~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Martin Pool
  • Date: 2009-10-29 06:11:23 UTC
  • mto: This revision was merged to the branch mainline in revision 4776.
  • Revision ID: mbp@sourcefrog.net-20091029061123-vyjm0si2r97rgsp2
one more the the

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2005-2011 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
25
25
import errno
26
26
import subprocess
27
27
import tempfile
 
28
import time
28
29
 
29
30
from bzrlib import (
 
31
    branch as _mod_branch,
30
32
    bzrdir,
31
 
    cmdline,
32
 
    cleanup,
 
33
    commands,
33
34
    errors,
34
35
    osutils,
35
36
    patiencediff,
37
38
    timestamp,
38
39
    views,
39
40
    )
40
 
 
41
 
from bzrlib.workingtree import WorkingTree
42
41
""")
43
42
 
44
 
from bzrlib.registry import (
45
 
    Registry,
 
43
from bzrlib.symbol_versioning import (
 
44
    deprecated_function,
46
45
    )
47
46
from bzrlib.trace import mutter, note, warning
48
47
 
49
48
 
50
 
class AtTemplate(string.Template):
51
 
    """Templating class that uses @ instead of $."""
52
 
 
53
 
    delimiter = '@'
54
 
 
55
 
 
56
49
# TODO: Rather than building a changeset object, we should probably
57
50
# invoke callbacks on an object.  That object can either accumulate a
58
51
# list, write them out directly, etc etc.
93
86
    if sequence_matcher is None:
94
87
        sequence_matcher = patiencediff.PatienceSequenceMatcher
95
88
    ud = patiencediff.unified_diff(oldlines, newlines,
96
 
                      fromfile=old_filename.encode(path_encoding, 'replace'),
97
 
                      tofile=new_filename.encode(path_encoding, 'replace'),
 
89
                      fromfile=old_filename.encode(path_encoding),
 
90
                      tofile=new_filename.encode(path_encoding),
98
91
                      sequencematcher=sequence_matcher)
99
92
 
100
93
    ud = list(ud)
284
277
                        new_abspath, e)
285
278
 
286
279
 
287
 
def get_trees_and_branches_to_diff_locked(
288
 
    path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
 
280
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
 
281
                                   apply_view=True):
289
282
    """Get the trees and specific files to diff given a list of paths.
290
283
 
291
284
    This method works out the trees to be diff'ed and the files of
302
295
    :param new_url:
303
296
        The url of the new branch or tree. If None, the tree to use is
304
297
        taken from the first path, if any, or the current working tree.
305
 
    :param add_cleanup:
306
 
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
307
 
        will register cleanups that must be run to unlock the trees, etc.
308
298
    :param apply_view:
309
299
        if True and a view is set, apply the view or check that the paths
310
300
        are within it
311
301
    :returns:
312
302
        a tuple of (old_tree, new_tree, old_branch, new_branch,
313
303
        specific_files, extra_trees) where extra_trees is a sequence of
314
 
        additional trees to search in for file-ids.  The trees and branches
315
 
        will be read-locked until the cleanups registered via the add_cleanup
316
 
        param are run.
 
304
        additional trees to search in for file-ids.
317
305
    """
318
306
    # Get the old and new revision specs
319
307
    old_revision_spec = None
342
330
        default_location = path_list[0]
343
331
        other_paths = path_list[1:]
344
332
 
345
 
    def lock_tree_or_branch(wt, br):
346
 
        if wt is not None:
347
 
            wt.lock_read()
348
 
            add_cleanup(wt.unlock)
349
 
        elif br is not None:
350
 
            br.lock_read()
351
 
            add_cleanup(br.unlock)
352
 
 
353
333
    # Get the old location
354
334
    specific_files = []
355
335
    if old_url is None:
356
336
        old_url = default_location
357
337
    working_tree, branch, relpath = \
358
338
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
359
 
    lock_tree_or_branch(working_tree, branch)
360
339
    if consider_relpath and relpath != '':
361
340
        if working_tree is not None and apply_view:
362
341
            views.check_path_in_view(working_tree, relpath)
370
349
    if new_url != old_url:
371
350
        working_tree, branch, relpath = \
372
351
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
373
 
        lock_tree_or_branch(working_tree, branch)
374
352
        if consider_relpath and relpath != '':
375
353
            if working_tree is not None and apply_view:
376
354
                views.check_path_in_view(working_tree, relpath)
381
359
 
382
360
    # Get the specific files (all files is None, no files is [])
383
361
    if make_paths_wt_relative and working_tree is not None:
384
 
        other_paths = working_tree.safe_relpath_files(
385
 
            other_paths,
 
362
        try:
 
363
            from bzrlib.builtins import safe_relpath_files
 
364
            other_paths = safe_relpath_files(working_tree, other_paths,
386
365
            apply_view=apply_view)
 
366
        except errors.FileInWrongBranch:
 
367
            raise errors.BzrCommandError("Files are in different branches")
387
368
    specific_files.extend(other_paths)
388
369
    if len(specific_files) == 0:
389
370
        specific_files = None
399
380
    extra_trees = None
400
381
    if working_tree is not None and working_tree not in (old_tree, new_tree):
401
382
        extra_trees = (working_tree,)
402
 
    return (old_tree, new_tree, old_branch, new_branch,
403
 
            specific_files, extra_trees)
 
383
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
404
384
 
405
385
 
406
386
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
422
402
                    old_label='a/', new_label='b/',
423
403
                    extra_trees=None,
424
404
                    path_encoding='utf8',
425
 
                    using=None,
426
 
                    format_cls=None):
 
405
                    using=None):
427
406
    """Show in text form the changes from one tree to another.
428
407
 
429
 
    :param to_file: The output stream.
430
 
    :param specific_files: Include only changes to these files - None for all
431
 
        changes.
432
 
    :param external_diff_options: If set, use an external GNU diff and pass 
433
 
        these options.
434
 
    :param extra_trees: If set, more Trees to use for looking up file ids
435
 
    :param path_encoding: If set, the path will be encoded as specified, 
436
 
        otherwise is supposed to be utf8
437
 
    :param format_cls: Formatter class (DiffTree subclass)
 
408
    to_file
 
409
        The output stream.
 
410
 
 
411
    specific_files
 
412
        Include only changes to these files - None for all changes.
 
413
 
 
414
    external_diff_options
 
415
        If set, use an external GNU diff and pass these options.
 
416
 
 
417
    extra_trees
 
418
        If set, more Trees to use for looking up file ids
 
419
 
 
420
    path_encoding
 
421
        If set, the path will be encoded as specified, otherwise is supposed
 
422
        to be utf8
438
423
    """
439
 
    if format_cls is None:
440
 
        format_cls = DiffTree
441
424
    old_tree.lock_read()
442
425
    try:
443
426
        if extra_trees is not None:
445
428
                tree.lock_read()
446
429
        new_tree.lock_read()
447
430
        try:
448
 
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
449
 
                                                   path_encoding,
450
 
                                                   external_diff_options,
451
 
                                                   old_label, new_label, using)
 
431
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
432
                                                 path_encoding,
 
433
                                                 external_diff_options,
 
434
                                                 old_label, new_label, using)
452
435
            return differ.show_diff(specific_files, extra_trees)
453
436
        finally:
454
437
            new_tree.unlock()
461
444
 
462
445
def _patch_header_date(tree, file_id, path):
463
446
    """Returns a timestamp suitable for use in a patch header."""
464
 
    try:
465
 
        mtime = tree.get_file_mtime(file_id, path)
466
 
    except errors.FileTimestampUnavailable:
467
 
        mtime = 0
 
447
    mtime = tree.get_file_mtime(file_id, path)
468
448
    return timestamp.format_patch_date(mtime)
469
449
 
470
450
 
665
645
        """
666
646
        def _get_text(tree, file_id, path):
667
647
            if file_id is not None:
668
 
                return tree.get_file_lines(file_id, path)
 
648
                return tree.get_file(file_id, path).readlines()
669
649
            else:
670
650
                return []
671
651
        try:
672
652
            from_text = _get_text(self.old_tree, from_file_id, from_path)
673
653
            to_text = _get_text(self.new_tree, to_file_id, to_path)
674
654
            self.text_differ(from_label, from_text, to_label, to_text,
675
 
                             self.to_file, path_encoding=self.path_encoding)
 
655
                             self.to_file)
676
656
        except errors.BinaryFile:
677
657
            self.to_file.write(
678
658
                  ("Binary files %s and %s differ\n" %
679
 
                  (from_label, to_label)).encode(self.path_encoding,'replace'))
 
659
                  (from_label, to_label)).encode(self.path_encoding))
680
660
        return self.CHANGED
681
661
 
682
662
 
691
671
    @classmethod
692
672
    def from_string(klass, command_string, old_tree, new_tree, to_file,
693
673
                    path_encoding='utf-8'):
694
 
        command_template = cmdline.split(command_string)
695
 
        if '@' not in command_string:
696
 
            command_template.extend(['@old_path', '@new_path'])
 
674
        command_template = commands.shlex_split_unicode(command_string)
 
675
        command_template.extend(['%(old_path)s', '%(new_path)s'])
697
676
        return klass(command_template, old_tree, new_tree, to_file,
698
677
                     path_encoding)
699
678
 
700
679
    @classmethod
701
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
680
    def make_from_diff_tree(klass, command_string):
702
681
        def from_diff_tree(diff_tree):
703
 
            full_command_string = [command_string]
704
 
            if external_diff_options is not None:
705
 
                full_command_string += ' ' + external_diff_options
706
 
            return klass.from_string(full_command_string, diff_tree.old_tree,
 
682
            return klass.from_string(command_string, diff_tree.old_tree,
707
683
                                     diff_tree.new_tree, diff_tree.to_file)
708
684
        return from_diff_tree
709
685
 
710
686
    def _get_command(self, old_path, new_path):
711
687
        my_map = {'old_path': old_path, 'new_path': new_path}
712
 
        command = [AtTemplate(t).substitute(my_map) for t in
713
 
                   self.command_template]
714
 
        if sys.platform == 'win32': # Popen doesn't accept unicode on win32
715
 
            command_encoded = []
716
 
            for c in command:
717
 
                if isinstance(c, unicode):
718
 
                    command_encoded.append(c.encode('mbcs'))
719
 
                else:
720
 
                    command_encoded.append(c)
721
 
            return command_encoded
722
 
        else:
723
 
            return command
 
688
        return [t % my_map for t in self.command_template]
724
689
 
725
690
    def _execute(self, old_path, new_path):
726
691
        command = self._get_command(old_path, new_path)
746
711
                raise
747
712
        return True
748
713
 
749
 
    @staticmethod
750
 
    def _fenc():
751
 
        """Returns safe encoding for passing file path to diff tool"""
752
 
        if sys.platform == 'win32':
753
 
            return 'mbcs'
754
 
        else:
755
 
            # Don't fallback to 'utf-8' because subprocess may not be able to
756
 
            # handle utf-8 correctly when locale is not utf-8.
757
 
            return sys.getfilesystemencoding() or 'ascii'
758
 
 
759
 
    def _is_safepath(self, path):
760
 
        """Return true if `path` may be able to pass to subprocess."""
761
 
        fenc = self._fenc()
762
 
        try:
763
 
            return path == path.encode(fenc).decode(fenc)
764
 
        except UnicodeError:
765
 
            return False
766
 
 
767
 
    def _safe_filename(self, prefix, relpath):
768
 
        """Replace unsafe character in `relpath` then join `self._root`,
769
 
        `prefix` and `relpath`."""
770
 
        fenc = self._fenc()
771
 
        # encoded_str.replace('?', '_') may break multibyte char.
772
 
        # So we should encode, decode, then replace(u'?', u'_')
773
 
        relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
774
 
        relpath_tmp = relpath_tmp.replace(u'?', u'_')
775
 
        return osutils.pathjoin(self._root, prefix, relpath_tmp)
776
 
 
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
 
            full_path = tree.abspath(tree.id2path(file_id))
781
 
            if self._is_safepath(full_path):
782
 
                return full_path
783
 
 
784
 
        full_path = self._safe_filename(prefix, relpath)
785
 
        if not force_temp and self._try_symlink_root(tree, prefix):
 
714
    def _write_file(self, file_id, tree, prefix, relpath):
 
715
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
716
        if self._try_symlink_root(tree, prefix):
786
717
            return full_path
787
718
        parent_dir = osutils.dirname(full_path)
788
719
        try:
799
730
                target.close()
800
731
        finally:
801
732
            source.close()
802
 
        try:
803
 
            mtime = tree.get_file_mtime(file_id)
804
 
        except errors.FileTimestampUnavailable:
805
 
            pass
806
 
        else:
807
 
            os.utime(full_path, (mtime, mtime))
808
 
        if not allow_write:
809
 
            osutils.make_readonly(full_path)
 
733
        osutils.make_readonly(full_path)
 
734
        mtime = tree.get_file_mtime(file_id)
 
735
        os.utime(full_path, (mtime, mtime))
810
736
        return full_path
811
737
 
812
 
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
813
 
                       allow_write_new=False):
 
738
    def _prepare_files(self, file_id, old_path, new_path):
814
739
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
815
 
                                         old_path, force_temp)
 
740
                                         old_path)
816
741
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
817
 
                                         new_path, force_temp,
818
 
                                         allow_write=allow_write_new)
 
742
                                         new_path)
819
743
        return old_disk_path, new_disk_path
820
744
 
821
745
    def finish(self):
829
753
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
830
754
        if (old_kind, new_kind) != ('file', 'file'):
831
755
            return DiffPath.CANNOT_DIFF
832
 
        (old_disk_path, new_disk_path) = self._prepare_files(
833
 
                                                file_id, old_path, new_path)
834
 
        self._execute(old_disk_path, new_disk_path)
835
 
 
836
 
    def edit_file(self, file_id):
837
 
        """Use this tool to edit a file.
838
 
 
839
 
        A temporary copy will be edited, and the new contents will be
840
 
        returned.
841
 
 
842
 
        :param file_id: The id of the file to edit.
843
 
        :return: The new contents of the file.
844
 
        """
845
 
        old_path = self.old_tree.id2path(file_id)
846
 
        new_path = self.new_tree.id2path(file_id)
847
 
        old_abs_path, new_abs_path = self._prepare_files(
848
 
                                            file_id, old_path, new_path,
849
 
                                            allow_write_new=True,
850
 
                                            force_temp=True)
851
 
        command = self._get_command(old_abs_path, new_abs_path)
852
 
        subprocess.call(command, cwd=self._root)
853
 
        new_file = open(new_abs_path, 'rb')
854
 
        try:
855
 
            return new_file.read()
856
 
        finally:
857
 
            new_file.close()
 
756
        self._prepare_files(file_id, old_path, new_path)
 
757
        self._execute(osutils.pathjoin('old', old_path),
 
758
                      osutils.pathjoin('new', new_path))
858
759
 
859
760
 
860
761
class DiffTree(object):
906
807
        """Factory for producing a DiffTree.
907
808
 
908
809
        Designed to accept options used by show_diff_trees.
909
 
 
910
810
        :param old_tree: The tree to show as old in the comparison
911
811
        :param new_tree: The tree to show as new in the comparison
912
812
        :param to_file: File to write comparisons to
918
818
        :param using: Commandline to use to invoke an external diff tool
919
819
        """
920
820
        if using is not None:
921
 
            extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
 
821
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
922
822
        else:
923
823
            extra_factories = []
924
824
        if external_diff_options:
925
825
            opts = external_diff_options.split()
926
 
            def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None):
927
 
                """:param path_encoding: not used but required
928
 
                        to match the signature of internal_diff.
929
 
                """
 
826
            def diff_file(olab, olines, nlab, nlines, to_file):
930
827
                external_diff(olab, olines, nlab, nlines, to_file, opts)
931
828
        else:
932
829
            diff_file = internal_diff
938
835
    def show_diff(self, specific_files, extra_trees=None):
939
836
        """Write tree diff to self.to_file
940
837
 
941
 
        :param specific_files: the specific files to compare (recursive)
 
838
        :param sepecific_files: the specific files to compare (recursive)
942
839
        :param extra_trees: extra trees to use for mapping paths to file_ids
943
840
        """
944
841
        try:
1034
931
            if error_path is None:
1035
932
                error_path = old_path
1036
933
            raise errors.NoDiffFound(error_path)
1037
 
 
1038
 
 
1039
 
format_registry = Registry()
1040
 
format_registry.register('default', DiffTree)