Guido van Rossum | f06ee5f | 1996-11-27 19:52:01 +0000 | [diff] [blame] | 1 | #! /usr/bin/env python |
Guido van Rossum | 6244867 | 1996-09-17 21:33:15 +0000 | [diff] [blame] | 2 | |
| 3 | """Classes to handle Unix style, MMDF style, and MH style mailboxes.""" |
| 4 | |
| 5 | |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 6 | import rfc822 |
Jack Jansen | 9715779 | 1995-10-23 13:59:53 +0000 | [diff] [blame] | 7 | import os |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 8 | |
| 9 | class _Mailbox: |
Guido van Rossum | 8ca8420 | 1998-03-26 20:56:10 +0000 | [diff] [blame] | 10 | |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 11 | def __init__(self, fp): |
| 12 | self.fp = fp |
| 13 | self.seekp = 0 |
| 14 | |
Guido van Rossum | 44a4d59 | 1996-01-25 18:40:41 +0000 | [diff] [blame] | 15 | def seek(self, pos, whence=0): |
| 16 | if whence==1: # Relative to current position |
| 17 | self.pos = self.pos + pos |
| 18 | if whence==2: # Relative to file's end |
| 19 | self.pos = self.stop + pos |
| 20 | else: # Default - absolute position |
| 21 | self.pos = self.start + pos |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 22 | |
| 23 | def next(self): |
| 24 | while 1: |
| 25 | self.fp.seek(self.seekp) |
| 26 | try: |
| 27 | self._search_start() |
| 28 | except EOFError: |
| 29 | self.seekp = self.fp.tell() |
| 30 | return None |
| 31 | start = self.fp.tell() |
| 32 | self._search_end() |
| 33 | self.seekp = stop = self.fp.tell() |
| 34 | if start <> stop: |
| 35 | break |
| 36 | return rfc822.Message(_Subfile(self.fp, start, stop)) |
| 37 | |
| 38 | class _Subfile: |
Guido van Rossum | 8ca8420 | 1998-03-26 20:56:10 +0000 | [diff] [blame] | 39 | |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 40 | def __init__(self, fp, start, stop): |
| 41 | self.fp = fp |
| 42 | self.start = start |
| 43 | self.stop = stop |
| 44 | self.pos = self.start |
| 45 | |
Guido van Rossum | 2922c6d | 1994-05-06 14:28:19 +0000 | [diff] [blame] | 46 | def read(self, length = None): |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 47 | if self.pos >= self.stop: |
| 48 | return '' |
Guido van Rossum | e50b0a4 | 1998-06-17 18:34:40 +0000 | [diff] [blame] | 49 | remaining = self.stop - self.pos |
| 50 | if length is None or length < 0: |
| 51 | length = remaining |
| 52 | elif length > remaining: |
| 53 | length = remaining |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 54 | self.fp.seek(self.pos) |
| 55 | self.pos = self.pos + length |
| 56 | return self.fp.read(length) |
| 57 | |
Guido van Rossum | 2922c6d | 1994-05-06 14:28:19 +0000 | [diff] [blame] | 58 | def readline(self, length = None): |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 59 | if self.pos >= self.stop: |
| 60 | return '' |
Guido van Rossum | 2922c6d | 1994-05-06 14:28:19 +0000 | [diff] [blame] | 61 | if length is None: |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 62 | length = self.stop - self.pos |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 63 | self.fp.seek(self.pos) |
| 64 | data = self.fp.readline(length) |
| 65 | if len(data) < length: |
| 66 | length = len(data) |
| 67 | self.pos = self.pos + length |
| 68 | return data |
| 69 | |
| 70 | def tell(self): |
| 71 | return self.pos - self.start |
| 72 | |
Jack Jansen | 9715779 | 1995-10-23 13:59:53 +0000 | [diff] [blame] | 73 | def seek(self, pos, whence=0): |
Guido van Rossum | 8ca8420 | 1998-03-26 20:56:10 +0000 | [diff] [blame] | 74 | if whence == 0: |
| 75 | self.pos = self.start + pos |
Jack Jansen | 9715779 | 1995-10-23 13:59:53 +0000 | [diff] [blame] | 76 | elif whence == 1: |
Guido van Rossum | 8ca8420 | 1998-03-26 20:56:10 +0000 | [diff] [blame] | 77 | self.pos = self.pos + pos |
Jack Jansen | 9715779 | 1995-10-23 13:59:53 +0000 | [diff] [blame] | 78 | elif whence == 2: |
Guido van Rossum | 8ca8420 | 1998-03-26 20:56:10 +0000 | [diff] [blame] | 79 | self.pos = self.stop + pos |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 80 | |
| 81 | def close(self): |
| 82 | pass |
| 83 | |
| 84 | class UnixMailbox(_Mailbox): |
Guido van Rossum | 8ca8420 | 1998-03-26 20:56:10 +0000 | [diff] [blame] | 85 | |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 86 | def _search_start(self): |
| 87 | while 1: |
| 88 | line = self.fp.readline() |
| 89 | if not line: |
| 90 | raise EOFError |
Guido van Rossum | fbe63de | 1998-04-03 16:04:05 +0000 | [diff] [blame] | 91 | if line[:5] == 'From ' and self._isrealfromline(line): |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 92 | return |
| 93 | |
| 94 | def _search_end(self): |
| 95 | while 1: |
| 96 | pos = self.fp.tell() |
| 97 | line = self.fp.readline() |
| 98 | if not line: |
| 99 | return |
Guido van Rossum | fbe63de | 1998-04-03 16:04:05 +0000 | [diff] [blame] | 100 | if line[:5] == 'From ' and self._isrealfromline(line): |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 101 | self.fp.seek(pos) |
| 102 | return |
| 103 | |
Guido van Rossum | fbe63de | 1998-04-03 16:04:05 +0000 | [diff] [blame] | 104 | # An overridable mechanism to test for From-line-ness. |
| 105 | # You can either specify a different regular expression |
| 106 | # or define a whole new _isrealfromline() method. |
| 107 | # Note that this only gets called for lines starting with |
| 108 | # the 5 characters "From ". |
| 109 | |
| 110 | _fromlinepattern = r"From \s*[^\s]+\s+\w\w\w\s+\w\w\w\s+\d?\d\s+" \ |
Guido van Rossum | 8ba4036 | 1998-07-02 23:05:32 +0000 | [diff] [blame] | 111 | r"\d?\d:\d\d(:\d\d)?(\s+[^\s]+)?\s+\d\d\d\d\s*$" |
Guido van Rossum | fbe63de | 1998-04-03 16:04:05 +0000 | [diff] [blame] | 112 | _regexp = None |
| 113 | |
| 114 | def _isrealfromline(self, line): |
| 115 | if not self._regexp: |
| 116 | import re |
| 117 | self._regexp = re.compile(self._fromlinepattern) |
| 118 | return self._regexp.match(line) |
| 119 | |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 120 | class MmdfMailbox(_Mailbox): |
Guido van Rossum | 8ca8420 | 1998-03-26 20:56:10 +0000 | [diff] [blame] | 121 | |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 122 | def _search_start(self): |
| 123 | while 1: |
| 124 | line = self.fp.readline() |
| 125 | if not line: |
| 126 | raise EOFError |
| 127 | if line[:5] == '\001\001\001\001\n': |
| 128 | return |
| 129 | |
| 130 | def _search_end(self): |
| 131 | while 1: |
| 132 | pos = self.fp.tell() |
| 133 | line = self.fp.readline() |
| 134 | if not line: |
| 135 | return |
| 136 | if line == '\001\001\001\001\n': |
| 137 | self.fp.seek(pos) |
| 138 | return |
| 139 | |
Jack Jansen | 9715779 | 1995-10-23 13:59:53 +0000 | [diff] [blame] | 140 | class MHMailbox: |
Jack Jansen | 9715779 | 1995-10-23 13:59:53 +0000 | [diff] [blame] | 141 | |
Guido van Rossum | 8ca8420 | 1998-03-26 20:56:10 +0000 | [diff] [blame] | 142 | def __init__(self, dirname): |
| 143 | import re |
| 144 | pat = re.compile('^[0-9][0-9]*$') |
| 145 | self.dirname = dirname |
| 146 | files = os.listdir(self.dirname) |
| 147 | self.boxes = [] |
| 148 | for f in files: |
| 149 | if pat.match(f): |
| 150 | self.boxes.append(f) |
| 151 | |
| 152 | def next(self): |
| 153 | if not self.boxes: |
| 154 | return None |
| 155 | fn = self.boxes[0] |
| 156 | del self.boxes[0] |
| 157 | fp = open(os.path.join(self.dirname, fn)) |
| 158 | return rfc822.Message(fp) |
Jack Jansen | 9715779 | 1995-10-23 13:59:53 +0000 | [diff] [blame] | 159 | |
| 160 | |
Guido van Rossum | fdf58fe | 1997-05-15 14:33:09 +0000 | [diff] [blame] | 161 | class BabylMailbox(_Mailbox): |
Guido van Rossum | fdf58fe | 1997-05-15 14:33:09 +0000 | [diff] [blame] | 162 | |
Guido van Rossum | 8ca8420 | 1998-03-26 20:56:10 +0000 | [diff] [blame] | 163 | def _search_start(self): |
| 164 | while 1: |
| 165 | line = self.fp.readline() |
| 166 | if not line: |
| 167 | raise EOFError |
| 168 | if line == '*** EOOH ***\n': |
| 169 | return |
| 170 | |
| 171 | def _search_end(self): |
| 172 | while 1: |
| 173 | pos = self.fp.tell() |
| 174 | line = self.fp.readline() |
| 175 | if not line: |
| 176 | return |
| 177 | if line == '\037\014\n': |
| 178 | self.fp.seek(pos) |
| 179 | return |
Guido van Rossum | fdf58fe | 1997-05-15 14:33:09 +0000 | [diff] [blame] | 180 | |
| 181 | |
Guido van Rossum | 6244867 | 1996-09-17 21:33:15 +0000 | [diff] [blame] | 182 | def _test(): |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 183 | import time |
| 184 | import sys |
| 185 | import string |
Guido van Rossum | 6244867 | 1996-09-17 21:33:15 +0000 | [diff] [blame] | 186 | import os |
Jack Jansen | 9715779 | 1995-10-23 13:59:53 +0000 | [diff] [blame] | 187 | |
Guido van Rossum | 6244867 | 1996-09-17 21:33:15 +0000 | [diff] [blame] | 188 | args = sys.argv[1:] |
| 189 | if not args: |
Guido van Rossum | 6dd8783 | 1997-05-08 23:11:52 +0000 | [diff] [blame] | 190 | for key in 'MAIL', 'LOGNAME', 'USER': |
| 191 | if os.environ.has_key(key): |
| 192 | mbox = os.environ[key] |
| 193 | break |
Guido van Rossum | 6244867 | 1996-09-17 21:33:15 +0000 | [diff] [blame] | 194 | else: |
Guido van Rossum | 6dd8783 | 1997-05-08 23:11:52 +0000 | [diff] [blame] | 195 | print "$MAIL, $LOGNAME nor $USER set -- who are you?" |
Guido van Rossum | 6244867 | 1996-09-17 21:33:15 +0000 | [diff] [blame] | 196 | return |
| 197 | else: |
| 198 | mbox = args[0] |
| 199 | if mbox[:1] == '+': |
| 200 | mbox = os.environ['HOME'] + '/Mail/' + mbox[1:] |
| 201 | elif not '/' in mbox: |
| 202 | mbox = '/usr/mail/' + mbox |
| 203 | if os.path.isdir(mbox): |
| 204 | mb = MHMailbox(mbox) |
| 205 | else: |
| 206 | fp = open(mbox, 'r') |
| 207 | mb = UnixMailbox(fp) |
Jack Jansen | 9715779 | 1995-10-23 13:59:53 +0000 | [diff] [blame] | 208 | |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 209 | msgs = [] |
| 210 | while 1: |
| 211 | msg = mb.next() |
Guido van Rossum | fbe63de | 1998-04-03 16:04:05 +0000 | [diff] [blame] | 212 | if msg is None: |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 213 | break |
| 214 | msgs.append(msg) |
Guido van Rossum | 9694fca | 1997-10-22 21:00:49 +0000 | [diff] [blame] | 215 | msg.fp = None |
Guido van Rossum | 6244867 | 1996-09-17 21:33:15 +0000 | [diff] [blame] | 216 | if len(args) > 1: |
| 217 | num = string.atoi(args[1]) |
Guido van Rossum | c7b6882 | 1994-04-28 09:53:33 +0000 | [diff] [blame] | 218 | print 'Message %d body:'%num |
| 219 | msg = msgs[num-1] |
| 220 | msg.rewindbody() |
| 221 | sys.stdout.write(msg.fp.read()) |
Guido van Rossum | 6244867 | 1996-09-17 21:33:15 +0000 | [diff] [blame] | 222 | else: |
| 223 | print 'Mailbox',mbox,'has',len(msgs),'messages:' |
| 224 | for msg in msgs: |
Guido van Rossum | fbe63de | 1998-04-03 16:04:05 +0000 | [diff] [blame] | 225 | f = msg.getheader('from') or "" |
| 226 | s = msg.getheader('subject') or "" |
| 227 | d = msg.getheader('date') or "" |
Guido van Rossum | 6244867 | 1996-09-17 21:33:15 +0000 | [diff] [blame] | 228 | print '%20.20s %18.18s %-30.30s'%(f, d[5:], s) |
| 229 | |
| 230 | |
| 231 | if __name__ == '__main__': |
| 232 | _test() |