← 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: 2010-08-13 19:08:57 UTC
  • mto: (5050.17.7 2.2)
  • mto: This revision was merged to the branch mainline in revision 5379.
  • Revision ID: john@arbash-meinel.com-20100813190857-mvzwnimrxvm0zimp
Lots of documentation updates.

We had a lot of http links pointing to the old domain. They should
all now be properly updated to the new domain. (only bazaar-vcs.org
entry left is for pqm, which seems to still reside at the old url.)

Also removed one 'TODO' doc entry about switching to binary xdelta, since
we basically did just that with groupcompress.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
 
1
# Copyright (C) 2005-2010 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
 
20
import string
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
 
    commands,
 
33
    cmdline,
 
34
    cleanup,
34
35
    errors,
35
36
    osutils,
36
37
    patiencediff,
38
39
    timestamp,
39
40
    views,
40
41
    )
 
42
 
 
43
from bzrlib.workingtree import WorkingTree
41
44
""")
42
45
 
 
46
from bzrlib.registry import (
 
47
    Registry,
 
48
    )
43
49
from bzrlib.symbol_versioning import (
44
50
    deprecated_function,
 
51
    deprecated_in,
45
52
    )
46
53
from bzrlib.trace import mutter, note, warning
47
54
 
48
55
 
 
56
class AtTemplate(string.Template):
 
57
    """Templating class that uses @ instead of $."""
 
58
 
 
59
    delimiter = '@'
 
60
 
 
61
 
49
62
# TODO: Rather than building a changeset object, we should probably
50
63
# invoke callbacks on an object.  That object can either accumulate a
51
64
# list, write them out directly, etc etc.
86
99
    if sequence_matcher is None:
87
100
        sequence_matcher = patiencediff.PatienceSequenceMatcher
88
101
    ud = patiencediff.unified_diff(oldlines, newlines,
89
 
                      fromfile=old_filename.encode(path_encoding),
90
 
                      tofile=new_filename.encode(path_encoding),
 
102
                      fromfile=old_filename.encode(path_encoding, 'replace'),
 
103
                      tofile=new_filename.encode(path_encoding, 'replace'),
91
104
                      sequencematcher=sequence_matcher)
92
105
 
93
106
    ud = list(ud)
171
184
 
172
185
        if not diff_opts:
173
186
            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')
174
192
        diffcmd = ['diff',
175
193
                   '--label', old_filename,
176
194
                   old_abspath,
272
290
                        new_abspath, e)
273
291
 
274
292
 
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.
 
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.
300
356
    """
301
357
    # Get the old and new revision specs
302
358
    old_revision_spec = None
325
381
        default_location = path_list[0]
326
382
        other_paths = path_list[1:]
327
383
 
 
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
 
328
392
    # Get the old location
329
393
    specific_files = []
330
394
    if old_url is None:
331
395
        old_url = default_location
332
396
    working_tree, branch, relpath = \
333
397
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
 
398
    lock_tree_or_branch(working_tree, branch)
334
399
    if consider_relpath and relpath != '':
335
400
        if working_tree is not None and apply_view:
336
401
            views.check_path_in_view(working_tree, relpath)
337
402
        specific_files.append(relpath)
338
403
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
 
404
    old_branch = branch
339
405
 
340
406
    # Get the new location
341
407
    if new_url is None:
343
409
    if new_url != old_url:
344
410
        working_tree, branch, relpath = \
345
411
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
 
412
        lock_tree_or_branch(working_tree, branch)
346
413
        if consider_relpath and relpath != '':
347
414
            if working_tree is not None and apply_view:
348
415
                views.check_path_in_view(working_tree, relpath)
349
416
            specific_files.append(relpath)
350
417
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
351
418
        basis_is_default=working_tree is None)
 
419
    new_branch = branch
352
420
 
353
421
    # Get the specific files (all files is None, no files is [])
354
422
    if make_paths_wt_relative and working_tree is not None:
373
441
    extra_trees = None
374
442
    if working_tree is not None and working_tree not in (old_tree, new_tree):
375
443
        extra_trees = (working_tree,)
376
 
    return old_tree, new_tree, specific_files, extra_trees
 
444
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
 
445
 
377
446
 
378
447
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
379
448
    if branch is None and tree is not None:
394
463
                    old_label='a/', new_label='b/',
395
464
                    extra_trees=None,
396
465
                    path_encoding='utf8',
397
 
                    using=None):
 
466
                    using=None,
 
467
                    format_cls=None):
398
468
    """Show in text form the changes from one tree to another.
399
469
 
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
 
470
    :param to_file: The output stream.
 
471
    :param specific_files:Include only changes to these files - None for all
 
472
        changes.
 
473
    :param external_diff_options: If set, use an external GNU diff and pass 
 
474
        these options.
 
475
    :param extra_trees: If set, more Trees to use for looking up file ids
 
476
    :param path_encoding: If set, the path will be encoded as specified, 
 
477
        otherwise is supposed to be utf8
 
478
    :param format_cls: Formatter class (DiffTree subclass)
415
479
    """
 
480
    if format_cls is None:
 
481
        format_cls = DiffTree
416
482
    old_tree.lock_read()
417
483
    try:
418
484
        if extra_trees is not None:
420
486
                tree.lock_read()
421
487
        new_tree.lock_read()
422
488
        try:
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)
 
489
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
 
490
                                                   path_encoding,
 
491
                                                   external_diff_options,
 
492
                                                   old_label, new_label, using)
427
493
            return differ.show_diff(specific_files, extra_trees)
428
494
        finally:
429
495
            new_tree.unlock()
436
502
 
437
503
def _patch_header_date(tree, file_id, path):
438
504
    """Returns a timestamp suitable for use in a patch header."""
439
 
    mtime = tree.get_file_mtime(file_id, path)
 
505
    try:
 
506
        mtime = tree.get_file_mtime(file_id, path)
 
507
    except errors.FileTimestampUnavailable:
 
508
        mtime = 0
440
509
    return timestamp.format_patch_date(mtime)
441
510
 
442
511
 
637
706
        """
638
707
        def _get_text(tree, file_id, path):
639
708
            if file_id is not None:
640
 
                return tree.get_file(file_id, path).readlines()
 
709
                return tree.get_file_lines(file_id, path)
641
710
            else:
642
711
                return []
643
712
        try:
644
713
            from_text = _get_text(self.old_tree, from_file_id, from_path)
645
714
            to_text = _get_text(self.new_tree, to_file_id, to_path)
646
715
            self.text_differ(from_label, from_text, to_label, to_text,
647
 
                             self.to_file)
 
716
                             self.to_file, path_encoding=self.path_encoding)
648
717
        except errors.BinaryFile:
649
718
            self.to_file.write(
650
719
                  ("Binary files %s and %s differ\n" %
651
 
                  (from_label, to_label)).encode(self.path_encoding))
 
720
                  (from_label, to_label)).encode(self.path_encoding,'replace'))
652
721
        return self.CHANGED
653
722
 
654
723
 
663
732
    @classmethod
664
733
    def from_string(klass, command_string, old_tree, new_tree, to_file,
665
734
                    path_encoding='utf-8'):
666
 
        command_template = commands.shlex_split_unicode(command_string)
667
 
        command_template.extend(['%(old_path)s', '%(new_path)s'])
 
735
        command_template = cmdline.split(command_string)
 
736
        if '@' not in command_string:
 
737
            command_template.extend(['@old_path', '@new_path'])
668
738
        return klass(command_template, old_tree, new_tree, to_file,
669
739
                     path_encoding)
670
740
 
677
747
 
678
748
    def _get_command(self, old_path, new_path):
679
749
        my_map = {'old_path': old_path, 'new_path': new_path}
680
 
        return [t % my_map for t in self.command_template]
 
750
        return [AtTemplate(t).substitute(my_map) for t in
 
751
                self.command_template]
681
752
 
682
753
    def _execute(self, old_path, new_path):
683
754
        command = self._get_command(old_path, new_path)
703
774
                raise
704
775
        return True
705
776
 
706
 
    def _write_file(self, file_id, tree, prefix, relpath):
 
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
        
707
782
        full_path = osutils.pathjoin(self._root, prefix, relpath)
708
 
        if self._try_symlink_root(tree, prefix):
 
783
        if not force_temp and self._try_symlink_root(tree, prefix):
709
784
            return full_path
710
785
        parent_dir = osutils.dirname(full_path)
711
786
        try:
722
797
                target.close()
723
798
        finally:
724
799
            source.close()
725
 
        osutils.make_readonly(full_path)
726
 
        mtime = tree.get_file_mtime(file_id)
727
 
        os.utime(full_path, (mtime, mtime))
 
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)
728
808
        return full_path
729
809
 
730
 
    def _prepare_files(self, file_id, old_path, new_path):
 
810
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
 
811
                       allow_write_new=False):
731
812
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
732
 
                                         old_path)
 
813
                                         old_path, force_temp)
733
814
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
734
 
                                         new_path)
 
815
                                         new_path, force_temp,
 
816
                                         allow_write=allow_write_new)
735
817
        return old_disk_path, new_disk_path
736
818
 
737
819
    def finish(self):
745
827
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
746
828
        if (old_kind, new_kind) != ('file', 'file'):
747
829
            return DiffPath.CANNOT_DIFF
748
 
        self._prepare_files(file_id, old_path, new_path)
749
 
        self._execute(osutils.pathjoin('old', old_path),
750
 
                      osutils.pathjoin('new', new_path))
 
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()
751
856
 
752
857
 
753
858
class DiffTree(object):
815
920
            extra_factories = []
816
921
        if external_diff_options:
817
922
            opts = external_diff_options.split()
818
 
            def diff_file(olab, olines, nlab, nlines, to_file):
 
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
                """
819
927
                external_diff(olab, olines, nlab, nlines, to_file, opts)
820
928
        else:
821
929
            diff_file = internal_diff
827
935
    def show_diff(self, specific_files, extra_trees=None):
828
936
        """Write tree diff to self.to_file
829
937
 
830
 
        :param sepecific_files: the specific files to compare (recursive)
 
938
        :param specific_files: the specific files to compare (recursive)
831
939
        :param extra_trees: extra trees to use for mapping paths to file_ids
832
940
        """
833
941
        try:
923
1031
            if error_path is None:
924
1032
                error_path = old_path
925
1033
            raise errors.NoDiffFound(error_path)
 
1034
 
 
1035
 
 
1036
format_registry = Registry()
 
1037
format_registry.register('default', DiffTree)