2
# -*- coding: UTF-8 -*-
1
# Copyright (C) 2005-2010 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
21
from errors import BzrError
24
def diff_trees(old_tree, new_tree):
25
"""Compute diff between two trees.
27
They may be in different branches and may be working or historical
30
Yields a sequence of (state, id, old_name, new_name, kind).
31
Each filename and each id is listed only once.
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(), """
31
branch as _mod_branch,
43
from bzrlib.workingtree import WorkingTree
46
from bzrlib.registry import (
49
from bzrlib.symbol_versioning import (
53
from bzrlib.trace import mutter, note, warning
56
class AtTemplate(string.Template):
57
"""Templating class that uses @ instead of $."""
62
# TODO: Rather than building a changeset object, we should probably
63
# invoke callbacks on an object. That object can either accumulate a
64
# list, write them out directly, etc etc.
67
class _PrematchedMatcher(difflib.SequenceMatcher):
68
"""Allow SequenceMatcher operations to use predetermined blocks"""
70
def __init__(self, matching_blocks):
71
difflib.SequenceMatcher(self, None, None)
72
self.matching_blocks = matching_blocks
76
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
77
allow_binary=False, sequence_matcher=None,
78
path_encoding='utf8'):
79
# FIXME: difflib is wrong if there is no trailing newline.
80
# The syntax used by patch seems to be "\ No newline at
81
# end of file" following the last diff line from that
82
# file. This is not trivial to insert into the
83
# unified_diff output and it might be better to just fix
84
# or replace that function.
86
# In the meantime we at least make sure the patch isn't
90
# Special workaround for Python2.3, where difflib fails if
91
# both sequences are empty.
92
if not oldlines and not newlines:
95
if allow_binary is False:
96
textfile.check_text_lines(oldlines)
97
textfile.check_text_lines(newlines)
99
if sequence_matcher is None:
100
sequence_matcher = patiencediff.PatienceSequenceMatcher
101
ud = patiencediff.unified_diff(oldlines, newlines,
102
fromfile=old_filename.encode(path_encoding),
103
tofile=new_filename.encode(path_encoding),
104
sequencematcher=sequence_matcher)
107
if len(ud) == 0: # Identical contents, nothing to do
109
# work-around for difflib being too smart for its own good
110
# if /dev/null is "1,0", patch won't recognize it as /dev/null
112
ud[2] = ud[2].replace('-1,0', '-0,0')
114
ud[2] = ud[2].replace('+1,0', '+0,0')
118
if not line.endswith('\n'):
119
to_file.write("\n\\ No newline at end of file\n")
123
def _spawn_external_diff(diffcmd, capture_errors=True):
124
"""Spawn the externall diff process, and return the child handle.
126
:param diffcmd: The command list to spawn
127
:param capture_errors: Capture stderr as well as setting LANG=C
128
and LC_ALL=C. This lets us read and understand the output of diff,
129
and respond to any errors.
130
:return: A Popen object.
34
## TODO: Compare files before diffing; only mention those that have changed
36
## TODO: Set nice names in the headers, maybe include diffstat
38
## TODO: Perhaps make this a generator rather than using
41
## TODO: Allow specifying a list of files to compare, rather than
42
## doing the whole tree? (Not urgent.)
44
## TODO: Allow diffing any two inventories, not just the
45
## current one against one. We mgiht need to specify two
46
## stores to look for the files if diffing two branches. That
47
## might imply this shouldn't be primarily a Branch method.
49
## XXX: This doesn't report on unknown files; that can be done
50
## from a separate method.
52
old_it = old_tree.list_files()
53
new_it = new_tree.list_files()
61
old_item = next(old_it)
62
new_item = next(new_it)
64
# We step through the two sorted iterators in parallel, trying to
67
while (old_item != None) or (new_item != None):
68
# OK, we still have some remaining on both, but they may be
71
old_name, old_class, old_kind, old_id = old_item
76
new_name, new_class, new_kind, new_id = new_item
80
mutter(" diff pairwise %r" % (old_item,))
81
mutter(" %r" % (new_item,))
84
# can't handle the old tree being a WorkingTree
85
assert old_class == 'V'
87
if new_item and (new_class != 'V'):
88
yield new_class, None, None, new_name, new_kind
89
new_item = next(new_it)
90
elif (not new_item) or (old_item and (old_name < new_name)):
91
mutter(" extra entry in old-tree sequence")
92
if new_tree.has_id(old_id):
93
# will be mentioned as renamed under new name
96
yield 'D', old_id, old_name, None, old_kind
97
old_item = next(old_it)
98
elif (not old_item) or (new_item and (new_name < old_name)):
99
mutter(" extra entry in new-tree sequence")
100
if old_tree.has_id(new_id):
101
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
103
yield 'A', new_id, None, new_name, new_kind
104
new_item = next(new_it)
105
elif old_id != new_id:
106
assert old_name == new_name
107
# both trees have a file of this name, but it is not the
108
# same file. in other words, the old filename has been
109
# overwritten by either a newly-added or a renamed file.
110
# (should we return something about the overwritten file?)
111
if old_tree.has_id(new_id):
112
# renaming, overlying a deleted file
113
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
115
yield 'A', new_id, None, new_name, new_kind
117
new_item = next(new_it)
118
old_item = next(old_it)
120
assert old_id == new_id
121
assert old_id != None
122
assert old_name == new_name
123
assert old_kind == new_kind
125
if old_kind == 'directory':
126
yield '.', new_id, old_name, new_name, new_kind
127
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
128
mutter(" file size has changed, must be different")
129
yield 'M', new_id, old_name, new_name, new_kind
130
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
131
mutter(" SHA1 indicates they're identical")
132
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
133
yield '.', new_id, old_name, new_name, new_kind
135
mutter(" quick compare shows different")
136
yield 'M', new_id, old_name, new_name, new_kind
138
new_item = next(new_it)
139
old_item = next(old_it)
143
def show_diff(b, revision, file_list):
144
import difflib, sys, types
147
old_tree = b.basis_tree()
133
# construct minimal environment
135
path = os.environ.get('PATH')
138
env['LANGUAGE'] = 'C' # on win32 only LANGUAGE has effect
141
stderr = subprocess.PIPE
149
old_tree = b.revision_tree(b.lookup_revision(revision))
151
new_tree = b.working_tree()
153
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
157
DEVNULL = '/dev/null'
158
# Windows users, don't panic about this filename -- it is a
159
# special signal to GNU patch that the file should be created or
160
# deleted respectively.
162
# TODO: Generation of pseudo-diffs for added/deleted files could
163
# be usefully made into a much faster special case.
165
# TODO: Better to return them in sorted order I think.
168
file_list = [b.relpath(f) for f in file_list]
170
# FIXME: If given a file list, compare only those files rather
171
# than comparing everything and then throwing stuff away.
173
for file_state, fid, old_name, new_name, kind in diff_trees(old_tree, new_tree):
175
if file_list and (new_name not in file_list):
178
# Don't show this by default; maybe do it if an option is passed
179
# idlabel = ' {%s}' % fid
182
# FIXME: Something about the diff format makes patch unhappy
183
# with newly-added files.
185
def diffit(oldlines, newlines, **kw):
187
# FIXME: difflib is wrong if there is no trailing newline.
188
# The syntax used by patch seems to be "\ No newline at
189
# end of file" following the last diff line from that
190
# file. This is not trivial to insert into the
191
# unified_diff output and it might be better to just fix
192
# or replace that function.
194
# In the meantime we at least make sure the patch isn't
198
# Special workaround for Python2.3, where difflib fails if
199
# both sequences are empty.
200
if not oldlines and not newlines:
147
pipe = subprocess.Popen(diffcmd,
148
stdin=subprocess.PIPE,
149
stdout=subprocess.PIPE,
153
if e.errno == errno.ENOENT:
154
raise errors.NoDiff(str(e))
160
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
162
"""Display a diff by calling out to the external diff program."""
163
# make sure our own output is properly ordered before the diff
166
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
167
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
168
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
169
newtmpf = os.fdopen(newtmp_fd, 'wb')
172
# TODO: perhaps a special case for comparing to or from the empty
173
# sequence; can just use /dev/null on Unix
175
# TODO: if either of the files being compared already exists as a
176
# regular named file (e.g. in the working directory) then we can
177
# compare directly to that, rather than copying it.
179
oldtmpf.writelines(oldlines)
180
newtmpf.writelines(newlines)
187
if sys.platform == 'win32':
188
# Popen doesn't do the proper encoding for external commands
189
# Since we are dealing with an ANSI api, use mbcs encoding
190
old_filename = old_filename.encode('mbcs')
191
new_filename = new_filename.encode('mbcs')
193
'--label', old_filename,
195
'--label', new_filename,
200
# diff only allows one style to be specified; they don't override.
201
# note that some of these take optargs, and the optargs can be
202
# directly appended to the options.
203
# this is only an approximate parser; it doesn't properly understand
205
for s in ['-c', '-u', '-C', '-U',
210
'-y', '--side-by-side',
222
diffcmd.extend(diff_opts)
224
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
225
out,err = pipe.communicate()
228
# internal_diff() adds a trailing newline, add one here for consistency
231
# 'diff' gives retcode == 2 for all sorts of errors
232
# one of those is 'Binary files differ'.
233
# Bad options could also be the problem.
234
# 'Binary files' is not a real error, so we suppress that error.
237
# Since we got here, we want to make sure to give an i18n error
238
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
239
out, err = pipe.communicate()
241
# Write out the new i18n diff response
242
to_file.write(out+'\n')
243
if pipe.returncode != 2:
244
raise errors.BzrError(
245
'external diff failed with exit code 2'
246
' when run with LANG=C and LC_ALL=C,'
247
' but not when run natively: %r' % (diffcmd,))
249
first_line = lang_c_out.split('\n', 1)[0]
250
# Starting with diffutils 2.8.4 the word "binary" was dropped.
251
m = re.match('^(binary )?files.*differ$', first_line, re.I)
253
raise errors.BzrError('external diff failed with exit code 2;'
254
' command: %r' % (diffcmd,))
256
# Binary files differ, just return
205
if oldlines and (oldlines[-1][-1] != '\n'):
208
if newlines and (newlines[-1][-1] != '\n'):
212
ud = difflib.unified_diff(oldlines, newlines, **kw)
213
sys.stdout.writelines(ud)
215
print "\\ No newline at end of file"
216
sys.stdout.write('\n')
259
# If we got to here, we haven't written out the output of diff
263
# returns 1 if files differ; that's OK
265
msg = 'signal %d' % (-rc)
267
msg = 'exit code %d' % rc
269
raise errors.BzrError('external diff failed with %s; command: %r'
274
oldtmpf.close() # and delete
276
# Clean up. Warn in case the files couldn't be deleted
277
# (in case windows still holds the file open, but not
278
# if the files have already been deleted)
280
os.remove(old_abspath)
282
if e.errno not in (errno.ENOENT,):
283
warning('Failed to delete temporary file: %s %s',
286
os.remove(new_abspath)
288
if e.errno not in (errno.ENOENT,):
289
warning('Failed to delete temporary file: %s %s',
293
@deprecated_function(deprecated_in((2, 2, 0)))
294
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
296
"""Get the trees and specific files to diff given a list of paths.
298
This method works out the trees to be diff'ed and the files of
299
interest within those trees.
302
the list of arguments passed to the diff command
303
:param revision_specs:
304
Zero, one or two RevisionSpecs from the diff command line,
305
saying what revisions to compare.
307
The url of the old branch or tree. If None, the tree to use is
308
taken from the first path, if any, or the current working tree.
310
The url of the new branch or tree. If None, the tree to use is
311
taken from the first path, if any, or the current working tree.
313
if True and a view is set, apply the view or check that the paths
316
a tuple of (old_tree, new_tree, old_branch, new_branch,
317
specific_files, extra_trees) where extra_trees is a sequence of
318
additional trees to search in for file-ids. The trees and branches
321
op = cleanup.OperationWithCleanups(get_trees_and_branches_to_diff_locked)
322
return op.run_simple(path_list, revision_specs, old_url, new_url,
323
op.add_cleanup, apply_view=apply_view)
326
def get_trees_and_branches_to_diff_locked(
327
path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
328
"""Get the trees and specific files to diff given a list of paths.
330
This method works out the trees to be diff'ed and the files of
331
interest within those trees.
334
the list of arguments passed to the diff command
335
:param revision_specs:
336
Zero, one or two RevisionSpecs from the diff command line,
337
saying what revisions to compare.
339
The url of the old branch or tree. If None, the tree to use is
340
taken from the first path, if any, or the current working tree.
342
The url of the new branch or tree. If None, the tree to use is
343
taken from the first path, if any, or the current working tree.
345
a callable like Command.add_cleanup. get_trees_and_branches_to_diff
346
will register cleanups that must be run to unlock the trees, etc.
348
if True and a view is set, apply the view or check that the paths
351
a tuple of (old_tree, new_tree, old_branch, new_branch,
352
specific_files, extra_trees) where extra_trees is a sequence of
353
additional trees to search in for file-ids. The trees and branches
354
will be read-locked until the cleanups registered via the add_cleanup
357
# Get the old and new revision specs
358
old_revision_spec = None
359
new_revision_spec = None
360
if revision_specs is not None:
361
if len(revision_specs) > 0:
362
old_revision_spec = revision_specs[0]
364
old_url = old_revision_spec.get_branch()
365
if len(revision_specs) > 1:
366
new_revision_spec = revision_specs[1]
368
new_url = new_revision_spec.get_branch()
371
make_paths_wt_relative = True
372
consider_relpath = True
373
if path_list is None or len(path_list) == 0:
374
# If no path is given, the current working tree is used
375
default_location = u'.'
376
consider_relpath = False
377
elif old_url is not None and new_url is not None:
378
other_paths = path_list
379
make_paths_wt_relative = False
381
default_location = path_list[0]
382
other_paths = path_list[1:]
384
def lock_tree_or_branch(wt, br):
387
add_cleanup(wt.unlock)
390
add_cleanup(br.unlock)
392
# Get the old location
395
old_url = default_location
396
working_tree, branch, relpath = \
397
bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
398
lock_tree_or_branch(working_tree, branch)
399
if consider_relpath and relpath != '':
400
if working_tree is not None and apply_view:
401
views.check_path_in_view(working_tree, relpath)
402
specific_files.append(relpath)
403
old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
406
# Get the new location
408
new_url = default_location
409
if new_url != old_url:
410
working_tree, branch, relpath = \
411
bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
412
lock_tree_or_branch(working_tree, branch)
413
if consider_relpath and relpath != '':
414
if working_tree is not None and apply_view:
415
views.check_path_in_view(working_tree, relpath)
416
specific_files.append(relpath)
417
new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
418
basis_is_default=working_tree is None)
421
# Get the specific files (all files is None, no files is [])
422
if make_paths_wt_relative and working_tree is not None:
424
from bzrlib.builtins import safe_relpath_files
425
other_paths = safe_relpath_files(working_tree, other_paths,
426
apply_view=apply_view)
427
except errors.FileInWrongBranch:
428
raise errors.BzrCommandError("Files are in different branches")
429
specific_files.extend(other_paths)
430
if len(specific_files) == 0:
431
specific_files = None
432
if (working_tree is not None and working_tree.supports_views()
434
view_files = working_tree.views.lookup_view()
436
specific_files = view_files
437
view_str = views.view_display_str(view_files)
438
note("*** Ignoring files outside view. View is %s" % view_str)
440
# Get extra trees that ought to be searched for file-ids
442
if working_tree is not None and working_tree not in (old_tree, new_tree):
443
extra_trees = (working_tree,)
444
return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
447
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
448
if branch is None and tree is not None:
450
if spec is None or spec.spec is None:
453
return tree.basis_tree()
455
return branch.basis_tree()
458
return spec.as_tree(branch)
461
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
462
external_diff_options=None,
463
old_label='a/', new_label='b/',
465
path_encoding='utf8',
468
"""Show in text form the changes from one tree to another.
470
:param to_file: The output stream.
471
:param specific_files:Include only changes to these files - None for all
473
:param external_diff_options: If set, use an external GNU diff and pass
475
:param extra_trees: If set, more Trees to use for looking up file ids
476
:param path_encoding: If set, the path will be encoded as specified,
477
otherwise is supposed to be utf8
478
:param format_cls: Formatter class (DiffTree subclass)
480
if format_cls is None:
481
format_cls = DiffTree
484
if extra_trees is not None:
485
for tree in extra_trees:
489
differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
491
external_diff_options,
492
old_label, new_label, using)
493
return differ.show_diff(specific_files, extra_trees)
496
if extra_trees is not None:
497
for tree in extra_trees:
503
def _patch_header_date(tree, file_id, path):
504
"""Returns a timestamp suitable for use in a patch header."""
506
mtime = tree.get_file_mtime(file_id, path)
507
except errors.FileTimestampUnavailable:
509
return timestamp.format_patch_date(mtime)
512
def get_executable_change(old_is_x, new_is_x):
513
descr = { True:"+x", False:"-x", None:"??" }
514
if old_is_x != new_is_x:
515
return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
520
class DiffPath(object):
521
"""Base type for command object that compare files"""
523
# The type or contents of the file were unsuitable for diffing
524
CANNOT_DIFF = 'CANNOT_DIFF'
525
# The file has changed in a semantic way
527
# The file content may have changed, but there is no semantic change
528
UNCHANGED = 'UNCHANGED'
530
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
533
:param old_tree: The tree to show as the old tree in the comparison
534
:param new_tree: The tree to show as new in the comparison
535
:param to_file: The file to write comparison data to
536
:param path_encoding: The character encoding to write paths in
538
self.old_tree = old_tree
539
self.new_tree = new_tree
540
self.to_file = to_file
541
self.path_encoding = path_encoding
547
def from_diff_tree(klass, diff_tree):
548
return klass(diff_tree.old_tree, diff_tree.new_tree,
549
diff_tree.to_file, diff_tree.path_encoding)
552
def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
553
for file_differ in differs:
554
result = file_differ.diff(file_id, old_path, new_path, old_kind,
556
if result is not DiffPath.CANNOT_DIFF:
559
return DiffPath.CANNOT_DIFF
562
class DiffKindChange(object):
563
"""Special differ for file kind changes.
565
Represents kind change as deletion + creation. Uses the other differs
568
def __init__(self, differs):
569
self.differs = differs
575
def from_diff_tree(klass, diff_tree):
576
return klass(diff_tree.differs)
578
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
579
"""Perform comparison
581
:param file_id: The file_id of the file to compare
582
:param old_path: Path of the file in the old tree
583
:param new_path: Path of the file in the new tree
584
:param old_kind: Old file-kind of the file
585
:param new_kind: New file-kind of the file
587
if None in (old_kind, new_kind):
588
return DiffPath.CANNOT_DIFF
589
result = DiffPath._diff_many(self.differs, file_id, old_path,
590
new_path, old_kind, None)
591
if result is DiffPath.CANNOT_DIFF:
593
return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
597
class DiffDirectory(DiffPath):
599
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
600
"""Perform comparison between two directories. (dummy)
603
if 'directory' not in (old_kind, new_kind):
604
return self.CANNOT_DIFF
605
if old_kind not in ('directory', None):
606
return self.CANNOT_DIFF
607
if new_kind not in ('directory', None):
608
return self.CANNOT_DIFF
612
class DiffSymlink(DiffPath):
614
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
615
"""Perform comparison between two symlinks
617
:param file_id: The file_id of the file to compare
618
:param old_path: Path of the file in the old tree
619
:param new_path: Path of the file in the new tree
620
:param old_kind: Old file-kind of the file
621
:param new_kind: New file-kind of the file
623
if 'symlink' not in (old_kind, new_kind):
624
return self.CANNOT_DIFF
625
if old_kind == 'symlink':
626
old_target = self.old_tree.get_symlink_target(file_id)
627
elif old_kind is None:
630
return self.CANNOT_DIFF
631
if new_kind == 'symlink':
632
new_target = self.new_tree.get_symlink_target(file_id)
633
elif new_kind is None:
636
return self.CANNOT_DIFF
637
return self.diff_symlink(old_target, new_target)
639
def diff_symlink(self, old_target, new_target):
640
if old_target is None:
641
self.to_file.write('=== target is %r\n' % new_target)
642
elif new_target is None:
643
self.to_file.write('=== target was %r\n' % old_target)
645
self.to_file.write('=== target changed %r => %r\n' %
646
(old_target, new_target))
650
class DiffText(DiffPath):
652
# GNU Patch uses the epoch date to detect files that are being added
653
# or removed in a diff.
654
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
656
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
657
old_label='', new_label='', text_differ=internal_diff):
658
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
659
self.text_differ = text_differ
660
self.old_label = old_label
661
self.new_label = new_label
662
self.path_encoding = path_encoding
664
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
665
"""Compare two files in unified diff format
667
:param file_id: The file_id of the file to compare
668
:param old_path: Path of the file in the old tree
669
:param new_path: Path of the file in the new tree
670
:param old_kind: Old file-kind of the file
671
:param new_kind: New file-kind of the file
673
if 'file' not in (old_kind, new_kind):
674
return self.CANNOT_DIFF
675
from_file_id = to_file_id = file_id
676
if old_kind == 'file':
677
old_date = _patch_header_date(self.old_tree, file_id, old_path)
678
elif old_kind is None:
679
old_date = self.EPOCH_DATE
682
return self.CANNOT_DIFF
683
if new_kind == 'file':
684
new_date = _patch_header_date(self.new_tree, file_id, new_path)
685
elif new_kind is None:
686
new_date = self.EPOCH_DATE
689
return self.CANNOT_DIFF
690
from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
691
to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
692
return self.diff_text(from_file_id, to_file_id, from_label, to_label,
695
def diff_text(self, from_file_id, to_file_id, from_label, to_label,
696
from_path=None, to_path=None):
697
"""Diff the content of given files in two trees
699
:param from_file_id: The id of the file in the from tree. If None,
700
the file is not present in the from tree.
701
:param to_file_id: The id of the file in the to tree. This may refer
702
to a different file from from_file_id. If None,
703
the file is not present in the to tree.
704
:param from_path: The path in the from tree or None if unknown.
705
:param to_path: The path in the to tree or None if unknown.
707
def _get_text(tree, file_id, path):
708
if file_id is not None:
709
return tree.get_file(file_id, path).readlines()
713
from_text = _get_text(self.old_tree, from_file_id, from_path)
714
to_text = _get_text(self.new_tree, to_file_id, to_path)
715
self.text_differ(from_label, from_text, to_label, to_text,
717
except errors.BinaryFile:
719
("Binary files %s and %s differ\n" %
720
(from_label, to_label)).encode(self.path_encoding))
724
class DiffFromTool(DiffPath):
726
def __init__(self, command_template, old_tree, new_tree, to_file,
727
path_encoding='utf-8'):
728
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
729
self.command_template = command_template
730
self._root = osutils.mkdtemp(prefix='bzr-diff-')
733
def from_string(klass, command_string, old_tree, new_tree, to_file,
734
path_encoding='utf-8'):
735
command_template = cmdline.split(command_string)
736
if '@' not in command_string:
737
command_template.extend(['@old_path', '@new_path'])
738
return klass(command_template, old_tree, new_tree, to_file,
742
def make_from_diff_tree(klass, command_string):
743
def from_diff_tree(diff_tree):
744
return klass.from_string(command_string, diff_tree.old_tree,
745
diff_tree.new_tree, diff_tree.to_file)
746
return from_diff_tree
748
def _get_command(self, old_path, new_path):
749
my_map = {'old_path': old_path, 'new_path': new_path}
750
return [AtTemplate(t).substitute(my_map) for t in
751
self.command_template]
753
def _execute(self, old_path, new_path):
754
command = self._get_command(old_path, new_path)
756
proc = subprocess.Popen(command, stdout=subprocess.PIPE,
759
if e.errno == errno.ENOENT:
760
raise errors.ExecutableMissing(command[0])
763
self.to_file.write(proc.stdout.read())
766
def _try_symlink_root(self, tree, prefix):
767
if (getattr(tree, 'abspath', None) is None
768
or not osutils.host_os_dereferences_symlinks()):
771
os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
773
if e.errno != errno.EEXIST:
777
def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
779
if not force_temp and isinstance(tree, WorkingTree):
780
return tree.abspath(tree.id2path(file_id))
218
if file_state in ['.', '?', 'I']:
220
elif file_state == 'A':
221
print '*** added %s %r' % (kind, new_name)
224
new_tree.get_file(fid).readlines(),
226
tofile=new_label + new_name + idlabel)
227
elif file_state == 'D':
228
assert isinstance(old_name, types.StringTypes)
229
print '*** deleted %s %r' % (kind, old_name)
231
diffit(old_tree.get_file(fid).readlines(), [],
232
fromfile=old_label + old_name + idlabel,
234
elif file_state in ['M', 'R']:
235
if file_state == 'M':
236
assert kind == 'file'
237
assert old_name == new_name
238
print '*** modified %s %r' % (kind, new_name)
239
elif file_state == 'R':
240
print '*** renamed %s %r => %r' % (kind, old_name, new_name)
243
diffit(old_tree.get_file(fid).readlines(),
244
new_tree.get_file(fid).readlines(),
245
fromfile=old_label + old_name + idlabel,
246
tofile=new_label + new_name)
248
raise BzrError("can't represent state %s {%s}" % (file_state, fid))
253
"""Describes changes from one tree to another.
262
(oldpath, newpath, id)
266
A path may occur in more than one list if it was e.g. deleted
267
under an old id and renamed into place in a new id.
269
Files are listed in either modified or renamed, not both. In
270
other words, renamed files may also be modified.
279
def compare_inventories(old_inv, new_inv):
280
"""Return a TreeDelta object describing changes between inventories.
282
This only describes changes in the shape of the tree, not the
285
This is an alternative to diff_trees() and should probably
286
eventually replace it.
288
old_ids = old_inv.id_set()
289
new_ids = new_inv.id_set()
292
delta.removed = [(old_inv.id2path(fid), fid) for fid in (old_ids - new_ids)]
295
delta.added = [(new_inv.id2path(fid), fid) for fid in (new_ids - old_ids)]
298
for fid in old_ids & new_ids:
299
old_ie = old_inv[fid]
300
new_ie = new_inv[fid]
301
old_path = old_inv.id2path(fid)
302
new_path = new_inv.id2path(fid)
304
if old_path != new_path:
305
delta.renamed.append((old_path, new_path, fid))
306
elif old_ie.text_sha1 != new_ie.text_sha1:
307
delta.modified.append((new_path, fid))
309
delta.modified.sort()
782
full_path = osutils.pathjoin(self._root, prefix, relpath)
783
if not force_temp and self._try_symlink_root(tree, prefix):
785
parent_dir = osutils.dirname(full_path)
787
os.makedirs(parent_dir)
789
if e.errno != errno.EEXIST:
791
source = tree.get_file(file_id, relpath)
793
target = open(full_path, 'wb')
795
osutils.pumpfile(source, target)
801
mtime = tree.get_file_mtime(file_id)
802
except errors.FileTimestampUnavailable:
805
os.utime(full_path, (mtime, mtime))
807
osutils.make_readonly(full_path)
810
def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
811
allow_write_new=False):
812
old_disk_path = self._write_file(file_id, self.old_tree, 'old',
813
old_path, force_temp)
814
new_disk_path = self._write_file(file_id, self.new_tree, 'new',
815
new_path, force_temp,
816
allow_write=allow_write_new)
817
return old_disk_path, new_disk_path
821
osutils.rmtree(self._root)
823
if e.errno != errno.ENOENT:
824
mutter("The temporary directory \"%s\" was not "
825
"cleanly removed: %s." % (self._root, e))
827
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
828
if (old_kind, new_kind) != ('file', 'file'):
829
return DiffPath.CANNOT_DIFF
830
(old_disk_path, new_disk_path) = self._prepare_files(
831
file_id, old_path, new_path)
832
self._execute(old_disk_path, new_disk_path)
834
def edit_file(self, file_id):
835
"""Use this tool to edit a file.
837
A temporary copy will be edited, and the new contents will be
840
:param file_id: The id of the file to edit.
841
:return: The new contents of the file.
843
old_path = self.old_tree.id2path(file_id)
844
new_path = self.new_tree.id2path(file_id)
845
new_abs_path = self._prepare_files(file_id, old_path, new_path,
846
allow_write_new=True,
848
command = self._get_command(osutils.pathjoin('old', old_path),
849
osutils.pathjoin('new', new_path))
850
subprocess.call(command, cwd=self._root)
851
new_file = open(new_abs_path, 'r')
853
return new_file.read()
858
class DiffTree(object):
859
"""Provides textual representations of the difference between two trees.
861
A DiffTree examines two trees and where a file-id has altered
862
between them, generates a textual representation of the difference.
863
DiffTree uses a sequence of DiffPath objects which are each
864
given the opportunity to handle a given altered fileid. The list
865
of DiffPath objects can be extended globally by appending to
866
DiffTree.diff_factories, or for a specific diff operation by
867
supplying the extra_factories option to the appropriate method.
870
# list of factories that can provide instances of DiffPath objects
871
# may be extended by plugins.
872
diff_factories = [DiffSymlink.from_diff_tree,
873
DiffDirectory.from_diff_tree]
875
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
876
diff_text=None, extra_factories=None):
879
:param old_tree: Tree to show as old in the comparison
880
:param new_tree: Tree to show as new in the comparison
881
:param to_file: File to write comparision to
882
:param path_encoding: Character encoding to write paths in
883
:param diff_text: DiffPath-type object to use as a last resort for
885
:param extra_factories: Factories of DiffPaths to try before any other
887
if diff_text is None:
888
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
889
'', '', internal_diff)
890
self.old_tree = old_tree
891
self.new_tree = new_tree
892
self.to_file = to_file
893
self.path_encoding = path_encoding
895
if extra_factories is not None:
896
self.differs.extend(f(self) for f in extra_factories)
897
self.differs.extend(f(self) for f in self.diff_factories)
898
self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
901
def from_trees_options(klass, old_tree, new_tree, to_file,
902
path_encoding, external_diff_options, old_label,
904
"""Factory for producing a DiffTree.
906
Designed to accept options used by show_diff_trees.
907
:param old_tree: The tree to show as old in the comparison
908
:param new_tree: The tree to show as new in the comparison
909
:param to_file: File to write comparisons to
910
:param path_encoding: Character encoding to use for writing paths
911
:param external_diff_options: If supplied, use the installed diff
912
binary to perform file comparison, using supplied options.
913
:param old_label: Prefix to use for old file labels
914
:param new_label: Prefix to use for new file labels
915
:param using: Commandline to use to invoke an external diff tool
917
if using is not None:
918
extra_factories = [DiffFromTool.make_from_diff_tree(using)]
921
if external_diff_options:
922
opts = external_diff_options.split()
923
def diff_file(olab, olines, nlab, nlines, to_file):
924
external_diff(olab, olines, nlab, nlines, to_file, opts)
926
diff_file = internal_diff
927
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
928
old_label, new_label, diff_file)
929
return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
932
def show_diff(self, specific_files, extra_trees=None):
933
"""Write tree diff to self.to_file
935
:param specific_files: the specific files to compare (recursive)
936
:param extra_trees: extra trees to use for mapping paths to file_ids
939
return self._show_diff(specific_files, extra_trees)
941
for differ in self.differs:
944
def _show_diff(self, specific_files, extra_trees):
945
# TODO: Generation of pseudo-diffs for added/deleted files could
946
# be usefully made into a much faster special case.
947
iterator = self.new_tree.iter_changes(self.old_tree,
948
specific_files=specific_files,
949
extra_trees=extra_trees,
950
require_versioned=True)
952
def changes_key(change):
953
old_path, new_path = change[1]
958
def get_encoded_path(path):
960
return path.encode(self.path_encoding, "replace")
961
for (file_id, paths, changed_content, versioned, parent, name, kind,
962
executable) in sorted(iterator, key=changes_key):
963
# The root does not get diffed, and items with no known kind (that
964
# is, missing) in both trees are skipped as well.
965
if parent == (None, None) or kind == (None, None):
967
oldpath, newpath = paths
968
oldpath_encoded = get_encoded_path(paths[0])
969
newpath_encoded = get_encoded_path(paths[1])
970
old_present = (kind[0] is not None and versioned[0])
971
new_present = (kind[1] is not None and versioned[1])
972
renamed = (parent[0], name[0]) != (parent[1], name[1])
974
properties_changed = []
975
properties_changed.extend(get_executable_change(executable[0], executable[1]))
977
if properties_changed:
978
prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
982
if (old_present, new_present) == (True, False):
983
self.to_file.write("=== removed %s '%s'\n" %
984
(kind[0], oldpath_encoded))
986
elif (old_present, new_present) == (False, True):
987
self.to_file.write("=== added %s '%s'\n" %
988
(kind[1], newpath_encoded))
991
self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
992
(kind[0], oldpath_encoded, newpath_encoded, prop_str))
994
# if it was produced by iter_changes, it must be
995
# modified *somehow*, either content or execute bit.
996
self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
997
newpath_encoded, prop_str))
999
self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1005
def diff(self, file_id, old_path, new_path):
1006
"""Perform a diff of a single file
1008
:param file_id: file-id of the file
1009
:param old_path: The path of the file in the old tree
1010
:param new_path: The path of the file in the new tree
1013
old_kind = self.old_tree.kind(file_id)
1014
except (errors.NoSuchId, errors.NoSuchFile):
1017
new_kind = self.new_tree.kind(file_id)
1018
except (errors.NoSuchId, errors.NoSuchFile):
1020
self._diff(file_id, old_path, new_path, old_kind, new_kind)
1023
def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
1024
result = DiffPath._diff_many(self.differs, file_id, old_path,
1025
new_path, old_kind, new_kind)
1026
if result is DiffPath.CANNOT_DIFF:
1027
error_path = new_path
1028
if error_path is None:
1029
error_path = old_path
1030
raise errors.NoDiffFound(error_path)
1033
format_registry = Registry()
1034
format_registry.register('default', DiffTree)
added
removed
bzrlib/diff.py
