« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 334
- compare with another revision
- download diff
- download tarball
- view history from revision 334
- Committer: Martin Pool
- Date: 2005-05-03 08:00:27 UTC
- Revision ID: mbp@sourcefrog.net-20050503080027-908edb5b39982198
doc
- files added:
- doc/faq.txt
- files removed:
- bzrlib/atomicfile.py
- 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, ImmutableSet
18
from sets import Set
19
19
20
20
from trace import mutter
21
from errors import BzrError
22
23
24
25
def _diff_one(oldlines, newlines, to_file, **kw):
26
import difflib
21
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.
32
"""
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
27
145
28
# FIXME: difflib is wrong if there is no trailing newline.
29
# The syntax used by patch seems to be "\ No newline at
30
# end of file" following the last diff line from that
31
# file. This is not trivial to insert into the
32
# unified_diff output and it might be better to just fix
33
# or replace that function.
34
35
# In the meantime we at least make sure the patch isn't
36
# mangled.
37
38
39
# Special workaround for Python2.3, where difflib fails if
40
# both sequences are empty.
41
if not oldlines and not newlines:
42
return
43
44
nonl = False
45
46
if oldlines and (oldlines[-1][-1] != '\n'):
47
oldlines[-1] += '\n'
48
nonl = True
49
if newlines and (newlines[-1][-1] != '\n'):
50
newlines[-1] += '\n'
51
nonl = True
52
53
ud = difflib.unified_diff(oldlines, newlines, **kw)
54
55
# work-around for difflib being too smart for its own good
56
# if /dev/null is "1,0", patch won't recognize it as /dev/null
57
if not oldlines:
58
ud = list(ud)
59
ud[2] = ud[2].replace('-1,0', '-0,0')
60
elif not newlines:
61
ud = list(ud)
62
ud[2] = ud[2].replace('+1,0', '+0,0')
63
64
to_file.writelines(ud)
65
if nonl:
66
print >>to_file, "\\ No newline at end of file"
67
print >>to_file
68
69
70
def show_diff(b, revision, specific_files):
71
import sys
72
73
146
if revision == None:
74
147
old_tree = b.basis_tree()
75
148
else:
89
162
# TODO: Generation of pseudo-diffs for added/deleted files could
90
163
# be usefully made into a much faster special case.
91
164
92
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
93
specific_files=specific_files)
94
95
for path, file_id, kind in delta.removed:
96
print '*** removed %s %r' % (kind, path)
97
if kind == 'file':
98
_diff_one(old_tree.get_file(file_id).readlines(),
99
[],
100
sys.stdout,
101
fromfile=old_label + path,
102
tofile=DEVNULL)
103
104
for path, file_id, kind in delta.added:
105
print '*** added %s %r' % (kind, path)
106
if kind == 'file':
107
_diff_one([],
108
new_tree.get_file(file_id).readlines(),
109
sys.stdout,
110
fromfile=DEVNULL,
111
tofile=new_label + path)
112
113
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
114
print '*** renamed %s %r => %r' % (kind, old_path, new_path)
115
if text_modified:
116
_diff_one(old_tree.get_file(file_id).readlines(),
117
new_tree.get_file(file_id).readlines(),
118
sys.stdout,
119
fromfile=old_label + old_path,
120
tofile=new_label + new_path)
121
122
for path, file_id, kind in delta.modified:
123
print '*** modified %s %r' % (kind, path)
124
if kind == 'file':
125
_diff_one(old_tree.get_file(file_id).readlines(),
126
new_tree.get_file(file_id).readlines(),
127
sys.stdout,
128
fromfile=old_label + path,
129
tofile=new_label + path)
130
131
132
133
class TreeDelta:
134
"""Describes changes from one tree to another.
135
136
Contains four lists:
137
138
added
139
(path, id, kind)
140
removed
141
(path, id, kind)
142
renamed
143
(oldpath, newpath, id, kind, text_modified)
144
modified
145
(path, id, kind)
146
unchanged
147
(path, id, kind)
148
149
Each id is listed only once.
150
151
Files that are both modified and renamed are listed only in
152
renamed, with the text_modified flag true.
153
154
The lists are normally sorted when the delta is created.
155
"""
156
def __init__(self):
157
self.added = []
158
self.removed = []
159
self.renamed = []
160
self.modified = []
161
self.unchanged = []
162
163
def show(self, to_file, show_ids=False, show_unchanged=False):
164
def show_list(files):
165
for path, fid, kind in files:
166
if kind == 'directory':
167
path += '/'
168
elif kind == 'symlink':
169
path += '@'
170
171
if show_ids:
172
print >>to_file, ' %-30s %s' % (path, fid)
173
else:
174
print >>to_file, ' ', path
175
176
if self.removed:
177
print >>to_file, 'removed:'
178
show_list(self.removed)
179
180
if self.added:
181
print >>to_file, 'added:'
182
show_list(self.added)
183
184
if self.renamed:
185
print >>to_file, 'renamed:'
186
for oldpath, newpath, fid, kind, text_modified in self.renamed:
187
if show_ids:
188
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
189
else:
190
print >>to_file, ' %s => %s' % (oldpath, newpath)
191
192
if self.modified:
193
print >>to_file, 'modified:'
194
show_list(self.modified)
195
196
if show_unchanged and self.unchanged:
197
print >>to_file, 'unchanged:'
198
show_list(self.unchanged)
199
200
201
202
def compare_trees(old_tree, new_tree, want_unchanged, specific_files=None):
203
"""Describe changes from one tree to another.
204
205
Returns a TreeDelta with details of added, modified, renamed, and
206
deleted entries.
207
208
The root entry is specifically exempt.
209
210
This only considers versioned files.
211
212
want_unchanged
213
If true, also list files unchanged from one version to
214
the next.
215
216
specific_files
217
If true, only check for changes to specified names or
218
files within them.
219
"""
220
221
from osutils import is_inside_any
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.
222
172
223
old_inv = old_tree.inventory
224
new_inv = new_tree.inventory
225
delta = TreeDelta()
226
mutter('start compare_trees')
227
228
# TODO: match for specific files can be rather smarter by finding
229
# the IDs of those files up front and then considering only that.
230
231
for file_id in old_tree:
232
if file_id in new_tree:
233
kind = old_inv.get_file_kind(file_id)
234
assert kind == new_inv.get_file_kind(file_id)
235
236
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
237
'invalid file kind %r' % kind
238
239
if kind == 'root_directory':
240
continue
241
242
old_path = old_inv.id2path(file_id)
243
new_path = new_inv.id2path(file_id)
244
245
if specific_files:
246
if (not is_inside_any(specific_files, old_path)
247
and not is_inside_any(specific_files, new_path)):
248
continue
249
250
if kind == 'file':
251
old_sha1 = old_tree.get_file_sha1(file_id)
252
new_sha1 = new_tree.get_file_sha1(file_id)
253
text_modified = (old_sha1 != new_sha1)
254
else:
255
## mutter("no text to check for %r %r" % (file_id, kind))
256
text_modified = False
257
258
# TODO: Can possibly avoid calculating path strings if the
259
# two files are unchanged and their names and parents are
260
# the same and the parents are unchanged all the way up.
261
# May not be worthwhile.
262
263
if old_path != new_path:
264
delta.renamed.append((old_path, new_path, file_id, kind,
265
text_modified))
266
elif text_modified:
267
delta.modified.append((new_path, file_id, kind))
268
elif want_unchanged:
269
delta.unchanged.append((new_path, file_id, kind))
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
# FIXME: Something about the diff format makes patch unhappy
183
# with newly-added files.
184
185
def diffit(oldlines, newlines, **kw):
186
187
# FIXME: difflib is wrong if there is no trailing newline.
188
# The syntax used by patch seems to be "\ No newline at
189
# end of file" following the last diff line from that
190
# file. This is not trivial to insert into the
191
# unified_diff output and it might be better to just fix
192
# or replace that function.
193
194
# In the meantime we at least make sure the patch isn't
195
# mangled.
196
197
198
# Special workaround for Python2.3, where difflib fails if
199
# both sequences are empty.
200
if not oldlines and not newlines:
201
return
202
203
nonl = False
204
205
if oldlines and (oldlines[-1][-1] != '\n'):
206
oldlines[-1] += '\n'
207
nonl = True
208
if newlines and (newlines[-1][-1] != '\n'):
209
newlines[-1] += '\n'
210
nonl = True
211
212
ud = difflib.unified_diff(oldlines, newlines, **kw)
213
sys.stdout.writelines(ud)
214
if nonl:
215
print "\\ No newline at end of file"
216
sys.stdout.write('\n')
217
218
if file_state in ['.', '?', 'I']:
219
continue
220
elif file_state == 'A':
221
print '*** added %s %r' % (kind, new_name)
222
if kind == 'file':
223
diffit([],
224
new_tree.get_file(fid).readlines(),
225
fromfile=DEVNULL,
226
tofile=new_label + new_name + idlabel)
227
elif file_state == 'D':
228
assert isinstance(old_name, types.StringTypes)
229
print '*** deleted %s %r' % (kind, old_name)
230
if kind == 'file':
231
diffit(old_tree.get_file(fid).readlines(), [],
232
fromfile=old_label + old_name + idlabel,
233
tofile=DEVNULL)
234
elif file_state in ['M', 'R']:
235
if file_state == 'M':
236
assert kind == 'file'
237
assert old_name == new_name
238
print '*** modified %s %r' % (kind, new_name)
239
elif file_state == 'R':
240
print '*** renamed %s %r => %r' % (kind, old_name, new_name)
241
242
if kind == 'file':
243
diffit(old_tree.get_file(fid).readlines(),
244
new_tree.get_file(fid).readlines(),
245
fromfile=old_label + old_name + idlabel,
246
tofile=new_label + new_name)
270
247
else:
271
old_path = old_inv.id2path(file_id)
272
if specific_files:
273
if not is_inside_any(specific_files, old_path):
274
continue
275
delta.removed.append((old_path, file_id, kind))
276
277
mutter('start looking for new files')
278
for file_id in new_inv:
279
if file_id in old_inv:
280
continue
281
new_path = new_inv.id2path(file_id)
282
if specific_files:
283
if not is_inside_any(specific_files, new_path):
284
continue
285
kind = new_inv.get_file_kind(file_id)
286
delta.added.append((new_path, file_id, kind))
287
288
delta.removed.sort()
289
delta.added.sort()
290
delta.renamed.sort()
291
delta.modified.sort()
292
delta.unchanged.sort()
293
294
return delta
248
bailout("can't represent state %s {%s}" % (file_state, fid))
249
250
Loggerhead is a web-based interface for Breezy
Version: 2.0.1