Lib/email/generator.py

   1 # Copyright (C) 2001-2006 Python Software Foundation
   2 # Author: Barry Warsaw
   3 # Contact: email-sig@python.org
   4
   5 """Classes to generate plain text from a message object tree."""
   6
   7 __all__ = ['Generator', 'DecodedGenerator']
   8
   9 import re
  10 import sys
  11 import time
  12 import random
  13 import warnings
  14
  15 from cStringIO import StringIO
  16 from email.header import Header
  17
  18 UNDERSCORE = '_'
  19 NL = '\n'
  20
  21 fcre = re.compile(r'^From ', re.MULTILINE)
  22
  23 def _is8bitstring(s):
  24     if isinstance(s, str):
  25         try:
  26             unicode(s, 'us-ascii')
  27         except UnicodeError:
  28             return True
  29     return False
  30
  31
  32 \f
  33 class Generator:
  34     """Generates output from a Message object tree.
  35
  36     This basic generator writes the message to the given file object as plain
  37     text.
  38     """
  39     #
  40     # Public interface
  41     #
  42
  43     def __init__(self, outfp, mangle_from_=True, maxheaderlen=78):
  44         """Create the generator for message flattening.
  45
  46         outfp is the output file-like object for writing the message to.  It
  47         must have a write() method.
  48
  49         Optional mangle_from_ is a flag that, when True (the default), escapes
  50         From_ lines in the body of the message by putting a `>' in front of
  51         them.
  52
  53         Optional maxheaderlen specifies the longest length for a non-continued
  54         header.  When a header line is longer (in characters, with tabs
  55         expanded to 8 spaces) than maxheaderlen, the header will split as
  56         defined in the Header class.  Set maxheaderlen to zero to disable
  57         header wrapping.  The default is 78, as recommended (but not required)
  58         by RFC 2822.
  59         """
  60         self._fp = outfp
  61         self._mangle_from_ = mangle_from_
  62         self._maxheaderlen = maxheaderlen
  63
  64     def write(self, s):
  65         # Just delegate to the file object
  66         self._fp.write(s)
  67
  68     def flatten(self, msg, unixfrom=False):
  69         """Print the message object tree rooted at msg to the output file
  70         specified when the Generator instance was created.
  71
  72         unixfrom is a flag that forces the printing of a Unix From_ delimiter
  73         before the first object in the message tree.  If the original message
  74         has no From_ delimiter, a `standard' one is crafted.  By default, this
  75         is False to inhibit the printing of any From_ delimiter.
  76
  77         Note that for subobjects, no From_ line is printed.
  78         """
  79         if unixfrom:
  80             ufrom = msg.get_unixfrom()
  81             if not ufrom:
  82                 ufrom = 'From nobody ' + time.ctime(time.time())
  83             print >> self._fp, ufrom
  84         self._write(msg)
  85
  86     def clone(self, fp):
  87         """Clone this generator with the exact same options."""
  88         return self.__class__(fp, self._mangle_from_, self._maxheaderlen)
  89
  90     #
  91     # Protected interface - undocumented ;/
  92     #
  93
  94     def _write(self, msg):
  95         # We can't write the headers yet because of the following scenario:
  96         # say a multipart message includes the boundary string somewhere in
  97         # its body.  We'd have to calculate the new boundary /before/ we write
  98         # the headers so that we can write the correct Content-Type:
  99         # parameter.
 100         #
 101         # The way we do this, so as to make the _handle_*() methods simpler,
 102         # is to cache any subpart writes into a StringIO.  The we write the
 103         # headers and the StringIO contents.  That way, subpart handlers can
 104         # Do The Right Thing, and can still modify the Content-Type: header if
 105         # necessary.
 106         oldfp = self._fp
 107         try:
 108             self._fp = sfp = StringIO()
 109             self._dispatch(msg)
 110         finally:
 111             self._fp = oldfp
 112         # Write the headers.  First we see if the message object wants to
 113         # handle that itself.  If not, we'll do it generically.
 114         meth = getattr(msg, '_write_headers', None)
 115         if meth is None:
 116             self._write_headers(msg)
 117         else:
 118             meth(self)
 119         self._fp.write(sfp.getvalue())
 120
 121     def _dispatch(self, msg):
 122         # Get the Content-Type: for the message, then try to dispatch to
 123         # self._handle_<maintype>_<subtype>().  If there's no handler for the
 124         # full MIME type, then dispatch to self._handle_<maintype>().  If
 125         # that's missing too, then dispatch to self._writeBody().
 126         main = msg.get_content_maintype()
 127         sub = msg.get_content_subtype()
 128         specific = UNDERSCORE.join((main, sub)).replace('-', '_')
 129         meth = getattr(self, '_handle_' + specific, None)
 130         if meth is None:
 131             generic = main.replace('-', '_')
 132             meth = getattr(self, '_handle_' + generic, None)
 133             if meth is None:
 134                 meth = self._writeBody
 135         meth(msg)
 136
 137     #
 138     # Default handlers
 139     #
 140
 141     def _write_headers(self, msg):
 142         for h, v in msg.items():
 143             print >> self._fp, '%s:' % h,
 144             if self._maxheaderlen == 0:
 145                 # Explicit no-wrapping
 146                 print >> self._fp, v
 147             elif isinstance(v, Header):
 148                 # Header instances know what to do
 149                 print >> self._fp, v.encode()
 150             elif _is8bitstring(v):
 151                 # If we have raw 8bit data in a byte string, we have no idea
 152                 # what the encoding is.  There is no safe way to split this
 153                 # string.  If it's ascii-subset, then we could do a normal
 154                 # ascii split, but if it's multibyte then we could break the
 155                 # string.  There's no way to know so the least harm seems to
 156                 # be to not split the string and risk it being too long.
 157                 print >> self._fp, v
 158             else:
 159                 # Header's got lots of smarts, so use it.
 160                 print >> self._fp, Header(
 161                     v, maxlinelen=self._maxheaderlen,
 162                     header_name=h, continuation_ws='\t').encode()
 163         # A blank line always separates headers from body
 164         print >> self._fp
 165
 166     #
 167     # Handlers for writing types and subtypes
 168     #
 169
 170     def _handle_text(self, msg):
 171         payload = msg.get_payload()
 172         if payload is None:
 173             return
 174         if not isinstance(payload, basestring):
 175             raise TypeError('string payload expected: %s' % type(payload))
 176         if self._mangle_from_:
 177             payload = fcre.sub('>From ', payload)
 178         self._fp.write(payload)
 179
 180     # Default body handler
 181     _writeBody = _handle_text
 182
 183     def _handle_multipart(self, msg):
 184         # The trick here is to write out each part separately, merge them all
 185         # together, and then make sure that the boundary we've chosen isn't
 186         # present in the payload.
 187         msgtexts = []
 188         subparts = msg.get_payload()
 189         if subparts is None:
 190             subparts = []
 191         elif isinstance(subparts, basestring):
 192             # e.g. a non-strict parse of a message with no starting boundary.
 193             self._fp.write(subparts)
 194             return
 195         elif not isinstance(subparts, list):
 196             # Scalar payload
 197             subparts = [subparts]
 198         for part in subparts:
 199             s = StringIO()
 200             g = self.clone(s)
 201             g.flatten(part, unixfrom=False)
 202             msgtexts.append(s.getvalue())
 203         # Now make sure the boundary we've selected doesn't appear in any of
 204         # the message texts.
 205         alltext = NL.join(msgtexts)
 206         # BAW: What about boundaries that are wrapped in double-quotes?
 207         boundary = msg.get_boundary(failobj=_make_boundary(alltext))
 208         # If we had to calculate a new boundary because the body text
 209         # contained that string, set the new boundary.  We don't do it
 210         # unconditionally because, while set_boundary() preserves order, it
 211         # doesn't preserve newlines/continuations in headers.  This is no big
 212         # deal in practice, but turns out to be inconvenient for the unittest
 213         # suite.
 214         if msg.get_boundary() <> boundary:
 215             msg.set_boundary(boundary)
 216         # If there's a preamble, write it out, with a trailing CRLF
 217         if msg.preamble is not None:
 218             print >> self._fp, msg.preamble
 219         # dash-boundary transport-padding CRLF
 220         print >> self._fp, '--' + boundary
 221         # body-part
 222         if msgtexts:
 223             self._fp.write(msgtexts.pop(0))
 224         # *encapsulation
 225         # --> delimiter transport-padding
 226         # --> CRLF body-part
 227         for body_part in msgtexts:
 228             # delimiter transport-padding CRLF
 229             print >> self._fp, '\n--' + boundary
 230             # body-part
 231             self._fp.write(body_part)
 232         # close-delimiter transport-padding
 233         self._fp.write('\n--' + boundary + '--')
 234         if msg.epilogue is not None:
 235             print >> self._fp
 236             self._fp.write(msg.epilogue)
 237
 238     def _handle_message_delivery_status(self, msg):
 239         # We can't just write the headers directly to self's file object
 240         # because this will leave an extra newline between the last header
 241         # block and the boundary.  Sigh.
 242         blocks = []
 243         for part in msg.get_payload():
 244             s = StringIO()
 245             g = self.clone(s)
 246             g.flatten(part, unixfrom=False)
 247             text = s.getvalue()
 248             lines = text.split('\n')
 249             # Strip off the unnecessary trailing empty line
 250             if lines and lines[-1] == '':
 251                 blocks.append(NL.join(lines[:-1]))
 252             else:
 253                 blocks.append(text)
 254         # Now join all the blocks with an empty line.  This has the lovely
 255         # effect of separating each block with an empty line, but not adding
 256         # an extra one after the last one.
 257         self._fp.write(NL.join(blocks))
 258
 259     def _handle_message(self, msg):
 260         s = StringIO()
 261         g = self.clone(s)
 262         # The payload of a message/rfc822 part should be a multipart sequence
 263         # of length 1.  The zeroth element of the list should be the Message
 264         # object for the subpart.  Extract that object, stringify it, and
 265         # write it out.
 266         g.flatten(msg.get_payload(0), unixfrom=False)
 267         self._fp.write(s.getvalue())
 268
 269
 270 \f
 271 _FMT = '[Non-text (%(type)s) part of message omitted, filename %(filename)s]'
 272
 273 class DecodedGenerator(Generator):
 274     """Generator a text representation of a message.
 275
 276     Like the Generator base class, except that non-text parts are substituted
 277     with a format string representing the part.
 278     """
 279     def __init__(self, outfp, mangle_from_=True, maxheaderlen=78, fmt=None):
 280         """Like Generator.__init__() except that an additional optional
 281         argument is allowed.
 282
 283         Walks through all subparts of a message.  If the subpart is of main
 284         type `text', then it prints the decoded payload of the subpart.
 285
 286         Otherwise, fmt is a format string that is used instead of the message
 287         payload.  fmt is expanded with the following keywords (in
 288         %(keyword)s format):
 289
 290         type       : Full MIME type of the non-text part
 291         maintype   : Main MIME type of the non-text part
 292         subtype    : Sub-MIME type of the non-text part
 293         filename   : Filename of the non-text part
 294         description: Description associated with the non-text part
 295         encoding   : Content transfer encoding of the non-text part
 296
 297         The default value for fmt is None, meaning
 298
 299         [Non-text (%(type)s) part of message omitted, filename %(filename)s]
 300         """
 301         Generator.__init__(self, outfp, mangle_from_, maxheaderlen)
 302         if fmt is None:
 303             self._fmt = _FMT
 304         else:
 305             self._fmt = fmt
 306
 307     def _dispatch(self, msg):
 308         for part in msg.walk():
 309             maintype = part.get_content_maintype()
 310             if maintype == 'text':
 311                 print >> self, part.get_payload(decode=True)
 312             elif maintype == 'multipart':
 313                 # Just skip this
 314                 pass
 315             else:
 316                 print >> self, self._fmt % {
 317                     'type'       : part.get_content_type(),
 318                     'maintype'   : part.get_content_maintype(),
 319                     'subtype'    : part.get_content_subtype(),
 320                     'filename'   : part.get_filename('[no filename]'),
 321                     'description': part.get('Content-Description',
 322                                             '[no description]'),
 323                     'encoding'   : part.get('Content-Transfer-Encoding',
 324                                             '[no encoding]'),
 325                     }
 326
 327
 328 \f
 329 # Helper
 330 _width = len(repr(sys.maxint-1))
 331 _fmt = '%%0%dd' % _width
 332
 333 def _make_boundary(text=None):
 334     # Craft a random boundary.  If text is given, ensure that the chosen
 335     # boundary doesn't appear in the text.
 336     token = random.randrange(sys.maxint)
 337     boundary = ('=' * 15) + (_fmt % token) + '=='
 338     if text is None:
 339         return boundary
 340     b = boundary
 341     counter = 0
 342     while True:
 343         cre = re.compile('^--' + re.escape(b) + '(--)?$', re.MULTILINE)
 344         if not cre.search(text):
 345             break
 346         b = boundary + '.' + str(counter)
 347         counter += 1
 348     return b