« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 561
- compare with another revision
- download diff
- download tarball
- view history from revision 561
- Committer: Martin Pool
- Date: 2005-05-26 03:10:53 UTC
- Revision ID: mbp@sourcefrog.net-20050526031053-e6a43ad83e2dc1ee
- add failing test for problem with stat after remove,
reported by William Dodé
reported by William Dodé
- files added:
- .rsyncexclude
- contrib
- contrib/bash
- contrib/zsh
- elementtree
- patches
- urlgrabber
- files removed:
- bzrlib/tests.py
- files renamed:
- bzr.py => 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
20
18
from trace import mutter
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
19
from errors import BzrError
20
21
22
23
def _diff_one(oldlines, newlines, to_file, **kw):
24
import difflib
25
26
# FIXME: difflib is wrong if there is no trailing newline.
27
# The syntax used by patch seems to be "\ No newline at
28
# end of file" following the last diff line from that
29
# file. This is not trivial to insert into the
30
# unified_diff output and it might be better to just fix
31
# or replace that function.
32
33
# In the meantime we at least make sure the patch isn't
34
# mangled.
35
36
37
# Special workaround for Python2.3, where difflib fails if
38
# both sequences are empty.
39
if not oldlines and not newlines:
40
return
41
42
nonl = False
43
44
if oldlines and (oldlines[-1][-1] != '\n'):
45
oldlines[-1] += '\n'
46
nonl = True
47
if newlines and (newlines[-1][-1] != '\n'):
48
newlines[-1] += '\n'
49
nonl = True
50
51
ud = difflib.unified_diff(oldlines, newlines, **kw)
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
def show_diff(b, revision, specific_files):
70
import sys
71
72
if revision == None:
73
old_tree = b.basis_tree()
74
else:
75
old_tree = b.revision_tree(b.lookup_revision(revision))
76
77
new_tree = b.working_tree()
78
79
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files)
80
81
82
83
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None):
84
"""Show in text form the changes from one tree to another.
85
86
to_files
87
If set, include only changes to these files.
88
"""
89
90
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
91
old_label = ''
92
new_label = ''
93
94
DEVNULL = '/dev/null'
95
# Windows users, don't panic about this filename -- it is a
96
# special signal to GNU patch that the file should be created or
97
# deleted respectively.
98
99
# TODO: Generation of pseudo-diffs for added/deleted files could
100
# be usefully made into a much faster special case.
101
102
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
103
specific_files=specific_files)
104
105
for path, file_id, kind in delta.removed:
106
print '*** removed %s %r' % (kind, path)
107
if kind == 'file':
108
_diff_one(old_tree.get_file(file_id).readlines(),
109
[],
110
to_file,
111
fromfile=old_label + path,
112
tofile=DEVNULL)
113
114
for path, file_id, kind in delta.added:
115
print '*** added %s %r' % (kind, path)
116
if kind == 'file':
117
_diff_one([],
118
new_tree.get_file(file_id).readlines(),
119
to_file,
120
fromfile=DEVNULL,
121
tofile=new_label + path)
122
123
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
124
print '*** renamed %s %r => %r' % (kind, old_path, new_path)
125
if text_modified:
126
_diff_one(old_tree.get_file(file_id).readlines(),
127
new_tree.get_file(file_id).readlines(),
128
to_file,
129
fromfile=old_label + old_path,
130
tofile=new_label + new_path)
131
132
for path, file_id, kind in delta.modified:
133
print '*** modified %s %r' % (kind, path)
134
if kind == 'file':
135
_diff_one(old_tree.get_file(file_id).readlines(),
136
new_tree.get_file(file_id).readlines(),
137
to_file,
138
fromfile=old_label + path,
139
tofile=new_label + path)
140
141
142
143
class TreeDelta(object):
144
"""Describes changes from one tree to another.
145
146
Contains four lists:
147
148
added
149
(path, id, kind)
150
removed
151
(path, id, kind)
152
renamed
153
(oldpath, newpath, id, kind, text_modified)
154
modified
155
(path, id, kind)
156
unchanged
157
(path, id, kind)
158
159
Each id is listed only once.
160
161
Files that are both modified and renamed are listed only in
162
renamed, with the text_modified flag true.
163
164
The lists are normally sorted when the delta is created.
165
"""
166
def __init__(self):
167
self.added = []
168
self.removed = []
169
self.renamed = []
170
self.modified = []
171
self.unchanged = []
172
173
174
def touches_file_id(self, file_id):
175
"""Return True if file_id is modified by this delta."""
176
for l in self.added, self.removed, self.modified:
177
for v in l:
178
if v[1] == file_id:
179
return True
180
for v in self.renamed:
181
if v[2] == file_id:
182
return True
183
return False
184
185
186
def show(self, to_file, show_ids=False, show_unchanged=False):
187
def show_list(files):
188
for path, fid, kind in files:
189
if kind == 'directory':
190
path += '/'
191
elif kind == 'symlink':
192
path += '@'
193
194
if show_ids:
195
print >>to_file, ' %-30s %s' % (path, fid)
196
else:
197
print >>to_file, ' ', path
198
199
if self.removed:
200
print >>to_file, 'removed:'
201
show_list(self.removed)
202
203
if self.added:
204
print >>to_file, 'added:'
205
show_list(self.added)
206
207
if self.renamed:
208
print >>to_file, 'renamed:'
209
for oldpath, newpath, fid, kind, text_modified in self.renamed:
210
if show_ids:
211
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
212
else:
213
print >>to_file, ' %s => %s' % (oldpath, newpath)
214
215
if self.modified:
216
print >>to_file, 'modified:'
217
show_list(self.modified)
218
219
if show_unchanged and self.unchanged:
220
print >>to_file, 'unchanged:'
221
show_list(self.unchanged)
222
223
224
225
def compare_trees(old_tree, new_tree, want_unchanged, specific_files=None):
226
"""Describe changes from one tree to another.
227
228
Returns a TreeDelta with details of added, modified, renamed, and
229
deleted entries.
230
231
The root entry is specifically exempt.
232
233
This only considers versioned files.
234
235
want_unchanged
236
If true, also list files unchanged from one version to
237
the next.
238
239
specific_files
240
If true, only check for changes to specified names or
241
files within them.
242
"""
243
244
from osutils import is_inside_any
245
246
old_inv = old_tree.inventory
247
new_inv = new_tree.inventory
248
delta = TreeDelta()
249
mutter('start compare_trees')
250
251
# TODO: match for specific files can be rather smarter by finding
252
# the IDs of those files up front and then considering only that.
253
254
for file_id in old_tree:
255
if file_id in new_tree:
256
kind = old_inv.get_file_kind(file_id)
257
assert kind == new_inv.get_file_kind(file_id)
258
259
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
260
'invalid file kind %r' % kind
261
262
if kind == 'root_directory':
263
continue
264
265
old_path = old_inv.id2path(file_id)
266
new_path = new_inv.id2path(file_id)
267
268
if specific_files:
269
if (not is_inside_any(specific_files, old_path)
270
and not is_inside_any(specific_files, new_path)):
271
continue
272
273
if kind == 'file':
274
old_sha1 = old_tree.get_file_sha1(file_id)
275
new_sha1 = new_tree.get_file_sha1(file_id)
276
text_modified = (old_sha1 != new_sha1)
277
else:
278
## mutter("no text to check for %r %r" % (file_id, kind))
279
text_modified = False
280
281
# TODO: Can possibly avoid calculating path strings if the
282
# two files are unchanged and their names and parents are
283
# the same and the parents are unchanged all the way up.
284
# May not be worthwhile.
285
286
if old_path != new_path:
287
delta.renamed.append((old_path, new_path, file_id, kind,
288
text_modified))
289
elif text_modified:
290
delta.modified.append((new_path, file_id, kind))
291
elif want_unchanged:
292
delta.unchanged.append((new_path, file_id, kind))
293
else:
294
old_path = old_inv.id2path(file_id)
295
if specific_files:
296
if not is_inside_any(specific_files, old_path):
297
continue
298
delta.removed.append((old_path, file_id, kind))
299
300
mutter('start looking for new files')
301
for file_id in new_inv:
302
if file_id in old_inv:
303
continue
304
new_path = new_inv.id2path(file_id)
305
if specific_files:
306
if not is_inside_any(specific_files, new_path):
307
continue
308
kind = new_inv.get_file_kind(file_id)
309
delta.added.append((new_path, file_id, kind))
310
311
delta.removed.sort()
312
delta.added.sort()
313
delta.renamed.sort()
314
delta.modified.sort()
315
delta.unchanged.sort()
316
317
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1