blob: 258b657445f888163beba3066c19ed4f05bce84a [file] [log] [blame]
Guido van Rossumf06ee5f1996-11-27 19:52:01 +00001#! /usr/bin/env python
Guido van Rossum62448671996-09-17 21:33:15 +00002
3"""Classes to handle Unix style, MMDF style, and MH style mailboxes."""
4
5
Guido van Rossumc7b68821994-04-28 09:53:33 +00006import rfc822
Jack Jansen97157791995-10-23 13:59:53 +00007import os
Guido van Rossumc7b68821994-04-28 09:53:33 +00008
Barry Warsawffd05ee2002-03-01 22:39:14 +00009__all__ = ["UnixMailbox","MmdfMailbox","MHMailbox","Maildir","BabylMailbox",
10 "PortableUnixMailbox"]
Skip Montanaro17ab1232001-01-24 06:27:27 +000011
Guido van Rossumc7b68821994-04-28 09:53:33 +000012class _Mailbox:
Guido van Rossum4bf12542002-09-12 05:08:00 +000013
Barry Warsaw81ad67c2001-01-31 22:13:15 +000014 def __init__(self, fp, factory=rfc822.Message):
Fred Drakedbbf76b2000-07-09 16:44:26 +000015 self.fp = fp
16 self.seekp = 0
Barry Warsaw81ad67c2001-01-31 22:13:15 +000017 self.factory = factory
Guido van Rossum8ca84201998-03-26 20:56:10 +000018
Fred Drake72987a42001-05-02 20:20:53 +000019 def __iter__(self):
Guido van Rossum93a696f2001-09-13 01:29:13 +000020 return iter(self.next, None)
Fred Drake72987a42001-05-02 20:20:53 +000021
Fred Drakedbbf76b2000-07-09 16:44:26 +000022 def next(self):
23 while 1:
24 self.fp.seek(self.seekp)
25 try:
26 self._search_start()
27 except EOFError:
28 self.seekp = self.fp.tell()
29 return None
30 start = self.fp.tell()
31 self._search_end()
32 self.seekp = stop = self.fp.tell()
Fred Drake8152d322000-12-12 23:20:45 +000033 if start != stop:
Fred Drakedbbf76b2000-07-09 16:44:26 +000034 break
Barry Warsaw81ad67c2001-01-31 22:13:15 +000035 return self.factory(_Subfile(self.fp, start, stop))
Guido van Rossumc7b68821994-04-28 09:53:33 +000036
Guido van Rossumc7b68821994-04-28 09:53:33 +000037
38class _Subfile:
Guido van Rossum4bf12542002-09-12 05:08:00 +000039
Fred Drakedbbf76b2000-07-09 16:44:26 +000040 def __init__(self, fp, start, stop):
41 self.fp = fp
42 self.start = start
43 self.stop = stop
44 self.pos = self.start
Guido van Rossum8ca84201998-03-26 20:56:10 +000045
Fred Drakedbbf76b2000-07-09 16:44:26 +000046 def read(self, length = None):
47 if self.pos >= self.stop:
48 return ''
49 remaining = self.stop - self.pos
50 if length is None or length < 0:
51 length = remaining
52 elif length > remaining:
53 length = remaining
54 self.fp.seek(self.pos)
55 data = self.fp.read(length)
56 self.pos = self.fp.tell()
57 return data
Guido van Rossumc7b68821994-04-28 09:53:33 +000058
Fred Drakedbbf76b2000-07-09 16:44:26 +000059 def readline(self, length = None):
60 if self.pos >= self.stop:
61 return ''
62 if length is None:
63 length = self.stop - self.pos
64 self.fp.seek(self.pos)
65 data = self.fp.readline(length)
66 self.pos = self.fp.tell()
67 return data
Guido van Rossumc7b68821994-04-28 09:53:33 +000068
Fred Drakedbbf76b2000-07-09 16:44:26 +000069 def readlines(self, sizehint = -1):
70 lines = []
71 while 1:
72 line = self.readline()
73 if not line:
74 break
75 lines.append(line)
76 if sizehint >= 0:
77 sizehint = sizehint - len(line)
78 if sizehint <= 0:
79 break
80 return lines
Guido van Rossumc7b68821994-04-28 09:53:33 +000081
Fred Drakedbbf76b2000-07-09 16:44:26 +000082 def tell(self):
83 return self.pos - self.start
Guido van Rossume256a0f1999-03-24 16:20:45 +000084
Fred Drakedbbf76b2000-07-09 16:44:26 +000085 def seek(self, pos, whence=0):
86 if whence == 0:
87 self.pos = self.start + pos
88 elif whence == 1:
89 self.pos = self.pos + pos
90 elif whence == 2:
91 self.pos = self.stop + pos
Guido van Rossumc7b68821994-04-28 09:53:33 +000092
Fred Drakedbbf76b2000-07-09 16:44:26 +000093 def close(self):
94 del self.fp
Guido van Rossumc7b68821994-04-28 09:53:33 +000095
Guido van Rossumc7b68821994-04-28 09:53:33 +000096
Barry Warsawffd05ee2002-03-01 22:39:14 +000097# Recommended to use PortableUnixMailbox instead!
Guido van Rossumc7b68821994-04-28 09:53:33 +000098class UnixMailbox(_Mailbox):
Guido van Rossum4bf12542002-09-12 05:08:00 +000099
Fred Drakedbbf76b2000-07-09 16:44:26 +0000100 def _search_start(self):
101 while 1:
102 pos = self.fp.tell()
103 line = self.fp.readline()
104 if not line:
105 raise EOFError
106 if line[:5] == 'From ' and self._isrealfromline(line):
107 self.fp.seek(pos)
108 return
Guido van Rossum8ca84201998-03-26 20:56:10 +0000109
Fred Drakedbbf76b2000-07-09 16:44:26 +0000110 def _search_end(self):
111 self.fp.readline() # Throw away header line
112 while 1:
113 pos = self.fp.tell()
114 line = self.fp.readline()
115 if not line:
116 return
117 if line[:5] == 'From ' and self._isrealfromline(line):
118 self.fp.seek(pos)
119 return
Guido van Rossumc7b68821994-04-28 09:53:33 +0000120
Barry Warsaw81ad67c2001-01-31 22:13:15 +0000121 # An overridable mechanism to test for From-line-ness. You can either
122 # specify a different regular expression or define a whole new
123 # _isrealfromline() method. Note that this only gets called for lines
124 # starting with the 5 characters "From ".
125 #
126 # BAW: According to
127 #http://home.netscape.com/eng/mozilla/2.0/relnotes/demo/content-length.html
128 # the only portable, reliable way to find message delimiters in a BSD (i.e
129 # Unix mailbox) style folder is to search for "\n\nFrom .*\n", or at the
130 # beginning of the file, "^From .*\n". While _fromlinepattern below seems
131 # like a good idea, in practice, there are too many variations for more
132 # strict parsing of the line to be completely accurate.
133 #
134 # _strict_isrealfromline() is the old version which tries to do stricter
135 # parsing of the From_ line. _portable_isrealfromline() simply returns
136 # true, since it's never called if the line doesn't already start with
137 # "From ".
138 #
139 # This algorithm, and the way it interacts with _search_start() and
140 # _search_end() may not be completely correct, because it doesn't check
141 # that the two characters preceding "From " are \n\n or the beginning of
142 # the file. Fixing this would require a more extensive rewrite than is
Barry Warsawda5628f2002-08-26 16:44:56 +0000143 # necessary. For convenience, we've added a PortableUnixMailbox class
144 # which uses the more lenient _fromlinepattern regular expression.
Guido van Rossumc7b68821994-04-28 09:53:33 +0000145
Fred Drakedbbf76b2000-07-09 16:44:26 +0000146 _fromlinepattern = r"From \s*[^\s]+\s+\w\w\w\s+\w\w\w\s+\d?\d\s+" \
147 r"\d?\d:\d\d(:\d\d)?(\s+[^\s]+)?\s+\d\d\d\d\s*$"
148 _regexp = None
Guido van Rossumfbe63de1998-04-03 16:04:05 +0000149
Barry Warsaw81ad67c2001-01-31 22:13:15 +0000150 def _strict_isrealfromline(self, line):
Fred Drakedbbf76b2000-07-09 16:44:26 +0000151 if not self._regexp:
152 import re
153 self._regexp = re.compile(self._fromlinepattern)
154 return self._regexp.match(line)
Guido van Rossumfbe63de1998-04-03 16:04:05 +0000155
Barry Warsaw81ad67c2001-01-31 22:13:15 +0000156 def _portable_isrealfromline(self, line):
Tim Petersbc0e9102002-04-04 22:55:58 +0000157 return True
Barry Warsaw81ad67c2001-01-31 22:13:15 +0000158
159 _isrealfromline = _strict_isrealfromline
160
161
162class PortableUnixMailbox(UnixMailbox):
163 _isrealfromline = UnixMailbox._portable_isrealfromline
164
Guido van Rossumfbe63de1998-04-03 16:04:05 +0000165
Guido van Rossumc7b68821994-04-28 09:53:33 +0000166class MmdfMailbox(_Mailbox):
Guido van Rossum4bf12542002-09-12 05:08:00 +0000167
Fred Drakedbbf76b2000-07-09 16:44:26 +0000168 def _search_start(self):
169 while 1:
170 line = self.fp.readline()
171 if not line:
172 raise EOFError
173 if line[:5] == '\001\001\001\001\n':
174 return
Guido van Rossum8ca84201998-03-26 20:56:10 +0000175
Fred Drakedbbf76b2000-07-09 16:44:26 +0000176 def _search_end(self):
177 while 1:
178 pos = self.fp.tell()
179 line = self.fp.readline()
180 if not line:
181 return
182 if line == '\001\001\001\001\n':
183 self.fp.seek(pos)
184 return
Guido van Rossumc7b68821994-04-28 09:53:33 +0000185
Guido van Rossumc7b68821994-04-28 09:53:33 +0000186
Jack Jansen97157791995-10-23 13:59:53 +0000187class MHMailbox:
Guido van Rossum4bf12542002-09-12 05:08:00 +0000188
Barry Warsaw81ad67c2001-01-31 22:13:15 +0000189 def __init__(self, dirname, factory=rfc822.Message):
Fred Drakedbbf76b2000-07-09 16:44:26 +0000190 import re
Guido van Rossum0707fea2000-08-10 03:05:26 +0000191 pat = re.compile('^[1-9][0-9]*$')
Fred Drakedbbf76b2000-07-09 16:44:26 +0000192 self.dirname = dirname
Sjoerd Mullenderd2653a92000-08-11 07:48:36 +0000193 # the three following lines could be combined into:
194 # list = map(long, filter(pat.match, os.listdir(self.dirname)))
195 list = os.listdir(self.dirname)
196 list = filter(pat.match, list)
Guido van Rossum0707fea2000-08-10 03:05:26 +0000197 list = map(long, list)
198 list.sort()
199 # This only works in Python 1.6 or later;
200 # before that str() added 'L':
201 self.boxes = map(str, list)
Raymond Hettingerb5ba8d72004-02-07 02:16:24 +0000202 self.boxes.reverse()
Barry Warsaw81ad67c2001-01-31 22:13:15 +0000203 self.factory = factory
Jack Jansen97157791995-10-23 13:59:53 +0000204
Fred Drake72987a42001-05-02 20:20:53 +0000205 def __iter__(self):
Guido van Rossum93a696f2001-09-13 01:29:13 +0000206 return iter(self.next, None)
Fred Drake72987a42001-05-02 20:20:53 +0000207
Fred Drakedbbf76b2000-07-09 16:44:26 +0000208 def next(self):
209 if not self.boxes:
210 return None
Raymond Hettingerb5ba8d72004-02-07 02:16:24 +0000211 fn = self.boxes.pop()
Fred Drakedbbf76b2000-07-09 16:44:26 +0000212 fp = open(os.path.join(self.dirname, fn))
Guido van Rossum4bf12542002-09-12 05:08:00 +0000213 msg = self.factory(fp)
214 try:
215 msg._mh_msgno = fn
216 except (AttributeError, TypeError):
217 pass
218 return msg
Guido van Rossum8ca84201998-03-26 20:56:10 +0000219
Guido van Rossum9a4d6371998-12-23 22:05:42 +0000220
221class Maildir:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000222 # Qmail directory mailbox
Guido van Rossum9a4d6371998-12-23 22:05:42 +0000223
Barry Warsaw81ad67c2001-01-31 22:13:15 +0000224 def __init__(self, dirname, factory=rfc822.Message):
Fred Drakedbbf76b2000-07-09 16:44:26 +0000225 self.dirname = dirname
Barry Warsaw81ad67c2001-01-31 22:13:15 +0000226 self.factory = factory
Guido van Rossum9a4d6371998-12-23 22:05:42 +0000227
Fred Drakedbbf76b2000-07-09 16:44:26 +0000228 # check for new mail
229 newdir = os.path.join(self.dirname, 'new')
Fred Draked9a8dec2000-09-22 18:41:50 +0000230 boxes = [os.path.join(newdir, f)
231 for f in os.listdir(newdir) if f[0] != '.']
Guido van Rossum9a4d6371998-12-23 22:05:42 +0000232
Fred Drakedbbf76b2000-07-09 16:44:26 +0000233 # Now check for current mail in this maildir
234 curdir = os.path.join(self.dirname, 'cur')
Fred Draked9a8dec2000-09-22 18:41:50 +0000235 boxes += [os.path.join(curdir, f)
236 for f in os.listdir(curdir) if f[0] != '.']
Raymond Hettingerb5ba8d72004-02-07 02:16:24 +0000237 boxes.reverse()
Fred Drakede3518e2000-10-23 13:37:01 +0000238 self.boxes = boxes
239
Fred Drake72987a42001-05-02 20:20:53 +0000240 def __iter__(self):
Guido van Rossum93a696f2001-09-13 01:29:13 +0000241 return iter(self.next, None)
Fred Drake72987a42001-05-02 20:20:53 +0000242
Fred Drakedbbf76b2000-07-09 16:44:26 +0000243 def next(self):
244 if not self.boxes:
245 return None
Raymond Hettingerb5ba8d72004-02-07 02:16:24 +0000246 fn = self.boxes.pop()
Fred Drakee108a022000-09-14 14:44:43 +0000247 fp = open(fn)
Barry Warsaw81ad67c2001-01-31 22:13:15 +0000248 return self.factory(fp)
Guido van Rossum9a4d6371998-12-23 22:05:42 +0000249
Guido van Rossum9a4d6371998-12-23 22:05:42 +0000250
Guido van Rossumfdf58fe1997-05-15 14:33:09 +0000251class BabylMailbox(_Mailbox):
Guido van Rossum4bf12542002-09-12 05:08:00 +0000252
Fred Drakedbbf76b2000-07-09 16:44:26 +0000253 def _search_start(self):
254 while 1:
255 line = self.fp.readline()
256 if not line:
257 raise EOFError
258 if line == '*** EOOH ***\n':
259 return
Guido van Rossumfdf58fe1997-05-15 14:33:09 +0000260
Fred Drakedbbf76b2000-07-09 16:44:26 +0000261 def _search_end(self):
262 while 1:
263 pos = self.fp.tell()
264 line = self.fp.readline()
265 if not line:
266 return
267 if line == '\037\014\n':
268 self.fp.seek(pos)
269 return
Guido van Rossumfdf58fe1997-05-15 14:33:09 +0000270
271
Guido van Rossum62448671996-09-17 21:33:15 +0000272def _test():
Fred Drakedbbf76b2000-07-09 16:44:26 +0000273 import sys
Jack Jansen97157791995-10-23 13:59:53 +0000274
Fred Drakedbbf76b2000-07-09 16:44:26 +0000275 args = sys.argv[1:]
276 if not args:
277 for key in 'MAILDIR', 'MAIL', 'LOGNAME', 'USER':
Raymond Hettinger54f02222002-06-01 14:18:47 +0000278 if key in os.environ:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000279 mbox = os.environ[key]
280 break
Fred Drake13a2c272000-02-10 17:17:14 +0000281 else:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000282 print "$MAIL, $LOGNAME nor $USER set -- who are you?"
283 return
284 else:
285 mbox = args[0]
286 if mbox[:1] == '+':
287 mbox = os.environ['HOME'] + '/Mail/' + mbox[1:]
288 elif not '/' in mbox:
Neil Schemenauer03f3ee62002-03-24 01:38:38 +0000289 if os.path.isfile('/var/mail/' + mbox):
290 mbox = '/var/mail/' + mbox
291 else:
292 mbox = '/usr/mail/' + mbox
Fred Drakedbbf76b2000-07-09 16:44:26 +0000293 if os.path.isdir(mbox):
294 if os.path.isdir(os.path.join(mbox, 'cur')):
295 mb = Maildir(mbox)
Fred Drake13a2c272000-02-10 17:17:14 +0000296 else:
Fred Drakedbbf76b2000-07-09 16:44:26 +0000297 mb = MHMailbox(mbox)
298 else:
299 fp = open(mbox, 'r')
Barry Warsawffd05ee2002-03-01 22:39:14 +0000300 mb = PortableUnixMailbox(fp)
Fred Drakedbbf76b2000-07-09 16:44:26 +0000301
302 msgs = []
303 while 1:
304 msg = mb.next()
305 if msg is None:
306 break
307 msgs.append(msg)
308 if len(args) <= 1:
309 msg.fp = None
310 if len(args) > 1:
Fred Drakede3518e2000-10-23 13:37:01 +0000311 num = int(args[1])
Fred Drakedbbf76b2000-07-09 16:44:26 +0000312 print 'Message %d body:'%num
313 msg = msgs[num-1]
314 msg.rewindbody()
315 sys.stdout.write(msg.fp.read())
316 else:
317 print 'Mailbox',mbox,'has',len(msgs),'messages:'
318 for msg in msgs:
319 f = msg.getheader('from') or ""
320 s = msg.getheader('subject') or ""
321 d = msg.getheader('date') or ""
Fred Drakecc4adf22000-09-30 23:59:04 +0000322 print '-%20.20s %20.20s %-30.30s'%(f, d[5:], s)
Guido van Rossum62448671996-09-17 21:33:15 +0000323
324
325if __name__ == '__main__':
Fred Drakedbbf76b2000-07-09 16:44:26 +0000326 _test()