1
# Copyright (C) 2005-2010 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
24
from bzrlib.lazy_import import lazy_import
25
lazy_import(globals(), """
32
branch as _mod_branch,
43
from bzrlib.workingtree import WorkingTree
46
from bzrlib.symbol_versioning import (
49
from bzrlib.trace import mutter, note, warning
52
class AtTemplate(string.Template):
53
"""Templating class that uses @ instead of $."""
58
# TODO: Rather than building a changeset object, we should probably
59
# invoke callbacks on an object. That object can either accumulate a
60
# list, write them out directly, etc etc.
63
class _PrematchedMatcher(difflib.SequenceMatcher):
64
"""Allow SequenceMatcher operations to use predetermined blocks"""
66
def __init__(self, matching_blocks):
67
difflib.SequenceMatcher(self, None, None)
68
self.matching_blocks = matching_blocks
72
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
73
allow_binary=False, sequence_matcher=None,
74
path_encoding='utf8'):
75
# FIXME: difflib is wrong if there is no trailing newline.
76
# The syntax used by patch seems to be "\ No newline at
77
# end of file" following the last diff line from that
78
# file. This is not trivial to insert into the
79
# unified_diff output and it might be better to just fix
80
# or replace that function.
82
# In the meantime we at least make sure the patch isn't
86
# Special workaround for Python2.3, where difflib fails if
87
# both sequences are empty.
88
if not oldlines and not newlines:
91
if allow_binary is False:
92
textfile.check_text_lines(oldlines)
93
textfile.check_text_lines(newlines)
95
if sequence_matcher is None:
96
sequence_matcher = patiencediff.PatienceSequenceMatcher
97
ud = patiencediff.unified_diff(oldlines, newlines,
98
fromfile=old_filename.encode(path_encoding),
99
tofile=new_filename.encode(path_encoding),
100
sequencematcher=sequence_matcher)
103
if len(ud) == 0: # Identical contents, nothing to do
105
# work-around for difflib being too smart for its own good
106
# if /dev/null is "1,0", patch won't recognize it as /dev/null
108
ud[2] = ud[2].replace('-1,0', '-0,0')
110
ud[2] = ud[2].replace('+1,0', '+0,0')
114
if not line.endswith('\n'):
115
to_file.write("\n\\ No newline at end of file\n")
119
def _spawn_external_diff(diffcmd, capture_errors=True):
120
"""Spawn the externall diff process, and return the child handle.
122
:param diffcmd: The command list to spawn
123
:param capture_errors: Capture stderr as well as setting LANG=C
124
and LC_ALL=C. This lets us read and understand the output of diff,
125
and respond to any errors.
126
:return: A Popen object.
129
# construct minimal environment
131
path = os.environ.get('PATH')
134
env['LANGUAGE'] = 'C' # on win32 only LANGUAGE has effect
137
stderr = subprocess.PIPE
143
pipe = subprocess.Popen(diffcmd,
144
stdin=subprocess.PIPE,
145
stdout=subprocess.PIPE,
149
if e.errno == errno.ENOENT:
150
raise errors.NoDiff(str(e))
156
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
158
"""Display a diff by calling out to the external diff program."""
159
# make sure our own output is properly ordered before the diff
162
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
163
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
164
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
165
newtmpf = os.fdopen(newtmp_fd, 'wb')
168
# TODO: perhaps a special case for comparing to or from the empty
169
# sequence; can just use /dev/null on Unix
171
# TODO: if either of the files being compared already exists as a
172
# regular named file (e.g. in the working directory) then we can
173
# compare directly to that, rather than copying it.
175
oldtmpf.writelines(oldlines)
176
newtmpf.writelines(newlines)
183
if sys.platform == 'win32':
184
# Popen doesn't do the proper encoding for external commands
185
# Since we are dealing with an ANSI api, use mbcs encoding
186
old_filename = old_filename.encode('mbcs')
187
new_filename = new_filename.encode('mbcs')
189
'--label', old_filename,
191
'--label', new_filename,
196
# diff only allows one style to be specified; they don't override.
197
# note that some of these take optargs, and the optargs can be
198
# directly appended to the options.
199
# this is only an approximate parser; it doesn't properly understand
201
for s in ['-c', '-u', '-C', '-U',
206
'-y', '--side-by-side',
218
diffcmd.extend(diff_opts)
220
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
221
out,err = pipe.communicate()
224
# internal_diff() adds a trailing newline, add one here for consistency
227
# 'diff' gives retcode == 2 for all sorts of errors
228
# one of those is 'Binary files differ'.
229
# Bad options could also be the problem.
230
# 'Binary files' is not a real error, so we suppress that error.
233
# Since we got here, we want to make sure to give an i18n error
234
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
235
out, err = pipe.communicate()
237
# Write out the new i18n diff response
238
to_file.write(out+'\n')
239
if pipe.returncode != 2:
240
raise errors.BzrError(
241
'external diff failed with exit code 2'
242
' when run with LANG=C and LC_ALL=C,'
243
' but not when run natively: %r' % (diffcmd,))
245
first_line = lang_c_out.split('\n', 1)[0]
246
# Starting with diffutils 2.8.4 the word "binary" was dropped.
247
m = re.match('^(binary )?files.*differ$', first_line, re.I)
249
raise errors.BzrError('external diff failed with exit code 2;'
250
' command: %r' % (diffcmd,))
252
# Binary files differ, just return
255
# If we got to here, we haven't written out the output of diff
259
# returns 1 if files differ; that's OK
261
msg = 'signal %d' % (-rc)
263
msg = 'exit code %d' % rc
265
raise errors.BzrError('external diff failed with %s; command: %r'
270
oldtmpf.close() # and delete
272
# Clean up. Warn in case the files couldn't be deleted
273
# (in case windows still holds the file open, but not
274
# if the files have already been deleted)
276
os.remove(old_abspath)
278
if e.errno not in (errno.ENOENT,):
279
warning('Failed to delete temporary file: %s %s',
282
os.remove(new_abspath)
284
if e.errno not in (errno.ENOENT,):
285
warning('Failed to delete temporary file: %s %s',
289
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
291
"""Get the trees and specific files to diff given a list of paths.
293
This method works out the trees to be diff'ed and the files of
294
interest within those trees.
297
the list of arguments passed to the diff command
298
:param revision_specs:
299
Zero, one or two RevisionSpecs from the diff command line,
300
saying what revisions to compare.
302
The url of the old branch or tree. If None, the tree to use is
303
taken from the first path, if any, or the current working tree.
305
The url of the new branch or tree. If None, the tree to use is
306
taken from the first path, if any, or the current working tree.
308
if True and a view is set, apply the view or check that the paths
311
a tuple of (old_tree, new_tree, old_branch, new_branch,
312
specific_files, extra_trees) where extra_trees is a sequence of
313
additional trees to search in for file-ids.
315
# Get the old and new revision specs
316
old_revision_spec = None
317
new_revision_spec = None
318
if revision_specs is not None:
319
if len(revision_specs) > 0:
320
old_revision_spec = revision_specs[0]
322
old_url = old_revision_spec.get_branch()
323
if len(revision_specs) > 1:
324
new_revision_spec = revision_specs[1]
326
new_url = new_revision_spec.get_branch()
329
make_paths_wt_relative = True
330
consider_relpath = True
331
if path_list is None or len(path_list) == 0:
332
# If no path is given, the current working tree is used
333
default_location = u'.'
334
consider_relpath = False
335
elif old_url is not None and new_url is not None:
336
other_paths = path_list
337
make_paths_wt_relative = False
339
default_location = path_list[0]
340
other_paths = path_list[1:]
342
# Get the old location
345
old_url = default_location
346
working_tree, branch, relpath = \
347
bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
348
if consider_relpath and relpath != '':
349
if working_tree is not None and apply_view:
350
views.check_path_in_view(working_tree, relpath)
351
specific_files.append(relpath)
352
old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
355
# Get the new location
357
new_url = default_location
358
if new_url != old_url:
359
working_tree, branch, relpath = \
360
bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
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
new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
366
basis_is_default=working_tree is None)
369
# Get the specific files (all files is None, no files is [])
370
if make_paths_wt_relative and working_tree is not None:
372
from bzrlib.builtins import safe_relpath_files
373
other_paths = safe_relpath_files(working_tree, other_paths,
374
apply_view=apply_view)
375
except errors.FileInWrongBranch:
376
raise errors.BzrCommandError("Files are in different branches")
377
specific_files.extend(other_paths)
378
if len(specific_files) == 0:
379
specific_files = None
380
if (working_tree is not None and working_tree.supports_views()
382
view_files = working_tree.views.lookup_view()
384
specific_files = view_files
385
view_str = views.view_display_str(view_files)
386
note("*** Ignoring files outside view. View is %s" % view_str)
388
# Get extra trees that ought to be searched for file-ids
390
if working_tree is not None and working_tree not in (old_tree, new_tree):
391
extra_trees = (working_tree,)
392
return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
395
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
396
if branch is None and tree is not None:
398
if spec is None or spec.spec is None:
401
return tree.basis_tree()
403
return branch.basis_tree()
406
return spec.as_tree(branch)
409
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
410
external_diff_options=None,
411
old_label='a/', new_label='b/',
413
path_encoding='utf8',
415
"""Show in text form the changes from one tree to another.
421
Include only changes to these files - None for all changes.
423
external_diff_options
424
If set, use an external GNU diff and pass these options.
427
If set, more Trees to use for looking up file ids
430
If set, the path will be encoded as specified, otherwise is supposed
435
if extra_trees is not None:
436
for tree in extra_trees:
440
differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
442
external_diff_options,
443
old_label, new_label, using)
444
return differ.show_diff(specific_files, extra_trees)
447
if extra_trees is not None:
448
for tree in extra_trees:
454
def _patch_header_date(tree, file_id, path):
455
"""Returns a timestamp suitable for use in a patch header."""
457
mtime = tree.get_file_mtime(file_id, path)
458
except errors.FileTimestampUnavailable:
460
return timestamp.format_patch_date(mtime)
463
def get_executable_change(old_is_x, new_is_x):
464
descr = { True:"+x", False:"-x", None:"??" }
465
if old_is_x != new_is_x:
466
return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
471
class DiffPath(object):
472
"""Base type for command object that compare files"""
474
# The type or contents of the file were unsuitable for diffing
475
CANNOT_DIFF = 'CANNOT_DIFF'
476
# The file has changed in a semantic way
478
# The file content may have changed, but there is no semantic change
479
UNCHANGED = 'UNCHANGED'
481
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
484
:param old_tree: The tree to show as the old tree in the comparison
485
:param new_tree: The tree to show as new in the comparison
486
:param to_file: The file to write comparison data to
487
:param path_encoding: The character encoding to write paths in
489
self.old_tree = old_tree
490
self.new_tree = new_tree
491
self.to_file = to_file
492
self.path_encoding = path_encoding
498
def from_diff_tree(klass, diff_tree):
499
return klass(diff_tree.old_tree, diff_tree.new_tree,
500
diff_tree.to_file, diff_tree.path_encoding)
503
def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
504
for file_differ in differs:
505
result = file_differ.diff(file_id, old_path, new_path, old_kind,
507
if result is not DiffPath.CANNOT_DIFF:
510
return DiffPath.CANNOT_DIFF
513
class DiffKindChange(object):
514
"""Special differ for file kind changes.
516
Represents kind change as deletion + creation. Uses the other differs
519
def __init__(self, differs):
520
self.differs = differs
526
def from_diff_tree(klass, diff_tree):
527
return klass(diff_tree.differs)
529
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
530
"""Perform comparison
532
:param file_id: The file_id of the file to compare
533
:param old_path: Path of the file in the old tree
534
:param new_path: Path of the file in the new tree
535
:param old_kind: Old file-kind of the file
536
:param new_kind: New file-kind of the file
538
if None in (old_kind, new_kind):
539
return DiffPath.CANNOT_DIFF
540
result = DiffPath._diff_many(self.differs, file_id, old_path,
541
new_path, old_kind, None)
542
if result is DiffPath.CANNOT_DIFF:
544
return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
548
class DiffDirectory(DiffPath):
550
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
551
"""Perform comparison between two directories. (dummy)
554
if 'directory' not in (old_kind, new_kind):
555
return self.CANNOT_DIFF
556
if old_kind not in ('directory', None):
557
return self.CANNOT_DIFF
558
if new_kind not in ('directory', None):
559
return self.CANNOT_DIFF
563
class DiffSymlink(DiffPath):
565
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
566
"""Perform comparison between two symlinks
568
:param file_id: The file_id of the file to compare
569
:param old_path: Path of the file in the old tree
570
:param new_path: Path of the file in the new tree
571
:param old_kind: Old file-kind of the file
572
:param new_kind: New file-kind of the file
574
if 'symlink' not in (old_kind, new_kind):
575
return self.CANNOT_DIFF
576
if old_kind == 'symlink':
577
old_target = self.old_tree.get_symlink_target(file_id)
578
elif old_kind is None:
581
return self.CANNOT_DIFF
582
if new_kind == 'symlink':
583
new_target = self.new_tree.get_symlink_target(file_id)
584
elif new_kind is None:
587
return self.CANNOT_DIFF
588
return self.diff_symlink(old_target, new_target)
590
def diff_symlink(self, old_target, new_target):
591
if old_target is None:
592
self.to_file.write('=== target is %r\n' % new_target)
593
elif new_target is None:
594
self.to_file.write('=== target was %r\n' % old_target)
596
self.to_file.write('=== target changed %r => %r\n' %
597
(old_target, new_target))
601
class DiffText(DiffPath):
603
# GNU Patch uses the epoch date to detect files that are being added
604
# or removed in a diff.
605
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
607
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
608
old_label='', new_label='', text_differ=internal_diff):
609
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
610
self.text_differ = text_differ
611
self.old_label = old_label
612
self.new_label = new_label
613
self.path_encoding = path_encoding
615
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
616
"""Compare two files in unified diff format
618
:param file_id: The file_id of the file to compare
619
:param old_path: Path of the file in the old tree
620
:param new_path: Path of the file in the new tree
621
:param old_kind: Old file-kind of the file
622
:param new_kind: New file-kind of the file
624
if 'file' not in (old_kind, new_kind):
625
return self.CANNOT_DIFF
626
from_file_id = to_file_id = file_id
627
if old_kind == 'file':
628
old_date = _patch_header_date(self.old_tree, file_id, old_path)
629
elif old_kind is None:
630
old_date = self.EPOCH_DATE
633
return self.CANNOT_DIFF
634
if new_kind == 'file':
635
new_date = _patch_header_date(self.new_tree, file_id, new_path)
636
elif new_kind is None:
637
new_date = self.EPOCH_DATE
640
return self.CANNOT_DIFF
641
from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
642
to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
643
return self.diff_text(from_file_id, to_file_id, from_label, to_label,
646
def diff_text(self, from_file_id, to_file_id, from_label, to_label,
647
from_path=None, to_path=None):
648
"""Diff the content of given files in two trees
650
:param from_file_id: The id of the file in the from tree. If None,
651
the file is not present in the from tree.
652
:param to_file_id: The id of the file in the to tree. This may refer
653
to a different file from from_file_id. If None,
654
the file is not present in the to tree.
655
:param from_path: The path in the from tree or None if unknown.
656
:param to_path: The path in the to tree or None if unknown.
658
def _get_text(tree, file_id, path):
659
if file_id is not None:
660
return tree.get_file(file_id, path).readlines()
664
from_text = _get_text(self.old_tree, from_file_id, from_path)
665
to_text = _get_text(self.new_tree, to_file_id, to_path)
666
self.text_differ(from_label, from_text, to_label, to_text,
668
except errors.BinaryFile:
670
("Binary files %s and %s differ\n" %
671
(from_label, to_label)).encode(self.path_encoding))
675
class DiffFromTool(DiffPath):
677
def __init__(self, command_template, old_tree, new_tree, to_file,
678
path_encoding='utf-8'):
679
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
680
self.command_template = command_template
681
self._root = osutils.mkdtemp(prefix='bzr-diff-')
684
def from_string(klass, command_string, old_tree, new_tree, to_file,
685
path_encoding='utf-8'):
686
command_template = cmdline.split(command_string)
687
if '@' not in command_string:
688
command_template.extend(['@old_path', '@new_path'])
689
return klass(command_template, old_tree, new_tree, to_file,
693
def make_from_diff_tree(klass, command_string):
694
def from_diff_tree(diff_tree):
695
return klass.from_string(command_string, diff_tree.old_tree,
696
diff_tree.new_tree, diff_tree.to_file)
697
return from_diff_tree
699
def _get_command(self, old_path, new_path):
700
my_map = {'old_path': old_path, 'new_path': new_path}
701
return [AtTemplate(t).substitute(my_map) for t in
702
self.command_template]
704
def _execute(self, old_path, new_path):
705
command = self._get_command(old_path, new_path)
707
proc = subprocess.Popen(command, stdout=subprocess.PIPE,
710
if e.errno == errno.ENOENT:
711
raise errors.ExecutableMissing(command[0])
714
self.to_file.write(proc.stdout.read())
717
def _try_symlink_root(self, tree, prefix):
718
if (getattr(tree, 'abspath', None) is None
719
or not osutils.host_os_dereferences_symlinks()):
722
os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
724
if e.errno != errno.EEXIST:
728
def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
730
if not force_temp and isinstance(tree, WorkingTree):
731
return tree.abspath(tree.id2path(file_id))
733
full_path = osutils.pathjoin(self._root, prefix, relpath)
734
if not force_temp and self._try_symlink_root(tree, prefix):
736
parent_dir = osutils.dirname(full_path)
738
os.makedirs(parent_dir)
740
if e.errno != errno.EEXIST:
742
source = tree.get_file(file_id, relpath)
744
target = open(full_path, 'wb')
746
osutils.pumpfile(source, target)
752
osutils.make_readonly(full_path)
754
mtime = tree.get_file_mtime(file_id)
755
except errors.FileTimestampUnavailable:
757
os.utime(full_path, (mtime, mtime))
760
def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
761
allow_write_new=False):
762
old_disk_path = self._write_file(file_id, self.old_tree, 'old',
763
old_path, force_temp)
764
new_disk_path = self._write_file(file_id, self.new_tree, 'new',
765
new_path, force_temp,
766
allow_write=allow_write_new)
767
return old_disk_path, new_disk_path
771
osutils.rmtree(self._root)
773
if e.errno != errno.ENOENT:
774
mutter("The temporary directory \"%s\" was not "
775
"cleanly removed: %s." % (self._root, e))
777
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
778
if (old_kind, new_kind) != ('file', 'file'):
779
return DiffPath.CANNOT_DIFF
780
(old_disk_path, new_disk_path) = self._prepare_files(
781
file_id, old_path, new_path)
782
self._execute(old_disk_path, new_disk_path)
784
def edit_file(self, file_id):
785
"""Use this tool to edit a file.
787
A temporary copy will be edited, and the new contents will be
790
:param file_id: The id of the file to edit.
791
:return: The new contents of the file.
793
old_path = self.old_tree.id2path(file_id)
794
new_path = self.new_tree.id2path(file_id)
795
new_abs_path = self._prepare_files(file_id, old_path, new_path,
796
allow_write_new=True,
798
command = self._get_command(osutils.pathjoin('old', old_path),
799
osutils.pathjoin('new', new_path))
800
subprocess.call(command, cwd=self._root)
801
new_file = open(new_abs_path, 'r')
803
return new_file.read()
808
class DiffTree(object):
809
"""Provides textual representations of the difference between two trees.
811
A DiffTree examines two trees and where a file-id has altered
812
between them, generates a textual representation of the difference.
813
DiffTree uses a sequence of DiffPath objects which are each
814
given the opportunity to handle a given altered fileid. The list
815
of DiffPath objects can be extended globally by appending to
816
DiffTree.diff_factories, or for a specific diff operation by
817
supplying the extra_factories option to the appropriate method.
820
# list of factories that can provide instances of DiffPath objects
821
# may be extended by plugins.
822
diff_factories = [DiffSymlink.from_diff_tree,
823
DiffDirectory.from_diff_tree]
825
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
826
diff_text=None, extra_factories=None):
829
:param old_tree: Tree to show as old in the comparison
830
:param new_tree: Tree to show as new in the comparison
831
:param to_file: File to write comparision to
832
:param path_encoding: Character encoding to write paths in
833
:param diff_text: DiffPath-type object to use as a last resort for
835
:param extra_factories: Factories of DiffPaths to try before any other
837
if diff_text is None:
838
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
839
'', '', internal_diff)
840
self.old_tree = old_tree
841
self.new_tree = new_tree
842
self.to_file = to_file
843
self.path_encoding = path_encoding
845
if extra_factories is not None:
846
self.differs.extend(f(self) for f in extra_factories)
847
self.differs.extend(f(self) for f in self.diff_factories)
848
self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
851
def from_trees_options(klass, old_tree, new_tree, to_file,
852
path_encoding, external_diff_options, old_label,
854
"""Factory for producing a DiffTree.
856
Designed to accept options used by show_diff_trees.
857
:param old_tree: The tree to show as old in the comparison
858
:param new_tree: The tree to show as new in the comparison
859
:param to_file: File to write comparisons to
860
:param path_encoding: Character encoding to use for writing paths
861
:param external_diff_options: If supplied, use the installed diff
862
binary to perform file comparison, using supplied options.
863
:param old_label: Prefix to use for old file labels
864
:param new_label: Prefix to use for new file labels
865
:param using: Commandline to use to invoke an external diff tool
867
if using is not None:
868
extra_factories = [DiffFromTool.make_from_diff_tree(using)]
871
if external_diff_options:
872
opts = external_diff_options.split()
873
def diff_file(olab, olines, nlab, nlines, to_file):
874
external_diff(olab, olines, nlab, nlines, to_file, opts)
876
diff_file = internal_diff
877
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
878
old_label, new_label, diff_file)
879
return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
882
def show_diff(self, specific_files, extra_trees=None):
883
"""Write tree diff to self.to_file
885
:param sepecific_files: the specific files to compare (recursive)
886
:param extra_trees: extra trees to use for mapping paths to file_ids
889
return self._show_diff(specific_files, extra_trees)
891
for differ in self.differs:
894
def _show_diff(self, specific_files, extra_trees):
895
# TODO: Generation of pseudo-diffs for added/deleted files could
896
# be usefully made into a much faster special case.
897
iterator = self.new_tree.iter_changes(self.old_tree,
898
specific_files=specific_files,
899
extra_trees=extra_trees,
900
require_versioned=True)
902
def changes_key(change):
903
old_path, new_path = change[1]
908
def get_encoded_path(path):
910
return path.encode(self.path_encoding, "replace")
911
for (file_id, paths, changed_content, versioned, parent, name, kind,
912
executable) in sorted(iterator, key=changes_key):
913
# The root does not get diffed, and items with no known kind (that
914
# is, missing) in both trees are skipped as well.
915
if parent == (None, None) or kind == (None, None):
917
oldpath, newpath = paths
918
oldpath_encoded = get_encoded_path(paths[0])
919
newpath_encoded = get_encoded_path(paths[1])
920
old_present = (kind[0] is not None and versioned[0])
921
new_present = (kind[1] is not None and versioned[1])
922
renamed = (parent[0], name[0]) != (parent[1], name[1])
924
properties_changed = []
925
properties_changed.extend(get_executable_change(executable[0], executable[1]))
927
if properties_changed:
928
prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
932
if (old_present, new_present) == (True, False):
933
self.to_file.write("=== removed %s '%s'\n" %
934
(kind[0], oldpath_encoded))
936
elif (old_present, new_present) == (False, True):
937
self.to_file.write("=== added %s '%s'\n" %
938
(kind[1], newpath_encoded))
941
self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
942
(kind[0], oldpath_encoded, newpath_encoded, prop_str))
944
# if it was produced by iter_changes, it must be
945
# modified *somehow*, either content or execute bit.
946
self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
947
newpath_encoded, prop_str))
949
self._diff(file_id, oldpath, newpath, kind[0], kind[1])
955
def diff(self, file_id, old_path, new_path):
956
"""Perform a diff of a single file
958
:param file_id: file-id of the file
959
:param old_path: The path of the file in the old tree
960
:param new_path: The path of the file in the new tree
963
old_kind = self.old_tree.kind(file_id)
964
except (errors.NoSuchId, errors.NoSuchFile):
967
new_kind = self.new_tree.kind(file_id)
968
except (errors.NoSuchId, errors.NoSuchFile):
970
self._diff(file_id, old_path, new_path, old_kind, new_kind)
973
def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
974
result = DiffPath._diff_many(self.differs, file_id, old_path,
975
new_path, old_kind, new_kind)
976
if result is DiffPath.CANNOT_DIFF:
977
error_path = new_path
978
if error_path is None:
979
error_path = old_path
980
raise errors.NoDiffFound(error_path)
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
