• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1"""Read/write support for Maildir, mbox, MH, Babyl, and MMDF mailboxes."""
2
3# Notes for authors of new mailbox subclasses:
4#
5# Remember to fsync() changes to disk before closing a modified file
6# or returning from a flush() method.  See functions _sync_flush() and
7# _sync_close().
8
9import os
10import time
11import calendar
12import socket
13import errno
14import copy
15import warnings
16import email
17import email.message
18import email.generator
19import io
20import contextlib
21from types import GenericAlias
22try:
23    import fcntl
24except ImportError:
25    fcntl = None
26
27__all__ = ['Mailbox', 'Maildir', 'mbox', 'MH', 'Babyl', 'MMDF',
28           'Message', 'MaildirMessage', 'mboxMessage', 'MHMessage',
29           'BabylMessage', 'MMDFMessage', 'Error', 'NoSuchMailboxError',
30           'NotEmptyError', 'ExternalClashError', 'FormatError']
31
32linesep = os.linesep.encode('ascii')
33
34class Mailbox:
35    """A group of messages in a particular place."""
36
37    def __init__(self, path, factory=None, create=True):
38        """Initialize a Mailbox instance."""
39        self._path = os.path.abspath(os.path.expanduser(path))
40        self._factory = factory
41
42    def add(self, message):
43        """Add message and return assigned key."""
44        raise NotImplementedError('Method must be implemented by subclass')
45
46    def remove(self, key):
47        """Remove the keyed message; raise KeyError if it doesn't exist."""
48        raise NotImplementedError('Method must be implemented by subclass')
49
50    def __delitem__(self, key):
51        self.remove(key)
52
53    def discard(self, key):
54        """If the keyed message exists, remove it."""
55        try:
56            self.remove(key)
57        except KeyError:
58            pass
59
60    def __setitem__(self, key, message):
61        """Replace the keyed message; raise KeyError if it doesn't exist."""
62        raise NotImplementedError('Method must be implemented by subclass')
63
64    def get(self, key, default=None):
65        """Return the keyed message, or default if it doesn't exist."""
66        try:
67            return self.__getitem__(key)
68        except KeyError:
69            return default
70
71    def __getitem__(self, key):
72        """Return the keyed message; raise KeyError if it doesn't exist."""
73        if not self._factory:
74            return self.get_message(key)
75        else:
76            with contextlib.closing(self.get_file(key)) as file:
77                return self._factory(file)
78
79    def get_message(self, key):
80        """Return a Message representation or raise a KeyError."""
81        raise NotImplementedError('Method must be implemented by subclass')
82
83    def get_string(self, key):
84        """Return a string representation or raise a KeyError.
85
86        Uses email.message.Message to create a 7bit clean string
87        representation of the message."""
88        return email.message_from_bytes(self.get_bytes(key)).as_string()
89
90    def get_bytes(self, key):
91        """Return a byte 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 __contains__(self, key):
136        """Return True if the keyed message exists, False otherwise."""
137        raise NotImplementedError('Method must be implemented by subclass')
138
139    def __len__(self):
140        """Return a count of messages in the mailbox."""
141        raise NotImplementedError('Method must be implemented by subclass')
142
143    def clear(self):
144        """Delete all messages."""
145        for key in self.keys():
146            self.discard(key)
147
148    def pop(self, key, default=None):
149        """Delete the keyed message and return it, or default."""
150        try:
151            result = self[key]
152        except KeyError:
153            return default
154        self.discard(key)
155        return result
156
157    def popitem(self):
158        """Delete an arbitrary (key, message) pair and return it."""
159        for key in self.iterkeys():
160            return (key, self.pop(key))     # This is only run once.
161        else:
162            raise KeyError('No messages in mailbox')
163
164    def update(self, arg=None):
165        """Change the messages that correspond to certain keys."""
166        if hasattr(arg, 'iteritems'):
167            source = arg.iteritems()
168        elif hasattr(arg, 'items'):
169            source = arg.items()
170        else:
171            source = arg
172        bad_key = False
173        for key, message in source:
174            try:
175                self[key] = message
176            except KeyError:
177                bad_key = True
178        if bad_key:
179            raise KeyError('No message with key(s)')
180
181    def flush(self):
182        """Write any pending changes to the disk."""
183        raise NotImplementedError('Method must be implemented by subclass')
184
185    def lock(self):
186        """Lock the mailbox."""
187        raise NotImplementedError('Method must be implemented by subclass')
188
189    def unlock(self):
190        """Unlock the mailbox if it is locked."""
191        raise NotImplementedError('Method must be implemented by subclass')
192
193    def close(self):
194        """Flush and close the mailbox."""
195        raise NotImplementedError('Method must be implemented by subclass')
196
197    def _string_to_bytes(self, message):
198        # If a message is not 7bit clean, we refuse to handle it since it
199        # likely came from reading invalid messages in text mode, and that way
200        # lies mojibake.
201        try:
202            return message.encode('ascii')
203        except UnicodeError:
204            raise ValueError("String input must be ASCII-only; "
205                "use bytes or a Message instead")
206
207    # Whether each message must end in a newline
208    _append_newline = False
209
210    def _dump_message(self, message, target, mangle_from_=False):
211        # This assumes the target file is open in binary mode.
212        """Dump message contents to target file."""
213        if isinstance(message, email.message.Message):
214            buffer = io.BytesIO()
215            gen = email.generator.BytesGenerator(buffer, mangle_from_, 0)
216            gen.flatten(message)
217            buffer.seek(0)
218            data = buffer.read()
219            data = data.replace(b'\n', linesep)
220            target.write(data)
221            if self._append_newline and not data.endswith(linesep):
222                # Make sure the message ends with a newline
223                target.write(linesep)
224        elif isinstance(message, (str, bytes, io.StringIO)):
225            if isinstance(message, io.StringIO):
226                warnings.warn("Use of StringIO input is deprecated, "
227                    "use BytesIO instead", DeprecationWarning, 3)
228                message = message.getvalue()
229            if isinstance(message, str):
230                message = self._string_to_bytes(message)
231            if mangle_from_:
232                message = message.replace(b'\nFrom ', b'\n>From ')
233            message = message.replace(b'\n', linesep)
234            target.write(message)
235            if self._append_newline and not message.endswith(linesep):
236                # Make sure the message ends with a newline
237                target.write(linesep)
238        elif hasattr(message, 'read'):
239            if hasattr(message, 'buffer'):
240                warnings.warn("Use of text mode files is deprecated, "
241                    "use a binary mode file instead", DeprecationWarning, 3)
242                message = message.buffer
243            lastline = None
244            while True:
245                line = message.readline()
246                # Universal newline support.
247                if line.endswith(b'\r\n'):
248                    line = line[:-2] + b'\n'
249                elif line.endswith(b'\r'):
250                    line = line[:-1] + b'\n'
251                if not line:
252                    break
253                if mangle_from_ and line.startswith(b'From '):
254                    line = b'>From ' + line[5:]
255                line = line.replace(b'\n', linesep)
256                target.write(line)
257                lastline = line
258            if self._append_newline and lastline and not lastline.endswith(linesep):
259                # Make sure the message ends with a newline
260                target.write(linesep)
261        else:
262            raise TypeError('Invalid message type: %s' % type(message))
263
264    __class_getitem__ = classmethod(GenericAlias)
265
266
267class Maildir(Mailbox):
268    """A qmail-style Maildir mailbox."""
269
270    colon = ':'
271
272    def __init__(self, dirname, factory=None, create=True):
273        """Initialize a Maildir instance."""
274        Mailbox.__init__(self, dirname, factory, create)
275        self._paths = {
276            'tmp': os.path.join(self._path, 'tmp'),
277            'new': os.path.join(self._path, 'new'),
278            'cur': os.path.join(self._path, 'cur'),
279            }
280        if not os.path.exists(self._path):
281            if create:
282                os.mkdir(self._path, 0o700)
283                for path in self._paths.values():
284                    os.mkdir(path, 0o700)
285            else:
286                raise NoSuchMailboxError(self._path)
287        self._toc = {}
288        self._toc_mtimes = {'cur': 0, 'new': 0}
289        self._last_read = 0         # Records last time we read cur/new
290        self._skewfactor = 0.1      # Adjust if os/fs clocks are skewing
291
292    def add(self, message):
293        """Add message and return assigned key."""
294        tmp_file = self._create_tmp()
295        try:
296            self._dump_message(message, tmp_file)
297        except BaseException:
298            tmp_file.close()
299            os.remove(tmp_file.name)
300            raise
301        _sync_close(tmp_file)
302        if isinstance(message, MaildirMessage):
303            subdir = message.get_subdir()
304            suffix = self.colon + message.get_info()
305            if suffix == self.colon:
306                suffix = ''
307        else:
308            subdir = 'new'
309            suffix = ''
310        uniq = os.path.basename(tmp_file.name).split(self.colon)[0]
311        dest = os.path.join(self._path, subdir, uniq + suffix)
312        if isinstance(message, MaildirMessage):
313            os.utime(tmp_file.name,
314                     (os.path.getatime(tmp_file.name), message.get_date()))
315        # No file modification should be done after the file is moved to its
316        # final position in order to prevent race conditions with changes
317        # from other programs
318        try:
319            try:
320                os.link(tmp_file.name, dest)
321            except (AttributeError, PermissionError):
322                os.rename(tmp_file.name, dest)
323            else:
324                os.remove(tmp_file.name)
325        except OSError as e:
326            os.remove(tmp_file.name)
327            if e.errno == errno.EEXIST:
328                raise ExternalClashError('Name clash with existing message: %s'
329                                         % dest)
330            else:
331                raise
332        return uniq
333
334    def remove(self, key):
335        """Remove the keyed message; raise KeyError if it doesn't exist."""
336        os.remove(os.path.join(self._path, self._lookup(key)))
337
338    def discard(self, key):
339        """If the keyed message exists, remove it."""
340        # This overrides an inapplicable implementation in the superclass.
341        try:
342            self.remove(key)
343        except (KeyError, FileNotFoundError):
344            pass
345
346    def __setitem__(self, key, message):
347        """Replace the keyed message; raise KeyError if it doesn't exist."""
348        old_subpath = self._lookup(key)
349        temp_key = self.add(message)
350        temp_subpath = self._lookup(temp_key)
351        if isinstance(message, MaildirMessage):
352            # temp's subdir and suffix were specified by message.
353            dominant_subpath = temp_subpath
354        else:
355            # temp's subdir and suffix were defaults from add().
356            dominant_subpath = old_subpath
357        subdir = os.path.dirname(dominant_subpath)
358        if self.colon in dominant_subpath:
359            suffix = self.colon + dominant_subpath.split(self.colon)[-1]
360        else:
361            suffix = ''
362        self.discard(key)
363        tmp_path = os.path.join(self._path, temp_subpath)
364        new_path = os.path.join(self._path, subdir, key + suffix)
365        if isinstance(message, MaildirMessage):
366            os.utime(tmp_path,
367                     (os.path.getatime(tmp_path), message.get_date()))
368        # No file modification should be done after the file is moved to its
369        # final position in order to prevent race conditions with changes
370        # from other programs
371        os.rename(tmp_path, new_path)
372
373    def get_message(self, key):
374        """Return a Message representation or raise a KeyError."""
375        subpath = self._lookup(key)
376        with open(os.path.join(self._path, subpath), 'rb') as f:
377            if self._factory:
378                msg = self._factory(f)
379            else:
380                msg = MaildirMessage(f)
381        subdir, name = os.path.split(subpath)
382        msg.set_subdir(subdir)
383        if self.colon in name:
384            msg.set_info(name.split(self.colon)[-1])
385        msg.set_date(os.path.getmtime(os.path.join(self._path, subpath)))
386        return msg
387
388    def get_bytes(self, key):
389        """Return a bytes representation or raise a KeyError."""
390        with open(os.path.join(self._path, self._lookup(key)), 'rb') as f:
391            return f.read().replace(linesep, b'\n')
392
393    def get_file(self, key):
394        """Return a file-like representation or raise a KeyError."""
395        f = open(os.path.join(self._path, self._lookup(key)), 'rb')
396        return _ProxyFile(f)
397
398    def iterkeys(self):
399        """Return an iterator over keys."""
400        self._refresh()
401        for key in self._toc:
402            try:
403                self._lookup(key)
404            except KeyError:
405                continue
406            yield key
407
408    def __contains__(self, key):
409        """Return True if the keyed message exists, False otherwise."""
410        self._refresh()
411        return key in self._toc
412
413    def __len__(self):
414        """Return a count of messages in the mailbox."""
415        self._refresh()
416        return len(self._toc)
417
418    def flush(self):
419        """Write any pending changes to disk."""
420        # Maildir changes are always written immediately, so there's nothing
421        # to do.
422        pass
423
424    def lock(self):
425        """Lock the mailbox."""
426        return
427
428    def unlock(self):
429        """Unlock the mailbox if it is locked."""
430        return
431
432    def close(self):
433        """Flush and close the mailbox."""
434        return
435
436    def list_folders(self):
437        """Return a list of folder names."""
438        result = []
439        for entry in os.listdir(self._path):
440            if len(entry) > 1 and entry[0] == '.' and \
441               os.path.isdir(os.path.join(self._path, entry)):
442                result.append(entry[1:])
443        return result
444
445    def get_folder(self, folder):
446        """Return a Maildir instance for the named folder."""
447        return Maildir(os.path.join(self._path, '.' + folder),
448                       factory=self._factory,
449                       create=False)
450
451    def add_folder(self, folder):
452        """Create a folder and return a Maildir instance representing it."""
453        path = os.path.join(self._path, '.' + folder)
454        result = Maildir(path, factory=self._factory)
455        maildirfolder_path = os.path.join(path, 'maildirfolder')
456        if not os.path.exists(maildirfolder_path):
457            os.close(os.open(maildirfolder_path, os.O_CREAT | os.O_WRONLY,
458                0o666))
459        return result
460
461    def remove_folder(self, folder):
462        """Delete the named folder, which must be empty."""
463        path = os.path.join(self._path, '.' + folder)
464        for entry in os.listdir(os.path.join(path, 'new')) + \
465                     os.listdir(os.path.join(path, 'cur')):
466            if len(entry) < 1 or entry[0] != '.':
467                raise NotEmptyError('Folder contains message(s): %s' % folder)
468        for entry in os.listdir(path):
469            if entry != 'new' and entry != 'cur' and entry != 'tmp' and \
470               os.path.isdir(os.path.join(path, entry)):
471                raise NotEmptyError("Folder contains subdirectory '%s': %s" %
472                                    (folder, entry))
473        for root, dirs, files in os.walk(path, topdown=False):
474            for entry in files:
475                os.remove(os.path.join(root, entry))
476            for entry in dirs:
477                os.rmdir(os.path.join(root, entry))
478        os.rmdir(path)
479
480    def clean(self):
481        """Delete old files in "tmp"."""
482        now = time.time()
483        for entry in os.listdir(os.path.join(self._path, 'tmp')):
484            path = os.path.join(self._path, 'tmp', entry)
485            if now - os.path.getatime(path) > 129600:   # 60 * 60 * 36
486                os.remove(path)
487
488    _count = 1  # This is used to generate unique file names.
489
490    def _create_tmp(self):
491        """Create a file in the tmp subdirectory and open and return it."""
492        now = time.time()
493        hostname = socket.gethostname()
494        if '/' in hostname:
495            hostname = hostname.replace('/', r'\057')
496        if ':' in hostname:
497            hostname = hostname.replace(':', r'\072')
498        uniq = "%s.M%sP%sQ%s.%s" % (int(now), int(now % 1 * 1e6), os.getpid(),
499                                    Maildir._count, hostname)
500        path = os.path.join(self._path, 'tmp', uniq)
501        try:
502            os.stat(path)
503        except FileNotFoundError:
504            Maildir._count += 1
505            try:
506                return _create_carefully(path)
507            except FileExistsError:
508                pass
509
510        # Fall through to here if stat succeeded or open raised EEXIST.
511        raise ExternalClashError('Name clash prevented file creation: %s' %
512                                 path)
513
514    def _refresh(self):
515        """Update table of contents mapping."""
516        # If it has been less than two seconds since the last _refresh() call,
517        # we have to unconditionally re-read the mailbox just in case it has
518        # been modified, because os.path.mtime() has a 2 sec resolution in the
519        # most common worst case (FAT) and a 1 sec resolution typically.  This
520        # results in a few unnecessary re-reads when _refresh() is called
521        # multiple times in that interval, but once the clock ticks over, we
522        # will only re-read as needed.  Because the filesystem might be being
523        # served by an independent system with its own clock, we record and
524        # compare with the mtimes from the filesystem.  Because the other
525        # system's clock might be skewing relative to our clock, we add an
526        # extra delta to our wait.  The default is one tenth second, but is an
527        # instance variable and so can be adjusted if dealing with a
528        # particularly skewed or irregular system.
529        if time.time() - self._last_read > 2 + self._skewfactor:
530            refresh = False
531            for subdir in self._toc_mtimes:
532                mtime = os.path.getmtime(self._paths[subdir])
533                if mtime > self._toc_mtimes[subdir]:
534                    refresh = True
535                self._toc_mtimes[subdir] = mtime
536            if not refresh:
537                return
538        # Refresh toc
539        self._toc = {}
540        for subdir in self._toc_mtimes:
541            path = self._paths[subdir]
542            for entry in os.listdir(path):
543                p = os.path.join(path, entry)
544                if os.path.isdir(p):
545                    continue
546                uniq = entry.split(self.colon)[0]
547                self._toc[uniq] = os.path.join(subdir, entry)
548        self._last_read = time.time()
549
550    def _lookup(self, key):
551        """Use TOC to return subpath for given key, or raise a KeyError."""
552        try:
553            if os.path.exists(os.path.join(self._path, self._toc[key])):
554                return self._toc[key]
555        except KeyError:
556            pass
557        self._refresh()
558        try:
559            return self._toc[key]
560        except KeyError:
561            raise KeyError('No message with key: %s' % key) from None
562
563    # This method is for backward compatibility only.
564    def next(self):
565        """Return the next message in a one-time iteration."""
566        if not hasattr(self, '_onetime_keys'):
567            self._onetime_keys = self.iterkeys()
568        while True:
569            try:
570                return self[next(self._onetime_keys)]
571            except StopIteration:
572                return None
573            except KeyError:
574                continue
575
576
577class _singlefileMailbox(Mailbox):
578    """A single-file mailbox."""
579
580    def __init__(self, path, factory=None, create=True):
581        """Initialize a single-file mailbox."""
582        Mailbox.__init__(self, path, factory, create)
583        try:
584            f = open(self._path, 'rb+')
585        except OSError as e:
586            if e.errno == errno.ENOENT:
587                if create:
588                    f = open(self._path, 'wb+')
589                else:
590                    raise NoSuchMailboxError(self._path)
591            elif e.errno in (errno.EACCES, errno.EROFS):
592                f = open(self._path, 'rb')
593            else:
594                raise
595        self._file = f
596        self._toc = None
597        self._next_key = 0
598        self._pending = False       # No changes require rewriting the file.
599        self._pending_sync = False  # No need to sync the file
600        self._locked = False
601        self._file_length = None    # Used to record mailbox size
602
603    def add(self, message):
604        """Add message and return assigned key."""
605        self._lookup()
606        self._toc[self._next_key] = self._append_message(message)
607        self._next_key += 1
608        # _append_message appends the message to the mailbox file. We
609        # don't need a full rewrite + rename, sync is enough.
610        self._pending_sync = True
611        return self._next_key - 1
612
613    def remove(self, key):
614        """Remove the keyed message; raise KeyError if it doesn't exist."""
615        self._lookup(key)
616        del self._toc[key]
617        self._pending = True
618
619    def __setitem__(self, key, message):
620        """Replace the keyed message; raise KeyError if it doesn't exist."""
621        self._lookup(key)
622        self._toc[key] = self._append_message(message)
623        self._pending = True
624
625    def iterkeys(self):
626        """Return an iterator over keys."""
627        self._lookup()
628        yield from self._toc.keys()
629
630    def __contains__(self, key):
631        """Return True if the keyed message exists, False otherwise."""
632        self._lookup()
633        return key in self._toc
634
635    def __len__(self):
636        """Return a count of messages in the mailbox."""
637        self._lookup()
638        return len(self._toc)
639
640    def lock(self):
641        """Lock the mailbox."""
642        if not self._locked:
643            _lock_file(self._file)
644            self._locked = True
645
646    def unlock(self):
647        """Unlock the mailbox if it is locked."""
648        if self._locked:
649            _unlock_file(self._file)
650            self._locked = False
651
652    def flush(self):
653        """Write any pending changes to disk."""
654        if not self._pending:
655            if self._pending_sync:
656                # Messages have only been added, so syncing the file
657                # is enough.
658                _sync_flush(self._file)
659                self._pending_sync = False
660            return
661
662        # In order to be writing anything out at all, self._toc must
663        # already have been generated (and presumably has been modified
664        # by adding or deleting an item).
665        assert self._toc is not None
666
667        # Check length of self._file; if it's changed, some other process
668        # has modified the mailbox since we scanned it.
669        self._file.seek(0, 2)
670        cur_len = self._file.tell()
671        if cur_len != self._file_length:
672            raise ExternalClashError('Size of mailbox file changed '
673                                     '(expected %i, found %i)' %
674                                     (self._file_length, cur_len))
675
676        new_file = _create_temporary(self._path)
677        try:
678            new_toc = {}
679            self._pre_mailbox_hook(new_file)
680            for key in sorted(self._toc.keys()):
681                start, stop = self._toc[key]
682                self._file.seek(start)
683                self._pre_message_hook(new_file)
684                new_start = new_file.tell()
685                while True:
686                    buffer = self._file.read(min(4096,
687                                                 stop - self._file.tell()))
688                    if not buffer:
689                        break
690                    new_file.write(buffer)
691                new_toc[key] = (new_start, new_file.tell())
692                self._post_message_hook(new_file)
693            self._file_length = new_file.tell()
694        except:
695            new_file.close()
696            os.remove(new_file.name)
697            raise
698        _sync_close(new_file)
699        # self._file is about to get replaced, so no need to sync.
700        self._file.close()
701        # Make sure the new file's mode is the same as the old file's
702        mode = os.stat(self._path).st_mode
703        os.chmod(new_file.name, mode)
704        try:
705            os.rename(new_file.name, self._path)
706        except FileExistsError:
707            os.remove(self._path)
708            os.rename(new_file.name, self._path)
709        self._file = open(self._path, 'rb+')
710        self._toc = new_toc
711        self._pending = False
712        self._pending_sync = False
713        if self._locked:
714            _lock_file(self._file, dotlock=False)
715
716    def _pre_mailbox_hook(self, f):
717        """Called before writing the mailbox to file f."""
718        return
719
720    def _pre_message_hook(self, f):
721        """Called before writing each message to file f."""
722        return
723
724    def _post_message_hook(self, f):
725        """Called after writing each message to file f."""
726        return
727
728    def close(self):
729        """Flush and close the mailbox."""
730        try:
731            self.flush()
732        finally:
733            try:
734                if self._locked:
735                    self.unlock()
736            finally:
737                self._file.close()  # Sync has been done by self.flush() above.
738
739    def _lookup(self, key=None):
740        """Return (start, stop) or raise KeyError."""
741        if self._toc is None:
742            self._generate_toc()
743        if key is not None:
744            try:
745                return self._toc[key]
746            except KeyError:
747                raise KeyError('No message with key: %s' % key) from None
748
749    def _append_message(self, message):
750        """Append message to mailbox and return (start, stop) offsets."""
751        self._file.seek(0, 2)
752        before = self._file.tell()
753        if len(self._toc) == 0 and not self._pending:
754            # This is the first message, and the _pre_mailbox_hook
755            # hasn't yet been called. If self._pending is True,
756            # messages have been removed, so _pre_mailbox_hook must
757            # have been called already.
758            self._pre_mailbox_hook(self._file)
759        try:
760            self._pre_message_hook(self._file)
761            offsets = self._install_message(message)
762            self._post_message_hook(self._file)
763        except BaseException:
764            self._file.truncate(before)
765            raise
766        self._file.flush()
767        self._file_length = self._file.tell()  # Record current length of mailbox
768        return offsets
769
770
771
772class _mboxMMDF(_singlefileMailbox):
773    """An mbox or MMDF mailbox."""
774
775    _mangle_from_ = True
776
777    def get_message(self, key):
778        """Return a Message representation or raise a KeyError."""
779        start, stop = self._lookup(key)
780        self._file.seek(start)
781        from_line = self._file.readline().replace(linesep, b'')
782        string = self._file.read(stop - self._file.tell())
783        msg = self._message_factory(string.replace(linesep, b'\n'))
784        msg.set_from(from_line[5:].decode('ascii'))
785        return msg
786
787    def get_string(self, key, from_=False):
788        """Return a string representation or raise a KeyError."""
789        return email.message_from_bytes(
790            self.get_bytes(key, from_)).as_string(unixfrom=from_)
791
792    def get_bytes(self, key, from_=False):
793        """Return a string representation or raise a KeyError."""
794        start, stop = self._lookup(key)
795        self._file.seek(start)
796        if not from_:
797            self._file.readline()
798        string = self._file.read(stop - self._file.tell())
799        return string.replace(linesep, b'\n')
800
801    def get_file(self, key, from_=False):
802        """Return a file-like representation or raise a KeyError."""
803        start, stop = self._lookup(key)
804        self._file.seek(start)
805        if not from_:
806            self._file.readline()
807        return _PartialFile(self._file, self._file.tell(), stop)
808
809    def _install_message(self, message):
810        """Format a message and blindly write to self._file."""
811        from_line = None
812        if isinstance(message, str):
813            message = self._string_to_bytes(message)
814        if isinstance(message, bytes) and message.startswith(b'From '):
815            newline = message.find(b'\n')
816            if newline != -1:
817                from_line = message[:newline]
818                message = message[newline + 1:]
819            else:
820                from_line = message
821                message = b''
822        elif isinstance(message, _mboxMMDFMessage):
823            author = message.get_from().encode('ascii')
824            from_line = b'From ' + author
825        elif isinstance(message, email.message.Message):
826            from_line = message.get_unixfrom()  # May be None.
827            if from_line is not None:
828                from_line = from_line.encode('ascii')
829        if from_line is None:
830            from_line = b'From MAILER-DAEMON ' + time.asctime(time.gmtime()).encode()
831        start = self._file.tell()
832        self._file.write(from_line + linesep)
833        self._dump_message(message, self._file, self._mangle_from_)
834        stop = self._file.tell()
835        return (start, stop)
836
837
838class mbox(_mboxMMDF):
839    """A classic mbox mailbox."""
840
841    _mangle_from_ = True
842
843    # All messages must end in a newline character, and
844    # _post_message_hooks outputs an empty line between messages.
845    _append_newline = True
846
847    def __init__(self, path, factory=None, create=True):
848        """Initialize an mbox mailbox."""
849        self._message_factory = mboxMessage
850        _mboxMMDF.__init__(self, path, factory, create)
851
852    def _post_message_hook(self, f):
853        """Called after writing each message to file f."""
854        f.write(linesep)
855
856    def _generate_toc(self):
857        """Generate key-to-(start, stop) table of contents."""
858        starts, stops = [], []
859        last_was_empty = False
860        self._file.seek(0)
861        while True:
862            line_pos = self._file.tell()
863            line = self._file.readline()
864            if line.startswith(b'From '):
865                if len(stops) < len(starts):
866                    if last_was_empty:
867                        stops.append(line_pos - len(linesep))
868                    else:
869                        # The last line before the "From " line wasn't
870                        # blank, but we consider it a start of a
871                        # message anyway.
872                        stops.append(line_pos)
873                starts.append(line_pos)
874                last_was_empty = False
875            elif not line:
876                if last_was_empty:
877                    stops.append(line_pos - len(linesep))
878                else:
879                    stops.append(line_pos)
880                break
881            elif line == linesep:
882                last_was_empty = True
883            else:
884                last_was_empty = False
885        self._toc = dict(enumerate(zip(starts, stops)))
886        self._next_key = len(self._toc)
887        self._file_length = self._file.tell()
888
889
890class MMDF(_mboxMMDF):
891    """An MMDF mailbox."""
892
893    def __init__(self, path, factory=None, create=True):
894        """Initialize an MMDF mailbox."""
895        self._message_factory = MMDFMessage
896        _mboxMMDF.__init__(self, path, factory, create)
897
898    def _pre_message_hook(self, f):
899        """Called before writing each message to file f."""
900        f.write(b'\001\001\001\001' + linesep)
901
902    def _post_message_hook(self, f):
903        """Called after writing each message to file f."""
904        f.write(linesep + b'\001\001\001\001' + linesep)
905
906    def _generate_toc(self):
907        """Generate key-to-(start, stop) table of contents."""
908        starts, stops = [], []
909        self._file.seek(0)
910        next_pos = 0
911        while True:
912            line_pos = next_pos
913            line = self._file.readline()
914            next_pos = self._file.tell()
915            if line.startswith(b'\001\001\001\001' + linesep):
916                starts.append(next_pos)
917                while True:
918                    line_pos = next_pos
919                    line = self._file.readline()
920                    next_pos = self._file.tell()
921                    if line == b'\001\001\001\001' + linesep:
922                        stops.append(line_pos - len(linesep))
923                        break
924                    elif not line:
925                        stops.append(line_pos)
926                        break
927            elif not line:
928                break
929        self._toc = dict(enumerate(zip(starts, stops)))
930        self._next_key = len(self._toc)
931        self._file.seek(0, 2)
932        self._file_length = self._file.tell()
933
934
935class MH(Mailbox):
936    """An MH mailbox."""
937
938    def __init__(self, path, factory=None, create=True):
939        """Initialize an MH instance."""
940        Mailbox.__init__(self, path, factory, create)
941        if not os.path.exists(self._path):
942            if create:
943                os.mkdir(self._path, 0o700)
944                os.close(os.open(os.path.join(self._path, '.mh_sequences'),
945                                 os.O_CREAT | os.O_EXCL | os.O_WRONLY, 0o600))
946            else:
947                raise NoSuchMailboxError(self._path)
948        self._locked = False
949
950    def add(self, message):
951        """Add message and return assigned key."""
952        keys = self.keys()
953        if len(keys) == 0:
954            new_key = 1
955        else:
956            new_key = max(keys) + 1
957        new_path = os.path.join(self._path, str(new_key))
958        f = _create_carefully(new_path)
959        closed = False
960        try:
961            if self._locked:
962                _lock_file(f)
963            try:
964                try:
965                    self._dump_message(message, f)
966                except BaseException:
967                    # Unlock and close so it can be deleted on Windows
968                    if self._locked:
969                        _unlock_file(f)
970                    _sync_close(f)
971                    closed = True
972                    os.remove(new_path)
973                    raise
974                if isinstance(message, MHMessage):
975                    self._dump_sequences(message, new_key)
976            finally:
977                if self._locked:
978                    _unlock_file(f)
979        finally:
980            if not closed:
981                _sync_close(f)
982        return new_key
983
984    def remove(self, key):
985        """Remove the keyed message; raise KeyError if it doesn't exist."""
986        path = os.path.join(self._path, str(key))
987        try:
988            f = open(path, 'rb+')
989        except OSError as e:
990            if e.errno == errno.ENOENT:
991                raise KeyError('No message with key: %s' % key)
992            else:
993                raise
994        else:
995            f.close()
996            os.remove(path)
997
998    def __setitem__(self, key, message):
999        """Replace the keyed message; raise KeyError if it doesn't exist."""
1000        path = os.path.join(self._path, str(key))
1001        try:
1002            f = open(path, 'rb+')
1003        except OSError as e:
1004            if e.errno == errno.ENOENT:
1005                raise KeyError('No message with key: %s' % key)
1006            else:
1007                raise
1008        try:
1009            if self._locked:
1010                _lock_file(f)
1011            try:
1012                os.close(os.open(path, os.O_WRONLY | os.O_TRUNC))
1013                self._dump_message(message, f)
1014                if isinstance(message, MHMessage):
1015                    self._dump_sequences(message, key)
1016            finally:
1017                if self._locked:
1018                    _unlock_file(f)
1019        finally:
1020            _sync_close(f)
1021
1022    def get_message(self, key):
1023        """Return a Message representation or raise a KeyError."""
1024        try:
1025            if self._locked:
1026                f = open(os.path.join(self._path, str(key)), 'rb+')
1027            else:
1028                f = open(os.path.join(self._path, str(key)), 'rb')
1029        except OSError as e:
1030            if e.errno == errno.ENOENT:
1031                raise KeyError('No message with key: %s' % key)
1032            else:
1033                raise
1034        with f:
1035            if self._locked:
1036                _lock_file(f)
1037            try:
1038                msg = MHMessage(f)
1039            finally:
1040                if self._locked:
1041                    _unlock_file(f)
1042        for name, key_list in self.get_sequences().items():
1043            if key in key_list:
1044                msg.add_sequence(name)
1045        return msg
1046
1047    def get_bytes(self, key):
1048        """Return a bytes representation or raise a KeyError."""
1049        try:
1050            if self._locked:
1051                f = open(os.path.join(self._path, str(key)), 'rb+')
1052            else:
1053                f = open(os.path.join(self._path, str(key)), 'rb')
1054        except OSError as e:
1055            if e.errno == errno.ENOENT:
1056                raise KeyError('No message with key: %s' % key)
1057            else:
1058                raise
1059        with f:
1060            if self._locked:
1061                _lock_file(f)
1062            try:
1063                return f.read().replace(linesep, b'\n')
1064            finally:
1065                if self._locked:
1066                    _unlock_file(f)
1067
1068    def get_file(self, key):
1069        """Return a file-like representation or raise a KeyError."""
1070        try:
1071            f = open(os.path.join(self._path, str(key)), 'rb')
1072        except OSError as e:
1073            if e.errno == errno.ENOENT:
1074                raise KeyError('No message with key: %s' % key)
1075            else:
1076                raise
1077        return _ProxyFile(f)
1078
1079    def iterkeys(self):
1080        """Return an iterator over keys."""
1081        return iter(sorted(int(entry) for entry in os.listdir(self._path)
1082                                      if entry.isdigit()))
1083
1084    def __contains__(self, key):
1085        """Return True if the keyed message exists, False otherwise."""
1086        return os.path.exists(os.path.join(self._path, str(key)))
1087
1088    def __len__(self):
1089        """Return a count of messages in the mailbox."""
1090        return len(list(self.iterkeys()))
1091
1092    def lock(self):
1093        """Lock the mailbox."""
1094        if not self._locked:
1095            self._file = open(os.path.join(self._path, '.mh_sequences'), 'rb+')
1096            _lock_file(self._file)
1097            self._locked = True
1098
1099    def unlock(self):
1100        """Unlock the mailbox if it is locked."""
1101        if self._locked:
1102            _unlock_file(self._file)
1103            _sync_close(self._file)
1104            del self._file
1105            self._locked = False
1106
1107    def flush(self):
1108        """Write any pending changes to the disk."""
1109        return
1110
1111    def close(self):
1112        """Flush and close the mailbox."""
1113        if self._locked:
1114            self.unlock()
1115
1116    def list_folders(self):
1117        """Return a list of folder names."""
1118        result = []
1119        for entry in os.listdir(self._path):
1120            if os.path.isdir(os.path.join(self._path, entry)):
1121                result.append(entry)
1122        return result
1123
1124    def get_folder(self, folder):
1125        """Return an MH instance for the named folder."""
1126        return MH(os.path.join(self._path, folder),
1127                  factory=self._factory, create=False)
1128
1129    def add_folder(self, folder):
1130        """Create a folder and return an MH instance representing it."""
1131        return MH(os.path.join(self._path, folder),
1132                  factory=self._factory)
1133
1134    def remove_folder(self, folder):
1135        """Delete the named folder, which must be empty."""
1136        path = os.path.join(self._path, folder)
1137        entries = os.listdir(path)
1138        if entries == ['.mh_sequences']:
1139            os.remove(os.path.join(path, '.mh_sequences'))
1140        elif entries == []:
1141            pass
1142        else:
1143            raise NotEmptyError('Folder not empty: %s' % self._path)
1144        os.rmdir(path)
1145
1146    def get_sequences(self):
1147        """Return a name-to-key-list dictionary to define each sequence."""
1148        results = {}
1149        with open(os.path.join(self._path, '.mh_sequences'), 'r', encoding='ASCII') as f:
1150            all_keys = set(self.keys())
1151            for line in f:
1152                try:
1153                    name, contents = line.split(':')
1154                    keys = set()
1155                    for spec in contents.split():
1156                        if spec.isdigit():
1157                            keys.add(int(spec))
1158                        else:
1159                            start, stop = (int(x) for x in spec.split('-'))
1160                            keys.update(range(start, stop + 1))
1161                    results[name] = [key for key in sorted(keys) \
1162                                         if key in all_keys]
1163                    if len(results[name]) == 0:
1164                        del results[name]
1165                except ValueError:
1166                    raise FormatError('Invalid sequence specification: %s' %
1167                                      line.rstrip())
1168        return results
1169
1170    def set_sequences(self, sequences):
1171        """Set sequences using the given name-to-key-list dictionary."""
1172        f = open(os.path.join(self._path, '.mh_sequences'), 'r+', encoding='ASCII')
1173        try:
1174            os.close(os.open(f.name, os.O_WRONLY | os.O_TRUNC))
1175            for name, keys in sequences.items():
1176                if len(keys) == 0:
1177                    continue
1178                f.write(name + ':')
1179                prev = None
1180                completing = False
1181                for key in sorted(set(keys)):
1182                    if key - 1 == prev:
1183                        if not completing:
1184                            completing = True
1185                            f.write('-')
1186                    elif completing:
1187                        completing = False
1188                        f.write('%s %s' % (prev, key))
1189                    else:
1190                        f.write(' %s' % key)
1191                    prev = key
1192                if completing:
1193                    f.write(str(prev) + '\n')
1194                else:
1195                    f.write('\n')
1196        finally:
1197            _sync_close(f)
1198
1199    def pack(self):
1200        """Re-name messages to eliminate numbering gaps. Invalidates keys."""
1201        sequences = self.get_sequences()
1202        prev = 0
1203        changes = []
1204        for key in self.iterkeys():
1205            if key - 1 != prev:
1206                changes.append((key, prev + 1))
1207                try:
1208                    os.link(os.path.join(self._path, str(key)),
1209                            os.path.join(self._path, str(prev + 1)))
1210                except (AttributeError, PermissionError):
1211                    os.rename(os.path.join(self._path, str(key)),
1212                              os.path.join(self._path, str(prev + 1)))
1213                else:
1214                    os.unlink(os.path.join(self._path, str(key)))
1215            prev += 1
1216        self._next_key = prev + 1
1217        if len(changes) == 0:
1218            return
1219        for name, key_list in sequences.items():
1220            for old, new in changes:
1221                if old in key_list:
1222                    key_list[key_list.index(old)] = new
1223        self.set_sequences(sequences)
1224
1225    def _dump_sequences(self, message, key):
1226        """Inspect a new MHMessage and update sequences appropriately."""
1227        pending_sequences = message.get_sequences()
1228        all_sequences = self.get_sequences()
1229        for name, key_list in all_sequences.items():
1230            if name in pending_sequences:
1231                key_list.append(key)
1232            elif key in key_list:
1233                del key_list[key_list.index(key)]
1234        for sequence in pending_sequences:
1235            if sequence not in all_sequences:
1236                all_sequences[sequence] = [key]
1237        self.set_sequences(all_sequences)
1238
1239
1240class Babyl(_singlefileMailbox):
1241    """An Rmail-style Babyl mailbox."""
1242
1243    _special_labels = frozenset({'unseen', 'deleted', 'filed', 'answered',
1244                                 'forwarded', 'edited', 'resent'})
1245
1246    def __init__(self, path, factory=None, create=True):
1247        """Initialize a Babyl mailbox."""
1248        _singlefileMailbox.__init__(self, path, factory, create)
1249        self._labels = {}
1250
1251    def add(self, message):
1252        """Add message and return assigned key."""
1253        key = _singlefileMailbox.add(self, message)
1254        if isinstance(message, BabylMessage):
1255            self._labels[key] = message.get_labels()
1256        return key
1257
1258    def remove(self, key):
1259        """Remove the keyed message; raise KeyError if it doesn't exist."""
1260        _singlefileMailbox.remove(self, key)
1261        if key in self._labels:
1262            del self._labels[key]
1263
1264    def __setitem__(self, key, message):
1265        """Replace the keyed message; raise KeyError if it doesn't exist."""
1266        _singlefileMailbox.__setitem__(self, key, message)
1267        if isinstance(message, BabylMessage):
1268            self._labels[key] = message.get_labels()
1269
1270    def get_message(self, key):
1271        """Return a Message representation or raise a KeyError."""
1272        start, stop = self._lookup(key)
1273        self._file.seek(start)
1274        self._file.readline()   # Skip b'1,' line specifying labels.
1275        original_headers = io.BytesIO()
1276        while True:
1277            line = self._file.readline()
1278            if line == b'*** EOOH ***' + linesep or not line:
1279                break
1280            original_headers.write(line.replace(linesep, b'\n'))
1281        visible_headers = io.BytesIO()
1282        while True:
1283            line = self._file.readline()
1284            if line == linesep or not line:
1285                break
1286            visible_headers.write(line.replace(linesep, b'\n'))
1287        # Read up to the stop, or to the end
1288        n = stop - self._file.tell()
1289        assert n >= 0
1290        body = self._file.read(n)
1291        body = body.replace(linesep, b'\n')
1292        msg = BabylMessage(original_headers.getvalue() + body)
1293        msg.set_visible(visible_headers.getvalue())
1294        if key in self._labels:
1295            msg.set_labels(self._labels[key])
1296        return msg
1297
1298    def get_bytes(self, key):
1299        """Return a string representation or raise a KeyError."""
1300        start, stop = self._lookup(key)
1301        self._file.seek(start)
1302        self._file.readline()   # Skip b'1,' line specifying labels.
1303        original_headers = io.BytesIO()
1304        while True:
1305            line = self._file.readline()
1306            if line == b'*** EOOH ***' + linesep or not line:
1307                break
1308            original_headers.write(line.replace(linesep, b'\n'))
1309        while True:
1310            line = self._file.readline()
1311            if line == linesep or not line:
1312                break
1313        headers = original_headers.getvalue()
1314        n = stop - self._file.tell()
1315        assert n >= 0
1316        data = self._file.read(n)
1317        data = data.replace(linesep, b'\n')
1318        return headers + data
1319
1320    def get_file(self, key):
1321        """Return a file-like representation or raise a KeyError."""
1322        return io.BytesIO(self.get_bytes(key).replace(b'\n', linesep))
1323
1324    def get_labels(self):
1325        """Return a list of user-defined labels in the mailbox."""
1326        self._lookup()
1327        labels = set()
1328        for label_list in self._labels.values():
1329            labels.update(label_list)
1330        labels.difference_update(self._special_labels)
1331        return list(labels)
1332
1333    def _generate_toc(self):
1334        """Generate key-to-(start, stop) table of contents."""
1335        starts, stops = [], []
1336        self._file.seek(0)
1337        next_pos = 0
1338        label_lists = []
1339        while True:
1340            line_pos = next_pos
1341            line = self._file.readline()
1342            next_pos = self._file.tell()
1343            if line == b'\037\014' + linesep:
1344                if len(stops) < len(starts):
1345                    stops.append(line_pos - len(linesep))
1346                starts.append(next_pos)
1347                labels = [label.strip() for label
1348                                        in self._file.readline()[1:].split(b',')
1349                                        if label.strip()]
1350                label_lists.append(labels)
1351            elif line == b'\037' or line == b'\037' + linesep:
1352                if len(stops) < len(starts):
1353                    stops.append(line_pos - len(linesep))
1354            elif not line:
1355                stops.append(line_pos - len(linesep))
1356                break
1357        self._toc = dict(enumerate(zip(starts, stops)))
1358        self._labels = dict(enumerate(label_lists))
1359        self._next_key = len(self._toc)
1360        self._file.seek(0, 2)
1361        self._file_length = self._file.tell()
1362
1363    def _pre_mailbox_hook(self, f):
1364        """Called before writing the mailbox to file f."""
1365        babyl = b'BABYL OPTIONS:' + linesep
1366        babyl += b'Version: 5' + linesep
1367        labels = self.get_labels()
1368        labels = (label.encode() for label in labels)
1369        babyl += b'Labels:' + b','.join(labels) + linesep
1370        babyl += b'\037'
1371        f.write(babyl)
1372
1373    def _pre_message_hook(self, f):
1374        """Called before writing each message to file f."""
1375        f.write(b'\014' + linesep)
1376
1377    def _post_message_hook(self, f):
1378        """Called after writing each message to file f."""
1379        f.write(linesep + b'\037')
1380
1381    def _install_message(self, message):
1382        """Write message contents and return (start, stop)."""
1383        start = self._file.tell()
1384        if isinstance(message, BabylMessage):
1385            special_labels = []
1386            labels = []
1387            for label in message.get_labels():
1388                if label in self._special_labels:
1389                    special_labels.append(label)
1390                else:
1391                    labels.append(label)
1392            self._file.write(b'1')
1393            for label in special_labels:
1394                self._file.write(b', ' + label.encode())
1395            self._file.write(b',,')
1396            for label in labels:
1397                self._file.write(b' ' + label.encode() + b',')
1398            self._file.write(linesep)
1399        else:
1400            self._file.write(b'1,,' + linesep)
1401        if isinstance(message, email.message.Message):
1402            orig_buffer = io.BytesIO()
1403            orig_generator = email.generator.BytesGenerator(orig_buffer, False, 0)
1404            orig_generator.flatten(message)
1405            orig_buffer.seek(0)
1406            while True:
1407                line = orig_buffer.readline()
1408                self._file.write(line.replace(b'\n', linesep))
1409                if line == b'\n' or not line:
1410                    break
1411            self._file.write(b'*** EOOH ***' + linesep)
1412            if isinstance(message, BabylMessage):
1413                vis_buffer = io.BytesIO()
1414                vis_generator = email.generator.BytesGenerator(vis_buffer, False, 0)
1415                vis_generator.flatten(message.get_visible())
1416                while True:
1417                    line = vis_buffer.readline()
1418                    self._file.write(line.replace(b'\n', linesep))
1419                    if line == b'\n' or not line:
1420                        break
1421            else:
1422                orig_buffer.seek(0)
1423                while True:
1424                    line = orig_buffer.readline()
1425                    self._file.write(line.replace(b'\n', linesep))
1426                    if line == b'\n' or not line:
1427                        break
1428            while True:
1429                buffer = orig_buffer.read(4096) # Buffer size is arbitrary.
1430                if not buffer:
1431                    break
1432                self._file.write(buffer.replace(b'\n', linesep))
1433        elif isinstance(message, (bytes, str, io.StringIO)):
1434            if isinstance(message, io.StringIO):
1435                warnings.warn("Use of StringIO input is deprecated, "
1436                    "use BytesIO instead", DeprecationWarning, 3)
1437                message = message.getvalue()
1438            if isinstance(message, str):
1439                message = self._string_to_bytes(message)
1440            body_start = message.find(b'\n\n') + 2
1441            if body_start - 2 != -1:
1442                self._file.write(message[:body_start].replace(b'\n', linesep))
1443                self._file.write(b'*** EOOH ***' + linesep)
1444                self._file.write(message[:body_start].replace(b'\n', linesep))
1445                self._file.write(message[body_start:].replace(b'\n', linesep))
1446            else:
1447                self._file.write(b'*** EOOH ***' + linesep + linesep)
1448                self._file.write(message.replace(b'\n', linesep))
1449        elif hasattr(message, 'readline'):
1450            if hasattr(message, 'buffer'):
1451                warnings.warn("Use of text mode files is deprecated, "
1452                    "use a binary mode file instead", DeprecationWarning, 3)
1453                message = message.buffer
1454            original_pos = message.tell()
1455            first_pass = True
1456            while True:
1457                line = message.readline()
1458                # Universal newline support.
1459                if line.endswith(b'\r\n'):
1460                    line = line[:-2] + b'\n'
1461                elif line.endswith(b'\r'):
1462                    line = line[:-1] + b'\n'
1463                self._file.write(line.replace(b'\n', linesep))
1464                if line == b'\n' or not line:
1465                    if first_pass:
1466                        first_pass = False
1467                        self._file.write(b'*** EOOH ***' + linesep)
1468                        message.seek(original_pos)
1469                    else:
1470                        break
1471            while True:
1472                line = message.readline()
1473                if not line:
1474                    break
1475                # Universal newline support.
1476                if line.endswith(b'\r\n'):
1477                    line = line[:-2] + linesep
1478                elif line.endswith(b'\r'):
1479                    line = line[:-1] + linesep
1480                elif line.endswith(b'\n'):
1481                    line = line[:-1] + linesep
1482                self._file.write(line)
1483        else:
1484            raise TypeError('Invalid message type: %s' % type(message))
1485        stop = self._file.tell()
1486        return (start, stop)
1487
1488
1489class Message(email.message.Message):
1490    """Message with mailbox-format-specific properties."""
1491
1492    def __init__(self, message=None):
1493        """Initialize a Message instance."""
1494        if isinstance(message, email.message.Message):
1495            self._become_message(copy.deepcopy(message))
1496            if isinstance(message, Message):
1497                message._explain_to(self)
1498        elif isinstance(message, bytes):
1499            self._become_message(email.message_from_bytes(message))
1500        elif isinstance(message, str):
1501            self._become_message(email.message_from_string(message))
1502        elif isinstance(message, io.TextIOWrapper):
1503            self._become_message(email.message_from_file(message))
1504        elif hasattr(message, "read"):
1505            self._become_message(email.message_from_binary_file(message))
1506        elif message is None:
1507            email.message.Message.__init__(self)
1508        else:
1509            raise TypeError('Invalid message type: %s' % type(message))
1510
1511    def _become_message(self, message):
1512        """Assume the non-format-specific state of message."""
1513        type_specific = getattr(message, '_type_specific_attributes', [])
1514        for name in message.__dict__:
1515            if name not in type_specific:
1516                self.__dict__[name] = message.__dict__[name]
1517
1518    def _explain_to(self, message):
1519        """Copy format-specific state to message insofar as possible."""
1520        if isinstance(message, Message):
1521            return  # There's nothing format-specific to explain.
1522        else:
1523            raise TypeError('Cannot convert to specified type')
1524
1525
1526class MaildirMessage(Message):
1527    """Message with Maildir-specific properties."""
1528
1529    _type_specific_attributes = ['_subdir', '_info', '_date']
1530
1531    def __init__(self, message=None):
1532        """Initialize a MaildirMessage instance."""
1533        self._subdir = 'new'
1534        self._info = ''
1535        self._date = time.time()
1536        Message.__init__(self, message)
1537
1538    def get_subdir(self):
1539        """Return 'new' or 'cur'."""
1540        return self._subdir
1541
1542    def set_subdir(self, subdir):
1543        """Set subdir to 'new' or 'cur'."""
1544        if subdir == 'new' or subdir == 'cur':
1545            self._subdir = subdir
1546        else:
1547            raise ValueError("subdir must be 'new' or 'cur': %s" % subdir)
1548
1549    def get_flags(self):
1550        """Return as a string the flags that are set."""
1551        if self._info.startswith('2,'):
1552            return self._info[2:]
1553        else:
1554            return ''
1555
1556    def set_flags(self, flags):
1557        """Set the given flags and unset all others."""
1558        self._info = '2,' + ''.join(sorted(flags))
1559
1560    def add_flag(self, flag):
1561        """Set the given flag(s) without changing others."""
1562        self.set_flags(''.join(set(self.get_flags()) | set(flag)))
1563
1564    def remove_flag(self, flag):
1565        """Unset the given string flag(s) without changing others."""
1566        if self.get_flags():
1567            self.set_flags(''.join(set(self.get_flags()) - set(flag)))
1568
1569    def get_date(self):
1570        """Return delivery date of message, in seconds since the epoch."""
1571        return self._date
1572
1573    def set_date(self, date):
1574        """Set delivery date of message, in seconds since the epoch."""
1575        try:
1576            self._date = float(date)
1577        except ValueError:
1578            raise TypeError("can't convert to float: %s" % date) from None
1579
1580    def get_info(self):
1581        """Get the message's "info" as a string."""
1582        return self._info
1583
1584    def set_info(self, info):
1585        """Set the message's "info" string."""
1586        if isinstance(info, str):
1587            self._info = info
1588        else:
1589            raise TypeError('info must be a string: %s' % type(info))
1590
1591    def _explain_to(self, message):
1592        """Copy Maildir-specific state to message insofar as possible."""
1593        if isinstance(message, MaildirMessage):
1594            message.set_flags(self.get_flags())
1595            message.set_subdir(self.get_subdir())
1596            message.set_date(self.get_date())
1597        elif isinstance(message, _mboxMMDFMessage):
1598            flags = set(self.get_flags())
1599            if 'S' in flags:
1600                message.add_flag('R')
1601            if self.get_subdir() == 'cur':
1602                message.add_flag('O')
1603            if 'T' in flags:
1604                message.add_flag('D')
1605            if 'F' in flags:
1606                message.add_flag('F')
1607            if 'R' in flags:
1608                message.add_flag('A')
1609            message.set_from('MAILER-DAEMON', time.gmtime(self.get_date()))
1610        elif isinstance(message, MHMessage):
1611            flags = set(self.get_flags())
1612            if 'S' not in flags:
1613                message.add_sequence('unseen')
1614            if 'R' in flags:
1615                message.add_sequence('replied')
1616            if 'F' in flags:
1617                message.add_sequence('flagged')
1618        elif isinstance(message, BabylMessage):
1619            flags = set(self.get_flags())
1620            if 'S' not in flags:
1621                message.add_label('unseen')
1622            if 'T' in flags:
1623                message.add_label('deleted')
1624            if 'R' in flags:
1625                message.add_label('answered')
1626            if 'P' in flags:
1627                message.add_label('forwarded')
1628        elif isinstance(message, Message):
1629            pass
1630        else:
1631            raise TypeError('Cannot convert to specified type: %s' %
1632                            type(message))
1633
1634
1635class _mboxMMDFMessage(Message):
1636    """Message with mbox- or MMDF-specific properties."""
1637
1638    _type_specific_attributes = ['_from']
1639
1640    def __init__(self, message=None):
1641        """Initialize an mboxMMDFMessage instance."""
1642        self.set_from('MAILER-DAEMON', True)
1643        if isinstance(message, email.message.Message):
1644            unixfrom = message.get_unixfrom()
1645            if unixfrom is not None and unixfrom.startswith('From '):
1646                self.set_from(unixfrom[5:])
1647        Message.__init__(self, message)
1648
1649    def get_from(self):
1650        """Return contents of "From " line."""
1651        return self._from
1652
1653    def set_from(self, from_, time_=None):
1654        """Set "From " line, formatting and appending time_ if specified."""
1655        if time_ is not None:
1656            if time_ is True:
1657                time_ = time.gmtime()
1658            from_ += ' ' + time.asctime(time_)
1659        self._from = from_
1660
1661    def get_flags(self):
1662        """Return as a string the flags that are set."""
1663        return self.get('Status', '') + self.get('X-Status', '')
1664
1665    def set_flags(self, flags):
1666        """Set the given flags and unset all others."""
1667        flags = set(flags)
1668        status_flags, xstatus_flags = '', ''
1669        for flag in ('R', 'O'):
1670            if flag in flags:
1671                status_flags += flag
1672                flags.remove(flag)
1673        for flag in ('D', 'F', 'A'):
1674            if flag in flags:
1675                xstatus_flags += flag
1676                flags.remove(flag)
1677        xstatus_flags += ''.join(sorted(flags))
1678        try:
1679            self.replace_header('Status', status_flags)
1680        except KeyError:
1681            self.add_header('Status', status_flags)
1682        try:
1683            self.replace_header('X-Status', xstatus_flags)
1684        except KeyError:
1685            self.add_header('X-Status', xstatus_flags)
1686
1687    def add_flag(self, flag):
1688        """Set the given flag(s) without changing others."""
1689        self.set_flags(''.join(set(self.get_flags()) | set(flag)))
1690
1691    def remove_flag(self, flag):
1692        """Unset the given string flag(s) without changing others."""
1693        if 'Status' in self or 'X-Status' in self:
1694            self.set_flags(''.join(set(self.get_flags()) - set(flag)))
1695
1696    def _explain_to(self, message):
1697        """Copy mbox- or MMDF-specific state to message insofar as possible."""
1698        if isinstance(message, MaildirMessage):
1699            flags = set(self.get_flags())
1700            if 'O' in flags:
1701                message.set_subdir('cur')
1702            if 'F' in flags:
1703                message.add_flag('F')
1704            if 'A' in flags:
1705                message.add_flag('R')
1706            if 'R' in flags:
1707                message.add_flag('S')
1708            if 'D' in flags:
1709                message.add_flag('T')
1710            del message['status']
1711            del message['x-status']
1712            maybe_date = ' '.join(self.get_from().split()[-5:])
1713            try:
1714                message.set_date(calendar.timegm(time.strptime(maybe_date,
1715                                                      '%a %b %d %H:%M:%S %Y')))
1716            except (ValueError, OverflowError):
1717                pass
1718        elif isinstance(message, _mboxMMDFMessage):
1719            message.set_flags(self.get_flags())
1720            message.set_from(self.get_from())
1721        elif isinstance(message, MHMessage):
1722            flags = set(self.get_flags())
1723            if 'R' not in flags:
1724                message.add_sequence('unseen')
1725            if 'A' in flags:
1726                message.add_sequence('replied')
1727            if 'F' in flags:
1728                message.add_sequence('flagged')
1729            del message['status']
1730            del message['x-status']
1731        elif isinstance(message, BabylMessage):
1732            flags = set(self.get_flags())
1733            if 'R' not in flags:
1734                message.add_label('unseen')
1735            if 'D' in flags:
1736                message.add_label('deleted')
1737            if 'A' in flags:
1738                message.add_label('answered')
1739            del message['status']
1740            del message['x-status']
1741        elif isinstance(message, Message):
1742            pass
1743        else:
1744            raise TypeError('Cannot convert to specified type: %s' %
1745                            type(message))
1746
1747
1748class mboxMessage(_mboxMMDFMessage):
1749    """Message with mbox-specific properties."""
1750
1751
1752class MHMessage(Message):
1753    """Message with MH-specific properties."""
1754
1755    _type_specific_attributes = ['_sequences']
1756
1757    def __init__(self, message=None):
1758        """Initialize an MHMessage instance."""
1759        self._sequences = []
1760        Message.__init__(self, message)
1761
1762    def get_sequences(self):
1763        """Return a list of sequences that include the message."""
1764        return self._sequences[:]
1765
1766    def set_sequences(self, sequences):
1767        """Set the list of sequences that include the message."""
1768        self._sequences = list(sequences)
1769
1770    def add_sequence(self, sequence):
1771        """Add sequence to list of sequences including the message."""
1772        if isinstance(sequence, str):
1773            if not sequence in self._sequences:
1774                self._sequences.append(sequence)
1775        else:
1776            raise TypeError('sequence type must be str: %s' % type(sequence))
1777
1778    def remove_sequence(self, sequence):
1779        """Remove sequence from the list of sequences including the message."""
1780        try:
1781            self._sequences.remove(sequence)
1782        except ValueError:
1783            pass
1784
1785    def _explain_to(self, message):
1786        """Copy MH-specific state to message insofar as possible."""
1787        if isinstance(message, MaildirMessage):
1788            sequences = set(self.get_sequences())
1789            if 'unseen' in sequences:
1790                message.set_subdir('cur')
1791            else:
1792                message.set_subdir('cur')
1793                message.add_flag('S')
1794            if 'flagged' in sequences:
1795                message.add_flag('F')
1796            if 'replied' in sequences:
1797                message.add_flag('R')
1798        elif isinstance(message, _mboxMMDFMessage):
1799            sequences = set(self.get_sequences())
1800            if 'unseen' not in sequences:
1801                message.add_flag('RO')
1802            else:
1803                message.add_flag('O')
1804            if 'flagged' in sequences:
1805                message.add_flag('F')
1806            if 'replied' in sequences:
1807                message.add_flag('A')
1808        elif isinstance(message, MHMessage):
1809            for sequence in self.get_sequences():
1810                message.add_sequence(sequence)
1811        elif isinstance(message, BabylMessage):
1812            sequences = set(self.get_sequences())
1813            if 'unseen' in sequences:
1814                message.add_label('unseen')
1815            if 'replied' in sequences:
1816                message.add_label('answered')
1817        elif isinstance(message, Message):
1818            pass
1819        else:
1820            raise TypeError('Cannot convert to specified type: %s' %
1821                            type(message))
1822
1823
1824class BabylMessage(Message):
1825    """Message with Babyl-specific properties."""
1826
1827    _type_specific_attributes = ['_labels', '_visible']
1828
1829    def __init__(self, message=None):
1830        """Initialize a BabylMessage instance."""
1831        self._labels = []
1832        self._visible = Message()
1833        Message.__init__(self, message)
1834
1835    def get_labels(self):
1836        """Return a list of labels on the message."""
1837        return self._labels[:]
1838
1839    def set_labels(self, labels):
1840        """Set the list of labels on the message."""
1841        self._labels = list(labels)
1842
1843    def add_label(self, label):
1844        """Add label to list of labels on the message."""
1845        if isinstance(label, str):
1846            if label not in self._labels:
1847                self._labels.append(label)
1848        else:
1849            raise TypeError('label must be a string: %s' % type(label))
1850
1851    def remove_label(self, label):
1852        """Remove label from the list of labels on the message."""
1853        try:
1854            self._labels.remove(label)
1855        except ValueError:
1856            pass
1857
1858    def get_visible(self):
1859        """Return a Message representation of visible headers."""
1860        return Message(self._visible)
1861
1862    def set_visible(self, visible):
1863        """Set the Message representation of visible headers."""
1864        self._visible = Message(visible)
1865
1866    def update_visible(self):
1867        """Update and/or sensibly generate a set of visible headers."""
1868        for header in self._visible.keys():
1869            if header in self:
1870                self._visible.replace_header(header, self[header])
1871            else:
1872                del self._visible[header]
1873        for header in ('Date', 'From', 'Reply-To', 'To', 'CC', 'Subject'):
1874            if header in self and header not in self._visible:
1875                self._visible[header] = self[header]
1876
1877    def _explain_to(self, message):
1878        """Copy Babyl-specific state to message insofar as possible."""
1879        if isinstance(message, MaildirMessage):
1880            labels = set(self.get_labels())
1881            if 'unseen' in labels:
1882                message.set_subdir('cur')
1883            else:
1884                message.set_subdir('cur')
1885                message.add_flag('S')
1886            if 'forwarded' in labels or 'resent' in labels:
1887                message.add_flag('P')
1888            if 'answered' in labels:
1889                message.add_flag('R')
1890            if 'deleted' in labels:
1891                message.add_flag('T')
1892        elif isinstance(message, _mboxMMDFMessage):
1893            labels = set(self.get_labels())
1894            if 'unseen' not in labels:
1895                message.add_flag('RO')
1896            else:
1897                message.add_flag('O')
1898            if 'deleted' in labels:
1899                message.add_flag('D')
1900            if 'answered' in labels:
1901                message.add_flag('A')
1902        elif isinstance(message, MHMessage):
1903            labels = set(self.get_labels())
1904            if 'unseen' in labels:
1905                message.add_sequence('unseen')
1906            if 'answered' in labels:
1907                message.add_sequence('replied')
1908        elif isinstance(message, BabylMessage):
1909            message.set_visible(self.get_visible())
1910            for label in self.get_labels():
1911                message.add_label(label)
1912        elif isinstance(message, Message):
1913            pass
1914        else:
1915            raise TypeError('Cannot convert to specified type: %s' %
1916                            type(message))
1917
1918
1919class MMDFMessage(_mboxMMDFMessage):
1920    """Message with MMDF-specific properties."""
1921
1922
1923class _ProxyFile:
1924    """A read-only wrapper of a file."""
1925
1926    def __init__(self, f, pos=None):
1927        """Initialize a _ProxyFile."""
1928        self._file = f
1929        if pos is None:
1930            self._pos = f.tell()
1931        else:
1932            self._pos = pos
1933
1934    def read(self, size=None):
1935        """Read bytes."""
1936        return self._read(size, self._file.read)
1937
1938    def read1(self, size=None):
1939        """Read bytes."""
1940        return self._read(size, self._file.read1)
1941
1942    def readline(self, size=None):
1943        """Read a line."""
1944        return self._read(size, self._file.readline)
1945
1946    def readlines(self, sizehint=None):
1947        """Read multiple lines."""
1948        result = []
1949        for line in self:
1950            result.append(line)
1951            if sizehint is not None:
1952                sizehint -= len(line)
1953                if sizehint <= 0:
1954                    break
1955        return result
1956
1957    def __iter__(self):
1958        """Iterate over lines."""
1959        while True:
1960            line = self.readline()
1961            if not line:
1962                return
1963            yield line
1964
1965    def tell(self):
1966        """Return the position."""
1967        return self._pos
1968
1969    def seek(self, offset, whence=0):
1970        """Change position."""
1971        if whence == 1:
1972            self._file.seek(self._pos)
1973        self._file.seek(offset, whence)
1974        self._pos = self._file.tell()
1975
1976    def close(self):
1977        """Close the file."""
1978        if hasattr(self, '_file'):
1979            try:
1980                if hasattr(self._file, 'close'):
1981                    self._file.close()
1982            finally:
1983                del self._file
1984
1985    def _read(self, size, read_method):
1986        """Read size bytes using read_method."""
1987        if size is None:
1988            size = -1
1989        self._file.seek(self._pos)
1990        result = read_method(size)
1991        self._pos = self._file.tell()
1992        return result
1993
1994    def __enter__(self):
1995        """Context management protocol support."""
1996        return self
1997
1998    def __exit__(self, *exc):
1999        self.close()
2000
2001    def readable(self):
2002        return self._file.readable()
2003
2004    def writable(self):
2005        return self._file.writable()
2006
2007    def seekable(self):
2008        return self._file.seekable()
2009
2010    def flush(self):
2011        return self._file.flush()
2012
2013    @property
2014    def closed(self):
2015        if not hasattr(self, '_file'):
2016            return True
2017        if not hasattr(self._file, 'closed'):
2018            return False
2019        return self._file.closed
2020
2021    __class_getitem__ = classmethod(GenericAlias)
2022
2023
2024class _PartialFile(_ProxyFile):
2025    """A read-only wrapper of part of a file."""
2026
2027    def __init__(self, f, start=None, stop=None):
2028        """Initialize a _PartialFile."""
2029        _ProxyFile.__init__(self, f, start)
2030        self._start = start
2031        self._stop = stop
2032
2033    def tell(self):
2034        """Return the position with respect to start."""
2035        return _ProxyFile.tell(self) - self._start
2036
2037    def seek(self, offset, whence=0):
2038        """Change position, possibly with respect to start or stop."""
2039        if whence == 0:
2040            self._pos = self._start
2041            whence = 1
2042        elif whence == 2:
2043            self._pos = self._stop
2044            whence = 1
2045        _ProxyFile.seek(self, offset, whence)
2046
2047    def _read(self, size, read_method):
2048        """Read size bytes using read_method, honoring start and stop."""
2049        remaining = self._stop - self._pos
2050        if remaining <= 0:
2051            return b''
2052        if size is None or size < 0 or size > remaining:
2053            size = remaining
2054        return _ProxyFile._read(self, size, read_method)
2055
2056    def close(self):
2057        # do *not* close the underlying file object for partial files,
2058        # since it's global to the mailbox object
2059        if hasattr(self, '_file'):
2060            del self._file
2061
2062
2063def _lock_file(f, dotlock=True):
2064    """Lock file f using lockf and dot locking."""
2065    dotlock_done = False
2066    try:
2067        if fcntl:
2068            try:
2069                fcntl.lockf(f, fcntl.LOCK_EX | fcntl.LOCK_NB)
2070            except OSError as e:
2071                if e.errno in (errno.EAGAIN, errno.EACCES, errno.EROFS):
2072                    raise ExternalClashError('lockf: lock unavailable: %s' %
2073                                             f.name)
2074                else:
2075                    raise
2076        if dotlock:
2077            try:
2078                pre_lock = _create_temporary(f.name + '.lock')
2079                pre_lock.close()
2080            except OSError as e:
2081                if e.errno in (errno.EACCES, errno.EROFS):
2082                    return  # Without write access, just skip dotlocking.
2083                else:
2084                    raise
2085            try:
2086                try:
2087                    os.link(pre_lock.name, f.name + '.lock')
2088                    dotlock_done = True
2089                except (AttributeError, PermissionError):
2090                    os.rename(pre_lock.name, f.name + '.lock')
2091                    dotlock_done = True
2092                else:
2093                    os.unlink(pre_lock.name)
2094            except FileExistsError:
2095                os.remove(pre_lock.name)
2096                raise ExternalClashError('dot lock unavailable: %s' %
2097                                         f.name)
2098    except:
2099        if fcntl:
2100            fcntl.lockf(f, fcntl.LOCK_UN)
2101        if dotlock_done:
2102            os.remove(f.name + '.lock')
2103        raise
2104
2105def _unlock_file(f):
2106    """Unlock file f using lockf and dot locking."""
2107    if fcntl:
2108        fcntl.lockf(f, fcntl.LOCK_UN)
2109    if os.path.exists(f.name + '.lock'):
2110        os.remove(f.name + '.lock')
2111
2112def _create_carefully(path):
2113    """Create a file if it doesn't exist and open for reading and writing."""
2114    fd = os.open(path, os.O_CREAT | os.O_EXCL | os.O_RDWR, 0o666)
2115    try:
2116        return open(path, 'rb+')
2117    finally:
2118        os.close(fd)
2119
2120def _create_temporary(path):
2121    """Create a temp file based on path and open for reading and writing."""
2122    return _create_carefully('%s.%s.%s.%s' % (path, int(time.time()),
2123                                              socket.gethostname(),
2124                                              os.getpid()))
2125
2126def _sync_flush(f):
2127    """Ensure changes to file f are physically on disk."""
2128    f.flush()
2129    if hasattr(os, 'fsync'):
2130        os.fsync(f.fileno())
2131
2132def _sync_close(f):
2133    """Close file f, ensuring all changes are physically on disk."""
2134    _sync_flush(f)
2135    f.close()
2136
2137
2138class Error(Exception):
2139    """Raised for module-specific errors."""
2140
2141class NoSuchMailboxError(Error):
2142    """The specified mailbox does not exist and won't be created."""
2143
2144class NotEmptyError(Error):
2145    """The specified mailbox is not empty and deletion was requested."""
2146
2147class ExternalClashError(Error):
2148    """Another process caused an action to fail."""
2149
2150class FormatError(Error):
2151    """A file appears to have an invalid format."""
2152