/Lib/mailbox.py

Large files files are truncated, but you can click here to view the full file

   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 rfc822
  22import StringIO
  23try:
  24    if sys.platform == 'os2emx':
  25        # OS/2 EMX fcntl() not adequate
  26        raise ImportError
  27    import fcntl
  28except ImportError:
  29    fcntl = None
  30
  31__all__ = [ 'Mailbox', 'Maildir', 'mbox', 'MH', 'Babyl', 'MMDF',
  32            'Message', 'MaildirMessage', 'mboxMessage', 'MHMessage',
  33            'BabylMessage', 'MMDFMessage', 'UnixMailbox',
  34            'PortableUnixMailbox', 'MmdfMailbox', 'MHMailbox', 'BabylMailbox' ]
  35
  36class Mailbox:
  37    """A group of messages in a particular place."""
  38
  39    def __init__(self, path, factory=None, create=True):
  40        """Initialize a Mailbox instance."""
  41        self._path = os.path.abspath(os.path.expanduser(path))
  42        self._factory = factory
  43
  44    def add(self, message):
  45        """Add message and return assigned key."""
  46        raise NotImplementedError('Method must be implemented by subclass')
  47
  48    def remove(self, key):
  49        """Remove the keyed message; raise KeyError if it doesn't exist."""
  50        raise NotImplementedError('Method must be implemented by subclass')
  51
  52    def __delitem__(self, key):
  53        self.remove(key)
  54
  55    def discard(self, key):
  56        """If the keyed message exists, remove it."""
  57        try:
  58            self.remove(key)
  59        except KeyError:
  60            pass
  61
  62    def __setitem__(self, key, message):
  63        """Replace the keyed message; raise KeyError if it doesn't exist."""
  64        raise NotImplementedError('Method must be implemented by subclass')
  65
  66    def get(self, key, default=None):
  67        """Return the keyed message, or default if it doesn't exist."""
  68        try:
  69            return self.__getitem__(key)
  70        except KeyError:
  71            return default
  72
  73    def __getitem__(self, key):
  74        """Return the keyed message; raise KeyError if it doesn't exist."""
  75        if not self._factory:
  76            return self.get_message(key)
  77        else:
  78            return self._factory(self.get_file(key))
  79
  80    def get_message(self, key):
  81        """Return a Message representation or raise a KeyError."""
  82        raise NotImplementedError('Method must be implemented by subclass')
  83
  84    def get_string(self, key):
  85        """Return a string representation or raise a KeyError."""
  86        raise NotImplementedError('Method must be implemented by subclass')
  87
  88    def get_file(self, key):
  89        """Return a file-like representation or raise a KeyError."""
  90        raise NotImplementedError('Method must be implemented by subclass')
  91
  92    def iterkeys(self):
  93        """Return an iterator over keys."""
  94        raise NotImplementedError('Method must be implemented by subclass')
  95
  96    def keys(self):
  97        """Return a list of keys."""
  98        return list(self.iterkeys())
  99
 100    def itervalues(self):
 101        """Return an iterator over all messages."""
 102        for key in self.iterkeys():
 103            try:
 104                value = self[key]
 105            except KeyError:
 106                continue
 107            yield value
 108
 109    def __iter__(self):
 110        return self.itervalues()
 111
 112    def values(self):
 113        """Return a list of messages. Memory intensive."""
 114        return list(self.itervalues())
 115
 116    def iteritems(self):
 117        """Return an iterator over (key, message) tuples."""
 118        for key in self.iterkeys():
 119            try:
 120                value = self[key]
 121            except KeyError:
 122                continue
 123            yield (key, value)
 124
 125    def items(self):
 126        """Return a list of (key, message) tuples. Memory intensive."""
 127        return list(self.iteritems())
 128
 129    def has_key(self, key):
 130        """Return True if the keyed message exists, False otherwise."""
 131        raise NotImplementedError('Method must be implemented by subclass')
 132
 133    def __contains__(self, key):
 134        return self.has_key(key)
 135
 136    def __len__(self):
 137        """Return a count of messages in the mailbox."""
 138        raise NotImplementedError('Method must be implemented by subclass')
 139
 140    def clear(self):
 141        """Delete all messages."""
 142        for key in self.iterkeys():
 143            self.discard(key)
 144
 145    def pop(self, key, default=None):
 146        """Delete the keyed message and return it, or default."""
 147        try:
 148            result = self[key]
 149        except KeyError:
 150            return default
 151        self.discard(key)
 152        return result
 153
 154    def popitem(self):
 155        """Delete an arbitrary (key, message) pair and return it."""
 156        for key in self.iterkeys():
 157            return (key, self.pop(key))     # This is only run once.
 158        else:
 159            raise KeyError('No messages in mailbox')
 160
 161    def update(self, arg=None):
 162        """Change the messages that correspond to certain keys."""
 163        if hasattr(arg, 'iteritems'):
 164            source = arg.iteritems()
 165        elif hasattr(arg, 'items'):
 166            source = arg.items()
 167        else:
 168            source = arg
 169        bad_key = False
 170        for key, message in source:
 171            try:
 172                self[key] = message
 173            except KeyError:
 174                bad_key = True
 175        if bad_key:
 176            raise KeyError('No message with key(s)')
 177
 178    def flush(self):
 179        """Write any pending changes to the disk."""
 180        raise NotImplementedError('Method must be implemented by subclass')
 181
 182    def lock(self):
 183        """Lock the mailbox."""
 184        raise NotImplementedError('Method must be implemented by subclass')
 185
 186    def unlock(self):
 187        """Unlock the mailbox if it is locked."""
 188        raise NotImplementedError('Method must be implemented by subclass')
 189
 190    def close(self):
 191        """Flush and close the mailbox."""
 192        raise NotImplementedError('Method must be implemented by subclass')
 193
 194    def _dump_message(self, message, target, mangle_from_=False):
 195        # Most files are opened in binary mode to allow predictable seeking.
 196        # To get native line endings on disk, the user-friendly \n line endings
 197        # used in strings and by email.Message are translated here.
 198        """Dump message contents to target file."""
 199        if isinstance(message, email.message.Message):
 200            buffer = StringIO.StringIO()
 201            gen = email.generator.Generator(buffer, mangle_from_, 0)
 202            gen.flatten(message)
 203            buffer.seek(0)
 204            target.write(buffer.read().replace('\n', os.linesep))
 205        elif isinstance(message, str):
 206            if mangle_from_:
 207                message = message.replace('\nFrom ', '\n>From ')
 208            message = message.replace('\n', os.linesep)
 209            target.write(message)
 210        elif hasattr(message, 'read'):
 211            while True:
 212                line = message.readline()
 213                if line == '':
 214                    break
 215                if mangle_from_ and line.startswith('From '):
 216                    line = '>From ' + line[5:]
 217                line = line.replace('\n', os.linesep)
 218                target.write(line)
 219        else:
 220            raise TypeError('Invalid message type: %s' % type(message))
 221
 222
 223class Maildir(Mailbox):
 224    """A qmail-style Maildir mailbox."""
 225
 226    colon = ':'
 227
 228    def __init__(self, dirname, factory=rfc822.Message, create=True):
 229        """Initialize a Maildir instance."""
 230        Mailbox.__init__(self, dirname, factory, create)
 231        if not os.path.exists(self._path):
 232            if create:
 233                os.mkdir(self._path, 0700)
 234                os.mkdir(os.path.join(self._path, 'tmp'), 0700)
 235                os.mkdir(os.path.join(self._path, 'new'), 0700)
 236                os.mkdir(os.path.join(self._path, 'cur'), 0700)
 237            else:
 238                raise NoSuchMailboxError(self._path)
 239        self._toc = {}
 240
 241    def add(self, message):
 242        """Add message and return assigned key."""
 243        tmp_file = self._create_tmp()
 244        try:
 245            self._dump_message(message, tmp_file)
 246        finally:
 247            _sync_close(tmp_file)
 248        if isinstance(message, MaildirMessage):
 249            subdir = message.get_subdir()
 250            suffix = self.colon + message.get_info()
 251            if suffix == self.colon:
 252                suffix = ''
 253        else:
 254            subdir = 'new'
 255            suffix = ''
 256        uniq = os.path.basename(tmp_file.name).split(self.colon)[0]
 257        dest = os.path.join(self._path, subdir, uniq + suffix)
 258        try:
 259            if hasattr(os, 'link'):
 260                os.link(tmp_file.name, dest)
 261                os.remove(tmp_file.name)
 262            else:
 263                os.rename(tmp_file.name, dest)
 264        except OSError, e:
 265            os.remove(tmp_file.name)
 266            if e.errno == errno.EEXIST:
 267                raise ExternalClashError('Name clash with existing message: %s'
 268                                         % dest)
 269            else:
 270                raise
 271        if isinstance(message, MaildirMessage):
 272            os.utime(dest, (os.path.getatime(dest), message.get_date()))
 273        return uniq
 274
 275    def remove(self, key):
 276        """Remove the keyed message; raise KeyError if it doesn't exist."""
 277        os.remove(os.path.join(self._path, self._lookup(key)))
 278
 279    def discard(self, key):
 280        """If the keyed message exists, remove it."""
 281        # This overrides an inapplicable implementation in the superclass.
 282        try:
 283            self.remove(key)
 284        except KeyError:
 285            pass
 286        except OSError, e:
 287            if e.errno != errno.ENOENT:
 288                raise
 289
 290    def __setitem__(self, key, message):
 291        """Replace the keyed message; raise KeyError if it doesn't exist."""
 292        old_subpath = self._lookup(key)
 293        temp_key = self.add(message)
 294        temp_subpath = self._lookup(temp_key)
 295        if isinstance(message, MaildirMessage):
 296            # temp's subdir and suffix were specified by message.
 297            dominant_subpath = temp_subpath
 298        else:
 299            # temp's subdir and suffix were defaults from add().
 300            dominant_subpath = old_subpath
 301        subdir = os.path.dirname(dominant_subpath)
 302        if self.colon in dominant_subpath:
 303            suffix = self.colon + dominant_subpath.split(self.colon)[-1]
 304        else:
 305            suffix = ''
 306        self.discard(key)
 307        new_path = os.path.join(self._path, subdir, key + suffix)
 308        os.rename(os.path.join(self._path, temp_subpath), new_path)
 309        if isinstance(message, MaildirMessage):
 310            os.utime(new_path, (os.path.getatime(new_path),
 311                                message.get_date()))
 312
 313    def get_message(self, key):
 314        """Return a Message representation or raise a KeyError."""
 315        subpath = self._lookup(key)
 316        f = open(os.path.join(self._path, subpath), 'r')
 317        try:
 318            if self._factory:
 319                msg = self._factory(f)
 320            else:
 321                msg = MaildirMessage(f)
 322        finally:
 323            f.close()
 324        subdir, name = os.path.split(subpath)
 325        msg.set_subdir(subdir)
 326        if self.colon in name:
 327            msg.set_info(name.split(self.colon)[-1])
 328        msg.set_date(os.path.getmtime(os.path.join(self._path, subpath)))
 329        return msg
 330
 331    def get_string(self, key):
 332        """Return a string representation or raise a KeyError."""
 333        f = open(os.path.join(self._path, self._lookup(key)), 'r')
 334        try:
 335            return f.read()
 336        finally:
 337            f.close()
 338
 339    def get_file(self, key):
 340        """Return a file-like representation or raise a KeyError."""
 341        f = open(os.path.join(self._path, self._lookup(key)), 'rb')
 342        return _ProxyFile(f)
 343
 344    def iterkeys(self):
 345        """Return an iterator over keys."""
 346        self._refresh()
 347        for key in self._toc:
 348            try:
 349                self._lookup(key)
 350            except KeyError:
 351                continue
 352            yield key
 353
 354    def has_key(self, key):
 355        """Return True if the keyed message exists, False otherwise."""
 356        self._refresh()
 357        return key in self._toc
 358
 359    def __len__(self):
 360        """Return a count of messages in the mailbox."""
 361        self._refresh()
 362        return len(self._toc)
 363
 364    def flush(self):
 365        """Write any pending changes to disk."""
 366        return  # Maildir changes are always written immediately.
 367
 368    def lock(self):
 369        """Lock the mailbox."""
 370        return
 371
 372    def unlock(self):
 373        """Unlock the mailbox if it is locked."""
 374        return
 375
 376    def close(self):
 377        """Flush and close the mailbox."""
 378        return
 379
 380    def list_folders(self):
 381        """Return a list of folder names."""
 382        result = []
 383        for entry in os.listdir(self._path):
 384            if len(entry) > 1 and entry[0] == '.' and \
 385               os.path.isdir(os.path.join(self._path, entry)):
 386                result.append(entry[1:])
 387        return result
 388
 389    def get_folder(self, folder):
 390        """Return a Maildir instance for the named folder."""
 391        return Maildir(os.path.join(self._path, '.' + folder),
 392                       factory=self._factory,
 393                       create=False)
 394
 395    def add_folder(self, folder):
 396        """Create a folder and return a Maildir instance representing it."""
 397        path = os.path.join(self._path, '.' + folder)
 398        result = Maildir(path, factory=self._factory)
 399        maildirfolder_path = os.path.join(path, 'maildirfolder')
 400        if not os.path.exists(maildirfolder_path):
 401            os.close(os.open(maildirfolder_path, os.O_CREAT | os.O_WRONLY,
 402                0666))
 403        return result
 404
 405    def remove_folder(self, folder):
 406        """Delete the named folder, which must be empty."""
 407        path = os.path.join(self._path, '.' + folder)
 408        for entry in os.listdir(os.path.join(path, 'new')) + \
 409                     os.listdir(os.path.join(path, 'cur')):
 410            if len(entry) < 1 or entry[0] != '.':
 411                raise NotEmptyError('Folder contains message(s): %s' % folder)
 412        for entry in os.listdir(path):
 413            if entry != 'new' and entry != 'cur' and entry != 'tmp' and \
 414               os.path.isdir(os.path.join(path, entry)):
 415                raise NotEmptyError("Folder contains subdirectory '%s': %s" %
 416                                    (folder, entry))
 417        for root, dirs, files in os.walk(path, topdown=False):
 418            for entry in files:
 419                os.remove(os.path.join(root, entry))
 420            for entry in dirs:
 421                os.rmdir(os.path.join(root, entry))
 422        os.rmdir(path)
 423
 424    def clean(self):
 425        """Delete old files in "tmp"."""
 426        now = time.time()
 427        for entry in os.listdir(os.path.join(self._path, 'tmp')):
 428            path = os.path.join(self._path, 'tmp', entry)
 429            if now - os.path.getatime(path) > 129600:   # 60 * 60 * 36
 430                os.remove(path)
 431
 432    _count = 1  # This is used to generate unique file names.
 433
 434    def _create_tmp(self):
 435        """Create a file in the tmp subdirectory and open and return it."""
 436        now = time.time()
 437        hostname = socket.gethostname()
 438        if '/' in hostname:
 439            hostname = hostname.replace('/', r'\057')
 440        if ':' in hostname:
 441            hostname = hostname.replace(':', r'\072')
 442        uniq = "%s.M%sP%sQ%s.%s" % (int(now), int(now % 1 * 1e6), os.getpid(),
 443                                    Maildir._count, hostname)
 444        path = os.path.join(self._path, 'tmp', uniq)
 445        try:
 446            os.stat(path)
 447        except OSError, e:
 448            if e.errno == errno.ENOENT:
 449                Maildir._count += 1
 450                try:
 451                    return _create_carefully(path)
 452                except OSError, e:
 453                    if e.errno != errno.EEXIST:
 454                        raise
 455            else:
 456                raise
 457
 458        # Fall through to here if stat succeeded or open raised EEXIST.
 459        raise ExternalClashError('Name clash prevented file creation: %s' %
 460                                 path)
 461
 462    def _refresh(self):
 463        """Update table of contents mapping."""
 464        self._toc = {}
 465        for subdir in ('new', 'cur'):
 466            subdir_path = os.path.join(self._path, subdir)
 467            for entry in os.listdir(subdir_path):
 468                p = os.path.join(subdir_path, entry)
 469                if os.path.isdir(p):
 470                    continue
 471                uniq = entry.split(self.colon)[0]
 472                self._toc[uniq] = os.path.join(subdir, entry)
 473
 474    def _lookup(self, key):
 475        """Use TOC to return subpath for given key, or raise a KeyError."""
 476        try:
 477            if os.path.exists(os.path.join(self._path, self._toc[key])):
 478                return self._toc[key]
 479        except KeyError:
 480            pass
 481        self._refresh()
 482        try:
 483            return self._toc[key]
 484        except KeyError:
 485            raise KeyError('No message with key: %s' % key)
 486
 487    # This method is for backward compatibility only.
 488    def next(self):
 489        """Return the next message in a one-time iteration."""
 490        if not hasattr(self, '_onetime_keys'):
 491            self._onetime_keys = self.iterkeys()
 492        while True:
 493            try:
 494                return self[self._onetime_keys.next()]
 495            except StopIteration:
 496                return None
 497            except KeyError:
 498                continue
 499
 500
 501class _singlefileMailbox(Mailbox):
 502    """A single-file mailbox."""
 503
 504    def __init__(self, path, factory=None, create=True):
 505        """Initialize a single-file mailbox."""
 506        Mailbox.__init__(self, path, factory, create)
 507        try:
 508            f = open(self._path, 'rb+')
 509        except IOError, e:
 510            if e.errno == errno.ENOENT:
 511                if create:
 512                    f = open(self._path, 'wb+')
 513                else:
 514                    raise NoSuchMailboxError(self._path)
 515            elif e.errno == errno.EACCES:
 516                f = open(self._path, 'rb')
 517            else:
 518                raise
 519        self._file = f
 520        self._toc = None
 521        self._next_key = 0
 522        self._pending = False   # No changes require rewriting the file.
 523        self._locked = False
 524        self._file_length = None        # Used to record mailbox size
 525
 526    def add(self, message):
 527        """Add message and return assigned key."""
 528        self._lookup()
 529        self._toc[self._next_key] = self._append_message(message)
 530        self._next_key += 1
 531        self._pending = True
 532        return self._next_key - 1
 533
 534    def remove(self, key):
 535        """Remove the keyed message; raise KeyError if it doesn't exist."""
 536        self._lookup(key)
 537        del self._toc[key]
 538        self._pending = True
 539
 540    def __setitem__(self, key, message):
 541        """Replace the keyed message; raise KeyError if it doesn't exist."""
 542        self._lookup(key)
 543        self._toc[key] = self._append_message(message)
 544        self._pending = True
 545
 546    def iterkeys(self):
 547        """Return an iterator over keys."""
 548        self._lookup()
 549        for key in self._toc.keys():
 550            yield key
 551
 552    def has_key(self, key):
 553        """Return True if the keyed message exists, False otherwise."""
 554        self._lookup()
 555        return key in self._toc
 556
 557    def __len__(self):
 558        """Return a count of messages in the mailbox."""
 559        self._lookup()
 560        return len(self._toc)
 561
 562    def lock(self):
 563        """Lock the mailbox."""
 564        if not self._locked:
 565            _lock_file(self._file)
 566            self._locked = True
 567
 568    def unlock(self):
 569        """Unlock the mailbox if it is locked."""
 570        if self._locked:
 571            _unlock_file(self._file)
 572            self._locked = False
 573
 574    def flush(self):
 575        """Write any pending changes to disk."""
 576        if not self._pending:
 577            return
 578
 579        # In order to be writing anything out at all, self._toc must
 580        # already have been generated (and presumably has been modified
 581        # by adding or deleting an item).
 582        assert self._toc is not None
 583
 584        # Check length of self._file; if it's changed, some other process
 585        # has modified the mailbox since we scanned it.
 586        self._file.seek(0, 2)
 587        cur_len = self._file.tell()
 588        if cur_len != self._file_length:
 589            raise ExternalClashError('Size of mailbox file changed '
 590                                     '(expected %i, found %i)' %
 591                                     (self._file_length, cur_len))
 592
 593        new_file = _create_temporary(self._path)
 594        try:
 595            new_toc = {}
 596            self._pre_mailbox_hook(new_file)
 597            for key in sorted(self._toc.keys()):
 598                start, stop = self._toc[key]
 599                self._file.seek(start)
 600                self._pre_message_hook(new_file)
 601                new_start = new_file.tell()
 602                while True:
 603                    buffer = self._file.read(min(4096,
 604                                                 stop - self._file.tell()))
 605                    if buffer == '':
 606                        break
 607                    new_file.write(buffer)
 608                new_toc[key] = (new_start, new_file.tell())
 609                self._post_message_hook(new_file)
 610        except:
 611            new_file.close()
 612            os.remove(new_file.name)
 613            raise
 614        _sync_close(new_file)
 615        # self._file is about to get replaced, so no need to sync.
 616        self._file.close()
 617        try:
 618            os.rename(new_file.name, self._path)
 619        except OSError, e:
 620            if e.errno == errno.EEXIST or \
 621              (os.name == 'os2' and e.errno == errno.EACCES):
 622                os.remove(self._path)
 623                os.rename(new_file.name, self._path)
 624            else:
 625                raise
 626        self._file = open(self._path, 'rb+')
 627        self._toc = new_toc
 628        self._pending = False
 629        if self._locked:
 630            _lock_file(self._file, dotlock=False)
 631
 632    def _pre_mailbox_hook(self, f):
 633        """Called before writing the mailbox to file f."""
 634        return
 635
 636    def _pre_message_hook(self, f):
 637        """Called before writing each message to file f."""
 638        return
 639
 640    def _post_message_hook(self, f):
 641        """Called after writing each message to file f."""
 642        return
 643
 644    def close(self):
 645        """Flush and close the mailbox."""
 646        self.flush()
 647        if self._locked:
 648            self.unlock()
 649        self._file.close()  # Sync has been done by self.flush() above.
 650
 651    def _lookup(self, key=None):
 652        """Return (start, stop) or raise KeyError."""
 653        if self._toc is None:
 654            self._generate_toc()
 655        if key is not None:
 656            try:
 657                return self._toc[key]
 658            except KeyError:
 659                raise KeyError('No message with key: %s' % key)
 660
 661    def _append_message(self, message):
 662        """Append message to mailbox and return (start, stop) offsets."""
 663        self._file.seek(0, 2)
 664        self._pre_message_hook(self._file)
 665        offsets = self._install_message(message)
 666        self._post_message_hook(self._file)
 667        self._file.flush()
 668        self._file_length = self._file.tell()  # Record current length of mailbox
 669        return offsets
 670
 671
 672
 673class _mboxMMDF(_singlefileMailbox):
 674    """An mbox or MMDF mailbox."""
 675
 676    _mangle_from_ = True
 677
 678    def get_message(self, key):
 679        """Return a Message representation or raise a KeyError."""
 680        start, stop = self._lookup(key)
 681        self._file.seek(start)
 682        from_line = self._file.readline().replace(os.linesep, '')
 683        string = self._file.read(stop - self._file.tell())
 684        msg = self._message_factory(string.replace(os.linesep, '\n'))
 685        msg.set_from(from_line[5:])
 686        return msg
 687
 688    def get_string(self, key, from_=False):
 689        """Return a string representation or raise a KeyError."""
 690        start, stop = self._lookup(key)
 691        self._file.seek(start)
 692        if not from_:
 693            self._file.readline()
 694        string = self._file.read(stop - self._file.tell())
 695        return string.replace(os.linesep, '\n')
 696
 697    def get_file(self, key, from_=False):
 698        """Return a file-like representation or raise a KeyError."""
 699        start, stop = self._lookup(key)
 700        self._file.seek(start)
 701        if not from_:
 702            self._file.readline()
 703        return _PartialFile(self._file, self._file.tell(), stop)
 704
 705    def _install_message(self, message):
 706        """Format a message and blindly write to self._file."""
 707        from_line = None
 708        if isinstance(message, str) and message.startswith('From '):
 709            newline = message.find('\n')
 710            if newline != -1:
 711                from_line = message[:newline]
 712                message = message[newline + 1:]
 713            else:
 714                from_line = message
 715                message = ''
 716        elif isinstance(message, _mboxMMDFMessage):
 717            from_line = 'From ' + message.get_from()
 718        elif isinstance(message, email.message.Message):
 719            from_line = message.get_unixfrom()  # May be None.
 720        if from_line is None:
 721            from_line = 'From MAILER-DAEMON %s' % time.asctime(time.gmtime())
 722        start = self._file.tell()
 723        self._file.write(from_line + os.linesep)
 724        self._dump_message(message, self._file, self._mangle_from_)
 725        stop = self._file.tell()
 726        return (start, stop)
 727
 728
 729class mbox(_mboxMMDF):
 730    """A classic mbox mailbox."""
 731
 732    _mangle_from_ = True
 733
 734    def __init__(self, path, factory=None, create=True):
 735        """Initialize an mbox mailbox."""
 736        self._message_factory = mboxMessage
 737        _mboxMMDF.__init__(self, path, factory, create)
 738
 739    def _pre_message_hook(self, f):
 740        """Called before writing each message to file f."""
 741        if f.tell() != 0:
 742            f.write(os.linesep)
 743
 744    def _generate_toc(self):
 745        """Generate key-to-(start, stop) table of contents."""
 746        starts, stops = [], []
 747        self._file.seek(0)
 748        while True:
 749            line_pos = self._file.tell()
 750            line = self._file.readline()
 751            if line.startswith('From '):
 752                if len(stops) < len(starts):
 753                    stops.append(line_pos - len(os.linesep))
 754                starts.append(line_pos)
 755            elif line == '':
 756                stops.append(line_pos)
 757                break
 758        self._toc = dict(enumerate(zip(starts, stops)))
 759        self._next_key = len(self._toc)
 760        self._file_length = self._file.tell()
 761
 762
 763class MMDF(_mboxMMDF):
 764    """An MMDF mailbox."""
 765
 766    def __init__(self, path, factory=None, create=True):
 767        """Initialize an MMDF mailbox."""
 768        self._message_factory = MMDFMessage
 769        _mboxMMDF.__init__(self, path, factory, create)
 770
 771    def _pre_message_hook(self, f):
 772        """Called before writing each message to file f."""
 773        f.write('\001\001\001\001' + os.linesep)
 774
 775    def _post_message_hook(self, f):
 776        """Called after writing each message to file f."""
 777        f.write(os.linesep + '\001\001\001\001' + os.linesep)
 778
 779    def _generate_toc(self):
 780        """Generate key-to-(start, stop) table of contents."""
 781        starts, stops = [], []
 782        self._file.seek(0)
 783        next_pos = 0
 784        while True:
 785            line_pos = next_pos
 786            line = self._file.readline()
 787            next_pos = self._file.tell()
 788            if line.startswith('\001\001\001\001' + os.linesep):
 789                starts.append(next_pos)
 790                while True:
 791                    line_pos = next_pos
 792                    line = self._file.readline()
 793                    next_pos = self._file.tell()
 794                    if line == '\001\001\001\001' + os.linesep:
 795                        stops.append(line_pos - len(os.linesep))
 796                        break
 797                    elif line == '':
 798                        stops.append(line_pos)
 799                        break
 800            elif line == '':
 801                break
 802        self._toc = dict(enumerate(zip(starts, stops)))
 803        self._next_key = len(self._toc)
 804        self._file.seek(0, 2)
 805        self._file_length = self._file.tell()
 806
 807
 808class MH(Mailbox):
 809    """An MH mailbox."""
 810
 811    def __init__(self, path, factory=None, create=True):
 812        """Initialize an MH instance."""
 813        Mailbox.__init__(self, path, factory, create)
 814        if not os.path.exists(self._path):
 815            if create:
 816                os.mkdir(self._path, 0700)
 817                os.close(os.open(os.path.join(self._path, '.mh_sequences'),
 818                                 os.O_CREAT | os.O_EXCL | os.O_WRONLY, 0600))
 819            else:
 820                raise NoSuchMailboxError(self._path)
 821        self._locked = False
 822
 823    def add(self, message):
 824        """Add message and return assigned key."""
 825        keys = self.keys()
 826        if len(keys) == 0:
 827            new_key = 1
 828        else:
 829            new_key = max(keys) + 1
 830        new_path = os.path.join(self._path, str(new_key))
 831        f = _create_carefully(new_path)
 832        try:
 833            if self._locked:
 834                _lock_file(f)
 835            try:
 836                self._dump_message(message, f)
 837                if isinstance(message, MHMessage):
 838                    self._dump_sequences(message, new_key)
 839            finally:
 840                if self._locked:
 841                    _unlock_file(f)
 842        finally:
 843            _sync_close(f)
 844        return new_key
 845
 846    def remove(self, key):
 847        """Remove the keyed message; raise KeyError if it doesn't exist."""
 848        path = os.path.join(self._path, str(key))
 849        try:
 850            f = open(path, 'rb+')
 851        except IOError, e:
 852            if e.errno == errno.ENOENT:
 853                raise KeyError('No message with key: %s' % key)
 854            else:
 855                raise
 856        try:
 857            if self._locked:
 858                _lock_file(f)
 859            try:
 860                f.close()
 861                os.remove(os.path.join(self._path, str(key)))
 862            finally:
 863                if self._locked:
 864                    _unlock_file(f)
 865        finally:
 866            f.close()
 867
 868    def __setitem__(self, key, message):
 869        """Replace the keyed message; raise KeyError if it doesn't exist."""
 870        path = os.path.join(self._path, str(key))
 871        try:
 872            f = open(path, 'rb+')
 873        except IOError, e:
 874            if e.errno == errno.ENOENT:
 875                raise KeyError('No message with key: %s' % key)
 876            else:
 877                raise
 878        try:
 879            if self._locked:
 880                _lock_file(f)
 881            try:
 882                os.close(os.open(path, os.O_WRONLY | os.O_TRUNC))
 883                self._dump_message(message, f)
 884                if isinstance(message, MHMessage):
 885                    self._dump_sequences(message, key)
 886            finally:
 887                if self._locked:
 888                    _unlock_file(f)
 889        finally:
 890            _sync_close(f)
 891
 892    def get_message(self, key):
 893        """Return a Message representation or raise a KeyError."""
 894        try:
 895            if self._locked:
 896                f = open(os.path.join(self._path, str(key)), 'r+')
 897            else:
 898                f = open(os.path.join(self._path, str(key)), 'r')
 899        except IOError, e:
 900            if e.errno == errno.ENOENT:
 901                raise KeyError('No message with key: %s' % key)
 902            else:
 903                raise
 904        try:
 905            if self._locked:
 906                _lock_file(f)
 907            try:
 908                msg = MHMessage(f)
 909            finally:
 910                if self._locked:
 911                    _unlock_file(f)
 912        finally:
 913            f.close()
 914        for name, key_list in self.get_sequences().iteritems():
 915            if key in key_list:
 916                msg.add_sequence(name)
 917        return msg
 918
 919    def get_string(self, key):
 920        """Return a string representation or raise a KeyError."""
 921        try:
 922            if self._locked:
 923                f = open(os.path.join(self._path, str(key)), 'r+')
 924            else:
 925                f = open(os.path.join(self._path, str(key)), 'r')
 926        except IOError, e:
 927            if e.errno == errno.ENOENT:
 928                raise KeyError('No message with key: %s' % key)
 929            else:
 930                raise
 931        try:
 932            if self._locked:
 933                _lock_file(f)
 934            try:
 935                return f.read()
 936            finally:
 937                if self._locked:
 938                    _unlock_file(f)
 939        finally:
 940            f.close()
 941
 942    def get_file(self, key):
 943        """Return a file-like representation or raise a KeyError."""
 944        try:
 945            f = open(os.path.join(self._path, str(key)), 'rb')
 946        except IOError, e:
 947            if e.errno == errno.ENOENT:
 948                raise KeyError('No message with key: %s' % key)
 949            else:
 950                raise
 951        return _ProxyFile(f)
 952
 953    def iterkeys(self):
 954        """Return an iterator over keys."""
 955        return iter(sorted(int(entry) for entry in os.listdir(self._path)
 956                                      if entry.isdigit()))
 957
 958    def has_key(self, key):
 959        """Return True if the keyed message exists, False otherwise."""
 960        return os.path.exists(os.path.join(self._path, str(key)))
 961
 962    def __len__(self):
 963        """Return a count of messages in the mailbox."""
 964        return len(list(self.iterkeys()))
 965
 966    def lock(self):
 967        """Lock the mailbox."""
 968        if not self._locked:
 969            self._file = open(os.path.join(self._path, '.mh_sequences'), 'rb+')
 970            _lock_file(self._file)
 971            self._locked = True
 972
 973    def unlock(self):
 974        """Unlock the mailbox if it is locked."""
 975        if self._locked:
 976            _unlock_file(self._file)
 977            _sync_close(self._file)
 978            del self._file
 979            self._locked = False
 980
 981    def flush(self):
 982        """Write any pending changes to the disk."""
 983        return
 984
 985    def close(self):
 986        """Flush and close the mailbox."""
 987        if self._locked:
 988            self.unlock()
 989
 990    def list_folders(self):
 991        """Return a list of folder names."""
 992        result = []
 993        for entry in os.listdir(self._path):
 994            if os.path.isdir(os.path.join(self._path, entry)):
 995                result.append(entry)
 996        return result
 997
 998    def get_folder(self, folder):
 999        """Return an MH instance for the named folder."""
1000        return MH(os.path.join(self._path, folder),
1001                  factory=self._factory, create=False)
1002
1003    def add_folder(self, folder):
1004        """Create a folder and return an MH instance representing it."""
1005        return MH(os.path.join(self._path, folder),
1006                  factory=self._factory)
1007
1008    def remove_folder(self, folder):
1009        """Delete the named folder, which must be empty."""
1010        path = os.path.join(self._path, folder)
1011        entries = os.listdir(path)
1012        if entries == ['.mh_sequences']:
1013            os.remove(os.path.join(path, '.mh_sequences'))
1014        elif entries == []:
1015            pass
1016        else:
1017            raise NotEmptyError('Folder not empty: %s' % self._path)
1018        os.rmdir(path)
1019
1020    def get_sequences(self):
1021        """Return a name-to-key-list dictionary to define each sequence."""
1022        results = {}
1023        f = open(os.path.join(self._path, '.mh_sequences'), 'r')
1024        try:
1025            all_keys = set(self.keys())
1026            for line in f:
1027                try:
1028                    name, contents = line.split(':')
1029                    keys = set()
1030                    for spec in contents.split():
1031                        if spec.isdigit():
1032                            keys.add(int(spec))
1033                        else:
1034                            start, stop = (int(x) for x in spec.split('-'))
1035                            keys.update(range(start, stop + 1))
1036                    results[name] = [key for key in sorted(keys) \
1037                                         if key in all_keys]
1038                    if len(results[name]) == 0:
1039                        del results[name]
1040                except ValueError:
1041                    raise FormatError('Invalid sequence specification: %s' %
1042                                      line.rstrip())
1043        finally:
1044            f.close()
1045        return results
1046
1047    def set_sequences(self, sequences):
1048        """Set sequences using the given name-to-key-list dictionary."""
1049        f = open(os.path.join(self._path, '.mh_sequences'), 'r+')
1050        try:
1051            os.close(os.open(f.name, os.O_WRONLY | os.O_TRUNC))
1052            for name, keys in sequences.iteritems():
1053                if len(keys) == 0:
1054                    continue
1055                f.write('%s:' % name)
1056                prev = None
1057                completing = False
1058                for key in sorted(set(keys)):
1059                    if key - 1 == prev:
1060                        if not completing:
1061                            completing = True
1062                            f.write('-')
1063                    elif completing:
1064                        completing = False
1065                        f.write('%s %s' % (prev, key))
1066                    else:
1067                        f.write(' %s' % key)
1068                    prev = key
1069                if completing:
1070                    f.write(str(prev) + '\n')
1071                else:
1072                    f.write('\n')
1073        finally:
1074            _sync_close(f)
1075
1076    def pack(self):
1077        """Re-name messages to eliminate numbering gaps. Invalidates keys."""
1078        sequences = self.get_sequences()
1079        prev = 0
1080        changes = []
1081        for key in self.iterkeys():
1082            if key - 1 != prev:
1083                changes.append((key, prev + 1))
1084                if hasattr(os, 'link'):
1085                    os.link(os.path.join(self._path, str(key)),
1086                            os.path.join(self._path, str(prev + 1)))
1087                    os.unlink(os.path.join(self._path, str(key)))
1088                else:
1089                    os.rename(os.path.join(self._path, str(key)),
1090                              os.path.join(self._path, str(prev + 1)))
1091            prev += 1
1092        self._next_key = prev + 1
1093        if len(changes) == 0:
1094            return
1095        for name, key_list in sequences.items():
1096            for old, new in changes:
1097                if old in key_list:
1098                    key_list[key_list.index(old)] = new
1099        self.set_sequences(sequences)
1100
1101    def _dump_sequences(self, message, key):
1102        """Inspect a new MHMessage and update sequences appropriately."""
1103        pending_sequences = message.get_sequences()
1104        all_sequences = self.get_sequences()
1105        for name, key_list in all_sequences.iteritems():
1106            if name in pending_sequences:
1107                key_list.append(key)
1108            elif key in key_list:
1109                del key_list[key_list.index(key)]
1110        for sequence in pending_sequences:
1111            if sequence not in all_sequences:
1112                all_sequences[sequence] = [key]
1113        self.set_sequences(all_sequences)
1114
1115
1116class Babyl(_singlefileMailbox):
1117    """An Rmail-style Babyl mailbox."""
1118
1119    _special_labels = frozenset(('unseen', 'deleted', 'filed', 'answered',
1120                                 'forwarded', 'edited', 'resent'))
1121
1122    def __init__(self, path, factory=None, create=True):
1123        """Initialize a Babyl mailbox."""
1124        _singlefileMailbox.__init__(self, path, factory, create)
1125        self._labels = {}
1126
1127    def add(self, message):
1128        """Add message and return assigned key."""
1129        key = _singlefileMailbox.add(self, message)
1130        if isinstance(message, BabylMessage):
1131            self._labels[key] = message.get_labels()
1132        return key
1133
1134    def remove(self, key):
1135        """Remove the keyed message; raise KeyError if it doesn't exist."""
1136        _singlefileMailbox.remove(self, key)
1137        if key in self._labels:
1138            del self._labels[key]
1139
1140    def __setitem__(self, key, message):
1141        """Replace the keyed message; raise KeyError if it doesn't exist."""
1142        _singlefileMailbox.__setitem__(self, key, message)
1143        if isinstance(message, BabylMessage):
1144            self._labels[key] = message.get_labels()
1145
1146    def get_message(self, key):
1147        """Return a Message representation or raise a KeyError."""
1148        start, stop = self._lookup(key)
1149        self._file.seek(start)
1150        self._file.readline()   # Skip '1,' line specifying labels.
1151        original_headers = StringIO.StringIO()
1152        while True:
1153            line = self._file.readline()
1154            if line == '*** EOOH ***' + os.linesep or line == '':
1155                break
1156            original_headers.write(line.replace(os.linesep, '\n'))
1157        visible_headers = StringIO.StringIO()
1158        while True:
1159            line = self._file.readline()
1160            if line == os.linesep or line == '':
1161                break
1162            visible_headers.write(line.replace(os.linesep, '\n'))
1163        body = self._file.read(stop - self._file.tell()).replace(os.linesep,
1164                                                                 '\n')
1165        msg = BabylMessage(original_headers.getvalue() + body)
1166        msg.set_visible(visible_headers.getvalue())
1167        if key in self._labels:
1168            msg.set_labels(self._labels[key])
1169        return msg
1170
1171    def get_string(self, key):
1172        """Return a string representation or raise a KeyError."""
1173        start, stop = self._lookup(key)
1174        self._file.seek(start)
1175        self._file.readline()   # Skip '1,' line specifying labels.
1176        original_headers = StringIO.StringIO()
1177        while True:
1178            line = self._file.readline()
1179            if line == '*** EOOH ***' + os.linesep or line == '':
1180                break
1181            original_headers.write(line.replace(os.linesep, '\n'))
1182        while True:
1183            line = self._file.readline()
1184            if line == os.linesep or line == '':
1185                break
1186        return original_headers.getvalue() + \
1187               self._file.read(stop - self._file.tell()).replace(os.linesep,
1188                                                                 '\n')
1189
1190    def get_file(self, key):
1191        """Return a file-like representation or raise a KeyError."""
1192        return StringIO.StringIO(self.get_string(key).replace('\n',
1193                                                              os.linesep))
1194
1195    def get_labels(self):
1196        """Return a list of user-defined labels in the mailbox."""
1197        self._lookup()
1198        labels = set()
1199        for label_list in self._labels.values():
1200            labels.update(label_list)
1201        labels.difference_update(self._special_labels)
1202        return list(labels)
1203
1204    def _generate_toc(self):
1205        """Generate key-to-(start, stop) table of contents."""
1206        starts, stops = [], []
1207        self._file.seek(0)
1208        next_pos = 0
1209        label_lists = []
1210        while True:
1211            line_pos = next_pos
1212            line = self._file.readline()
1213            next_pos = self._file.tell()
1214            if line == '\037\014' + os.linesep:
1215                if len(stops) < len(starts):
1216                    stops.append(line_pos - len(os.linesep))
1217                starts.append(next_pos)
1218                labels = [label.strip() for label
1219                                        in self._file.readline()[1:].split(',')
1220                                        if label.strip() != '']
1221                label_lists.append(labels)
1222            elif line == '\037' or line == '\037' + os.linesep:
1223                if len(stops) < len(starts):
1224                    stops.append(line_pos - len(os.linesep))
1225            elif line == '':
1226                stops.append(line_pos - len(os.linesep))
1227                break
1228        self._toc = dict(enumerate(zip(starts, stops)))
1229        self._labels = dict(enumerate(label_lists))
1230        self._next_key = len(self._toc)
1231        self._file.seek(0, 2)
1232        self._file_length = self._file.tell()
1233
1234    def _pre_mailbox_hook(self, f):
1235        """Called before writing the mailbox to file f."""
1236        f.write('BABYL OPTIONS:%sVersion: 5%sLabels:%s%s\037' %
1237                (os.linesep, os.linesep, ','.join(self.get_labels()),
1238                 os.linesep))
1239
1240    def _pre_message_hook(self, f):
1241        """Called before writing each message to file f."""
1242        f.write('\014' + os.linesep)
1243
1244    def _post_message_hook(self, f):
1245        """Called after writing each message to file f."""
1246        f.write(os.linesep + '\037')
1247
1248    def _install_message(self, message):
1249        """Write message contents and return (start, stop)."""
1250        start = self._file.tell()
1251        if isinstance(message, BabylMessage):
1252            special_labels = []
1253            labels = []
1254            for label in message.get_labels():
1255                if label in self._special_labels:
1256                    special_labels.append(label)
1257                else:
1258                    labels.append(label)
1259            self._file.write('1')
1260            for label in special_labels:
1261                self._file.write(', ' + label)
1262            self._file.write(',,')
1263            for label in labels:
1264                self._file.write(' ' + label + ',')
1265            self._file.write(os.linesep)
1266        else:
1267            self._file.write('1,,' + os.linesep)
1268        if isinstance(message, email.message.Message):
1269            orig_buffer = StringIO.StringIO()
1270            orig_generator = email.generator.Generator(orig_buffer, False, 0)
1271            orig_generator.flatten(message)
1272            orig_buffer.seek(0)
1273            while True:
1274                line = orig_buffer.readline()
1275                self._file.write(line.replace('\n', os.linesep))
1276                if line == '\n' or line == '':
1277                    break
1278            self._file.write('*** EOOH ***' + os.linesep)
1279            if isinstance(message, BabylMessage):
1280                vis_buffer = StringIO.StringIO()
1281                vis_generator = email.generator.Generator(vis_buffer, False, 0)
1282                vis_generator.flatten(message.get_visible())
1283                while True:
1284                    line = vis_buffer.readline()
1285                    self._file.write(line.replace('\n', os.linesep))
1286                    if line == '\n' or line == '':
1287                        break
1288            else:
1289                orig_buffer.seek(0)
1290                while True:
1291                    line = orig_buffer.readline()
1292                    self._file.write(line.replace('\n', os.linesep))
1293                    if line == '\n' or line == '':
1294                        break
1295            while True:
1296                buffer = orig_buffer.read(4096) # Buffer size is arbitrary.
1297                if buffer == '':
1298                    break
1299                self._file.write(buffer.replace('\n', os.linesep))
1300        elif isinstance(message, str):
1301            body_start = message.find('\n\n') + 2
1302            if body_start - 2 != -1:
1303                self._file.write(message[:body_start].replace('\n',
1304                                                              os.linesep))
1305                self._file.write('*** EOOH ***' + os.linesep)
1306                self._file.write(message[:body_start].replace('\n',
1307                                                              os.linesep))
1308                self._file.write(message[body_start:].replace('\n',
1309                                                              os.linesep))
1310            else:
1311                self._file.write('*** EOOH ***' + os.linesep + os.linesep)
1312                self._file.write(message.replace('\n', os.linesep))
1313        elif hasattr(message, 'readline'):
1314            original_pos = message.tell()
1315            first_pass = True
1316            while True:
1317                line = message.readline()
1318                self._file.write(line.replace('\n', os.linesep))
1319                if line == '\n' or line == '':
1320                    self._file.write('*** EOOH ***' + os.linesep)
1321                    if first_pass:
1322                        first_pass = False
1323                        message.seek(original_pos)
1324                    else:
1325                        break
1326            while True:
1327                buffer = message.read(4096)     # Buffer size is arbitrary.
1328                if buffer == '':
1329                    break
1330                self._file.write(buffer.replace('\n', os.linesep))
1331        else:
1332            raise TypeError('Invalid message type: %s' % type(message))
1333        stop = self._file.tell()
1334        return (start, stop)
1335
1336
1337class Message(email.message.Message):
1338    """Message with mailbox-format-specific properties."""
1339
1340    def __init__(self, message=None):
1341        """Initialize a Message instance."""
1342        if isinstance(message, email.message.Message):
1343            self._become_message(copy.deepcopy(message))
1344            if isinstance(message, Message):
1345                message._explain_to(self)
1346        elif isinstance(message, str):
1347            self._become_message(email.message_from_string(message))
1348        elif hasattr(message, "read"):
1349            self._become_message(email.message_from_file(message))
1350        elif message is None:
1351            email.message.Message.__init__(self)
1352        else:
1353            raise TypeError('Invalid message type: %s' % type(message))
1354
1355    def _become_message(self, message):
1356        """Assume the non-format-specific state of message."""
1357        for name in ('_headers', '_unixfrom', '_payload', '_charset',
1358                     'preamble', 'epilogue', 'defects', '_default_type'):
1359            self.__dict__[name] = message.__dict__[name]
1360
1361    def _explain_to(self, message):
1362        """Copy format-specific state to message insofar as possible."""
1363        if isinstance(message, Message):
1364            return  # There's nothing format-specific to explain.
1365        else:
1366            raise TypeError('Cannot convert to spe…