« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 439
- compare with another revision
- download diff
- download tarball
- view history from revision 439
- Committer: Martin Pool
- Date: 2005-05-10 07:06:05 UTC
- Revision ID: mbp@sourcefrog.net-20050510070605-0a2453ae5db6fc3c
- new command 'bzr added'
- files added:
- bzrlib/statcache.py
- files removed:
- bzr-man.py
- bzrlib/selftest
- effbot
- effbot/org
- patches
- plugins
- plugins/changeset
- tools
- files renamed:
- contrib/bash/bzr.simple => contrib/bash/bzr
- files modified:
- .bzrignore
added
removed
bzrlib/diff.py
15
15
# along with this program; if not, write to the Free Software
16
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
17
18
from sets import Set
19
18
20
from trace import mutter
19
21
from errors import BzrError
20
22
21
23
22
# TODO: Rather than building a changeset object, we should probably
23
# invoke callbacks on an object. That object can either accumulate a
24
# list, write them out directly, etc etc.
25
26
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
27
import difflib
28
29
# FIXME: difflib is wrong if there is no trailing newline.
30
# The syntax used by patch seems to be "\ No newline at
31
# end of file" following the last diff line from that
32
# file. This is not trivial to insert into the
33
# unified_diff output and it might be better to just fix
34
# or replace that function.
35
36
# In the meantime we at least make sure the patch isn't
37
# mangled.
38
39
40
# Special workaround for Python2.3, where difflib fails if
41
# both sequences are empty.
42
if not oldlines and not newlines:
43
return
44
45
nonl = False
46
47
if oldlines and (oldlines[-1][-1] != '\n'):
48
oldlines[-1] += '\n'
49
nonl = True
50
if newlines and (newlines[-1][-1] != '\n'):
51
newlines[-1] += '\n'
52
nonl = True
53
54
ud = difflib.unified_diff(oldlines, newlines,
55
fromfile=old_label, tofile=new_label)
56
57
# work-around for difflib being too smart for its own good
58
# if /dev/null is "1,0", patch won't recognize it as /dev/null
59
if not oldlines:
60
ud = list(ud)
61
ud[2] = ud[2].replace('-1,0', '-0,0')
62
elif not newlines:
63
ud = list(ud)
64
ud[2] = ud[2].replace('+1,0', '+0,0')
65
66
for line in ud:
67
to_file.write(line)
68
if nonl:
69
print >>to_file, "\\ No newline at end of file"
70
print >>to_file
71
72
73
74
75
def external_diff(old_label, oldlines, new_label, newlines, to_file,
76
diff_opts):
77
"""Display a diff by calling out to the external diff program."""
78
import sys
79
80
if to_file != sys.stdout:
81
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
82
to_file)
83
84
# make sure our own output is properly ordered before the diff
85
to_file.flush()
86
87
from tempfile import NamedTemporaryFile
88
import os
89
90
oldtmpf = NamedTemporaryFile()
91
newtmpf = NamedTemporaryFile()
92
93
try:
94
# TODO: perhaps a special case for comparing to or from the empty
95
# sequence; can just use /dev/null on Unix
96
97
# TODO: if either of the files being compared already exists as a
98
# regular named file (e.g. in the working directory) then we can
99
# compare directly to that, rather than copying it.
100
101
oldtmpf.writelines(oldlines)
102
newtmpf.writelines(newlines)
103
104
oldtmpf.flush()
105
newtmpf.flush()
106
107
if not diff_opts:
108
diff_opts = []
109
diffcmd = ['diff',
110
'--label', old_label,
111
oldtmpf.name,
112
'--label', new_label,
113
newtmpf.name]
114
115
# diff only allows one style to be specified; they don't override.
116
# note that some of these take optargs, and the optargs can be
117
# directly appended to the options.
118
# this is only an approximate parser; it doesn't properly understand
119
# the grammar.
120
for s in ['-c', '-u', '-C', '-U',
121
'-e', '--ed',
122
'-q', '--brief',
123
'--normal',
124
'-n', '--rcs',
125
'-y', '--side-by-side',
126
'-D', '--ifdef']:
127
for j in diff_opts:
128
if j.startswith(s):
129
break
130
else:
131
continue
132
break
133
else:
134
diffcmd.append('-u')
135
136
if diff_opts:
137
diffcmd.extend(diff_opts)
138
139
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
140
141
if rc != 0 and rc != 1:
142
# returns 1 if files differ; that's OK
143
if rc < 0:
144
msg = 'signal %d' % (-rc)
145
else:
146
msg = 'exit code %d' % rc
147
148
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
149
finally:
150
oldtmpf.close() # and delete
151
newtmpf.close()
152
153
154
155
def show_diff(b, revision, specific_files, external_diff_options=None):
156
"""Shortcut for showing the diff to the working tree.
157
158
b
159
Branch.
160
161
revision
162
None for each, or otherwise the old revision to compare against.
163
164
The more general form is show_diff_trees(), where the caller
165
supplies any two trees.
24
def diff_trees(old_tree, new_tree):
25
"""Compute diff between two trees.
26
27
They may be in different branches and may be working or historical
28
trees.
29
30
Yields a sequence of (state, id, old_name, new_name, kind).
31
Each filename and each id is listed only once.
166
32
"""
167
import sys
168
33
34
## TODO: Compare files before diffing; only mention those that have changed
35
36
## TODO: Set nice names in the headers, maybe include diffstat
37
38
## TODO: Perhaps make this a generator rather than using
39
## a callback object?
40
41
## TODO: Allow specifying a list of files to compare, rather than
42
## doing the whole tree? (Not urgent.)
43
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.
48
49
## XXX: This doesn't report on unknown files; that can be done
50
## from a separate method.
51
52
old_it = old_tree.list_files()
53
new_it = new_tree.list_files()
54
55
def next(it):
56
try:
57
return it.next()
58
except StopIteration:
59
return None
60
61
old_item = next(old_it)
62
new_item = next(new_it)
63
64
# We step through the two sorted iterators in parallel, trying to
65
# keep them lined up.
66
67
while (old_item != None) or (new_item != None):
68
# OK, we still have some remaining on both, but they may be
69
# out of step.
70
if old_item != None:
71
old_name, old_class, old_kind, old_id = old_item
72
else:
73
old_name = None
74
75
if new_item != None:
76
new_name, new_class, new_kind, new_id = new_item
77
else:
78
new_name = None
79
80
mutter(" diff pairwise %r" % (old_item,))
81
mutter(" %r" % (new_item,))
82
83
if old_item:
84
# can't handle the old tree being a WorkingTree
85
assert old_class == 'V'
86
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
94
pass
95
else:
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
102
else:
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
114
else:
115
yield 'A', new_id, None, new_name, new_kind
116
117
new_item = next(new_it)
118
old_item = next(old_it)
119
else:
120
assert old_id == new_id
121
assert old_id != None
122
assert old_name == new_name
123
assert old_kind == new_kind
124
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
134
else:
135
mutter(" quick compare shows different")
136
yield 'M', new_id, old_name, new_name, new_kind
137
138
new_item = next(new_it)
139
old_item = next(old_it)
140
141
142
143
def show_diff(b, revision, file_list):
144
import difflib, sys, types
145
169
146
if revision == None:
170
147
old_tree = b.basis_tree()
171
148
else:
173
150
174
151
new_tree = b.working_tree()
175
152
176
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
177
external_diff_options)
178
179
180
181
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
182
external_diff_options=None):
183
"""Show in text form the changes from one tree to another.
184
185
to_files
186
If set, include only changes to these files.
187
188
external_diff_options
189
If set, use an external GNU diff and pass these options.
190
"""
191
192
153
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
193
154
old_label = ''
194
155
new_label = ''
201
162
# TODO: Generation of pseudo-diffs for added/deleted files could
202
163
# be usefully made into a much faster special case.
203
164
204
if external_diff_options:
205
assert isinstance(external_diff_options, basestring)
206
opts = external_diff_options.split()
207
def diff_file(olab, olines, nlab, nlines, to_file):
208
external_diff(olab, olines, nlab, nlines, to_file, opts)
209
else:
210
diff_file = internal_diff
165
# TODO: Better to return them in sorted order I think.
166
167
if file_list:
168
file_list = [b.relpath(f) for f in file_list]
169
170
# FIXME: If given a file list, compare only those files rather
171
# than comparing everything and then throwing stuff away.
211
172
212
213
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
214
specific_files=specific_files)
215
216
for path, file_id, kind in delta.removed:
217
print >>to_file, '*** removed %s %r' % (kind, path)
218
if kind == 'file':
219
diff_file(old_label + path,
220
old_tree.get_file(file_id).readlines(),
221
DEVNULL,
222
[],
223
to_file)
224
225
for path, file_id, kind in delta.added:
226
print >>to_file, '*** added %s %r' % (kind, path)
227
if kind == 'file':
228
diff_file(DEVNULL,
229
[],
230
new_label + path,
231
new_tree.get_file(file_id).readlines(),
232
to_file)
233
234
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
235
print >>to_file, '*** renamed %s %r => %r' % (kind, old_path, new_path)
236
if text_modified:
237
diff_file(old_label + old_path,
238
old_tree.get_file(file_id).readlines(),
239
new_label + new_path,
240
new_tree.get_file(file_id).readlines(),
241
to_file)
242
243
for path, file_id, kind in delta.modified:
244
print >>to_file, '*** modified %s %r' % (kind, path)
245
if kind == 'file':
246
diff_file(old_label + path,
247
old_tree.get_file(file_id).readlines(),
248
new_label + path,
249
new_tree.get_file(file_id).readlines(),
250
to_file)
251
252
253
254
class TreeDelta(object):
173
for file_state, fid, old_name, new_name, kind in diff_trees(old_tree, new_tree):
174
175
if file_list and (new_name not in file_list):
176
continue
177
178
# Don't show this by default; maybe do it if an option is passed
179
# idlabel = ' {%s}' % fid
180
idlabel = ''
181
182
# FIXME: Something about the diff format makes patch unhappy
183
# with newly-added files.
184
185
def diffit(oldlines, newlines, **kw):
186
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.
193
194
# In the meantime we at least make sure the patch isn't
195
# mangled.
196
197
198
# Special workaround for Python2.3, where difflib fails if
199
# both sequences are empty.
200
if not oldlines and not newlines:
201
return
202
203
nonl = False
204
205
if oldlines and (oldlines[-1][-1] != '\n'):
206
oldlines[-1] += '\n'
207
nonl = True
208
if newlines and (newlines[-1][-1] != '\n'):
209
newlines[-1] += '\n'
210
nonl = True
211
212
ud = difflib.unified_diff(oldlines, newlines, **kw)
213
sys.stdout.writelines(ud)
214
if nonl:
215
print "\\ No newline at end of file"
216
sys.stdout.write('\n')
217
218
if file_state in ['.', '?', 'I']:
219
continue
220
elif file_state == 'A':
221
print '*** added %s %r' % (kind, new_name)
222
if kind == 'file':
223
diffit([],
224
new_tree.get_file(fid).readlines(),
225
fromfile=DEVNULL,
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)
230
if kind == 'file':
231
diffit(old_tree.get_file(fid).readlines(), [],
232
fromfile=old_label + old_name + idlabel,
233
tofile=DEVNULL)
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)
241
242
if kind == 'file':
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)
247
else:
248
raise BzrError("can't represent state %s {%s}" % (file_state, fid))
249
250
251
252
class TreeDelta:
255
253
"""Describes changes from one tree to another.
256
254
257
255
Contains four lists:
258
256
259
257
added
260
(path, id, kind)
258
(path, id)
261
259
removed
262
(path, id, kind)
260
(path, id)
263
261
renamed
264
(oldpath, newpath, id, kind, text_modified)
262
(oldpath, newpath, id)
265
263
modified
266
(path, id, kind)
267
unchanged
268
(path, id, kind)
269
270
Each id is listed only once.
271
272
Files that are both modified and renamed are listed only in
273
renamed, with the text_modified flag true.
274
275
Files are only considered renamed if their name has changed or
276
their parent directory has changed. Renaming a directory
277
does not count as renaming all its contents.
278
279
The lists are normally sorted when the delta is created.
264
(path, id)
265
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.
268
269
Files are listed in either modified or renamed, not both. In
270
other words, renamed files may also be modified.
280
271
"""
281
272
def __init__(self):
282
273
self.added = []
283
274
self.removed = []
284
275
self.renamed = []
285
276
self.modified = []
286
self.unchanged = []
287
288
def __eq__(self, other):
289
if not isinstance(other, TreeDelta):
290
return False
291
return self.added == other.added \
292
and self.removed == other.removed \
293
and self.renamed == other.renamed \
294
and self.modified == other.modified \
295
and self.unchanged == other.unchanged
296
297
def __ne__(self, other):
298
return not (self == other)
299
300
def __repr__(self):
301
return "TreeDelta(added=%r, removed=%r, renamed=%r, modified=%r," \
302
" unchanged=%r)" % (self.added, self.removed, self.renamed,
303
self.modified, self.unchanged)
304
305
def has_changed(self):
306
changes = len(self.added) + len(self.removed) + len(self.renamed)
307
changes += len(self.modified)
308
return (changes != 0)
309
310
def touches_file_id(self, file_id):
311
"""Return True if file_id is modified by this delta."""
312
for l in self.added, self.removed, self.modified:
313
for v in l:
314
if v[1] == file_id:
315
return True
316
for v in self.renamed:
317
if v[2] == file_id:
318
return True
319
return False
320
321
322
def show(self, to_file, show_ids=False, show_unchanged=False):
323
def show_list(files):
324
for path, fid, kind in files:
325
if kind == 'directory':
326
path += '/'
327
elif kind == 'symlink':
328
path += '@'
329
330
if show_ids:
331
print >>to_file, ' %-30s %s' % (path, fid)
332
else:
333
print >>to_file, ' ', path
334
335
if self.removed:
336
print >>to_file, 'removed:'
337
show_list(self.removed)
338
339
if self.added:
340
print >>to_file, 'added:'
341
show_list(self.added)
342
343
if self.renamed:
344
print >>to_file, 'renamed:'
345
for oldpath, newpath, fid, kind, text_modified in self.renamed:
346
if show_ids:
347
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
348
else:
349
print >>to_file, ' %s => %s' % (oldpath, newpath)
350
351
if self.modified:
352
print >>to_file, 'modified:'
353
show_list(self.modified)
354
355
if show_unchanged and self.unchanged:
356
print >>to_file, 'unchanged:'
357
show_list(self.unchanged)
358
359
360
361
def compare_trees(old_tree, new_tree, want_unchanged=False, specific_files=None):
362
"""Describe changes from one tree to another.
363
364
Returns a TreeDelta with details of added, modified, renamed, and
365
deleted entries.
366
367
The root entry is specifically exempt.
368
369
This only considers versioned files.
370
371
want_unchanged
372
If true, also list files unchanged from one version to
373
the next.
374
375
specific_files
376
If true, only check for changes to specified names or
377
files within them.
277
278
279
def compare_inventories(old_inv, new_inv):
280
"""Return a TreeDelta object describing changes between inventories.
281
282
This only describes changes in the shape of the tree, not the
283
actual texts.
284
285
This is an alternative to diff_trees() and should probably
286
eventually replace it.
378
287
"""
379
380
from osutils import is_inside_any
381
382
old_inv = old_tree.inventory
383
new_inv = new_tree.inventory
288
old_ids = old_inv.id_set()
289
new_ids = new_inv.id_set()
384
290
delta = TreeDelta()
385
mutter('start compare_trees')
386
387
# TODO: match for specific files can be rather smarter by finding
388
# the IDs of those files up front and then considering only that.
389
390
for file_id in old_tree:
391
if file_id in new_tree:
392
old_ie = old_inv[file_id]
393
new_ie = new_inv[file_id]
394
395
kind = old_ie.kind
396
assert kind == new_ie.kind
397
398
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
399
'invalid file kind %r' % kind
400
401
if kind == 'root_directory':
402
continue
403
404
if specific_files:
405
if (not is_inside_any(specific_files, old_inv.id2path(file_id))
406
and not is_inside_any(specific_files, new_inv.id2path(file_id))):
407
continue
408
409
if kind == 'file':
410
old_sha1 = old_tree.get_file_sha1(file_id)
411
new_sha1 = new_tree.get_file_sha1(file_id)
412
text_modified = (old_sha1 != new_sha1)
413
else:
414
## mutter("no text to check for %r %r" % (file_id, kind))
415
text_modified = False
416
417
# TODO: Can possibly avoid calculating path strings if the
418
# two files are unchanged and their names and parents are
419
# the same and the parents are unchanged all the way up.
420
# May not be worthwhile.
421
422
if (old_ie.name != new_ie.name
423
or old_ie.parent_id != new_ie.parent_id):
424
delta.renamed.append((old_inv.id2path(file_id),
425
new_inv.id2path(file_id),
426
file_id, kind,
427
text_modified))
428
elif text_modified:
429
delta.modified.append((new_inv.id2path(file_id), file_id, kind))
430
elif want_unchanged:
431
delta.unchanged.append((new_inv.id2path(file_id), file_id, kind))
432
else:
433
kind = old_inv.get_file_kind(file_id)
434
old_path = old_inv.id2path(file_id)
435
if specific_files:
436
if not is_inside_any(specific_files, old_path):
437
continue
438
delta.removed.append((old_path, file_id, kind))
439
440
mutter('start looking for new files')
441
for file_id in new_inv:
442
if file_id in old_inv:
443
continue
444
new_path = new_inv.id2path(file_id)
445
if specific_files:
446
if not is_inside_any(specific_files, new_path):
447
continue
448
kind = new_inv.get_file_kind(file_id)
449
delta.added.append((new_path, file_id, kind))
450
291
292
delta.removed = [(old_inv.id2path(fid), fid) for fid in (old_ids - new_ids)]
451
293
delta.removed.sort()
294
295
delta.added = [(new_inv.id2path(fid), fid) for fid in (new_ids - old_ids)]
452
296
delta.added.sort()
453
delta.renamed.sort()
297
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)
303
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))
308
454
309
delta.modified.sort()
455
delta.unchanged.sort()
310
delta.renamed.sort()
456
311
457
312
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1