comparison: scripts/dtrx
scripts/dtrx
- branch
- trunk
- changeset 30
- 1015bbd6dc5e
- parent 29
- 5fad99c17221
- child 31
- c3a2760d1c3a
equal
deleted
inserted
replaced
| 29:5fad99c17221 | 30:1015bbd6dc5e |
|---|---|
| 20 import errno | 20 import errno |
| 21 import logging | 21 import logging |
| 22 import mimetypes | 22 import mimetypes |
| 23 import optparse | 23 import optparse |
| 24 import os | 24 import os |
| 25 import re | |
| 25 import stat | 26 import stat |
| 26 import subprocess | 27 import subprocess |
| 27 import sys | 28 import sys |
| 28 import tempfile | 29 import tempfile |
| 29 import textwrap | 30 import textwrap |
| 546 | 547 |
| 547 handlers = [FlatHandler, OverwriteHandler, MatchHandler, EmptyHandler, | 548 handlers = [FlatHandler, OverwriteHandler, MatchHandler, EmptyHandler, |
| 548 BombHandler] | 549 BombHandler] |
| 549 | 550 |
| 550 class ExtractorBuilder(object): | 551 class ExtractorBuilder(object): |
| 551 extractor_map = {} | 552 extractor_map = {'tar': (TarExtractor, None), |
| 552 for args in ((TarExtractor, None, 'x-tar'), | 553 'zip': (ZipExtractor, None), |
| 553 (ZipExtractor, None, 'zip', 'x-msdos-program'), | 554 'deb': (DebExtractor, DebMetadataExtractor), |
| 554 (DebExtractor, DebMetadataExtractor, 'x-debian-package'), | 555 'rpm': (RPMExtractor, None), |
| 555 (RPMExtractor, None, 'x-redhat-package-manager', 'x-rpm'), | 556 'cpio': (CpioExtractor, None), |
| 556 (CpioExtractor, None, 'x-cpio'), | 557 'gem': (GemExtractor, GemMetadataExtractor), |
| 557 (GemExtractor, GemMetadataExtractor, 'x-ruby-gem')): | 558 'compress': (CompressionExtractor, None)} |
| 558 for entry in args[2:]: | 559 |
| 559 if '/' not in entry: | 560 mimetype_map = {} |
| 560 entry = 'application/' + entry | 561 for mapping in (('tar', 'x-tar'), |
| 561 extractor_map[entry] = args[0:2] | 562 ('zip', 'x-msdos-program', 'zip'), |
| 563 ('deb', 'x-debian-package'), | |
| 564 ('rpm', 'x-redhat-package-manager', 'x-rpm'), | |
| 565 ('cpio', 'x-cpio'), | |
| 566 ('gem', 'x-ruby-gem')): | |
| 567 for mimetype in mapping[1:]: | |
| 568 if '/' not in mimetype: | |
| 569 mimetype = 'application/' + mimetype | |
| 570 mimetype_map[mimetype] = mapping[0] | |
| 571 | |
| 572 magic_mime_map = {} | |
| 573 for mapping in (('deb', 'Debian binary package'), | |
| 574 ('cpio', 'cpio archive'), | |
| 575 ('tar', 'POSIX tar archive'), | |
| 576 ('zip', 'Zip archive'), | |
| 577 ('rpm', 'RPM')): | |
| 578 for pattern in mapping[1:]: | |
| 579 magic_mime_map[re.compile(pattern)] = mapping[0] | |
| 580 | |
| 581 magic_encoding_map = {} | |
| 582 for mapping in (('bzip2', 'bzip2 compressed'), | |
| 583 ('gzip', 'gzip compressed')): | |
| 584 for pattern in mapping[1:]: | |
| 585 magic_encoding_map[re.compile(pattern)] = mapping[0] | |
| 562 | 586 |
| 563 def __init__(self, filename, options): | 587 def __init__(self, filename, options): |
| 564 self.filename = filename | 588 self.filename = filename |
| 565 self.options = options | 589 self.options = options |
| 566 self.mimetype, self.encoding = mimetypes.guess_type(self.filename) | 590 |
| 591 def build_extractor(self, archive_type, encoding): | |
| 592 extractors = self.extractor_map[archive_type] | |
| 593 if self.options.metadata and (extractors[1] is not None): | |
| 594 extractor = extractors[1] | |
| 595 else: | |
| 596 extractor = extractors[0] | |
| 597 return extractor(self.filename, encoding) | |
| 567 | 598 |
| 568 def get_extractor(self): | 599 def get_extractor(self): |
| 569 extractor = self.find_extractor() | 600 for func_name in ('mimetype', 'magic'): |
| 570 if extractor is None: | 601 archive_type, encoding = getattr(self, 'try_by_' + func_name)() |
| 571 raise ExtractorError("not a known archive type") | 602 if archive_type is not None: |
| 572 return extractor(self.filename, self.encoding) | 603 yield self.build_extractor(archive_type, encoding) |
| 573 | 604 |
| 574 def find_extractor(self): | 605 def try_by_mimetype(self): |
| 575 extractor = None | 606 mimetype, encoding = mimetypes.guess_type(self.filename) |
| 576 try: | 607 try: |
| 577 extractors = self.extractor_map[self.mimetype] | 608 return self.mimetype_map[mimetype], encoding |
| 578 if self.options.metadata and (extractors[1] is not None): | |
| 579 extractor = extractors[1] | |
| 580 else: | |
| 581 extractor = extractors[0] | |
| 582 except KeyError: | 609 except KeyError: |
| 583 if self.encoding: | 610 if encoding: |
| 584 extractor = CompressionExtractor | 611 return 'compress', encoding |
| 585 return extractor | 612 return None, None |
| 613 | |
| 614 def try_by_magic(self): | |
| 615 process = subprocess.Popen(['file', '-z', self.filename], | |
| 616 stdout=subprocess.PIPE) | |
| 617 status = process.wait() | |
| 618 if status != 0: | |
| 619 return None, None | |
| 620 output = process.stdout.readline() | |
| 621 process.stdout.close() | |
| 622 if output.startswith('%s: ' % self.filename): | |
| 623 output = output[len(self.filename) + 2:] | |
| 624 results = [None, None] | |
| 625 for index, mapping in enumerate((self.magic_mime_map, | |
| 626 self.magic_encoding_map)): | |
| 627 for regexp, result in mapping.items(): | |
| 628 if regexp.search(output): | |
| 629 results[index] = result | |
| 630 break | |
| 631 return results | |
| 586 | 632 |
| 587 | 633 |
| 588 class ExtractorApplication(object): | 634 class ExtractorApplication(object): |
| 589 def __init__(self, arguments): | 635 def __init__(self, arguments): |
| 590 self.parse_options(arguments) | 636 self.parse_options(arguments) |
| 636 handler.setLevel(10 * (self.options.quiet - self.options.verbose)) | 682 handler.setLevel(10 * (self.options.quiet - self.options.verbose)) |
| 637 formatter = logging.Formatter("dtrx: %(levelname)s: %(message)s") | 683 formatter = logging.Formatter("dtrx: %(levelname)s: %(message)s") |
| 638 handler.setFormatter(formatter) | 684 handler.setFormatter(formatter) |
| 639 self.logger.addHandler(handler) | 685 self.logger.addHandler(handler) |
| 640 | 686 |
| 641 def get_extractor(self): | |
| 642 builder = ExtractorBuilder(self.current_filename, self.options) | |
| 643 self.current_extractor = builder.get_extractor() | |
| 644 | |
| 645 def get_handler(self): | 687 def get_handler(self): |
| 646 for var_name in ('type', 'name'): | 688 for var_name in ('type', 'name'): |
| 647 exec('content_%s = self.current_extractor.content_%s' % | 689 exec('content_%s = self.current_extractor.content_%s' % |
| 648 (var_name, var_name)) | 690 (var_name, var_name)) |
| 649 if content_type == ONE_ENTRY: | 691 if content_type == ONE_ENTRY: |
| 681 self.successes.append(self.current_filename) | 723 self.successes.append(self.current_filename) |
| 682 else: | 724 else: |
| 683 self.failures.append(self.current_filename) | 725 self.failures.append(self.current_filename) |
| 684 | 726 |
| 685 def extract(self): | 727 def extract(self): |
| 686 while self.archives: | 728 success = (self.report(self.current_extractor.extract) and |
| 687 self.current_directory, filenames = self.archives.popitem() | 729 self.report(self.get_handler) and |
| 688 for filename in filenames: | 730 self.report(self.current_handler.handle)) |
| 689 os.chdir(self.current_directory) | 731 if success: |
| 690 self.current_filename = filename | 732 self.recurse() |
| 691 success = (self.report(self.get_extractor) and | 733 return success |
| 692 self.report(self.current_extractor.extract) and | |
| 693 self.report(self.get_handler) and | |
| 694 self.report(self.current_handler.handle)) | |
| 695 if success: | |
| 696 self.recurse() | |
| 697 self.record_status(success) | |
| 698 self.options.one_entry_policy.permanent_policy = EXTRACT_WRAP | |
| 699 | 734 |
| 700 def show_contents(self): | 735 def show_contents(self): |
| 701 for filename in self.current_extractor.get_filenames(): | 736 for filename in self.current_extractor.get_filenames(): |
| 702 print filename | 737 print filename |
| 703 | 738 |
| 704 def show_list(self): | 739 def make_list(self): |
| 705 filenames = self.archives.values()[0] | 740 if len(self.archives.values()[0]) == 1: |
| 706 if len(filenames) > 1: | 741 def show_list(): |
| 707 header = "%s:\n" | 742 return self.report(self.show_contents) |
| 708 else: | 743 else: |
| 709 header = None | 744 def show_list(): |
| 710 for filename in filenames: | 745 if self.current_filename == self.filenames[0]: |
| 711 if header: | 746 print "%s:\n" % (self.current_filename,), |
| 712 print header % (filename,), | 747 else: |
| 713 header = "\n%s:\n" | 748 print "\n%s:\n" % (self.current_filename,), |
| 714 self.current_filename = filename | 749 return self.report(self.show_contents) |
| 715 success = (self.report(self.get_extractor) and | 750 return show_list |
| 716 self.report(self.show_contents)) | |
| 717 self.record_status(success) | |
| 718 | 751 |
| 719 def run(self): | 752 def run(self): |
| 720 if self.options.show_list: | 753 if self.options.show_list: |
| 721 self.show_list() | 754 action_function = self.make_list() |
| 722 else: | 755 else: |
| 723 self.extract() | 756 action_function = self.extract |
| 757 while self.archives: | |
| 758 self.current_directory, self.filenames = self.archives.popitem() | |
| 759 os.chdir(self.current_directory) | |
| 760 for filename in self.filenames: | |
| 761 self.current_filename = filename | |
| 762 builder = ExtractorBuilder(self.current_filename, self.options) | |
| 763 for extractor in builder.get_extractor(): | |
| 764 self.current_extractor = extractor | |
| 765 success = action_function() | |
| 766 if success: | |
| 767 self.record_status(success) | |
| 768 break | |
| 769 else: | |
| 770 self.record_status(success=False) | |
| 771 self.options.one_entry_policy.permanent_policy = EXTRACT_WRAP | |
| 724 if self.failures: | 772 if self.failures: |
| 725 return 1 | 773 return 1 |
| 726 return 0 | 774 return 0 |
| 727 | 775 |
| 728 | 776 |
