1
# Copyright (C) 2005-2011 Canonical Ltd.
2
# -*- coding: UTF-8 -*-
3
4
# This program is free software; you can redistribute it and/or modify
4
5
# it under the terms of the GNU General Public License as published by
5
6
# the Free Software Foundation; either version 2 of the License, or
6
7
# (at your option) any later version.
8
9
# This program is distributed in the hope that it will be useful,
9
10
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
11
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
12
# GNU General Public License for more details.
13
14
# You should have received a copy of the GNU General Public License
14
15
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23
from bzrlib.lazy_import import lazy_import
24
lazy_import(globals(), """
41
from bzrlib.workingtree import WorkingTree
42
from bzrlib.i18n import gettext
45
from bzrlib.registry import (
48
from bzrlib.trace import mutter, note, warning
51
class AtTemplate(string.Template):
52
"""Templating class that uses @ instead of $."""
57
# TODO: Rather than building a changeset object, we should probably
58
# invoke callbacks on an object. That object can either accumulate a
59
# list, write them out directly, etc etc.
62
class _PrematchedMatcher(difflib.SequenceMatcher):
63
"""Allow SequenceMatcher operations to use predetermined blocks"""
65
def __init__(self, matching_blocks):
66
difflib.SequenceMatcher(self, None, None)
67
self.matching_blocks = matching_blocks
71
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
72
allow_binary=False, sequence_matcher=None,
73
path_encoding='utf8'):
74
# FIXME: difflib is wrong if there is no trailing newline.
75
# The syntax used by patch seems to be "\ No newline at
76
# end of file" following the last diff line from that
77
# file. This is not trivial to insert into the
78
# unified_diff output and it might be better to just fix
79
# or replace that function.
81
# In the meantime we at least make sure the patch isn't
85
# Special workaround for Python2.3, where difflib fails if
86
# both sequences are empty.
87
if not oldlines and not newlines:
90
if allow_binary is False:
91
textfile.check_text_lines(oldlines)
92
textfile.check_text_lines(newlines)
94
if sequence_matcher is None:
95
sequence_matcher = patiencediff.PatienceSequenceMatcher
96
ud = patiencediff.unified_diff(oldlines, newlines,
97
fromfile=old_filename.encode(path_encoding, 'replace'),
98
tofile=new_filename.encode(path_encoding, 'replace'),
99
sequencematcher=sequence_matcher)
102
if len(ud) == 0: # Identical contents, nothing to do
104
# work-around for difflib being too smart for its own good
105
# if /dev/null is "1,0", patch won't recognize it as /dev/null
107
ud[2] = ud[2].replace('-1,0', '-0,0')
109
ud[2] = ud[2].replace('+1,0', '+0,0')
113
if not line.endswith('\n'):
114
to_file.write("\n\\ No newline at end of file\n")
118
def _spawn_external_diff(diffcmd, capture_errors=True):
119
"""Spawn the externall diff process, and return the child handle.
121
:param diffcmd: The command list to spawn
122
:param capture_errors: Capture stderr as well as setting LANG=C
123
and LC_ALL=C. This lets us read and understand the output of diff,
124
and respond to any errors.
125
:return: A Popen object.
128
# construct minimal environment
130
path = os.environ.get('PATH')
133
env['LANGUAGE'] = 'C' # on win32 only LANGUAGE has effect
136
stderr = subprocess.PIPE
142
pipe = subprocess.Popen(diffcmd,
143
stdin=subprocess.PIPE,
144
stdout=subprocess.PIPE,
148
if e.errno == errno.ENOENT:
149
raise errors.NoDiff(str(e))
155
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
157
"""Display a diff by calling out to the external diff program."""
158
# make sure our own output is properly ordered before the diff
161
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
162
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
163
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
164
newtmpf = os.fdopen(newtmp_fd, 'wb')
167
# TODO: perhaps a special case for comparing to or from the empty
168
# sequence; can just use /dev/null on Unix
170
# TODO: if either of the files being compared already exists as a
171
# regular named file (e.g. in the working directory) then we can
172
# compare directly to that, rather than copying it.
174
oldtmpf.writelines(oldlines)
175
newtmpf.writelines(newlines)
182
if sys.platform == 'win32':
183
# Popen doesn't do the proper encoding for external commands
184
# Since we are dealing with an ANSI api, use mbcs encoding
185
old_filename = old_filename.encode('mbcs')
186
new_filename = new_filename.encode('mbcs')
188
'--label', old_filename,
190
'--label', new_filename,
195
# diff only allows one style to be specified; they don't override.
196
# note that some of these take optargs, and the optargs can be
197
# directly appended to the options.
198
# this is only an approximate parser; it doesn't properly understand
200
for s in ['-c', '-u', '-C', '-U',
205
'-y', '--side-by-side',
217
diffcmd.extend(diff_opts)
219
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
220
out,err = pipe.communicate()
223
# internal_diff() adds a trailing newline, add one here for consistency
226
# 'diff' gives retcode == 2 for all sorts of errors
227
# one of those is 'Binary files differ'.
228
# Bad options could also be the problem.
229
# 'Binary files' is not a real error, so we suppress that error.
232
# Since we got here, we want to make sure to give an i18n error
233
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
234
out, err = pipe.communicate()
236
# Write out the new i18n diff response
237
to_file.write(out+'\n')
238
if pipe.returncode != 2:
239
raise errors.BzrError(
240
'external diff failed with exit code 2'
241
' when run with LANG=C and LC_ALL=C,'
242
' but not when run natively: %r' % (diffcmd,))
244
first_line = lang_c_out.split('\n', 1)[0]
245
# Starting with diffutils 2.8.4 the word "binary" was dropped.
246
m = re.match('^(binary )?files.*differ$', first_line, re.I)
248
raise errors.BzrError('external diff failed with exit code 2;'
249
' command: %r' % (diffcmd,))
251
# Binary files differ, just return
254
# If we got to here, we haven't written out the output of diff
258
# returns 1 if files differ; that's OK
260
msg = 'signal %d' % (-rc)
262
msg = 'exit code %d' % rc
264
raise errors.BzrError('external diff failed with %s; command: %r'
269
oldtmpf.close() # and delete
271
# Clean up. Warn in case the files couldn't be deleted
272
# (in case windows still holds the file open, but not
273
# if the files have already been deleted)
275
os.remove(old_abspath)
277
if e.errno not in (errno.ENOENT,):
278
warning('Failed to delete temporary file: %s %s',
281
os.remove(new_abspath)
283
if e.errno not in (errno.ENOENT,):
284
warning('Failed to delete temporary file: %s %s',
288
def get_trees_and_branches_to_diff_locked(
289
path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
290
"""Get the trees and specific files to diff given a list of paths.
292
This method works out the trees to be diff'ed and the files of
293
interest within those trees.
296
the list of arguments passed to the diff command
297
:param revision_specs:
298
Zero, one or two RevisionSpecs from the diff command line,
299
saying what revisions to compare.
301
The url of the old branch or tree. If None, the tree to use is
302
taken from the first path, if any, or the current working tree.
304
The url of the new branch or tree. If None, the tree to use is
305
taken from the first path, if any, or the current working tree.
307
a callable like Command.add_cleanup. get_trees_and_branches_to_diff
308
will register cleanups that must be run to unlock the trees, etc.
310
if True and a view is set, apply the view or check that the paths
313
a tuple of (old_tree, new_tree, old_branch, new_branch,
314
specific_files, extra_trees) where extra_trees is a sequence of
315
additional trees to search in for file-ids. The trees and branches
316
will be read-locked until the cleanups registered via the add_cleanup
319
# Get the old and new revision specs
320
old_revision_spec = None
321
new_revision_spec = None
322
if revision_specs is not None:
323
if len(revision_specs) > 0:
324
old_revision_spec = revision_specs[0]
326
old_url = old_revision_spec.get_branch()
327
if len(revision_specs) > 1:
328
new_revision_spec = revision_specs[1]
330
new_url = new_revision_spec.get_branch()
333
make_paths_wt_relative = True
334
consider_relpath = True
335
if path_list is None or len(path_list) == 0:
336
# If no path is given, the current working tree is used
337
default_location = u'.'
338
consider_relpath = False
339
elif old_url is not None and new_url is not None:
340
other_paths = path_list
341
make_paths_wt_relative = False
343
default_location = path_list[0]
344
other_paths = path_list[1:]
346
def lock_tree_or_branch(wt, br):
349
add_cleanup(wt.unlock)
352
add_cleanup(br.unlock)
354
# Get the old location
357
old_url = default_location
358
working_tree, branch, relpath = \
359
controldir.ControlDir.open_containing_tree_or_branch(old_url)
360
lock_tree_or_branch(working_tree, branch)
361
if consider_relpath and relpath != '':
362
if working_tree is not None and apply_view:
363
views.check_path_in_view(working_tree, relpath)
364
specific_files.append(relpath)
365
old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
368
# Get the new location
370
new_url = default_location
371
if new_url != old_url:
372
working_tree, branch, relpath = \
373
controldir.ControlDir.open_containing_tree_or_branch(new_url)
374
lock_tree_or_branch(working_tree, branch)
375
if consider_relpath and relpath != '':
376
if working_tree is not None and apply_view:
377
views.check_path_in_view(working_tree, relpath)
378
specific_files.append(relpath)
379
new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
380
basis_is_default=working_tree is None)
383
# Get the specific files (all files is None, no files is [])
384
if make_paths_wt_relative and working_tree is not None:
385
other_paths = working_tree.safe_relpath_files(
387
apply_view=apply_view)
388
specific_files.extend(other_paths)
389
if len(specific_files) == 0:
390
specific_files = None
391
if (working_tree is not None and working_tree.supports_views()
393
view_files = working_tree.views.lookup_view()
395
specific_files = view_files
396
view_str = views.view_display_str(view_files)
397
note(gettext("*** Ignoring files outside view. View is %s") % view_str)
399
# Get extra trees that ought to be searched for file-ids
401
if working_tree is not None and working_tree not in (old_tree, new_tree):
402
extra_trees = (working_tree,)
403
return (old_tree, new_tree, old_branch, new_branch,
404
specific_files, extra_trees)
407
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
408
if branch is None and tree is not None:
410
if spec is None or spec.spec is None:
413
return tree.basis_tree()
415
return branch.basis_tree()
418
return spec.as_tree(branch)
421
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
422
external_diff_options=None,
423
old_label='a/', new_label='b/',
425
path_encoding='utf8',
428
"""Show in text form the changes from one tree to another.
430
:param to_file: The output stream.
431
:param specific_files: Include only changes to these files - None for all
433
:param external_diff_options: If set, use an external GNU diff and pass
435
:param extra_trees: If set, more Trees to use for looking up file ids
436
:param path_encoding: If set, the path will be encoded as specified,
437
otherwise is supposed to be utf8
438
:param format_cls: Formatter class (DiffTree subclass)
440
if format_cls is None:
441
format_cls = DiffTree
444
if extra_trees is not None:
445
for tree in extra_trees:
449
differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
451
external_diff_options,
452
old_label, new_label, using)
453
return differ.show_diff(specific_files, extra_trees)
456
if extra_trees is not None:
457
for tree in extra_trees:
463
def _patch_header_date(tree, file_id, path):
464
"""Returns a timestamp suitable for use in a patch header."""
466
mtime = tree.get_file_mtime(file_id, path)
467
except errors.FileTimestampUnavailable:
469
return timestamp.format_patch_date(mtime)
472
def get_executable_change(old_is_x, new_is_x):
473
descr = { True:"+x", False:"-x", None:"??" }
474
if old_is_x != new_is_x:
475
return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
480
class DiffPath(object):
481
"""Base type for command object that compare files"""
483
# The type or contents of the file were unsuitable for diffing
484
CANNOT_DIFF = 'CANNOT_DIFF'
485
# The file has changed in a semantic way
487
# The file content may have changed, but there is no semantic change
488
UNCHANGED = 'UNCHANGED'
490
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
493
:param old_tree: The tree to show as the old tree in the comparison
494
:param new_tree: The tree to show as new in the comparison
495
:param to_file: The file to write comparison data to
496
:param path_encoding: The character encoding to write paths in
498
self.old_tree = old_tree
499
self.new_tree = new_tree
500
self.to_file = to_file
501
self.path_encoding = path_encoding
507
def from_diff_tree(klass, diff_tree):
508
return klass(diff_tree.old_tree, diff_tree.new_tree,
509
diff_tree.to_file, diff_tree.path_encoding)
512
def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
513
for file_differ in differs:
514
result = file_differ.diff(file_id, old_path, new_path, old_kind,
516
if result is not DiffPath.CANNOT_DIFF:
519
return DiffPath.CANNOT_DIFF
522
class DiffKindChange(object):
523
"""Special differ for file kind changes.
525
Represents kind change as deletion + creation. Uses the other differs
528
def __init__(self, differs):
529
self.differs = differs
535
def from_diff_tree(klass, diff_tree):
536
return klass(diff_tree.differs)
538
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
539
"""Perform comparison
541
:param file_id: The file_id of the file to compare
542
:param old_path: Path of the file in the old tree
543
:param new_path: Path of the file in the new tree
544
:param old_kind: Old file-kind of the file
545
:param new_kind: New file-kind of the file
547
if None in (old_kind, new_kind):
548
return DiffPath.CANNOT_DIFF
549
result = DiffPath._diff_many(self.differs, file_id, old_path,
550
new_path, old_kind, None)
551
if result is DiffPath.CANNOT_DIFF:
553
return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
557
class DiffDirectory(DiffPath):
559
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
560
"""Perform comparison between two directories. (dummy)
563
if 'directory' not in (old_kind, new_kind):
564
return self.CANNOT_DIFF
565
if old_kind not in ('directory', None):
566
return self.CANNOT_DIFF
567
if new_kind not in ('directory', None):
568
return self.CANNOT_DIFF
572
class DiffSymlink(DiffPath):
574
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
575
"""Perform comparison between two symlinks
577
:param file_id: The file_id of the file to compare
578
:param old_path: Path of the file in the old tree
579
:param new_path: Path of the file in the new tree
580
:param old_kind: Old file-kind of the file
581
:param new_kind: New file-kind of the file
583
if 'symlink' not in (old_kind, new_kind):
584
return self.CANNOT_DIFF
585
if old_kind == 'symlink':
586
old_target = self.old_tree.get_symlink_target(file_id)
587
elif old_kind is None:
590
return self.CANNOT_DIFF
591
if new_kind == 'symlink':
592
new_target = self.new_tree.get_symlink_target(file_id)
593
elif new_kind is None:
596
return self.CANNOT_DIFF
597
return self.diff_symlink(old_target, new_target)
599
def diff_symlink(self, old_target, new_target):
600
if old_target is None:
601
self.to_file.write('=== target is %r\n' % new_target)
602
elif new_target is None:
603
self.to_file.write('=== target was %r\n' % old_target)
605
self.to_file.write('=== target changed %r => %r\n' %
606
(old_target, new_target))
610
class DiffText(DiffPath):
612
# GNU Patch uses the epoch date to detect files that are being added
613
# or removed in a diff.
614
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
616
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
617
old_label='', new_label='', text_differ=internal_diff):
618
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
619
self.text_differ = text_differ
620
self.old_label = old_label
621
self.new_label = new_label
622
self.path_encoding = path_encoding
624
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
625
"""Compare two files in unified diff format
627
:param file_id: The file_id of the file to compare
628
:param old_path: Path of the file in the old tree
629
:param new_path: Path of the file in the new tree
630
:param old_kind: Old file-kind of the file
631
:param new_kind: New file-kind of the file
633
if 'file' not in (old_kind, new_kind):
634
return self.CANNOT_DIFF
635
from_file_id = to_file_id = file_id
636
if old_kind == 'file':
637
old_date = _patch_header_date(self.old_tree, file_id, old_path)
638
elif old_kind is None:
639
old_date = self.EPOCH_DATE
642
return self.CANNOT_DIFF
643
if new_kind == 'file':
644
new_date = _patch_header_date(self.new_tree, file_id, new_path)
645
elif new_kind is None:
646
new_date = self.EPOCH_DATE
649
return self.CANNOT_DIFF
650
from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
651
to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
652
return self.diff_text(from_file_id, to_file_id, from_label, to_label,
655
def diff_text(self, from_file_id, to_file_id, from_label, to_label,
656
from_path=None, to_path=None):
657
"""Diff the content of given files in two trees
659
:param from_file_id: The id of the file in the from tree. If None,
660
the file is not present in the from tree.
661
:param to_file_id: The id of the file in the to tree. This may refer
662
to a different file from from_file_id. If None,
663
the file is not present in the to tree.
664
:param from_path: The path in the from tree or None if unknown.
665
:param to_path: The path in the to tree or None if unknown.
667
def _get_text(tree, file_id, path):
668
if file_id is not None:
669
return tree.get_file_lines(file_id, path)
673
from_text = _get_text(self.old_tree, from_file_id, from_path)
674
to_text = _get_text(self.new_tree, to_file_id, to_path)
675
self.text_differ(from_label, from_text, to_label, to_text,
676
self.to_file, path_encoding=self.path_encoding)
677
except errors.BinaryFile:
679
("Binary files %s and %s differ\n" %
680
(from_label, to_label)).encode(self.path_encoding,'replace'))
684
class DiffFromTool(DiffPath):
686
def __init__(self, command_template, old_tree, new_tree, to_file,
687
path_encoding='utf-8'):
688
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
689
self.command_template = command_template
690
self._root = osutils.mkdtemp(prefix='bzr-diff-')
693
def from_string(klass, command_string, old_tree, new_tree, to_file,
694
path_encoding='utf-8'):
695
command_template = cmdline.split(command_string)
696
if '@' not in command_string:
697
command_template.extend(['@old_path', '@new_path'])
698
return klass(command_template, old_tree, new_tree, to_file,
702
def make_from_diff_tree(klass, command_string, external_diff_options=None):
703
def from_diff_tree(diff_tree):
704
full_command_string = [command_string]
705
if external_diff_options is not None:
706
full_command_string += ' ' + external_diff_options
707
return klass.from_string(full_command_string, diff_tree.old_tree,
708
diff_tree.new_tree, diff_tree.to_file)
709
return from_diff_tree
711
def _get_command(self, old_path, new_path):
712
my_map = {'old_path': old_path, 'new_path': new_path}
713
command = [AtTemplate(t).substitute(my_map) for t in
714
self.command_template]
715
if sys.platform == 'win32': # Popen doesn't accept unicode on win32
718
if isinstance(c, unicode):
719
command_encoded.append(c.encode('mbcs'))
721
command_encoded.append(c)
722
return command_encoded
726
def _execute(self, old_path, new_path):
727
command = self._get_command(old_path, new_path)
729
proc = subprocess.Popen(command, stdout=subprocess.PIPE,
732
if e.errno == errno.ENOENT:
733
raise errors.ExecutableMissing(command[0])
736
self.to_file.write(proc.stdout.read())
739
def _try_symlink_root(self, tree, prefix):
740
if (getattr(tree, 'abspath', None) is None
741
or not osutils.host_os_dereferences_symlinks()):
744
os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
746
if e.errno != errno.EEXIST:
752
"""Returns safe encoding for passing file path to diff tool"""
753
if sys.platform == 'win32':
756
# Don't fallback to 'utf-8' because subprocess may not be able to
757
# handle utf-8 correctly when locale is not utf-8.
758
return sys.getfilesystemencoding() or 'ascii'
760
def _is_safepath(self, path):
761
"""Return true if `path` may be able to pass to subprocess."""
764
return path == path.encode(fenc).decode(fenc)
768
def _safe_filename(self, prefix, relpath):
769
"""Replace unsafe character in `relpath` then join `self._root`,
770
`prefix` and `relpath`."""
772
# encoded_str.replace('?', '_') may break multibyte char.
773
# So we should encode, decode, then replace(u'?', u'_')
774
relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
775
relpath_tmp = relpath_tmp.replace(u'?', u'_')
776
return osutils.pathjoin(self._root, prefix, relpath_tmp)
778
def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
780
if not force_temp and isinstance(tree, WorkingTree):
781
full_path = tree.abspath(tree.id2path(file_id))
782
if self._is_safepath(full_path):
785
full_path = self._safe_filename(prefix, relpath)
786
if not force_temp and self._try_symlink_root(tree, prefix):
788
parent_dir = osutils.dirname(full_path)
790
os.makedirs(parent_dir)
792
if e.errno != errno.EEXIST:
794
source = tree.get_file(file_id, relpath)
796
target = open(full_path, 'wb')
798
osutils.pumpfile(source, target)
804
mtime = tree.get_file_mtime(file_id)
805
except errors.FileTimestampUnavailable:
808
os.utime(full_path, (mtime, mtime))
810
osutils.make_readonly(full_path)
813
def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
814
allow_write_new=False):
815
old_disk_path = self._write_file(file_id, self.old_tree, 'old',
816
old_path, force_temp)
817
new_disk_path = self._write_file(file_id, self.new_tree, 'new',
818
new_path, force_temp,
819
allow_write=allow_write_new)
820
return old_disk_path, new_disk_path
824
osutils.rmtree(self._root)
826
if e.errno != errno.ENOENT:
827
mutter("The temporary directory \"%s\" was not "
828
"cleanly removed: %s." % (self._root, e))
830
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
831
if (old_kind, new_kind) != ('file', 'file'):
832
return DiffPath.CANNOT_DIFF
833
(old_disk_path, new_disk_path) = self._prepare_files(
834
file_id, old_path, new_path)
835
self._execute(old_disk_path, new_disk_path)
837
def edit_file(self, file_id):
838
"""Use this tool to edit a file.
840
A temporary copy will be edited, and the new contents will be
843
:param file_id: The id of the file to edit.
844
:return: The new contents of the file.
846
old_path = self.old_tree.id2path(file_id)
847
new_path = self.new_tree.id2path(file_id)
848
old_abs_path, new_abs_path = self._prepare_files(
849
file_id, old_path, new_path,
850
allow_write_new=True,
852
command = self._get_command(old_abs_path, new_abs_path)
853
subprocess.call(command, cwd=self._root)
854
new_file = open(new_abs_path, 'rb')
856
return new_file.read()
861
class DiffTree(object):
862
"""Provides textual representations of the difference between two trees.
864
A DiffTree examines two trees and where a file-id has altered
865
between them, generates a textual representation of the difference.
866
DiffTree uses a sequence of DiffPath objects which are each
867
given the opportunity to handle a given altered fileid. The list
868
of DiffPath objects can be extended globally by appending to
869
DiffTree.diff_factories, or for a specific diff operation by
870
supplying the extra_factories option to the appropriate method.
873
# list of factories that can provide instances of DiffPath objects
874
# may be extended by plugins.
875
diff_factories = [DiffSymlink.from_diff_tree,
876
DiffDirectory.from_diff_tree]
878
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
879
diff_text=None, extra_factories=None):
882
:param old_tree: Tree to show as old in the comparison
883
:param new_tree: Tree to show as new in the comparison
884
:param to_file: File to write comparision to
885
:param path_encoding: Character encoding to write paths in
886
:param diff_text: DiffPath-type object to use as a last resort for
888
:param extra_factories: Factories of DiffPaths to try before any other
890
if diff_text is None:
891
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
892
'', '', internal_diff)
893
self.old_tree = old_tree
894
self.new_tree = new_tree
895
self.to_file = to_file
896
self.path_encoding = path_encoding
898
if extra_factories is not None:
899
self.differs.extend(f(self) for f in extra_factories)
900
self.differs.extend(f(self) for f in self.diff_factories)
901
self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
904
def from_trees_options(klass, old_tree, new_tree, to_file,
905
path_encoding, external_diff_options, old_label,
907
"""Factory for producing a DiffTree.
909
Designed to accept options used by show_diff_trees.
911
:param old_tree: The tree to show as old in the comparison
912
:param new_tree: The tree to show as new in the comparison
913
:param to_file: File to write comparisons to
914
:param path_encoding: Character encoding to use for writing paths
915
:param external_diff_options: If supplied, use the installed diff
916
binary to perform file comparison, using supplied options.
917
:param old_label: Prefix to use for old file labels
918
:param new_label: Prefix to use for new file labels
919
:param using: Commandline to use to invoke an external diff tool
921
if using is not None:
922
extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
925
if external_diff_options:
926
opts = external_diff_options.split()
927
def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None):
928
""":param path_encoding: not used but required
929
to match the signature of internal_diff.
931
external_diff(olab, olines, nlab, nlines, to_file, opts)
933
diff_file = internal_diff
934
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
935
old_label, new_label, diff_file)
936
return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
939
def show_diff(self, specific_files, extra_trees=None):
940
"""Write tree diff to self.to_file
942
:param specific_files: the specific files to compare (recursive)
943
:param extra_trees: extra trees to use for mapping paths to file_ids
946
return self._show_diff(specific_files, extra_trees)
948
for differ in self.differs:
951
def _show_diff(self, specific_files, extra_trees):
952
# TODO: Generation of pseudo-diffs for added/deleted files could
953
# be usefully made into a much faster special case.
954
iterator = self.new_tree.iter_changes(self.old_tree,
955
specific_files=specific_files,
956
extra_trees=extra_trees,
957
require_versioned=True)
959
def changes_key(change):
960
old_path, new_path = change[1]
965
def get_encoded_path(path):
967
return path.encode(self.path_encoding, "replace")
968
for (file_id, paths, changed_content, versioned, parent, name, kind,
969
executable) in sorted(iterator, key=changes_key):
970
# The root does not get diffed, and items with no known kind (that
971
# is, missing) in both trees are skipped as well.
972
if parent == (None, None) or kind == (None, None):
974
oldpath, newpath = paths
975
oldpath_encoded = get_encoded_path(paths[0])
976
newpath_encoded = get_encoded_path(paths[1])
977
old_present = (kind[0] is not None and versioned[0])
978
new_present = (kind[1] is not None and versioned[1])
979
renamed = (parent[0], name[0]) != (parent[1], name[1])
981
properties_changed = []
982
properties_changed.extend(get_executable_change(executable[0], executable[1]))
984
if properties_changed:
985
prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
989
if (old_present, new_present) == (True, False):
990
self.to_file.write("=== removed %s '%s'\n" %
991
(kind[0], oldpath_encoded))
993
elif (old_present, new_present) == (False, True):
994
self.to_file.write("=== added %s '%s'\n" %
995
(kind[1], newpath_encoded))
998
self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
999
(kind[0], oldpath_encoded, newpath_encoded, prop_str))
1001
# if it was produced by iter_changes, it must be
1002
# modified *somehow*, either content or execute bit.
1003
self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
1004
newpath_encoded, prop_str))
1006
self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1012
def diff(self, file_id, old_path, new_path):
1013
"""Perform a diff of a single file
1015
:param file_id: file-id of the file
1016
:param old_path: The path of the file in the old tree
1017
:param new_path: The path of the file in the new tree
1020
old_kind = self.old_tree.kind(file_id)
1021
except (errors.NoSuchId, errors.NoSuchFile):
1024
new_kind = self.new_tree.kind(file_id)
1025
except (errors.NoSuchId, errors.NoSuchFile):
1027
self._diff(file_id, old_path, new_path, old_kind, new_kind)
1030
def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
1031
result = DiffPath._diff_many(self.differs, file_id, old_path,
1032
new_path, old_kind, new_kind)
1033
if result is DiffPath.CANNOT_DIFF:
1034
error_path = new_path
1035
if error_path is None:
1036
error_path = old_path
1037
raise errors.NoDiffFound(error_path)
1040
format_registry = Registry()
1041
format_registry.register('default', DiffTree)
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20
from trace import mutter
27
def diff_trees(old_tree, new_tree):
28
"""Compute diff between two trees.
30
They may be in different branches and may be working or historical
33
Yields a sequence of (state, id, old_name, new_name, kind).
34
Each filename and each id is listed only once.
37
## TODO: Compare files before diffing; only mention those that have changed
39
## TODO: Set nice names in the headers, maybe include diffstat
41
## TODO: Perhaps make this a generator rather than using
44
## TODO: Allow specifying a list of files to compare, rather than
45
## doing the whole tree? (Not urgent.)
47
## TODO: Allow diffing any two inventories, not just the
48
## current one against one. We mgiht need to specify two
49
## stores to look for the files if diffing two branches. That
50
## might imply this shouldn't be primarily a Branch method.
52
## XXX: This doesn't report on unknown files; that can be done
53
## from a separate method.
55
old_it = old_tree.list_files()
56
new_it = new_tree.list_files()
64
old_item = next(old_it)
65
new_item = next(new_it)
67
# We step through the two sorted iterators in parallel, trying to
70
while (old_item != None) or (new_item != None):
71
# OK, we still have some remaining on both, but they may be
74
old_name, old_class, old_kind, old_id = old_item
79
new_name, new_class, new_kind, new_id = new_item
83
mutter(" diff pairwise %r" % (old_item,))
84
mutter(" %r" % (new_item,))
87
# can't handle the old tree being a WorkingTree
88
assert old_class == 'V'
90
if new_item and (new_class != 'V'):
91
yield new_class, None, None, new_name, new_kind
92
new_item = next(new_it)
93
elif (not new_item) or (old_item and (old_name < new_name)):
94
mutter(" extra entry in old-tree sequence")
95
if new_tree.has_id(old_id):
96
# will be mentioned as renamed under new name
99
yield 'D', old_id, old_name, None, old_kind
100
old_item = next(old_it)
101
elif (not old_item) or (new_item and (new_name < old_name)):
102
mutter(" extra entry in new-tree sequence")
103
if old_tree.has_id(new_id):
104
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
106
yield 'A', new_id, None, new_name, new_kind
107
new_item = next(new_it)
108
elif old_id != new_id:
109
assert old_name == new_name
110
# both trees have a file of this name, but it is not the
111
# same file. in other words, the old filename has been
112
# overwritten by either a newly-added or a renamed file.
113
# (should we return something about the overwritten file?)
114
if old_tree.has_id(new_id):
115
# renaming, overlying a deleted file
116
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
118
yield 'A', new_id, None, new_name, new_kind
120
new_item = next(new_it)
121
old_item = next(old_it)
123
assert old_id == new_id
124
assert old_name == new_name
125
assert old_kind == new_kind
127
if old_kind == 'directory':
128
yield '.', new_id, old_name, new_name, new_kind
129
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
130
mutter(" file size has changed, must be different")
131
yield 'M', new_id, old_name, new_name, new_kind
132
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
133
mutter(" SHA1 indicates they're identical")
134
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
135
yield '.', new_id, old_name, new_name, new_kind
137
mutter(" quick compare shows different")
138
yield 'M', new_id, old_name, new_name, new_kind
140
new_item = next(new_it)
141
old_item = next(old_it)
added
removed
bzrlib/diff.py
