~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

[merge] bzr.dev 2255, resolve conflicts, update copyrights

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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
16
 
17
 
import difflib
18
17
import os
19
18
import re
20
 
import shutil
21
19
import sys
22
20
 
23
21
from bzrlib.lazy_import import lazy_import
28
26
import time
29
27
 
30
28
from bzrlib import (
31
 
    branch as _mod_branch,
32
 
    bzrdir,
33
 
    commands,
34
29
    errors,
35
30
    osutils,
36
31
    patiencediff,
37
32
    textfile,
38
 
    timestamp,
39
 
    views,
40
33
    )
41
34
""")
42
35
 
 
36
# compatability - plugins import compare_trees from diff!!!
 
37
# deprecated as of 0.10
 
38
from bzrlib.delta import compare_trees
43
39
from bzrlib.symbol_versioning import (
44
 
    deprecated_function,
45
 
    )
46
 
from bzrlib.trace import mutter, note, warning
 
40
        deprecated_function,
 
41
        zero_eight,
 
42
        )
 
43
from bzrlib.trace import mutter, warning
47
44
 
48
45
 
49
46
# TODO: Rather than building a changeset object, we should probably
50
47
# invoke callbacks on an object.  That object can either accumulate a
51
48
# list, write them out directly, etc etc.
52
49
 
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
 
 
63
50
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
64
51
                  allow_binary=False, sequence_matcher=None,
65
52
                  path_encoding='utf8'):
78
65
    # both sequences are empty.
79
66
    if not oldlines and not newlines:
80
67
        return
81
 
 
 
68
    
82
69
    if allow_binary is False:
83
70
        textfile.check_text_lines(oldlines)
84
71
        textfile.check_text_lines(newlines)
91
78
                      sequencematcher=sequence_matcher)
92
79
 
93
80
    ud = list(ud)
94
 
    if len(ud) == 0: # Identical contents, nothing to do
95
 
        return
96
81
    # work-around for difflib being too smart for its own good
97
82
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
98
83
    if not oldlines:
99
84
        ud[2] = ud[2].replace('-1,0', '-0,0')
100
85
    elif not newlines:
101
86
        ud[2] = ud[2].replace('+1,0', '+0,0')
 
87
    # work around for difflib emitting random spaces after the label
 
88
    ud[0] = ud[0][:-2] + '\n'
 
89
    ud[1] = ud[1][:-2] + '\n'
102
90
 
103
91
    for line in ud:
104
92
        to_file.write(line)
105
93
        if not line.endswith('\n'):
106
94
            to_file.write("\n\\ No newline at end of file\n")
107
 
    to_file.write('\n')
 
95
    print >>to_file
 
96
 
 
97
 
 
98
def _set_lang_C():
 
99
    """Set the env vars LANG=C and LC_ALL=C."""
 
100
    osutils.set_or_unset_env('LANG', 'C')
 
101
    osutils.set_or_unset_env('LC_ALL', 'C')
 
102
    osutils.set_or_unset_env('LC_CTYPE', None)
 
103
    osutils.set_or_unset_env('LANGUAGE', None)
108
104
 
109
105
 
110
106
def _spawn_external_diff(diffcmd, capture_errors=True):
117
113
    :return: A Popen object.
118
114
    """
119
115
    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'
 
116
        if sys.platform == 'win32':
 
117
            # Win32 doesn't support preexec_fn, but that is
 
118
            # okay, because it doesn't support LANG and LC_ALL either.
 
119
            preexec_fn = None
 
120
        else:
 
121
            preexec_fn = _set_lang_C
128
122
        stderr = subprocess.PIPE
129
123
    else:
130
 
        env = None
 
124
        preexec_fn = None
131
125
        stderr = None
132
126
 
133
127
    try:
135
129
                                stdin=subprocess.PIPE,
136
130
                                stdout=subprocess.PIPE,
137
131
                                stderr=stderr,
138
 
                                env=env)
 
132
                                preexec_fn=preexec_fn)
139
133
    except OSError, e:
140
134
        if e.errno == errno.ENOENT:
141
135
            raise errors.NoDiff(str(e))
171
165
 
172
166
        if not diff_opts:
173
167
            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')
179
168
        diffcmd = ['diff',
180
169
                   '--label', old_filename,
181
170
                   old_abspath,
204
193
            break
205
194
        else:
206
195
            diffcmd.append('-u')
207
 
 
 
196
                  
208
197
        if diff_opts:
209
198
            diffcmd.extend(diff_opts)
210
199
 
211
200
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
212
201
        out,err = pipe.communicate()
213
202
        rc = pipe.returncode
214
 
 
 
203
        
215
204
        # internal_diff() adds a trailing newline, add one here for consistency
216
205
        out += '\n'
217
206
        if rc == 2:
252
241
                msg = 'signal %d' % (-rc)
253
242
            else:
254
243
                msg = 'exit code %d' % rc
255
 
 
256
 
            raise errors.BzrError('external diff failed with %s; command: %r'
 
244
                
 
245
            raise errors.BzrError('external diff failed with %s; command: %r' 
257
246
                                  % (rc, diffcmd))
258
247
 
259
248
 
277
266
                        new_abspath, e)
278
267
 
279
268
 
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
 
269
@deprecated_function(zero_eight)
 
270
def show_diff(b, from_spec, specific_files, external_diff_options=None,
 
271
              revision2=None, output=None, b2=None):
 
272
    """Shortcut for showing the diff to the working tree.
 
273
 
 
274
    Please use show_diff_trees instead.
 
275
 
 
276
    b
 
277
        Branch.
 
278
 
 
279
    revision
 
280
        None for 'basis tree', or otherwise the old revision to compare against.
 
281
    
 
282
    The more general form is show_diff_trees(), where the caller
 
283
    supplies any two trees.
 
284
    """
 
285
    if output is None:
 
286
        output = sys.stdout
 
287
 
 
288
    if from_spec is None:
 
289
        old_tree = b.bzrdir.open_workingtree()
 
290
        if b2 is None:
 
291
            old_tree = old_tree = old_tree.basis_tree()
 
292
    else:
 
293
        old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
 
294
 
 
295
    if revision2 is None:
 
296
        if b2 is None:
 
297
            new_tree = b.bzrdir.open_workingtree()
 
298
        else:
 
299
            new_tree = b2.bzrdir.open_workingtree()
 
300
    else:
 
301
        new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
 
302
 
 
303
    return show_diff_trees(old_tree, new_tree, output, specific_files,
 
304
                           external_diff_options)
 
305
 
 
306
 
 
307
def diff_cmd_helper(tree, specific_files, external_diff_options, 
 
308
                    old_revision_spec=None, new_revision_spec=None,
 
309
                    revision_specs=None,
 
310
                    old_label='a/', new_label='b/'):
 
311
    """Helper for cmd_diff.
 
312
 
 
313
    :param tree:
 
314
        A WorkingTree
 
315
 
 
316
    :param specific_files:
 
317
        The specific files to compare, or None
 
318
 
 
319
    :param external_diff_options:
 
320
        If non-None, run an external diff, and pass it these options
 
321
 
 
322
    :param old_revision_spec:
 
323
        If None, use basis tree as old revision, otherwise use the tree for
 
324
        the specified revision. 
 
325
 
 
326
    :param new_revision_spec:
 
327
        If None, use working tree as new revision, otherwise use the tree for
 
328
        the specified revision.
 
329
    
 
330
    :param revision_specs: 
 
331
        Zero, one or two RevisionSpecs from the command line, saying what revisions 
 
332
        to compare.  This can be passed as an alternative to the old_revision_spec 
 
333
        and new_revision_spec parameters.
 
334
 
 
335
    The more general form is show_diff_trees(), where the caller
 
336
    supplies any two trees.
 
337
    """
 
338
 
 
339
    # TODO: perhaps remove the old parameters old_revision_spec and
 
340
    # new_revision_spec, since this is only really for use from cmd_diff and
 
341
    # it now always passes through a sequence of revision_specs -- mbp
 
342
    # 20061221
 
343
 
 
344
    def spec_tree(spec):
 
345
        if tree:
 
346
            revision = spec.in_store(tree.branch)
 
347
        else:
 
348
            revision = spec.in_store(None)
 
349
        revision_id = revision.rev_id
 
350
        branch = revision.branch
 
351
        return branch.repository.revision_tree(revision_id)
 
352
 
309
353
    if revision_specs is not None:
 
354
        assert (old_revision_spec is None
 
355
                and new_revision_spec is None)
310
356
        if len(revision_specs) > 0:
311
357
            old_revision_spec = revision_specs[0]
312
 
            if old_url is None:
313
 
                old_url = old_revision_spec.get_branch()
314
358
        if len(revision_specs) > 1:
315
359
            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)
 
360
 
 
361
    if old_revision_spec is None:
 
362
        old_tree = tree.basis_tree()
 
363
    else:
 
364
        old_tree = spec_tree(old_revision_spec)
 
365
 
 
366
    if (new_revision_spec is None
 
367
        or new_revision_spec.spec is None):
 
368
        new_tree = tree
 
369
    else:
 
370
        new_tree = spec_tree(new_revision_spec)
 
371
 
 
372
    if new_tree is not tree:
 
373
        extra_trees = (tree,)
 
374
    else:
 
375
        extra_trees = None
 
376
 
 
377
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
 
378
                           external_diff_options,
 
379
                           old_label=old_label, new_label=new_label,
 
380
                           extra_trees=extra_trees)
395
381
 
396
382
 
397
383
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
398
384
                    external_diff_options=None,
399
385
                    old_label='a/', new_label='b/',
400
 
                    extra_trees=None,
401
 
                    path_encoding='utf8',
402
 
                    using=None):
 
386
                    extra_trees=None):
403
387
    """Show in text form the changes from one tree to another.
404
388
 
405
 
    to_file
406
 
        The output stream.
407
 
 
408
 
    specific_files
409
 
        Include only changes to these files - None for all changes.
 
389
    to_files
 
390
        If set, include only changes to these files.
410
391
 
411
392
    external_diff_options
412
393
        If set, use an external GNU diff and pass these options.
413
394
 
414
395
    extra_trees
415
396
        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
420
397
    """
421
398
    old_tree.lock_read()
422
399
    try:
423
 
        if extra_trees is not None:
424
 
            for tree in extra_trees:
425
 
                tree.lock_read()
426
400
        new_tree.lock_read()
427
401
        try:
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)
 
402
            return _show_diff_trees(old_tree, new_tree, to_file,
 
403
                                    specific_files, external_diff_options,
 
404
                                    old_label=old_label, new_label=new_label,
 
405
                                    extra_trees=extra_trees)
433
406
        finally:
434
407
            new_tree.unlock()
435
 
            if extra_trees is not None:
436
 
                for tree in extra_trees:
437
 
                    tree.unlock()
438
408
    finally:
439
409
        old_tree.unlock()
440
410
 
441
411
 
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):
 
412
def _show_diff_trees(old_tree, new_tree, to_file,
 
413
                     specific_files, external_diff_options, 
 
414
                     old_label='a/', new_label='b/', extra_trees=None):
587
415
 
588
416
    # GNU Patch uses the epoch date to detect files that are being added
589
417
    # or removed in a diff.
590
418
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
591
419
 
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.
 
420
    # TODO: Generation of pseudo-diffs for added/deleted files could
 
421
    # be usefully made into a much faster special case.
 
422
 
 
423
    if external_diff_options:
 
424
        assert isinstance(external_diff_options, basestring)
 
425
        opts = external_diff_options.split()
 
426
        def diff_file(olab, olines, nlab, nlines, to_file):
 
427
            external_diff(olab, olines, nlab, nlines, to_file, opts)
 
428
    else:
 
429
        diff_file = internal_diff
 
430
    
 
431
    delta = new_tree.changes_from(old_tree,
 
432
        specific_files=specific_files,
 
433
        extra_trees=extra_trees, require_versioned=True)
 
434
 
 
435
    has_changes = 0
 
436
    for path, file_id, kind in delta.removed:
 
437
        has_changes = 1
 
438
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
439
        old_name = '%s%s\t%s' % (old_label, path,
 
440
                                 _patch_header_date(old_tree, file_id, path))
 
441
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
442
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
443
                                         new_name, None, None, to_file)
 
444
    for path, file_id, kind in delta.added:
 
445
        has_changes = 1
 
446
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
447
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
448
        new_name = '%s%s\t%s' % (new_label, path,
 
449
                                 _patch_header_date(new_tree, file_id, path))
 
450
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
451
                                         old_name, None, None, to_file, 
 
452
                                         reverse=True)
 
453
    for (old_path, new_path, file_id, kind,
 
454
         text_modified, meta_modified) in delta.renamed:
 
455
        has_changes = 1
 
456
        prop_str = get_prop_change(meta_modified)
 
457
        print >>to_file, '=== renamed %s %r => %r%s' % (
 
458
                    kind, old_path.encode('utf8'),
 
459
                    new_path.encode('utf8'), prop_str)
 
460
        old_name = '%s%s\t%s' % (old_label, old_path,
 
461
                                 _patch_header_date(old_tree, file_id,
 
462
                                                    old_path))
 
463
        new_name = '%s%s\t%s' % (new_label, new_path,
 
464
                                 _patch_header_date(new_tree, file_id,
 
465
                                                    new_path))
 
466
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
467
                                    new_name, new_tree,
 
468
                                    text_modified, kind, to_file, diff_file)
 
469
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
 
470
        has_changes = 1
 
471
        prop_str = get_prop_change(meta_modified)
 
472
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
473
        old_name = '%s%s\t%s' % (old_label, path,
 
474
                                 _patch_header_date(old_tree, file_id, path))
 
475
        new_name = '%s%s\t%s' % (new_label, path,
 
476
                                 _patch_header_date(new_tree, file_id, path))
 
477
        if text_modified:
 
478
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
479
                                        new_name, new_tree,
 
480
                                        True, kind, to_file, diff_file)
 
481
 
 
482
    return has_changes
 
483
 
 
484
 
 
485
def _patch_header_date(tree, file_id, path):
 
486
    """Returns a timestamp suitable for use in a patch header."""
 
487
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
 
488
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
 
489
 
 
490
 
 
491
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
492
    """Complain if paths are not in either inventory or tree.
 
493
 
 
494
    It's OK with the files exist in either tree's inventory, or 
 
495
    if they exist in the tree but are not versioned.
 
496
    
 
497
    This can be used by operations such as bzr status that can accept
 
498
    unknown or ignored files.
768
499
    """
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)
 
500
    mutter("check paths: %r", paths)
 
501
    if not paths:
 
502
        return
 
503
    s = old_tree.filter_unversioned_files(paths)
 
504
    s = new_tree.filter_unversioned_files(s)
 
505
    s = [path for path in s if not new_tree.has_filename(path)]
 
506
    if s:
 
507
        raise errors.PathsDoNotExist(sorted(s))
 
508
 
 
509
 
 
510
def get_prop_change(meta_modified):
 
511
    if meta_modified:
 
512
        return " (properties changed)"
 
513
    else:
 
514
        return  ""
 
515
 
 
516
 
 
517
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
518
                                new_path, new_tree, text_modified,
 
519
                                kind, to_file, diff_file):
 
520
    if text_modified:
 
521
        new_entry = new_tree.inventory[file_id]
 
522
        old_tree.inventory[file_id].diff(diff_file,
 
523
                                         old_path, old_tree,
 
524
                                         new_path, new_entry, 
 
525
                                         new_tree, to_file)