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
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.
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()
19
from bzrlib.delta import compare_trees
20
from bzrlib.errors import BzrError
21
import bzrlib.errors as errors
22
from bzrlib.patiencediff import unified_diff
23
import bzrlib.patiencediff
24
from bzrlib.symbol_versioning import *
25
from bzrlib.textfile import check_text_lines
26
from bzrlib.trace import mutter
29
# TODO: Rather than building a changeset object, we should probably
30
# invoke callbacks on an object. That object can either accumulate a
31
# list, write them out directly, etc etc.
33
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
34
allow_binary=False, sequence_matcher=None,
35
path_encoding='utf8'):
36
# FIXME: difflib is wrong if there is no trailing newline.
37
# The syntax used by patch seems to be "\ No newline at
38
# end of file" following the last diff line from that
39
# file. This is not trivial to insert into the
40
# unified_diff output and it might be better to just fix
41
# or replace that function.
43
# In the meantime we at least make sure the patch isn't
47
# Special workaround for Python2.3, where difflib fails if
48
# both sequences are empty.
49
if not oldlines and not newlines:
52
if allow_binary is False:
53
check_text_lines(oldlines)
54
check_text_lines(newlines)
56
if sequence_matcher is None:
57
sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
58
ud = unified_diff(oldlines, newlines,
59
fromfile=old_filename.encode(path_encoding),
60
tofile=new_filename.encode(path_encoding),
61
sequencematcher=sequence_matcher)
64
# work-around for difflib being too smart for its own good
65
# if /dev/null is "1,0", patch won't recognize it as /dev/null
67
ud[2] = ud[2].replace('-1,0', '-0,0')
69
ud[2] = ud[2].replace('+1,0', '+0,0')
70
# work around for difflib emitting random spaces after the label
71
ud[0] = ud[0][:-2] + '\n'
72
ud[1] = ud[1][:-2] + '\n'
76
if not line.endswith('\n'):
77
to_file.write("\n\\ No newline at end of file\n")
81
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
83
"""Display a diff by calling out to the external diff program."""
86
if to_file != sys.stdout:
87
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
90
# make sure our own output is properly ordered before the diff
93
from tempfile import NamedTemporaryFile
96
oldtmpf = NamedTemporaryFile()
97
newtmpf = NamedTemporaryFile()
100
# TODO: perhaps a special case for comparing to or from the empty
101
# sequence; can just use /dev/null on Unix
103
# TODO: if either of the files being compared already exists as a
104
# regular named file (e.g. in the working directory) then we can
105
# compare directly to that, rather than copying it.
107
oldtmpf.writelines(oldlines)
108
newtmpf.writelines(newlines)
116
'--label', old_filename,
118
'--label', new_filename,
121
# diff only allows one style to be specified; they don't override.
122
# note that some of these take optargs, and the optargs can be
123
# directly appended to the options.
124
# this is only an approximate parser; it doesn't properly understand
126
for s in ['-c', '-u', '-C', '-U',
131
'-y', '--side-by-side',
143
diffcmd.extend(diff_opts)
145
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
147
if rc != 0 and rc != 1:
148
# returns 1 if files differ; that's OK
150
msg = 'signal %d' % (-rc)
152
msg = 'exit code %d' % rc
154
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
156
oldtmpf.close() # and delete
160
@deprecated_function(zero_eight)
161
def show_diff(b, from_spec, specific_files, external_diff_options=None,
162
revision2=None, output=None, b2=None):
163
"""Shortcut for showing the diff to the working tree.
165
Please use show_diff_trees instead.
171
None for 'basis tree', or otherwise the old revision to compare against.
173
The more general form is show_diff_trees(), where the caller
174
supplies any two trees.
180
if from_spec is None:
181
old_tree = b.bzrdir.open_workingtree()
183
old_tree = old_tree = old_tree.basis_tree()
185
old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
187
if revision2 is None:
189
new_tree = b.bzrdir.open_workingtree()
191
new_tree = b2.bzrdir.open_workingtree()
193
new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
195
return show_diff_trees(old_tree, new_tree, output, specific_files,
196
external_diff_options)
199
def diff_cmd_helper(tree, specific_files, external_diff_options,
200
old_revision_spec=None, new_revision_spec=None,
201
old_label='a/', new_label='b/'):
202
"""Helper for cmd_diff.
208
The specific files to compare, or None
210
external_diff_options
211
If non-None, run an external diff, and pass it these options
214
If None, use basis tree as old revision, otherwise use the tree for
215
the specified revision.
218
If None, use working tree as new revision, otherwise use the tree for
219
the specified revision.
221
The more general form is show_diff_trees(), where the caller
222
supplies any two trees.
227
revision_id = spec.in_store(tree.branch).rev_id
228
return tree.branch.repository.revision_tree(revision_id)
229
if old_revision_spec is None:
230
old_tree = tree.basis_tree()
232
old_tree = spec_tree(old_revision_spec)
234
if new_revision_spec is None:
237
new_tree = spec_tree(new_revision_spec)
239
return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
240
external_diff_options,
241
old_label=old_label, new_label=new_label)
244
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
245
external_diff_options=None,
246
old_label='a/', new_label='b/'):
247
"""Show in text form the changes from one tree to another.
250
If set, include only changes to these files.
252
external_diff_options
253
If set, use an external GNU diff and pass these options.
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):
147
old_tree = b.basis_tree()
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.
259
return _show_diff_trees(old_tree, new_tree, to_file,
260
specific_files, external_diff_options,
261
old_label=old_label, new_label=new_label)
268
def _show_diff_trees(old_tree, new_tree, to_file,
269
specific_files, external_diff_options,
270
old_label='a/', new_label='b/' ):
272
# GNU Patch uses the epoch date to detect files that are being added
273
# or removed in a diff.
274
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
162
276
# TODO: Generation of pseudo-diffs for added/deleted files could
163
277
# 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:
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
bailout("can't represent state %s {%s}" % (file_state, fid))
279
_raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
281
if external_diff_options:
282
assert isinstance(external_diff_options, basestring)
283
opts = external_diff_options.split()
284
def diff_file(olab, olines, nlab, nlines, to_file):
285
external_diff(olab, olines, nlab, nlines, to_file, opts)
287
diff_file = internal_diff
289
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
290
specific_files=specific_files)
293
for path, file_id, kind in delta.removed:
295
print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
296
old_name = '%s%s\t%s' % (old_label, path,
297
_patch_header_date(old_tree, file_id, path))
298
new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
299
old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
300
new_name, None, None, to_file)
301
for path, file_id, kind in delta.added:
303
print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
304
old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
305
new_name = '%s%s\t%s' % (new_label, path,
306
_patch_header_date(new_tree, file_id, path))
307
new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
308
old_name, None, None, to_file,
310
for (old_path, new_path, file_id, kind,
311
text_modified, meta_modified) in delta.renamed:
313
prop_str = get_prop_change(meta_modified)
314
print >>to_file, '=== renamed %s %r => %r%s' % (
315
kind, old_path.encode('utf8'),
316
new_path.encode('utf8'), prop_str)
317
old_name = '%s%s\t%s' % (old_label, old_path,
318
_patch_header_date(old_tree, file_id,
320
new_name = '%s%s\t%s' % (new_label, new_path,
321
_patch_header_date(new_tree, file_id,
323
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
325
text_modified, kind, to_file, diff_file)
326
for path, file_id, kind, text_modified, meta_modified in delta.modified:
328
prop_str = get_prop_change(meta_modified)
329
print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
330
old_name = '%s%s\t%s' % (old_label, path,
331
_patch_header_date(old_tree, file_id, path))
332
new_name = '%s%s\t%s' % (new_label, path,
333
_patch_header_date(new_tree, file_id, path))
335
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
337
True, kind, to_file, diff_file)
342
def _patch_header_date(tree, file_id, path):
343
"""Returns a timestamp suitable for use in a patch header."""
344
tm = time.gmtime(tree.get_file_mtime(file_id, path))
345
return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
348
def _raise_if_doubly_unversioned(specific_files, old_tree, new_tree):
349
"""Complain if paths are not versioned in either tree."""
350
if not specific_files:
352
old_unversioned = old_tree.filter_unversioned_files(specific_files)
353
new_unversioned = new_tree.filter_unversioned_files(specific_files)
354
unversioned = old_unversioned.intersection(new_unversioned)
356
raise errors.PathsNotVersionedError(sorted(unversioned))
359
def _raise_if_nonexistent(paths, old_tree, new_tree):
360
"""Complain if paths are not in either inventory or tree.
362
It's OK with the files exist in either tree's inventory, or
363
if they exist in the tree but are not versioned.
365
This can be used by operations such as bzr status that can accept
366
unknown or ignored files.
368
mutter("check paths: %r", paths)
371
s = old_tree.filter_unversioned_files(paths)
372
s = new_tree.filter_unversioned_files(s)
373
s = [path for path in s if not new_tree.has_filename(path)]
375
raise errors.PathsDoNotExist(sorted(s))
378
def get_prop_change(meta_modified):
380
return " (properties changed)"
385
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
386
new_path, new_tree, text_modified,
387
kind, to_file, diff_file):
389
new_entry = new_tree.inventory[file_id]
390
old_tree.inventory[file_id].diff(diff_file,
added
removed
bzrlib/diff.py
