14
15
# along with this program; if not, write to the Free Software
15
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from bzrlib.delta import compare_trees
18
from bzrlib.errors import BzrError
19
from bzrlib.symbol_versioning import *
20
from bzrlib.trace import mutter
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.
26
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file):
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.
36
# In the meantime we at least make sure the patch isn't
40
# Special workaround for Python2.3, where difflib fails if
41
# both sequences are empty.
42
if not oldlines and not newlines:
45
ud = difflib.unified_diff(oldlines, newlines,
46
fromfile=old_filename+'\t',
47
tofile=new_filename+'\t')
50
# work-around for difflib being too smart for its own good
51
# if /dev/null is "1,0", patch won't recognize it as /dev/null
53
ud[2] = ud[2].replace('-1,0', '-0,0')
55
ud[2] = ud[2].replace('+1,0', '+0,0')
56
# work around for difflib emitting random spaces after the label
57
ud[0] = ud[0][:-2] + '\n'
58
ud[1] = ud[1][:-2] + '\n'
62
if not line.endswith('\n'):
63
to_file.write("\n\\ No newline at end of file\n")
67
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
69
"""Display a diff by calling out to the external diff program."""
72
if to_file != sys.stdout:
73
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
76
# make sure our own output is properly ordered before the diff
79
from tempfile import NamedTemporaryFile
82
oldtmpf = NamedTemporaryFile()
83
newtmpf = NamedTemporaryFile()
86
# TODO: perhaps a special case for comparing to or from the empty
87
# sequence; can just use /dev/null on Unix
89
# TODO: if either of the files being compared already exists as a
90
# regular named file (e.g. in the working directory) then we can
91
# compare directly to that, rather than copying it.
93
oldtmpf.writelines(oldlines)
94
newtmpf.writelines(newlines)
102
'--label', old_filename+'\t',
104
'--label', new_filename+'\t',
107
# diff only allows one style to be specified; they don't override.
108
# note that some of these take optargs, and the optargs can be
109
# directly appended to the options.
110
# this is only an approximate parser; it doesn't properly understand
112
for s in ['-c', '-u', '-C', '-U',
117
'-y', '--side-by-side',
129
diffcmd.extend(diff_opts)
131
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
133
if rc != 0 and rc != 1:
134
# returns 1 if files differ; that's OK
136
msg = 'signal %d' % (-rc)
138
msg = 'exit code %d' % rc
140
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
142
oldtmpf.close() # and delete
146
@deprecated_function(zero_eight)
147
def show_diff(b, from_spec, specific_files, external_diff_options=None,
148
revision2=None, output=None, b2=None):
149
"""Shortcut for showing the diff to the working tree.
151
Please use show_diff_trees instead.
157
None for 'basis tree', or otherwise the old revision to compare against.
159
The more general form is show_diff_trees(), where the caller
160
supplies any two trees.
166
if from_spec is None:
167
old_tree = b.bzrdir.open_workingtree()
169
old_tree = old_tree = old_tree.basis_tree()
171
old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
173
if revision2 is None:
175
new_tree = b.bzrdir.open_workingtree()
177
new_tree = b2.bzrdir.open_workingtree()
179
new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
181
return show_diff_trees(old_tree, new_tree, output, specific_files,
182
external_diff_options)
185
def diff_cmd_helper(tree, specific_files, external_diff_options,
186
old_revision_spec=None, new_revision_spec=None):
187
"""Helper for cmd_diff.
193
The specific files to compare, or None
195
external_diff_options
196
If non-None, run an external diff, and pass it these options
199
If None, use basis tree as old revision, otherwise use the tree for
200
the specified revision.
203
If None, use working tree as new revision, otherwise use the tree for
204
the specified revision.
206
The more general form is show_diff_trees(), where the caller
207
supplies any two trees.
212
revision_id = spec.in_store(tree.branch).rev_id
213
return tree.branch.repository.revision_tree(revision_id)
214
if old_revision_spec is None:
215
old_tree = tree.basis_tree()
217
old_tree = spec_tree(old_revision_spec)
219
if new_revision_spec is None:
222
new_tree = spec_tree(new_revision_spec)
224
return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
225
external_diff_options)
228
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
229
external_diff_options=None):
230
"""Show in text form the changes from one tree to another.
233
If set, include only changes to these files.
235
external_diff_options
236
If set, use an external GNU diff and pass these options.
20
from trace import mutter
21
from errors import BzrError
24
def diff_trees(old_tree, new_tree):
25
"""Compute diff between two trees.
27
They may be in different branches and may be working or historical
30
Yields a sequence of (state, id, old_name, new_name, kind).
31
Each filename and each id is listed only once.
34
## TODO: Compare files before diffing; only mention those that have changed
36
## TODO: Set nice names in the headers, maybe include diffstat
38
## TODO: Perhaps make this a generator rather than using
41
## TODO: Allow specifying a list of files to compare, rather than
42
## doing the whole tree? (Not urgent.)
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.
49
## XXX: This doesn't report on unknown files; that can be done
50
## from a separate method.
52
old_it = old_tree.list_files()
53
new_it = new_tree.list_files()
243
return _show_diff_trees(old_tree, new_tree, to_file,
244
specific_files, external_diff_options)
251
def _show_diff_trees(old_tree, new_tree, to_file,
252
specific_files, external_diff_options):
254
# TODO: Options to control putting on a prefix or suffix, perhaps
255
# as a format string?
61
old_item = next(old_it)
62
new_item = next(new_it)
64
# We step through the two sorted iterators in parallel, trying to
67
while (old_item != None) or (new_item != None):
68
# OK, we still have some remaining on both, but they may be
71
old_name, old_class, old_kind, old_id = old_item
76
new_name, new_class, new_kind, new_id = new_item
80
mutter(" diff pairwise %r" % (old_item,))
81
mutter(" %r" % (new_item,))
84
# can't handle the old tree being a WorkingTree
85
assert old_class == 'V'
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
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
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
115
yield 'A', new_id, None, new_name, new_kind
117
new_item = next(new_it)
118
old_item = next(old_it)
120
assert old_id == new_id
121
assert old_id != None
122
assert old_name == new_name
123
assert old_kind == new_kind
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
135
mutter(" quick compare shows different")
136
yield 'M', new_id, old_name, new_name, new_kind
138
new_item = next(new_it)
139
old_item = next(old_it)
143
def show_diff(b, revision, file_list):
144
import difflib, sys, types
147
old_tree = b.basis_tree()
149
old_tree = b.revision_tree(b.lookup_revision(revision))
151
new_tree = b.working_tree()
153
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
259
157
DEVNULL = '/dev/null'
260
158
# Windows users, don't panic about this filename -- it is a
264
162
# TODO: Generation of pseudo-diffs for added/deleted files could
265
163
# be usefully made into a much faster special case.
267
if external_diff_options:
268
assert isinstance(external_diff_options, basestring)
269
opts = external_diff_options.split()
270
def diff_file(olab, olines, nlab, nlines, to_file):
271
external_diff(olab, olines, nlab, nlines, to_file, opts)
273
diff_file = internal_diff
276
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
277
specific_files=specific_files)
280
for path, file_id, kind in delta.removed:
282
print >>to_file, '=== removed %s %r' % (kind, old_label + path)
283
old_tree.inventory[file_id].diff(diff_file, old_label + path, old_tree,
284
DEVNULL, None, None, to_file)
285
for path, file_id, kind in delta.added:
287
print >>to_file, '=== added %s %r' % (kind, new_label + path)
288
new_tree.inventory[file_id].diff(diff_file, new_label + path, new_tree,
289
DEVNULL, None, None, to_file,
291
for (old_path, new_path, file_id, kind,
292
text_modified, meta_modified) in delta.renamed:
294
prop_str = get_prop_change(meta_modified)
295
print >>to_file, '=== renamed %s %r => %r%s' % (
296
kind, old_label + old_path, new_label + new_path, prop_str)
297
_maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
298
new_label, new_path, new_tree,
299
text_modified, kind, to_file, diff_file)
300
for path, file_id, kind, text_modified, meta_modified in delta.modified:
302
prop_str = get_prop_change(meta_modified)
303
print >>to_file, '=== modified %s %r%s' % (kind, old_label + path,
306
_maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
307
new_label, path, new_tree,
308
True, kind, to_file, diff_file)
312
def get_prop_change(meta_modified):
314
return " (properties changed)"
319
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
320
new_label, new_path, new_tree, text_modified,
321
kind, to_file, diff_file):
323
new_entry = new_tree.inventory[file_id]
324
old_tree.inventory[file_id].diff(diff_file,
325
old_label + old_path, old_tree,
326
new_label + new_path, new_entry,
165
# TODO: Better to return them in sorted order I think.
168
file_list = [b.relpath(f) for f in file_list]
170
# FIXME: If given a file list, compare only those files rather
171
# than comparing everything and then throwing stuff away.
173
for file_state, fid, old_name, new_name, kind in diff_trees(old_tree, new_tree):
175
if file_list and (new_name not in file_list):
178
# Don't show this by default; maybe do it if an option is passed
179
# idlabel = ' {%s}' % fid
182
# FIXME: Something about the diff format makes patch unhappy
183
# with newly-added files.
185
def diffit(oldlines, newlines, **kw):
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.
194
# In the meantime we at least make sure the patch isn't
198
# Special workaround for Python2.3, where difflib fails if
199
# both sequences are empty.
200
if not oldlines and not newlines:
205
if oldlines and (oldlines[-1][-1] != '\n'):
208
if newlines and (newlines[-1][-1] != '\n'):
212
ud = difflib.unified_diff(oldlines, newlines, **kw)
213
sys.stdout.writelines(ud)
215
print "\\ No newline at end of file"
216
sys.stdout.write('\n')
218
if file_state in ['.', '?', 'I']:
220
elif file_state == 'A':
221
print '*** added %s %r' % (kind, new_name)
224
new_tree.get_file(fid).readlines(),
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)
231
diffit(old_tree.get_file(fid).readlines(), [],
232
fromfile=old_label + old_name + idlabel,
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)
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)
248
raise BzrError("can't represent state %s {%s}" % (file_state, fid))
253
"""Describes changes from one tree to another.
262
(oldpath, newpath, id)
266
A path may occur in more than one list if it was e.g. deleted
267
under an old id and renamed into place in a new id.
269
Files are listed in either modified or renamed, not both. In
270
other words, renamed files may also be modified.
279
def compare_inventories(old_inv, new_inv):
280
"""Return a TreeDelta object describing changes between inventories.
282
This only describes changes in the shape of the tree, not the
285
This is an alternative to diff_trees() and should probably
286
eventually replace it.
288
old_ids = old_inv.id_set()
289
new_ids = new_inv.id_set()
292
delta.removed = [(old_inv.id2path(fid), fid) for fid in (old_ids - new_ids)]
295
delta.added = [(new_inv.id2path(fid), fid) for fid in (new_ids - old_ids)]
298
for fid in old_ids & new_ids:
299
old_ie = old_inv[fid]
300
new_ie = new_inv[fid]
301
old_path = old_inv.id2path(fid)
302
new_path = new_inv.id2path(fid)
304
if old_path != new_path:
305
delta.renamed.append((old_path, new_path, fid))
306
elif old_ie.text_sha1 != new_ie.text_sha1:
307
delta.modified.append((new_path, fid))
309
delta.modified.sort()
added
removed
bzrlib/diff.py
