8, because that only matches
  86                 # data that *IS* UTF-8:
  87                 encodings = [enc for enc in ('utf-8',
  88                                              locale_encoding, # can be None
  89                                              'latin-1') # fallback encoding
  90                              if enc]
  91         for enc in encodings:
  92             try:
  93                 decoded = unicode(data, enc, self.error_handler)
  94                 self.successful_encoding = enc
  95                 # Return decoded, removing BOMs.
  96                 return decoded.replace(u'\ufeff', u'')
  97             except (UnicodeError, LookupError), err:
  98                 error = err # in Python 3, the <exception instance> is
  99                             # local to the except clause
 100         raise UnicodeError(
 101             'Unable to decode input data.  Tried the following encodings: '
 102             '%s.\n(%s)' % (', '.join([repr(enc) for enc in encodings]),
 103                          ErrorString(error)))
 104
 105     coding_slug = re.compile(b("coding[:=]\s*([-\w.]+)"))
 106     """Encoding declaration pattern."""
 107
 108     byte_order_marks = ((codecs.BOM_UTF8, 'utf-8'), # actually 'utf-8-sig'
 109                         (codecs.BOM_UTF16_BE, 'utf-16-be'),
 110                         (codecs.BOM_UTF16_LE, 'utf-16-le'),)
 111     """Sequence of (start_bytes, encoding) tuples for encoding detection.
 112     The first bytes of input data are checked against the start_bytes strings.
 113     A match indicates the given encoding."""
 114
 115     def determine_encoding_from_data(self, data):
 116         """
 117         Try to determine the encoding of `data` by looking *in* `data`.
 118         Check for a byte order mark (BOM) or an encoding declaration.
 119         """
 120         # check for a byte order mark:
 121         for start_bytes, encoding in self.byte_order_marks:
 122             if data.startswith(start_bytes):
 123                 return encoding
 124         # check for an encoding declaration pattern in first 2 lines of file:
 125         for line in data.splitlines()[:2]:
 126             match = self.coding_slug.search(line)
 127             if match:
 128                 return match.group(1).decode('ascii')
 129         return None
 130
 131
 132 class Output(TransformSpec):
 133
 134     """
 135     Abstract base class for output wrappers.
 136     """
 137
 138     component_type = 'output'
 139
 140     default_destination_path = None
 141
 142     def __init__(self, destination=None, destination_path=None,
 143                  encoding=None, error_handler='strict'):
 144         self.encoding = encoding
 145         """Text encoding for the output destination."""
 146
 147         self.error_handler = error_handler or 'strict'
 148         """Text encoding error handler."""
 149
 150         self.destination = destination
 151         """The destination for output data."""
 152
 153         self.destination_path = destination_path
 154         """A text reference to the destination."""
 155
 156         if not destination_path:
 157             self.destination_path = self.default_destination_path
 158
 159     def __repr__(self):
 160         return ('%s: destination=%r, destination_path=%r'
 161                 % (self.__class__, self.destination, self.destination_path))
 162
 163     def write(self, data):
 164         """`data` is a Unicode string, to be encoded by `self.encode`."""
 165         raise NotImplementedError
 166
 167     def encode(self, data):
 168         if self.encoding and self.encoding.lower() == 'unicode':
 169             assert isinstance(data, unicode), (
 170                 'the encoding given is "unicode" but the output is not '
 171                 'a Unicode string')
 172             return data
 173         if not isinstance(data, unicode):
 174             # Non-unicode (e.g. binary) output.
 175             return data
 176         else:
 177             return data.encode(self.encoding, self.error_handler)
 178
 179
 180 class FileInput(Input):
 181
 182     """
 183     Input for single, simple file-like objects.
 184     """
 185     def __init__(self, source=None, source_path=None,
 186                  encoding=None, error_handler='strict',
 187                  autoclose=True, handle_io_errors=True, mode='rU'):
 188         """
 189         :Parameters:
 190             - `source`: either a file-like object (which is read directly), or
 191               `None` (which implies `sys.stdin` if no `source_path` given).
 192             - `source_path`: a path to a file, which is opened and then read.
 193             - `encoding`: the expected text encoding of the input file.
 194             - `error_handler`: the encoding error handler to use.
 195             - `autoclose`: close automatically after read (except when
 196               `sys.stdin` is the source).
 197             - `handle_io_errors`: summarize I/O errors here, and exit?
 198             - `mode`: how the file is to be opened (see standard function
 199               `open`). The default 'rU' provides universal newline support
 200               for text files.
 201         """
 202         Input.__init__(self, source, source_path, encoding, error_handler)
 203         self.autoclose = autoclose
 204         self.handle_io_errors = handle_io_errors
 205         self._stderr = ErrorOutput()
 206
 207         if source is None:
 208             if source_path:
 209                 # Specify encoding in Python 3
 210                 if sys.version_info >= (3,0):
 211                     kwargs = {'encoding': self.encoding,
 212                               'errors': self.error_handler}
 213                 else:
 214                     kwargs = {}
 215
 216                 try:
 217                     self.source = open(source_path, mode, **kwargs)
 218                 except IOError, error:
 219                     if not handle_io_errors:
 220                         raise
 221                     print >>self._stderr, ErrorString(error)
 222                     print >>self._stderr, (u'Unable to open source'
 223                         u" file for reading ('%s'). Exiting." % source_path)
 224                     sys.exit(1)
 225             else:
 226                 self.source = sys.stdin
 227         if not source_path:
 228             try:
 229                 self.source_path = self.source.name
 230             except AttributeError:
 231                 pass
 232
 233     def read(self):
 234         """
 235         Read and decode a single file and return the data (Unicode string).
 236         """
 237         try:
 238             data = self.source.read()
 239         finally:
 240             if self.autoclose:
 241                 self.close()
 242         return self.decode(data)
 243
 244     def readlines(self):
 245         """
 246         Return lines of a single file as list of Unicode strings.
 247         """
 248         try:
 249             lines = self.source.readlines()
 250         finally:
 251             if self.autoclose:
 252                 self.close()
 253         return [self.decode(line) for line in lines]
 254
 255     def close(self):
 256         if self.source is not sys.stdin:
 257             self.source.close()
 258
 259
 260 class FileOutput(Output):
 261
 262     """
 263     Output for single, simple file-like objects.
 264     """
 265
 266     def __init__(self, destination=None, destination_path=None,
 267                  encoding=None, error_handler='strict', autoclose=True,
 268                  handle_io_errors=True):
 269         """
 270         :Parameters:
 271             - `destination`: either a file-like object (which is written
 272               directly) or `None` (which implies `sys.stdout` if no
 273               `destination_path` given).
 274             - `destination_path`: a path to a file, which is opened and then
 275               written.
 276             - `autoclose`: close automatically after write (except when
 277               `sys.stdout` or `sys.stderr` is the destination).
 278         """
 279         Output.__init__(self, destination, destination_path,
 280                         encoding, error_handler)
 281         self.opened = True
 282         self.autoclose = autoclose
 283         self.handle_io_errors = handle_io_errors
 284         self._stderr = ErrorOutput()
 285         if destination is None:
 286             if destination_path:
 287                 self.opened = False
 288             else:
 289                 self.destination = sys.stdout
 290         if not destination_path:
 291             try:
 292                 self.destination_path = self.destination.name
 293             except AttributeError:
 294                 pass
 295
 296     def open(self):
 297         # Specify encoding in Python 3.
 298         # (Do not use binary mode ('wb') as this prevents the
 299         # conversion of newlines to the system specific default.)
 300         if sys.version_info >= (3,0):
 301             kwargs = {'encoding': self.encoding,
 302                       'errors': self.error_handler}
 303         else:
 304             kwargs = {}
 305
 306         try:
 307             self.destination = open(self.destination_path, 'w', **kwargs)
 308         except IOError, error:
 309             if not self.handle_io_errors:
 310                 raise
 311             print >>self._stderr, ErrorString(error)
 312             print >>self._stderr, (u'Unable to open destination file'
 313                 u" for writing ('%s').  Exiting." % self.destination_path)
 314             sys.exit(1)
 315         self.opened = True
 316
 317     def write(self, data):
 318         """Encode `data`, write it to a single file, and return it.
 319
 320         In Python 3, a (unicode) string is returned.
 321         """
 322         if sys.version_info >= (3,0):
 323             output = data # in py3k, write expects a (Unicode) string
 324         else:
 325             output = self.encode(data)
 326         if not self.opened:
 327             self.open()
 328         try:
 329             self.destination.write(output)
 330         finally:
 331             if self.autoclose:
 332                 self.close()
 333         return output
 334
 335     def close(self):
 336         if self.destination not in (sys.stdout, sys.stderr):
 337             self.destination.close()
 338             self.opened = False
 339
 340
 341 class BinaryFileOutput(FileOutput):
 342     """
 343     A version of docutils.io.FileOutput which writes to a binary file.
 344     """
 345     def open(self):
 346         try:
 347             self.destination = open(self.destination_path, 'wb')
 348         except IOError, error:
 349             if not self.handle_io_errors:
 350                 raise
 351             print >>self._stderr, ErrorString(error)
 352             print >>self._stderr, (u'Unable to open destination file'
 353                 u" for writing ('%s').  Exiting." % self.destination_path)
 354             sys.exit(1)
 355         self.opened = True
 356
 357
 358 class StringInput(Input):
 359
 360     """
 361     Direct string input.
 362     """
 363
 364     default_source_path = '<string>'
 365
 366     def read(self):
 367         """Decode and return the source string."""
 368         return self.decode(self.source)
 369
 370
 371 class StringOutput(Output):
 372
 373     """
 374     Direct string output.
 375     """
 376
 377     default_destination_path = '<string>'
 378
 379     def write(self, data):
 380         """Encode `data`, store it in `self.destination`, and return it."""
 381         self.destination = self.encode(data)
 382         return self.destination
 383
 384
 385 class NullInput(Input):
 386
 387     """
 388     Degenerate input: read nothing.
 389     """
 390
 391     default_source_path = 'null input'
 392
 393     def read(self):
 394         """Return a null string."""
 395         return u''
 396
 397
 398 class NullOutput(Output):
 399
 400     """
 401     Degenerate output: write nothing.
 402     """
 403
 404     default_destination_path = 'null output'
 405
 406     def write(self, data):
 407         """Do nothing ([don't even] send data to the bit bucket)."""
 408         pass
 409
 410
 411 class DocTreeInput(Input):
 412
 413     """
 414     Adapter for document tree input.
 415
 416     The document tree must be passed in the ``source`` parameter.
 417     """
 418
 419     default_source_path = 'doctree input'
 420
 421     def read(self):
 422         """Return the document tree."""
 423         return self.source