2
# -*- coding: UTF-8 -*-
1
# Copyright (C) 2005-2011 Canonical Ltd.
4
3
# This program is free software; you can redistribute it and/or modify
5
4
# it under the terms of the GNU General Public License as published by
6
5
# the Free Software Foundation; either version 2 of the License, or
7
6
# (at your option) any later version.
9
8
# This program is distributed in the hope that it will be useful,
10
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
11
# GNU General Public License for more details.
14
13
# You should have received a copy of the GNU General Public License
15
14
# along with this program; if not, write to the Free Software
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20
from trace import mutter
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
17
from __future__ import absolute_import
25
from bzrlib.lazy_import import lazy_import
26
lazy_import(globals(), """
43
from bzrlib.workingtree import WorkingTree
44
from bzrlib.i18n import gettext
47
from bzrlib.registry import (
50
from bzrlib.trace import mutter, note, warning
52
DEFAULT_CONTEXT_AMOUNT = 3
54
class AtTemplate(string.Template):
55
"""Templating class that uses @ instead of $."""
60
# TODO: Rather than building a changeset object, we should probably
61
# invoke callbacks on an object. That object can either accumulate a
62
# list, write them out directly, etc etc.
65
class _PrematchedMatcher(difflib.SequenceMatcher):
66
"""Allow SequenceMatcher operations to use predetermined blocks"""
68
def __init__(self, matching_blocks):
69
difflib.SequenceMatcher(self, None, None)
70
self.matching_blocks = matching_blocks
74
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
75
allow_binary=False, sequence_matcher=None,
76
path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
77
# FIXME: difflib is wrong if there is no trailing newline.
78
# The syntax used by patch seems to be "\ No newline at
79
# end of file" following the last diff line from that
80
# file. This is not trivial to insert into the
81
# unified_diff output and it might be better to just fix
82
# or replace that function.
84
# In the meantime we at least make sure the patch isn't
88
# Special workaround for Python2.3, where difflib fails if
89
# both sequences are empty.
90
if not oldlines and not newlines:
93
if allow_binary is False:
94
textfile.check_text_lines(oldlines)
95
textfile.check_text_lines(newlines)
97
if sequence_matcher is None:
98
sequence_matcher = patiencediff.PatienceSequenceMatcher
99
ud = patiencediff.unified_diff(oldlines, newlines,
100
fromfile=old_filename.encode(path_encoding, 'replace'),
101
tofile=new_filename.encode(path_encoding, 'replace'),
102
n=context_lines, sequencematcher=sequence_matcher)
105
if len(ud) == 0: # Identical contents, nothing to do
107
# work-around for difflib being too smart for its own good
108
# if /dev/null is "1,0", patch won't recognize it as /dev/null
110
ud[2] = ud[2].replace('-1,0', '-0,0')
112
ud[2] = ud[2].replace('+1,0', '+0,0')
116
if not line.endswith('\n'):
117
to_file.write("\n\\ No newline at end of file\n")
121
def _spawn_external_diff(diffcmd, capture_errors=True):
122
"""Spawn the externall diff process, and return the child handle.
124
:param diffcmd: The command list to spawn
125
:param capture_errors: Capture stderr as well as setting LANG=C
126
and LC_ALL=C. This lets us read and understand the output of diff,
127
and respond to any errors.
128
:return: A Popen object.
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()
131
# construct minimal environment
133
path = os.environ.get('PATH')
136
env['LANGUAGE'] = 'C' # on win32 only LANGUAGE has effect
139
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:
145
pipe = subprocess.Popen(diffcmd,
146
stdin=subprocess.PIPE,
147
stdout=subprocess.PIPE,
151
if e.errno == errno.ENOENT:
152
raise errors.NoDiff(str(e))
158
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
160
"""Display a diff by calling out to the external diff program."""
161
# make sure our own output is properly ordered before the diff
164
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
165
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
166
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
167
newtmpf = os.fdopen(newtmp_fd, 'wb')
170
# TODO: perhaps a special case for comparing to or from the empty
171
# sequence; can just use /dev/null on Unix
173
# TODO: if either of the files being compared already exists as a
174
# regular named file (e.g. in the working directory) then we can
175
# compare directly to that, rather than copying it.
177
oldtmpf.writelines(oldlines)
178
newtmpf.writelines(newlines)
185
if sys.platform == 'win32':
186
# Popen doesn't do the proper encoding for external commands
187
# Since we are dealing with an ANSI api, use mbcs encoding
188
old_filename = old_filename.encode('mbcs')
189
new_filename = new_filename.encode('mbcs')
191
'--label', old_filename,
193
'--label', new_filename,
198
# diff only allows one style to be specified; they don't override.
199
# note that some of these take optargs, and the optargs can be
200
# directly appended to the options.
201
# this is only an approximate parser; it doesn't properly understand
203
for s in ['-c', '-u', '-C', '-U',
208
'-y', '--side-by-side',
220
diffcmd.extend(diff_opts)
222
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
223
out,err = pipe.communicate()
226
# internal_diff() adds a trailing newline, add one here for consistency
229
# 'diff' gives retcode == 2 for all sorts of errors
230
# one of those is 'Binary files differ'.
231
# Bad options could also be the problem.
232
# 'Binary files' is not a real error, so we suppress that error.
235
# Since we got here, we want to make sure to give an i18n error
236
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
237
out, err = pipe.communicate()
239
# Write out the new i18n diff response
240
to_file.write(out+'\n')
241
if pipe.returncode != 2:
242
raise errors.BzrError(
243
'external diff failed with exit code 2'
244
' when run with LANG=C and LC_ALL=C,'
245
' but not when run natively: %r' % (diffcmd,))
247
first_line = lang_c_out.split('\n', 1)[0]
248
# Starting with diffutils 2.8.4 the word "binary" was dropped.
249
m = re.match('^(binary )?files.*differ$', first_line, re.I)
251
raise errors.BzrError('external diff failed with exit code 2;'
252
' command: %r' % (diffcmd,))
254
# Binary files differ, just return
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()
257
# If we got to here, we haven't written out the output of diff
261
# returns 1 if files differ; that's OK
263
msg = 'signal %d' % (-rc)
265
msg = 'exit code %d' % rc
267
raise errors.BzrError('external diff failed with %s; command: %r'
272
oldtmpf.close() # and delete
274
# Clean up. Warn in case the files couldn't be deleted
275
# (in case windows still holds the file open, but not
276
# if the files have already been deleted)
278
os.remove(old_abspath)
280
if e.errno not in (errno.ENOENT,):
281
warning('Failed to delete temporary file: %s %s',
284
os.remove(new_abspath)
286
if e.errno not in (errno.ENOENT,):
287
warning('Failed to delete temporary file: %s %s',
291
def get_trees_and_branches_to_diff_locked(
292
path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
293
"""Get the trees and specific files to diff given a list of paths.
295
This method works out the trees to be diff'ed and the files of
296
interest within those trees.
299
the list of arguments passed to the diff command
300
:param revision_specs:
301
Zero, one or two RevisionSpecs from the diff command line,
302
saying what revisions to compare.
304
The url of the old branch or tree. If None, the tree to use is
305
taken from the first path, if any, or the current working tree.
307
The url of the new branch or tree. If None, the tree to use is
308
taken from the first path, if any, or the current working tree.
310
a callable like Command.add_cleanup. get_trees_and_branches_to_diff
311
will register cleanups that must be run to unlock the trees, etc.
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
319
will be read-locked until the cleanups registered via the add_cleanup
322
# Get the old and new revision specs
323
old_revision_spec = None
324
new_revision_spec = None
325
if revision_specs is not None:
326
if len(revision_specs) > 0:
327
old_revision_spec = revision_specs[0]
329
old_url = old_revision_spec.get_branch()
330
if len(revision_specs) > 1:
331
new_revision_spec = revision_specs[1]
333
new_url = new_revision_spec.get_branch()
336
make_paths_wt_relative = True
337
consider_relpath = True
338
if path_list is None or len(path_list) == 0:
339
# If no path is given, the current working tree is used
340
default_location = u'.'
341
consider_relpath = False
342
elif old_url is not None and new_url is not None:
343
other_paths = path_list
344
make_paths_wt_relative = False
346
default_location = path_list[0]
347
other_paths = path_list[1:]
349
def lock_tree_or_branch(wt, br):
352
add_cleanup(wt.unlock)
355
add_cleanup(br.unlock)
357
# Get the old location
360
old_url = default_location
361
working_tree, branch, relpath = \
362
controldir.ControlDir.open_containing_tree_or_branch(old_url)
363
lock_tree_or_branch(working_tree, branch)
364
if consider_relpath and relpath != '':
365
if working_tree is not None and apply_view:
366
views.check_path_in_view(working_tree, relpath)
367
specific_files.append(relpath)
368
old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
371
# Get the new location
373
new_url = default_location
374
if new_url != old_url:
375
working_tree, branch, relpath = \
376
controldir.ControlDir.open_containing_tree_or_branch(new_url)
377
lock_tree_or_branch(working_tree, branch)
378
if consider_relpath and relpath != '':
379
if working_tree is not None and apply_view:
380
views.check_path_in_view(working_tree, relpath)
381
specific_files.append(relpath)
382
new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
383
basis_is_default=working_tree is None)
386
# Get the specific files (all files is None, no files is [])
387
if make_paths_wt_relative and working_tree is not None:
388
other_paths = working_tree.safe_relpath_files(
390
apply_view=apply_view)
391
specific_files.extend(other_paths)
392
if len(specific_files) == 0:
393
specific_files = None
394
if (working_tree is not None and working_tree.supports_views()
396
view_files = working_tree.views.lookup_view()
398
specific_files = view_files
399
view_str = views.view_display_str(view_files)
400
note(gettext("*** Ignoring files outside view. View is %s") % view_str)
402
# Get extra trees that ought to be searched for file-ids
404
if working_tree is not None and working_tree not in (old_tree, new_tree):
405
extra_trees = (working_tree,)
406
return (old_tree, new_tree, old_branch, new_branch,
407
specific_files, extra_trees)
410
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
411
if branch is None and tree is not None:
413
if spec is None or spec.spec is None:
416
return tree.basis_tree()
418
return branch.basis_tree()
421
return spec.as_tree(branch)
424
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
425
external_diff_options=None,
426
old_label='a/', new_label='b/',
428
path_encoding='utf8',
431
context=DEFAULT_CONTEXT_AMOUNT):
432
"""Show in text form the changes from one tree to another.
434
:param to_file: The output stream.
435
:param specific_files: Include only changes to these files - None for all
437
:param external_diff_options: If set, use an external GNU diff and pass
439
:param extra_trees: If set, more Trees to use for looking up file ids
440
:param path_encoding: If set, the path will be encoded as specified,
441
otherwise is supposed to be utf8
442
:param format_cls: Formatter class (DiffTree subclass)
445
context = DEFAULT_CONTEXT_AMOUNT
446
if format_cls is None:
447
format_cls = DiffTree
450
if extra_trees is not None:
451
for tree in extra_trees:
455
differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
457
external_diff_options,
458
old_label, new_label, using,
459
context_lines=context)
460
return differ.show_diff(specific_files, extra_trees)
463
if extra_trees is not None:
464
for tree in extra_trees:
470
def _patch_header_date(tree, file_id, path):
471
"""Returns a timestamp suitable for use in a patch header."""
473
mtime = tree.get_file_mtime(file_id, path)
474
except errors.FileTimestampUnavailable:
476
return timestamp.format_patch_date(mtime)
479
def get_executable_change(old_is_x, new_is_x):
480
descr = { True:"+x", False:"-x", None:"??" }
481
if old_is_x != new_is_x:
482
return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
487
class DiffPath(object):
488
"""Base type for command object that compare files"""
490
# The type or contents of the file were unsuitable for diffing
491
CANNOT_DIFF = 'CANNOT_DIFF'
492
# The file has changed in a semantic way
494
# The file content may have changed, but there is no semantic change
495
UNCHANGED = 'UNCHANGED'
497
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
500
:param old_tree: The tree to show as the old tree in the comparison
501
:param new_tree: The tree to show as new in the comparison
502
:param to_file: The file to write comparison data to
503
:param path_encoding: The character encoding to write paths in
505
self.old_tree = old_tree
506
self.new_tree = new_tree
507
self.to_file = to_file
508
self.path_encoding = path_encoding
514
def from_diff_tree(klass, diff_tree):
515
return klass(diff_tree.old_tree, diff_tree.new_tree,
516
diff_tree.to_file, diff_tree.path_encoding)
519
def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
520
for file_differ in differs:
521
result = file_differ.diff(file_id, old_path, new_path, old_kind,
523
if result is not DiffPath.CANNOT_DIFF:
526
return DiffPath.CANNOT_DIFF
529
class DiffKindChange(object):
530
"""Special differ for file kind changes.
532
Represents kind change as deletion + creation. Uses the other differs
535
def __init__(self, differs):
536
self.differs = differs
542
def from_diff_tree(klass, diff_tree):
543
return klass(diff_tree.differs)
545
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
546
"""Perform comparison
548
:param file_id: The file_id of the file to compare
549
:param old_path: Path of the file in the old tree
550
:param new_path: Path of the file in the new tree
551
:param old_kind: Old file-kind of the file
552
:param new_kind: New file-kind of the file
554
if None in (old_kind, new_kind):
555
return DiffPath.CANNOT_DIFF
556
result = DiffPath._diff_many(self.differs, file_id, old_path,
557
new_path, old_kind, None)
558
if result is DiffPath.CANNOT_DIFF:
560
return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
564
class DiffDirectory(DiffPath):
566
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
567
"""Perform comparison between two directories. (dummy)
570
if 'directory' not in (old_kind, new_kind):
571
return self.CANNOT_DIFF
572
if old_kind not in ('directory', None):
573
return self.CANNOT_DIFF
574
if new_kind not in ('directory', None):
575
return self.CANNOT_DIFF
579
class DiffSymlink(DiffPath):
581
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
582
"""Perform comparison between two symlinks
584
:param file_id: The file_id of the file to compare
585
:param old_path: Path of the file in the old tree
586
:param new_path: Path of the file in the new tree
587
:param old_kind: Old file-kind of the file
588
:param new_kind: New file-kind of the file
590
if 'symlink' not in (old_kind, new_kind):
591
return self.CANNOT_DIFF
592
if old_kind == 'symlink':
593
old_target = self.old_tree.get_symlink_target(file_id)
594
elif old_kind is None:
597
return self.CANNOT_DIFF
598
if new_kind == 'symlink':
599
new_target = self.new_tree.get_symlink_target(file_id)
600
elif new_kind is None:
603
return self.CANNOT_DIFF
604
return self.diff_symlink(old_target, new_target)
606
def diff_symlink(self, old_target, new_target):
607
if old_target is None:
608
self.to_file.write('=== target is %r\n' % new_target)
609
elif new_target is None:
610
self.to_file.write('=== target was %r\n' % old_target)
612
self.to_file.write('=== target changed %r => %r\n' %
613
(old_target, new_target))
617
class DiffText(DiffPath):
619
# GNU Patch uses the epoch date to detect files that are being added
620
# or removed in a diff.
621
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
623
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
624
old_label='', new_label='', text_differ=internal_diff,
625
context_lines=DEFAULT_CONTEXT_AMOUNT):
626
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
627
self.text_differ = text_differ
628
self.old_label = old_label
629
self.new_label = new_label
630
self.path_encoding = path_encoding
631
self.context_lines = context_lines
633
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
634
"""Compare two files in unified diff format
636
:param file_id: The file_id of the file to compare
637
:param old_path: Path of the file in the old tree
638
:param new_path: Path of the file in the new tree
639
:param old_kind: Old file-kind of the file
640
:param new_kind: New file-kind of the file
642
if 'file' not in (old_kind, new_kind):
643
return self.CANNOT_DIFF
644
from_file_id = to_file_id = file_id
645
if old_kind == 'file':
646
old_date = _patch_header_date(self.old_tree, file_id, old_path)
647
elif old_kind is None:
648
old_date = self.EPOCH_DATE
651
return self.CANNOT_DIFF
652
if new_kind == 'file':
653
new_date = _patch_header_date(self.new_tree, file_id, new_path)
654
elif new_kind is None:
655
new_date = self.EPOCH_DATE
658
return self.CANNOT_DIFF
659
from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
660
to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
661
return self.diff_text(from_file_id, to_file_id, from_label, to_label,
664
def diff_text(self, from_file_id, to_file_id, from_label, to_label,
665
from_path=None, to_path=None):
666
"""Diff the content of given files in two trees
668
:param from_file_id: The id of the file in the from tree. If None,
669
the file is not present in the from tree.
670
:param to_file_id: The id of the file in the to tree. This may refer
671
to a different file from from_file_id. If None,
672
the file is not present in the to tree.
673
:param from_path: The path in the from tree or None if unknown.
674
:param to_path: The path in the to tree or None if unknown.
676
def _get_text(tree, file_id, path):
677
if file_id is not None:
678
return tree.get_file_lines(file_id, path)
682
from_text = _get_text(self.old_tree, from_file_id, from_path)
683
to_text = _get_text(self.new_tree, to_file_id, to_path)
684
self.text_differ(from_label, from_text, to_label, to_text,
685
self.to_file, path_encoding=self.path_encoding,
686
context_lines=self.context_lines)
687
except errors.BinaryFile:
689
("Binary files %s and %s differ\n" %
690
(from_label, to_label)).encode(self.path_encoding,'replace'))
694
class DiffFromTool(DiffPath):
696
def __init__(self, command_template, old_tree, new_tree, to_file,
697
path_encoding='utf-8'):
698
DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
699
self.command_template = command_template
700
self._root = osutils.mkdtemp(prefix='bzr-diff-')
703
def from_string(klass, command_string, old_tree, new_tree, to_file,
704
path_encoding='utf-8'):
705
command_template = cmdline.split(command_string)
706
if '@' not in command_string:
707
command_template.extend(['@old_path', '@new_path'])
708
return klass(command_template, old_tree, new_tree, to_file,
712
def make_from_diff_tree(klass, command_string, external_diff_options=None):
713
def from_diff_tree(diff_tree):
714
full_command_string = [command_string]
715
if external_diff_options is not None:
716
full_command_string += ' ' + external_diff_options
717
return klass.from_string(full_command_string, diff_tree.old_tree,
718
diff_tree.new_tree, diff_tree.to_file)
719
return from_diff_tree
721
def _get_command(self, old_path, new_path):
722
my_map = {'old_path': old_path, 'new_path': new_path}
723
command = [AtTemplate(t).substitute(my_map) for t in
724
self.command_template]
725
if sys.platform == 'win32': # Popen doesn't accept unicode on win32
728
if isinstance(c, unicode):
729
command_encoded.append(c.encode('mbcs'))
731
command_encoded.append(c)
732
return command_encoded
736
def _execute(self, old_path, new_path):
737
command = self._get_command(old_path, new_path)
739
proc = subprocess.Popen(command, stdout=subprocess.PIPE,
742
if e.errno == errno.ENOENT:
743
raise errors.ExecutableMissing(command[0])
746
self.to_file.write(proc.stdout.read())
749
def _try_symlink_root(self, tree, prefix):
750
if (getattr(tree, 'abspath', None) is None
751
or not osutils.host_os_dereferences_symlinks()):
754
os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
756
if e.errno != errno.EEXIST:
762
"""Returns safe encoding for passing file path to diff tool"""
763
if sys.platform == 'win32':
766
# Don't fallback to 'utf-8' because subprocess may not be able to
767
# handle utf-8 correctly when locale is not utf-8.
768
return sys.getfilesystemencoding() or 'ascii'
770
def _is_safepath(self, path):
771
"""Return true if `path` may be able to pass to subprocess."""
774
return path == path.encode(fenc).decode(fenc)
778
def _safe_filename(self, prefix, relpath):
779
"""Replace unsafe character in `relpath` then join `self._root`,
780
`prefix` and `relpath`."""
782
# encoded_str.replace('?', '_') may break multibyte char.
783
# So we should encode, decode, then replace(u'?', u'_')
784
relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
785
relpath_tmp = relpath_tmp.replace(u'?', u'_')
786
return osutils.pathjoin(self._root, prefix, relpath_tmp)
788
def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
790
if not force_temp and isinstance(tree, WorkingTree):
791
full_path = tree.abspath(tree.id2path(file_id))
792
if self._is_safepath(full_path):
795
full_path = self._safe_filename(prefix, relpath)
796
if not force_temp and self._try_symlink_root(tree, prefix):
798
parent_dir = osutils.dirname(full_path)
800
os.makedirs(parent_dir)
802
if e.errno != errno.EEXIST:
804
source = tree.get_file(file_id, relpath)
806
target = open(full_path, 'wb')
808
osutils.pumpfile(source, target)
814
mtime = tree.get_file_mtime(file_id)
815
except errors.FileTimestampUnavailable:
818
os.utime(full_path, (mtime, mtime))
820
osutils.make_readonly(full_path)
823
def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
824
allow_write_new=False):
825
old_disk_path = self._write_file(file_id, self.old_tree, 'old',
826
old_path, force_temp)
827
new_disk_path = self._write_file(file_id, self.new_tree, 'new',
828
new_path, force_temp,
829
allow_write=allow_write_new)
830
return old_disk_path, new_disk_path
834
osutils.rmtree(self._root)
836
if e.errno != errno.ENOENT:
837
mutter("The temporary directory \"%s\" was not "
838
"cleanly removed: %s." % (self._root, e))
840
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
841
if (old_kind, new_kind) != ('file', 'file'):
842
return DiffPath.CANNOT_DIFF
843
(old_disk_path, new_disk_path) = self._prepare_files(
844
file_id, old_path, new_path)
845
self._execute(old_disk_path, new_disk_path)
847
def edit_file(self, file_id):
848
"""Use this tool to edit a file.
850
A temporary copy will be edited, and the new contents will be
853
:param file_id: The id of the file to edit.
854
:return: The new contents of the file.
856
old_path = self.old_tree.id2path(file_id)
857
new_path = self.new_tree.id2path(file_id)
858
old_abs_path, new_abs_path = self._prepare_files(
859
file_id, old_path, new_path,
860
allow_write_new=True,
862
command = self._get_command(old_abs_path, new_abs_path)
863
subprocess.call(command, cwd=self._root)
864
new_file = open(new_abs_path, 'rb')
866
return new_file.read()
871
class DiffTree(object):
872
"""Provides textual representations of the difference between two trees.
874
A DiffTree examines two trees and where a file-id has altered
875
between them, generates a textual representation of the difference.
876
DiffTree uses a sequence of DiffPath objects which are each
877
given the opportunity to handle a given altered fileid. The list
878
of DiffPath objects can be extended globally by appending to
879
DiffTree.diff_factories, or for a specific diff operation by
880
supplying the extra_factories option to the appropriate method.
883
# list of factories that can provide instances of DiffPath objects
884
# may be extended by plugins.
885
diff_factories = [DiffSymlink.from_diff_tree,
886
DiffDirectory.from_diff_tree]
888
def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
889
diff_text=None, extra_factories=None):
892
:param old_tree: Tree to show as old in the comparison
893
:param new_tree: Tree to show as new in the comparison
894
:param to_file: File to write comparision to
895
:param path_encoding: Character encoding to write paths in
896
:param diff_text: DiffPath-type object to use as a last resort for
898
:param extra_factories: Factories of DiffPaths to try before any other
900
if diff_text is None:
901
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
902
'', '', internal_diff)
903
self.old_tree = old_tree
904
self.new_tree = new_tree
905
self.to_file = to_file
906
self.path_encoding = path_encoding
908
if extra_factories is not None:
909
self.differs.extend(f(self) for f in extra_factories)
910
self.differs.extend(f(self) for f in self.diff_factories)
911
self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
914
def from_trees_options(klass, old_tree, new_tree, to_file,
915
path_encoding, external_diff_options, old_label,
916
new_label, using, context_lines):
917
"""Factory for producing a DiffTree.
919
Designed to accept options used by show_diff_trees.
921
:param old_tree: The tree to show as old in the comparison
922
:param new_tree: The tree to show as new in the comparison
923
:param to_file: File to write comparisons to
924
:param path_encoding: Character encoding to use for writing paths
925
:param external_diff_options: If supplied, use the installed diff
926
binary to perform file comparison, using supplied options.
927
:param old_label: Prefix to use for old file labels
928
:param new_label: Prefix to use for new file labels
929
:param using: Commandline to use to invoke an external diff tool
931
if using is not None:
932
extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
935
if external_diff_options:
936
opts = external_diff_options.split()
937
def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None, context_lines=None):
938
""":param path_encoding: not used but required
939
to match the signature of internal_diff.
941
external_diff(olab, olines, nlab, nlines, to_file, opts)
943
diff_file = internal_diff
944
diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
945
old_label, new_label, diff_file, context_lines=context_lines)
946
return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
949
def show_diff(self, specific_files, extra_trees=None):
950
"""Write tree diff to self.to_file
952
:param specific_files: the specific files to compare (recursive)
953
:param extra_trees: extra trees to use for mapping paths to file_ids
956
return self._show_diff(specific_files, extra_trees)
958
for differ in self.differs:
961
def _show_diff(self, specific_files, extra_trees):
962
# TODO: Generation of pseudo-diffs for added/deleted files could
963
# be usefully made into a much faster special case.
964
iterator = self.new_tree.iter_changes(self.old_tree,
965
specific_files=specific_files,
966
extra_trees=extra_trees,
967
require_versioned=True)
969
def changes_key(change):
970
old_path, new_path = change[1]
975
def get_encoded_path(path):
977
return path.encode(self.path_encoding, "replace")
978
for (file_id, paths, changed_content, versioned, parent, name, kind,
979
executable) in sorted(iterator, key=changes_key):
980
# The root does not get diffed, and items with no known kind (that
981
# is, missing) in both trees are skipped as well.
982
if parent == (None, None) or kind == (None, None):
984
oldpath, newpath = paths
985
oldpath_encoded = get_encoded_path(paths[0])
986
newpath_encoded = get_encoded_path(paths[1])
987
old_present = (kind[0] is not None and versioned[0])
988
new_present = (kind[1] is not None and versioned[1])
989
renamed = (parent[0], name[0]) != (parent[1], name[1])
991
properties_changed = []
992
properties_changed.extend(get_executable_change(executable[0], executable[1]))
994
if properties_changed:
995
prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
999
if (old_present, new_present) == (True, False):
1000
self.to_file.write("=== removed %s '%s'\n" %
1001
(kind[0], oldpath_encoded))
1003
elif (old_present, new_present) == (False, True):
1004
self.to_file.write("=== added %s '%s'\n" %
1005
(kind[1], newpath_encoded))
1008
self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
1009
(kind[0], oldpath_encoded, newpath_encoded, prop_str))
1011
# if it was produced by iter_changes, it must be
1012
# modified *somehow*, either content or execute bit.
1013
self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
1014
newpath_encoded, prop_str))
1016
self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1022
def diff(self, file_id, old_path, new_path):
1023
"""Perform a diff of a single file
1025
:param file_id: file-id of the file
1026
:param old_path: The path of the file in the old tree
1027
:param new_path: The path of the file in the new tree
1030
old_kind = self.old_tree.kind(file_id)
1031
except (errors.NoSuchId, errors.NoSuchFile):
1034
new_kind = self.new_tree.kind(file_id)
1035
except (errors.NoSuchId, errors.NoSuchFile):
1037
self._diff(file_id, old_path, new_path, old_kind, new_kind)
1040
def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
1041
result = DiffPath._diff_many(self.differs, file_id, old_path,
1042
new_path, old_kind, new_kind)
1043
if result is DiffPath.CANNOT_DIFF:
1044
error_path = new_path
1045
if error_path is None:
1046
error_path = old_path
1047
raise errors.NoDiffFound(error_path)
1050
format_registry = Registry()
1051
format_registry.register('default', DiffTree)
added
removed
bzrlib/diff.py
