← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: John Arbash Meinel
  • Date: 2009-03-27 22:29:55 UTC
  • mto: (3735.39.2 clean)
  • mto: This revision was merged to the branch mainline in revision 4280.
  • Revision ID: john@arbash-meinel.com-20090327222955-utifmfm888zerixt
Implement apply_delta_to_source which doesn't have to malloc another string.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# Copyright (C) 2005-2010 Canonical Ltd.
 
1
# Copyright (C) 2004, 2005, 2006 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 string
 
20
import shutil
21
21
import sys
22
22
 
23
23
from bzrlib.lazy_import import lazy_import
30
30
from bzrlib import (
31
31
    branch as _mod_branch,
32
32
    bzrdir,
33
 
    cmdline,
34
 
    cleanup,
 
33
    commands,
35
34
    errors,
36
35
    osutils,
37
36
    patiencediff,
39
38
    timestamp,
40
39
    views,
41
40
    )
42
 
 
43
 
from bzrlib.workingtree import WorkingTree
44
41
""")
45
42
 
46
 
from bzrlib.registry import (
47
 
    Registry,
48
 
    )
49
43
from bzrlib.symbol_versioning import (
50
44
    deprecated_function,
51
 
    deprecated_in,
52
45
    )
53
46
from bzrlib.trace import mutter, note, warning
54
47
 
55
48
 
56
 
class AtTemplate(string.Template):
57
 
    """Templating class that uses @ instead of $."""
58
 
 
59
 
    delimiter = '@'
60
 
 
61
 
 
62
49
# TODO: Rather than building a changeset object, we should probably
63
50
# invoke callbacks on an object.  That object can either accumulate a
64
51
# list, write them out directly, etc etc.
99
86
    if sequence_matcher is None:
100
87
        sequence_matcher = patiencediff.PatienceSequenceMatcher
101
88
    ud = patiencediff.unified_diff(oldlines, newlines,
102
 
                      fromfile=old_filename.encode(path_encoding, 'replace'),
103
 
                      tofile=new_filename.encode(path_encoding, 'replace'),
 
89
                      fromfile=old_filename.encode(path_encoding),
 
90
                      tofile=new_filename.encode(path_encoding),
104
91
                      sequencematcher=sequence_matcher)
105
92
 
106
93
    ud = list(ud)
184
171
 
185
172
        if not diff_opts:
186
173
            diff_opts = []
187
 
        if sys.platform == 'win32':
188
 
            # Popen doesn't do the proper encoding for external commands
189
 
            # Since we are dealing with an ANSI api, use mbcs encoding
190
 
            old_filename = old_filename.encode('mbcs')
191
 
            new_filename = new_filename.encode('mbcs')
192
174
        diffcmd = ['diff',
193
175
                   '--label', old_filename,
194
176
                   old_abspath,
290
272
                        new_abspath, e)
291
273
 
292
274
 
293
 
@deprecated_function(deprecated_in((2, 2, 0)))
294
 
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
295
 
                                   apply_view=True):
296
 
    """Get the trees and specific files to diff given a list of paths.
297
 
 
298
 
    This method works out the trees to be diff'ed and the files of
299
 
    interest within those trees.
300
 
 
301
 
    :param path_list:
302
 
        the list of arguments passed to the diff command
303
 
    :param revision_specs:
304
 
        Zero, one or two RevisionSpecs from the diff command line,
305
 
        saying what revisions to compare.
306
 
    :param old_url:
307
 
        The url of the old branch or tree. If None, the tree to use is
308
 
        taken from the first path, if any, or the current working tree.
309
 
    :param new_url:
310
 
        The url of the new branch or tree. If None, the tree to use is
311
 
        taken from the first path, if any, or the current working tree.
312
 
    :param apply_view:
313
 
        if True and a view is set, apply the view or check that the paths
314
 
        are within it
315
 
    :returns:
316
 
        a tuple of (old_tree, new_tree, old_branch, new_branch,
317
 
        specific_files, extra_trees) where extra_trees is a sequence of
318
 
        additional trees to search in for file-ids.  The trees and branches
319
 
        are not locked.
320
 
    """
321
 
    op = cleanup.OperationWithCleanups(get_trees_and_branches_to_diff_locked)
322
 
    return op.run_simple(path_list, revision_specs, old_url, new_url,
323
 
            op.add_cleanup, apply_view=apply_view)
324
 
    
325
 
 
326
 
def get_trees_and_branches_to_diff_locked(
327
 
    path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
328
 
    """Get the trees and specific files to diff given a list of paths.
329
 
 
330
 
    This method works out the trees to be diff'ed and the files of
331
 
    interest within those trees.
332
 
 
333
 
    :param path_list:
334
 
        the list of arguments passed to the diff command
335
 
    :param revision_specs:
336
 
        Zero, one or two RevisionSpecs from the diff command line,
337
 
        saying what revisions to compare.
338
 
    :param old_url:
339
 
        The url of the old branch or tree. If None, the tree to use is
340
 
        taken from the first path, if any, or the current working tree.
341
 
    :param new_url:
342
 
        The url of the new branch or tree. If None, the tree to use is
343
 
        taken from the first path, if any, or the current working tree.
344
 
    :param add_cleanup:
345
 
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
346
 
        will register cleanups that must be run to unlock the trees, etc.
347
 
    :param apply_view:
348
 
        if True and a view is set, apply the view or check that the paths
349
 
        are within it
350
 
    :returns:
351
 
        a tuple of (old_tree, new_tree, old_branch, new_branch,
352
 
        specific_files, extra_trees) where extra_trees is a sequence of
353
 
        additional trees to search in for file-ids.  The trees and branches
354
 
        will be read-locked until the cleanups registered via the add_cleanup
355
 
        param are run.
 
275
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url,
 
276
    apply_view=True):
 
277
    """Get the trees and specific files to diff given a list of paths.
 
278
 
 
279
    This method works out the trees to be diff'ed and the files of
 
280
    interest within those trees.
 
281
 
 
282
    :param path_list:
 
283
        the list of arguments passed to the diff command
 
284
    :param revision_specs:
 
285
        Zero, one or two RevisionSpecs from the diff command line,
 
286
        saying what revisions to compare.
 
287
    :param old_url:
 
288
        The url of the old branch or tree. If None, the tree to use is
 
289
        taken from the first path, if any, or the current working tree.
 
290
    :param new_url:
 
291
        The url of the new branch or tree. If None, the tree to use is
 
292
        taken from the first path, if any, or the current working tree.
 
293
    :param apply_view:
 
294
        if True and a view is set, apply the view or check that the paths
 
295
        are within it
 
296
    :returns:
 
297
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
298
        extra_trees is a sequence of additional trees to search in for
 
299
        file-ids.
356
300
    """
357
301
    # Get the old and new revision specs
358
302
    old_revision_spec = None
381
325
        default_location = path_list[0]
382
326
        other_paths = path_list[1:]
383
327
 
384
 
    def lock_tree_or_branch(wt, br):
385
 
        if wt is not None:
386
 
            wt.lock_read()
387
 
            add_cleanup(wt.unlock)
388
 
        elif br is not None:
389
 
            br.lock_read()
390
 
            add_cleanup(br.unlock)
391
 
 
392
328
    # Get the old location
393
329
    specific_files = []
394
330
    if old_url is None:
395
331
        old_url = default_location
396
332
    working_tree, branch, relpath = \
397
333
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
398
 
    lock_tree_or_branch(working_tree, branch)
399
334
    if consider_relpath and relpath != '':
400
335
        if working_tree is not None and apply_view:
401
336
            views.check_path_in_view(working_tree, relpath)
402
337
        specific_files.append(relpath)
403
338
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
404
 
    old_branch = branch
405
339
 
406
340
    # Get the new location
407
341
    if new_url is None:
409
343
    if new_url != old_url:
410
344
        working_tree, branch, relpath = \
411
345
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
412
 
        lock_tree_or_branch(working_tree, branch)
413
346
        if consider_relpath and relpath != '':
414
347
            if working_tree is not None and apply_view:
415
348
                views.check_path_in_view(working_tree, relpath)
416
349
            specific_files.append(relpath)
417
350
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
418
351
        basis_is_default=working_tree is None)
419
 
    new_branch = branch
420
352
 
421
353
    # Get the specific files (all files is None, no files is [])
422
354
    if make_paths_wt_relative and working_tree is not None:
423
 
        other_paths = working_tree.safe_relpath_files(
424
 
            other_paths,
 
355
        try:
 
356
            from bzrlib.builtins import safe_relpath_files
 
357
            other_paths = safe_relpath_files(working_tree, other_paths,
425
358
            apply_view=apply_view)
 
359
        except errors.FileInWrongBranch:
 
360
            raise errors.BzrCommandError("Files are in different branches")
426
361
    specific_files.extend(other_paths)
427
362
    if len(specific_files) == 0:
428
363
        specific_files = None
432
367
            if view_files:
433
368
                specific_files = view_files
434
369
                view_str = views.view_display_str(view_files)
435
 
                note("*** Ignoring files outside view. View is %s" % view_str)
 
370
                note("*** ignoring files outside view: %s" % view_str)
436
371
 
437
372
    # Get extra trees that ought to be searched for file-ids
438
373
    extra_trees = None
439
374
    if working_tree is not None and working_tree not in (old_tree, new_tree):
440
375
        extra_trees = (working_tree,)
441
 
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
442
 
 
 
376
    return old_tree, new_tree, specific_files, extra_trees
443
377
 
444
378
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
445
379
    if branch is None and tree is not None:
460
394
                    old_label='a/', new_label='b/',
461
395
                    extra_trees=None,
462
396
                    path_encoding='utf8',
463
 
                    using=None,
464
 
                    format_cls=None):
 
397
                    using=None):
465
398
    """Show in text form the changes from one tree to another.
466
399
 
467
 
    :param to_file: The output stream.
468
 
    :param specific_files:Include only changes to these files - None for all
469
 
        changes.
470
 
    :param external_diff_options: If set, use an external GNU diff and pass 
471
 
        these options.
472
 
    :param extra_trees: If set, more Trees to use for looking up file ids
473
 
    :param path_encoding: If set, the path will be encoded as specified, 
474
 
        otherwise is supposed to be utf8
475
 
    :param format_cls: Formatter class (DiffTree subclass)
 
400
    to_file
 
401
        The output stream.
 
402
 
 
403
    specific_files
 
404
        Include only changes to these files - None for all changes.
 
405
 
 
406
    external_diff_options
 
407
        If set, use an external GNU diff and pass these options.
 
408
 
 
409
    extra_trees
 
410
        If set, more Trees to use for looking up file ids
 
411
 
 
412
    path_encoding
 
413
        If set, the path will be encoded as specified, otherwise is supposed
 
414
        to be utf8
476
415
    """
477
 
    if format_cls is None:
478
 
        format_cls = DiffTree
479
416
    old_tree.lock_read()
480
417
    try:
481
418
        if extra_trees is not None:
483
420
                tree.lock_read()
484
421
        new_tree.lock_read()
485
422
        try:
486
 
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
487
 
                                                   path_encoding,
488
 
                                                   external_diff_options,
489
 
                                                   old_label, new_label, using)
 
423
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
424
                                                 path_encoding,
 
425
                                                 external_diff_options,
 
426
                                                 old_label, new_label, using)
490
427
            return differ.show_diff(specific_files, extra_trees)
491
428
        finally:
492
429
            new_tree.unlock()
499
436
 
500
437
def _patch_header_date(tree, file_id, path):
501
438
    """Returns a timestamp suitable for use in a patch header."""
502
 
    try:
503
 
        mtime = tree.get_file_mtime(file_id, path)
504
 
    except errors.FileTimestampUnavailable:
505
 
        mtime = 0
 
439
    mtime = tree.get_file_mtime(file_id, path)
506
440
    return timestamp.format_patch_date(mtime)
507
441
 
508
442
 
686
620
            return self.CANNOT_DIFF
687
621
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
688
622
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
689
 
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
690
 
            old_path, new_path)
 
623
        return self.diff_text(from_file_id, to_file_id, from_label, to_label)
691
624
 
692
 
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
693
 
        from_path=None, to_path=None):
 
625
    def diff_text(self, from_file_id, to_file_id, from_label, to_label):
694
626
        """Diff the content of given files in two trees
695
627
 
696
628
        :param from_file_id: The id of the file in the from tree.  If None,
698
630
        :param to_file_id: The id of the file in the to tree.  This may refer
699
631
            to a different file from from_file_id.  If None,
700
632
            the file is not present in the to tree.
701
 
        :param from_path: The path in the from tree or None if unknown.
702
 
        :param to_path: The path in the to tree or None if unknown.
703
633
        """
704
 
        def _get_text(tree, file_id, path):
 
634
        def _get_text(tree, file_id):
705
635
            if file_id is not None:
706
 
                return tree.get_file_lines(file_id, path)
 
636
                return tree.get_file(file_id).readlines()
707
637
            else:
708
638
                return []
709
639
        try:
710
 
            from_text = _get_text(self.old_tree, from_file_id, from_path)
711
 
            to_text = _get_text(self.new_tree, to_file_id, to_path)
 
640
            from_text = _get_text(self.old_tree, from_file_id)
 
641
            to_text = _get_text(self.new_tree, to_file_id)
712
642
            self.text_differ(from_label, from_text, to_label, to_text,
713
 
                             self.to_file, path_encoding=self.path_encoding)
 
643
                             self.to_file)
714
644
        except errors.BinaryFile:
715
645
            self.to_file.write(
716
646
                  ("Binary files %s and %s differ\n" %
717
 
                  (from_label, to_label)).encode(self.path_encoding,'replace'))
 
647
                  (from_label, to_label)).encode(self.path_encoding))
718
648
        return self.CHANGED
719
649
 
720
650
 
729
659
    @classmethod
730
660
    def from_string(klass, command_string, old_tree, new_tree, to_file,
731
661
                    path_encoding='utf-8'):
732
 
        command_template = cmdline.split(command_string)
733
 
        if '@' not in command_string:
734
 
            command_template.extend(['@old_path', '@new_path'])
 
662
        command_template = commands.shlex_split_unicode(command_string)
 
663
        command_template.extend(['%(old_path)s', '%(new_path)s'])
735
664
        return klass(command_template, old_tree, new_tree, to_file,
736
665
                     path_encoding)
737
666
 
738
667
    @classmethod
739
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
668
    def make_from_diff_tree(klass, command_string):
740
669
        def from_diff_tree(diff_tree):
741
 
            full_command_string = [command_string]
742
 
            if external_diff_options is not None:
743
 
                full_command_string += ' ' + external_diff_options
744
 
            return klass.from_string(full_command_string, diff_tree.old_tree,
 
670
            return klass.from_string(command_string, diff_tree.old_tree,
745
671
                                     diff_tree.new_tree, diff_tree.to_file)
746
672
        return from_diff_tree
747
673
 
748
674
    def _get_command(self, old_path, new_path):
749
675
        my_map = {'old_path': old_path, 'new_path': new_path}
750
 
        return [AtTemplate(t).substitute(my_map) for t in
751
 
                self.command_template]
 
676
        return [t % my_map for t in self.command_template]
752
677
 
753
678
    def _execute(self, old_path, new_path):
754
679
        command = self._get_command(old_path, new_path)
774
699
                raise
775
700
        return True
776
701
 
777
 
    def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
778
 
                    allow_write=False):
779
 
        if not force_temp and isinstance(tree, WorkingTree):
780
 
            return tree.abspath(tree.id2path(file_id))
781
 
        
 
702
    def _write_file(self, file_id, tree, prefix, relpath):
782
703
        full_path = osutils.pathjoin(self._root, prefix, relpath)
783
 
        if not force_temp and self._try_symlink_root(tree, prefix):
 
704
        if self._try_symlink_root(tree, prefix):
784
705
            return full_path
785
706
        parent_dir = osutils.dirname(full_path)
786
707
        try:
797
718
                target.close()
798
719
        finally:
799
720
            source.close()
800
 
        try:
801
 
            mtime = tree.get_file_mtime(file_id)
802
 
        except errors.FileTimestampUnavailable:
803
 
            pass
804
 
        else:
805
 
            os.utime(full_path, (mtime, mtime))
806
 
        if not allow_write:
807
 
            osutils.make_readonly(full_path)
 
721
        osutils.make_readonly(full_path)
 
722
        mtime = tree.get_file_mtime(file_id)
 
723
        os.utime(full_path, (mtime, mtime))
808
724
        return full_path
809
725
 
810
 
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
811
 
                       allow_write_new=False):
 
726
    def _prepare_files(self, file_id, old_path, new_path):
812
727
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
813
 
                                         old_path, force_temp)
 
728
                                         old_path)
814
729
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
815
 
                                         new_path, force_temp,
816
 
                                         allow_write=allow_write_new)
 
730
                                         new_path)
817
731
        return old_disk_path, new_disk_path
818
732
 
819
733
    def finish(self):
820
 
        try:
821
 
            osutils.rmtree(self._root)
822
 
        except OSError, e:
823
 
            if e.errno != errno.ENOENT:
824
 
                mutter("The temporary directory \"%s\" was not "
825
 
                        "cleanly removed: %s." % (self._root, e))
 
734
        osutils.rmtree(self._root)
826
735
 
827
736
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
828
737
        if (old_kind, new_kind) != ('file', 'file'):
829
738
            return DiffPath.CANNOT_DIFF
830
 
        (old_disk_path, new_disk_path) = self._prepare_files(
831
 
                                                file_id, old_path, new_path)
832
 
        self._execute(old_disk_path, new_disk_path)
833
 
 
834
 
    def edit_file(self, file_id):
835
 
        """Use this tool to edit a file.
836
 
 
837
 
        A temporary copy will be edited, and the new contents will be
838
 
        returned.
839
 
 
840
 
        :param file_id: The id of the file to edit.
841
 
        :return: The new contents of the file.
842
 
        """
843
 
        old_path = self.old_tree.id2path(file_id)
844
 
        new_path = self.new_tree.id2path(file_id)
845
 
        new_abs_path = self._prepare_files(file_id, old_path, new_path,
846
 
                                           allow_write_new=True,
847
 
                                           force_temp=True)[1]
848
 
        command = self._get_command(osutils.pathjoin('old', old_path),
849
 
                                    osutils.pathjoin('new', new_path))
850
 
        subprocess.call(command, cwd=self._root)
851
 
        new_file = open(new_abs_path, 'r')
852
 
        try:
853
 
            return new_file.read()
854
 
        finally:
855
 
            new_file.close()
 
739
        self._prepare_files(file_id, old_path, new_path)
 
740
        self._execute(osutils.pathjoin('old', old_path),
 
741
                      osutils.pathjoin('new', new_path))
856
742
 
857
743
 
858
744
class DiffTree(object):
915
801
        :param using: Commandline to use to invoke an external diff tool
916
802
        """
917
803
        if using is not None:
918
 
            extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
 
804
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
919
805
        else:
920
806
            extra_factories = []
921
807
        if external_diff_options:
922
808
            opts = external_diff_options.split()
923
 
            def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None):
924
 
                """:param path_encoding: not used but required
925
 
                        to match the signature of internal_diff.
926
 
                """
 
809
            def diff_file(olab, olines, nlab, nlines, to_file):
927
810
                external_diff(olab, olines, nlab, nlines, to_file, opts)
928
811
        else:
929
812
            diff_file = internal_diff
935
818
    def show_diff(self, specific_files, extra_trees=None):
936
819
        """Write tree diff to self.to_file
937
820
 
938
 
        :param specific_files: the specific files to compare (recursive)
 
821
        :param sepecific_files: the specific files to compare (recursive)
939
822
        :param extra_trees: extra trees to use for mapping paths to file_ids
940
823
        """
941
824
        try:
999
882
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
1000
883
                                   newpath_encoded, prop_str))
1001
884
            if changed_content:
1002
 
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
 
885
                self.diff(file_id, oldpath, newpath)
1003
886
                has_changes = 1
1004
887
            if renamed:
1005
888
                has_changes = 1
1020
903
            new_kind = self.new_tree.kind(file_id)
1021
904
        except (errors.NoSuchId, errors.NoSuchFile):
1022
905
            new_kind = None
1023
 
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
1024
 
 
1025
 
 
1026
 
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
906
 
1027
907
        result = DiffPath._diff_many(self.differs, file_id, old_path,
1028
908
                                       new_path, old_kind, new_kind)
1029
909
        if result is DiffPath.CANNOT_DIFF:
1031
911
            if error_path is None:
1032
912
                error_path = old_path
1033
913
            raise errors.NoDiffFound(error_path)
1034
 
 
1035
 
 
1036
 
format_registry = Registry()
1037
 
format_registry.register('default', DiffTree)