« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 6
- compare with another revision
- download diff
- download tarball
- view history from revision 6
- Committer: mbp at sourcefrog
- Date: 2005-03-09 04:51:05 UTC
- Revision ID: mbp@sourcefrog.net-20050309045105-d02cd410a115da2c
import all docs from arch
- files added:
- bzrlib/tests.py
- files removed:
- .bzrignore
- contrib
- contrib/bash
- contrib/zsh
- elementtree
- notes
- patches
- urlgrabber
- files renamed:
- bzrlib/commands.py => bzr.py
- files modified:
- README
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
from errors import BzrError
20
21
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
"""Display a diff by calling out to the external diff program."""
72
import sys
73
74
if to_file != sys.stdout:
75
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
76
to_file)
77
78
from tempfile import NamedTemporaryFile
79
from os import system
80
81
oldtmpf = NamedTemporaryFile()
82
newtmpf = NamedTemporaryFile()
83
84
try:
85
# TODO: perhaps a special case for comparing to or from the empty
86
# sequence; can just use /dev/null on Unix
87
88
# TODO: if either of the files being compared already exists as a
89
# regular named file (e.g. in the working directory) then we can
90
# compare directly to that, rather than copying it.
91
92
# TODO: Set the labels appropriately
93
94
oldtmpf.writelines(oldlines)
95
newtmpf.writelines(newlines)
96
97
oldtmpf.flush()
98
newtmpf.flush()
99
100
system('diff -u --label %s %s --label %s %s' % (old_label, oldtmpf.name, new_label, newtmpf.name))
101
finally:
102
oldtmpf.close() # and delete
103
newtmpf.close()
104
105
106
107
def diff_file(old_label, oldlines, new_label, newlines, to_file):
108
if False:
109
differ = external_diff
110
else:
111
differ = internal_diff
112
113
differ(old_label, oldlines, new_label, newlines, to_file)
114
115
116
117
def show_diff(b, revision, specific_files):
118
import sys
119
120
if revision == None:
121
old_tree = b.basis_tree()
122
else:
123
old_tree = b.revision_tree(b.lookup_revision(revision))
124
125
new_tree = b.working_tree()
126
127
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files)
128
129
130
131
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None):
132
"""Show in text form the changes from one tree to another.
133
134
to_files
135
If set, include only changes to these files.
136
"""
137
138
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
139
old_label = ''
140
new_label = ''
141
142
DEVNULL = '/dev/null'
143
# Windows users, don't panic about this filename -- it is a
144
# special signal to GNU patch that the file should be created or
145
# deleted respectively.
146
147
# TODO: Generation of pseudo-diffs for added/deleted files could
148
# be usefully made into a much faster special case.
149
150
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
151
specific_files=specific_files)
152
153
for path, file_id, kind in delta.removed:
154
print '*** removed %s %r' % (kind, path)
155
if kind == 'file':
156
diff_file(old_label + path,
157
old_tree.get_file(file_id).readlines(),
158
DEVNULL,
159
[],
160
to_file)
161
162
for path, file_id, kind in delta.added:
163
print '*** added %s %r' % (kind, path)
164
if kind == 'file':
165
diff_file(DEVNULL,
166
[],
167
new_label + path,
168
new_tree.get_file(file_id).readlines(),
169
to_file)
170
171
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
172
print '*** renamed %s %r => %r' % (kind, old_path, new_path)
173
if text_modified:
174
diff_file(old_label + old_path,
175
old_tree.get_file(file_id).readlines(),
176
new_label + new_path,
177
new_tree.get_file(file_id).readlines(),
178
to_file)
179
180
for path, file_id, kind in delta.modified:
181
print '*** modified %s %r' % (kind, path)
182
if kind == 'file':
183
diff_file(old_label + path,
184
old_tree.get_file(file_id).readlines(),
185
new_label + path,
186
new_tree.get_file(file_id).readlines(),
187
to_file)
188
189
190
191
class TreeDelta(object):
192
"""Describes changes from one tree to another.
193
194
Contains four lists:
195
196
added
197
(path, id, kind)
198
removed
199
(path, id, kind)
200
renamed
201
(oldpath, newpath, id, kind, text_modified)
202
modified
203
(path, id, kind)
204
unchanged
205
(path, id, kind)
206
207
Each id is listed only once.
208
209
Files that are both modified and renamed are listed only in
210
renamed, with the text_modified flag true.
211
212
The lists are normally sorted when the delta is created.
213
"""
214
def __init__(self):
215
self.added = []
216
self.removed = []
217
self.renamed = []
218
self.modified = []
219
self.unchanged = []
220
221
222
def touches_file_id(self, file_id):
223
"""Return True if file_id is modified by this delta."""
224
for l in self.added, self.removed, self.modified:
225
for v in l:
226
if v[1] == file_id:
227
return True
228
for v in self.renamed:
229
if v[2] == file_id:
230
return True
231
return False
232
233
234
def show(self, to_file, show_ids=False, show_unchanged=False):
235
def show_list(files):
236
for path, fid, kind in files:
237
if kind == 'directory':
238
path += '/'
239
elif kind == 'symlink':
240
path += '@'
241
242
if show_ids:
243
print >>to_file, ' %-30s %s' % (path, fid)
244
else:
245
print >>to_file, ' ', path
246
247
if self.removed:
248
print >>to_file, 'removed:'
249
show_list(self.removed)
250
251
if self.added:
252
print >>to_file, 'added:'
253
show_list(self.added)
254
255
if self.renamed:
256
print >>to_file, 'renamed:'
257
for oldpath, newpath, fid, kind, text_modified in self.renamed:
258
if show_ids:
259
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
260
else:
261
print >>to_file, ' %s => %s' % (oldpath, newpath)
262
263
if self.modified:
264
print >>to_file, 'modified:'
265
show_list(self.modified)
266
267
if show_unchanged and self.unchanged:
268
print >>to_file, 'unchanged:'
269
show_list(self.unchanged)
270
271
272
273
def compare_trees(old_tree, new_tree, want_unchanged, specific_files=None):
274
"""Describe changes from one tree to another.
275
276
Returns a TreeDelta with details of added, modified, renamed, and
277
deleted entries.
278
279
The root entry is specifically exempt.
280
281
This only considers versioned files.
282
283
want_unchanged
284
If true, also list files unchanged from one version to
285
the next.
286
287
specific_files
288
If true, only check for changes to specified names or
289
files within them.
290
"""
291
292
from osutils import is_inside_any
293
294
old_inv = old_tree.inventory
295
new_inv = new_tree.inventory
296
delta = TreeDelta()
297
mutter('start compare_trees')
298
299
# TODO: match for specific files can be rather smarter by finding
300
# the IDs of those files up front and then considering only that.
301
302
for file_id in old_tree:
303
if file_id in new_tree:
304
kind = old_inv.get_file_kind(file_id)
305
assert kind == new_inv.get_file_kind(file_id)
306
307
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
308
'invalid file kind %r' % kind
309
310
if kind == 'root_directory':
311
continue
312
313
old_path = old_inv.id2path(file_id)
314
new_path = new_inv.id2path(file_id)
315
316
if specific_files:
317
if (not is_inside_any(specific_files, old_path)
318
and not is_inside_any(specific_files, new_path)):
319
continue
320
321
if kind == 'file':
322
old_sha1 = old_tree.get_file_sha1(file_id)
323
new_sha1 = new_tree.get_file_sha1(file_id)
324
text_modified = (old_sha1 != new_sha1)
325
else:
326
## mutter("no text to check for %r %r" % (file_id, kind))
327
text_modified = False
328
329
# TODO: Can possibly avoid calculating path strings if the
330
# two files are unchanged and their names and parents are
331
# the same and the parents are unchanged all the way up.
332
# May not be worthwhile.
333
334
if old_path != new_path:
335
delta.renamed.append((old_path, new_path, file_id, kind,
336
text_modified))
337
elif text_modified:
338
delta.modified.append((new_path, file_id, kind))
339
elif want_unchanged:
340
delta.unchanged.append((new_path, file_id, kind))
341
else:
342
kind = old_inv.get_file_kind(file_id)
343
old_path = old_inv.id2path(file_id)
344
if specific_files:
345
if not is_inside_any(specific_files, old_path):
346
continue
347
delta.removed.append((old_path, file_id, kind))
348
349
mutter('start looking for new files')
350
for file_id in new_inv:
351
if file_id in old_inv:
352
continue
353
new_path = new_inv.id2path(file_id)
354
if specific_files:
355
if not is_inside_any(specific_files, new_path):
356
continue
357
kind = new_inv.get_file_kind(file_id)
358
delta.added.append((new_path, file_id, kind))
359
360
delta.removed.sort()
361
delta.added.sort()
362
delta.renamed.sort()
363
delta.modified.sort()
364
delta.unchanged.sort()
365
366
return delta
21
22
23
24
25
26
27
def diff_trees(old_tree, new_tree):
28
"""Compute diff between two trees.
29
30
They may be in different branches and may be working or historical
31
trees.
32
33
Yields a sequence of (state, id, old_name, new_name, kind).
34
Each filename and each id is listed only once.
35
"""
36
37
## TODO: Compare files before diffing; only mention those that have changed
38
39
## TODO: Set nice names in the headers, maybe include diffstat
40
41
## TODO: Perhaps make this a generator rather than using
42
## a callback object?
43
44
## TODO: Allow specifying a list of files to compare, rather than
45
## doing the whole tree? (Not urgent.)
46
47
## TODO: Allow diffing any two inventories, not just the
48
## current one against one. We mgiht need to specify two
49
## stores to look for the files if diffing two branches. That
50
## might imply this shouldn't be primarily a Branch method.
51
52
## XXX: This doesn't report on unknown files; that can be done
53
## from a separate method.
54
55
old_it = old_tree.list_files()
56
new_it = new_tree.list_files()
57
58
def next(it):
59
try:
60
return it.next()
61
except StopIteration:
62
return None
63
64
old_item = next(old_it)
65
new_item = next(new_it)
66
67
# We step through the two sorted iterators in parallel, trying to
68
# keep them lined up.
69
70
while (old_item != None) or (new_item != None):
71
# OK, we still have some remaining on both, but they may be
72
# out of step.
73
if old_item != None:
74
old_name, old_class, old_kind, old_id = old_item
75
else:
76
old_name = None
77
78
if new_item != None:
79
new_name, new_class, new_kind, new_id = new_item
80
else:
81
new_name = None
82
83
mutter(" diff pairwise %r" % (old_item,))
84
mutter(" %r" % (new_item,))
85
86
if old_item:
87
# can't handle the old tree being a WorkingTree
88
assert old_class == 'V'
89
90
if new_item and (new_class != 'V'):
91
yield new_class, None, None, new_name, new_kind
92
new_item = next(new_it)
93
elif (not new_item) or (old_item and (old_name < new_name)):
94
mutter(" extra entry in old-tree sequence")
95
if new_tree.has_id(old_id):
96
# will be mentioned as renamed under new name
97
pass
98
else:
99
yield 'D', old_id, old_name, None, old_kind
100
old_item = next(old_it)
101
elif (not old_item) or (new_item and (new_name < old_name)):
102
mutter(" extra entry in new-tree sequence")
103
if old_tree.has_id(new_id):
104
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
105
else:
106
yield 'A', new_id, None, new_name, new_kind
107
new_item = next(new_it)
108
elif old_id != new_id:
109
assert old_name == new_name
110
# both trees have a file of this name, but it is not the
111
# same file. in other words, the old filename has been
112
# overwritten by either a newly-added or a renamed file.
113
# (should we return something about the overwritten file?)
114
if old_tree.has_id(new_id):
115
# renaming, overlying a deleted file
116
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
117
else:
118
yield 'A', new_id, None, new_name, new_kind
119
120
new_item = next(new_it)
121
old_item = next(old_it)
122
else:
123
assert old_id == new_id
124
assert old_name == new_name
125
assert old_kind == new_kind
126
127
if old_kind == 'directory':
128
yield '.', new_id, old_name, new_name, new_kind
129
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
130
mutter(" file size has changed, must be different")
131
yield 'M', new_id, old_name, new_name, new_kind
132
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
133
mutter(" SHA1 indicates they're identical")
134
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
135
yield '.', new_id, old_name, new_name, new_kind
136
else:
137
mutter(" quick compare shows different")
138
yield 'M', new_id, old_name, new_name, new_kind
139
140
new_item = next(new_it)
141
old_item = next(old_it)
142
143
Loggerhead is a web-based interface for Breezy
Version: 2.0.1