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
44
from bzrlib.registry import (
47
from bzrlib.symbol_versioning import (
51
from bzrlib.trace import mutter, note, warning
54
class AtTemplate(string.Template):
55
"""Templating class that uses @ instead of $."""
60
# TODO: Rather than building a changeset object, we should probably
61
# invoke callbacks on an object. That object can either accumulate a
62
# list, write them out directly, etc etc.
65
class _PrematchedMatcher(difflib.SequenceMatcher):
66
"""Allow SequenceMatcher operations to use predetermined blocks"""
68
def __init__(self, matching_blocks):
69
difflib.SequenceMatcher(self, None, None)
70
self.matching_blocks = matching_blocks
74
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
75
allow_binary=False, sequence_matcher=None,
76
path_encoding='utf8'):
77
# FIXME: difflib is wrong if there is no trailing newline.
78
# The syntax used by patch seems to be "\ No newline at
79
# end of file" following the last diff line from that
80
# file. This is not trivial to insert into the
81
# unified_diff output and it might be better to just fix
82
# or replace that function.
84
# In the meantime we at least make sure the patch isn't
88
# Special workaround for Python2.3, where difflib fails if
89
# both sequences are empty.
90
if not oldlines and not newlines:
93
if allow_binary is False:
94
textfile.check_text_lines(oldlines)
95
textfile.check_text_lines(newlines)
97
if sequence_matcher is None:
98
sequence_matcher = patiencediff.PatienceSequenceMatcher
99
ud = patiencediff.unified_diff(oldlines, newlines,
100
fromfile=old_filename.encode(path_encoding, 'replace'),
101
tofile=new_filename.encode(path_encoding, 'replace'),
102
sequencematcher=sequence_matcher)
105
if len(ud) == 0: # Identical contents, nothing to do
107
# work-around for difflib being too smart for its own good
108
# if /dev/null is "1,0", patch won't recognize it as /dev/null
110
ud[2] = ud[2].replace('-1,0', '-0,0')
112
ud[2] = ud[2].replace('+1,0', '+0,0')
116
if not line.endswith('\n'):
117
to_file.write("\n\\ No newline at end of file\n")
121
def _spawn_external_diff(diffcmd, capture_errors=True):
122
"""Spawn the externall diff process, and return the child handle.
124
:param diffcmd: The command list to spawn
125
:param capture_errors: Capture stderr as well as setting LANG=C
126
and LC_ALL=C. This lets us read and understand the output of diff,
127
and respond to any errors.
128
:return: A Popen object.
131
# construct minimal environment
133
path = os.environ.get('PATH')
136
env['LANGUAGE'] = 'C' # on win32 only LANGUAGE has effect
139
stderr = subprocess.PIPE
145
pipe = subprocess.Popen(diffcmd,
146
stdin=subprocess.PIPE,
147
stdout=subprocess.PIPE,
151
if e.errno == errno.ENOENT:
152
raise errors.NoDiff(str(e))
158
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
160
"""Display a diff by calling out to the external diff program."""
161
# make sure our own output is properly ordered before the diff
164
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
165
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
166
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
167
newtmpf = os.fdopen(newtmp_fd, 'wb')
170
# TODO: perhaps a special case for comparing to or from the empty
171
# sequence; can just use /dev/null on Unix
173
# TODO: if either of the files being compared already exists as a
174
# regular named file (e.g. in the working directory) then we can
175
# compare directly to that, rather than copying it.
177
oldtmpf.writelines(oldlines)
178
newtmpf.writelines(newlines)
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')
191
'--label', old_filename,
193
'--label', new_filename,
198
# diff only allows one style to be specified; they don't override.
199
# note that some of these take optargs, and the optargs can be
200
# directly appended to the options.
201
# this is only an approximate parser; it doesn't properly understand
203
for s in ['-c', '-u', '-C', '-U',
208
'-y', '--side-by-side',
220
diffcmd.extend(diff_opts)
222
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
223
out,err = pipe.communicate()
226
# internal_diff() adds a trailing newline, add one here for consistency
229
# 'diff' gives retcode == 2 for all sorts of errors
230
# one of those is 'Binary files differ'.
231
# Bad options could also be the problem.
232
# 'Binary files' is not a real error, so we suppress that error.
235
# Since we got here, we want to make sure to give an i18n error
236
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
237
out, err = pipe.communicate()
239
# Write out the new i18n diff response
240
to_file.write(out+'\n')
241
if pipe.returncode != 2:
242
raise errors.BzrError(
243
'external diff failed with exit code 2'
244
' when run with LANG=C and LC_ALL=C,'
245
' but not when run natively: %r' % (diffcmd,))
247
first_line = lang_c_out.split('\n', 1)[0]
248
# Starting with diffutils 2.8.4 the word "binary" was dropped.
249
m = re.match('^(binary )?files.*differ$', first_line, re.I)
251
raise errors.BzrError('external diff failed with exit code 2;'
252
' command: %r' % (diffcmd,))
254
# Binary files differ, just return
257
# If we got to here, we haven't written out the output of diff
261
# returns 1 if files differ; that's OK
263
msg = 'signal %d' % (-rc)
265
msg = 'exit code %d' % rc
267
raise errors.BzrError('external diff failed with %s; command: %r'
272
oldtmpf.close() # and delete
274
# Clean up. Warn in case the files couldn't be deleted
275
# (in case windows still holds the file open, but not
276
# if the files have already been deleted)
278
os.remove(old_abspath)
280
if e.errno not in (errno.ENOENT,):
281
warning('Failed to delete temporary file: %s %s',
284
os.remove(new_abspath)
286
if e.errno not in (errno.ENOENT,):
287
warning('Failed to delete temporary file: %s %s',
291
@deprecated_function(deprecated_in((2, 2, 0)))
292
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
294
"""Get the trees and specific files to diff given a list of paths.
296
This method works out the trees to be diff'ed and the files of
297
interest within those trees.
300
the list of arguments passed to the diff command
301
:param revision_specs:
302
Zero, one or two RevisionSpecs from the diff command line,
303
saying what revisions to compare.
305
The url of the old branch or tree. If None, the tree to use is
306
taken from the first path, if any, or the current working tree.
308
The url of the new branch or tree. If None, the tree to use is
309
taken from the first path, if any, or the current working tree.
311
if True and a view is set, apply the view or check that the paths
314
a tuple of (old_tree, new_tree, old_branch, new_branch,
315
specific_files, extra_trees) where extra_trees is a sequence of
316
additional trees to search in for file-ids. The trees and branches
319
op = cleanup.OperationWithCleanups(get_trees_and_branches_to_diff_locked)
320
return op.run_simple(path_list, revision_specs, old_url, new_url,
321
op.add_cleanup, apply_view=apply_view)
324
def get_trees_and_branches_to_diff_locked(
325
path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
326
"""Get the trees and specific files to diff given a list of paths.
328
This method works out the trees to be diff'ed and the files of
329
interest within those trees.
332
the list of arguments passed to the diff command
333
:param revision_specs:
334
Zero, one or two RevisionSpecs from the diff command line,
335
saying what revisions to compare.
337
The url of the old branch or tree. If None, the tree to use is
338
taken from the first path, if any, or the current working tree.
340
The url of the new branch or tree. If None, the tree to use is
341
taken from the first path, if any, or the current working tree.
343
a callable like Command.add_cleanup. get_trees_and_branches_to_diff
344
will register cleanups that must be run to unlock the trees, etc.
346
if True and a view is set, apply the view or check that the paths
349
a tuple of (old_tree, new_tree, old_branch, new_branch,
350
specific_files, extra_trees) where extra_trees is a sequence of
351
additional trees to search in for file-ids. The trees and branches
352
will be read-locked until the cleanups registered via the add_cleanup
355
# Get the old and new revision specs
356
old_revision_spec = None
357
new_revision_spec = None
358
if revision_specs is not None:
359
if len(revision_specs) > 0:
360
old_revision_spec = revision_specs[0]
362
old_url = old_revision_spec.get_branch()
363
if len(revision_specs) > 1:
364
new_revision_spec = revision_specs[1]
366
new_url = new_revision_spec.get_branch()
369
make_paths_wt_relative = True
370
consider_relpath = True
371
if path_list is None or len(path_list) == 0:
372
# If no path is given, the current working tree is used
373
default_location = u'.'
374
consider_relpath = False
375
elif old_url is not None and new_url is not None:
376
other_paths = path_list
377
make_paths_wt_relative = False
379
default_location = path_list[0]
380
other_paths = path_list[1:]
382
def lock_tree_or_branch(wt, br):
385
add_cleanup(wt.unlock)
388
add_cleanup(br.unlock)
390
# Get the old location
393
old_url = default_location
394
working_tree, branch, relpath = \
395
bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
396
lock_tree_or_branch(working_tree, branch)
397
if consider_relpath and relpath != '':
398
if working_tree is not None and apply_view:
399
views.check_path_in_view(working_tree, relpath)
400
specific_files.append(relpath)
401
old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
404
# Get the new location
406
new_url = default_location
407
if new_url != old_url:
408
working_tree, branch, relpath = \
409
bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
410
lock_tree_or_branch(working_tree, branch)
411
if consider_relpath and relpath != '':
412
if working_tree is not None and apply_view:
413
views.check_path_in_view(working_tree, relpath)
414
specific_files.append(relpath)
415
new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
416
basis_is_default=working_tree is None)
419
# Get the specific files (all files is None, no files is [])
420
if make_paths_wt_relative and working_tree is not None:
421
other_paths = working_tree.safe_relpath_files(
423
apply_view=apply_view)
424
specific_files.extend(other_paths)
425
if len(specific_files) == 0:
426
specific_files = None
427
if (working_tree is not None and working_tree.supports_views()
429
view_files = working_tree.views.lookup_view()
431
specific_files = view_files
432
view_str = views.view_display_str(view_files)
433
note("*** Ignoring files outside view. View is %s" % view_str)
435
# Get extra trees that ought to be searched for file-ids
437
if working_tree is not None and working_tree not in (old_tree, new_tree):
438
extra_trees = (working_tree,)
439
return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
442
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
443
if branch is None and tree is not None:
445
if spec is None or spec.spec is None:
448
return tree.basis_tree()
450
return branch.basis_tree()
453
return spec.as_tree(branch)
456
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
457
external_diff_options=None,
458
old_label='a/', new_label='b/',
460
path_encoding='utf8',
463
"""Show in text form the changes from one tree to another.
465
:param to_file: The output stream.
466
:param specific_files:Include only changes to these files - None for all
468
:param external_diff_options: If set, use an external GNU diff and pass
470
:param extra_trees: If set, more Trees to use for looking up file ids
471
:param path_encoding: If set, the path will be encoded as specified,
472
otherwise is supposed to be utf8
473
:param format_cls: Formatter class (DiffTree subclass)
475
if format_cls is None:
476
format_cls = DiffTree
479
if extra_trees is not None:
480
for tree in extra_trees:
484
differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
486
external_diff_options,
487
old_label, new_label, using)
488
return differ.show_diff(specific_files, extra_trees)
491
if extra_trees is not None:
492
for tree in extra_trees:
498
def _patch_header_date(tree, file_id, path):
499
"""Returns a timestamp suitable for use in a patch header."""
501
mtime = tree.get_file_mtime(file_id, path)
502
except errors.FileTimestampUnavailable:
504
return timestamp.format_patch_date(mtime)
507
def get_executable_change(old_is_x, new_is_x):
508
descr = { True:"+x", False:"-x", None:"??" }
509
if old_is_x != new_is_x:
510
return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
515
class DiffPath(object):
516
"""Base type for command object that compare files"""
518
# The type or contents of the file were unsuitable for diffing
519
CANNOT_DIFF = 'CANNOT_DIFF'
520
# The file has changed in a semantic way
522
# The file content may have changed, but there is no semantic change
523
UNCHANGED = 'UNCHANGED'
525
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
528
:param old_tree: The tree to show as the old tree in the comparison
529
:param new_tree: The tree to show as new in the comparison
530
:param to_file: The file to write comparison data to
531
:param path_encoding: The character encoding to write paths in
533
self.old_tree = old_tree
534
self.new_tree = new_tree
535
self.to_file = to_file
536
self.path_encoding = path_encoding
542
def from_diff_tree(klass, diff_tree):
543
return klass(diff_tree.old_tree, diff_tree.new_tree,
544
diff_tree.to_file, diff_tree.path_encoding)
547
def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
548
for file_differ in differs:
549
result = file_differ.diff(file_id, old_path, new_path, old_kind,
551
if result is not DiffPath.CANNOT_DIFF:
554
return DiffPath.CANNOT_DIFF
557
class DiffKindChange(object):
558
"""Special differ for file kind changes.
560
Represents kind change as deletion + creation. Uses the other differs
563
def __init__(self, differs):
564
self.differs = differs
570
def from_diff_tree(klass, diff_tree):
571
return klass(diff_tree.differs)
573
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
574
"""Perform comparison
576
:param file_id: The file_id of the file to compare
577
:param old_path: Path of the file in the old tree
578
:param new_path: Path of the file in the new tree
579
:param old_kind: Old file-kind of the file
580
:param new_kind: New file-kind of the file
582
if None in (old_kind, new_kind):
583
return DiffPath.CANNOT_DIFF
584
result = DiffPath._diff_many(self.differs, file_id, old_path,
585
new_path, old_kind, None)
586
if result is DiffPath.CANNOT_DIFF:
588
return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
592
class DiffDirectory(DiffPath):
594
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
595
"""Perform comparison between two directories. (dummy)
598
if 'directory' not in (old_kind, new_kind):
599
return self.CANNOT_DIFF
600
if old_kind not in ('directory', None):
601
return self.CANNOT_DIFF
602
if new_kind not in ('directory', None):
603
return self.CANNOT_DIFF
607
class DiffSymlink(DiffPath):
609
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
610
"""Perform comparison between two symlinks
612
:param file_id: The file_id of the file to compare
613
:param old_path: Path of the file in the old tree
614
:param new_path: Path of the file in the new tree
615
:param old_kind: Old file-kind of the file
616
:param new_kind: New file-kind of the file
618
if 'symlink' not in (old_kind, new_kind):
619
return self.CANNOT_DIFF
620
if old_kind == 'symlink':
621
old_target = self.old_tree.get_symlink_target(file_id)
622
elif old_kind is None:
625
return self.CANNOT_DIFF
626
if new_kind == 'symlink':
627
new_target = self.new_tree.get_symlink_target(file_id)
628
elif new_kind is None:
631
return self.CANNOT_DIFF
632
return self.diff_symlink(old_target, new_target)
634
def diff_symlink(self, old_target, new_target):
635
if old_target is None:
636
self.to_file.write('=== target is %r\n' % new_target)
637
elif new_target is None:
638
self.to_file.write('=== target was %r\n' % old_target)
640
self.to_file.write('=== target changed %r => %r\n' %
641
(old_target, new_target))
645
class DiffText(DiffPath):
647
# GNU Patch uses the epoch date to detect files that are being added
648
# or removed in a diff.
649
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
651
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
652
old_label='', new_label='', text_differ=internal_diff):
653
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
654
self.text_differ = text_differ
655
self.old_label = old_label
656
self.new_label = new_label
657
self.path_encoding = path_encoding
659
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
660
"""Compare two files in unified diff format
662
:param file_id: The file_id of the file to compare
663
:param old_path: Path of the file in the old tree
664
:param new_path: Path of the file in the new tree
665
:param old_kind: Old file-kind of the file
666
:param new_kind: New file-kind of the file
668
if 'file' not in (old_kind, new_kind):
669
return self.CANNOT_DIFF
670
from_file_id = to_file_id = file_id
671
if old_kind == 'file':
672
old_date = _patch_header_date(self.old_tree, file_id, old_path)
673
elif old_kind is None:
674
old_date = self.EPOCH_DATE
677
return self.CANNOT_DIFF
678
if new_kind == 'file':
679
new_date = _patch_header_date(self.new_tree, file_id, new_path)
680
elif new_kind is None:
681
new_date = self.EPOCH_DATE
684
return self.CANNOT_DIFF
685
from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
686
to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
687
return self.diff_text(from_file_id, to_file_id, from_label, to_label,
690
def diff_text(self, from_file_id, to_file_id, from_label, to_label,
691
from_path=None, to_path=None):
692
"""Diff the content of given files in two trees
694
:param from_file_id: The id of the file in the from tree. If None,
695
the file is not present in the from tree.
696
:param to_file_id: The id of the file in the to tree. This may refer
697
to a different file from from_file_id. If None,
698
the file is not present in the to tree.
699
:param from_path: The path in the from tree or None if unknown.
700
:param to_path: The path in the to tree or None if unknown.
702
def _get_text(tree, file_id, path):
703
if file_id is not None:
704
return tree.get_file_lines(file_id, path)
708
from_text = _get_text(self.old_tree, from_file_id, from_path)
709
to_text = _get_text(self.new_tree, to_file_id, to_path)
710
self.text_differ(from_label, from_text, to_label, to_text,
711
self.to_file, path_encoding=self.path_encoding)
712
except errors.BinaryFile:
714
("Binary files %s and %s differ\n" %
715
(from_label, to_label)).encode(self.path_encoding,'replace'))
719
class DiffFromTool(DiffPath):
721
def __init__(self, command_template, old_tree, new_tree, to_file,
722
path_encoding='utf-8'):
723
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
724
self.command_template = command_template
725
self._root = osutils.mkdtemp(prefix='bzr-diff-')
728
def from_string(klass, command_string, old_tree, new_tree, to_file,
729
path_encoding='utf-8'):
730
command_template = cmdline.split(command_string)
731
if '@' not in command_string:
732
command_template.extend(['@old_path', '@new_path'])
733
return klass(command_template, old_tree, new_tree, to_file,
737
def make_from_diff_tree(klass, command_string, external_diff_options=None):
738
def from_diff_tree(diff_tree):
739
full_command_string = [command_string]
740
if external_diff_options is not None:
741
full_command_string += ' ' + external_diff_options
742
return klass.from_string(full_command_string, diff_tree.old_tree,
743
diff_tree.new_tree, diff_tree.to_file)
744
return from_diff_tree
746
def _get_command(self, old_path, new_path):
747
my_map = {'old_path': old_path, 'new_path': new_path}
748
return [AtTemplate(t).substitute(my_map) for t in
749
self.command_template]
751
def _execute(self, old_path, new_path):
752
command = self._get_command(old_path, new_path)
754
proc = subprocess.Popen(command, stdout=subprocess.PIPE,
757
if e.errno == errno.ENOENT:
758
raise errors.ExecutableMissing(command[0])
761
self.to_file.write(proc.stdout.read())
764
def _try_symlink_root(self, tree, prefix):
765
if (getattr(tree, 'abspath', None) is None
766
or not osutils.host_os_dereferences_symlinks()):
769
os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
771
if e.errno != errno.EEXIST:
775
def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
777
if not force_temp and isinstance(tree, WorkingTree):
778
return tree.abspath(tree.id2path(file_id))
780
full_path = osutils.pathjoin(self._root, prefix, relpath)
781
if not force_temp and self._try_symlink_root(tree, prefix):
783
parent_dir = osutils.dirname(full_path)
785
os.makedirs(parent_dir)
787
if e.errno != errno.EEXIST:
789
source = tree.get_file(file_id, relpath)
791
target = open(full_path, 'wb')
793
osutils.pumpfile(source, target)
799
mtime = tree.get_file_mtime(file_id)
800
except errors.FileTimestampUnavailable:
803
os.utime(full_path, (mtime, mtime))
805
osutils.make_readonly(full_path)
808
def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
809
allow_write_new=False):
810
old_disk_path = self._write_file(file_id, self.old_tree, 'old',
811
old_path, force_temp)
812
new_disk_path = self._write_file(file_id, self.new_tree, 'new',
813
new_path, force_temp,
814
allow_write=allow_write_new)
815
return old_disk_path, new_disk_path
819
osutils.rmtree(self._root)
821
if e.errno != errno.ENOENT:
822
mutter("The temporary directory \"%s\" was not "
823
"cleanly removed: %s." % (self._root, e))
825
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
826
if (old_kind, new_kind) != ('file', 'file'):
827
return DiffPath.CANNOT_DIFF
828
(old_disk_path, new_disk_path) = self._prepare_files(
829
file_id, old_path, new_path)
830
self._execute(old_disk_path, new_disk_path)
832
def edit_file(self, file_id):
833
"""Use this tool to edit a file.
835
A temporary copy will be edited, and the new contents will be
838
:param file_id: The id of the file to edit.
839
:return: The new contents of the file.
841
old_path = self.old_tree.id2path(file_id)
842
new_path = self.new_tree.id2path(file_id)
843
new_abs_path = self._prepare_files(file_id, old_path, new_path,
844
allow_write_new=True,
846
command = self._get_command(osutils.pathjoin('old', old_path),
847
osutils.pathjoin('new', new_path))
848
subprocess.call(command, cwd=self._root)
849
new_file = open(new_abs_path, 'r')
851
return new_file.read()
856
class DiffTree(object):
857
"""Provides textual representations of the difference between two trees.
859
A DiffTree examines two trees and where a file-id has altered
860
between them, generates a textual representation of the difference.
861
DiffTree uses a sequence of DiffPath objects which are each
862
given the opportunity to handle a given altered fileid. The list
863
of DiffPath objects can be extended globally by appending to
864
DiffTree.diff_factories, or for a specific diff operation by
865
supplying the extra_factories option to the appropriate method.
868
# list of factories that can provide instances of DiffPath objects
869
# may be extended by plugins.
870
diff_factories = [DiffSymlink.from_diff_tree,
871
DiffDirectory.from_diff_tree]
873
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
874
diff_text=None, extra_factories=None):
877
:param old_tree: Tree to show as old in the comparison
878
:param new_tree: Tree to show as new in the comparison
879
:param to_file: File to write comparision to
880
:param path_encoding: Character encoding to write paths in
881
:param diff_text: DiffPath-type object to use as a last resort for
883
:param extra_factories: Factories of DiffPaths to try before any other
885
if diff_text is None:
886
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
887
'', '', internal_diff)
888
self.old_tree = old_tree
889
self.new_tree = new_tree
890
self.to_file = to_file
891
self.path_encoding = path_encoding
893
if extra_factories is not None:
894
self.differs.extend(f(self) for f in extra_factories)
895
self.differs.extend(f(self) for f in self.diff_factories)
896
self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
899
def from_trees_options(klass, old_tree, new_tree, to_file,
900
path_encoding, external_diff_options, old_label,
902
"""Factory for producing a DiffTree.
904
Designed to accept options used by show_diff_trees.
905
:param old_tree: The tree to show as old in the comparison
906
:param new_tree: The tree to show as new in the comparison
907
:param to_file: File to write comparisons to
908
:param path_encoding: Character encoding to use for writing paths
909
:param external_diff_options: If supplied, use the installed diff
910
binary to perform file comparison, using supplied options.
911
:param old_label: Prefix to use for old file labels
912
:param new_label: Prefix to use for new file labels
913
:param using: Commandline to use to invoke an external diff tool
915
if using is not None:
916
extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
919
if external_diff_options:
920
opts = external_diff_options.split()
921
def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None):
922
""":param path_encoding: not used but required
923
to match the signature of internal_diff.
925
external_diff(olab, olines, nlab, nlines, to_file, opts)
927
diff_file = internal_diff
928
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
929
old_label, new_label, diff_file)
930
return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
933
def show_diff(self, specific_files, extra_trees=None):
934
"""Write tree diff to self.to_file
936
:param specific_files: the specific files to compare (recursive)
937
:param extra_trees: extra trees to use for mapping paths to file_ids
940
return self._show_diff(specific_files, extra_trees)
942
for differ in self.differs:
945
def _show_diff(self, specific_files, extra_trees):
946
# TODO: Generation of pseudo-diffs for added/deleted files could
947
# be usefully made into a much faster special case.
948
iterator = self.new_tree.iter_changes(self.old_tree,
949
specific_files=specific_files,
950
extra_trees=extra_trees,
951
require_versioned=True)
953
def changes_key(change):
954
old_path, new_path = change[1]
959
def get_encoded_path(path):
961
return path.encode(self.path_encoding, "replace")
962
for (file_id, paths, changed_content, versioned, parent, name, kind,
963
executable) in sorted(iterator, key=changes_key):
964
# The root does not get diffed, and items with no known kind (that
965
# is, missing) in both trees are skipped as well.
966
if parent == (None, None) or kind == (None, None):
968
oldpath, newpath = paths
969
oldpath_encoded = get_encoded_path(paths[0])
970
newpath_encoded = get_encoded_path(paths[1])
971
old_present = (kind[0] is not None and versioned[0])
972
new_present = (kind[1] is not None and versioned[1])
973
renamed = (parent[0], name[0]) != (parent[1], name[1])
975
properties_changed = []
976
properties_changed.extend(get_executable_change(executable[0], executable[1]))
978
if properties_changed:
979
prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
983
if (old_present, new_present) == (True, False):
984
self.to_file.write("=== removed %s '%s'\n" %
985
(kind[0], oldpath_encoded))
987
elif (old_present, new_present) == (False, True):
988
self.to_file.write("=== added %s '%s'\n" %
989
(kind[1], newpath_encoded))
992
self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
993
(kind[0], oldpath_encoded, newpath_encoded, prop_str))
995
# if it was produced by iter_changes, it must be
996
# modified *somehow*, either content or execute bit.
997
self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
998
newpath_encoded, prop_str))
1000
self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1006
def diff(self, file_id, old_path, new_path):
1007
"""Perform a diff of a single file
1009
:param file_id: file-id of the file
1010
:param old_path: The path of the file in the old tree
1011
:param new_path: The path of the file in the new tree
1014
old_kind = self.old_tree.kind(file_id)
1015
except (errors.NoSuchId, errors.NoSuchFile):
1018
new_kind = self.new_tree.kind(file_id)
1019
except (errors.NoSuchId, errors.NoSuchFile):
1021
self._diff(file_id, old_path, new_path, old_kind, new_kind)
1024
def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
1025
result = DiffPath._diff_many(self.differs, file_id, old_path,
1026
new_path, old_kind, new_kind)
1027
if result is DiffPath.CANNOT_DIFF:
1028
error_path = new_path
1029
if error_path is None:
1030
error_path = old_path
1031
raise errors.NoDiffFound(error_path)
1034
format_registry = Registry()
1035
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
