To get this branch, use:
bzr branch
http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar
« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 450
- compare with another revision
- download diff
- download tarball
- view history from revision 450
- Committer: Martin Pool
- Date: 2005-05-11 01:06:07 UTC
- Revision ID: mbp@sourcefrog.net-20050511010607-2475d3b03b3b4b0e
- Use urlgrabber by default
- files added:
- bzrlib/statcache.py
- files removed:
- bzr-man.py
- bzrlib/selftest
- contrib/emacs
- 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 bzrlib.trace import mutter
19
from bzrlib.errors import BzrError
20
from bzrlib.delta import compare_trees
21
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
ud = difflib.unified_diff(oldlines, newlines,
46
fromfile=old_label, tofile=new_label)
47
48
# work-around for difflib being too smart for its own good
49
# if /dev/null is "1,0", patch won't recognize it as /dev/null
50
if not oldlines:
51
ud = list(ud)
52
ud[2] = ud[2].replace('-1,0', '-0,0')
53
elif not newlines:
54
ud = list(ud)
55
ud[2] = ud[2].replace('+1,0', '+0,0')
56
57
for line in ud:
58
to_file.write(line)
59
if not line.endswith('\n'):
60
to_file.write("\n\\ No newline at end of file\n")
61
print >>to_file
62
63
64
65
66
def external_diff(old_label, oldlines, new_label, newlines, to_file,
67
diff_opts):
68
"""Display a diff by calling out to the external diff program."""
69
import sys
70
71
if to_file != sys.stdout:
72
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
73
to_file)
74
75
# make sure our own output is properly ordered before the diff
76
to_file.flush()
77
78
from tempfile import NamedTemporaryFile
79
import os
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
oldtmpf.writelines(oldlines)
93
newtmpf.writelines(newlines)
94
95
oldtmpf.flush()
96
newtmpf.flush()
97
98
if not diff_opts:
99
diff_opts = []
100
diffcmd = ['diff',
101
'--label', old_label,
102
oldtmpf.name,
103
'--label', new_label,
104
newtmpf.name]
105
106
# diff only allows one style to be specified; they don't override.
107
# note that some of these take optargs, and the optargs can be
108
# directly appended to the options.
109
# this is only an approximate parser; it doesn't properly understand
110
# the grammar.
111
for s in ['-c', '-u', '-C', '-U',
112
'-e', '--ed',
113
'-q', '--brief',
114
'--normal',
115
'-n', '--rcs',
116
'-y', '--side-by-side',
117
'-D', '--ifdef']:
118
for j in diff_opts:
119
if j.startswith(s):
120
break
121
else:
122
continue
123
break
124
else:
125
diffcmd.append('-u')
126
127
if diff_opts:
128
diffcmd.extend(diff_opts)
129
130
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
131
132
if rc != 0 and rc != 1:
133
# returns 1 if files differ; that's OK
134
if rc < 0:
135
msg = 'signal %d' % (-rc)
136
else:
137
msg = 'exit code %d' % rc
138
139
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
140
finally:
141
oldtmpf.close() # and delete
142
newtmpf.close()
143
144
145
146
def show_diff(b, revision, specific_files, external_diff_options=None):
147
"""Shortcut for showing the diff to the working tree.
148
149
b
150
Branch.
151
152
revision
153
None for each, or otherwise the old revision to compare against.
154
155
The more general form is show_diff_trees(), where the caller
156
supplies any two trees.
18
from sets import Set
19
20
from trace import mutter
21
from errors import BzrError
22
23
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.
157
32
"""
158
import sys
159
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
160
146
if revision == None:
161
147
old_tree = b.basis_tree()
162
148
else:
164
150
165
151
new_tree = b.working_tree()
166
152
167
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
168
external_diff_options)
169
170
171
172
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
173
external_diff_options=None):
174
"""Show in text form the changes from one tree to another.
175
176
to_files
177
If set, include only changes to these files.
178
179
external_diff_options
180
If set, use an external GNU diff and pass these options.
181
"""
182
183
153
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
184
154
old_label = ''
185
155
new_label = ''
192
162
# TODO: Generation of pseudo-diffs for added/deleted files could
193
163
# be usefully made into a much faster special case.
194
164
195
if external_diff_options:
196
assert isinstance(external_diff_options, basestring)
197
opts = external_diff_options.split()
198
def diff_file(olab, olines, nlab, nlines, to_file):
199
external_diff(olab, olines, nlab, nlines, to_file, opts)
200
else:
201
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.
202
172
203
204
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
205
specific_files=specific_files)
206
207
for path, file_id, kind in delta.removed:
208
print >>to_file, '*** removed %s %r' % (kind, path)
209
if kind == 'file':
210
diff_file(old_label + path,
211
old_tree.get_file(file_id).readlines(),
212
DEVNULL,
213
[],
214
to_file)
215
216
for path, file_id, kind in delta.added:
217
print >>to_file, '*** added %s %r' % (kind, path)
218
if kind == 'file':
219
diff_file(DEVNULL,
220
[],
221
new_label + path,
222
new_tree.get_file(file_id).readlines(),
223
to_file)
224
225
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
226
print >>to_file, '*** renamed %s %r => %r' % (kind, old_path, new_path)
227
if text_modified:
228
diff_file(old_label + old_path,
229
old_tree.get_file(file_id).readlines(),
230
new_label + new_path,
231
new_tree.get_file(file_id).readlines(),
232
to_file)
233
234
for path, file_id, kind in delta.modified:
235
print >>to_file, '*** modified %s %r' % (kind, path)
236
if kind == 'file':
237
diff_file(old_label + path,
238
old_tree.get_file(file_id).readlines(),
239
new_label + path,
240
new_tree.get_file(file_id).readlines(),
241
to_file)
242
243
244
245
246
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
def diffit(oldlines, newlines, **kw):
183
184
# FIXME: difflib is wrong if there is no trailing newline.
185
# The syntax used by patch seems to be "\ No newline at
186
# end of file" following the last diff line from that
187
# file. This is not trivial to insert into the
188
# unified_diff output and it might be better to just fix
189
# or replace that function.
190
191
# In the meantime we at least make sure the patch isn't
192
# mangled.
193
194
195
# Special workaround for Python2.3, where difflib fails if
196
# both sequences are empty.
197
if not oldlines and not newlines:
198
return
199
200
nonl = False
201
202
if oldlines and (oldlines[-1][-1] != '\n'):
203
oldlines[-1] += '\n'
204
nonl = True
205
if newlines and (newlines[-1][-1] != '\n'):
206
newlines[-1] += '\n'
207
nonl = True
208
209
ud = difflib.unified_diff(oldlines, newlines, **kw)
210
211
# work-around for difflib being too smart for its own good
212
# if /dev/null is "1,0", patch won't recognize it as /dev/null
213
if not oldlines:
214
ud = list(ud)
215
ud[2] = ud[2].replace('-1,0', '-0,0')
216
elif not newlines:
217
ud = list(ud)
218
ud[2] = ud[2].replace('+1,0', '+0,0')
219
220
sys.stdout.writelines(ud)
221
if nonl:
222
print "\\ No newline at end of file"
223
sys.stdout.write('\n')
224
225
if file_state in ['.', '?', 'I']:
226
continue
227
elif file_state == 'A':
228
print '*** added %s %r' % (kind, new_name)
229
if kind == 'file':
230
diffit([],
231
new_tree.get_file(fid).readlines(),
232
fromfile=DEVNULL,
233
tofile=new_label + new_name + idlabel)
234
elif file_state == 'D':
235
assert isinstance(old_name, types.StringTypes)
236
print '*** deleted %s %r' % (kind, old_name)
237
if kind == 'file':
238
diffit(old_tree.get_file(fid).readlines(), [],
239
fromfile=old_label + old_name + idlabel,
240
tofile=DEVNULL)
241
elif file_state in ['M', 'R']:
242
if file_state == 'M':
243
assert kind == 'file'
244
assert old_name == new_name
245
print '*** modified %s %r' % (kind, new_name)
246
elif file_state == 'R':
247
print '*** renamed %s %r => %r' % (kind, old_name, new_name)
248
249
if kind == 'file':
250
diffit(old_tree.get_file(fid).readlines(),
251
new_tree.get_file(fid).readlines(),
252
fromfile=old_label + old_name + idlabel,
253
tofile=new_label + new_name)
254
else:
255
raise BzrError("can't represent state %s {%s}" % (file_state, fid))
256
257
258
259
class TreeDelta:
260
"""Describes changes from one tree to another.
261
262
Contains four lists:
263
264
added
265
(path, id)
266
removed
267
(path, id)
268
renamed
269
(oldpath, newpath, id)
270
modified
271
(path, id)
272
273
A path may occur in more than one list if it was e.g. deleted
274
under an old id and renamed into place in a new id.
275
276
Files are listed in either modified or renamed, not both. In
277
other words, renamed files may also be modified.
278
"""
279
def __init__(self):
280
self.added = []
281
self.removed = []
282
self.renamed = []
283
self.modified = []
284
285
286
def compare_inventories(old_inv, new_inv):
287
"""Return a TreeDelta object describing changes between inventories.
288
289
This only describes changes in the shape of the tree, not the
290
actual texts.
291
292
This is an alternative to diff_trees() and should probably
293
eventually replace it.
294
"""
295
old_ids = old_inv.id_set()
296
new_ids = new_inv.id_set()
297
delta = TreeDelta()
298
299
delta.removed = [(old_inv.id2path(fid), fid) for fid in (old_ids - new_ids)]
300
delta.removed.sort()
301
302
delta.added = [(new_inv.id2path(fid), fid) for fid in (new_ids - old_ids)]
303
delta.added.sort()
304
305
for fid in old_ids & new_ids:
306
old_ie = old_inv[fid]
307
new_ie = new_inv[fid]
308
old_path = old_inv.id2path(fid)
309
new_path = new_inv.id2path(fid)
310
311
if old_path != new_path:
312
delta.renamed.append((old_path, new_path, fid))
313
elif old_ie.text_sha1 != new_ie.text_sha1:
314
delta.modified.append((new_path, fid))
315
316
delta.modified.sort()
317
delta.renamed.sort()
318
319
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.2.dev0