« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 856
- compare with another revision
- download diff
- download tarball
- view history from revision 856
- Committer: Martin Pool
- Date: 2005-07-07 10:31:36 UTC
- Revision ID: mbp@sourcefrog.net-20050707103135-9b4d911d8df6e880
- fix pwk help
- files added:
- bzr-man.py
- bzrlib/selftest
- effbot
- effbot/org
- patches
- plugins
- plugins/changeset
- tools
- files removed:
- bzrlib/tests.py
- files renamed:
- contrib/bash/bzr => contrib/bash/bzr.simple
- 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
20
18
from trace import mutter
21
19
from errors import BzrError
22
20
23
21
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
This only compares the versioned files, paying no attention to
31
files which are ignored or unknown. Those can only be present in
32
working trees and can be reported on separately.
33
34
Yields a sequence of (state, id, old_name, new_name, kind).
35
Each filename and each id is listed only once.
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.
36
166
"""
37
## TODO: Allow specifying a list of files to compare, rather than
38
## doing the whole tree? (Not urgent.)
39
40
## TODO: Allow diffing any two inventories, not just the
41
## current one against one. We mgiht need to specify two
42
## stores to look for the files if diffing two branches. That
43
## might imply this shouldn't be primarily a Branch method.
44
45
sha_match_cnt = modified_cnt = 0
46
47
old_it = old_tree.list_files()
48
new_it = new_tree.list_files()
49
50
def next(it):
51
try:
52
return it.next()
53
except StopIteration:
54
return None
55
56
old_item = next(old_it)
57
new_item = next(new_it)
58
59
# We step through the two sorted iterators in parallel, trying to
60
# keep them lined up.
61
62
while (old_item != None) or (new_item != None):
63
# OK, we still have some remaining on both, but they may be
64
# out of step.
65
if old_item != None:
66
old_name, old_class, old_kind, old_id = old_item
67
else:
68
old_name = None
69
70
if new_item != None:
71
new_name, new_class, new_kind, new_id = new_item
72
else:
73
new_name = None
74
75
if old_item:
76
# can't handle the old tree being a WorkingTree
77
assert old_class == 'V'
78
79
if new_item and (new_class != 'V'):
80
yield new_class, None, None, new_name, new_kind
81
new_item = next(new_it)
82
elif (not new_item) or (old_item and (old_name < new_name)):
83
if new_tree.has_id(old_id):
84
# will be mentioned as renamed under new name
85
pass
86
else:
87
yield 'D', old_id, old_name, None, old_kind
88
old_item = next(old_it)
89
elif (not old_item) or (new_item and (new_name < old_name)):
90
if old_tree.has_id(new_id):
91
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
92
else:
93
yield 'A', new_id, None, new_name, new_kind
94
new_item = next(new_it)
95
elif old_id != new_id:
96
assert old_name == new_name
97
# both trees have a file of this name, but it is not the
98
# same file. in other words, the old filename has been
99
# overwritten by either a newly-added or a renamed file.
100
# (should we return something about the overwritten file?)
101
if old_tree.has_id(new_id):
102
# renaming, overlying a deleted file
103
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
104
else:
105
yield 'A', new_id, None, new_name, new_kind
106
107
new_item = next(new_it)
108
old_item = next(old_it)
109
else:
110
assert old_id == new_id
111
assert old_id != None
112
assert old_name == new_name
113
assert old_kind == new_kind
114
115
if old_kind == 'directory':
116
yield '.', new_id, old_name, new_name, new_kind
117
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
118
sha_match_cnt += 1
119
yield '.', new_id, old_name, new_name, new_kind
120
else:
121
modified_cnt += 1
122
yield 'M', new_id, old_name, new_name, new_kind
123
124
new_item = next(new_it)
125
old_item = next(old_it)
126
127
128
mutter("diff finished: %d SHA matches, %d modified"
129
% (sha_match_cnt, modified_cnt))
130
131
132
133
def show_diff(b, revision, file_list):
134
import difflib, sys, types
135
167
import sys
168
136
169
if revision == None:
137
170
old_tree = b.basis_tree()
138
171
else:
140
173
141
174
new_tree = b.working_tree()
142
175
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
143
192
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
144
193
old_label = ''
145
194
new_label = ''
152
201
# TODO: Generation of pseudo-diffs for added/deleted files could
153
202
# be usefully made into a much faster special case.
154
203
155
# TODO: Better to return them in sorted order I think.
156
157
if file_list:
158
file_list = [b.relpath(f) for f in file_list]
159
160
# FIXME: If given a file list, compare only those files rather
161
# than comparing everything and then throwing stuff away.
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
162
211
163
for file_state, fid, old_name, new_name, kind in diff_trees(old_tree, new_tree):
164
165
if file_list and (new_name not in file_list):
166
continue
167
168
# Don't show this by default; maybe do it if an option is passed
169
# idlabel = ' {%s}' % fid
170
idlabel = ''
171
172
def diffit(oldlines, newlines, **kw):
173
174
# FIXME: difflib is wrong if there is no trailing newline.
175
# The syntax used by patch seems to be "\ No newline at
176
# end of file" following the last diff line from that
177
# file. This is not trivial to insert into the
178
# unified_diff output and it might be better to just fix
179
# or replace that function.
180
181
# In the meantime we at least make sure the patch isn't
182
# mangled.
183
184
185
# Special workaround for Python2.3, where difflib fails if
186
# both sequences are empty.
187
if not oldlines and not newlines:
188
return
189
190
nonl = False
191
192
if oldlines and (oldlines[-1][-1] != '\n'):
193
oldlines[-1] += '\n'
194
nonl = True
195
if newlines and (newlines[-1][-1] != '\n'):
196
newlines[-1] += '\n'
197
nonl = True
198
199
ud = difflib.unified_diff(oldlines, newlines, **kw)
200
201
# work-around for difflib being too smart for its own good
202
# if /dev/null is "1,0", patch won't recognize it as /dev/null
203
if not oldlines:
204
ud = list(ud)
205
ud[2] = ud[2].replace('-1,0', '-0,0')
206
elif not newlines:
207
ud = list(ud)
208
ud[2] = ud[2].replace('+1,0', '+0,0')
209
210
sys.stdout.writelines(ud)
211
if nonl:
212
print "\\ No newline at end of file"
213
sys.stdout.write('\n')
214
215
if file_state in ['.', '?', 'I']:
216
continue
217
elif file_state == 'A':
218
print '*** added %s %r' % (kind, new_name)
219
if kind == 'file':
220
diffit([],
221
new_tree.get_file(fid).readlines(),
222
fromfile=DEVNULL,
223
tofile=new_label + new_name + idlabel)
224
elif file_state == 'D':
225
assert isinstance(old_name, types.StringTypes)
226
print '*** deleted %s %r' % (kind, old_name)
227
if kind == 'file':
228
diffit(old_tree.get_file(fid).readlines(), [],
229
fromfile=old_label + old_name + idlabel,
230
tofile=DEVNULL)
231
elif file_state in ['M', 'R']:
232
if file_state == 'M':
233
assert kind == 'file'
234
assert old_name == new_name
235
print '*** modified %s %r' % (kind, new_name)
236
elif file_state == 'R':
237
print '*** renamed %s %r => %r' % (kind, old_name, new_name)
238
239
if kind == 'file':
240
diffit(old_tree.get_file(fid).readlines(),
241
new_tree.get_file(fid).readlines(),
242
fromfile=old_label + old_name + idlabel,
243
tofile=new_label + new_name)
244
else:
245
raise BzrError("can't represent state %s {%s}" % (file_state, fid))
246
247
248
249
class TreeDelta:
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):
250
255
"""Describes changes from one tree to another.
251
256
252
257
Contains four lists:
253
258
254
259
added
255
(path, id)
260
(path, id, kind)
256
261
removed
257
(path, id)
262
(path, id, kind)
258
263
renamed
259
(oldpath, newpath, id, text_modified)
264
(oldpath, newpath, id, kind, text_modified)
260
265
modified
261
(path, id)
266
(path, id, kind)
262
267
unchanged
263
(path, id)
268
(path, id, kind)
264
269
265
270
Each id is listed only once.
266
271
267
272
Files that are both modified and renamed are listed only in
268
273
renamed, with the text_modified flag true.
269
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
270
279
The lists are normally sorted when the delta is created.
271
280
"""
272
281
def __init__(self):
276
285
self.modified = []
277
286
self.unchanged = []
278
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
279
322
def show(self, to_file, show_ids=False, show_unchanged=False):
280
323
def show_list(files):
281
for path, fid in files:
324
for path, fid, kind in files:
325
if kind == 'directory':
326
path += '/'
327
elif kind == 'symlink':
328
path += '@'
329
282
330
if show_ids:
283
331
print >>to_file, ' %-30s %s' % (path, fid)
284
332
else:
285
333
print >>to_file, ' ', path
286
334
287
335
if self.removed:
288
print >>to_file, 'removed files:'
336
print >>to_file, 'removed:'
289
337
show_list(self.removed)
290
338
291
339
if self.added:
292
print >>to_file, 'added files:'
340
print >>to_file, 'added:'
293
341
show_list(self.added)
294
342
295
343
if self.renamed:
296
print >>to_file, 'renamed files:'
297
for oldpath, newpath, fid, text_modified in self.renamed:
344
print >>to_file, 'renamed:'
345
for oldpath, newpath, fid, kind, text_modified in self.renamed:
298
346
if show_ids:
299
347
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
300
348
else:
301
349
print >>to_file, ' %s => %s' % (oldpath, newpath)
302
350
303
351
if self.modified:
304
print >>to_file, 'modified files:'
352
print >>to_file, 'modified:'
305
353
show_list(self.modified)
306
354
307
355
if show_unchanged and self.unchanged:
308
print >>to_file, 'unchanged files:'
356
print >>to_file, 'unchanged:'
309
357
show_list(self.unchanged)
310
358
311
359
312
360
313
def compare_trees(old_tree, new_tree, want_unchanged):
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.
378
"""
379
380
from osutils import is_inside_any
381
314
382
old_inv = old_tree.inventory
315
383
new_inv = new_tree.inventory
316
384
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
317
390
for file_id in old_tree:
318
391
if file_id in new_tree:
319
old_path = old_inv.id2path(file_id)
320
new_path = new_inv.id2path(file_id)
321
322
392
kind = old_inv.get_file_kind(file_id)
393
assert kind == new_inv.get_file_kind(file_id)
394
323
395
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
324
396
'invalid file kind %r' % kind
397
398
if kind == 'root_directory':
399
continue
400
401
old_path = old_inv.id2path(file_id)
402
new_path = new_inv.id2path(file_id)
403
404
old_ie = old_inv[file_id]
405
new_ie = new_inv[file_id]
406
407
if specific_files:
408
if (not is_inside_any(specific_files, old_path)
409
and not is_inside_any(specific_files, new_path)):
410
continue
411
325
412
if kind == 'file':
326
413
old_sha1 = old_tree.get_file_sha1(file_id)
327
414
new_sha1 = new_tree.get_file_sha1(file_id)
335
422
# the same and the parents are unchanged all the way up.
336
423
# May not be worthwhile.
337
424
338
if old_path != new_path:
339
delta.renamed.append((old_path, new_path, file_id, text_modified))
425
if (old_ie.name != new_ie.name
426
or old_ie.parent_id != new_ie.parent_id):
427
delta.renamed.append((old_path, new_path, file_id, kind,
428
text_modified))
340
429
elif text_modified:
341
delta.modified.append((new_path, file_id))
430
delta.modified.append((new_path, file_id, kind))
342
431
elif want_unchanged:
343
delta.unchanged.append((new_path, file_id))
432
delta.unchanged.append((new_path, file_id, kind))
344
433
else:
345
delta.removed.append((old_inv.id2path(file_id), file_id))
434
kind = old_inv.get_file_kind(file_id)
435
old_path = old_inv.id2path(file_id)
436
if specific_files:
437
if not is_inside_any(specific_files, old_path):
438
continue
439
delta.removed.append((old_path, file_id, kind))
440
441
mutter('start looking for new files')
346
442
for file_id in new_inv:
347
443
if file_id in old_inv:
348
444
continue
349
delta.added.append((new_inv.id2path(file_id), file_id))
445
new_path = new_inv.id2path(file_id)
446
if specific_files:
447
if not is_inside_any(specific_files, new_path):
448
continue
449
kind = new_inv.get_file_kind(file_id)
450
delta.added.append((new_path, file_id, kind))
350
451
351
452
delta.removed.sort()
352
453
delta.added.sort()
353
454
delta.renamed.sort()
354
455
delta.modified.sort()
456
delta.unchanged.sort()
355
457
356
458
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1