scripts/dtrx

#!/usr/bin/env python
#
# dtrx -- Intelligently extract various archive types.
# Copyright (c) 2006, 2007, 2008 Brett Smith <brettcsmith@brettcsmith.org>.
#
# This program is free software; you can redistribute it and/or modify it
# under the terms of the GNU General Public License as published by the
# Free Software Foundation; either version 3 of the License, or (at your
# option) any later version.
#
# This program is distributed in the hope that it will be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General
# Public License for more details.
#
# You should have received a copy of the GNU General Public License along
# with this program; if not, see <http://www.gnu.org/licenses/>.

# Python 2.3 string methods: 'rfind', 'rindex', 'rjust', 'rstrip'

import errno
import logging
import mimetypes
import optparse
import os
import re
import shutil
import signal
import stat
import subprocess
import sys
import tempfile
import textwrap
import traceback

from sets import Set as set

VERSION = "6.2"
VERSION_BANNER = """dtrx version %s
Copyright (c) 2006, 2007, 2008 Brett Smith <brettcsmith@brettcsmith.org>

This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 3 of the License, or (at your
option) any later version.

This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General
Public License for more details.""" % (VERSION,)

MATCHING_DIRECTORY = 1
ONE_ENTRY_KNOWN = 2
BOMB = 3
EMPTY = 4
ONE_ENTRY_FILE = 'file'
ONE_ENTRY_DIRECTORY = 'directory'

ONE_ENTRY_UNKNOWN = [ONE_ENTRY_FILE, ONE_ENTRY_DIRECTORY]

EXTRACT_HERE = 1
EXTRACT_WRAP = 2
EXTRACT_RENAME = 3

RECURSE_ALWAYS = 1
RECURSE_ONCE = 2
RECURSE_NOT_NOW = 3
RECURSE_NEVER = 4
RECURSE_LIST = 5

mimetypes.encodings_map.setdefault('.bz2', 'bzip2')
mimetypes.encodings_map.setdefault('.lzma', 'lzma')
mimetypes.types_map.setdefault('.gem', 'application/x-ruby-gem')

logger = logging.getLogger('dtrx-log')

class FilenameChecker(object):
    free_func = os.open
    free_args = (os.O_CREAT | os.O_EXCL,)
    free_close = os.close

    def __init__(self, original_name):
        self.original_name = original_name

    def is_free(self, filename):
        try:
            result = self.free_func(filename, *self.free_args)
        except OSError, error:
            if error.errno == errno.EEXIST:
                return False
            raise
        if self.free_close:
            self.free_close(result)
        return True

    def create(self):
        fd, filename = tempfile.mkstemp(prefix=self.original_name + '.',
                                        dir='.')
        os.close(fd)
        return filename

    def check(self):
        for suffix in [''] + ['.%s' % (x,) for x in range(1, 10)]:
            filename = '%s%s' % (self.original_name, suffix)
            if self.is_free(filename):
                return filename
        return self.create()


class DirectoryChecker(FilenameChecker):
    free_func = os.mkdir
    free_args = ()
    free_close = None

    def create(self):
        return tempfile.mkdtemp(prefix=self.original_name + '.', dir='.')


class ExtractorError(Exception):
    pass


class ExtractorUnusable(Exception):
    pass


EXTRACTION_ERRORS = (ExtractorError, ExtractorUnusable, OSError, IOError)

class BaseExtractor(object):
    decoders = {'bzip2': 'bzcat', 'gzip': 'zcat', 'compress': 'zcat',
                'lzma': 'lzcat'}
    name_checker = DirectoryChecker

    def __init__(self, filename, encoding):
        if encoding and (not self.decoders.has_key(encoding)):
            raise ValueError("unrecognized encoding %s" % (encoding,))
        self.filename = os.path.realpath(filename)
        self.encoding = encoding
        self.file_count = 0
        self.included_archives = []
        self.target = None
        self.content_type = None
        self.content_name = None
        self.pipes = []
        self.stderr = tempfile.TemporaryFile()
        self.exit_codes = []
        try:
            self.archive = open(filename, 'r')
        except (IOError, OSError), error:
            raise ExtractorError("could not open %s: %s" %
                                 (filename, error.strerror))
        if encoding:
            self.pipe([self.decoders[encoding]], "decoding")
        self.prepare()

    def pipe(self, command, description="extraction"):
        self.pipes.append((command, description))

    def first_bad_exit_code(self):
        for index, code in enumerate(self.exit_codes):
            if code != 0:
                return index
        return None

    def run_pipes(self, final_stdout=None):
        if not self.pipes:
            return
        elif final_stdout is None:
            # FIXME: Buffering this might be dumb.
            final_stdout = tempfile.TemporaryFile()
        num_pipes = len(self.pipes)
        last_pipe = num_pipes - 1
        processes = []
        for index, command in enumerate([pipe[0] for pipe in self.pipes]):
            if index == 0:
                stdin = self.archive
            else:
                stdin = processes[-1].stdout
            if index == last_pipe:
                stdout = final_stdout
            else:
                stdout = subprocess.PIPE
            try:
                processes.append(subprocess.Popen(command, stdin=stdin,
                                                  stdout=stdout,
                                                  stderr=self.stderr))
            except OSError, error:
                if error.errno == errno.ENOENT:
                    raise ExtractorUnusable("could not run %s" % (command[0],))
                raise
        self.exit_codes = [pipe.wait() for pipe in processes]
        self.archive.close()
        for index in range(last_pipe):
            processes[index].stdout.close()
        self.archive = final_stdout

    def prepare(self):
        pass

    def check_included_archives(self):
        if (self.content_name is None) or (not self.content_name.endswith('/')):
            self.included_root = './'
        else:
            self.included_root = self.content_name
        start_index = len(self.included_root)
        for path, dirname, filenames in os.walk(self.included_root):
            self.file_count += len(filenames)
            path = path[start_index:]
            for filename in filenames:
                if (ExtractorBuilder.try_by_mimetype(filename) or
                    ExtractorBuilder.try_by_extension(filename)):
                    self.included_archives.append(os.path.join(path, filename))

    def check_contents(self):
        if not self.contents:
            self.content_type = EMPTY
        elif len(self.contents) == 1:
            if self.basename() == self.contents[0]:
                self.content_type = MATCHING_DIRECTORY
            elif os.path.isdir(self.contents[0]):
                self.content_type = ONE_ENTRY_DIRECTORY
            else:
                self.content_type = ONE_ENTRY_FILE
            self.content_name = self.contents[0]
            if os.path.isdir(self.contents[0]):
                self.content_name += '/'
        else:
            self.content_type = BOMB
        self.check_included_archives()

    def basename(self):
        pieces = os.path.basename(self.filename).split('.')
        extension = '.' + pieces[-1]
        if mimetypes.encodings_map.has_key(extension):
            pieces.pop()
            extension = '.' + pieces[-1]
        if (mimetypes.types_map.has_key(extension) or
            mimetypes.common_types.has_key(extension) or
            mimetypes.suffix_map.has_key(extension)):
            pieces.pop()
        return '.'.join(pieces)

    def get_stderr(self):
        self.stderr.seek(0, 0)
        errors = self.stderr.read(-1)
        self.stderr.close()
        return errors

    def check_success(self, got_output):
        error_index = self.first_bad_exit_code()
        if (not got_output) and (error_index is not None):
            command = ' '.join(self.pipes[error_index][0])
            raise ExtractorError("%s error: '%s' returned status code %s" %
                                 (self.pipes[error_index][1], command,
                                  self.exit_codes[error_index]))
        
    def extract_archive(self):
        self.pipe(self.extract_pipe)
        self.run_pipes()

    def extract(self):
        try:
            self.target = tempfile.mkdtemp(prefix='.dtrx-', dir='.')
        except (OSError, IOError), error:
            raise ExtractorError("cannot extract here: %s" % (error.strerror,))
        old_path = os.path.realpath(os.curdir)
        os.chdir(self.target)
        try:
            self.archive.seek(0, 0)
            self.extract_archive()
            self.contents = os.listdir('.')
            self.check_contents()
            self.check_success(self.content_type != EMPTY)
        except EXTRACTION_ERRORS:
            self.archive.close()
            os.chdir(old_path)
            shutil.rmtree(self.target, ignore_errors=True)
            raise
        self.archive.close()
        os.chdir(old_path)

    def get_filenames(self):
        self.pipe(self.list_pipe, "listing")
        self.run_pipes()
        self.check_success(False)
        self.archive.seek(0, 0)
        while True:
            line = self.archive.readline()
            if not line:
                self.archive.close()
                return
            yield line.rstrip('\n')
    

class CompressionExtractor(BaseExtractor):
    file_type = 'compressed file'
    name_checker = FilenameChecker

    def basename(self):
        pieces = os.path.basename(self.filename).split('.')
        extension = '.' + pieces[-1]
        if mimetypes.encodings_map.has_key(extension):
            pieces.pop()
        return '.'.join(pieces)

    def get_filenames(self):
        # This code used to just immediately yield the basename, under the
        # assumption that that would be the filename.  However, if that
        # happens, dtrx -l will report this as a valid result for files with
        # compression extensions, even if those files shouldn't actually be
        # handled this way.  So, we call out to the file command to do a quick
        # check and make sure this actually looks like a compressed file.
        if 'compress' not in [match[0] for match in
                              ExtractorBuilder.try_by_magic(self.filename)]:
            raise ExtractorError("doesn't look like a compressed file")
        yield self.basename()

    def extract(self):
        self.content_type = ONE_ENTRY_KNOWN
        self.content_name = self.basename()
        self.contents = None
        self.included_root = './'
        try:
            output_fd, self.target = tempfile.mkstemp(prefix='.dtrx-', dir='.')
        except (OSError, IOError), error:
            raise ExtractorError("cannot extract here: %s" % (error.strerror,))
        self.run_pipes(output_fd)
        os.close(output_fd)
        try:
            self.check_success(os.stat(self.target)[stat.ST_SIZE] > 0)
        except EXTRACTION_ERRORS:
            os.unlink(self.target)
            raise
            
class TarExtractor(BaseExtractor):
    file_type = 'tar file'
    extract_pipe = ['tar', '-x']
    list_pipe = ['tar', '-t']
        
        
class CpioExtractor(BaseExtractor):
    file_type = 'cpio file'
    extract_pipe = ['cpio', '-i', '--make-directories', '--quiet',
                   '--no-absolute-filenames']
    list_pipe = ['cpio', '-t', '--quiet']


class RPMExtractor(CpioExtractor):
    file_type = 'RPM'

    def prepare(self):
        self.pipe(['rpm2cpio', '-'], "rpm2cpio")

    def basename(self):
        pieces = os.path.basename(self.filename).split('.')
        if len(pieces) == 1:
            return pieces[0]
        elif pieces[-1] != 'rpm':
            return BaseExtractor.basename(self)
        pieces.pop()
        if len(pieces) == 1:
            return pieces[0]
        elif len(pieces[-1]) < 8:
            pieces.pop()
        return '.'.join(pieces)

    def check_contents(self):
        self.check_included_archives()
        self.content_type = BOMB


class DebExtractor(TarExtractor):
    file_type = 'Debian package'

    def prepare(self):
        self.pipe(['ar', 'p', self.filename, 'data.tar.gz'],
                  "data.tar.gz extraction")
        self.pipe(['zcat'], "data.tar.gz decompression")

    def basename(self):
        pieces = os.path.basename(self.filename).split('_')
        if len(pieces) == 1:
            return pieces[0]
        last_piece = pieces.pop()
        if (len(last_piece) > 10) or (not last_piece.endswith('.deb')):
            return BaseExtractor.basename(self)
        return '_'.join(pieces)

    def check_contents(self):
        self.check_included_archives()
        self.content_type = BOMB


class DebMetadataExtractor(DebExtractor):
    def prepare(self):
        self.pipe(['ar', 'p', self.filename, 'control.tar.gz'],
                  "control.tar.gz extraction")
        self.pipe(['zcat'], "control.tar.gz decompression")


class GemExtractor(TarExtractor):
    file_type = 'Ruby gem'

    def prepare(self):
        self.pipe(['tar', '-xO', 'data.tar.gz'], "data.tar.gz extraction")
        self.pipe(['zcat'], "data.tar.gz decompression")

    def check_contents(self):
        self.check_included_archives()
        self.content_type = BOMB


class GemMetadataExtractor(CompressionExtractor):
    file_type = 'Ruby gem'

    def prepare(self):
        self.pipe(['tar', '-xO', 'metadata.gz'], "metadata.gz extraction")
        self.pipe(['zcat'], "metadata.gz decompression")

    def basename(self):
        return os.path.basename(self.filename) + '-metadata.txt'


class NoPipeExtractor(BaseExtractor):
    # Some extraction tools won't accept the archive from stdin.  With
    # these, the piping infrastructure we normally set up generally doesn't
    # work, at least at first.  We can still use most of it; we just don't
    # want to seed self.archive with the archive file, since that sucks up
    # memory.  So instead we seed it with /dev/null, and specify the
    # filename on the command line as necessary.  We also open the actual
    # file with os.open, to make sure we can actually do it (permissions
    # are good, etc.).  This class doesn't do anything by itself; it's just
    # meant to be a base class for extractors that rely on these dumb
    # tools.
    def __init__(self, filename, encoding):
        os.close(os.open(filename, os.O_RDONLY))
        BaseExtractor.__init__(self, '/dev/null', None)
        self.filename = os.path.realpath(filename)

    def extract_archive(self):
        self.extract_pipe = self.extract_command + [self.filename]
        BaseExtractor.extract_archive(self)

    def get_filenames(self):
        self.list_pipe = self.list_command + [self.filename]
        return BaseExtractor.get_filenames(self)


class ZipExtractor(NoPipeExtractor):
    file_type = 'Zip file'
    extract_command = ['unzip', '-q']
    list_command = ['zipinfo', '-1']


class SevenExtractor(NoPipeExtractor):
    file_type = '7z file'
    extract_command = ['7z', 'x']
    list_command = ['7z', 'l']
    border_re = re.compile('^[- ]+$')

    def get_filenames(self):
        fn_index = None
        for line in NoPipeExtractor.get_filenames(self):
            if self.border_re.match(line):
                if fn_index is not None:
                    break
                else:
                    fn_index = line.rindex(' ') + 1
            elif fn_index is not None:
                yield line[fn_index:]
        self.archive.close()
        

class CABExtractor(NoPipeExtractor):
    file_type = 'CAB archive'
    extract_command = ['cabextract', '-q']
    list_command = ['cabextract', '-l']
    border_re = re.compile(r'^[-\+]+$')

    def get_filenames(self):
        fn_index = None
        filenames = NoPipeExtractor.get_filenames(self)
        for line in filenames:
            if self.border_re.match(line):
                break
        for line in filenames:
            try:
                yield line.split(' | ', 2)[2]
            except IndexError:
                break
        self.archive.close()


class ShieldExtractor(NoPipeExtractor):
    file_type = 'InstallShield archive'
    extract_command = ['unshield', 'x']
    list_command = ['unshield', 'l']
    prefix_re = re.compile(r'^\s+\d+\s+')
    end_re = re.compile(r'^\s+-+\s+-+\s*$')

    def get_filenames(self):
        for line in NoPipeExtractor.get_filenames(self):
            if self.end_re.match(line):
                break
            else:
                match = self.prefix_re.match(line)
                if match:
                    yield line[match.end():]
        self.archive.close()

    def basename(self):
        result = NoPipeExtractor.basename(self)
        if result.endswith('.hdr'):
            result = result[:-4]
        return result


class BaseHandler(object):
    def __init__(self, extractor, options):
        self.extractor = extractor
        self.options = options
        self.target = None

    def handle(self):
        command = 'find'
        status = subprocess.call(['find', self.extractor.target, '-type', 'd',
                                  '-exec', 'chmod', 'u+rwx', '{}', ';'])
        if status == 0:
            command = 'chmod'
            status = subprocess.call(['chmod', '-R', 'u+rwX',
                                      self.extractor.target])
        if status != 0:
            return "%s returned with exit status %s" % (command, status)
        return self.organize()

    def set_target(self, target, checker):
        self.target = checker(target).check()
        if self.target != target:
            logger.warning("extracting %s to %s" %
                           (self.extractor.filename, self.target))


# The "where to extract" table, with options and archive types.
# This dictates the contents of each can_handle method.
#
#         Flat           Overwrite            None
# File    basename       basename             FilenameChecked
# Match   .              .                    tempdir + checked
# Bomb    .              basename             DirectoryChecked

class FlatHandler(BaseHandler):
    def can_handle(contents, options):
        return ((options.flat and (contents != ONE_ENTRY_KNOWN)) or
                (options.overwrite and (contents == MATCHING_DIRECTORY)))
    can_handle = staticmethod(can_handle)

    def organize(self):
        self.target = '.'
        for curdir, dirs, filenames in os.walk(self.extractor.target,
                                               topdown=False):
            path_parts = curdir.split(os.sep)
            if path_parts[0] == '.':
                del path_parts[1]
            else:
                del path_parts[0]
            newdir = os.path.join(*path_parts)
            if not os.path.isdir(newdir):
                os.makedirs(newdir)
            for filename in filenames:
                os.rename(os.path.join(curdir, filename),
                          os.path.join(newdir, filename))
            os.rmdir(curdir)


class OverwriteHandler(BaseHandler):
    def can_handle(contents, options):
        return ((options.flat and (contents == ONE_ENTRY_KNOWN)) or
                (options.overwrite and (contents != MATCHING_DIRECTORY)))
    can_handle = staticmethod(can_handle)

    def organize(self):
        self.target = self.extractor.basename()
        if os.path.isdir(self.target):
            shutil.rmtree(self.target)
        os.rename(self.extractor.target, self.target)
        

class MatchHandler(BaseHandler):
    def can_handle(contents, options):
        return ((contents == MATCHING_DIRECTORY) or
                ((contents in ONE_ENTRY_UNKNOWN) and
                 options.one_entry_policy.ok_for_match()))
    can_handle = staticmethod(can_handle)

    def organize(self):
        source = os.path.join(self.extractor.target,
                              os.listdir(self.extractor.target)[0])
        if os.path.isdir(source):
            checker = DirectoryChecker
        else:
            checker = FilenameChecker
        if self.options.one_entry_policy == EXTRACT_HERE:
            destination = self.extractor.content_name.rstrip('/')
        else:
            destination = self.extractor.basename()
        self.set_target(destination, checker)
        if os.path.isdir(self.extractor.target):
            os.rename(source, self.target)
            os.rmdir(self.extractor.target)
        else:
            os.rename(self.extractor.target, self.target)
        self.extractor.included_root = './'


class EmptyHandler(object):
    def can_handle(contents, options):
        return contents == EMPTY
    can_handle = staticmethod(can_handle)

    def __init__(self, extractor, options): pass
    def handle(self): pass


class BombHandler(BaseHandler):
    def can_handle(contents, options):
        return True
    can_handle = staticmethod(can_handle)

    def organize(self):
        basename = self.extractor.basename()
        self.set_target(basename, self.extractor.name_checker)
        os.rename(self.extractor.target, self.target)

        
class BasePolicy(object):
    def __init__(self, options):
        self.current_policy = None
        if options.batch:
            self.permanent_policy = self.answers['']
        else:
            self.permanent_policy = None

    def ask_question(self, question):
        question = question + self.choices
        while True:
            print "\n".join(question)
            try:
                answer = raw_input(self.prompt)
            except EOFError:
                return self.answers['']
            try:
                return self.answers[answer.lower()]
            except KeyError:
                print

    def __cmp__(self, other):
        return cmp(self.current_policy, other)
    

class OneEntryPolicy(BasePolicy):
    answers = {'h': EXTRACT_HERE, 'i': EXTRACT_WRAP, 'r': EXTRACT_RENAME,
               '': EXTRACT_WRAP}
    choices = ["You can:",
               " * extract it Inside another directory",
               " * extract it and Rename the directory",
               " * extract it Here"]
    prompt = "What do you want to do?  (I/r/h) "

    def __init__(self, options):
        BasePolicy.__init__(self, options)
        if options.flat:
            default = 'h'
        elif options.one_entry_default is not None:
            default = options.one_entry_default.lower()
        else:
            return
        if 'here'.startswith(default):
            self.permanent_policy = EXTRACT_HERE
        elif 'rename'.startswith(default):
            self.permanent_policy = EXTRACT_RENAME
        elif 'inside'.startswith(default):
            self.permanent_policy = EXTRACT_WRAP
        elif default is not None:
            raise ValueError("bad value %s for default policy" % (default,))

    def prep(self, archive_filename, extractor):
        question = ["%s contains one %s, but it has a weird name." %
                    (archive_filename, extractor.content_type)]
        question.append(" Expected: " + extractor.basename())
        question.append("   Actual: " + extractor.content_name)
        self.current_policy = (self.permanent_policy or
                               self.ask_question(question))

    def ok_for_match(self):
        return self.current_policy in (EXTRACT_RENAME, EXTRACT_HERE)


class RecursionPolicy(BasePolicy):
    answers = {'o': RECURSE_ONCE, 'a': RECURSE_ALWAYS, 'n': RECURSE_NOT_NOW,
               'v': RECURSE_NEVER, 'l': RECURSE_LIST, '': RECURSE_NOT_NOW}
    choices = ["You can:",
               " * Always extract included archives",
               " * extract included archives this Once",
               " * choose Not to extract included archives",
               " * neVer extract included archives",
               " * List included archives"]
    prompt = "What do you want to do?  (a/o/N/v/l) "

    def __init__(self, options):
        BasePolicy.__init__(self, options)
        if options.show_list:
            self.permanent_policy = RECURSE_NEVER
        elif options.recursive:
            self.permanent_policy = RECURSE_ALWAYS

    def prep(self, current_filename, target, extractor):
        archive_count = len(extractor.included_archives)
        if (self.permanent_policy is not None) or (archive_count == 0):
            self.current_policy = self.permanent_policy or RECURSE_NOT_NOW
            return
        question = (("%s contains %s other archive file(s), " +
                     "out of %s file(s) total.") %
                    (current_filename, archive_count, extractor.file_count))
        question = textwrap.wrap(question)
        if target == '.':
            target = ''
        included_root = extractor.included_root
        if included_root == './':
            included_root = ''
        while True:
            self.current_policy = self.ask_question(question)
            if self.current_policy != RECURSE_LIST:
                break
            print ("\n%s\n" %
                   '\n'.join([os.path.join(target, included_root, filename)
                              for filename in extractor.included_archives]))
        if self.current_policy in (RECURSE_ALWAYS, RECURSE_NEVER):
            self.permanent_policy = self.current_policy

    def ok_to_recurse(self):
        return self.current_policy in (RECURSE_ALWAYS, RECURSE_ONCE)
            

class ExtractorBuilder(object):
    extractor_map = {'tar': {'extractor': TarExtractor,
                             'mimetypes': ('x-tar',),
                             'extensions': ('tar',),
                             'magic': ('POSIX tar archive',)},
                     'zip': {'extractor': ZipExtractor,
                             'mimetypes': ('zip',),
                             'extensions': ('zip',),
                             'magic': ('(Zip|ZIP self-extracting) archive',)},
                     'rpm': {'extractor': RPMExtractor,
                             'mimetypes': ('x-redhat-package-manager', 'x-rpm'),
                             'extensions': ('rpm',),
                             'magic': ('RPM',)},
                     'deb': {'extractor': DebExtractor,
                             'metadata': DebMetadataExtractor,
                             'mimetypes': ('x-debian-package',),
                             'extensions': ('deb',),
                             'magic': ('Debian binary package',)},
                     'cpio': {'extractor': CpioExtractor,
                              'mimetypes': ('x-cpio',),
                              'extensions': ('cpio',),
                              'magic': ('cpio archive',)},
                     'gem': {'extractor': GemExtractor,
                             'metadata': GemMetadataExtractor,
                             'mimetypes': ('x-ruby-gem',),
                             'extensions': ('gem',)},
                     '7z': {'extractor': SevenExtractor,
                             'mimetypes': ('x-7z-compressed',),
                             'extensions': ('7z',),
                             'magic': ('7-zip archive',)},
                     'cab': {'extractor': CABExtractor,
                             'mimetypes': ('x-cab',),
                             'extensions': ('cab',),
                             'magic': ('Microsoft Cabinet Archive',)},
                     'shield': {'extractor': ShieldExtractor,
                                'mimetypes': ('x-cab',),
                                'extensions': ('cab', 'hdr'),
                                'magic': ('InstallShield CAB',)},
                     'compress': {'extractor': CompressionExtractor}
                     }

    mimetype_map = {}
    magic_mime_map = {}
    extension_map = {}
    for ext_name, ext_info in extractor_map.items():
        for mimetype in ext_info.get('mimetypes', ()):
            if '/' not in mimetype:
                mimetype = 'application/' + mimetype
            mimetype_map[mimetype] = ext_name
        for magic_re in ext_info.get('magic', ()):
            magic_mime_map[re.compile(magic_re)] = ext_name
        for extension in ext_info.get('extensions', ()):
            extension_map.setdefault(extension, []).append((ext_name, None))

    for mapping in (('tar', 'bzip2', 'tar.bz2'),
                    ('tar', 'gzip', 'tar.gz', 'tgz'),
                    ('compress', 'gzip', 'Z', 'gz'),
                    ('compress', 'bzip2', 'bz2'),
                    ('compress', 'lzma', 'lzma')):
        for extension in mapping[2:]:
            extension_map.setdefault(extension, []).append(mapping[:2])

    magic_encoding_map = {}
    for mapping in (('bzip2', 'bzip2 compressed'),
                    ('gzip', 'gzip compressed')):
        for pattern in mapping[1:]:
            magic_encoding_map[re.compile(pattern)] = mapping[0]

    def __init__(self, filename, options):
        self.filename = filename
        self.options = options

    def build_extractor(self, archive_type, encoding):
        extractors = self.extractor_map[archive_type]
        if self.options.metadata and extractors.has_key('metadata'):
            extractor = extractors['metadata']
        else:
            extractor = extractors['extractor']
        return extractor(self.filename, encoding)

    def get_extractor(self):
        tried_types = set()
        # As smart as it is, the magic test can't go first, because at least
        # on my system it just recognizes gem files as tar files.  I guess
        # it's possible for the opposite problem to occur -- where the mimetype
        # or extension suggests something less than ideal -- but it seems less
        # likely so I'm sticking with this.
        for func_name in ('mimetype', 'extension', 'magic'):
            logger.debug("getting extractors by %s" % (func_name,))
            extractor_types = \
                            getattr(self, 'try_by_' + func_name)(self.filename)
            logger.debug("done getting extractors")
            for ext_args in extractor_types:
                if ext_args in tried_types:
                    continue
                tried_types.add(ext_args)
                logger.debug("trying %s extractor from %s" %
                             (ext_args, func_name))
                yield self.build_extractor(*ext_args)

    def try_by_mimetype(cls, filename):
        mimetype, encoding = mimetypes.guess_type(filename)
        try:
            return [(cls.mimetype_map[mimetype], encoding)]
        except KeyError:
            if encoding:
                return [('compress', encoding)]
        return []
    try_by_mimetype = classmethod(try_by_mimetype)

    def magic_map_matches(cls, output, magic_map):
        return [result for regexp, result in magic_map.items()
                if regexp.search(output)]
    magic_map_matches = classmethod(magic_map_matches)
        
    def try_by_magic(cls, filename):
        process = subprocess.Popen(['file', '-z', filename],
                                   stdout=subprocess.PIPE)
        status = process.wait()
        if status != 0:
            return []
        output = process.stdout.readline()
        process.stdout.close()
        if output.startswith('%s: ' % filename):
            output = output[len(filename) + 2:]
        mimes = cls.magic_map_matches(output, cls.magic_mime_map)
        encodings = cls.magic_map_matches(output, cls.magic_encoding_map)
        if mimes and not encodings:
            encodings = [None]
        elif encodings and not mimes:
            mimes = ['compress']
        return [(m, e) for m in mimes for e in encodings]
    try_by_magic = classmethod(try_by_magic)

    def try_by_extension(cls, filename):
        parts = filename.split('.')[-2:]
        results = []
        while parts:
            results.extend(cls.extension_map.get('.'.join(parts), []))
            del parts[0]
        return results
    try_by_extension = classmethod(try_by_extension)


class BaseAction(object):
    def __init__(self, options, filenames):
        self.options = options
        self.filenames = filenames
        self.target = None
        
    def report(self, function, *args):
        try:
            error = function(*args)
        except EXTRACTION_ERRORS, exception:
            error = str(exception)
            logger.debug(''.join(traceback.format_exception(*sys.exc_info())))
        return error


class ExtractionAction(BaseAction):
    handlers = [FlatHandler, OverwriteHandler, MatchHandler, EmptyHandler,
                BombHandler]

    def __init__(self, options, filenames):
        BaseAction.__init__(self, options, filenames)
        self.did_print = False

    def get_handler(self, extractor):
        if extractor.content_type in ONE_ENTRY_UNKNOWN:
            self.options.one_entry_policy.prep(self.current_filename,
                                               extractor)
        for handler in self.handlers:
            if handler.can_handle(extractor.content_type, self.options):
                logger.debug("using %s handler" % (handler.__name__,))
                self.current_handler = handler(extractor, self.options)
                break

    def show_extraction(self, extractor):
        if self.options.log_level > logging.INFO:
            return
        elif self.did_print:
            print
        else:
            self.did_print = True
        print "%s:" % (self.current_filename,)
        if extractor.contents is None:
            print self.current_handler.target
            return
        def reverser(x, y):
            return cmp(y, x)
        if self.current_handler.target == '.':
            filenames = extractor.contents
            filenames.sort(reverser)
        else:
            filenames = [self.current_handler.target]
        pathjoin = os.path.join
        isdir = os.path.isdir
        while filenames:
            filename = filenames.pop()
            if isdir(filename):
                print "%s/" % (filename,)
                new_filenames = os.listdir(filename)
                new_filenames.sort(reverser)
                filenames.extend([pathjoin(filename, new_filename)
                                  for new_filename in new_filenames])
            else:
                print filename

    def run(self, filename, extractor):
        self.current_filename = filename
        error = (self.report(extractor.extract) or
                 self.report(self.get_handler, extractor) or
                 self.report(self.current_handler.handle) or
                 self.report(self.show_extraction, extractor))
        if not error:
            self.target = self.current_handler.target
        return error


class ListAction(BaseAction):
    def __init__(self, options, filenames):
        BaseAction.__init__(self, options, filenames)
        self.count = 0

    def get_list(self, extractor):
        # Note: The reason I'm getting all the filenames up front is
        # because if we run into trouble partway through the archive, we'll
        # try another extractor.  So before we display anything we have to
        # be sure this one is successful.  We maybe don't have to be quite
        # this conservative but this is the easy way out for now.
        self.filelist = list(extractor.get_filenames())

    def show_list(self, filename):
        self.count += 1
        if len(self.filenames) != 1:
            if self.count > 1:
                print
            print "%s:" % (filename,)
        print '\n'.join(self.filelist)

    def run(self, filename, extractor):
        return (self.report(self.get_list, extractor) or
                self.report(self.show_list, filename))


class ExtractorApplication(object):
    def __init__(self, arguments):
        for signal_num in (signal.SIGINT, signal.SIGTERM):
            signal.signal(signal_num, self.abort)
        signal.signal(signal.SIGPIPE, signal.SIG_DFL)
        self.parse_options(arguments)
        self.setup_logger()
        self.successes = []
        self.failures = []

    def clean_destination(self, dest_name):
        try:
            os.unlink(dest_name)
        except OSError, error:
            if error.errno == errno.EISDIR:
                shutil.rmtree(dest_name, ignore_errors=True)

    def abort(self, signal_num, frame):
        signal.signal(signal_num, signal.SIG_IGN)
        print
        logger.debug("traceback:\n" +
                     ''.join(traceback.format_stack(frame)).rstrip())
        logger.debug("got signal %s" % (signal_num,))
        try:
            basename = self.current_extractor.target
        except AttributeError:
            basename = None
        if basename is not None:
            logger.debug("cleaning up %s" % (basename,))
            clean_targets = set([os.path.realpath('.')])
            if hasattr(self, 'current_directory'):
                clean_targets.add(os.path.realpath(self.current_directory))
            for directory in clean_targets:
                self.clean_destination(os.path.join(directory, basename))
        sys.exit(1)

    def parse_options(self, arguments):
        parser = optparse.OptionParser(
            usage="%prog [options] archive [archive2 ...]",
            description="Intelligent archive extractor",
            version=VERSION_BANNER
            )
        parser.add_option('-l', '-t', '--list', '--table', dest='show_list',
                          action='store_true', default=False,
                          help="list contents of archives on standard output")
        parser.add_option('-m', '--metadata', dest='metadata',
                          action='store_true', default=False,
                          help="extract metadata from a .deb/.gem")
        parser.add_option('-r', '--recursive', dest='recursive',
                          action='store_true', default=False,
                          help="extract archives contained in the ones listed")
        parser.add_option('--one', '--one-entry', dest='one_entry_default',
                          default=None,
                          help=("specify extraction policy for one-entry " +
                                "archives: inside/rename/here"))
        parser.add_option('-n', '--noninteractive', dest='batch',
                          action='store_true', default=False,
                          help="don't ask how to handle special cases")
        parser.add_option('-o', '--overwrite', dest='overwrite',
                          action='store_true', default=False,
                          help="overwrite any existing target output")
        parser.add_option('-f', '--flat', '--no-directory', dest='flat',
                          action='store_true', default=False,
                          help="extract everything to the current directory")
        parser.add_option('-v', '--verbose', dest='verbose',
                          action='count', default=0,
                          help="be verbose/print debugging information")
        parser.add_option('-q', '--quiet', dest='quiet',
                          action='count', default=3,
                          help="suppress warning/error messages")
        self.options, filenames = parser.parse_args(arguments)
        if not filenames:
            parser.error("you did not list any archives")
        # This makes WARNING is the default.
        self.options.log_level = (10 * (self.options.quiet -
                                        self.options.verbose))
        try:
            self.options.one_entry_policy = OneEntryPolicy(self.options)
        except ValueError:
            parser.error("invalid value for --one-entry option")
        self.options.recursion_policy = RecursionPolicy(self.options)
        self.archives = {os.path.realpath(os.curdir): filenames}

    def setup_logger(self):
        logging.getLogger().setLevel(self.options.log_level)
        handler = logging.StreamHandler()
        handler.setLevel(self.options.log_level)
        formatter = logging.Formatter("dtrx: %(levelname)s: %(message)s")
        handler.setFormatter(formatter)
        logger.addHandler(handler)
        logger.debug("logger is set up")

    def recurse(self, filename, extractor, action):
        self.options.recursion_policy.prep(filename, action.target, extractor)
        if self.options.recursion_policy.ok_to_recurse():
            for filename in extractor.included_archives:
                logger.debug("recursing with %s archive" %
                             (extractor.content_type,))
                tail_path, basename = os.path.split(filename)
                path_args = [self.current_directory, extractor.included_root,
                             tail_path]
                logger.debug("included root: %s" % (extractor.included_root,))
                logger.debug("tail path: %s" % (tail_path,))
                if os.path.isdir(action.target):
                    logger.debug("action target: %s" % (action.target,))
                    path_args.insert(1, action.target)
                directory = os.path.join(*path_args)
                self.archives.setdefault(directory, []).append(basename)

    def check_file(self, filename):
        try:
            result = os.stat(filename)
        except OSError, error:
            return error.strerror
        if stat.S_ISDIR(result.st_mode):
            return "cannot work with a directory"

    def show_stderr(self, logger_func, stderr):
        if stderr:
            logger_func("Error output from this process:\n" +
                        stderr.rstrip('\n'))

    def try_extractors(self, filename, builder):
        errors = []
        for extractor in builder:
            self.current_extractor = extractor  # For the abort() method.
            error = self.action.run(filename, extractor)
            if error:
                errors.append((extractor.file_type, extractor.encoding, error,
                               extractor.get_stderr()))
                if extractor.target is not None:
                    self.clean_destination(extractor.target)
            else:
                self.show_stderr(logger.warn, extractor.get_stderr())
                self.recurse(filename, extractor, self.action)
                return
        logger.error("could not handle %s" % (filename,))
        if not errors:
            logger.error("not a known archive type")
            return True
        for file_type, encoding, error, stderr in errors:
            message = ["treating as", file_type, "failed:", error]
            if encoding:
                message.insert(1, "%s-encoded" % (encoding,))
            logger.error(' '.join(message))
            self.show_stderr(logger.error, stderr)
        return True
        
    def run(self):
        if self.options.show_list:
            action = ListAction
        else:
            action = ExtractionAction
        self.action = action(self.options, self.archives.values()[0])
        while self.archives:
            self.current_directory, self.filenames = self.archives.popitem()
            os.chdir(self.current_directory)
            for filename in self.filenames:
                builder = ExtractorBuilder(filename, self.options)
                error = (self.check_file(filename) or
                         self.try_extractors(filename, builder.get_extractor()))
                if error:
                    if error != True:
                        logger.error("%s: %s" % (filename, error))
                    self.failures.append(filename)
                else:
                    self.successes.append(filename)
            self.options.one_entry_policy.permanent_policy = EXTRACT_WRAP
        if self.failures:
            return 1
        return 0


if __name__ == '__main__':
    app = ExtractorApplication(sys.argv[1:])
    sys.exit(app.run())