« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 706
- compare with another revision
- download diff
- download tarball
- view history from revision 706
- Committer: Martin Pool
- Date: 2005-06-20 03:29:21 UTC
- Revision ID: mbp@sourcefrog.net-20050620032921-240d0ade23545055
- remove redundant precursor check
- files added:
- bzr-man.py
- patches
- 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
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
23
import difflib
24
25
# FIXME: difflib is wrong if there is no trailing newline.
26
# The syntax used by patch seems to be "\ No newline at
27
# end of file" following the last diff line from that
28
# file. This is not trivial to insert into the
29
# unified_diff output and it might be better to just fix
30
# or replace that function.
31
32
# In the meantime we at least make sure the patch isn't
33
# mangled.
34
35
36
# Special workaround for Python2.3, where difflib fails if
37
# both sequences are empty.
38
if not oldlines and not newlines:
39
return
40
41
nonl = False
42
43
if oldlines and (oldlines[-1][-1] != '\n'):
44
oldlines[-1] += '\n'
45
nonl = True
46
if newlines and (newlines[-1][-1] != '\n'):
47
newlines[-1] += '\n'
48
nonl = True
49
50
ud = difflib.unified_diff(oldlines, newlines,
51
fromfile=old_label, tofile=new_label)
52
53
# work-around for difflib being too smart for its own good
54
# if /dev/null is "1,0", patch won't recognize it as /dev/null
55
if not oldlines:
56
ud = list(ud)
57
ud[2] = ud[2].replace('-1,0', '-0,0')
58
elif not newlines:
59
ud = list(ud)
60
ud[2] = ud[2].replace('+1,0', '+0,0')
61
62
to_file.writelines(ud)
63
if nonl:
64
print >>to_file, "\\ No newline at end of file"
65
print >>to_file
66
67
68
69
70
def external_diff(old_label, oldlines, new_label, newlines, to_file,
71
diff_opts):
72
"""Display a diff by calling out to the external diff program."""
73
import sys
74
75
if to_file != sys.stdout:
76
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
77
to_file)
78
79
# make sure our own output is properly ordered before the diff
80
to_file.flush()
81
82
from tempfile import NamedTemporaryFile
83
import os
84
85
oldtmpf = NamedTemporaryFile()
86
newtmpf = NamedTemporaryFile()
87
88
try:
89
# TODO: perhaps a special case for comparing to or from the empty
90
# sequence; can just use /dev/null on Unix
91
92
# TODO: if either of the files being compared already exists as a
93
# regular named file (e.g. in the working directory) then we can
94
# compare directly to that, rather than copying it.
95
96
oldtmpf.writelines(oldlines)
97
newtmpf.writelines(newlines)
98
99
oldtmpf.flush()
100
newtmpf.flush()
101
102
if not diff_opts:
103
diff_opts = []
104
diffcmd = ['diff',
105
'--label', old_label,
106
oldtmpf.name,
107
'--label', new_label,
108
newtmpf.name]
109
110
# diff only allows one style to be specified; they don't override.
111
# note that some of these take optargs, and the optargs can be
112
# directly appended to the options.
113
# this is only an approximate parser; it doesn't properly understand
114
# the grammar.
115
for s in ['-c', '-u', '-C', '-U',
116
'-e', '--ed',
117
'-q', '--brief',
118
'--normal',
119
'-n', '--rcs',
120
'-y', '--side-by-side',
121
'-D', '--ifdef']:
122
for j in diff_opts:
123
if j.startswith(s):
124
break
125
else:
126
continue
127
break
128
else:
129
diffcmd.append('-u')
130
131
if diff_opts:
132
diffcmd.extend(diff_opts)
133
134
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
135
136
if rc != 0 and rc != 1:
137
# returns 1 if files differ; that's OK
138
if rc < 0:
139
msg = 'signal %d' % (-rc)
140
else:
141
msg = 'exit code %d' % rc
142
143
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
144
finally:
145
oldtmpf.close() # and delete
146
newtmpf.close()
147
148
149
150
def show_diff(b, revision, specific_files, external_diff_options=None):
151
"""Shortcut for showing the diff to the working tree.
152
153
b
154
Branch.
155
156
revision
157
None for each, or otherwise the old revision to compare against.
158
159
The more general form is show_diff_trees(), where the caller
160
supplies any two trees.
36
161
"""
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
162
import sys
163
136
164
if revision == None:
137
165
old_tree = b.basis_tree()
138
166
else:
140
168
141
169
new_tree = b.working_tree()
142
170
171
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
172
external_diff_options)
173
174
175
176
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
177
external_diff_options=None):
178
"""Show in text form the changes from one tree to another.
179
180
to_files
181
If set, include only changes to these files.
182
183
external_diff_options
184
If set, use an external GNU diff and pass these options.
185
"""
186
143
187
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
144
188
old_label = ''
145
189
new_label = ''
152
196
# TODO: Generation of pseudo-diffs for added/deleted files could
153
197
# be usefully made into a much faster special case.
154
198
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.
199
if external_diff_options:
200
assert isinstance(external_diff_options, basestring)
201
opts = external_diff_options.split()
202
def diff_file(olab, olines, nlab, nlines, to_file):
203
external_diff(olab, olines, nlab, nlines, to_file, opts)
204
else:
205
diff_file = internal_diff
162
206
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:
207
208
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
209
specific_files=specific_files)
210
211
for path, file_id, kind in delta.removed:
212
print >>to_file, '*** removed %s %r' % (kind, path)
213
if kind == 'file':
214
diff_file(old_label + path,
215
old_tree.get_file(file_id).readlines(),
216
DEVNULL,
217
[],
218
to_file)
219
220
for path, file_id, kind in delta.added:
221
print >>to_file, '*** added %s %r' % (kind, path)
222
if kind == 'file':
223
diff_file(DEVNULL,
224
[],
225
new_label + path,
226
new_tree.get_file(file_id).readlines(),
227
to_file)
228
229
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
230
print >>to_file, '*** renamed %s %r => %r' % (kind, old_path, new_path)
231
if text_modified:
232
diff_file(old_label + old_path,
233
old_tree.get_file(file_id).readlines(),
234
new_label + new_path,
235
new_tree.get_file(file_id).readlines(),
236
to_file)
237
238
for path, file_id, kind in delta.modified:
239
print >>to_file, '*** modified %s %r' % (kind, path)
240
if kind == 'file':
241
diff_file(old_label + path,
242
old_tree.get_file(file_id).readlines(),
243
new_label + path,
244
new_tree.get_file(file_id).readlines(),
245
to_file)
246
247
248
249
class TreeDelta(object):
250
250
"""Describes changes from one tree to another.
251
251
252
252
Contains four lists:
253
253
254
254
added
255
(path, id)
255
(path, id, kind)
256
256
removed
257
(path, id)
257
(path, id, kind)
258
258
renamed
259
(oldpath, newpath, id, text_modified)
259
(oldpath, newpath, id, kind, text_modified)
260
260
modified
261
(path, id)
261
(path, id, kind)
262
262
unchanged
263
(path, id)
263
(path, id, kind)
264
264
265
265
Each id is listed only once.
266
266
276
276
self.modified = []
277
277
self.unchanged = []
278
278
279
def __repr__(self):
280
return "TreeDelta(added=%r, removed=%r, renamed=%r, modified=%r," \
281
" unchanged=%r)" % (self.added, self.removed, self.renamed,
282
self.modified, self.unchanged)
283
284
def has_changed(self):
285
changes = len(self.added) + len(self.removed) + len(self.renamed)
286
changes += len(self.modified)
287
return (changes != 0)
288
289
def touches_file_id(self, file_id):
290
"""Return True if file_id is modified by this delta."""
291
for l in self.added, self.removed, self.modified:
292
for v in l:
293
if v[1] == file_id:
294
return True
295
for v in self.renamed:
296
if v[2] == file_id:
297
return True
298
return False
299
300
279
301
def show(self, to_file, show_ids=False, show_unchanged=False):
280
302
def show_list(files):
281
for path, fid in files:
303
for path, fid, kind in files:
304
if kind == 'directory':
305
path += '/'
306
elif kind == 'symlink':
307
path += '@'
308
282
309
if show_ids:
283
310
print >>to_file, ' %-30s %s' % (path, fid)
284
311
else:
285
312
print >>to_file, ' ', path
286
313
287
314
if self.removed:
288
print >>to_file, 'removed files:'
315
print >>to_file, 'removed:'
289
316
show_list(self.removed)
290
317
291
318
if self.added:
292
print >>to_file, 'added files:'
319
print >>to_file, 'added:'
293
320
show_list(self.added)
294
321
295
322
if self.renamed:
296
print >>to_file, 'renamed files:'
297
for oldpath, newpath, fid, text_modified in self.renamed:
323
print >>to_file, 'renamed:'
324
for oldpath, newpath, fid, kind, text_modified in self.renamed:
298
325
if show_ids:
299
326
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
300
327
else:
301
328
print >>to_file, ' %s => %s' % (oldpath, newpath)
302
329
303
330
if self.modified:
304
print >>to_file, 'modified files:'
331
print >>to_file, 'modified:'
305
332
show_list(self.modified)
306
333
307
334
if show_unchanged and self.unchanged:
308
print >>to_file, 'unchanged files:'
335
print >>to_file, 'unchanged:'
309
336
show_list(self.unchanged)
310
337
311
338
312
339
313
def compare_trees(old_tree, new_tree, want_unchanged):
340
def compare_trees(old_tree, new_tree, want_unchanged, specific_files=None):
341
"""Describe changes from one tree to another.
342
343
Returns a TreeDelta with details of added, modified, renamed, and
344
deleted entries.
345
346
The root entry is specifically exempt.
347
348
This only considers versioned files.
349
350
want_unchanged
351
If true, also list files unchanged from one version to
352
the next.
353
354
specific_files
355
If true, only check for changes to specified names or
356
files within them.
357
"""
358
359
from osutils import is_inside_any
360
314
361
old_inv = old_tree.inventory
315
362
new_inv = new_tree.inventory
316
363
delta = TreeDelta()
364
mutter('start compare_trees')
365
366
# TODO: match for specific files can be rather smarter by finding
367
# the IDs of those files up front and then considering only that.
368
317
369
for file_id in old_tree:
318
370
if file_id in new_tree:
319
old_path = old_inv.id2path(file_id)
320
new_path = new_inv.id2path(file_id)
321
322
371
kind = old_inv.get_file_kind(file_id)
372
assert kind == new_inv.get_file_kind(file_id)
373
323
374
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
324
375
'invalid file kind %r' % kind
376
377
if kind == 'root_directory':
378
continue
379
380
old_path = old_inv.id2path(file_id)
381
new_path = new_inv.id2path(file_id)
382
383
if specific_files:
384
if (not is_inside_any(specific_files, old_path)
385
and not is_inside_any(specific_files, new_path)):
386
continue
387
325
388
if kind == 'file':
326
389
old_sha1 = old_tree.get_file_sha1(file_id)
327
390
new_sha1 = new_tree.get_file_sha1(file_id)
336
399
# May not be worthwhile.
337
400
338
401
if old_path != new_path:
339
delta.renamed.append((old_path, new_path, file_id, text_modified))
402
delta.renamed.append((old_path, new_path, file_id, kind,
403
text_modified))
340
404
elif text_modified:
341
delta.modified.append((new_path, file_id))
405
delta.modified.append((new_path, file_id, kind))
342
406
elif want_unchanged:
343
delta.unchanged.append((new_path, file_id))
407
delta.unchanged.append((new_path, file_id, kind))
344
408
else:
345
delta.removed.append((old_inv.id2path(file_id), file_id))
409
kind = old_inv.get_file_kind(file_id)
410
old_path = old_inv.id2path(file_id)
411
if specific_files:
412
if not is_inside_any(specific_files, old_path):
413
continue
414
delta.removed.append((old_path, file_id, kind))
415
416
mutter('start looking for new files')
346
417
for file_id in new_inv:
347
418
if file_id in old_inv:
348
419
continue
349
delta.added.append((new_inv.id2path(file_id), file_id))
420
new_path = new_inv.id2path(file_id)
421
if specific_files:
422
if not is_inside_any(specific_files, new_path):
423
continue
424
kind = new_inv.get_file_kind(file_id)
425
delta.added.append((new_path, file_id, kind))
350
426
351
427
delta.removed.sort()
352
428
delta.added.sort()
353
429
delta.renamed.sort()
354
430
delta.modified.sort()
431
delta.unchanged.sort()
355
432
356
433
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1