2
# -*- coding: UTF-8 -*-
1
# Copyright (C) 2005-2011 Canonical Ltd.
4
3
# This program is free software; you can redistribute it and/or modify
5
4
# it under the terms of the GNU General Public License as published by
6
5
# the Free Software Foundation; either version 2 of the License, or
7
6
# (at your option) any later version.
9
8
# This program is distributed in the hope that it will be useful,
10
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
11
# GNU General Public License for more details.
14
13
# You should have received a copy of the GNU General Public License
15
14
# along with this program; if not, write to the Free Software
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_id != None
125
assert old_name == new_name
126
assert old_kind == new_kind
128
if old_kind == 'directory':
129
yield '.', new_id, old_name, new_name, new_kind
130
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
131
mutter(" file size has changed, must be different")
132
yield 'M', new_id, old_name, new_name, new_kind
133
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
134
mutter(" SHA1 indicates they're identical")
135
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
136
yield '.', new_id, old_name, new_name, new_kind
138
mutter(" quick compare shows different")
139
yield 'M', new_id, old_name, new_name, new_kind
141
new_item = next(new_it)
142
old_item = next(old_it)
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
from __future__ import absolute_import
25
from bzrlib.lazy_import import lazy_import
26
lazy_import(globals(), """
43
from bzrlib.workingtree import WorkingTree
44
from bzrlib.i18n import gettext
47
from bzrlib.registry import (
50
from bzrlib.trace import mutter, note, warning
53
class AtTemplate(string.Template):
54
"""Templating class that uses @ instead of $."""
59
# TODO: Rather than building a changeset object, we should probably
60
# invoke callbacks on an object. That object can either accumulate a
61
# list, write them out directly, etc etc.
64
class _PrematchedMatcher(difflib.SequenceMatcher):
65
"""Allow SequenceMatcher operations to use predetermined blocks"""
67
def __init__(self, matching_blocks):
68
difflib.SequenceMatcher(self, None, None)
69
self.matching_blocks = matching_blocks
73
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
74
allow_binary=False, sequence_matcher=None,
75
path_encoding='utf8'):
76
# FIXME: difflib is wrong if there is no trailing newline.
77
# The syntax used by patch seems to be "\ No newline at
78
# end of file" following the last diff line from that
79
# file. This is not trivial to insert into the
80
# unified_diff output and it might be better to just fix
81
# or replace that function.
83
# In the meantime we at least make sure the patch isn't
87
# Special workaround for Python2.3, where difflib fails if
88
# both sequences are empty.
89
if not oldlines and not newlines:
92
if allow_binary is False:
93
textfile.check_text_lines(oldlines)
94
textfile.check_text_lines(newlines)
96
if sequence_matcher is None:
97
sequence_matcher = patiencediff.PatienceSequenceMatcher
98
ud = patiencediff.unified_diff(oldlines, newlines,
99
fromfile=old_filename.encode(path_encoding, 'replace'),
100
tofile=new_filename.encode(path_encoding, 'replace'),
101
sequencematcher=sequence_matcher)
104
if len(ud) == 0: # Identical contents, nothing to do
106
# work-around for difflib being too smart for its own good
107
# if /dev/null is "1,0", patch won't recognize it as /dev/null
109
ud[2] = ud[2].replace('-1,0', '-0,0')
111
ud[2] = ud[2].replace('+1,0', '+0,0')
115
if not line.endswith('\n'):
116
to_file.write("\n\\ No newline at end of file\n")
120
def _spawn_external_diff(diffcmd, capture_errors=True):
121
"""Spawn the externall diff process, and return the child handle.
123
:param diffcmd: The command list to spawn
124
:param capture_errors: Capture stderr as well as setting LANG=C
125
and LC_ALL=C. This lets us read and understand the output of diff,
126
and respond to any errors.
127
:return: A Popen object.
130
# construct minimal environment
132
path = os.environ.get('PATH')
135
env['LANGUAGE'] = 'C' # on win32 only LANGUAGE has effect
138
stderr = subprocess.PIPE
144
pipe = subprocess.Popen(diffcmd,
145
stdin=subprocess.PIPE,
146
stdout=subprocess.PIPE,
150
if e.errno == errno.ENOENT:
151
raise errors.NoDiff(str(e))
157
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
159
"""Display a diff by calling out to the external diff program."""
160
# make sure our own output is properly ordered before the diff
163
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
164
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
165
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
166
newtmpf = os.fdopen(newtmp_fd, 'wb')
169
# TODO: perhaps a special case for comparing to or from the empty
170
# sequence; can just use /dev/null on Unix
172
# TODO: if either of the files being compared already exists as a
173
# regular named file (e.g. in the working directory) then we can
174
# compare directly to that, rather than copying it.
176
oldtmpf.writelines(oldlines)
177
newtmpf.writelines(newlines)
184
if sys.platform == 'win32':
185
# Popen doesn't do the proper encoding for external commands
186
# Since we are dealing with an ANSI api, use mbcs encoding
187
old_filename = old_filename.encode('mbcs')
188
new_filename = new_filename.encode('mbcs')
190
'--label', old_filename,
192
'--label', new_filename,
197
# diff only allows one style to be specified; they don't override.
198
# note that some of these take optargs, and the optargs can be
199
# directly appended to the options.
200
# this is only an approximate parser; it doesn't properly understand
202
for s in ['-c', '-u', '-C', '-U',
207
'-y', '--side-by-side',
219
diffcmd.extend(diff_opts)
221
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
222
out,err = pipe.communicate()
225
# internal_diff() adds a trailing newline, add one here for consistency
228
# 'diff' gives retcode == 2 for all sorts of errors
229
# one of those is 'Binary files differ'.
230
# Bad options could also be the problem.
231
# 'Binary files' is not a real error, so we suppress that error.
234
# Since we got here, we want to make sure to give an i18n error
235
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
236
out, err = pipe.communicate()
238
# Write out the new i18n diff response
239
to_file.write(out+'\n')
240
if pipe.returncode != 2:
241
raise errors.BzrError(
242
'external diff failed with exit code 2'
243
' when run with LANG=C and LC_ALL=C,'
244
' but not when run natively: %r' % (diffcmd,))
246
first_line = lang_c_out.split('\n', 1)[0]
247
# Starting with diffutils 2.8.4 the word "binary" was dropped.
248
m = re.match('^(binary )?files.*differ$', first_line, re.I)
250
raise errors.BzrError('external diff failed with exit code 2;'
251
' command: %r' % (diffcmd,))
253
# Binary files differ, just return
256
# If we got to here, we haven't written out the output of diff
260
# returns 1 if files differ; that's OK
262
msg = 'signal %d' % (-rc)
264
msg = 'exit code %d' % rc
266
raise errors.BzrError('external diff failed with %s; command: %r'
271
oldtmpf.close() # and delete
273
# Clean up. Warn in case the files couldn't be deleted
274
# (in case windows still holds the file open, but not
275
# if the files have already been deleted)
277
os.remove(old_abspath)
279
if e.errno not in (errno.ENOENT,):
280
warning('Failed to delete temporary file: %s %s',
283
os.remove(new_abspath)
285
if e.errno not in (errno.ENOENT,):
286
warning('Failed to delete temporary file: %s %s',
290
def get_trees_and_branches_to_diff_locked(
291
path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
292
"""Get the trees and specific files to diff given a list of paths.
294
This method works out the trees to be diff'ed and the files of
295
interest within those trees.
298
the list of arguments passed to the diff command
299
:param revision_specs:
300
Zero, one or two RevisionSpecs from the diff command line,
301
saying what revisions to compare.
303
The url of the old branch or tree. If None, the tree to use is
304
taken from the first path, if any, or the current working tree.
306
The url of the new branch or tree. If None, the tree to use is
307
taken from the first path, if any, or the current working tree.
309
a callable like Command.add_cleanup. get_trees_and_branches_to_diff
310
will register cleanups that must be run to unlock the trees, etc.
312
if True and a view is set, apply the view or check that the paths
315
a tuple of (old_tree, new_tree, old_branch, new_branch,
316
specific_files, extra_trees) where extra_trees is a sequence of
317
additional trees to search in for file-ids. The trees and branches
318
will be read-locked until the cleanups registered via the add_cleanup
321
# Get the old and new revision specs
322
old_revision_spec = None
323
new_revision_spec = None
324
if revision_specs is not None:
325
if len(revision_specs) > 0:
326
old_revision_spec = revision_specs[0]
328
old_url = old_revision_spec.get_branch()
329
if len(revision_specs) > 1:
330
new_revision_spec = revision_specs[1]
332
new_url = new_revision_spec.get_branch()
335
make_paths_wt_relative = True
336
consider_relpath = True
337
if path_list is None or len(path_list) == 0:
338
# If no path is given, the current working tree is used
339
default_location = u'.'
340
consider_relpath = False
341
elif old_url is not None and new_url is not None:
342
other_paths = path_list
343
make_paths_wt_relative = False
345
default_location = path_list[0]
346
other_paths = path_list[1:]
348
def lock_tree_or_branch(wt, br):
351
add_cleanup(wt.unlock)
354
add_cleanup(br.unlock)
356
# Get the old location
359
old_url = default_location
360
working_tree, branch, relpath = \
361
controldir.ControlDir.open_containing_tree_or_branch(old_url)
362
lock_tree_or_branch(working_tree, branch)
363
if consider_relpath and relpath != '':
364
if working_tree is not None and apply_view:
365
views.check_path_in_view(working_tree, relpath)
366
specific_files.append(relpath)
367
old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
370
# Get the new location
372
new_url = default_location
373
if new_url != old_url:
374
working_tree, branch, relpath = \
375
controldir.ControlDir.open_containing_tree_or_branch(new_url)
376
lock_tree_or_branch(working_tree, branch)
377
if consider_relpath and relpath != '':
378
if working_tree is not None and apply_view:
379
views.check_path_in_view(working_tree, relpath)
380
specific_files.append(relpath)
381
new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
382
basis_is_default=working_tree is None)
385
# Get the specific files (all files is None, no files is [])
386
if make_paths_wt_relative and working_tree is not None:
387
other_paths = working_tree.safe_relpath_files(
389
apply_view=apply_view)
390
specific_files.extend(other_paths)
391
if len(specific_files) == 0:
392
specific_files = None
393
if (working_tree is not None and working_tree.supports_views()
395
view_files = working_tree.views.lookup_view()
397
specific_files = view_files
398
view_str = views.view_display_str(view_files)
399
note(gettext("*** Ignoring files outside view. View is %s") % view_str)
401
# Get extra trees that ought to be searched for file-ids
403
if working_tree is not None and working_tree not in (old_tree, new_tree):
404
extra_trees = (working_tree,)
405
return (old_tree, new_tree, old_branch, new_branch,
406
specific_files, extra_trees)
409
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
410
if branch is None and tree is not None:
412
if spec is None or spec.spec is None:
415
return tree.basis_tree()
417
return branch.basis_tree()
420
return spec.as_tree(branch)
423
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
424
external_diff_options=None,
425
old_label='a/', new_label='b/',
427
path_encoding='utf8',
430
"""Show in text form the changes from one tree to another.
432
:param to_file: The output stream.
433
:param specific_files: Include only changes to these files - None for all
435
:param external_diff_options: If set, use an external GNU diff and pass
437
:param extra_trees: If set, more Trees to use for looking up file ids
438
:param path_encoding: If set, the path will be encoded as specified,
439
otherwise is supposed to be utf8
440
:param format_cls: Formatter class (DiffTree subclass)
442
if format_cls is None:
443
format_cls = DiffTree
446
if extra_trees is not None:
447
for tree in extra_trees:
451
differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
453
external_diff_options,
454
old_label, new_label, using)
455
return differ.show_diff(specific_files, extra_trees)
458
if extra_trees is not None:
459
for tree in extra_trees:
465
def _patch_header_date(tree, file_id, path):
466
"""Returns a timestamp suitable for use in a patch header."""
468
mtime = tree.get_file_mtime(file_id, path)
469
except errors.FileTimestampUnavailable:
471
return timestamp.format_patch_date(mtime)
474
def get_executable_change(old_is_x, new_is_x):
475
descr = { True:"+x", False:"-x", None:"??" }
476
if old_is_x != new_is_x:
477
return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
482
class DiffPath(object):
483
"""Base type for command object that compare files"""
485
# The type or contents of the file were unsuitable for diffing
486
CANNOT_DIFF = 'CANNOT_DIFF'
487
# The file has changed in a semantic way
489
# The file content may have changed, but there is no semantic change
490
UNCHANGED = 'UNCHANGED'
492
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
495
:param old_tree: The tree to show as the old tree in the comparison
496
:param new_tree: The tree to show as new in the comparison
497
:param to_file: The file to write comparison data to
498
:param path_encoding: The character encoding to write paths in
500
self.old_tree = old_tree
501
self.new_tree = new_tree
502
self.to_file = to_file
503
self.path_encoding = path_encoding
509
def from_diff_tree(klass, diff_tree):
510
return klass(diff_tree.old_tree, diff_tree.new_tree,
511
diff_tree.to_file, diff_tree.path_encoding)
514
def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
515
for file_differ in differs:
516
result = file_differ.diff(file_id, old_path, new_path, old_kind,
518
if result is not DiffPath.CANNOT_DIFF:
521
return DiffPath.CANNOT_DIFF
524
class DiffKindChange(object):
525
"""Special differ for file kind changes.
527
Represents kind change as deletion + creation. Uses the other differs
530
def __init__(self, differs):
531
self.differs = differs
537
def from_diff_tree(klass, diff_tree):
538
return klass(diff_tree.differs)
540
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
541
"""Perform comparison
543
:param file_id: The file_id of the file to compare
544
:param old_path: Path of the file in the old tree
545
:param new_path: Path of the file in the new tree
546
:param old_kind: Old file-kind of the file
547
:param new_kind: New file-kind of the file
549
if None in (old_kind, new_kind):
550
return DiffPath.CANNOT_DIFF
551
result = DiffPath._diff_many(self.differs, file_id, old_path,
552
new_path, old_kind, None)
553
if result is DiffPath.CANNOT_DIFF:
555
return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
559
class DiffDirectory(DiffPath):
561
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
562
"""Perform comparison between two directories. (dummy)
565
if 'directory' not in (old_kind, new_kind):
566
return self.CANNOT_DIFF
567
if old_kind not in ('directory', None):
568
return self.CANNOT_DIFF
569
if new_kind not in ('directory', None):
570
return self.CANNOT_DIFF
574
class DiffSymlink(DiffPath):
576
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
577
"""Perform comparison between two symlinks
579
:param file_id: The file_id of the file to compare
580
:param old_path: Path of the file in the old tree
581
:param new_path: Path of the file in the new tree
582
:param old_kind: Old file-kind of the file
583
:param new_kind: New file-kind of the file
585
if 'symlink' not in (old_kind, new_kind):
586
return self.CANNOT_DIFF
587
if old_kind == 'symlink':
588
old_target = self.old_tree.get_symlink_target(file_id)
589
elif old_kind is None:
592
return self.CANNOT_DIFF
593
if new_kind == 'symlink':
594
new_target = self.new_tree.get_symlink_target(file_id)
595
elif new_kind is None:
598
return self.CANNOT_DIFF
599
return self.diff_symlink(old_target, new_target)
601
def diff_symlink(self, old_target, new_target):
602
if old_target is None:
603
self.to_file.write('=== target is %r\n' % new_target)
604
elif new_target is None:
605
self.to_file.write('=== target was %r\n' % old_target)
607
self.to_file.write('=== target changed %r => %r\n' %
608
(old_target, new_target))
612
class DiffText(DiffPath):
614
# GNU Patch uses the epoch date to detect files that are being added
615
# or removed in a diff.
616
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
618
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
619
old_label='', new_label='', text_differ=internal_diff):
620
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
621
self.text_differ = text_differ
622
self.old_label = old_label
623
self.new_label = new_label
624
self.path_encoding = path_encoding
626
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
627
"""Compare two files in unified diff format
629
:param file_id: The file_id of the file to compare
630
:param old_path: Path of the file in the old tree
631
:param new_path: Path of the file in the new tree
632
:param old_kind: Old file-kind of the file
633
:param new_kind: New file-kind of the file
635
if 'file' not in (old_kind, new_kind):
636
return self.CANNOT_DIFF
637
from_file_id = to_file_id = file_id
638
if old_kind == 'file':
639
old_date = _patch_header_date(self.old_tree, file_id, old_path)
640
elif old_kind is None:
641
old_date = self.EPOCH_DATE
644
return self.CANNOT_DIFF
645
if new_kind == 'file':
646
new_date = _patch_header_date(self.new_tree, file_id, new_path)
647
elif new_kind is None:
648
new_date = self.EPOCH_DATE
651
return self.CANNOT_DIFF
652
from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
653
to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
654
return self.diff_text(from_file_id, to_file_id, from_label, to_label,
657
def diff_text(self, from_file_id, to_file_id, from_label, to_label,
658
from_path=None, to_path=None):
659
"""Diff the content of given files in two trees
661
:param from_file_id: The id of the file in the from tree. If None,
662
the file is not present in the from tree.
663
:param to_file_id: The id of the file in the to tree. This may refer
664
to a different file from from_file_id. If None,
665
the file is not present in the to tree.
666
:param from_path: The path in the from tree or None if unknown.
667
:param to_path: The path in the to tree or None if unknown.
669
def _get_text(tree, file_id, path):
670
if file_id is not None:
671
return tree.get_file_lines(file_id, path)
675
from_text = _get_text(self.old_tree, from_file_id, from_path)
676
to_text = _get_text(self.new_tree, to_file_id, to_path)
677
self.text_differ(from_label, from_text, to_label, to_text,
678
self.to_file, path_encoding=self.path_encoding)
679
except errors.BinaryFile:
681
("Binary files %s and %s differ\n" %
682
(from_label, to_label)).encode(self.path_encoding,'replace'))
686
class DiffFromTool(DiffPath):
688
def __init__(self, command_template, old_tree, new_tree, to_file,
689
path_encoding='utf-8'):
690
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
691
self.command_template = command_template
692
self._root = osutils.mkdtemp(prefix='bzr-diff-')
695
def from_string(klass, command_string, old_tree, new_tree, to_file,
696
path_encoding='utf-8'):
697
command_template = cmdline.split(command_string)
698
if '@' not in command_string:
699
command_template.extend(['@old_path', '@new_path'])
700
return klass(command_template, old_tree, new_tree, to_file,
704
def make_from_diff_tree(klass, command_string, external_diff_options=None):
705
def from_diff_tree(diff_tree):
706
full_command_string = [command_string]
707
if external_diff_options is not None:
708
full_command_string += ' ' + external_diff_options
709
return klass.from_string(full_command_string, diff_tree.old_tree,
710
diff_tree.new_tree, diff_tree.to_file)
711
return from_diff_tree
713
def _get_command(self, old_path, new_path):
714
my_map = {'old_path': old_path, 'new_path': new_path}
715
command = [AtTemplate(t).substitute(my_map) for t in
716
self.command_template]
717
if sys.platform == 'win32': # Popen doesn't accept unicode on win32
720
if isinstance(c, unicode):
721
command_encoded.append(c.encode('mbcs'))
723
command_encoded.append(c)
724
return command_encoded
728
def _execute(self, old_path, new_path):
729
command = self._get_command(old_path, new_path)
731
proc = subprocess.Popen(command, stdout=subprocess.PIPE,
734
if e.errno == errno.ENOENT:
735
raise errors.ExecutableMissing(command[0])
738
self.to_file.write(proc.stdout.read())
741
def _try_symlink_root(self, tree, prefix):
742
if (getattr(tree, 'abspath', None) is None
743
or not osutils.host_os_dereferences_symlinks()):
746
os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
748
if e.errno != errno.EEXIST:
754
"""Returns safe encoding for passing file path to diff tool"""
755
if sys.platform == 'win32':
758
# Don't fallback to 'utf-8' because subprocess may not be able to
759
# handle utf-8 correctly when locale is not utf-8.
760
return sys.getfilesystemencoding() or 'ascii'
762
def _is_safepath(self, path):
763
"""Return true if `path` may be able to pass to subprocess."""
766
return path == path.encode(fenc).decode(fenc)
770
def _safe_filename(self, prefix, relpath):
771
"""Replace unsafe character in `relpath` then join `self._root`,
772
`prefix` and `relpath`."""
774
# encoded_str.replace('?', '_') may break multibyte char.
775
# So we should encode, decode, then replace(u'?', u'_')
776
relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
777
relpath_tmp = relpath_tmp.replace(u'?', u'_')
778
return osutils.pathjoin(self._root, prefix, relpath_tmp)
780
def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
782
if not force_temp and isinstance(tree, WorkingTree):
783
full_path = tree.abspath(tree.id2path(file_id))
784
if self._is_safepath(full_path):
787
full_path = self._safe_filename(prefix, relpath)
788
if not force_temp and self._try_symlink_root(tree, prefix):
790
parent_dir = osutils.dirname(full_path)
792
os.makedirs(parent_dir)
794
if e.errno != errno.EEXIST:
796
source = tree.get_file(file_id, relpath)
798
target = open(full_path, 'wb')
800
osutils.pumpfile(source, target)
806
mtime = tree.get_file_mtime(file_id)
807
except errors.FileTimestampUnavailable:
810
os.utime(full_path, (mtime, mtime))
812
osutils.make_readonly(full_path)
815
def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
816
allow_write_new=False):
817
old_disk_path = self._write_file(file_id, self.old_tree, 'old',
818
old_path, force_temp)
819
new_disk_path = self._write_file(file_id, self.new_tree, 'new',
820
new_path, force_temp,
821
allow_write=allow_write_new)
822
return old_disk_path, new_disk_path
826
osutils.rmtree(self._root)
828
if e.errno != errno.ENOENT:
829
mutter("The temporary directory \"%s\" was not "
830
"cleanly removed: %s." % (self._root, e))
832
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
833
if (old_kind, new_kind) != ('file', 'file'):
834
return DiffPath.CANNOT_DIFF
835
(old_disk_path, new_disk_path) = self._prepare_files(
836
file_id, old_path, new_path)
837
self._execute(old_disk_path, new_disk_path)
839
def edit_file(self, file_id):
840
"""Use this tool to edit a file.
842
A temporary copy will be edited, and the new contents will be
845
:param file_id: The id of the file to edit.
846
:return: The new contents of the file.
848
old_path = self.old_tree.id2path(file_id)
849
new_path = self.new_tree.id2path(file_id)
850
old_abs_path, new_abs_path = self._prepare_files(
851
file_id, old_path, new_path,
852
allow_write_new=True,
854
command = self._get_command(old_abs_path, new_abs_path)
855
subprocess.call(command, cwd=self._root)
856
new_file = open(new_abs_path, 'rb')
858
return new_file.read()
863
class DiffTree(object):
864
"""Provides textual representations of the difference between two trees.
866
A DiffTree examines two trees and where a file-id has altered
867
between them, generates a textual representation of the difference.
868
DiffTree uses a sequence of DiffPath objects which are each
869
given the opportunity to handle a given altered fileid. The list
870
of DiffPath objects can be extended globally by appending to
871
DiffTree.diff_factories, or for a specific diff operation by
872
supplying the extra_factories option to the appropriate method.
875
# list of factories that can provide instances of DiffPath objects
876
# may be extended by plugins.
877
diff_factories = [DiffSymlink.from_diff_tree,
878
DiffDirectory.from_diff_tree]
880
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
881
diff_text=None, extra_factories=None):
884
:param old_tree: Tree to show as old in the comparison
885
:param new_tree: Tree to show as new in the comparison
886
:param to_file: File to write comparision to
887
:param path_encoding: Character encoding to write paths in
888
:param diff_text: DiffPath-type object to use as a last resort for
890
:param extra_factories: Factories of DiffPaths to try before any other
892
if diff_text is None:
893
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
894
'', '', internal_diff)
895
self.old_tree = old_tree
896
self.new_tree = new_tree
897
self.to_file = to_file
898
self.path_encoding = path_encoding
900
if extra_factories is not None:
901
self.differs.extend(f(self) for f in extra_factories)
902
self.differs.extend(f(self) for f in self.diff_factories)
903
self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
906
def from_trees_options(klass, old_tree, new_tree, to_file,
907
path_encoding, external_diff_options, old_label,
909
"""Factory for producing a DiffTree.
911
Designed to accept options used by show_diff_trees.
913
:param old_tree: The tree to show as old in the comparison
914
:param new_tree: The tree to show as new in the comparison
915
:param to_file: File to write comparisons to
916
:param path_encoding: Character encoding to use for writing paths
917
:param external_diff_options: If supplied, use the installed diff
918
binary to perform file comparison, using supplied options.
919
:param old_label: Prefix to use for old file labels
920
:param new_label: Prefix to use for new file labels
921
:param using: Commandline to use to invoke an external diff tool
923
if using is not None:
924
extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
927
if external_diff_options:
928
opts = external_diff_options.split()
929
def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None):
930
""":param path_encoding: not used but required
931
to match the signature of internal_diff.
933
external_diff(olab, olines, nlab, nlines, to_file, opts)
935
diff_file = internal_diff
936
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
937
old_label, new_label, diff_file)
938
return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
941
def show_diff(self, specific_files, extra_trees=None):
942
"""Write tree diff to self.to_file
944
:param specific_files: the specific files to compare (recursive)
945
:param extra_trees: extra trees to use for mapping paths to file_ids
948
return self._show_diff(specific_files, extra_trees)
950
for differ in self.differs:
953
def _show_diff(self, specific_files, extra_trees):
954
# TODO: Generation of pseudo-diffs for added/deleted files could
955
# be usefully made into a much faster special case.
956
iterator = self.new_tree.iter_changes(self.old_tree,
957
specific_files=specific_files,
958
extra_trees=extra_trees,
959
require_versioned=True)
961
def changes_key(change):
962
old_path, new_path = change[1]
967
def get_encoded_path(path):
969
return path.encode(self.path_encoding, "replace")
970
for (file_id, paths, changed_content, versioned, parent, name, kind,
971
executable) in sorted(iterator, key=changes_key):
972
# The root does not get diffed, and items with no known kind (that
973
# is, missing) in both trees are skipped as well.
974
if parent == (None, None) or kind == (None, None):
976
oldpath, newpath = paths
977
oldpath_encoded = get_encoded_path(paths[0])
978
newpath_encoded = get_encoded_path(paths[1])
979
old_present = (kind[0] is not None and versioned[0])
980
new_present = (kind[1] is not None and versioned[1])
981
renamed = (parent[0], name[0]) != (parent[1], name[1])
983
properties_changed = []
984
properties_changed.extend(get_executable_change(executable[0], executable[1]))
986
if properties_changed:
987
prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
991
if (old_present, new_present) == (True, False):
992
self.to_file.write("=== removed %s '%s'\n" %
993
(kind[0], oldpath_encoded))
995
elif (old_present, new_present) == (False, True):
996
self.to_file.write("=== added %s '%s'\n" %
997
(kind[1], newpath_encoded))
1000
self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
1001
(kind[0], oldpath_encoded, newpath_encoded, prop_str))
1003
# if it was produced by iter_changes, it must be
1004
# modified *somehow*, either content or execute bit.
1005
self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
1006
newpath_encoded, prop_str))
1008
self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1014
def diff(self, file_id, old_path, new_path):
1015
"""Perform a diff of a single file
1017
:param file_id: file-id of the file
1018
:param old_path: The path of the file in the old tree
1019
:param new_path: The path of the file in the new tree
1022
old_kind = self.old_tree.kind(file_id)
1023
except (errors.NoSuchId, errors.NoSuchFile):
1026
new_kind = self.new_tree.kind(file_id)
1027
except (errors.NoSuchId, errors.NoSuchFile):
1029
self._diff(file_id, old_path, new_path, old_kind, new_kind)
1032
def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
1033
result = DiffPath._diff_many(self.differs, file_id, old_path,
1034
new_path, old_kind, new_kind)
1035
if result is DiffPath.CANNOT_DIFF:
1036
error_path = new_path
1037
if error_path is None:
1038
error_path = old_path
1039
raise errors.NoDiffFound(error_path)
1042
format_registry = Registry()
1043
format_registry.register('default', DiffTree)
added
removed
bzrlib/diff.py
