~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

Show diffs side-by-side

added added

removed removed

Lines of Context:
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
 
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
16
 
17
 
import errno
 
17
import difflib
18
18
import os
19
19
import re
 
20
import shutil
 
21
import sys
 
22
 
 
23
from bzrlib.lazy_import import lazy_import
 
24
lazy_import(globals(), """
 
25
import errno
20
26
import subprocess
21
 
import sys
22
27
import tempfile
23
28
import time
24
29
 
25
 
# compatability - plugins import compare_trees from diff!!!
26
 
# deprecated as of 0.10
27
 
from bzrlib.delta import compare_trees
28
 
from bzrlib.errors import BzrError
29
 
import bzrlib.errors as errors
30
 
import bzrlib.osutils
31
 
from bzrlib.patiencediff import unified_diff
32
 
import bzrlib.patiencediff
33
 
from bzrlib.symbol_versioning import (deprecated_function,
34
 
        zero_eight)
35
 
from bzrlib.textfile import check_text_lines
36
 
from bzrlib.trace import mutter, warning
 
30
from bzrlib import (
 
31
    branch as _mod_branch,
 
32
    bzrdir,
 
33
    commands,
 
34
    errors,
 
35
    osutils,
 
36
    patiencediff,
 
37
    textfile,
 
38
    timestamp,
 
39
    views,
 
40
    )
 
41
""")
 
42
 
 
43
from bzrlib.symbol_versioning import (
 
44
    deprecated_function,
 
45
    )
 
46
from bzrlib.trace import mutter, note, warning
37
47
 
38
48
 
39
49
# TODO: Rather than building a changeset object, we should probably
40
50
# invoke callbacks on an object.  That object can either accumulate a
41
51
# list, write them out directly, etc etc.
42
52
 
 
53
 
 
54
class _PrematchedMatcher(difflib.SequenceMatcher):
 
55
    """Allow SequenceMatcher operations to use predetermined blocks"""
 
56
 
 
57
    def __init__(self, matching_blocks):
 
58
        difflib.SequenceMatcher(self, None, None)
 
59
        self.matching_blocks = matching_blocks
 
60
        self.opcodes = None
 
61
 
 
62
 
43
63
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
44
64
                  allow_binary=False, sequence_matcher=None,
45
65
                  path_encoding='utf8'):
58
78
    # both sequences are empty.
59
79
    if not oldlines and not newlines:
60
80
        return
61
 
    
 
81
 
62
82
    if allow_binary is False:
63
 
        check_text_lines(oldlines)
64
 
        check_text_lines(newlines)
 
83
        textfile.check_text_lines(oldlines)
 
84
        textfile.check_text_lines(newlines)
65
85
 
66
86
    if sequence_matcher is None:
67
 
        sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
68
 
    ud = unified_diff(oldlines, newlines,
 
87
        sequence_matcher = patiencediff.PatienceSequenceMatcher
 
88
    ud = patiencediff.unified_diff(oldlines, newlines,
69
89
                      fromfile=old_filename.encode(path_encoding),
70
90
                      tofile=new_filename.encode(path_encoding),
71
91
                      sequencematcher=sequence_matcher)
72
92
 
73
93
    ud = list(ud)
 
94
    if len(ud) == 0: # Identical contents, nothing to do
 
95
        return
74
96
    # work-around for difflib being too smart for its own good
75
97
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
76
98
    if not oldlines:
77
99
        ud[2] = ud[2].replace('-1,0', '-0,0')
78
100
    elif not newlines:
79
101
        ud[2] = ud[2].replace('+1,0', '+0,0')
80
 
    # work around for difflib emitting random spaces after the label
81
 
    ud[0] = ud[0][:-2] + '\n'
82
 
    ud[1] = ud[1][:-2] + '\n'
83
102
 
84
103
    for line in ud:
85
104
        to_file.write(line)
86
105
        if not line.endswith('\n'):
87
106
            to_file.write("\n\\ No newline at end of file\n")
88
 
    print >>to_file
 
107
    to_file.write('\n')
 
108
 
 
109
 
 
110
def _spawn_external_diff(diffcmd, capture_errors=True):
 
111
    """Spawn the externall diff process, and return the child handle.
 
112
 
 
113
    :param diffcmd: The command list to spawn
 
114
    :param capture_errors: Capture stderr as well as setting LANG=C
 
115
        and LC_ALL=C. This lets us read and understand the output of diff,
 
116
        and respond to any errors.
 
117
    :return: A Popen object.
 
118
    """
 
119
    if capture_errors:
 
120
        # construct minimal environment
 
121
        env = {}
 
122
        path = os.environ.get('PATH')
 
123
        if path is not None:
 
124
            env['PATH'] = path
 
125
        env['LANGUAGE'] = 'C'   # on win32 only LANGUAGE has effect
 
126
        env['LANG'] = 'C'
 
127
        env['LC_ALL'] = 'C'
 
128
        stderr = subprocess.PIPE
 
129
    else:
 
130
        env = None
 
131
        stderr = None
 
132
 
 
133
    try:
 
134
        pipe = subprocess.Popen(diffcmd,
 
135
                                stdin=subprocess.PIPE,
 
136
                                stdout=subprocess.PIPE,
 
137
                                stderr=stderr,
 
138
                                env=env)
 
139
    except OSError, e:
 
140
        if e.errno == errno.ENOENT:
 
141
            raise errors.NoDiff(str(e))
 
142
        raise
 
143
 
 
144
    return pipe
89
145
 
90
146
 
91
147
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
115
171
 
116
172
        if not diff_opts:
117
173
            diff_opts = []
 
174
        if sys.platform == 'win32':
 
175
            # Popen doesn't do the proper encoding for external commands
 
176
            # Since we are dealing with an ANSI api, use mbcs encoding
 
177
            old_filename = old_filename.encode('mbcs')
 
178
            new_filename = new_filename.encode('mbcs')
118
179
        diffcmd = ['diff',
119
180
                   '--label', old_filename,
120
181
                   old_abspath,
143
204
            break
144
205
        else:
145
206
            diffcmd.append('-u')
146
 
                  
 
207
 
147
208
        if diff_opts:
148
209
            diffcmd.extend(diff_opts)
149
210
 
150
 
        try:
151
 
            pipe = subprocess.Popen(diffcmd,
152
 
                                    stdin=subprocess.PIPE,
153
 
                                    stdout=subprocess.PIPE)
154
 
        except OSError, e:
155
 
            if e.errno == errno.ENOENT:
156
 
                raise errors.NoDiff(str(e))
157
 
            raise
158
 
        pipe.stdin.close()
 
211
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
 
212
        out,err = pipe.communicate()
 
213
        rc = pipe.returncode
159
214
 
160
 
        first_line = pipe.stdout.readline()
161
 
        to_file.write(first_line)
162
 
        bzrlib.osutils.pumpfile(pipe.stdout, to_file)
163
 
        rc = pipe.wait()
164
 
        
 
215
        # internal_diff() adds a trailing newline, add one here for consistency
 
216
        out += '\n'
165
217
        if rc == 2:
166
218
            # 'diff' gives retcode == 2 for all sorts of errors
167
219
            # one of those is 'Binary files differ'.
168
220
            # Bad options could also be the problem.
169
 
            # 'Binary files' is not a real error, so we suppress that error
170
 
            m = re.match('^binary files.*differ$', first_line, re.I)
171
 
            if not m:
172
 
                raise BzrError('external diff failed with exit code 2;'
173
 
                               ' command: %r' % (diffcmd,))
174
 
        elif rc not in (0, 1):
 
221
            # 'Binary files' is not a real error, so we suppress that error.
 
222
            lang_c_out = out
 
223
 
 
224
            # Since we got here, we want to make sure to give an i18n error
 
225
            pipe = _spawn_external_diff(diffcmd, capture_errors=False)
 
226
            out, err = pipe.communicate()
 
227
 
 
228
            # Write out the new i18n diff response
 
229
            to_file.write(out+'\n')
 
230
            if pipe.returncode != 2:
 
231
                raise errors.BzrError(
 
232
                               'external diff failed with exit code 2'
 
233
                               ' when run with LANG=C and LC_ALL=C,'
 
234
                               ' but not when run natively: %r' % (diffcmd,))
 
235
 
 
236
            first_line = lang_c_out.split('\n', 1)[0]
 
237
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
 
238
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
 
239
            if m is None:
 
240
                raise errors.BzrError('external diff failed with exit code 2;'
 
241
                                      ' command: %r' % (diffcmd,))
 
242
            else:
 
243
                # Binary files differ, just return
 
244
                return
 
245
 
 
246
        # If we got to here, we haven't written out the output of diff
 
247
        # do so now
 
248
        to_file.write(out)
 
249
        if rc not in (0, 1):
175
250
            # returns 1 if files differ; that's OK
176
251
            if rc < 0:
177
252
                msg = 'signal %d' % (-rc)
178
253
            else:
179
254
                msg = 'exit code %d' % rc
180
 
                
181
 
            raise BzrError('external diff failed with %s; command: %r' 
182
 
                           % (rc, diffcmd))
183
 
 
184
 
        # internal_diff() adds a trailing newline, add one here for consistency
185
 
        to_file.write('\n')
 
255
 
 
256
            raise errors.BzrError('external diff failed with %s; command: %r'
 
257
                                  % (rc, diffcmd))
 
258
 
186
259
 
187
260
    finally:
188
261
        oldtmpf.close()                 # and delete
204
277
                        new_abspath, e)
205
278
 
206
279
 
207
 
@deprecated_function(zero_eight)
208
 
def show_diff(b, from_spec, specific_files, external_diff_options=None,
209
 
              revision2=None, output=None, b2=None):
210
 
    """Shortcut for showing the diff to the working tree.
211
 
 
212
 
    Please use show_diff_trees instead.
213
 
 
214
 
    b
215
 
        Branch.
216
 
 
217
 
    revision
218
 
        None for 'basis tree', or otherwise the old revision to compare against.
219
 
    
220
 
    The more general form is show_diff_trees(), where the caller
221
 
    supplies any two trees.
222
 
    """
223
 
    if output is None:
224
 
        output = sys.stdout
225
 
 
226
 
    if from_spec is None:
227
 
        old_tree = b.bzrdir.open_workingtree()
228
 
        if b2 is None:
229
 
            old_tree = old_tree = old_tree.basis_tree()
230
 
    else:
231
 
        old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
232
 
 
233
 
    if revision2 is None:
234
 
        if b2 is None:
235
 
            new_tree = b.bzrdir.open_workingtree()
236
 
        else:
237
 
            new_tree = b2.bzrdir.open_workingtree()
238
 
    else:
239
 
        new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
240
 
 
241
 
    return show_diff_trees(old_tree, new_tree, output, specific_files,
242
 
                           external_diff_options)
243
 
 
244
 
 
245
 
def diff_cmd_helper(tree, specific_files, external_diff_options, 
246
 
                    old_revision_spec=None, new_revision_spec=None,
247
 
                    old_label='a/', new_label='b/'):
248
 
    """Helper for cmd_diff.
249
 
 
250
 
   tree 
251
 
        A WorkingTree
252
 
 
253
 
    specific_files
254
 
        The specific files to compare, or None
255
 
 
256
 
    external_diff_options
257
 
        If non-None, run an external diff, and pass it these options
258
 
 
259
 
    old_revision_spec
260
 
        If None, use basis tree as old revision, otherwise use the tree for
261
 
        the specified revision. 
262
 
 
263
 
    new_revision_spec
264
 
        If None, use working tree as new revision, otherwise use the tree for
265
 
        the specified revision.
266
 
    
267
 
    The more general form is show_diff_trees(), where the caller
268
 
    supplies any two trees.
269
 
    """
270
 
    def spec_tree(spec):
271
 
        if tree:
272
 
            revision = spec.in_store(tree.branch)
273
 
        else:
274
 
            revision = spec.in_store(None)
275
 
        revision_id = revision.rev_id
276
 
        branch = revision.branch
277
 
        return branch.repository.revision_tree(revision_id)
278
 
    if old_revision_spec is None:
279
 
        old_tree = tree.basis_tree()
280
 
    else:
281
 
        old_tree = spec_tree(old_revision_spec)
282
 
 
283
 
    if new_revision_spec is None:
284
 
        new_tree = tree
285
 
    else:
286
 
        new_tree = spec_tree(new_revision_spec)
287
 
    if new_tree is not tree:
288
 
        extra_trees = (tree,)
289
 
    else:
290
 
        extra_trees = None
291
 
 
292
 
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
293
 
                           external_diff_options,
294
 
                           old_label=old_label, new_label=new_label,
295
 
                           extra_trees=extra_trees)
 
280
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url,
 
281
    apply_view=True):
 
282
    """Get the trees and specific files to diff given a list of paths.
 
283
 
 
284
    This method works out the trees to be diff'ed and the files of
 
285
    interest within those trees.
 
286
 
 
287
    :param path_list:
 
288
        the list of arguments passed to the diff command
 
289
    :param revision_specs:
 
290
        Zero, one or two RevisionSpecs from the diff command line,
 
291
        saying what revisions to compare.
 
292
    :param old_url:
 
293
        The url of the old branch or tree. If None, the tree to use is
 
294
        taken from the first path, if any, or the current working tree.
 
295
    :param new_url:
 
296
        The url of the new branch or tree. If None, the tree to use is
 
297
        taken from the first path, if any, or the current working tree.
 
298
    :param apply_view:
 
299
        if True and a view is set, apply the view or check that the paths
 
300
        are within it
 
301
    :returns:
 
302
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
303
        extra_trees is a sequence of additional trees to search in for
 
304
        file-ids.
 
305
    """
 
306
    # Get the old and new revision specs
 
307
    old_revision_spec = None
 
308
    new_revision_spec = None
 
309
    if revision_specs is not None:
 
310
        if len(revision_specs) > 0:
 
311
            old_revision_spec = revision_specs[0]
 
312
            if old_url is None:
 
313
                old_url = old_revision_spec.get_branch()
 
314
        if len(revision_specs) > 1:
 
315
            new_revision_spec = revision_specs[1]
 
316
            if new_url is None:
 
317
                new_url = new_revision_spec.get_branch()
 
318
 
 
319
    other_paths = []
 
320
    make_paths_wt_relative = True
 
321
    consider_relpath = True
 
322
    if path_list is None or len(path_list) == 0:
 
323
        # If no path is given, the current working tree is used
 
324
        default_location = u'.'
 
325
        consider_relpath = False
 
326
    elif old_url is not None and new_url is not None:
 
327
        other_paths = path_list
 
328
        make_paths_wt_relative = False
 
329
    else:
 
330
        default_location = path_list[0]
 
331
        other_paths = path_list[1:]
 
332
 
 
333
    # Get the old location
 
334
    specific_files = []
 
335
    if old_url is None:
 
336
        old_url = default_location
 
337
    working_tree, branch, relpath = \
 
338
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
 
339
    if consider_relpath and relpath != '':
 
340
        if working_tree is not None and apply_view:
 
341
            views.check_path_in_view(working_tree, relpath)
 
342
        specific_files.append(relpath)
 
343
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
 
344
 
 
345
    # Get the new location
 
346
    if new_url is None:
 
347
        new_url = default_location
 
348
    if new_url != old_url:
 
349
        working_tree, branch, relpath = \
 
350
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
 
351
        if consider_relpath and relpath != '':
 
352
            if working_tree is not None and apply_view:
 
353
                views.check_path_in_view(working_tree, relpath)
 
354
            specific_files.append(relpath)
 
355
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
 
356
        basis_is_default=working_tree is None)
 
357
 
 
358
    # Get the specific files (all files is None, no files is [])
 
359
    if make_paths_wt_relative and working_tree is not None:
 
360
        try:
 
361
            from bzrlib.builtins import safe_relpath_files
 
362
            other_paths = safe_relpath_files(working_tree, other_paths,
 
363
            apply_view=apply_view)
 
364
        except errors.FileInWrongBranch:
 
365
            raise errors.BzrCommandError("Files are in different branches")
 
366
    specific_files.extend(other_paths)
 
367
    if len(specific_files) == 0:
 
368
        specific_files = None
 
369
        if (working_tree is not None and working_tree.supports_views()
 
370
            and apply_view):
 
371
            view_files = working_tree.views.lookup_view()
 
372
            if view_files:
 
373
                specific_files = view_files
 
374
                view_str = views.view_display_str(view_files)
 
375
                note("*** Ignoring files outside view. View is %s" % view_str)
 
376
 
 
377
    # Get extra trees that ought to be searched for file-ids
 
378
    extra_trees = None
 
379
    if working_tree is not None and working_tree not in (old_tree, new_tree):
 
380
        extra_trees = (working_tree,)
 
381
    return old_tree, new_tree, specific_files, extra_trees
 
382
 
 
383
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
 
384
    if branch is None and tree is not None:
 
385
        branch = tree.branch
 
386
    if spec is None or spec.spec is None:
 
387
        if basis_is_default:
 
388
            if tree is not None:
 
389
                return tree.basis_tree()
 
390
            else:
 
391
                return branch.basis_tree()
 
392
        else:
 
393
            return tree
 
394
    return spec.as_tree(branch)
296
395
 
297
396
 
298
397
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
299
398
                    external_diff_options=None,
300
399
                    old_label='a/', new_label='b/',
301
 
                    extra_trees=None):
 
400
                    extra_trees=None,
 
401
                    path_encoding='utf8',
 
402
                    using=None):
302
403
    """Show in text form the changes from one tree to another.
303
404
 
304
 
    to_files
305
 
        If set, include only changes to these files.
 
405
    to_file
 
406
        The output stream.
 
407
 
 
408
    specific_files
 
409
        Include only changes to these files - None for all changes.
306
410
 
307
411
    external_diff_options
308
412
        If set, use an external GNU diff and pass these options.
309
413
 
310
414
    extra_trees
311
415
        If set, more Trees to use for looking up file ids
 
416
 
 
417
    path_encoding
 
418
        If set, the path will be encoded as specified, otherwise is supposed
 
419
        to be utf8
312
420
    """
313
421
    old_tree.lock_read()
314
422
    try:
 
423
        if extra_trees is not None:
 
424
            for tree in extra_trees:
 
425
                tree.lock_read()
315
426
        new_tree.lock_read()
316
427
        try:
317
 
            return _show_diff_trees(old_tree, new_tree, to_file,
318
 
                                    specific_files, external_diff_options,
319
 
                                    old_label=old_label, new_label=new_label,
320
 
                                    extra_trees=extra_trees)
 
428
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
429
                                                 path_encoding,
 
430
                                                 external_diff_options,
 
431
                                                 old_label, new_label, using)
 
432
            return differ.show_diff(specific_files, extra_trees)
321
433
        finally:
322
434
            new_tree.unlock()
 
435
            if extra_trees is not None:
 
436
                for tree in extra_trees:
 
437
                    tree.unlock()
323
438
    finally:
324
439
        old_tree.unlock()
325
440
 
326
441
 
327
 
def _show_diff_trees(old_tree, new_tree, to_file,
328
 
                     specific_files, external_diff_options, 
329
 
                     old_label='a/', new_label='b/', extra_trees=None):
 
442
def _patch_header_date(tree, file_id, path):
 
443
    """Returns a timestamp suitable for use in a patch header."""
 
444
    mtime = tree.get_file_mtime(file_id, path)
 
445
    return timestamp.format_patch_date(mtime)
 
446
 
 
447
 
 
448
def get_executable_change(old_is_x, new_is_x):
 
449
    descr = { True:"+x", False:"-x", None:"??" }
 
450
    if old_is_x != new_is_x:
 
451
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
452
    else:
 
453
        return []
 
454
 
 
455
 
 
456
class DiffPath(object):
 
457
    """Base type for command object that compare files"""
 
458
 
 
459
    # The type or contents of the file were unsuitable for diffing
 
460
    CANNOT_DIFF = 'CANNOT_DIFF'
 
461
    # The file has changed in a semantic way
 
462
    CHANGED = 'CHANGED'
 
463
    # The file content may have changed, but there is no semantic change
 
464
    UNCHANGED = 'UNCHANGED'
 
465
 
 
466
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
 
467
        """Constructor.
 
468
 
 
469
        :param old_tree: The tree to show as the old tree in the comparison
 
470
        :param new_tree: The tree to show as new in the comparison
 
471
        :param to_file: The file to write comparison data to
 
472
        :param path_encoding: The character encoding to write paths in
 
473
        """
 
474
        self.old_tree = old_tree
 
475
        self.new_tree = new_tree
 
476
        self.to_file = to_file
 
477
        self.path_encoding = path_encoding
 
478
 
 
479
    def finish(self):
 
480
        pass
 
481
 
 
482
    @classmethod
 
483
    def from_diff_tree(klass, diff_tree):
 
484
        return klass(diff_tree.old_tree, diff_tree.new_tree,
 
485
                     diff_tree.to_file, diff_tree.path_encoding)
 
486
 
 
487
    @staticmethod
 
488
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
 
489
        for file_differ in differs:
 
490
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
491
                                      new_kind)
 
492
            if result is not DiffPath.CANNOT_DIFF:
 
493
                return result
 
494
        else:
 
495
            return DiffPath.CANNOT_DIFF
 
496
 
 
497
 
 
498
class DiffKindChange(object):
 
499
    """Special differ for file kind changes.
 
500
 
 
501
    Represents kind change as deletion + creation.  Uses the other differs
 
502
    to do this.
 
503
    """
 
504
    def __init__(self, differs):
 
505
        self.differs = differs
 
506
 
 
507
    def finish(self):
 
508
        pass
 
509
 
 
510
    @classmethod
 
511
    def from_diff_tree(klass, diff_tree):
 
512
        return klass(diff_tree.differs)
 
513
 
 
514
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
515
        """Perform comparison
 
516
 
 
517
        :param file_id: The file_id of the file to compare
 
518
        :param old_path: Path of the file in the old tree
 
519
        :param new_path: Path of the file in the new tree
 
520
        :param old_kind: Old file-kind of the file
 
521
        :param new_kind: New file-kind of the file
 
522
        """
 
523
        if None in (old_kind, new_kind):
 
524
            return DiffPath.CANNOT_DIFF
 
525
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
526
                                       new_path, old_kind, None)
 
527
        if result is DiffPath.CANNOT_DIFF:
 
528
            return result
 
529
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
530
                                     None, new_kind)
 
531
 
 
532
 
 
533
class DiffDirectory(DiffPath):
 
534
 
 
535
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
536
        """Perform comparison between two directories.  (dummy)
 
537
 
 
538
        """
 
539
        if 'directory' not in (old_kind, new_kind):
 
540
            return self.CANNOT_DIFF
 
541
        if old_kind not in ('directory', None):
 
542
            return self.CANNOT_DIFF
 
543
        if new_kind not in ('directory', None):
 
544
            return self.CANNOT_DIFF
 
545
        return self.CHANGED
 
546
 
 
547
 
 
548
class DiffSymlink(DiffPath):
 
549
 
 
550
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
551
        """Perform comparison between two symlinks
 
552
 
 
553
        :param file_id: The file_id of the file to compare
 
554
        :param old_path: Path of the file in the old tree
 
555
        :param new_path: Path of the file in the new tree
 
556
        :param old_kind: Old file-kind of the file
 
557
        :param new_kind: New file-kind of the file
 
558
        """
 
559
        if 'symlink' not in (old_kind, new_kind):
 
560
            return self.CANNOT_DIFF
 
561
        if old_kind == 'symlink':
 
562
            old_target = self.old_tree.get_symlink_target(file_id)
 
563
        elif old_kind is None:
 
564
            old_target = None
 
565
        else:
 
566
            return self.CANNOT_DIFF
 
567
        if new_kind == 'symlink':
 
568
            new_target = self.new_tree.get_symlink_target(file_id)
 
569
        elif new_kind is None:
 
570
            new_target = None
 
571
        else:
 
572
            return self.CANNOT_DIFF
 
573
        return self.diff_symlink(old_target, new_target)
 
574
 
 
575
    def diff_symlink(self, old_target, new_target):
 
576
        if old_target is None:
 
577
            self.to_file.write('=== target is %r\n' % new_target)
 
578
        elif new_target is None:
 
579
            self.to_file.write('=== target was %r\n' % old_target)
 
580
        else:
 
581
            self.to_file.write('=== target changed %r => %r\n' %
 
582
                              (old_target, new_target))
 
583
        return self.CHANGED
 
584
 
 
585
 
 
586
class DiffText(DiffPath):
330
587
 
331
588
    # GNU Patch uses the epoch date to detect files that are being added
332
589
    # or removed in a diff.
333
590
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
334
591
 
335
 
    # TODO: Generation of pseudo-diffs for added/deleted files could
336
 
    # be usefully made into a much faster special case.
337
 
 
338
 
    if external_diff_options:
339
 
        assert isinstance(external_diff_options, basestring)
340
 
        opts = external_diff_options.split()
341
 
        def diff_file(olab, olines, nlab, nlines, to_file):
342
 
            external_diff(olab, olines, nlab, nlines, to_file, opts)
343
 
    else:
344
 
        diff_file = internal_diff
345
 
    
346
 
    delta = new_tree.changes_from(old_tree,
347
 
        specific_files=specific_files,
348
 
        extra_trees=extra_trees, require_versioned=True)
349
 
 
350
 
    has_changes = 0
351
 
    for path, file_id, kind in delta.removed:
352
 
        has_changes = 1
353
 
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
354
 
        old_name = '%s%s\t%s' % (old_label, path,
355
 
                                 _patch_header_date(old_tree, file_id, path))
356
 
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
357
 
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
358
 
                                         new_name, None, None, to_file)
359
 
    for path, file_id, kind in delta.added:
360
 
        has_changes = 1
361
 
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
362
 
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
363
 
        new_name = '%s%s\t%s' % (new_label, path,
364
 
                                 _patch_header_date(new_tree, file_id, path))
365
 
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
366
 
                                         old_name, None, None, to_file, 
367
 
                                         reverse=True)
368
 
    for (old_path, new_path, file_id, kind,
369
 
         text_modified, meta_modified) in delta.renamed:
370
 
        has_changes = 1
371
 
        prop_str = get_prop_change(meta_modified)
372
 
        print >>to_file, '=== renamed %s %r => %r%s' % (
373
 
                    kind, old_path.encode('utf8'),
374
 
                    new_path.encode('utf8'), prop_str)
375
 
        old_name = '%s%s\t%s' % (old_label, old_path,
376
 
                                 _patch_header_date(old_tree, file_id,
377
 
                                                    old_path))
378
 
        new_name = '%s%s\t%s' % (new_label, new_path,
379
 
                                 _patch_header_date(new_tree, file_id,
380
 
                                                    new_path))
381
 
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
382
 
                                    new_name, new_tree,
383
 
                                    text_modified, kind, to_file, diff_file)
384
 
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
385
 
        has_changes = 1
386
 
        prop_str = get_prop_change(meta_modified)
387
 
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
388
 
        old_name = '%s%s\t%s' % (old_label, path,
389
 
                                 _patch_header_date(old_tree, file_id, path))
390
 
        new_name = '%s%s\t%s' % (new_label, path,
391
 
                                 _patch_header_date(new_tree, file_id, path))
392
 
        if text_modified:
393
 
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
394
 
                                        new_name, new_tree,
395
 
                                        True, kind, to_file, diff_file)
396
 
 
397
 
    return has_changes
398
 
 
399
 
 
400
 
def _patch_header_date(tree, file_id, path):
401
 
    """Returns a timestamp suitable for use in a patch header."""
402
 
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
403
 
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
404
 
 
405
 
 
406
 
def _raise_if_nonexistent(paths, old_tree, new_tree):
407
 
    """Complain if paths are not in either inventory or tree.
408
 
 
409
 
    It's OK with the files exist in either tree's inventory, or 
410
 
    if they exist in the tree but are not versioned.
411
 
    
412
 
    This can be used by operations such as bzr status that can accept
413
 
    unknown or ignored files.
 
592
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
593
                 old_label='', new_label='', text_differ=internal_diff):
 
594
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
595
        self.text_differ = text_differ
 
596
        self.old_label = old_label
 
597
        self.new_label = new_label
 
598
        self.path_encoding = path_encoding
 
599
 
 
600
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
601
        """Compare two files in unified diff format
 
602
 
 
603
        :param file_id: The file_id of the file to compare
 
604
        :param old_path: Path of the file in the old tree
 
605
        :param new_path: Path of the file in the new tree
 
606
        :param old_kind: Old file-kind of the file
 
607
        :param new_kind: New file-kind of the file
 
608
        """
 
609
        if 'file' not in (old_kind, new_kind):
 
610
            return self.CANNOT_DIFF
 
611
        from_file_id = to_file_id = file_id
 
612
        if old_kind == 'file':
 
613
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
 
614
        elif old_kind is None:
 
615
            old_date = self.EPOCH_DATE
 
616
            from_file_id = None
 
617
        else:
 
618
            return self.CANNOT_DIFF
 
619
        if new_kind == 'file':
 
620
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
 
621
        elif new_kind is None:
 
622
            new_date = self.EPOCH_DATE
 
623
            to_file_id = None
 
624
        else:
 
625
            return self.CANNOT_DIFF
 
626
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
627
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
628
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
 
629
            old_path, new_path)
 
630
 
 
631
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
 
632
        from_path=None, to_path=None):
 
633
        """Diff the content of given files in two trees
 
634
 
 
635
        :param from_file_id: The id of the file in the from tree.  If None,
 
636
            the file is not present in the from tree.
 
637
        :param to_file_id: The id of the file in the to tree.  This may refer
 
638
            to a different file from from_file_id.  If None,
 
639
            the file is not present in the to tree.
 
640
        :param from_path: The path in the from tree or None if unknown.
 
641
        :param to_path: The path in the to tree or None if unknown.
 
642
        """
 
643
        def _get_text(tree, file_id, path):
 
644
            if file_id is not None:
 
645
                return tree.get_file(file_id, path).readlines()
 
646
            else:
 
647
                return []
 
648
        try:
 
649
            from_text = _get_text(self.old_tree, from_file_id, from_path)
 
650
            to_text = _get_text(self.new_tree, to_file_id, to_path)
 
651
            self.text_differ(from_label, from_text, to_label, to_text,
 
652
                             self.to_file)
 
653
        except errors.BinaryFile:
 
654
            self.to_file.write(
 
655
                  ("Binary files %s and %s differ\n" %
 
656
                  (from_label, to_label)).encode(self.path_encoding))
 
657
        return self.CHANGED
 
658
 
 
659
 
 
660
class DiffFromTool(DiffPath):
 
661
 
 
662
    def __init__(self, command_template, old_tree, new_tree, to_file,
 
663
                 path_encoding='utf-8'):
 
664
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
665
        self.command_template = command_template
 
666
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
 
667
 
 
668
    @classmethod
 
669
    def from_string(klass, command_string, old_tree, new_tree, to_file,
 
670
                    path_encoding='utf-8'):
 
671
        command_template = commands.shlex_split_unicode(command_string)
 
672
        command_template.extend(['%(old_path)s', '%(new_path)s'])
 
673
        return klass(command_template, old_tree, new_tree, to_file,
 
674
                     path_encoding)
 
675
 
 
676
    @classmethod
 
677
    def make_from_diff_tree(klass, command_string):
 
678
        def from_diff_tree(diff_tree):
 
679
            return klass.from_string(command_string, diff_tree.old_tree,
 
680
                                     diff_tree.new_tree, diff_tree.to_file)
 
681
        return from_diff_tree
 
682
 
 
683
    def _get_command(self, old_path, new_path):
 
684
        my_map = {'old_path': old_path, 'new_path': new_path}
 
685
        return [t % my_map for t in self.command_template]
 
686
 
 
687
    def _execute(self, old_path, new_path):
 
688
        command = self._get_command(old_path, new_path)
 
689
        try:
 
690
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
 
691
                                    cwd=self._root)
 
692
        except OSError, e:
 
693
            if e.errno == errno.ENOENT:
 
694
                raise errors.ExecutableMissing(command[0])
 
695
            else:
 
696
                raise
 
697
        self.to_file.write(proc.stdout.read())
 
698
        return proc.wait()
 
699
 
 
700
    def _try_symlink_root(self, tree, prefix):
 
701
        if (getattr(tree, 'abspath', None) is None
 
702
            or not osutils.host_os_dereferences_symlinks()):
 
703
            return False
 
704
        try:
 
705
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
 
706
        except OSError, e:
 
707
            if e.errno != errno.EEXIST:
 
708
                raise
 
709
        return True
 
710
 
 
711
    def _write_file(self, file_id, tree, prefix, relpath):
 
712
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
713
        if self._try_symlink_root(tree, prefix):
 
714
            return full_path
 
715
        parent_dir = osutils.dirname(full_path)
 
716
        try:
 
717
            os.makedirs(parent_dir)
 
718
        except OSError, e:
 
719
            if e.errno != errno.EEXIST:
 
720
                raise
 
721
        source = tree.get_file(file_id, relpath)
 
722
        try:
 
723
            target = open(full_path, 'wb')
 
724
            try:
 
725
                osutils.pumpfile(source, target)
 
726
            finally:
 
727
                target.close()
 
728
        finally:
 
729
            source.close()
 
730
        osutils.make_readonly(full_path)
 
731
        mtime = tree.get_file_mtime(file_id)
 
732
        os.utime(full_path, (mtime, mtime))
 
733
        return full_path
 
734
 
 
735
    def _prepare_files(self, file_id, old_path, new_path):
 
736
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
737
                                         old_path)
 
738
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
739
                                         new_path)
 
740
        return old_disk_path, new_disk_path
 
741
 
 
742
    def finish(self):
 
743
        try:
 
744
            osutils.rmtree(self._root)
 
745
        except OSError, e:
 
746
            if e.errno != errno.ENOENT:
 
747
                mutter("The temporary directory \"%s\" was not "
 
748
                        "cleanly removed: %s." % (self._root, e))
 
749
 
 
750
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
751
        if (old_kind, new_kind) != ('file', 'file'):
 
752
            return DiffPath.CANNOT_DIFF
 
753
        self._prepare_files(file_id, old_path, new_path)
 
754
        self._execute(osutils.pathjoin('old', old_path),
 
755
                      osutils.pathjoin('new', new_path))
 
756
 
 
757
 
 
758
class DiffTree(object):
 
759
    """Provides textual representations of the difference between two trees.
 
760
 
 
761
    A DiffTree examines two trees and where a file-id has altered
 
762
    between them, generates a textual representation of the difference.
 
763
    DiffTree uses a sequence of DiffPath objects which are each
 
764
    given the opportunity to handle a given altered fileid. The list
 
765
    of DiffPath objects can be extended globally by appending to
 
766
    DiffTree.diff_factories, or for a specific diff operation by
 
767
    supplying the extra_factories option to the appropriate method.
414
768
    """
415
 
    mutter("check paths: %r", paths)
416
 
    if not paths:
417
 
        return
418
 
    s = old_tree.filter_unversioned_files(paths)
419
 
    s = new_tree.filter_unversioned_files(s)
420
 
    s = [path for path in s if not new_tree.has_filename(path)]
421
 
    if s:
422
 
        raise errors.PathsDoNotExist(sorted(s))
423
 
 
424
 
 
425
 
def get_prop_change(meta_modified):
426
 
    if meta_modified:
427
 
        return " (properties changed)"
428
 
    else:
429
 
        return  ""
430
 
 
431
 
 
432
 
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
433
 
                                new_path, new_tree, text_modified,
434
 
                                kind, to_file, diff_file):
435
 
    if text_modified:
436
 
        new_entry = new_tree.inventory[file_id]
437
 
        old_tree.inventory[file_id].diff(diff_file,
438
 
                                         old_path, old_tree,
439
 
                                         new_path, new_entry, 
440
 
                                         new_tree, to_file)
 
769
 
 
770
    # list of factories that can provide instances of DiffPath objects
 
771
    # may be extended by plugins.
 
772
    diff_factories = [DiffSymlink.from_diff_tree,
 
773
                      DiffDirectory.from_diff_tree]
 
774
 
 
775
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
776
                 diff_text=None, extra_factories=None):
 
777
        """Constructor
 
778
 
 
779
        :param old_tree: Tree to show as old in the comparison
 
780
        :param new_tree: Tree to show as new in the comparison
 
781
        :param to_file: File to write comparision to
 
782
        :param path_encoding: Character encoding to write paths in
 
783
        :param diff_text: DiffPath-type object to use as a last resort for
 
784
            diffing text files.
 
785
        :param extra_factories: Factories of DiffPaths to try before any other
 
786
            DiffPaths"""
 
787
        if diff_text is None:
 
788
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
789
                                 '', '',  internal_diff)
 
790
        self.old_tree = old_tree
 
791
        self.new_tree = new_tree
 
792
        self.to_file = to_file
 
793
        self.path_encoding = path_encoding
 
794
        self.differs = []
 
795
        if extra_factories is not None:
 
796
            self.differs.extend(f(self) for f in extra_factories)
 
797
        self.differs.extend(f(self) for f in self.diff_factories)
 
798
        self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
 
799
 
 
800
    @classmethod
 
801
    def from_trees_options(klass, old_tree, new_tree, to_file,
 
802
                           path_encoding, external_diff_options, old_label,
 
803
                           new_label, using):
 
804
        """Factory for producing a DiffTree.
 
805
 
 
806
        Designed to accept options used by show_diff_trees.
 
807
        :param old_tree: The tree to show as old in the comparison
 
808
        :param new_tree: The tree to show as new in the comparison
 
809
        :param to_file: File to write comparisons to
 
810
        :param path_encoding: Character encoding to use for writing paths
 
811
        :param external_diff_options: If supplied, use the installed diff
 
812
            binary to perform file comparison, using supplied options.
 
813
        :param old_label: Prefix to use for old file labels
 
814
        :param new_label: Prefix to use for new file labels
 
815
        :param using: Commandline to use to invoke an external diff tool
 
816
        """
 
817
        if using is not None:
 
818
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
 
819
        else:
 
820
            extra_factories = []
 
821
        if external_diff_options:
 
822
            opts = external_diff_options.split()
 
823
            def diff_file(olab, olines, nlab, nlines, to_file):
 
824
                external_diff(olab, olines, nlab, nlines, to_file, opts)
 
825
        else:
 
826
            diff_file = internal_diff
 
827
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
828
                             old_label, new_label, diff_file)
 
829
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
 
830
                     extra_factories)
 
831
 
 
832
    def show_diff(self, specific_files, extra_trees=None):
 
833
        """Write tree diff to self.to_file
 
834
 
 
835
        :param sepecific_files: the specific files to compare (recursive)
 
836
        :param extra_trees: extra trees to use for mapping paths to file_ids
 
837
        """
 
838
        try:
 
839
            return self._show_diff(specific_files, extra_trees)
 
840
        finally:
 
841
            for differ in self.differs:
 
842
                differ.finish()
 
843
 
 
844
    def _show_diff(self, specific_files, extra_trees):
 
845
        # TODO: Generation of pseudo-diffs for added/deleted files could
 
846
        # be usefully made into a much faster special case.
 
847
        iterator = self.new_tree.iter_changes(self.old_tree,
 
848
                                               specific_files=specific_files,
 
849
                                               extra_trees=extra_trees,
 
850
                                               require_versioned=True)
 
851
        has_changes = 0
 
852
        def changes_key(change):
 
853
            old_path, new_path = change[1]
 
854
            path = new_path
 
855
            if path is None:
 
856
                path = old_path
 
857
            return path
 
858
        def get_encoded_path(path):
 
859
            if path is not None:
 
860
                return path.encode(self.path_encoding, "replace")
 
861
        for (file_id, paths, changed_content, versioned, parent, name, kind,
 
862
             executable) in sorted(iterator, key=changes_key):
 
863
            # The root does not get diffed, and items with no known kind (that
 
864
            # is, missing) in both trees are skipped as well.
 
865
            if parent == (None, None) or kind == (None, None):
 
866
                continue
 
867
            oldpath, newpath = paths
 
868
            oldpath_encoded = get_encoded_path(paths[0])
 
869
            newpath_encoded = get_encoded_path(paths[1])
 
870
            old_present = (kind[0] is not None and versioned[0])
 
871
            new_present = (kind[1] is not None and versioned[1])
 
872
            renamed = (parent[0], name[0]) != (parent[1], name[1])
 
873
 
 
874
            properties_changed = []
 
875
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
 
876
 
 
877
            if properties_changed:
 
878
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
 
879
            else:
 
880
                prop_str = ""
 
881
 
 
882
            if (old_present, new_present) == (True, False):
 
883
                self.to_file.write("=== removed %s '%s'\n" %
 
884
                                   (kind[0], oldpath_encoded))
 
885
                newpath = oldpath
 
886
            elif (old_present, new_present) == (False, True):
 
887
                self.to_file.write("=== added %s '%s'\n" %
 
888
                                   (kind[1], newpath_encoded))
 
889
                oldpath = newpath
 
890
            elif renamed:
 
891
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
892
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
 
893
            else:
 
894
                # if it was produced by iter_changes, it must be
 
895
                # modified *somehow*, either content or execute bit.
 
896
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
897
                                   newpath_encoded, prop_str))
 
898
            if changed_content:
 
899
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
 
900
                has_changes = 1
 
901
            if renamed:
 
902
                has_changes = 1
 
903
        return has_changes
 
904
 
 
905
    def diff(self, file_id, old_path, new_path):
 
906
        """Perform a diff of a single file
 
907
 
 
908
        :param file_id: file-id of the file
 
909
        :param old_path: The path of the file in the old tree
 
910
        :param new_path: The path of the file in the new tree
 
911
        """
 
912
        try:
 
913
            old_kind = self.old_tree.kind(file_id)
 
914
        except (errors.NoSuchId, errors.NoSuchFile):
 
915
            old_kind = None
 
916
        try:
 
917
            new_kind = self.new_tree.kind(file_id)
 
918
        except (errors.NoSuchId, errors.NoSuchFile):
 
919
            new_kind = None
 
920
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
 
921
 
 
922
 
 
923
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
924
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
925
                                       new_path, old_kind, new_kind)
 
926
        if result is DiffPath.CANNOT_DIFF:
 
927
            error_path = new_path
 
928
            if error_path is None:
 
929
                error_path = old_path
 
930
            raise errors.NoDiffFound(error_path)