1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
1
# Copyright (C) 2005-2011 Canonical Ltd.
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
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
from __future__ import absolute_import
23
25
from bzrlib.lazy_import import lazy_import
43
from bzrlib.workingtree import WorkingTree
44
from bzrlib.i18n import gettext
43
from bzrlib.symbol_versioning import (
47
from bzrlib.registry import (
46
50
from bzrlib.trace import mutter, note, warning
52
DEFAULT_CONTEXT_AMOUNT = 3
54
class AtTemplate(string.Template):
55
"""Templating class that uses @ instead of $."""
49
60
# TODO: Rather than building a changeset object, we should probably
50
61
# invoke callbacks on an object. That object can either accumulate a
63
74
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
64
75
allow_binary=False, sequence_matcher=None,
65
path_encoding='utf8'):
76
path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
66
77
# FIXME: difflib is wrong if there is no trailing newline.
67
78
# The syntax used by patch seems to be "\ No newline at
68
79
# end of file" following the last diff line from that
86
97
if sequence_matcher is None:
87
98
sequence_matcher = patiencediff.PatienceSequenceMatcher
88
99
ud = patiencediff.unified_diff(oldlines, newlines,
89
fromfile=old_filename.encode(path_encoding),
90
tofile=new_filename.encode(path_encoding),
91
sequencematcher=sequence_matcher)
100
fromfile=old_filename.encode(path_encoding, 'replace'),
101
tofile=new_filename.encode(path_encoding, 'replace'),
102
n=context_lines, sequencematcher=sequence_matcher)
94
105
if len(ud) == 0: # Identical contents, nothing to do
172
183
if not diff_opts:
185
if sys.platform == 'win32':
186
# Popen doesn't do the proper encoding for external commands
187
# Since we are dealing with an ANSI api, use mbcs encoding
188
old_filename = old_filename.encode('mbcs')
189
new_filename = new_filename.encode('mbcs')
174
190
diffcmd = ['diff',
175
191
'--label', old_filename,
275
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url,
291
def get_trees_and_branches_to_diff_locked(
292
path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
277
293
"""Get the trees and specific files to diff given a list of paths.
279
295
This method works out the trees to be diff'ed and the files of
291
307
The url of the new branch or tree. If None, the tree to use is
292
308
taken from the first path, if any, or the current working tree.
310
a callable like Command.add_cleanup. get_trees_and_branches_to_diff
311
will register cleanups that must be run to unlock the trees, etc.
293
312
:param apply_view:
294
313
if True and a view is set, apply the view or check that the paths
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
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
will be read-locked until the cleanups registered via the add_cleanup
301
322
# Get the old and new revision specs
302
323
old_revision_spec = None
325
346
default_location = path_list[0]
326
347
other_paths = path_list[1:]
349
def lock_tree_or_branch(wt, br):
352
add_cleanup(wt.unlock)
355
add_cleanup(br.unlock)
328
357
# Get the old location
329
358
specific_files = []
330
359
if old_url is None:
331
360
old_url = default_location
332
361
working_tree, branch, relpath = \
333
bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
362
controldir.ControlDir.open_containing_tree_or_branch(old_url)
363
lock_tree_or_branch(working_tree, branch)
334
364
if consider_relpath and relpath != '':
335
365
if working_tree is not None and apply_view:
336
366
views.check_path_in_view(working_tree, relpath)
337
367
specific_files.append(relpath)
338
368
old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
340
371
# Get the new location
341
372
if new_url is None:
342
373
new_url = default_location
343
374
if new_url != old_url:
344
375
working_tree, branch, relpath = \
345
bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
376
controldir.ControlDir.open_containing_tree_or_branch(new_url)
377
lock_tree_or_branch(working_tree, branch)
346
378
if consider_relpath and relpath != '':
347
379
if working_tree is not None and apply_view:
348
380
views.check_path_in_view(working_tree, relpath)
349
381
specific_files.append(relpath)
350
382
new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
351
383
basis_is_default=working_tree is None)
353
386
# Get the specific files (all files is None, no files is [])
354
387
if make_paths_wt_relative and working_tree is not None:
356
from bzrlib.builtins import safe_relpath_files
357
other_paths = safe_relpath_files(working_tree, other_paths,
388
other_paths = working_tree.safe_relpath_files(
358
390
apply_view=apply_view)
359
except errors.FileInWrongBranch:
360
raise errors.BzrCommandError("Files are in different branches")
361
391
specific_files.extend(other_paths)
362
392
if len(specific_files) == 0:
363
393
specific_files = None
368
398
specific_files = view_files
369
399
view_str = views.view_display_str(view_files)
370
note("*** Ignoring files outside view. View is %s" % view_str)
400
note(gettext("*** Ignoring files outside view. View is %s") % view_str)
372
402
# Get extra trees that ought to be searched for file-ids
373
403
extra_trees = None
374
404
if working_tree is not None and working_tree not in (old_tree, new_tree):
375
405
extra_trees = (working_tree,)
376
return old_tree, new_tree, specific_files, extra_trees
406
return (old_tree, new_tree, old_branch, new_branch,
407
specific_files, extra_trees)
378
410
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
379
411
if branch is None and tree is not None:
394
426
old_label='a/', new_label='b/',
395
427
extra_trees=None,
396
428
path_encoding='utf8',
431
context=DEFAULT_CONTEXT_AMOUNT):
398
432
"""Show in text form the changes from one tree to another.
404
Include only changes to these files - None for all changes.
406
external_diff_options
407
If set, use an external GNU diff and pass these options.
410
If set, more Trees to use for looking up file ids
413
If set, the path will be encoded as specified, otherwise is supposed
434
:param to_file: The output stream.
435
:param specific_files: Include only changes to these files - None for all
437
:param external_diff_options: If set, use an external GNU diff and pass
439
:param extra_trees: If set, more Trees to use for looking up file ids
440
:param path_encoding: If set, the path will be encoded as specified,
441
otherwise is supposed to be utf8
442
:param format_cls: Formatter class (DiffTree subclass)
445
context = DEFAULT_CONTEXT_AMOUNT
446
if format_cls is None:
447
format_cls = DiffTree
416
448
old_tree.lock_read()
418
450
if extra_trees is not None:
421
453
new_tree.lock_read()
423
differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
425
external_diff_options,
426
old_label, new_label, using)
455
differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
457
external_diff_options,
458
old_label, new_label, using,
459
context_lines=context)
427
460
return differ.show_diff(specific_files, extra_trees)
429
462
new_tree.unlock()
437
470
def _patch_header_date(tree, file_id, path):
438
471
"""Returns a timestamp suitable for use in a patch header."""
439
mtime = tree.get_file_mtime(file_id, path)
473
mtime = tree.get_file_mtime(file_id, path)
474
except errors.FileTimestampUnavailable:
440
476
return timestamp.format_patch_date(mtime)
584
620
# or removed in a diff.
585
621
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
587
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
588
old_label='', new_label='', text_differ=internal_diff):
623
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
624
old_label='', new_label='', text_differ=internal_diff,
625
context_lines=DEFAULT_CONTEXT_AMOUNT):
589
626
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
590
627
self.text_differ = text_differ
591
628
self.old_label = old_label
592
629
self.new_label = new_label
593
630
self.path_encoding = path_encoding
631
self.context_lines = context_lines
595
633
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
596
634
"""Compare two files in unified diff format
638
676
def _get_text(tree, file_id, path):
639
677
if file_id is not None:
640
return tree.get_file(file_id, path).readlines()
678
return tree.get_file_lines(file_id, path)
644
682
from_text = _get_text(self.old_tree, from_file_id, from_path)
645
683
to_text = _get_text(self.new_tree, to_file_id, to_path)
646
684
self.text_differ(from_label, from_text, to_label, to_text,
685
self.to_file, path_encoding=self.path_encoding,
686
context_lines=self.context_lines)
648
687
except errors.BinaryFile:
649
688
self.to_file.write(
650
689
("Binary files %s and %s differ\n" %
651
(from_label, to_label)).encode(self.path_encoding))
690
(from_label, to_label)).encode(self.path_encoding,'replace'))
652
691
return self.CHANGED
664
703
def from_string(klass, command_string, old_tree, new_tree, to_file,
665
704
path_encoding='utf-8'):
666
command_template = commands.shlex_split_unicode(command_string)
667
command_template.extend(['%(old_path)s', '%(new_path)s'])
705
command_template = cmdline.split(command_string)
706
if '@' not in command_string:
707
command_template.extend(['@old_path', '@new_path'])
668
708
return klass(command_template, old_tree, new_tree, to_file,
672
def make_from_diff_tree(klass, command_string):
712
def make_from_diff_tree(klass, command_string, external_diff_options=None):
673
713
def from_diff_tree(diff_tree):
674
return klass.from_string(command_string, diff_tree.old_tree,
714
full_command_string = [command_string]
715
if external_diff_options is not None:
716
full_command_string += ' ' + external_diff_options
717
return klass.from_string(full_command_string, diff_tree.old_tree,
675
718
diff_tree.new_tree, diff_tree.to_file)
676
719
return from_diff_tree
678
721
def _get_command(self, old_path, new_path):
679
722
my_map = {'old_path': old_path, 'new_path': new_path}
680
return [t % my_map for t in self.command_template]
723
command = [AtTemplate(t).substitute(my_map) for t in
724
self.command_template]
725
if sys.platform == 'win32': # Popen doesn't accept unicode on win32
728
if isinstance(c, unicode):
729
command_encoded.append(c.encode('mbcs'))
731
command_encoded.append(c)
732
return command_encoded
682
736
def _execute(self, old_path, new_path):
683
737
command = self._get_command(old_path, new_path)
706
def _write_file(self, file_id, tree, prefix, relpath):
707
full_path = osutils.pathjoin(self._root, prefix, relpath)
708
if self._try_symlink_root(tree, prefix):
762
"""Returns safe encoding for passing file path to diff tool"""
763
if sys.platform == 'win32':
766
# Don't fallback to 'utf-8' because subprocess may not be able to
767
# handle utf-8 correctly when locale is not utf-8.
768
return sys.getfilesystemencoding() or 'ascii'
770
def _is_safepath(self, path):
771
"""Return true if `path` may be able to pass to subprocess."""
774
return path == path.encode(fenc).decode(fenc)
778
def _safe_filename(self, prefix, relpath):
779
"""Replace unsafe character in `relpath` then join `self._root`,
780
`prefix` and `relpath`."""
782
# encoded_str.replace('?', '_') may break multibyte char.
783
# So we should encode, decode, then replace(u'?', u'_')
784
relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
785
relpath_tmp = relpath_tmp.replace(u'?', u'_')
786
return osutils.pathjoin(self._root, prefix, relpath_tmp)
788
def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
790
if not force_temp and isinstance(tree, WorkingTree):
791
full_path = tree.abspath(tree.id2path(file_id))
792
if self._is_safepath(full_path):
795
full_path = self._safe_filename(prefix, relpath)
796
if not force_temp and self._try_symlink_root(tree, prefix):
710
798
parent_dir = osutils.dirname(full_path)
725
osutils.make_readonly(full_path)
726
mtime = tree.get_file_mtime(file_id)
727
os.utime(full_path, (mtime, mtime))
814
mtime = tree.get_file_mtime(file_id)
815
except errors.FileTimestampUnavailable:
818
os.utime(full_path, (mtime, mtime))
820
osutils.make_readonly(full_path)
730
def _prepare_files(self, file_id, old_path, new_path):
823
def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
824
allow_write_new=False):
731
825
old_disk_path = self._write_file(file_id, self.old_tree, 'old',
826
old_path, force_temp)
733
827
new_disk_path = self._write_file(file_id, self.new_tree, 'new',
828
new_path, force_temp,
829
allow_write=allow_write_new)
735
830
return old_disk_path, new_disk_path
737
832
def finish(self):
745
840
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
746
841
if (old_kind, new_kind) != ('file', 'file'):
747
842
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))
843
(old_disk_path, new_disk_path) = self._prepare_files(
844
file_id, old_path, new_path)
845
self._execute(old_disk_path, new_disk_path)
847
def edit_file(self, file_id):
848
"""Use this tool to edit a file.
850
A temporary copy will be edited, and the new contents will be
853
:param file_id: The id of the file to edit.
854
:return: The new contents of the file.
856
old_path = self.old_tree.id2path(file_id)
857
new_path = self.new_tree.id2path(file_id)
858
old_abs_path, new_abs_path = self._prepare_files(
859
file_id, old_path, new_path,
860
allow_write_new=True,
862
command = self._get_command(old_abs_path, new_abs_path)
863
subprocess.call(command, cwd=self._root)
864
new_file = open(new_abs_path, 'rb')
866
return new_file.read()
753
871
class DiffTree(object):
796
914
def from_trees_options(klass, old_tree, new_tree, to_file,
797
915
path_encoding, external_diff_options, old_label,
916
new_label, using, context_lines):
799
917
"""Factory for producing a DiffTree.
801
919
Designed to accept options used by show_diff_trees.
802
921
:param old_tree: The tree to show as old in the comparison
803
922
:param new_tree: The tree to show as new in the comparison
804
923
:param to_file: File to write comparisons to
810
929
:param using: Commandline to use to invoke an external diff tool
812
931
if using is not None:
813
extra_factories = [DiffFromTool.make_from_diff_tree(using)]
932
extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
815
934
extra_factories = []
816
935
if external_diff_options:
817
936
opts = external_diff_options.split()
818
def diff_file(olab, olines, nlab, nlines, to_file):
937
def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None, context_lines=None):
938
""":param path_encoding: not used but required
939
to match the signature of internal_diff.
819
941
external_diff(olab, olines, nlab, nlines, to_file, opts)
821
943
diff_file = internal_diff
822
944
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
823
old_label, new_label, diff_file)
945
old_label, new_label, diff_file, context_lines=context_lines)
824
946
return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
827
949
def show_diff(self, specific_files, extra_trees=None):
828
950
"""Write tree diff to self.to_file
830
:param sepecific_files: the specific files to compare (recursive)
952
:param specific_files: the specific files to compare (recursive)
831
953
:param extra_trees: extra trees to use for mapping paths to file_ids
added
removed
bzrlib/diff.py
