← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Jelmer Vernooij
  • Date: 2009-02-10 04:10:44 UTC
  • mto: This revision was merged to the branch mainline in revision 3995.
  • Revision ID: jelmer@samba.org-20090210041044-42lmb09hskt9lt9l
Review from Ian.

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
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
 
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
16
 
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,
38
37
    textfile,
39
38
    timestamp,
40
 
    views,
41
39
    )
42
 
 
43
 
from bzrlib.workingtree import WorkingTree
44
40
""")
45
41
 
46
 
from bzrlib.registry import (
47
 
    Registry,
48
 
    )
49
42
from bzrlib.symbol_versioning import (
50
 
    deprecated_function,
51
 
    deprecated_in,
52
 
    )
53
 
from bzrlib.trace import mutter, note, warning
54
 
 
55
 
 
56
 
class AtTemplate(string.Template):
57
 
    """Templating class that uses @ instead of $."""
58
 
 
59
 
    delimiter = '@'
 
43
        deprecated_function,
 
44
        one_three
 
45
        )
 
46
from bzrlib.trace import warning
60
47
 
61
48
 
62
49
# TODO: Rather than building a changeset object, we should probably
91
78
    # both sequences are empty.
92
79
    if not oldlines and not newlines:
93
80
        return
94
 
 
 
81
    
95
82
    if allow_binary is False:
96
83
        textfile.check_text_lines(oldlines)
97
84
        textfile.check_text_lines(newlines)
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,
217
199
            break
218
200
        else:
219
201
            diffcmd.append('-u')
220
 
 
 
202
                  
221
203
        if diff_opts:
222
204
            diffcmd.extend(diff_opts)
223
205
 
224
206
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
225
207
        out,err = pipe.communicate()
226
208
        rc = pipe.returncode
227
 
 
 
209
        
228
210
        # internal_diff() adds a trailing newline, add one here for consistency
229
211
        out += '\n'
230
212
        if rc == 2:
265
247
                msg = 'signal %d' % (-rc)
266
248
            else:
267
249
                msg = 'exit code %d' % rc
268
 
 
269
 
            raise errors.BzrError('external diff failed with %s; command: %r'
 
250
                
 
251
            raise errors.BzrError('external diff failed with %s; command: %r' 
270
252
                                  % (rc, diffcmd))
271
253
 
272
254
 
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
    """Get the trees and specific files to diff given a list of paths.
 
277
 
 
278
    This method works out the trees to be diff'ed and the files of
 
279
    interest within those trees.
 
280
 
 
281
    :param path_list:
 
282
        the list of arguments passed to the diff command
 
283
    :param revision_specs:
 
284
        Zero, one or two RevisionSpecs from the diff command line,
 
285
        saying what revisions to compare.
 
286
    :param old_url:
 
287
        The url of the old branch or tree. If None, the tree to use is
 
288
        taken from the first path, if any, or the current working tree.
 
289
    :param new_url:
 
290
        The url of the new branch or tree. If None, the tree to use is
 
291
        taken from the first path, if any, or the current working tree.
 
292
    :returns:
 
293
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
294
        extra_trees is a sequence of additional trees to search in for
 
295
        file-ids.
356
296
    """
357
297
    # Get the old and new revision specs
358
298
    old_revision_spec = None
381
321
        default_location = path_list[0]
382
322
        other_paths = path_list[1:]
383
323
 
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
324
    # Get the old location
393
325
    specific_files = []
394
326
    if old_url is None:
395
327
        old_url = default_location
396
328
    working_tree, branch, relpath = \
397
329
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
398
 
    lock_tree_or_branch(working_tree, branch)
399
330
    if consider_relpath and relpath != '':
400
 
        if working_tree is not None and apply_view:
401
 
            views.check_path_in_view(working_tree, relpath)
402
331
        specific_files.append(relpath)
403
332
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
404
 
    old_branch = branch
405
333
 
406
334
    # Get the new location
407
335
    if new_url is None:
409
337
    if new_url != old_url:
410
338
        working_tree, branch, relpath = \
411
339
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
412
 
        lock_tree_or_branch(working_tree, branch)
413
340
        if consider_relpath and relpath != '':
414
 
            if working_tree is not None and apply_view:
415
 
                views.check_path_in_view(working_tree, relpath)
416
341
            specific_files.append(relpath)
417
342
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
418
343
        basis_is_default=working_tree is None)
419
 
    new_branch = branch
420
344
 
421
345
    # Get the specific files (all files is None, no files is [])
422
346
    if make_paths_wt_relative and working_tree is not None:
423
 
        try:
424
 
            from bzrlib.builtins import safe_relpath_files
425
 
            other_paths = safe_relpath_files(working_tree, other_paths,
426
 
            apply_view=apply_view)
427
 
        except errors.FileInWrongBranch:
428
 
            raise errors.BzrCommandError("Files are in different branches")
 
347
        other_paths = _relative_paths_in_tree(working_tree, other_paths)
429
348
    specific_files.extend(other_paths)
430
349
    if len(specific_files) == 0:
431
350
        specific_files = None
432
 
        if (working_tree is not None and working_tree.supports_views()
433
 
            and apply_view):
434
 
            view_files = working_tree.views.lookup_view()
435
 
            if view_files:
436
 
                specific_files = view_files
437
 
                view_str = views.view_display_str(view_files)
438
 
                note("*** Ignoring files outside view. View is %s" % view_str)
439
351
 
440
352
    # Get extra trees that ought to be searched for file-ids
441
353
    extra_trees = None
442
354
    if working_tree is not None and working_tree not in (old_tree, new_tree):
443
355
        extra_trees = (working_tree,)
444
 
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
 
356
    return old_tree, new_tree, specific_files, extra_trees
445
357
 
446
358
 
447
359
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
458
370
    return spec.as_tree(branch)
459
371
 
460
372
 
 
373
def _relative_paths_in_tree(tree, paths):
 
374
    """Get the relative paths within a working tree.
 
375
 
 
376
    Each path may be either an absolute path or a path relative to the
 
377
    current working directory.
 
378
    """
 
379
    result = []
 
380
    for filename in paths:
 
381
        try:
 
382
            result.append(tree.relpath(osutils.dereference_path(filename)))
 
383
        except errors.PathNotChild:
 
384
            raise errors.BzrCommandError("Files are in different branches")
 
385
    return result
 
386
 
 
387
 
461
388
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
462
389
                    external_diff_options=None,
463
390
                    old_label='a/', new_label='b/',
464
391
                    extra_trees=None,
465
392
                    path_encoding='utf8',
466
 
                    using=None,
467
 
                    format_cls=None):
 
393
                    using=None):
468
394
    """Show in text form the changes from one tree to another.
469
395
 
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)
 
396
    to_file
 
397
        The output stream.
 
398
 
 
399
    specific_files
 
400
        Include only changes to these files - None for all changes.
 
401
 
 
402
    external_diff_options
 
403
        If set, use an external GNU diff and pass these options.
 
404
 
 
405
    extra_trees
 
406
        If set, more Trees to use for looking up file ids
 
407
 
 
408
    path_encoding
 
409
        If set, the path will be encoded as specified, otherwise is supposed
 
410
        to be utf8
479
411
    """
480
 
    if format_cls is None:
481
 
        format_cls = DiffTree
482
412
    old_tree.lock_read()
483
413
    try:
484
414
        if extra_trees is not None:
486
416
                tree.lock_read()
487
417
        new_tree.lock_read()
488
418
        try:
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)
 
419
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
420
                                                 path_encoding,
 
421
                                                 external_diff_options,
 
422
                                                 old_label, new_label, using)
493
423
            return differ.show_diff(specific_files, extra_trees)
494
424
        finally:
495
425
            new_tree.unlock()
502
432
 
503
433
def _patch_header_date(tree, file_id, path):
504
434
    """Returns a timestamp suitable for use in a patch header."""
505
 
    try:
506
 
        mtime = tree.get_file_mtime(file_id, path)
507
 
    except errors.FileTimestampUnavailable:
508
 
        mtime = 0
 
435
    mtime = tree.get_file_mtime(file_id, path)
509
436
    return timestamp.format_patch_date(mtime)
510
437
 
511
438
 
 
439
@deprecated_function(one_three)
 
440
def get_prop_change(meta_modified):
 
441
    if meta_modified:
 
442
        return " (properties changed)"
 
443
    else:
 
444
        return  ""
 
445
 
512
446
def get_executable_change(old_is_x, new_is_x):
513
447
    descr = { True:"+x", False:"-x", None:"??" }
514
448
    if old_is_x != new_is_x:
689
623
            return self.CANNOT_DIFF
690
624
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
691
625
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
692
 
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
693
 
            old_path, new_path)
 
626
        return self.diff_text(from_file_id, to_file_id, from_label, to_label)
694
627
 
695
 
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
696
 
        from_path=None, to_path=None):
 
628
    def diff_text(self, from_file_id, to_file_id, from_label, to_label):
697
629
        """Diff the content of given files in two trees
698
630
 
699
631
        :param from_file_id: The id of the file in the from tree.  If None,
701
633
        :param to_file_id: The id of the file in the to tree.  This may refer
702
634
            to a different file from from_file_id.  If None,
703
635
            the file is not present in the to tree.
704
 
        :param from_path: The path in the from tree or None if unknown.
705
 
        :param to_path: The path in the to tree or None if unknown.
706
636
        """
707
 
        def _get_text(tree, file_id, path):
 
637
        def _get_text(tree, file_id):
708
638
            if file_id is not None:
709
 
                return tree.get_file_lines(file_id, path)
 
639
                return tree.get_file(file_id).readlines()
710
640
            else:
711
641
                return []
712
642
        try:
713
 
            from_text = _get_text(self.old_tree, from_file_id, from_path)
714
 
            to_text = _get_text(self.new_tree, to_file_id, to_path)
 
643
            from_text = _get_text(self.old_tree, from_file_id)
 
644
            to_text = _get_text(self.new_tree, to_file_id)
715
645
            self.text_differ(from_label, from_text, to_label, to_text,
716
 
                             self.to_file, path_encoding=self.path_encoding)
 
646
                             self.to_file)
717
647
        except errors.BinaryFile:
718
648
            self.to_file.write(
719
649
                  ("Binary files %s and %s differ\n" %
720
 
                  (from_label, to_label)).encode(self.path_encoding,'replace'))
 
650
                  (from_label, to_label)).encode(self.path_encoding))
721
651
        return self.CHANGED
722
652
 
723
653
 
732
662
    @classmethod
733
663
    def from_string(klass, command_string, old_tree, new_tree, to_file,
734
664
                    path_encoding='utf-8'):
735
 
        command_template = cmdline.split(command_string)
736
 
        if '@' not in command_string:
737
 
            command_template.extend(['@old_path', '@new_path'])
 
665
        command_template = commands.shlex_split_unicode(command_string)
 
666
        command_template.extend(['%(old_path)s', '%(new_path)s'])
738
667
        return klass(command_template, old_tree, new_tree, to_file,
739
668
                     path_encoding)
740
669
 
747
676
 
748
677
    def _get_command(self, old_path, new_path):
749
678
        my_map = {'old_path': old_path, 'new_path': new_path}
750
 
        return [AtTemplate(t).substitute(my_map) for t in
751
 
                self.command_template]
 
679
        return [t % my_map for t in self.command_template]
752
680
 
753
681
    def _execute(self, old_path, new_path):
754
682
        command = self._get_command(old_path, new_path)
774
702
                raise
775
703
        return True
776
704
 
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
 
        
 
705
    def _write_file(self, file_id, tree, prefix, relpath):
782
706
        full_path = osutils.pathjoin(self._root, prefix, relpath)
783
 
        if not force_temp and self._try_symlink_root(tree, prefix):
 
707
        if self._try_symlink_root(tree, prefix):
784
708
            return full_path
785
709
        parent_dir = osutils.dirname(full_path)
786
710
        try:
797
721
                target.close()
798
722
        finally:
799
723
            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)
 
724
        osutils.make_readonly(full_path)
 
725
        mtime = tree.get_file_mtime(file_id)
 
726
        os.utime(full_path, (mtime, mtime))
808
727
        return full_path
809
728
 
810
 
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
811
 
                       allow_write_new=False):
 
729
    def _prepare_files(self, file_id, old_path, new_path):
812
730
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
813
 
                                         old_path, force_temp)
 
731
                                         old_path)
814
732
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
815
 
                                         new_path, force_temp,
816
 
                                         allow_write=allow_write_new)
 
733
                                         new_path)
817
734
        return old_disk_path, new_disk_path
818
735
 
819
736
    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))
 
737
        osutils.rmtree(self._root)
826
738
 
827
739
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
828
740
        if (old_kind, new_kind) != ('file', 'file'):
829
741
            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()
 
742
        self._prepare_files(file_id, old_path, new_path)
 
743
        self._execute(osutils.pathjoin('old', old_path),
 
744
                      osutils.pathjoin('new', new_path))
856
745
 
857
746
 
858
747
class DiffTree(object):
920
809
            extra_factories = []
921
810
        if external_diff_options:
922
811
            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
 
                """
 
812
            def diff_file(olab, olines, nlab, nlines, to_file):
927
813
                external_diff(olab, olines, nlab, nlines, to_file, opts)
928
814
        else:
929
815
            diff_file = internal_diff
935
821
    def show_diff(self, specific_files, extra_trees=None):
936
822
        """Write tree diff to self.to_file
937
823
 
938
 
        :param specific_files: the specific files to compare (recursive)
 
824
        :param sepecific_files: the specific files to compare (recursive)
939
825
        :param extra_trees: extra trees to use for mapping paths to file_ids
940
826
        """
941
827
        try:
999
885
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
1000
886
                                   newpath_encoded, prop_str))
1001
887
            if changed_content:
1002
 
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
 
888
                self.diff(file_id, oldpath, newpath)
1003
889
                has_changes = 1
1004
890
            if renamed:
1005
891
                has_changes = 1
1020
906
            new_kind = self.new_tree.kind(file_id)
1021
907
        except (errors.NoSuchId, errors.NoSuchFile):
1022
908
            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):
 
909
 
1027
910
        result = DiffPath._diff_many(self.differs, file_id, old_path,
1028
911
                                       new_path, old_kind, new_kind)
1029
912
        if result is DiffPath.CANNOT_DIFF:
1031
914
            if error_path is None:
1032
915
                error_path = old_path
1033
916
            raise errors.NoDiffFound(error_path)
1034
 
 
1035
 
 
1036
 
format_registry = Registry()
1037
 
format_registry.register('default', DiffTree)