~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Jelmer Vernooij
  • Date: 2011-09-26 15:21:01 UTC
  • mfrom: (6165.2.3 avoid-revision-history)
  • mto: This revision was merged to the branch mainline in revision 6216.
  • Revision ID: jelmer@samba.org-20110926152101-afcxw1hikybyivfd
merge avoid-revision-history.

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
 
1
# Copyright (C) 2005-2011 Canonical Ltd.
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
17
17
import difflib
18
18
import os
19
19
import re
20
 
import shutil
21
20
import string
22
21
import sys
23
22
 
26
25
import errno
27
26
import subprocess
28
27
import tempfile
29
 
import time
30
28
 
31
29
from bzrlib import (
32
 
    branch as _mod_branch,
33
30
    bzrdir,
34
 
    commands,
 
31
    cmdline,
 
32
    cleanup,
35
33
    errors,
36
34
    osutils,
37
35
    patiencediff,
41
39
    )
42
40
 
43
41
from bzrlib.workingtree import WorkingTree
 
42
from bzrlib.i18n import gettext
44
43
""")
45
44
 
46
 
from bzrlib.symbol_versioning import (
47
 
    deprecated_function,
 
45
from bzrlib.registry import (
 
46
    Registry,
48
47
    )
49
48
from bzrlib.trace import mutter, note, warning
50
49
 
95
94
    if sequence_matcher is None:
96
95
        sequence_matcher = patiencediff.PatienceSequenceMatcher
97
96
    ud = patiencediff.unified_diff(oldlines, newlines,
98
 
                      fromfile=old_filename.encode(path_encoding),
99
 
                      tofile=new_filename.encode(path_encoding),
 
97
                      fromfile=old_filename.encode(path_encoding, 'replace'),
 
98
                      tofile=new_filename.encode(path_encoding, 'replace'),
100
99
                      sequencematcher=sequence_matcher)
101
100
 
102
101
    ud = list(ud)
286
285
                        new_abspath, e)
287
286
 
288
287
 
289
 
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
290
 
                                   apply_view=True):
 
288
def get_trees_and_branches_to_diff_locked(
 
289
    path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
291
290
    """Get the trees and specific files to diff given a list of paths.
292
291
 
293
292
    This method works out the trees to be diff'ed and the files of
304
303
    :param new_url:
305
304
        The url of the new branch or tree. If None, the tree to use is
306
305
        taken from the first path, if any, or the current working tree.
 
306
    :param add_cleanup:
 
307
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
 
308
        will register cleanups that must be run to unlock the trees, etc.
307
309
    :param apply_view:
308
310
        if True and a view is set, apply the view or check that the paths
309
311
        are within it
310
312
    :returns:
311
313
        a tuple of (old_tree, new_tree, old_branch, new_branch,
312
314
        specific_files, extra_trees) where extra_trees is a sequence of
313
 
        additional trees to search in for file-ids.
 
315
        additional trees to search in for file-ids.  The trees and branches
 
316
        will be read-locked until the cleanups registered via the add_cleanup
 
317
        param are run.
314
318
    """
315
319
    # Get the old and new revision specs
316
320
    old_revision_spec = None
339
343
        default_location = path_list[0]
340
344
        other_paths = path_list[1:]
341
345
 
 
346
    def lock_tree_or_branch(wt, br):
 
347
        if wt is not None:
 
348
            wt.lock_read()
 
349
            add_cleanup(wt.unlock)
 
350
        elif br is not None:
 
351
            br.lock_read()
 
352
            add_cleanup(br.unlock)
 
353
 
342
354
    # Get the old location
343
355
    specific_files = []
344
356
    if old_url is None:
345
357
        old_url = default_location
346
358
    working_tree, branch, relpath = \
347
359
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
 
360
    lock_tree_or_branch(working_tree, branch)
348
361
    if consider_relpath and relpath != '':
349
362
        if working_tree is not None and apply_view:
350
363
            views.check_path_in_view(working_tree, relpath)
358
371
    if new_url != old_url:
359
372
        working_tree, branch, relpath = \
360
373
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
 
374
        lock_tree_or_branch(working_tree, branch)
361
375
        if consider_relpath and relpath != '':
362
376
            if working_tree is not None and apply_view:
363
377
                views.check_path_in_view(working_tree, relpath)
368
382
 
369
383
    # Get the specific files (all files is None, no files is [])
370
384
    if make_paths_wt_relative and working_tree is not None:
371
 
        try:
372
 
            from bzrlib.builtins import safe_relpath_files
373
 
            other_paths = safe_relpath_files(working_tree, other_paths,
 
385
        other_paths = working_tree.safe_relpath_files(
 
386
            other_paths,
374
387
            apply_view=apply_view)
375
 
        except errors.FileInWrongBranch:
376
 
            raise errors.BzrCommandError("Files are in different branches")
377
388
    specific_files.extend(other_paths)
378
389
    if len(specific_files) == 0:
379
390
        specific_files = None
383
394
            if view_files:
384
395
                specific_files = view_files
385
396
                view_str = views.view_display_str(view_files)
386
 
                note("*** Ignoring files outside view. View is %s" % view_str)
 
397
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
387
398
 
388
399
    # Get extra trees that ought to be searched for file-ids
389
400
    extra_trees = None
390
401
    if working_tree is not None and working_tree not in (old_tree, new_tree):
391
402
        extra_trees = (working_tree,)
392
 
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
 
403
    return (old_tree, new_tree, old_branch, new_branch,
 
404
            specific_files, extra_trees)
393
405
 
394
406
 
395
407
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
411
423
                    old_label='a/', new_label='b/',
412
424
                    extra_trees=None,
413
425
                    path_encoding='utf8',
414
 
                    using=None):
 
426
                    using=None,
 
427
                    format_cls=None):
415
428
    """Show in text form the changes from one tree to another.
416
429
 
417
 
    to_file
418
 
        The output stream.
419
 
 
420
 
    specific_files
421
 
        Include only changes to these files - None for all changes.
422
 
 
423
 
    external_diff_options
424
 
        If set, use an external GNU diff and pass these options.
425
 
 
426
 
    extra_trees
427
 
        If set, more Trees to use for looking up file ids
428
 
 
429
 
    path_encoding
430
 
        If set, the path will be encoded as specified, otherwise is supposed
431
 
        to be utf8
 
430
    :param to_file: The output stream.
 
431
    :param specific_files: Include only changes to these files - None for all
 
432
        changes.
 
433
    :param external_diff_options: If set, use an external GNU diff and pass 
 
434
        these options.
 
435
    :param extra_trees: If set, more Trees to use for looking up file ids
 
436
    :param path_encoding: If set, the path will be encoded as specified, 
 
437
        otherwise is supposed to be utf8
 
438
    :param format_cls: Formatter class (DiffTree subclass)
432
439
    """
 
440
    if format_cls is None:
 
441
        format_cls = DiffTree
433
442
    old_tree.lock_read()
434
443
    try:
435
444
        if extra_trees is not None:
437
446
                tree.lock_read()
438
447
        new_tree.lock_read()
439
448
        try:
440
 
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
441
 
                                                 path_encoding,
442
 
                                                 external_diff_options,
443
 
                                                 old_label, new_label, using)
 
449
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
 
450
                                                   path_encoding,
 
451
                                                   external_diff_options,
 
452
                                                   old_label, new_label, using)
444
453
            return differ.show_diff(specific_files, extra_trees)
445
454
        finally:
446
455
            new_tree.unlock()
657
666
        """
658
667
        def _get_text(tree, file_id, path):
659
668
            if file_id is not None:
660
 
                return tree.get_file(file_id, path).readlines()
 
669
                return tree.get_file_lines(file_id, path)
661
670
            else:
662
671
                return []
663
672
        try:
664
673
            from_text = _get_text(self.old_tree, from_file_id, from_path)
665
674
            to_text = _get_text(self.new_tree, to_file_id, to_path)
666
675
            self.text_differ(from_label, from_text, to_label, to_text,
667
 
                             self.to_file)
 
676
                             self.to_file, path_encoding=self.path_encoding)
668
677
        except errors.BinaryFile:
669
678
            self.to_file.write(
670
679
                  ("Binary files %s and %s differ\n" %
671
 
                  (from_label, to_label)).encode(self.path_encoding))
 
680
                  (from_label, to_label)).encode(self.path_encoding,'replace'))
672
681
        return self.CHANGED
673
682
 
674
683
 
683
692
    @classmethod
684
693
    def from_string(klass, command_string, old_tree, new_tree, to_file,
685
694
                    path_encoding='utf-8'):
686
 
        command_template = commands.shlex_split_unicode(command_string)
 
695
        command_template = cmdline.split(command_string)
687
696
        if '@' not in command_string:
688
697
            command_template.extend(['@old_path', '@new_path'])
689
698
        return klass(command_template, old_tree, new_tree, to_file,
690
699
                     path_encoding)
691
700
 
692
701
    @classmethod
693
 
    def make_from_diff_tree(klass, command_string):
 
702
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
694
703
        def from_diff_tree(diff_tree):
695
 
            return klass.from_string(command_string, diff_tree.old_tree,
 
704
            full_command_string = [command_string]
 
705
            if external_diff_options is not None:
 
706
                full_command_string += ' ' + external_diff_options
 
707
            return klass.from_string(full_command_string, diff_tree.old_tree,
696
708
                                     diff_tree.new_tree, diff_tree.to_file)
697
709
        return from_diff_tree
698
710
 
699
711
    def _get_command(self, old_path, new_path):
700
712
        my_map = {'old_path': old_path, 'new_path': new_path}
701
 
        return [AtTemplate(t).substitute(my_map) for t in
702
 
                self.command_template]
 
713
        command = [AtTemplate(t).substitute(my_map) for t in
 
714
                   self.command_template]
 
715
        if sys.platform == 'win32': # Popen doesn't accept unicode on win32
 
716
            command_encoded = []
 
717
            for c in command:
 
718
                if isinstance(c, unicode):
 
719
                    command_encoded.append(c.encode('mbcs'))
 
720
                else:
 
721
                    command_encoded.append(c)
 
722
            return command_encoded
 
723
        else:
 
724
            return command
703
725
 
704
726
    def _execute(self, old_path, new_path):
705
727
        command = self._get_command(old_path, new_path)
725
747
                raise
726
748
        return True
727
749
 
 
750
    @staticmethod
 
751
    def _fenc():
 
752
        """Returns safe encoding for passing file path to diff tool"""
 
753
        if sys.platform == 'win32':
 
754
            return 'mbcs'
 
755
        else:
 
756
            # Don't fallback to 'utf-8' because subprocess may not be able to
 
757
            # handle utf-8 correctly when locale is not utf-8.
 
758
            return sys.getfilesystemencoding() or 'ascii'
 
759
 
 
760
    def _is_safepath(self, path):
 
761
        """Return true if `path` may be able to pass to subprocess."""
 
762
        fenc = self._fenc()
 
763
        try:
 
764
            return path == path.encode(fenc).decode(fenc)
 
765
        except UnicodeError:
 
766
            return False
 
767
 
 
768
    def _safe_filename(self, prefix, relpath):
 
769
        """Replace unsafe character in `relpath` then join `self._root`,
 
770
        `prefix` and `relpath`."""
 
771
        fenc = self._fenc()
 
772
        # encoded_str.replace('?', '_') may break multibyte char.
 
773
        # So we should encode, decode, then replace(u'?', u'_')
 
774
        relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
 
775
        relpath_tmp = relpath_tmp.replace(u'?', u'_')
 
776
        return osutils.pathjoin(self._root, prefix, relpath_tmp)
 
777
 
728
778
    def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
729
779
                    allow_write=False):
730
780
        if not force_temp and isinstance(tree, WorkingTree):
731
 
            return tree.abspath(tree.id2path(file_id))
732
 
        
733
 
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
781
            full_path = tree.abspath(tree.id2path(file_id))
 
782
            if self._is_safepath(full_path):
 
783
                return full_path
 
784
 
 
785
        full_path = self._safe_filename(prefix, relpath)
734
786
        if not force_temp and self._try_symlink_root(tree, prefix):
735
787
            return full_path
736
788
        parent_dir = osutils.dirname(full_path)
748
800
                target.close()
749
801
        finally:
750
802
            source.close()
 
803
        try:
 
804
            mtime = tree.get_file_mtime(file_id)
 
805
        except errors.FileTimestampUnavailable:
 
806
            pass
 
807
        else:
 
808
            os.utime(full_path, (mtime, mtime))
751
809
        if not allow_write:
752
810
            osutils.make_readonly(full_path)
753
 
        try:
754
 
            mtime = tree.get_file_mtime(file_id)
755
 
        except errors.FileTimestampUnavailable:
756
 
            mtime = 0
757
 
        os.utime(full_path, (mtime, mtime))
758
811
        return full_path
759
812
 
760
813
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
792
845
        """
793
846
        old_path = self.old_tree.id2path(file_id)
794
847
        new_path = self.new_tree.id2path(file_id)
795
 
        new_abs_path = self._prepare_files(file_id, old_path, new_path,
796
 
                                           allow_write_new=True,
797
 
                                           force_temp=True)[1]
798
 
        command = self._get_command(osutils.pathjoin('old', old_path),
799
 
                                    osutils.pathjoin('new', new_path))
 
848
        old_abs_path, new_abs_path = self._prepare_files(
 
849
                                            file_id, old_path, new_path,
 
850
                                            allow_write_new=True,
 
851
                                            force_temp=True)
 
852
        command = self._get_command(old_abs_path, new_abs_path)
800
853
        subprocess.call(command, cwd=self._root)
801
 
        new_file = open(new_abs_path, 'r')
 
854
        new_file = open(new_abs_path, 'rb')
802
855
        try:
803
856
            return new_file.read()
804
857
        finally:
854
907
        """Factory for producing a DiffTree.
855
908
 
856
909
        Designed to accept options used by show_diff_trees.
 
910
 
857
911
        :param old_tree: The tree to show as old in the comparison
858
912
        :param new_tree: The tree to show as new in the comparison
859
913
        :param to_file: File to write comparisons to
865
919
        :param using: Commandline to use to invoke an external diff tool
866
920
        """
867
921
        if using is not None:
868
 
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
 
922
            extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
869
923
        else:
870
924
            extra_factories = []
871
925
        if external_diff_options:
872
926
            opts = external_diff_options.split()
873
 
            def diff_file(olab, olines, nlab, nlines, to_file):
 
927
            def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None):
 
928
                """:param path_encoding: not used but required
 
929
                        to match the signature of internal_diff.
 
930
                """
874
931
                external_diff(olab, olines, nlab, nlines, to_file, opts)
875
932
        else:
876
933
            diff_file = internal_diff
882
939
    def show_diff(self, specific_files, extra_trees=None):
883
940
        """Write tree diff to self.to_file
884
941
 
885
 
        :param sepecific_files: the specific files to compare (recursive)
 
942
        :param specific_files: the specific files to compare (recursive)
886
943
        :param extra_trees: extra trees to use for mapping paths to file_ids
887
944
        """
888
945
        try:
978
1035
            if error_path is None:
979
1036
                error_path = old_path
980
1037
            raise errors.NoDiffFound(error_path)
 
1038
 
 
1039
 
 
1040
format_registry = Registry()
 
1041
format_registry.register('default', DiffTree)