1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
1
# Copyright (C) 2005-2014 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
110
121
def _spawn_external_diff(diffcmd, capture_errors=True):
111
"""Spawn the externall diff process, and return the child handle.
122
"""Spawn the external diff process, and return the child handle.
113
124
:param diffcmd: The command list to spawn
114
125
:param capture_errors: Capture stderr as well as setting LANG=C
157
# diff style options as of GNU diff v3.2
158
style_option_list = ['-c', '-C', '--context',
160
'-f', '--forward-ed',
164
'-u', '-U', '--unified',
165
'-y', '--side-by-side',
168
def default_style_unified(diff_opts):
169
"""Default to unified diff style if alternative not specified in diff_opts.
171
diff only allows one style to be specified; they don't override.
172
Note that some of these take optargs, and the optargs can be
173
directly appended to the options.
174
This is only an approximate parser; it doesn't properly understand
177
:param diff_opts: List of options for external (GNU) diff.
178
:return: List of options with default style=='unified'.
180
for s in style_option_list:
188
diff_opts.append('-u')
147
192
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
172
217
if not diff_opts:
219
if sys.platform == 'win32':
220
# Popen doesn't do the proper encoding for external commands
221
# Since we are dealing with an ANSI api, use mbcs encoding
222
old_filename = old_filename.encode('mbcs')
223
new_filename = new_filename.encode('mbcs')
174
224
diffcmd = ['diff',
175
225
'--label', old_filename,
182
# diff only allows one style to be specified; they don't override.
183
# note that some of these take optargs, and the optargs can be
184
# directly appended to the options.
185
# this is only an approximate parser; it doesn't properly understand
187
for s in ['-c', '-u', '-C', '-U',
192
'-y', '--side-by-side',
232
diff_opts = default_style_unified(diff_opts)
204
235
diffcmd.extend(diff_opts)
249
280
msg = 'exit code %d' % rc
251
282
raise errors.BzrError('external diff failed with %s; command: %r'
256
287
oldtmpf.close() # and delete
258
# Clean up. Warn in case the files couldn't be deleted
259
# (in case windows still holds the file open, but not
260
# if the files have already been deleted)
262
os.remove(old_abspath)
264
if e.errno not in (errno.ENOENT,):
265
warning('Failed to delete temporary file: %s %s',
268
os.remove(new_abspath)
270
if e.errno not in (errno.ENOENT,):
271
warning('Failed to delete temporary file: %s %s',
275
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url,
291
# Warn in case the file couldn't be deleted (in case windows still
292
# holds the file open, but not if the files have already been
297
if e.errno not in (errno.ENOENT,):
298
warning('Failed to delete temporary file: %s %s', path, e)
304
def get_trees_and_branches_to_diff_locked(
305
path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
277
306
"""Get the trees and specific files to diff given a list of paths.
279
308
This method works out the trees to be diff'ed and the files of
291
320
The url of the new branch or tree. If None, the tree to use is
292
321
taken from the first path, if any, or the current working tree.
323
a callable like Command.add_cleanup. get_trees_and_branches_to_diff
324
will register cleanups that must be run to unlock the trees, etc.
293
325
:param apply_view:
294
326
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
329
a tuple of (old_tree, new_tree, old_branch, new_branch,
330
specific_files, extra_trees) where extra_trees is a sequence of
331
additional trees to search in for file-ids. The trees and branches
332
will be read-locked until the cleanups registered via the add_cleanup
301
335
# Get the old and new revision specs
302
336
old_revision_spec = None
325
359
default_location = path_list[0]
326
360
other_paths = path_list[1:]
362
def lock_tree_or_branch(wt, br):
365
add_cleanup(wt.unlock)
368
add_cleanup(br.unlock)
328
370
# Get the old location
329
371
specific_files = []
330
372
if old_url is None:
331
373
old_url = default_location
332
374
working_tree, branch, relpath = \
333
bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
375
controldir.ControlDir.open_containing_tree_or_branch(old_url)
376
lock_tree_or_branch(working_tree, branch)
334
377
if consider_relpath and relpath != '':
335
378
if working_tree is not None and apply_view:
336
379
views.check_path_in_view(working_tree, relpath)
337
380
specific_files.append(relpath)
338
381
old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
340
384
# Get the new location
341
385
if new_url is None:
342
386
new_url = default_location
343
387
if new_url != old_url:
344
388
working_tree, branch, relpath = \
345
bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
389
controldir.ControlDir.open_containing_tree_or_branch(new_url)
390
lock_tree_or_branch(working_tree, branch)
346
391
if consider_relpath and relpath != '':
347
392
if working_tree is not None and apply_view:
348
393
views.check_path_in_view(working_tree, relpath)
349
394
specific_files.append(relpath)
350
395
new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
351
396
basis_is_default=working_tree is None)
353
399
# Get the specific files (all files is None, no files is [])
354
400
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,
401
other_paths = working_tree.safe_relpath_files(
358
403
apply_view=apply_view)
359
except errors.FileInWrongBranch:
360
raise errors.BzrCommandError("Files are in different branches")
361
404
specific_files.extend(other_paths)
362
405
if len(specific_files) == 0:
363
406
specific_files = None
368
411
specific_files = view_files
369
412
view_str = views.view_display_str(view_files)
370
note("*** Ignoring files outside view. View is %s" % view_str)
413
note(gettext("*** Ignoring files outside view. View is %s") % view_str)
372
415
# Get extra trees that ought to be searched for file-ids
373
416
extra_trees = None
374
417
if working_tree is not None and working_tree not in (old_tree, new_tree):
375
418
extra_trees = (working_tree,)
376
return old_tree, new_tree, specific_files, extra_trees
419
return (old_tree, new_tree, old_branch, new_branch,
420
specific_files, extra_trees)
378
423
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
379
424
if branch is None and tree is not None:
394
439
old_label='a/', new_label='b/',
395
440
extra_trees=None,
396
441
path_encoding='utf8',
444
context=DEFAULT_CONTEXT_AMOUNT):
398
445
"""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
447
:param to_file: The output stream.
448
:param specific_files: Include only changes to these files - None for all
450
:param external_diff_options: If set, use an external GNU diff and pass
452
:param extra_trees: If set, more Trees to use for looking up file ids
453
:param path_encoding: If set, the path will be encoded as specified,
454
otherwise is supposed to be utf8
455
:param format_cls: Formatter class (DiffTree subclass)
458
context = DEFAULT_CONTEXT_AMOUNT
459
if format_cls is None:
460
format_cls = DiffTree
416
461
old_tree.lock_read()
418
463
if extra_trees is not None:
421
466
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)
468
differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
470
external_diff_options,
471
old_label, new_label, using,
472
context_lines=context)
427
473
return differ.show_diff(specific_files, extra_trees)
429
475
new_tree.unlock()
437
483
def _patch_header_date(tree, file_id, path):
438
484
"""Returns a timestamp suitable for use in a patch header."""
439
mtime = tree.get_file_mtime(file_id, path)
486
mtime = tree.get_file_mtime(file_id, path)
487
except errors.FileTimestampUnavailable:
440
489
return timestamp.format_patch_date(mtime)
584
633
# or removed in a diff.
585
634
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):
636
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
637
old_label='', new_label='', text_differ=internal_diff,
638
context_lines=DEFAULT_CONTEXT_AMOUNT):
589
639
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
590
640
self.text_differ = text_differ
591
641
self.old_label = old_label
592
642
self.new_label = new_label
593
643
self.path_encoding = path_encoding
644
self.context_lines = context_lines
595
646
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
596
647
"""Compare two files in unified diff format
638
689
def _get_text(tree, file_id, path):
639
690
if file_id is not None:
640
return tree.get_file(file_id, path).readlines()
691
return tree.get_file_lines(file_id, path)
644
695
from_text = _get_text(self.old_tree, from_file_id, from_path)
645
696
to_text = _get_text(self.new_tree, to_file_id, to_path)
646
697
self.text_differ(from_label, from_text, to_label, to_text,
698
self.to_file, path_encoding=self.path_encoding,
699
context_lines=self.context_lines)
648
700
except errors.BinaryFile:
649
701
self.to_file.write(
650
702
("Binary files %s and %s differ\n" %
651
(from_label, to_label)).encode(self.path_encoding))
703
(from_label, to_label)).encode(self.path_encoding,'replace'))
652
704
return self.CHANGED
664
716
def from_string(klass, command_string, old_tree, new_tree, to_file,
665
717
path_encoding='utf-8'):
666
command_template = commands.shlex_split_unicode(command_string)
667
command_template.extend(['%(old_path)s', '%(new_path)s'])
718
command_template = cmdline.split(command_string)
719
if '@' not in command_string:
720
command_template.extend(['@old_path', '@new_path'])
668
721
return klass(command_template, old_tree, new_tree, to_file,
672
def make_from_diff_tree(klass, command_string):
725
def make_from_diff_tree(klass, command_string, external_diff_options=None):
673
726
def from_diff_tree(diff_tree):
674
return klass.from_string(command_string, diff_tree.old_tree,
727
full_command_string = [command_string]
728
if external_diff_options is not None:
729
full_command_string += ' ' + external_diff_options
730
return klass.from_string(full_command_string, diff_tree.old_tree,
675
731
diff_tree.new_tree, diff_tree.to_file)
676
732
return from_diff_tree
678
734
def _get_command(self, old_path, new_path):
679
735
my_map = {'old_path': old_path, 'new_path': new_path}
680
return [t % my_map for t in self.command_template]
736
command = [AtTemplate(t).substitute(my_map) for t in
737
self.command_template]
738
if sys.platform == 'win32': # Popen doesn't accept unicode on win32
741
if isinstance(c, unicode):
742
command_encoded.append(c.encode('mbcs'))
744
command_encoded.append(c)
745
return command_encoded
682
749
def _execute(self, old_path, new_path):
683
750
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):
775
"""Returns safe encoding for passing file path to diff tool"""
776
if sys.platform == 'win32':
779
# Don't fallback to 'utf-8' because subprocess may not be able to
780
# handle utf-8 correctly when locale is not utf-8.
781
return sys.getfilesystemencoding() or 'ascii'
783
def _is_safepath(self, path):
784
"""Return true if `path` may be able to pass to subprocess."""
787
return path == path.encode(fenc).decode(fenc)
791
def _safe_filename(self, prefix, relpath):
792
"""Replace unsafe character in `relpath` then join `self._root`,
793
`prefix` and `relpath`."""
795
# encoded_str.replace('?', '_') may break multibyte char.
796
# So we should encode, decode, then replace(u'?', u'_')
797
relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
798
relpath_tmp = relpath_tmp.replace(u'?', u'_')
799
return osutils.pathjoin(self._root, prefix, relpath_tmp)
801
def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
803
if not force_temp and isinstance(tree, WorkingTree):
804
full_path = tree.abspath(tree.id2path(file_id))
805
if self._is_safepath(full_path):
808
full_path = self._safe_filename(prefix, relpath)
809
if not force_temp and self._try_symlink_root(tree, prefix):
710
811
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))
827
mtime = tree.get_file_mtime(file_id)
828
except errors.FileTimestampUnavailable:
831
os.utime(full_path, (mtime, mtime))
833
osutils.make_readonly(full_path)
730
def _prepare_files(self, file_id, old_path, new_path):
836
def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
837
allow_write_new=False):
731
838
old_disk_path = self._write_file(file_id, self.old_tree, 'old',
839
old_path, force_temp)
733
840
new_disk_path = self._write_file(file_id, self.new_tree, 'new',
841
new_path, force_temp,
842
allow_write=allow_write_new)
735
843
return old_disk_path, new_disk_path
737
845
def finish(self):
745
853
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
746
854
if (old_kind, new_kind) != ('file', 'file'):
747
855
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))
856
(old_disk_path, new_disk_path) = self._prepare_files(
857
file_id, old_path, new_path)
858
self._execute(old_disk_path, new_disk_path)
860
def edit_file(self, file_id):
861
"""Use this tool to edit a file.
863
A temporary copy will be edited, and the new contents will be
866
:param file_id: The id of the file to edit.
867
:return: The new contents of the file.
869
old_path = self.old_tree.id2path(file_id)
870
new_path = self.new_tree.id2path(file_id)
871
old_abs_path, new_abs_path = self._prepare_files(
872
file_id, old_path, new_path,
873
allow_write_new=True,
875
command = self._get_command(old_abs_path, new_abs_path)
876
subprocess.call(command, cwd=self._root)
877
new_file = open(new_abs_path, 'rb')
879
return new_file.read()
753
884
class DiffTree(object):
796
927
def from_trees_options(klass, old_tree, new_tree, to_file,
797
928
path_encoding, external_diff_options, old_label,
929
new_label, using, context_lines):
799
930
"""Factory for producing a DiffTree.
801
932
Designed to accept options used by show_diff_trees.
802
934
:param old_tree: The tree to show as old in the comparison
803
935
:param new_tree: The tree to show as new in the comparison
804
936
:param to_file: File to write comparisons to
810
942
:param using: Commandline to use to invoke an external diff tool
812
944
if using is not None:
813
extra_factories = [DiffFromTool.make_from_diff_tree(using)]
945
extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
815
947
extra_factories = []
816
948
if external_diff_options:
817
949
opts = external_diff_options.split()
818
def diff_file(olab, olines, nlab, nlines, to_file):
950
def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None, context_lines=None):
951
""":param path_encoding: not used but required
952
to match the signature of internal_diff.
819
954
external_diff(olab, olines, nlab, nlines, to_file, opts)
821
956
diff_file = internal_diff
822
957
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
823
old_label, new_label, diff_file)
958
old_label, new_label, diff_file, context_lines=context_lines)
824
959
return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
827
962
def show_diff(self, specific_files, extra_trees=None):
828
963
"""Write tree diff to self.to_file
830
:param sepecific_files: the specific files to compare (recursive)
965
:param specific_files: the specific files to compare (recursive)
831
966
:param extra_trees: extra trees to use for mapping paths to file_ids
added
removed
bzrlib/diff.py
