2
# -*- coding: UTF-8 -*-
1
# Copyright (C) 2004, 2005, 2006 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
15
# 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.
23
from bzrlib.lazy_import import lazy_import
24
lazy_import(globals(), """
31
branch as _mod_branch,
42
from bzrlib.symbol_versioning import (
46
from bzrlib.trace import warning
49
# TODO: Rather than building a changeset object, we should probably
50
# invoke callbacks on an object. That object can either accumulate a
51
# list, write them out directly, etc etc.
54
class _PrematchedMatcher(difflib.SequenceMatcher):
55
"""Allow SequenceMatcher operations to use predetermined blocks"""
57
def __init__(self, matching_blocks):
58
difflib.SequenceMatcher(self, None, None)
59
self.matching_blocks = matching_blocks
63
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
64
allow_binary=False, sequence_matcher=None,
65
path_encoding='utf8'):
66
# FIXME: difflib is wrong if there is no trailing newline.
67
# The syntax used by patch seems to be "\ No newline at
68
# end of file" following the last diff line from that
69
# file. This is not trivial to insert into the
70
# unified_diff output and it might be better to just fix
71
# or replace that function.
73
# In the meantime we at least make sure the patch isn't
77
# Special workaround for Python2.3, where difflib fails if
78
# both sequences are empty.
79
if not oldlines and not newlines:
82
if allow_binary is False:
83
textfile.check_text_lines(oldlines)
84
textfile.check_text_lines(newlines)
86
if sequence_matcher is None:
87
sequence_matcher = patiencediff.PatienceSequenceMatcher
88
ud = patiencediff.unified_diff(oldlines, newlines,
89
fromfile=old_filename.encode(path_encoding),
90
tofile=new_filename.encode(path_encoding),
91
sequencematcher=sequence_matcher)
94
if len(ud) == 0: # Identical contents, nothing to do
96
# work-around for difflib being too smart for its own good
97
# if /dev/null is "1,0", patch won't recognize it as /dev/null
99
ud[2] = ud[2].replace('-1,0', '-0,0')
101
ud[2] = ud[2].replace('+1,0', '+0,0')
102
# work around for difflib emitting random spaces after the label
103
ud[0] = ud[0][:-2] + '\n'
104
ud[1] = ud[1][:-2] + '\n'
108
if not line.endswith('\n'):
109
to_file.write("\n\\ No newline at end of file\n")
113
def _spawn_external_diff(diffcmd, capture_errors=True):
114
"""Spawn the externall diff process, and return the child handle.
116
:param diffcmd: The command list to spawn
117
:param capture_errors: Capture stderr as well as setting LANG=C
118
and LC_ALL=C. This lets us read and understand the output of diff,
119
and respond to any errors.
120
: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()
123
# construct minimal environment
125
path = os.environ.get('PATH')
128
env['LANGUAGE'] = 'C' # on win32 only LANGUAGE has effect
131
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:
137
pipe = subprocess.Popen(diffcmd,
138
stdin=subprocess.PIPE,
139
stdout=subprocess.PIPE,
143
if e.errno == errno.ENOENT:
144
raise errors.NoDiff(str(e))
150
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
152
"""Display a diff by calling out to the external diff program."""
153
# make sure our own output is properly ordered before the diff
156
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
157
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
158
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
159
newtmpf = os.fdopen(newtmp_fd, 'wb')
162
# TODO: perhaps a special case for comparing to or from the empty
163
# sequence; can just use /dev/null on Unix
165
# TODO: if either of the files being compared already exists as a
166
# regular named file (e.g. in the working directory) then we can
167
# compare directly to that, rather than copying it.
169
oldtmpf.writelines(oldlines)
170
newtmpf.writelines(newlines)
178
'--label', old_filename,
180
'--label', new_filename,
185
# diff only allows one style to be specified; they don't override.
186
# note that some of these take optargs, and the optargs can be
187
# directly appended to the options.
188
# this is only an approximate parser; it doesn't properly understand
190
for s in ['-c', '-u', '-C', '-U',
195
'-y', '--side-by-side',
207
diffcmd.extend(diff_opts)
209
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
210
out,err = pipe.communicate()
213
# internal_diff() adds a trailing newline, add one here for consistency
216
# 'diff' gives retcode == 2 for all sorts of errors
217
# one of those is 'Binary files differ'.
218
# Bad options could also be the problem.
219
# 'Binary files' is not a real error, so we suppress that error.
222
# Since we got here, we want to make sure to give an i18n error
223
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
224
out, err = pipe.communicate()
226
# Write out the new i18n diff response
227
to_file.write(out+'\n')
228
if pipe.returncode != 2:
229
raise errors.BzrError(
230
'external diff failed with exit code 2'
231
' when run with LANG=C and LC_ALL=C,'
232
' but not when run natively: %r' % (diffcmd,))
234
first_line = lang_c_out.split('\n', 1)[0]
235
# Starting with diffutils 2.8.4 the word "binary" was dropped.
236
m = re.match('^(binary )?files.*differ$', first_line, re.I)
238
raise errors.BzrError('external diff failed with exit code 2;'
239
' command: %r' % (diffcmd,))
241
# 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')
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()
244
# If we got to here, we haven't written out the output of diff
248
# returns 1 if files differ; that's OK
250
msg = 'signal %d' % (-rc)
252
msg = 'exit code %d' % rc
254
raise errors.BzrError('external diff failed with %s; command: %r'
259
oldtmpf.close() # and delete
261
# Clean up. Warn in case the files couldn't be deleted
262
# (in case windows still holds the file open, but not
263
# if the files have already been deleted)
265
os.remove(old_abspath)
267
if e.errno not in (errno.ENOENT,):
268
warning('Failed to delete temporary file: %s %s',
271
os.remove(new_abspath)
273
if e.errno not in (errno.ENOENT,):
274
warning('Failed to delete temporary file: %s %s',
278
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url):
279
"""Get the trees and specific files to diff given a list of paths.
281
This method works out the trees to be diff'ed and the files of
282
interest within those trees.
285
the list of arguments passed to the diff command
286
:param revision_specs:
287
Zero, one or two RevisionSpecs from the diff command line,
288
saying what revisions to compare.
290
The url of the old branch or tree. If None, the tree to use is
291
taken from the first path, if any, or the current working tree.
293
The url of the new branch or tree. If None, the tree to use is
294
taken from the first path, if any, or the current working tree.
296
a tuple of (old_tree, new_tree, specific_files, extra_trees) where
297
extra_trees is a sequence of additional trees to search in for
300
# Get the old and new revision specs
301
old_revision_spec = None
302
new_revision_spec = None
303
if revision_specs is not None:
304
if len(revision_specs) > 0:
305
old_revision_spec = revision_specs[0]
307
old_url = old_revision_spec.get_branch()
308
if len(revision_specs) > 1:
309
new_revision_spec = revision_specs[1]
311
new_url = new_revision_spec.get_branch()
314
make_paths_wt_relative = True
315
consider_relpath = True
316
if path_list is None or len(path_list) == 0:
317
# If no path is given, the current working tree is used
318
default_location = u'.'
319
consider_relpath = False
320
elif old_url is not None and new_url is not None:
321
other_paths = path_list
322
make_paths_wt_relative = False
324
default_location = path_list[0]
325
other_paths = path_list[1:]
327
# Get the old location
330
old_url = default_location
331
working_tree, branch, relpath = \
332
bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
333
if consider_relpath and relpath != '':
334
specific_files.append(relpath)
335
old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
337
# Get the new location
339
new_url = default_location
340
if new_url != old_url:
341
working_tree, branch, relpath = \
342
bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
343
if consider_relpath and relpath != '':
344
specific_files.append(relpath)
345
new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
346
basis_is_default=working_tree is None)
348
# Get the specific files (all files is None, no files is [])
349
if make_paths_wt_relative and working_tree is not None:
350
other_paths = _relative_paths_in_tree(working_tree, other_paths)
351
specific_files.extend(other_paths)
352
if len(specific_files) == 0:
353
specific_files = None
355
# Get extra trees that ought to be searched for file-ids
357
if working_tree is not None and working_tree not in (old_tree, new_tree):
358
extra_trees = (working_tree,)
359
return old_tree, new_tree, specific_files, extra_trees
362
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
363
if branch is None and tree is not None:
365
if spec is None or spec.spec is None:
368
return tree.basis_tree()
370
return branch.basis_tree()
373
return spec.as_tree(branch)
376
def _relative_paths_in_tree(tree, paths):
377
"""Get the relative paths within a working tree.
379
Each path may be either an absolute path or a path relative to the
380
current working directory.
383
for filename in paths:
385
result.append(tree.relpath(osutils.dereference_path(filename)))
386
except errors.PathNotChild:
387
raise errors.BzrCommandError("Files are in different branches")
391
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
392
external_diff_options=None,
393
old_label='a/', new_label='b/',
395
path_encoding='utf8',
397
"""Show in text form the changes from one tree to another.
403
Include only changes to these files - None for all changes.
405
external_diff_options
406
If set, use an external GNU diff and pass these options.
409
If set, more Trees to use for looking up file ids
412
If set, the path will be encoded as specified, otherwise is supposed
417
if extra_trees is not None:
418
for tree in extra_trees:
422
differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
424
external_diff_options,
425
old_label, new_label, using)
426
return differ.show_diff(specific_files, extra_trees)
429
if extra_trees is not None:
430
for tree in extra_trees:
436
def _patch_header_date(tree, file_id, path):
437
"""Returns a timestamp suitable for use in a patch header."""
438
mtime = tree.get_file_mtime(file_id, path)
439
return timestamp.format_patch_date(mtime)
442
@deprecated_function(one_three)
443
def get_prop_change(meta_modified):
445
return " (properties changed)"
449
def get_executable_change(old_is_x, new_is_x):
450
descr = { True:"+x", False:"-x", None:"??" }
451
if old_is_x != new_is_x:
452
return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
457
class DiffPath(object):
458
"""Base type for command object that compare files"""
460
# The type or contents of the file were unsuitable for diffing
461
CANNOT_DIFF = 'CANNOT_DIFF'
462
# The file has changed in a semantic way
464
# The file content may have changed, but there is no semantic change
465
UNCHANGED = 'UNCHANGED'
467
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
470
:param old_tree: The tree to show as the old tree in the comparison
471
:param new_tree: The tree to show as new in the comparison
472
:param to_file: The file to write comparison data to
473
:param path_encoding: The character encoding to write paths in
475
self.old_tree = old_tree
476
self.new_tree = new_tree
477
self.to_file = to_file
478
self.path_encoding = path_encoding
484
def from_diff_tree(klass, diff_tree):
485
return klass(diff_tree.old_tree, diff_tree.new_tree,
486
diff_tree.to_file, diff_tree.path_encoding)
489
def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
490
for file_differ in differs:
491
result = file_differ.diff(file_id, old_path, new_path, old_kind,
493
if result is not DiffPath.CANNOT_DIFF:
496
return DiffPath.CANNOT_DIFF
499
class DiffKindChange(object):
500
"""Special differ for file kind changes.
502
Represents kind change as deletion + creation. Uses the other differs
505
def __init__(self, differs):
506
self.differs = differs
512
def from_diff_tree(klass, diff_tree):
513
return klass(diff_tree.differs)
515
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
516
"""Perform comparison
518
:param file_id: The file_id of the file to compare
519
:param old_path: Path of the file in the old tree
520
:param new_path: Path of the file in the new tree
521
:param old_kind: Old file-kind of the file
522
:param new_kind: New file-kind of the file
524
if None in (old_kind, new_kind):
525
return DiffPath.CANNOT_DIFF
526
result = DiffPath._diff_many(self.differs, file_id, old_path,
527
new_path, old_kind, None)
528
if result is DiffPath.CANNOT_DIFF:
530
return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
534
class DiffDirectory(DiffPath):
536
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
537
"""Perform comparison between two directories. (dummy)
540
if 'directory' not in (old_kind, new_kind):
541
return self.CANNOT_DIFF
542
if old_kind not in ('directory', None):
543
return self.CANNOT_DIFF
544
if new_kind not in ('directory', None):
545
return self.CANNOT_DIFF
549
class DiffSymlink(DiffPath):
551
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
552
"""Perform comparison between two symlinks
554
:param file_id: The file_id of the file to compare
555
:param old_path: Path of the file in the old tree
556
:param new_path: Path of the file in the new tree
557
:param old_kind: Old file-kind of the file
558
:param new_kind: New file-kind of the file
560
if 'symlink' not in (old_kind, new_kind):
561
return self.CANNOT_DIFF
562
if old_kind == 'symlink':
563
old_target = self.old_tree.get_symlink_target(file_id)
564
elif old_kind is None:
567
return self.CANNOT_DIFF
568
if new_kind == 'symlink':
569
new_target = self.new_tree.get_symlink_target(file_id)
570
elif new_kind is None:
573
return self.CANNOT_DIFF
574
return self.diff_symlink(old_target, new_target)
576
def diff_symlink(self, old_target, new_target):
577
if old_target is None:
578
self.to_file.write('=== target is %r\n' % new_target)
579
elif new_target is None:
580
self.to_file.write('=== target was %r\n' % old_target)
582
self.to_file.write('=== target changed %r => %r\n' %
583
(old_target, new_target))
587
class DiffText(DiffPath):
589
# GNU Patch uses the epoch date to detect files that are being added
590
# or removed in a diff.
591
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
593
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
594
old_label='', new_label='', text_differ=internal_diff):
595
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
596
self.text_differ = text_differ
597
self.old_label = old_label
598
self.new_label = new_label
599
self.path_encoding = path_encoding
601
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
602
"""Compare two files in unified diff format
604
:param file_id: The file_id of the file to compare
605
:param old_path: Path of the file in the old tree
606
:param new_path: Path of the file in the new tree
607
:param old_kind: Old file-kind of the file
608
:param new_kind: New file-kind of the file
610
if 'file' not in (old_kind, new_kind):
611
return self.CANNOT_DIFF
612
from_file_id = to_file_id = file_id
613
if old_kind == 'file':
614
old_date = _patch_header_date(self.old_tree, file_id, old_path)
615
elif old_kind is None:
616
old_date = self.EPOCH_DATE
619
return self.CANNOT_DIFF
620
if new_kind == 'file':
621
new_date = _patch_header_date(self.new_tree, file_id, new_path)
622
elif new_kind is None:
623
new_date = self.EPOCH_DATE
626
return self.CANNOT_DIFF
627
from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
628
to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
629
return self.diff_text(from_file_id, to_file_id, from_label, to_label)
631
def diff_text(self, from_file_id, to_file_id, from_label, to_label):
632
"""Diff the content of given files in two trees
634
:param from_file_id: The id of the file in the from tree. If None,
635
the file is not present in the from tree.
636
:param to_file_id: The id of the file in the to tree. This may refer
637
to a different file from from_file_id. If None,
638
the file is not present in the to tree.
640
def _get_text(tree, file_id):
641
if file_id is not None:
642
return tree.get_file(file_id).readlines()
646
from_text = _get_text(self.old_tree, from_file_id)
647
to_text = _get_text(self.new_tree, to_file_id)
648
self.text_differ(from_label, from_text, to_label, to_text,
650
except errors.BinaryFile:
652
("Binary files %s and %s differ\n" %
653
(from_label, to_label)).encode(self.path_encoding))
657
class DiffFromTool(DiffPath):
659
def __init__(self, command_template, old_tree, new_tree, to_file,
660
path_encoding='utf-8'):
661
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
662
self.command_template = command_template
663
self._root = osutils.mkdtemp(prefix='bzr-diff-')
666
def from_string(klass, command_string, old_tree, new_tree, to_file,
667
path_encoding='utf-8'):
668
command_template = commands.shlex_split_unicode(command_string)
669
command_template.extend(['%(old_path)s', '%(new_path)s'])
670
return klass(command_template, old_tree, new_tree, to_file,
674
def make_from_diff_tree(klass, command_string):
675
def from_diff_tree(diff_tree):
676
return klass.from_string(command_string, diff_tree.old_tree,
677
diff_tree.new_tree, diff_tree.to_file)
678
return from_diff_tree
680
def _get_command(self, old_path, new_path):
681
my_map = {'old_path': old_path, 'new_path': new_path}
682
return [t % my_map for t in self.command_template]
684
def _execute(self, old_path, new_path):
685
command = self._get_command(old_path, new_path)
687
proc = subprocess.Popen(command, stdout=subprocess.PIPE,
690
if e.errno == errno.ENOENT:
691
raise errors.ExecutableMissing(command[0])
694
self.to_file.write(proc.stdout.read())
697
def _try_symlink_root(self, tree, prefix):
698
if (getattr(tree, 'abspath', None) is None
699
or not osutils.host_os_dereferences_symlinks()):
702
os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
704
if e.errno != errno.EEXIST:
708
def _write_file(self, file_id, tree, prefix, relpath):
709
full_path = osutils.pathjoin(self._root, prefix, relpath)
710
if self._try_symlink_root(tree, prefix):
712
parent_dir = osutils.dirname(full_path)
714
os.makedirs(parent_dir)
716
if e.errno != errno.EEXIST:
718
source = tree.get_file(file_id, relpath)
720
target = open(full_path, 'wb')
722
osutils.pumpfile(source, target)
727
osutils.make_readonly(full_path)
728
mtime = tree.get_file_mtime(file_id)
729
os.utime(full_path, (mtime, mtime))
732
def _prepare_files(self, file_id, old_path, new_path):
733
old_disk_path = self._write_file(file_id, self.old_tree, 'old',
735
new_disk_path = self._write_file(file_id, self.new_tree, 'new',
737
return old_disk_path, new_disk_path
740
osutils.rmtree(self._root)
742
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
743
if (old_kind, new_kind) != ('file', 'file'):
744
return DiffPath.CANNOT_DIFF
745
self._prepare_files(file_id, old_path, new_path)
746
self._execute(osutils.pathjoin('old', old_path),
747
osutils.pathjoin('new', new_path))
750
class DiffTree(object):
751
"""Provides textual representations of the difference between two trees.
753
A DiffTree examines two trees and where a file-id has altered
754
between them, generates a textual representation of the difference.
755
DiffTree uses a sequence of DiffPath objects which are each
756
given the opportunity to handle a given altered fileid. The list
757
of DiffPath objects can be extended globally by appending to
758
DiffTree.diff_factories, or for a specific diff operation by
759
supplying the extra_factories option to the appropriate method.
762
# list of factories that can provide instances of DiffPath objects
763
# may be extended by plugins.
764
diff_factories = [DiffSymlink.from_diff_tree,
765
DiffDirectory.from_diff_tree]
767
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
768
diff_text=None, extra_factories=None):
771
:param old_tree: Tree to show as old in the comparison
772
:param new_tree: Tree to show as new in the comparison
773
:param to_file: File to write comparision to
774
:param path_encoding: Character encoding to write paths in
775
:param diff_text: DiffPath-type object to use as a last resort for
777
:param extra_factories: Factories of DiffPaths to try before any other
779
if diff_text is None:
780
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
781
'', '', internal_diff)
782
self.old_tree = old_tree
783
self.new_tree = new_tree
784
self.to_file = to_file
785
self.path_encoding = path_encoding
787
if extra_factories is not None:
788
self.differs.extend(f(self) for f in extra_factories)
789
self.differs.extend(f(self) for f in self.diff_factories)
790
self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
793
def from_trees_options(klass, old_tree, new_tree, to_file,
794
path_encoding, external_diff_options, old_label,
796
"""Factory for producing a DiffTree.
798
Designed to accept options used by show_diff_trees.
799
:param old_tree: The tree to show as old in the comparison
800
:param new_tree: The tree to show as new in the comparison
801
:param to_file: File to write comparisons to
802
:param path_encoding: Character encoding to use for writing paths
803
:param external_diff_options: If supplied, use the installed diff
804
binary to perform file comparison, using supplied options.
805
:param old_label: Prefix to use for old file labels
806
:param new_label: Prefix to use for new file labels
807
:param using: Commandline to use to invoke an external diff tool
809
if using is not None:
810
extra_factories = [DiffFromTool.make_from_diff_tree(using)]
813
if external_diff_options:
814
opts = external_diff_options.split()
815
def diff_file(olab, olines, nlab, nlines, to_file):
816
external_diff(olab, olines, nlab, nlines, to_file, opts)
818
diff_file = internal_diff
819
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
820
old_label, new_label, diff_file)
821
return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
824
def show_diff(self, specific_files, extra_trees=None):
825
"""Write tree diff to self.to_file
827
:param sepecific_files: the specific files to compare (recursive)
828
:param extra_trees: extra trees to use for mapping paths to file_ids
831
return self._show_diff(specific_files, extra_trees)
833
for differ in self.differs:
836
def _show_diff(self, specific_files, extra_trees):
837
# TODO: Generation of pseudo-diffs for added/deleted files could
838
# be usefully made into a much faster special case.
839
iterator = self.new_tree.iter_changes(self.old_tree,
840
specific_files=specific_files,
841
extra_trees=extra_trees,
842
require_versioned=True)
844
def changes_key(change):
845
old_path, new_path = change[1]
850
def get_encoded_path(path):
852
return path.encode(self.path_encoding, "replace")
853
for (file_id, paths, changed_content, versioned, parent, name, kind,
854
executable) in sorted(iterator, key=changes_key):
855
# The root does not get diffed, and items with no known kind (that
856
# is, missing) in both trees are skipped as well.
857
if parent == (None, None) or kind == (None, None):
859
oldpath, newpath = paths
860
oldpath_encoded = get_encoded_path(paths[0])
861
newpath_encoded = get_encoded_path(paths[1])
862
old_present = (kind[0] is not None and versioned[0])
863
new_present = (kind[1] is not None and versioned[1])
864
renamed = (parent[0], name[0]) != (parent[1], name[1])
866
properties_changed = []
867
properties_changed.extend(get_executable_change(executable[0], executable[1]))
869
if properties_changed:
870
prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
874
if (old_present, new_present) == (True, False):
875
self.to_file.write("=== removed %s '%s'\n" %
876
(kind[0], oldpath_encoded))
878
elif (old_present, new_present) == (False, True):
879
self.to_file.write("=== added %s '%s'\n" %
880
(kind[1], newpath_encoded))
883
self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
884
(kind[0], oldpath_encoded, newpath_encoded, prop_str))
886
# if it was produced by iter_changes, it must be
887
# modified *somehow*, either content or execute bit.
888
self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
889
newpath_encoded, prop_str))
891
self.diff(file_id, oldpath, newpath)
897
def diff(self, file_id, old_path, new_path):
898
"""Perform a diff of a single file
900
:param file_id: file-id of the file
901
:param old_path: The path of the file in the old tree
902
:param new_path: The path of the file in the new tree
905
old_kind = self.old_tree.kind(file_id)
906
except (errors.NoSuchId, errors.NoSuchFile):
909
new_kind = self.new_tree.kind(file_id)
910
except (errors.NoSuchId, errors.NoSuchFile):
913
result = DiffPath._diff_many(self.differs, file_id, old_path,
914
new_path, old_kind, new_kind)
915
if result is DiffPath.CANNOT_DIFF:
916
error_path = new_path
917
if error_path is None:
918
error_path = old_path
919
raise errors.NoDiffFound(error_path)
added
removed
bzrlib/diff.py
