1#! /usr/bin/env python
2
3"""Read/write support for Maildir, mbox, MH, Babyl, and MMDF mailboxes."""
4
5# Notes for authors of new mailbox subclasses:
6#
7# Remember to fsync() changes to disk before closing a modified file
8# or returning from a flush() method.  See functions _sync_flush() and
9# _sync_close().
10
11import sys
12import os
13import time
14import calendar
15import socket
16import errno
17import copy
18import email
19import email.message
20import email.generator
21import StringIO
22try:
23    if sys.platform == 'os2emx':
24        # OS/2 EMX fcntl() not adequate
25        raise ImportError
26    import fcntl
27except ImportError:
28    fcntl = None
29
30import warnings
31with warnings.catch_warnings():
32    if sys.py3kwarning:
33        warnings.filterwarnings("ignore", ".*rfc822 has been removed",
34                                DeprecationWarning)
35    import rfc822
36
37__all__ = [ 'Mailbox', 'Maildir', 'mbox', 'MH', 'Babyl', 'MMDF',
38            'Message', 'MaildirMessage', 'mboxMessage', 'MHMessage',
39            'BabylMessage', 'MMDFMessage', 'UnixMailbox',
40            'PortableUnixMailbox', 'MmdfMailbox', 'MHMailbox', 'BabylMailbox' ]
41
42class Mailbox:
43    """A group of messages in a particular place."""
44
45    def __init__(self, path, factory=None, create=True):
46        """Initialize a Mailbox instance."""
47        self._path = os.path.abspath(os.path.expanduser(path))
48        self._factory = factory
49
50    def add(self, message):
51        """Add message and return assigned key."""
52        raise NotImplementedError('Method must be implemented by subclass')
53
54    def remove(self, key):
55        """Remove the keyed message; raise KeyError if it doesn't exist."""
56        raise NotImplementedError('Method must be implemented by subclass')
57
58    def __delitem__(self, key):
59        self.remove(key)
60
61    def discard(self, key):
62        """If the keyed message exists, remove it."""
63        try:
64            self.remove(key)
65        except KeyError:
66            pass
67
68    def __setitem__(self, key, message):
69        """Replace the keyed message; raise KeyError if it doesn't exist."""
70        raise NotImplementedError('Method must be implemented by subclass')
71
72    def get(self, key, default=None):
73        """Return the keyed message, or default if it doesn't exist."""
74        try:
75            return self.__getitem__(key)
76        except KeyError:
77            return default
78
79    def __getitem__(self, key):
80        """Return the keyed message; raise KeyError if it doesn't exist."""
81        if not self._factory:
82            return self.get_message(key)
83        else:
84            return self._factory(self.get_file(key))
85
86    def get_message(self, key):
87        """Return a Message representation or raise a KeyError."""
88        raise NotImplementedError('Method must be implemented by subclass')
89
90    def get_string(self, key):
91        """Return a string representation or raise a KeyError."""
92        raise NotImplementedError('Method must be implemented by subclass')
93
94    def get_file(self, key):
95        """Return a file-like representation or raise a KeyError."""
96        raise NotImplementedError('Method must be implemented by subclass')
97
98    def iterkeys(self):
99        """Return an iterator over keys."""
100        raise NotImplementedError('Method must be implemented by subclass')
101
102    def keys(self):
103        """Return a list of keys."""
104        return list(self.iterkeys())
105
106    def itervalues(self):
107        """Return an iterator over all messages."""
108        for key in self.iterkeys():
109            try:
110                value = self[key]
111            except KeyError:
112                continue
113            yield value
114
115    def __iter__(self):
116        return self.itervalues()
117
118    def values(self):
119        """Return a list of messages. Memory intensive."""
120        return list(self.itervalues())
121
122    def iteritems(self):
123        """Return an iterator over (key, message) tuples."""
124        for key in self.iterkeys():
125            try:
126                value = self[key]
127            except KeyError:
128                continue
129            yield (key, value)
130
131    def items(self):
132        """Return a list of (key, message) tuples. Memory intensive."""
133        return list(self.iteritems())
134
135    def has_key(self, key):
136        """Return True if the keyed message exists, False otherwise."""
137        raise NotImplementedError('Method must be implemented by subclass')
138
139    def __contains__(self, key):
140        return self.has_key(key)
141
142    def __len__(self):
143        """Return a count of messages in the mailbox."""
144        raise NotImplementedError('Method must be implemented by subclass')
145
146    def clear(self):
147        """Delete all messages."""
148        for key in self.iterkeys():
149            self.discard(key)
150
151    def pop(self, key, default=None):
152        """Delete the keyed message and return it, or default."""
153        try:
154            result = self[key]
155        except KeyError:
156            return default
157        self.discard(key)
158        return result
159
160    def popitem(self):
161        """Delete an arbitrary (key, message) pair and return it."""
162        for key in self.iterkeys():
163            return (key, self.pop(key))     # This is only run once.
164        else:
165            raise KeyError('No messages in mailbox')
166
167    def update(self, arg=None):
168        """Change the messages that correspond to certain keys."""
169        if hasattr(arg, 'iteritems'):
170            source = arg.iteritems()
171        elif hasattr(arg, 'items'):
172            source = arg.items()
173        else:
174            source = arg
175        bad_key = False
176        for key, message in source:
177            try:
178                self[key] = message
179            except KeyError:
180                bad_key = True
181        if bad_key:
182            raise KeyError('No message with key(s)')
183
184    def flush(self):
185        """Write any pending changes to the disk."""
186        raise NotImplementedError('Method must be implemented by subclass')
187
188    def lock(self):
189        """Lock the mailbox."""
190        raise NotImplementedError('Method must be implemented by subclass')
191
192    def unlock(self):
193        """Unlock the mailbox if it is locked."""
194        raise NotImplementedError('Method must be implemented by subclass')
195
196    def close(self):
197        """Flush and close the mailbox."""
198        raise NotImplementedError('Method must be implemented by subclass')
199
200    # Whether each message must end in a newline
201    _append_newline = False
202
203    def _dump_message(self, message, target, mangle_from_=False):
204        # Most files are opened in binary mode to allow predictable seeking.
205        # To get native line endings on disk, the user-friendly \n line endings
206        # used in strings and by email.Message are translated here.
207        """Dump message contents to target file."""
208        if isinstance(message, email.message.Message):
209            buffer = StringIO.StringIO()
210            gen = email.generator.Generator(buffer, mangle_from_, 0)
211            gen.flatten(message)
212            buffer.seek(0)
213            data = buffer.read().replace('\n', os.linesep)
214            target.write(data)
215            if self._append_newline and not data.endswith(os.linesep):
216                # Make sure the message ends with a newline
217                target.write(os.linesep)
218        elif isinstance(message, str):
219            if mangle_from_:
220                message = message.replace('\nFrom ', '\n>From ')
221            message = message.replace('\n', os.linesep)
222            target.write(message)
223            if self._append_newline and not message.endswith(os.linesep):
224                # Make sure the message ends with a newline
225                target.write(os.linesep)
226        elif hasattr(message, 'read'):
227            lastline = None
228            while True:
229                line = message.readline()
230                if line == '':
231                    break
232                if mangle_from_ and line.startswith('From '):
233                    line = '>From ' + line[5:]
234                line = line.replace('\n', os.linesep)
235                target.write(line)
236                lastline = line
237            if self._append_newline and lastline and not lastline.endswith(os.linesep):
238                # Make sure the message ends with a newline
239                target.write(os.linesep)
240        else:
241            raise TypeError('Invalid message type: %s' % type(message))
242
243
244class Maildir(Mailbox):
245    """A qmail-style Maildir mailbox."""
246
247    colon = ':'
248
249    def __init__(self, dirname, factory=rfc822.Message, create=True):
250        """Initialize a Maildir instance."""
251        Mailbox.__init__(self, dirname, factory, create)
252        self._paths = {
253            'tmp': os.path.join(self._path, 'tmp'),
254            'new': os.path.join(self._path, 'new'),
255            'cur': os.path.join(self._path, 'cur'),
256            }
257        if not os.path.exists(self._path):
258            if create:
259                os.mkdir(self._path, 0700)
260                for path in self._paths.values():
261                    os.mkdir(path, 0o700)
262            else:
263                raise NoSuchMailboxError(self._path)
264        self._toc = {}
265        self._toc_mtimes = {'cur': 0, 'new': 0}
266        self._last_read = 0         # Records last time we read cur/new
267        self._skewfactor = 0.1      # Adjust if os/fs clocks are skewing
268
269    def add(self, message):
270        """Add message and return assigned key."""
271        tmp_file = self._create_tmp()
272        try:
273            self._dump_message(message, tmp_file)
274        except BaseException:
275            tmp_file.close()
276            os.remove(tmp_file.name)
277            raise
278        _sync_close(tmp_file)
279        if isinstance(message, MaildirMessage):
280            subdir = message.get_subdir()
281            suffix = self.colon + message.get_info()
282            if suffix == self.colon:
283                suffix = ''
284        else:
285            subdir = 'new'
286            suffix = ''
287        uniq = os.path.basename(tmp_file.name).split(self.colon)[0]
288        dest = os.path.join(self._path, subdir, uniq + suffix)
289        try:
290            if hasattr(os, 'link'):
291                os.link(tmp_file.name, dest)
292                os.remove(tmp_file.name)
293            else:
294                os.rename(tmp_file.name, dest)
295        except OSError, e:
296            os.remove(tmp_file.name)
297            if e.errno == errno.EEXIST:
298                raise ExternalClashError('Name clash with existing message: %s'
299                                         % dest)
300            else:
301                raise
302        if isinstance(message, MaildirMessage):
303            os.utime(dest, (os.path.getatime(dest), message.get_date()))
304        return uniq
305
306    def remove(self, key):
307        """Remove the keyed message; raise KeyError if it doesn't exist."""
308        os.remove(os.path.join(self._path, self._lookup(key)))
309
310    def discard(self, key):
311        """If the keyed message exists, remove it."""
312        # This overrides an inapplicable implementation in the superclass.
313        try:
314            self.remove(key)
315        except KeyError:
316            pass
317        except OSError, e:
318            if e.errno != errno.ENOENT:
319                raise
320
321    def __setitem__(self, key, message):
322        """Replace the keyed message; raise KeyError if it doesn't exist."""
323        old_subpath = self._lookup(key)
324        temp_key = self.add(message)
325        temp_subpath = self._lookup(temp_key)
326        if isinstance(message, MaildirMessage):
327            # temp's subdir and suffix were specified by message.
328            dominant_subpath = temp_subpath
329        else:
330            # temp's subdir and suffix were defaults from add().
331            dominant_subpath = old_subpath
332        subdir = os.path.dirname(dominant_subpath)
333        if self.colon in dominant_subpath:
334            suffix = self.colon + dominant_subpath.split(self.colon)[-1]
335        else:
336            suffix = ''
337        self.discard(key)
338        new_path = os.path.join(self._path, subdir, key + suffix)
339        os.rename(os.path.join(self._path, temp_subpath), new_path)
340        if isinstance(message, MaildirMessage):
341            os.utime(new_path, (os.path.getatime(new_path),
342                                message.get_date()))
343
344    def get_message(self, key):
345        """Return a Message representation or raise a KeyError."""
346        subpath = self._lookup(key)
347        f = open(os.path.join(self._path, subpath), 'r')
348        try:
349            if self._factory:
350                msg = self._factory(f)
351            else:
352                msg = MaildirMessage(f)
353        finally:
354            f.close()
355        subdir, name = os.path.split(subpath)
356        msg.set_subdir(subdir)
357        if self.colon in name:
358            msg.set_info(name.split(self.colon)[-1])
359        msg.set_date(os.path.getmtime(os.path.join(self._path, subpath)))
360        return msg
361
362    def get_string(self, key):
363        """Return a string representation or raise a KeyError."""
364        f = open(os.path.join(self._path, self._lookup(key)), 'r')
365        try:
366            return f.read()
367        finally:
368            f.close()
369
370    def get_file(self, key):
371        """Return a file-like representation or raise a KeyError."""
372        f = open(os.path.join(self._path, self._lookup(key)), 'rb')
373        return _ProxyFile(f)
374
375    def iterkeys(self):
376        """Return an iterator over keys."""
377        self._refresh()
378        for key in self._toc:
379            try:
380                self._lookup(key)
381            except KeyError:
382                continue
383            yield key
384
385    def has_key(self, key):
386        """Return True if the keyed message exists, False otherwise."""
387        self._refresh()
388        return key in self._toc
389
390    def __len__(self):
391        """Return a count of messages in the mailbox."""
392        self._refresh()
393        return len(self._toc)
394
395    def flush(self):
396        """Write any pending changes to disk."""
397        # Maildir changes are always written immediately, so there's nothing
398        # to do.
399        pass
400
401    def lock(self):
402        """Lock the mailbox."""
403        return
404
405    def unlock(self):
406        """Unlock the mailbox if it is locked."""
407        return
408
409    def close(self):
410        """Flush and close the mailbox."""
411        return
412
413    def list_folders(self):
414        """Return a list of folder names."""
415        result = []
416        for entry in os.listdir(self._path):
417            if len(entry) > 1 and entry[0] == '.' and \
418               os.path.isdir(os.path.join(self._path, entry)):
419                result.append(entry[1:])
420        return result
421
422    def get_folder(self, folder):
423        """Return a Maildir instance for the named folder."""
424        return Maildir(os.path.join(self._path, '.' + folder),
425                       factory=self._factory,
426                       create=False)
427
428    def add_folder(self, folder):
429        """Create a folder and return a Maildir instance representing it."""
430        path = os.path.join(self._path, '.' + folder)
431        result = Maildir(path, factory=self._factory)
432        maildirfolder_path = os.path.join(path, 'maildirfolder')
433        if not os.path.exists(maildirfolder_path):
434            os.close(os.open(maildirfolder_path, os.O_CREAT | os.O_WRONLY,
435                0666))
436        return result
437
438    def remove_folder(self, folder):
439        """Delete the named folder, which must be empty."""
440        path = os.path.join(self._path, '.' + folder)
441        for entry in os.listdir(os.path.join(path, 'new')) + \
442                     os.listdir(os.path.join(path, 'cur')):
443            if len(entry) < 1 or entry[0] != '.':
444                raise NotEmptyError('Folder contains message(s): %s' % folder)
445        for entry in os.listdir(path):
446            if entry != 'new' and entry != 'cur' and entry != 'tmp' and \
447               os.path.isdir(os.path.join(path, entry)):
448                raise NotEmptyError("Folder contains subdirectory '%s': %s" %
449                                    (folder, entry))
450        for root, dirs, files in os.walk(path, topdown=False):
451            for entry in files:
452                os.remove(os.path.join(root, entry))
453            for entry in dirs:
454                os.rmdir(os.path.join(root, entry))
455        os.rmdir(path)
456
457    def clean(self):
458        """Delete old files in "tmp"."""
459        now = time.time()
460        for entry in os.listdir(os.path.join(self._path, 'tmp')):
461            path = os.path.join(self._path, 'tmp', entry)
462            if now - os.path.getatime(path) > 129600:   # 60 * 60 * 36
463                os.remove(path)
464
465    _count = 1  # This is used to generate unique file names.
466
467    def _create_tmp(self):
468        """Create a file in the tmp subdirectory and open and return it."""
469        now = time.time()
470        hostname = socket.gethostname()
471        if '/' in hostname:
472            hostname = hostname.replace('/', r'\057')
473        if ':' in hostname:
474            hostname = hostname.replace(':', r'\072')
475        uniq = "%s.M%sP%sQ%s.%s" % (int(now), int(now % 1 * 1e6), os.getpid(),
476                                    Maildir._count, hostname)
477        path = os.path.join(self._path, 'tmp', uniq)
478        try:
479            os.stat(path)
480        except OSError, e:
481            if e.errno == errno.ENOENT:
482                Maildir._count += 1
483                try:
484                    return _create_carefully(path)
485                except OSError, e:
486                    if e.errno != errno.EEXIST:
487                        raise
488            else:
489                raise
490
491        # Fall through to here if stat succeeded or open raised EEXIST.
492        raise ExternalClashError('Name clash prevented file creation: %s' %
493                                 path)
494
495    def _refresh(self):
496        """Update table of contents mapping."""
497        # If it has been less than two seconds since the last _refresh() call,
498        # we have to unconditionally re-read the mailbox just in case it has
499        # been modified, because os.path.mtime() has a 2 sec resolution in the
500        # most common worst case (FAT) and a 1 sec resolution typically.  This
501        # results in a few unnecessary re-reads when _refresh() is called
502        # multiple times in that interval, but once the clock ticks over, we
503        # will only re-read as needed.  Because the filesystem might be being
504        # served by an independent system with its own clock, we record and
505        # compare with the mtimes from the filesystem.  Because the other
506        # system's clock might be skewing relative to our clock, we add an
507        # extra delta to our wait.  The default is one tenth second, but is an
508        # instance variable and so can be adjusted if dealing with a
509        # particularly skewed or irregular system.
510        if time.time() - self._last_read > 2 + self._skewfactor:
511            refresh = False
512            for subdir in self._toc_mtimes:
513                mtime = os.path.getmtime(self._paths[subdir])
514                if mtime > self._toc_mtimes[subdir]:
515                    refresh = True
516                self._toc_mtimes[subdir] = mtime
517            if not refresh:
518                return
519        # Refresh toc
520        self._toc = {}
521        for subdir in self._toc_mtimes:
522            path = self._paths[subdir]
523            for entry in os.listdir(path):
524                p = os.path.join(path, entry)
525                if os.path.isdir(p):
526                    continue
527                uniq = entry.split(self.colon)[0]
528                self._toc[uniq] = os.path.join(subdir, entry)
529        self._last_read = time.time()
530
531    def _lookup(self, key):
532        """Use TOC to return subpath for given key, or raise a KeyError."""
533        try:
534            if os.path.exists(os.path.join(self._path, self._toc[key])):
535                return self._toc[key]
536        except KeyError:
537            pass
538        self._refresh()
539        try:
540            return self._toc[key]
541        except KeyError:
542            raise KeyError('No message with key: %s' % key)
543
544    # This method is for backward compatibility only.
545    def next(self):
546        """Return the next message in a one-time iteration."""
547        if not hasattr(self, '_onetime_keys'):
548            self._onetime_keys = self.iterkeys()
549        while True:
550            try:
551                return self[self._onetime_keys.next()]
552            except StopIteration:
553                return None
554            except KeyError:
555                continue
556
557
558class _singlefileMailbox(Mailbox):
559    """A single-file mailbox."""
560
561    def __init__(self, path, factory=None, create=True):
562        """Initialize a single-file mailbox."""
563        Mailbox.__init__(self, path, factory, create)
564        try:
565            f = open(self._path, 'rb+')
566        except IOError, e:
567            if e.errno == errno.ENOENT:
568                if create:
569                    f = open(self._path, 'wb+')
570                else:
571                    raise NoSuchMailboxError(self._path)
572            elif e.errno in (errno.EACCES, errno.EROFS):
573                f = open(self._path, 'rb')
574            else:
575                raise
576        self._file = f
577        self._toc = None
578        self._next_key = 0
579        self._pending = False       # No changes require rewriting the file.
580        self._pending_sync = False  # No need to sync the file
581        self._locked = False
582        self._file_length = None    # Used to record mailbox size
583
584    def add(self, message):
585        """Add message and return assigned key."""
586        self._lookup()
587        self._toc[self._next_key] = self._append_message(message)
588        self._next_key += 1
589        # _append_message appends the message to the mailbox file. We
590        # don't need a full rewrite + rename, sync is enough.
591        self._pending_sync = True
592        return self._next_key - 1
593
594    def remove(self, key):
595        """Remove the keyed message; raise KeyError if it doesn't exist."""
596        self._lookup(key)
597        del self._toc[key]
598        self._pending = True
599
600    def __setitem__(self, key, message):
601        """Replace the keyed message; raise KeyError if it doesn't exist."""
602        self._lookup(key)
603        self._toc[key] = self._append_message(message)
604        self._pending = True
605
606    def iterkeys(self):
607        """Return an iterator over keys."""
608        self._lookup()
609        for key in self._toc.keys():
610            yield key
611
612    def has_key(self, key):
613        """Return True if the keyed message exists, False otherwise."""
614        self._lookup()
615        return key in self._toc
616
617    def __len__(self):
618        """Return a count of messages in the mailbox."""
619        self._lookup()
620        return len(self._toc)
621
622    def lock(self):
623        """Lock the mailbox."""
624        if not self._locked:
625            _lock_file(self._file)
626            self._locked = True
627
628    def unlock(self):
629        """Unlock the mailbox if it is locked."""
630        if self._locked:
631            _unlock_file(self._file)
632            self._locked = False
633
634    def flush(self):
635        """Write any pending changes to disk."""
636        if not self._pending:
637            if self._pending_sync:
638                # Messages have only been added, so syncing the file
639                # is enough.
640                _sync_flush(self._file)
641                self._pending_sync = False
642            return
643
644        # In order to be writing anything out at all, self._toc must
645        # already have been generated (and presumably has been modified
646        # by adding or deleting an item).
647        assert self._toc is not None
648
649        # Check length of self._file; if it's changed, some other process
650        # has modified the mailbox since we scanned it.
651        self._file.seek(0, 2)
652        cur_len = self._file.tell()
653        if cur_len != self._file_length:
654            raise ExternalClashError('Size of mailbox file changed '
655                                     '(expected %i, found %i)' %
656                                     (self._file_length, cur_len))
657
658        new_file = _create_temporary(self._path)
659        try:
660            new_toc = {}
661            self._pre_mailbox_hook(new_file)
662            for key in sorted(self._toc.keys()):
663                start, stop = self._toc[key]
664                self._file.seek(start)
665                self._pre_message_hook(new_file)
666                new_start = new_file.tell()
667                while True:
668                    buffer = self._file.read(min(4096,
669                                                 stop - self._file.tell()))
670                    if buffer == '':
671                        break
672                    new_file.write(buffer)
673                new_toc[key] = (new_start, new_file.tell())
674                self._post_message_hook(new_file)
675            self._file_length = new_file.tell()
676        except:
677            new_file.close()
678            os.remove(new_file.name)
679            raise
680        _sync_close(new_file)
681        # self._file is about to get replaced, so no need to sync.
682        self._file.close()
683        # Make sure the new file's mode is the same as the old file's
684        mode = os.stat(self._path).st_mode
685        os.chmod(new_file.name, mode)
686        try:
687            os.rename(new_file.name, self._path)
688        except OSError, e:
689            if e.errno == errno.EEXIST or \
690              (os.name == 'os2' and e.errno == errno.EACCES):
691                os.remove(self._path)
692                os.rename(new_file.name, self._path)
693            else:
694                raise
695        self._file = open(self._path, 'rb+')
696        self._toc = new_toc
697        self._pending = False
698        self._pending_sync = False
699        if self._locked:
700            _lock_file(self._file, dotlock=False)
701
702    def _pre_mailbox_hook(self, f):
703        """Called before writing the mailbox to file f."""
704        return
705
706    def _pre_message_hook(self, f):
707        """Called before writing each message to file f."""
708        return
709
710    def _post_message_hook(self, f):
711        """Called after writing each message to file f."""
712        return
713
714    def close(self):
715        """Flush and close the mailbox."""
716        self.flush()
717        if self._locked:
718            self.unlock()
719        self._file.close()  # Sync has been done by self.flush() above.
720
721    def _lookup(self, key=None):
722        """Return (start, stop) or raise KeyError."""
723        if self._toc is None:
724            self._generate_toc()
725        if key is not None:
726            try:
727                return self._toc[key]
728            except KeyError:
729                raise KeyError('No message with key: %s' % key)
730
731    def _append_message(self, message):
732        """Append message to mailbox and return (start, stop) offsets."""
733        self._file.seek(0, 2)
734        before = self._file.tell()
735        if len(self._toc) == 0 and not self._pending:
736            # This is the first message, and the _pre_mailbox_hook
737            # hasn't yet been called. If self._pending is True,
738            # messages have been removed, so _pre_mailbox_hook must
739            # have been called already.
740            self._pre_mailbox_hook(self._file)
741        try:
742            self._pre_message_hook(self._file)
743            offsets = self._install_message(message)
744            self._post_message_hook(self._file)
745        except BaseException:
746            self._file.truncate(before)
747            raise
748        self._file.flush()
749        self._file_length = self._file.tell()  # Record current length of mailbox
750        return offsets
751
752
753
754class _mboxMMDF(_singlefileMailbox):
755    """An mbox or MMDF mailbox."""
756
757    _mangle_from_ = True
758
759    def get_message(self, key):
760        """Return a Message representation or raise a KeyError."""
761        start, stop = self._lookup(key)
762        self._file.seek(start)
763        from_line = self._file.readline().replace(os.linesep, '')
764        string = self._file.read(stop - self._file.tell())
765        msg = self._message_factory(string.replace(os.linesep, '\n'))
766        msg.set_from(from_line[5:])
767        return msg
768
769    def get_string(self, key, from_=False):
770        """Return a string representation or raise a KeyError."""
771        start, stop = self._lookup(key)
772        self._file.seek(start)
773        if not from_:
774            self._file.readline()
775        string = self._file.read(stop - self._file.tell())
776        return string.replace(os.linesep, '\n')
777
778    def get_file(self, key, from_=False):
779        """Return a file-like representation or raise a KeyError."""
780        start, stop = self._lookup(key)
781        self._file.seek(start)
782        if not from_:
783            self._file.readline()
784        return _PartialFile(self._file, self._file.tell(), stop)
785
786    def _install_message(self, message):
787        """Format a message and blindly write to self._file."""
788        from_line = None
789        if isinstance(message, str) and message.startswith('From '):
790            newline = message.find('\n')
791            if newline != -1:
792                from_line = message[:newline]
793                message = message[newline + 1:]
794            else:
795                from_line = message
796                message = ''
797        elif isinstance(message, _mboxMMDFMessage):
798            from_line = 'From ' + message.get_from()
799        elif isinstance(message, email.message.Message):
800            from_line = message.get_unixfrom()  # May be None.
801        if from_line is None:
802            from_line = 'From MAILER-DAEMON %s' % time.asctime(time.gmtime())
803        start = self._file.tell()
804        self._file.write(from_line + os.linesep)
805        self._dump_message(message, self._file, self._mangle_from_)
806        stop = self._file.tell()
807        return (start, stop)
808
809
810class mbox(_mboxMMDF):
811    """A classic mbox mailbox."""
812
813    _mangle_from_ = True
814
815    # All messages must end in a newline character, and
816    # _post_message_hooks outputs an empty line between messages.
817    _append_newline = True
818
819    def __init__(self, path, factory=None, create=True):
820        """Initialize an mbox mailbox."""
821        self._message_factory = mboxMessage
822        _mboxMMDF.__init__(self, path, factory, create)
823
824    def _post_message_hook(self, f):
825        """Called after writing each message to file f."""
826        f.write(os.linesep)
827
828    def _generate_toc(self):
829        """Generate key-to-(start, stop) table of contents."""
830        starts, stops = [], []
831        last_was_empty = False
832        self._file.seek(0)
833        while True:
834            line_pos = self._file.tell()
835            line = self._file.readline()
836            if line.startswith('From '):
837                if len(stops) < len(starts):
838                    if last_was_empty:
839                        stops.append(line_pos - len(os.linesep))
840                    else:
841                        # The last line before the "From " line wasn't
842                        # blank, but we consider it a start of a
843                        # message anyway.
844                        stops.append(line_pos)
845                starts.append(line_pos)
846                last_was_empty = False
847            elif not line:
848                if last_was_empty:
849                    stops.append(line_pos - len(os.linesep))
850                else:
851                    stops.append(line_pos)
852                break
853            elif line == os.linesep:
854                last_was_empty = True
855            else:
856                last_was_empty = False
857        self._toc = dict(enumerate(zip(starts, stops)))
858        self._next_key = len(self._toc)
859        self._file_length = self._file.tell()
860
861
862class MMDF(_mboxMMDF):
863    """An MMDF mailbox."""
864
865    def __init__(self, path, factory=None, create=True):
866        """Initialize an MMDF mailbox."""
867        self._message_factory = MMDFMessage
868        _mboxMMDF.__init__(self, path, factory, create)
869
870    def _pre_message_hook(self, f):
871        """Called before writing each message to file f."""
872        f.write('\001\001\001\001' + os.linesep)
873
874    def _post_message_hook(self, f):
875        """Called after writing each message to file f."""
876        f.write(os.linesep + '\001\001\001\001' + os.linesep)
877
878    def _generate_toc(self):
879        """Generate key-to-(start, stop) table of contents."""
880        starts, stops = [], []
881        self._file.seek(0)
882        next_pos = 0
883        while True:
884            line_pos = next_pos
885            line = self._file.readline()
886            next_pos = self._file.tell()
887            if line.startswith('\001\001\001\001' + os.linesep):
888                starts.append(next_pos)
889                while True:
890                    line_pos = next_pos
891                    line = self._file.readline()
892                    next_pos = self._file.tell()
893                    if line == '\001\001\001\001' + os.linesep:
894                        stops.append(line_pos - len(os.linesep))
895                        break
896                    elif line == '':
897                        stops.append(line_pos)
898                        break
899            elif line == '':
900                break
901        self._toc = dict(enumerate(zip(starts, stops)))
902        self._next_key = len(self._toc)
903        self._file.seek(0, 2)
904        self._file_length = self._file.tell()
905
906
907class MH(Mailbox):
908    """An MH mailbox."""
909
910    def __init__(self, path, factory=None, create=True):
911        """Initialize an MH instance."""
912        Mailbox.__init__(self, path, factory, create)
913        if not os.path.exists(self._path):
914            if create:
915                os.mkdir(self._path, 0700)
916                os.close(os.open(os.path.join(self._path, '.mh_sequences'),
917                                 os.O_CREAT | os.O_EXCL | os.O_WRONLY, 0600))
918            else:
919                raise NoSuchMailboxError(self._path)
920        self._locked = False
921
922    def add(self, message):
923        """Add message and return assigned key."""
924        keys = self.keys()
925        if len(keys) == 0:
926            new_key = 1
927        else:
928            new_key = max(keys) + 1
929        new_path = os.path.join(self._path, str(new_key))
930        f = _create_carefully(new_path)
931        closed = False
932        try:
933            if self._locked:
934                _lock_file(f)
935            try:
936                try:
937                    self._dump_message(message, f)
938                except BaseException:
939                    # Unlock and close so it can be deleted on Windows
940                    if self._locked:
941                        _unlock_file(f)
942                    _sync_close(f)
943                    closed = True
944                    os.remove(new_path)
945                    raise
946                if isinstance(message, MHMessage):
947                    self._dump_sequences(message, new_key)
948            finally:
949                if self._locked:
950                    _unlock_file(f)
951        finally:
952            if not closed:
953                _sync_close(f)
954        return new_key
955
956    def remove(self, key):
957        """Remove the keyed message; raise KeyError if it doesn't exist."""
958        path = os.path.join(self._path, str(key))
959        try:
960            f = open(path, 'rb+')
961        except IOError, e:
962            if e.errno == errno.ENOENT:
963                raise KeyError('No message with key: %s' % key)
964            else:
965                raise
966        else:
967            f.close()
968            os.remove(path)
969
970    def __setitem__(self, key, message):
971        """Replace the keyed message; raise KeyError if it doesn't exist."""
972        path = os.path.join(self._path, str(key))
973        try:
974            f = open(path, 'rb+')
975        except IOError, e:
976            if e.errno == errno.ENOENT:
977                raise KeyError('No message with key: %s' % key)
978            else:
979                raise
980        try:
981            if self._locked:
982                _lock_file(f)
983            try:
984                os.close(os.open(path, os.O_WRONLY | os.O_TRUNC))
985                self._dump_message(message, f)
986                if isinstance(message, MHMessage):
987                    self._dump_sequences(message, key)
988            finally:
989                if self._locked:
990                    _unlock_file(f)
991        finally:
992            _sync_close(f)
993
994    def get_message(self, key):
995        """Return a Message representation or raise a KeyError."""
996        try:
997            if self._locked:
998                f = open(os.path.join(self._path, str(key)), 'r+')
999            else:
1000                f = open(os.path.join(self._path, str(key)), 'r')
1001        except IOError, e:
1002            if e.errno == errno.ENOENT:
1003                raise KeyError('No message with key: %s' % key)
1004            else:
1005                raise
1006        try:
1007            if self._locked:
1008                _lock_file(f)
1009            try:
1010                msg = MHMessage(f)
1011            finally:
1012                if self._locked:
1013                    _unlock_file(f)
1014        finally:
1015            f.close()
1016        for name, key_list in self.get_sequences().iteritems():
1017            if key in key_list:
1018                msg.add_sequence(name)
1019        return msg
1020
1021    def get_string(self, key):
1022        """Return a string representation or raise a KeyError."""
1023        try:
1024            if self._locked:
1025                f = open(os.path.join(self._path, str(key)), 'r+')
1026            else:
1027                f = open(os.path.join(self._path, str(key)), 'r')
1028        except IOError, e:
1029            if e.errno == errno.ENOENT:
1030                raise KeyError('No message with key: %s' % key)
1031            else:
1032                raise
1033        try:
1034            if self._locked:
1035                _lock_file(f)
1036            try:
1037                return f.read()
1038            finally:
1039                if self._locked:
1040                    _unlock_file(f)
1041        finally:
1042            f.close()
1043
1044    def get_file(self, key):
1045        """Return a file-like representation or raise a KeyError."""
1046        try:
1047            f = open(os.path.join(self._path, str(key)), 'rb')
1048        except IOError, e:
1049            if e.errno == errno.ENOENT:
1050                raise KeyError('No message with key: %s' % key)
1051            else:
1052                raise
1053        return _ProxyFile(f)
1054
1055    def iterkeys(self):
1056        """Return an iterator over keys."""
1057        return iter(sorted(int(entry) for entry in os.listdir(self._path)
1058                                      if entry.isdigit()))
1059
1060    def has_key(self, key):
1061        """Return True if the keyed message exists, False otherwise."""
1062        return os.path.exists(os.path.join(self._path, str(key)))
1063
1064    def __len__(self):
1065        """Return a count of messages in the mailbox."""
1066        return len(list(self.iterkeys()))
1067
1068    def lock(self):
1069        """Lock the mailbox."""
1070        if not self._locked:
1071            self._file = open(os.path.join(self._path, '.mh_sequences'), 'rb+')
1072            _lock_file(self._file)
1073            self._locked = True
1074
1075    def unlock(self):
1076        """Unlock the mailbox if it is locked."""
1077        if self._locked:
1078            _unlock_file(self._file)
1079            _sync_close(self._file)
1080            del self._file
1081            self._locked = False
1082
1083    def flush(self):
1084        """Write any pending changes to the disk."""
1085        return
1086
1087    def close(self):
1088        """Flush and close the mailbox."""
1089        if self._locked:
1090            self.unlock()
1091
1092    def list_folders(self):
1093        """Return a list of folder names."""
1094        result = []
1095        for entry in os.listdir(self._path):
1096            if os.path.isdir(os.path.join(self._path, entry)):
1097                result.append(entry)
1098        return result
1099
1100    def get_folder(self, folder):
1101        """Return an MH instance for the named folder."""
1102        return MH(os.path.join(self._path, folder),
1103                  factory=self._factory, create=False)
1104
1105    def add_folder(self, folder):
1106        """Create a folder and return an MH instance representing it."""
1107        return MH(os.path.join(self._path, folder),
1108                  factory=self._factory)
1109
1110    def remove_folder(self, folder):
1111        """Delete the named folder, which must be empty."""
1112        path = os.path.join(self._path, folder)
1113        entries = os.listdir(path)
1114        if entries == ['.mh_sequences']:
1115            os.remove(os.path.join(path, '.mh_sequences'))
1116        elif entries == []:
1117            pass
1118        else:
1119            raise NotEmptyError('Folder not empty: %s' % self._path)
1120        os.rmdir(path)
1121
1122    def get_sequences(self):
1123        """Return a name-to-key-list dictionary to define each sequence."""
1124        results = {}
1125        f = open(os.path.join(self._path, '.mh_sequences'), 'r')
1126        try:
1127            all_keys = set(self.keys())
1128            for line in f:
1129                try:
1130                    name, contents = line.split(':')
1131                    keys = set()
1132                    for spec in contents.split():
1133                        if spec.isdigit():
1134                            keys.add(int(spec))
1135                        else:
1136                            start, stop = (int(x) for x in spec.split('-'))
1137                            keys.update(range(start, stop + 1))
1138                    results[name] = [key for key in sorted(keys) \
1139                                         if key in all_keys]
1140                    if len(results[name]) == 0:
1141                        del results[name]
1142                except ValueError:
1143                    raise FormatError('Invalid sequence specification: %s' %
1144                                      line.rstrip())
1145        finally:
1146            f.close()
1147        return results
1148
1149    def set_sequences(self, sequences):
1150        """Set sequences using the given name-to-key-list dictionary."""
1151        f = open(os.path.join(self._path, '.mh_sequences'), 'r+')
1152        try:
1153            os.close(os.open(f.name, os.O_WRONLY | os.O_TRUNC))
1154            for name, keys in sequences.iteritems():
1155                if len(keys) == 0:
1156                    continue
1157                f.write('%s:' % name)
1158                prev = None
1159                completing = False
1160                for key in sorted(set(keys)):
1161                    if key - 1 == prev:
1162                        if not completing:
1163                            completing = True
1164                            f.write('-')
1165                    elif completing:
1166                        completing = False
1167                        f.write('%s %s' % (prev, key))
1168                    else:
1169                        f.write(' %s' % key)
1170                    prev = key
1171                if completing:
1172                    f.write(str(prev) + '\n')
1173                else:
1174                    f.write('\n')
1175        finally:
1176            _sync_close(f)
1177
1178    def pack(self):
1179        """Re-name messages to eliminate numbering gaps. Invalidates keys."""
1180        sequences = self.get_sequences()
1181        prev = 0
1182        changes = []
1183        for key in self.iterkeys():
1184            if key - 1 != prev:
1185                changes.append((key, prev + 1))
1186                if hasattr(os, 'link'):
1187                    os.link(os.path.join(self._path, str(key)),
1188                            os.path.join(self._path, str(prev + 1)))
1189                    os.unlink(os.path.join(self._path, str(key)))
1190                else:
1191                    os.rename(os.path.join(self._path, str(key)),
1192                              os.path.join(self._path, str(prev + 1)))
1193            prev += 1
1194        self._next_key = prev + 1
1195        if len(changes) == 0:
1196            return
1197        for name, key_list in sequences.items():
1198            for old, new in changes:
1199                if old in key_list:
1200                    key_list[key_list.index(old)] = new
1201        self.set_sequences(sequences)
1202
1203    def _dump_sequences(self, message, key):
1204        """Inspect a new MHMessage and update sequences appropriately."""
1205        pending_sequences = message.get_sequences()
1206        all_sequences = self.get_sequences()
1207        for name, key_list in all_sequences.iteritems():
1208            if name in pending_sequences:
1209                key_list.append(key)
1210            elif key in key_list:
1211                del key_list[key_list.index(key)]
1212        for sequence in pending_sequences:
1213            if sequence not in all_sequences:
1214                all_sequences[sequence] = [key]
1215        self.set_sequences(all_sequences)
1216
1217
1218class Babyl(_singlefileMailbox):
1219    """An Rmail-style Babyl mailbox."""
1220
1221    _special_labels = frozenset(('unseen', 'deleted', 'filed', 'answered',
1222                                 'forwarded', 'edited', 'resent'))
1223
1224    def __init__(self, path, factory=None, create=True):
1225        """Initialize a Babyl mailbox."""
1226        _singlefileMailbox.__init__(self, path, factory, create)
1227        self._labels = {}
1228
1229    def add(self, message):
1230        """Add message and return assigned key."""
1231        key = _singlefileMailbox.add(self, message)
1232        if isinstance(message, BabylMessage):
1233            self._labels[key] = message.get_labels()
1234        return key
1235
1236    def remove(self, key):
1237        """Remove the keyed message; raise KeyError if it doesn't exist."""
1238        _singlefileMailbox.remove(self, key)
1239        if key in self._labels:
1240            del self._labels[key]
1241
1242    def __setitem__(self, key, message):
1243        """Replace the keyed message; raise KeyError if it doesn't exist."""
1244        _singlefileMailbox.__setitem__(self, key, message)
1245        if isinstance(message, BabylMessage):
1246            self._labels[key] = message.get_labels()
1247
1248    def get_message(self, key):
1249        """Return a Message representation or raise a KeyError."""
1250        start, stop = self._lookup(key)
1251        self._file.seek(start)
1252        self._file.readline()   # Skip '1,' line specifying labels.
1253        original_headers = StringIO.StringIO()
1254        while True:
1255            line = self._file.readline()
1256            if line == '*** EOOH ***' + os.linesep or line == '':
1257                break
1258            original_headers.write(line.replace(os.linesep, '\n'))
1259        visible_headers = StringIO.StringIO()
1260        while True:
1261            line = self._file.readline()
1262            if line == os.linesep or line == '':
1263                break
1264            visible_headers.write(line.replace(os.linesep, '\n'))
1265        body = self._file.read(stop - self._file.tell()).replace(os.linesep,
1266                                                                 '\n')
1267        msg = BabylMessage(original_headers.getvalue() + body)
1268        msg.set_visible(visible_headers.getvalue())
1269        if key in self._labels:
1270            msg.set_labels(self._labels[key])
1271        return msg
1272
1273    def get_string(self, key):
1274        """Return a string representation or raise a KeyError."""
1275        start, stop = self._lookup(key)
1276        self._file.seek(start)
1277        self._file.readline()   # Skip '1,' line specifying labels.
1278        original_headers = StringIO.StringIO()
1279        while True:
1280            line = self._file.readline()
1281            if line == '*** EOOH ***' + os.linesep or line == '':
1282                break
1283            original_headers.write(line.replace(os.linesep, '\n'))
1284        while True:
1285            line = self._file.readline()
1286            if line == os.linesep or line == '':
1287                break
1288        return original_headers.getvalue() + \
1289               self._file.read(stop - self._file.tell()).replace(os.linesep,
1290                                                                 '\n')
1291
1292    def get_file(self, key):
1293        """Return a file-like representation or raise a KeyError."""
1294        return StringIO.StringIO(self.get_string(key).replace('\n',
1295                                                              os.linesep))
1296
1297    def get_labels(self):
1298        """Return a list of user-defined labels in the mailbox."""
1299        self._lookup()
1300        labels = set()
1301        for label_list in self._labels.values():
1302            labels.update(label_list)
1303        labels.difference_update(self._special_labels)
1304        return list(labels)
1305
1306    def _generate_toc(self):
1307        """Generate key-to-(start, stop) table of contents."""
1308        starts, stops = [], []
1309        self._file.seek(0)
1310        next_pos = 0
1311        label_lists = []
1312        while True:
1313            line_pos = next_pos
1314            line = self._file.readline()
1315            next_pos = self._file.tell()
1316            if line == '\037\014' + os.linesep:
1317                if len(stops) < len(starts):
1318                    stops.append(line_pos - len(os.linesep))
1319                starts.append(next_pos)
1320                labels = [label.strip() for label
1321                                        in self._file.readline()[1:].split(',')
1322                                        if label.strip() != '']
1323                label_lists.append(labels)
1324            elif line == '\037' or line == '\037' + os.linesep:
1325                if len(stops) < len(starts):
1326                    stops.append(line_pos - len(os.linesep))
1327            elif line == '':
1328                stops.append(line_pos - len(os.linesep))
1329                break
1330        self._toc = dict(enumerate(zip(starts, stops)))
1331        self._labels = dict(enumerate(label_lists))
1332        self._next_key = len(self._toc)
1333        self._file.seek(0, 2)
1334        self._file_length = self._file.tell()
1335
1336    def _pre_mailbox_hook(self, f):
1337        """Called before writing the mailbox to file f."""
1338        f.write('BABYL OPTIONS:%sVersion: 5%sLabels:%s%s\037' %
1339                (os.linesep, os.linesep, ','.join(self.get_labels()),
1340                 os.linesep))
1341
1342    def _pre_message_hook(self, f):
1343        """Called before writing each message to file f."""
1344        f.write('\014' + os.linesep)
1345
1346    def _post_message_hook(self, f):
1347        """Called after writing each message to file f."""
1348        f.write(os.linesep + '\037')
1349
1350    def _install_message(self, message):
1351        """Write message contents and return (start, stop)."""
1352        start = self._file.tell()
1353        if isinstance(message, BabylMessage):
1354            special_labels = []
1355            labels = []
1356            for label in message.get_labels():
1357                if label in self._special_labels:
1358                    special_labels.append(label)
1359                else:
1360                    labels.append(label)
1361            self._file.write('1')
1362            for label in special_labels:
1363                self._file.write(', ' + label)
1364            self._file.write(',,')
1365            for label in labels:
1366                self._file.write(' ' + label + ',')
1367            self._file.write(os.linesep)
1368        else:
1369            self._file.write('1,,' + os.linesep)
1370        if isinstance(message, email.message.Message):
1371            orig_buffer = StringIO.StringIO()
1372            orig_generator = email.generator.Generator(orig_buffer, False, 0)
1373            orig_generator.flatten(message)
1374            orig_buffer.seek(0)
1375            while True:
1376                line = orig_buffer.readline()
1377                self._file.write(line.replace('\n', os.linesep))
1378                if line == '\n' or line == '':
1379                    break
1380            self._file.write('*** EOOH ***' + os.linesep)
1381            if isinstance(message, BabylMessage):
1382                vis_buffer = StringIO.StringIO()
1383                vis_generator = email.generator.Generator(vis_buffer, False, 0)
1384                vis_generator.flatten(message.get_visible())
1385                while True:
1386                    line = vis_buffer.readline()
1387                    self._file.write(line.replace('\n', os.linesep))
1388                    if line == '\n' or line == '':
1389                        break
1390            else:
1391                orig_buffer.seek(0)
1392                while True:
1393                    line = orig_buffer.readline()
1394                    self._file.write(line.replace('\n', os.linesep))
1395                    if line == '\n' or line == '':
1396                        break
1397            while True:
1398                buffer = orig_buffer.read(4096) # Buffer size is arbitrary.
1399                if buffer == '':
1400                    break
1401                self._file.write(buffer.replace('\n', os.linesep))
1402        elif isinstance(message, str):
1403            body_start = message.find('\n\n') + 2
1404            if body_start - 2 != -1:
1405                self._file.write(message[:body_start].replace('\n',
1406                                                              os.linesep))
1407                self._file.write('*** EOOH ***' + os.linesep)
1408                self._file.write(message[:body_start].replace('\n',
1409                                                              os.linesep))
1410                self._file.write(message[body_start:].replace('\n',
1411                                                              os.linesep))
1412            else:
1413                self._file.write('*** EOOH ***' + os.linesep + os.linesep)
1414                self._file.write(message.replace('\n', os.linesep))
1415        elif hasattr(message, 'readline'):
1416            original_pos = message.tell()
1417            first_pass = True
1418            while True:
1419                line = message.readline()
1420                self._file.write(line.replace('\n', os.linesep))
1421                if line == '\n' or line == '':
1422                    if first_pass:
1423                        first_pass = False
1424                        self._file.write('*** EOOH ***' + os.linesep)
1425                        message.seek(original_pos)
1426                    else:
1427                        break
1428            while True:
1429                buffer = message.read(4096)     # Buffer size is arbitrary.
1430                if buffer == '':
1431                    break
1432                self._file.write(buffer.replace('\n', os.linesep))
1433        else:
1434            raise TypeError('Invalid message type: %s' % type(message))
1435        stop = self._file.tell()
1436        return (start, stop)
1437
1438
1439class Message(email.message.Message):
1440    """Message with mailbox-format-specific properties."""
1441
1442    def __init__(self, message=None):
1443        """Initialize a Message instance."""
1444        if isinstance(message, email.message.Message):
1445            self._become_message(copy.deepcopy(message))
1446            if isinstance(message, Message):
1447                message._explain_to(self)
1448        elif isinstance(message, str):
1449            self._become_message(email.message_from_string(message))
1450        elif hasattr(message, "read"):
1451            self._become_message(email.message_from_file(message))
1452        elif message is None:
1453            email.message.Message.__init__(self)
1454        else:
1455            raise TypeError('Invalid message type: %s' % type(message))
1456
1457    def _become_message(self, message):
1458        """Assume the non-format-specific state of message."""
1459        for name in ('_headers', '_unixfrom', '_payload', '_charset',
1460                     'preamble', 'epilogue', 'defects', '_default_type'):
1461            self.__dict__[name] = message.__dict__[name]
1462
1463    def _explain_to(self, message):
1464        """Copy format-specific state to message insofar as possible."""
1465        if isinstance(message, Message):
1466            return  # There's nothing format-specific to explain.
1467        else:
1468            raise TypeError('Cannot convert to specified type')
1469
1470
1471class MaildirMessage(Message):
1472    """Message with Maildir-specific properties."""
1473
1474    def __init__(self, message=None):
1475        """Initialize a MaildirMessage instance."""
1476        self._subdir = 'new'
1477        self._info = ''
1478        self._date = time.time()
1479        Message.__init__(self, message)
1480
1481    def get_subdir(self):
1482        """Return 'new' or 'cur'."""
1483        return self._subdir
1484
1485    def set_subdir(self, subdir):
1486        """Set subdir to 'new' or 'cur'."""
1487        if subdir == 'new' or subdir == 'cur':
1488            self._subdir = subdir
1489        else:
1490            raise ValueError("subdir must be 'new' or 'cur': %s" % subdir)
1491
1492    def get_flags(self):
1493        """Return as a string the flags that are set."""
1494        if self._info.startswith('2,'):
1495            return self._info[2:]
1496        else:
1497            return ''
1498
1499    def set_flags(self, flags):
1500        """Set the given flags and unset all others."""
1501        self._info = '2,' + ''.join(sorted(flags))
1502
1503    def add_flag(self, flag):
1504        """Set the given flag(s) without changing others."""
1505        self.set_flags(''.join(set(self.get_flags()) | set(flag)))
1506
1507    def remove_flag(self, flag):
1508        """Unset the given string flag(s) without changing others."""
1509        if self.get_flags() != '':
1510            self.set_flags(''.join(set(self.get_flags()) - set(flag)))
1511
1512    def get_date(self):
1513        """Return delivery date of message, in seconds since the epoch."""
1514        return self._date
1515
1516    def set_date(self, date):
1517        """Set delivery date of message, in seconds since the epoch."""
1518        try:
1519            self._date = float(date)
1520        except ValueError:
1521            raise TypeError("can't convert to float: %s" % date)
1522
1523    def get_info(self):
1524        """Get the message's "info" as a string."""
1525        return self._info
1526
1527    def set_info(self, info):
1528        """Set the message's "info" string."""
1529        if isinstance(info, str):
1530            self._info = info
1531        else:
1532            raise TypeError('info must be a string: %s' % type(info))
1533
1534    def _explain_to(self, message):
1535        """Copy Maildir-specific state to message insofar as possible."""
1536        if isinstance(message, MaildirMessage):
1537            message.set_flags(self.get_flags())
1538            message.set_subdir(self.get_subdir())
1539            message.set_date(self.get_date())
1540        elif isinstance(message, _mboxMMDFMessage):
1541            flags = set(self.get_flags())
1542            if 'S' in flags:
1543                message.add_flag('R')
1544            if self.get_subdir() == 'cur':
1545                message.add_flag('O')
1546            if 'T' in flags:
1547                message.add_flag('D')
1548            if 'F' in flags:
1549                message.add_flag('F')
1550            if 'R' in flags:
1551                message.add_flag('A')
1552            message.set_from('MAILER-DAEMON', time.gmtime(self.get_date()))
1553        elif isinstance(message, MHMessage):
1554            flags = set(self.get_flags())
1555            if 'S' not in flags:
1556                message.add_sequence('unseen')
1557            if 'R' in flags:
1558                message.add_sequence('replied')
1559            if 'F' in flags:
1560                message.add_sequence('flagged')
1561        elif isinstance(message, BabylMessage):
1562            flags = set(self.get_flags())
1563            if 'S' not in flags:
1564                message.add_label('unseen')
1565            if 'T' in flags:
1566                message.add_label('deleted')
1567            if 'R' in flags:
1568                message.add_label('answered')
1569            if 'P' in flags:
1570                message.add_label('forwarded')
1571        elif isinstance(message, Message):
1572            pass
1573        else:
1574            raise TypeError('Cannot convert to specified type: %s' %
1575                            type(message))
1576
1577
1578class _mboxMMDFMessage(Message):
1579    """Message with mbox- or MMDF-specific properties."""
1580
1581    def __init__(self, message=None):
1582        """Initialize an mboxMMDFMessage instance."""
1583        self.set_from('MAILER-DAEMON', True)
1584        if isinstance(message, email.message.Message):
1585            unixfrom = message.get_unixfrom()
1586            if unixfrom is not None and unixfrom.startswith('From '):
1587                self.set_from(unixfrom[5:])
1588        Message.__init__(self, message)
1589
1590    def get_from(self):
1591        """Return contents of "From " line."""
1592        return self._from
1593
1594    def set_from(self, from_, time_=None):
1595        """Set "From " line, formatting and appending time_ if specified."""
1596        if time_ is not None:
1597            if time_ is True:
1598                time_ = time.gmtime()
1599            from_ += ' ' + time.asctime(time_)
1600        self._from = from_
1601
1602    def get_flags(self):
1603        """Return as a string the flags that are set."""
1604        return self.get('Status', '') + self.get('X-Status', '')
1605
1606    def set_flags(self, flags):
1607        """Set the given flags and unset all others."""
1608        flags = set(flags)
1609        status_flags, xstatus_flags = '', ''
1610        for flag in ('R', 'O'):
1611            if flag in flags:
1612                status_flags += flag
1613                flags.remove(flag)
1614        for flag in ('D', 'F', 'A'):
1615            if flag in flags:
1616                xstatus_flags += flag
1617                flags.remove(flag)
1618        xstatus_flags += ''.join(sorted(flags))
1619        try:
1620            self.replace_header('Status', status_flags)
1621        except KeyError:
1622            self.add_header('Status', status_flags)
1623        try:
1624            self.replace_header('X-Status', xstatus_flags)
1625        except KeyError:
1626            self.add_header('X-Status', xstatus_flags)
1627
1628    def add_flag(self, flag):
1629        """Set the given flag(s) without changing others."""
1630        self.set_flags(''.join(set(self.get_flags()) | set(flag)))
1631
1632    def remove_flag(self, flag):
1633        """Unset the given string flag(s) without changing others."""
1634        if 'Status' in self or 'X-Status' in self:
1635            self.set_flags(''.join(set(self.get_flags()) - set(flag)))
1636
1637    def _explain_to(self, message):
1638        """Copy mbox- or MMDF-specific state to message insofar as possible."""
1639        if isinstance(message, MaildirMessage):
1640            flags = set(self.get_flags())
1641            if 'O' in flags:
1642                message.set_subdir('cur')
1643            if 'F' in flags:
1644                message.add_flag('F')
1645            if 'A' in flags:
1646                message.add_flag('R')
1647            if 'R' in flags:
1648                message.add_flag('S')
1649            if 'D' in flags:
1650                message.add_flag('T')
1651            del message['status']
1652            del message['x-status']
1653            maybe_date = ' '.join(self.get_from().split()[-5:])
1654            try:
1655                message.set_date(calendar.timegm(time.strptime(maybe_date,
1656                                                      '%a %b %d %H:%M:%S %Y')))
1657            except (ValueError, OverflowError):
1658                pass
1659        elif isinstance(message, _mboxMMDFMessage):
1660            message.set_flags(self.get_flags())
1661            message.set_from(self.get_from())
1662        elif isinstance(message, MHMessage):
1663            flags = set(self.get_flags())
1664            if 'R' not in flags:
1665                message.add_sequence('unseen')
1666            if 'A' in flags:
1667                message.add_sequence('replied')
1668            if 'F' in flags:
1669                message.add_sequence('flagged')
1670            del message['status']
1671            del message['x-status']
1672        elif isinstance(message, BabylMessage):
1673            flags = set(self.get_flags())
1674            if 'R' not in flags:
1675                message.add_label('unseen')
1676            if 'D' in flags:
1677                message.add_label('deleted')
1678            if 'A' in flags:
1679                message.add_label('answered')
1680            del message['status']
1681            del message['x-status']
1682        elif isinstance(message, Message):
1683            pass
1684        else:
1685            raise TypeError('Cannot convert to specified type: %s' %
1686                            type(message))
1687
1688
1689class mboxMessage(_mboxMMDFMessage):
1690    """Message with mbox-specific properties."""
1691
1692
1693class MHMessage(Message):
1694    """Message with MH-specific properties."""
1695
1696    def __init__(self, message=None):
1697        """Initialize an MHMessage instance."""
1698        self._sequences = []
1699        Message.__init__(self, message)
1700
1701    def get_sequences(self):
1702        """Return a list of sequences that include the message."""
1703        return self._sequences[:]
1704
1705    def set_sequences(self, sequences):
1706        """Set the list of sequences that include the message."""
1707        self._sequences = list(sequences)
1708
1709    def add_sequence(self, sequence):
1710        """Add sequence to list of sequences including the message."""
1711        if isinstance(sequence, str):
1712            if not sequence in self._sequences:
1713                self._sequences.append(sequence)
1714        else:
1715            raise TypeError('sequence must be a string: %s' % type(sequence))
1716
1717    def remove_sequence(self, sequence):
1718        """Remove sequence from the list of sequences including the message."""
1719        try:
1720            self._sequences.remove(sequence)
1721        except ValueError:
1722            pass
1723
1724    def _explain_to(self, message):
1725        """Copy MH-specific state to message insofar as possible."""
1726        if isinstance(message, MaildirMessage):
1727            sequences = set(self.get_sequences())
1728            if 'unseen' in sequences:
1729                message.set_subdir('cur')
1730            else:
1731                message.set_subdir('cur')
1732                message.add_flag('S')
1733            if 'flagged' in sequences:
1734                message.add_flag('F')
1735            if 'replied' in sequences:
1736                message.add_flag('R')
1737        elif isinstance(message, _mboxMMDFMessage):
1738            sequences = set(self.get_sequences())
1739            if 'unseen' not in sequences:
1740                message.add_flag('RO')
1741            else:
1742                message.add_flag('O')
1743            if 'flagged' in sequences:
1744                message.add_flag('F')
1745            if 'replied' in sequences:
1746                message.add_flag('A')
1747        elif isinstance(message, MHMessage):
1748            for sequence in self.get_sequences():
1749                message.add_sequence(sequence)
1750        elif isinstance(message, BabylMessage):
1751            sequences = set(self.get_sequences())
1752            if 'unseen' in sequences:
1753                message.add_label('unseen')
1754            if 'replied' in sequences:
1755                message.add_label('answered')
1756        elif isinstance(message, Message):
1757            pass
1758        else:
1759            raise TypeError('Cannot convert to specified type: %s' %
1760                            type(message))
1761
1762
1763class BabylMessage(Message):
1764    """Message with Babyl-specific properties."""
1765
1766    def __init__(self, message=None):
1767        """Initialize an BabylMessage instance."""
1768        self._labels = []
1769        self._visible = Message()
1770        Message.__init__(self, message)
1771
1772    def get_labels(self):
1773        """Return a list of labels on the message."""
1774        return self._labels[:]
1775
1776    def set_labels(self, labels):
1777        """Set the list of labels on the message."""
1778        self._labels = list(labels)
1779
1780    def add_label(self, label):
1781        """Add label to list of labels on the message."""
1782        if isinstance(label, str):
1783            if label not in self._labels:
1784                self._labels.append(label)
1785        else:
1786            raise TypeError('label must be a string: %s' % type(label))
1787
1788    def remove_label(self, label):
1789        """Remove label from the list of labels on the message."""
1790        try:
1791            self._labels.remove(label)
1792        except ValueError:
1793            pass
1794
1795    def get_visible(self):
1796        """Return a Message representation of visible headers."""
1797        return Message(self._visible)
1798
1799    def set_visible(self, visible):
1800        """Set the Message representation of visible headers."""
1801        self._visible = Message(visible)
1802
1803    def update_visible(self):
1804        """Update and/or sensibly generate a set of visible headers."""
1805        for header in self._visible.keys():
1806            if header in self:
1807                self._visible.replace_header(header, self[header])
1808            else:
1809                del self._visible[header]
1810        for header in ('Date', 'From', 'Reply-To', 'To', 'CC', 'Subject'):
1811            if header in self and header not in self._visible:
1812                self._visible[header] = self[header]
1813
1814    def _explain_to(self, message):
1815        """Copy Babyl-specific state to message insofar as possible."""
1816        if isinstance(message, MaildirMessage):
1817            labels = set(self.get_labels())
1818            if 'unseen' in labels:
1819                message.set_subdir('cur')
1820            else:
1821                message.set_subdir('cur')
1822                message.add_flag('S')
1823            if 'forwarded' in labels or 'resent' in labels:
1824                message.add_flag('P')
1825            if 'answered' in labels:
1826                message.add_flag('R')
1827            if 'deleted' in labels:
1828                message.add_flag('T')
1829        elif isinstance(message, _mboxMMDFMessage):
1830            labels = set(self.get_labels())
1831            if 'unseen' not in labels:
1832                message.add_flag('RO')
1833            else:
1834                message.add_flag('O')
1835            if 'deleted' in labels:
1836                message.add_flag('D')
1837            if 'answered' in labels:
1838                message.add_flag('A')
1839        elif isinstance(message, MHMessage):
1840            labels = set(self.get_labels())
1841            if 'unseen' in labels:
1842                message.add_sequence('unseen')
1843            if 'answered' in labels:
1844                message.add_sequence('replied')
1845        elif isinstance(message, BabylMessage):
1846            message.set_visible(self.get_visible())
1847            for label in self.get_labels():
1848                message.add_label(label)
1849        elif isinstance(message, Message):
1850            pass
1851        else:
1852            raise TypeError('Cannot convert to specified type: %s' %
1853                            type(message))
1854
1855
1856class MMDFMessage(_mboxMMDFMessage):
1857    """Message with MMDF-specific properties."""
1858
1859
1860class _ProxyFile:
1861    """A read-only wrapper of a file."""
1862
1863    def __init__(self, f, pos=None):
1864        """Initialize a _ProxyFile."""
1865        self._file = f
1866        if pos is None:
1867            self._pos = f.tell()
1868        else:
1869            self._pos = pos
1870
1871    def read(self, size=None):
1872        """Read bytes."""
1873        return self._read(size, self._file.read)
1874
1875    def readline(self, size=None):
1876        """Read a line."""
1877        return self._read(size, self._file.readline)
1878
1879    def readlines(self, sizehint=None):
1880        """Read multiple lines."""
1881        result = []
1882        for line in self:
1883            result.append(line)
1884            if sizehint is not None:
1885                sizehint -= len(line)
1886                if sizehint <= 0:
1887                    break
1888        return result
1889
1890    def __iter__(self):
1891        """Iterate over lines."""
1892        return iter(self.readline, "")
1893
1894    def tell(self):
1895        """Return the position."""
1896        return self._pos
1897
1898    def seek(self, offset, whence=0):
1899        """Change position."""
1900        if whence == 1:
1901            self._file.seek(self._pos)
1902        self._file.seek(offset, whence)
1903        self._pos = self._file.tell()
1904
1905    def close(self):
1906        """Close the file."""
1907        if hasattr(self, '_file'):
1908            if hasattr(self._file, 'close'):
1909                self._file.close()
1910            del self._file
1911
1912    def _read(self, size, read_method):
1913        """Read size bytes using read_method."""
1914        if size is None:
1915            size = -1
1916        self._file.seek(self._pos)
1917        result = read_method(size)
1918        self._pos = self._file.tell()
1919        return result
1920
1921
1922class _PartialFile(_ProxyFile):
1923    """A read-only wrapper of part of a file."""
1924
1925    def __init__(self, f, start=None, stop=None):
1926        """Initialize a _PartialFile."""
1927        _ProxyFile.__init__(self, f, start)
1928        self._start = start
1929        self._stop = stop
1930
1931    def tell(self):
1932        """Return the position with respect to start."""
1933        return _ProxyFile.tell(self) - self._start
1934
1935    def seek(self, offset, whence=0):
1936        """Change position, possibly with respect to start or stop."""
1937        if whence == 0:
1938            self._pos = self._start
1939            whence = 1
1940        elif whence == 2:
1941            self._pos = self._stop
1942            whence = 1
1943        _ProxyFile.seek(self, offset, whence)
1944
1945    def _read(self, size, read_method):
1946        """Read size bytes using read_method, honoring start and stop."""
1947        remaining = self._stop - self._pos
1948        if remaining <= 0:
1949            return ''
1950        if size is None or size < 0 or size > remaining:
1951            size = remaining
1952        return _ProxyFile._read(self, size, read_method)
1953
1954    def close(self):
1955        # do *not* close the underlying file object for partial files,
1956        # since it's global to the mailbox object
1957        if hasattr(self, '_file'):
1958            del self._file
1959
1960
1961def _lock_file(f, dotlock=True):
1962    """Lock file f using lockf and dot locking."""
1963    dotlock_done = False
1964    try:
1965        if fcntl:
1966            try:
1967                fcntl.lockf(f, fcntl.LOCK_EX | fcntl.LOCK_NB)
1968            except IOError, e:
1969                if e.errno in (errno.EAGAIN, errno.EACCES, errno.EROFS):
1970                    raise ExternalClashError('lockf: lock unavailable: %s' %
1971                                             f.name)
1972                else:
1973                    raise
1974        if dotlock:
1975            try:
1976                pre_lock = _create_temporary(f.name + '.lock')
1977                pre_lock.close()
1978            except IOError, e:
1979                if e.errno in (errno.EACCES, errno.EROFS):
1980                    return  # Without write access, just skip dotlocking.
1981                else:
1982                    raise
1983            try:
1984                if hasattr(os, 'link'):
1985                    os.link(pre_lock.name, f.name + '.lock')
1986                    dotlock_done = True
1987                    os.unlink(pre_lock.name)
1988                else:
1989                    os.rename(pre_lock.name, f.name + '.lock')
1990                    dotlock_done = True
1991            except OSError, e:
1992                if e.errno == errno.EEXIST or \
1993                  (os.name == 'os2' and e.errno == errno.EACCES):
1994                    os.remove(pre_lock.name)
1995                    raise ExternalClashError('dot lock unavailable: %s' %
1996                                             f.name)
1997                else:
1998                    raise
1999    except:
2000        if fcntl:
2001            fcntl.lockf(f, fcntl.LOCK_UN)
2002        if dotlock_done:
2003            os.remove(f.name + '.lock')
2004        raise
2005
2006def _unlock_file(f):
2007    """Unlock file f using lockf and dot locking."""
2008    if fcntl:
2009        fcntl.lockf(f, fcntl.LOCK_UN)
2010    if os.path.exists(f.name + '.lock'):
2011        os.remove(f.name + '.lock')
2012
2013def _create_carefully(path):
2014    """Create a file if it doesn't exist and open for reading and writing."""
2015    fd = os.open(path, os.O_CREAT | os.O_EXCL | os.O_RDWR, 0666)
2016    try:
2017        return open(path, 'rb+')
2018    finally:
2019        os.close(fd)
2020
2021def _create_temporary(path):
2022    """Create a temp file based on path and open for reading and writing."""
2023    return _create_carefully('%s.%s.%s.%s' % (path, int(time.time()),
2024                                              socket.gethostname(),
2025                                              os.getpid()))
2026
2027def _sync_flush(f):
2028    """Ensure changes to file f are physically on disk."""
2029    f.flush()
2030    if hasattr(os, 'fsync'):
2031        os.fsync(f.fileno())
2032
2033def _sync_close(f):
2034    """Close file f, ensuring all changes are physically on disk."""
2035    _sync_flush(f)
2036    f.close()
2037
2038## Start: classes from the original module (for backward compatibility).
2039
2040# Note that the Maildir class, whose name is unchanged, itself offers a next()
2041# method for backward compatibility.
2042
2043class _Mailbox:
2044
2045    def __init__(self, fp, factory=rfc822.Message):
2046        self.fp = fp
2047        self.seekp = 0
2048        self.factory = factory
2049
2050    def __iter__(self):
2051        return iter(self.next, None)
2052
2053    def next(self):
2054        while 1:
2055            self.fp.seek(self.seekp)
2056            try:
2057                self._search_start()
2058            except EOFError:
2059                self.seekp = self.fp.tell()
2060                return None
2061            start = self.fp.tell()
2062            self._search_end()
2063            self.seekp = stop = self.fp.tell()
2064            if start != stop:
2065                break
2066        return self.factory(_PartialFile(self.fp, start, stop))
2067
2068# Recommended to use PortableUnixMailbox instead!
2069class UnixMailbox(_Mailbox):
2070
2071    def _search_start(self):
2072        while 1:
2073            pos = self.fp.tell()
2074            line = self.fp.readline()
2075            if not line:
2076                raise EOFError
2077            if line[:5] == 'From ' and self._isrealfromline(line):
2078                self.fp.seek(pos)
2079                return
2080
2081    def _search_end(self):
2082        self.fp.readline()      # Throw away header line
2083        while 1:
2084            pos = self.fp.tell()
2085            line = self.fp.readline()
2086            if not line:
2087                return
2088            if line[:5] == 'From ' and self._isrealfromline(line):
2089                self.fp.seek(pos)
2090                return
2091
2092    # An overridable mechanism to test for From-line-ness.  You can either
2093    # specify a different regular expression or define a whole new
2094    # _isrealfromline() method.  Note that this only gets called for lines
2095    # starting with the 5 characters "From ".
2096    #
2097    # BAW: According to
2098    #http://home.netscape.com/eng/mozilla/2.0/relnotes/demo/content-length.html
2099    # the only portable, reliable way to find message delimiters in a BSD (i.e
2100    # Unix mailbox) style folder is to search for "\n\nFrom .*\n", or at the
2101    # beginning of the file, "^From .*\n".  While _fromlinepattern below seems
2102    # like a good idea, in practice, there are too many variations for more
2103    # strict parsing of the line to be completely accurate.
2104    #
2105    # _strict_isrealfromline() is the old version which tries to do stricter
2106    # parsing of the From_ line.  _portable_isrealfromline() simply returns
2107    # true, since it's never called if the line doesn't already start with
2108    # "From ".
2109    #
2110    # This algorithm, and the way it interacts with _search_start() and
2111    # _search_end() may not be completely correct, because it doesn't check
2112    # that the two characters preceding "From " are \n\n or the beginning of
2113    # the file.  Fixing this would require a more extensive rewrite than is
2114    # necessary.  For convenience, we've added a PortableUnixMailbox class
2115    # which does no checking of the format of the 'From' line.
2116
2117    _fromlinepattern = (r"From \s*[^\s]+\s+\w\w\w\s+\w\w\w\s+\d?\d\s+"
2118                        r"\d?\d:\d\d(:\d\d)?(\s+[^\s]+)?\s+\d\d\d\d\s*"
2119                        r"[^\s]*\s*"
2120                        "$")
2121    _regexp = None
2122
2123    def _strict_isrealfromline(self, line):
2124        if not self._regexp:
2125            import re
2126            self._regexp = re.compile(self._fromlinepattern)
2127        return self._regexp.match(line)
2128
2129    def _portable_isrealfromline(self, line):
2130        return True
2131
2132    _isrealfromline = _strict_isrealfromline
2133
2134
2135class PortableUnixMailbox(UnixMailbox):
2136    _isrealfromline = UnixMailbox._portable_isrealfromline
2137
2138
2139class MmdfMailbox(_Mailbox):
2140
2141    def _search_start(self):
2142        while 1:
2143            line = self.fp.readline()
2144            if not line:
2145                raise EOFError
2146            if line[:5] == '\001\001\001\001\n':
2147                return
2148
2149    def _search_end(self):
2150        while 1:
2151            pos = self.fp.tell()
2152            line = self.fp.readline()
2153            if not line:
2154                return
2155            if line == '\001\001\001\001\n':
2156                self.fp.seek(pos)
2157                return
2158
2159
2160class MHMailbox:
2161
2162    def __init__(self, dirname, factory=rfc822.Message):
2163        import re
2164        pat = re.compile('^[1-9][0-9]*$')
2165        self.dirname = dirname
2166        # the three following lines could be combined into:
2167        # list = map(long, filter(pat.match, os.listdir(self.dirname)))
2168        list = os.listdir(self.dirname)
2169        list = filter(pat.match, list)
2170        list = map(long, list)
2171        list.sort()
2172        # This only works in Python 1.6 or later;
2173        # before that str() added 'L':
2174        self.boxes = map(str, list)
2175        self.boxes.reverse()
2176        self.factory = factory
2177
2178    def __iter__(self):
2179        return iter(self.next, None)
2180
2181    def next(self):
2182        if not self.boxes:
2183            return None
2184        fn = self.boxes.pop()
2185        fp = open(os.path.join(self.dirname, fn))
2186        msg = self.factory(fp)
2187        try:
2188            msg._mh_msgno = fn
2189        except (AttributeError, TypeError):
2190            pass
2191        return msg
2192
2193
2194class BabylMailbox(_Mailbox):
2195
2196    def _search_start(self):
2197        while 1:
2198            line = self.fp.readline()
2199            if not line:
2200                raise EOFError
2201            if line == '*** EOOH ***\n':
2202                return
2203
2204    def _search_end(self):
2205        while 1:
2206            pos = self.fp.tell()
2207            line = self.fp.readline()
2208            if not line:
2209                return
2210            if line == '\037\014\n' or line == '\037':
2211                self.fp.seek(pos)
2212                return
2213
2214## End: classes from the original module (for backward compatibility).
2215
2216
2217class Error(Exception):
2218    """Raised for module-specific errors."""
2219
2220class NoSuchMailboxError(Error):
2221    """The specified mailbox does not exist and won't be created."""
2222
2223class NotEmptyError(Error):
2224    """The specified mailbox is not empty and deletion was requested."""
2225
2226class ExternalClashError(Error):
2227    """Another process caused an action to fail."""
2228
2229class FormatError(Error):
2230    """A file appears to have an invalid format."""
2231