/brz/remove-bazaar

To get this branch, use:
bzr branch http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Robert Collins
  • Date: 2010-05-06 11:08:10 UTC
  • mto: This revision was merged to the branch mainline in revision 5223.
  • Revision ID: robertc@robertcollins.net-20100506110810-h3j07fh5gmw54s25
Cleaner matcher matching revised unlocking protocol.

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
 
1
# Copyright (C) 2005-2010 Canonical Ltd.
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
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 difflib
17
18
import os
18
19
import re
 
20
import string
19
21
import sys
20
22
 
21
23
from bzrlib.lazy_import import lazy_import
26
28
import time
27
29
 
28
30
from bzrlib import (
 
31
    branch as _mod_branch,
 
32
    bzrdir,
 
33
    cmdline,
 
34
    cleanup,
29
35
    errors,
30
36
    osutils,
31
37
    patiencediff,
32
38
    textfile,
33
39
    timestamp,
 
40
    views,
34
41
    )
 
42
 
 
43
from bzrlib.workingtree import WorkingTree
35
44
""")
36
45
 
37
 
# compatability - plugins import compare_trees from diff!!!
38
 
# deprecated as of 0.10
39
 
from bzrlib.delta import compare_trees
 
46
from bzrlib.registry import (
 
47
    Registry,
 
48
    )
40
49
from bzrlib.symbol_versioning import (
41
 
        deprecated_function,
42
 
        zero_eight,
43
 
        )
44
 
from bzrlib.trace import mutter, warning
 
50
    deprecated_function,
 
51
    deprecated_in,
 
52
    )
 
53
from bzrlib.trace import mutter, note, warning
 
54
 
 
55
 
 
56
class AtTemplate(string.Template):
 
57
    """Templating class that uses @ instead of $."""
 
58
 
 
59
    delimiter = '@'
45
60
 
46
61
 
47
62
# TODO: Rather than building a changeset object, we should probably
48
63
# invoke callbacks on an object.  That object can either accumulate a
49
64
# list, write them out directly, etc etc.
50
65
 
 
66
 
 
67
class _PrematchedMatcher(difflib.SequenceMatcher):
 
68
    """Allow SequenceMatcher operations to use predetermined blocks"""
 
69
 
 
70
    def __init__(self, matching_blocks):
 
71
        difflib.SequenceMatcher(self, None, None)
 
72
        self.matching_blocks = matching_blocks
 
73
        self.opcodes = None
 
74
 
 
75
 
51
76
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
52
77
                  allow_binary=False, sequence_matcher=None,
53
78
                  path_encoding='utf8'):
66
91
    # both sequences are empty.
67
92
    if not oldlines and not newlines:
68
93
        return
69
 
    
 
94
 
70
95
    if allow_binary is False:
71
96
        textfile.check_text_lines(oldlines)
72
97
        textfile.check_text_lines(newlines)
79
104
                      sequencematcher=sequence_matcher)
80
105
 
81
106
    ud = list(ud)
 
107
    if len(ud) == 0: # Identical contents, nothing to do
 
108
        return
82
109
    # work-around for difflib being too smart for its own good
83
110
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
84
111
    if not oldlines:
85
112
        ud[2] = ud[2].replace('-1,0', '-0,0')
86
113
    elif not newlines:
87
114
        ud[2] = ud[2].replace('+1,0', '+0,0')
88
 
    # work around for difflib emitting random spaces after the label
89
 
    ud[0] = ud[0][:-2] + '\n'
90
 
    ud[1] = ud[1][:-2] + '\n'
91
115
 
92
116
    for line in ud:
93
117
        to_file.write(line)
94
118
        if not line.endswith('\n'):
95
119
            to_file.write("\n\\ No newline at end of file\n")
96
 
    print >>to_file
 
120
    to_file.write('\n')
97
121
 
98
122
 
99
123
def _spawn_external_diff(diffcmd, capture_errors=True):
160
184
 
161
185
        if not diff_opts:
162
186
            diff_opts = []
 
187
        if sys.platform == 'win32':
 
188
            # Popen doesn't do the proper encoding for external commands
 
189
            # Since we are dealing with an ANSI api, use mbcs encoding
 
190
            old_filename = old_filename.encode('mbcs')
 
191
            new_filename = new_filename.encode('mbcs')
163
192
        diffcmd = ['diff',
164
193
                   '--label', old_filename,
165
194
                   old_abspath,
188
217
            break
189
218
        else:
190
219
            diffcmd.append('-u')
191
 
                  
 
220
 
192
221
        if diff_opts:
193
222
            diffcmd.extend(diff_opts)
194
223
 
195
224
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
196
225
        out,err = pipe.communicate()
197
226
        rc = pipe.returncode
198
 
        
 
227
 
199
228
        # internal_diff() adds a trailing newline, add one here for consistency
200
229
        out += '\n'
201
230
        if rc == 2:
236
265
                msg = 'signal %d' % (-rc)
237
266
            else:
238
267
                msg = 'exit code %d' % rc
239
 
                
240
 
            raise errors.BzrError('external diff failed with %s; command: %r' 
 
268
 
 
269
            raise errors.BzrError('external diff failed with %s; command: %r'
241
270
                                  % (rc, diffcmd))
242
271
 
243
272
 
261
290
                        new_abspath, e)
262
291
 
263
292
 
264
 
@deprecated_function(zero_eight)
265
 
def show_diff(b, from_spec, specific_files, external_diff_options=None,
266
 
              revision2=None, output=None, b2=None):
267
 
    """Shortcut for showing the diff to the working tree.
268
 
 
269
 
    Please use show_diff_trees instead.
270
 
 
271
 
    b
272
 
        Branch.
273
 
 
274
 
    revision
275
 
        None for 'basis tree', or otherwise the old revision to compare against.
276
 
    
277
 
    The more general form is show_diff_trees(), where the caller
278
 
    supplies any two trees.
279
 
    """
280
 
    if output is None:
281
 
        output = sys.stdout
282
 
 
283
 
    if from_spec is None:
284
 
        old_tree = b.bzrdir.open_workingtree()
285
 
        if b2 is None:
286
 
            old_tree = old_tree = old_tree.basis_tree()
287
 
    else:
288
 
        old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
289
 
 
290
 
    if revision2 is None:
291
 
        if b2 is None:
292
 
            new_tree = b.bzrdir.open_workingtree()
293
 
        else:
294
 
            new_tree = b2.bzrdir.open_workingtree()
295
 
    else:
296
 
        new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
297
 
 
298
 
    return show_diff_trees(old_tree, new_tree, output, specific_files,
299
 
                           external_diff_options)
300
 
 
301
 
 
302
 
def diff_cmd_helper(tree, specific_files, external_diff_options, 
303
 
                    old_revision_spec=None, new_revision_spec=None,
304
 
                    revision_specs=None,
305
 
                    old_label='a/', new_label='b/'):
306
 
    """Helper for cmd_diff.
307
 
 
308
 
    :param tree:
309
 
        A WorkingTree
310
 
 
311
 
    :param specific_files:
312
 
        The specific files to compare, or None
313
 
 
314
 
    :param external_diff_options:
315
 
        If non-None, run an external diff, and pass it these options
316
 
 
317
 
    :param old_revision_spec:
318
 
        If None, use basis tree as old revision, otherwise use the tree for
319
 
        the specified revision. 
320
 
 
321
 
    :param new_revision_spec:
322
 
        If None, use working tree as new revision, otherwise use the tree for
323
 
        the specified revision.
324
 
    
325
 
    :param revision_specs: 
326
 
        Zero, one or two RevisionSpecs from the command line, saying what revisions 
327
 
        to compare.  This can be passed as an alternative to the old_revision_spec 
328
 
        and new_revision_spec parameters.
329
 
 
330
 
    The more general form is show_diff_trees(), where the caller
331
 
    supplies any two trees.
332
 
    """
333
 
 
334
 
    # TODO: perhaps remove the old parameters old_revision_spec and
335
 
    # new_revision_spec, since this is only really for use from cmd_diff and
336
 
    # it now always passes through a sequence of revision_specs -- mbp
337
 
    # 20061221
338
 
 
339
 
    def spec_tree(spec):
340
 
        if tree:
341
 
            revision = spec.in_store(tree.branch)
342
 
        else:
343
 
            revision = spec.in_store(None)
344
 
        revision_id = revision.rev_id
345
 
        branch = revision.branch
346
 
        return branch.repository.revision_tree(revision_id)
347
 
 
 
293
@deprecated_function(deprecated_in((2, 2, 0)))
 
294
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
 
295
                                   apply_view=True):
 
296
    """Get the trees and specific files to diff given a list of paths.
 
297
 
 
298
    This method works out the trees to be diff'ed and the files of
 
299
    interest within those trees.
 
300
 
 
301
    :param path_list:
 
302
        the list of arguments passed to the diff command
 
303
    :param revision_specs:
 
304
        Zero, one or two RevisionSpecs from the diff command line,
 
305
        saying what revisions to compare.
 
306
    :param old_url:
 
307
        The url of the old branch or tree. If None, the tree to use is
 
308
        taken from the first path, if any, or the current working tree.
 
309
    :param new_url:
 
310
        The url of the new branch or tree. If None, the tree to use is
 
311
        taken from the first path, if any, or the current working tree.
 
312
    :param apply_view:
 
313
        if True and a view is set, apply the view or check that the paths
 
314
        are within it
 
315
    :returns:
 
316
        a tuple of (old_tree, new_tree, old_branch, new_branch,
 
317
        specific_files, extra_trees) where extra_trees is a sequence of
 
318
        additional trees to search in for file-ids.  The trees and branches
 
319
        are not locked.
 
320
    """
 
321
    op = cleanup.OperationWithCleanups(get_trees_and_branches_to_diff_locked)
 
322
    return op.run_simple(path_list, revision_specs, old_url, new_url,
 
323
            op.add_cleanup, apply_view=apply_view)
 
324
    
 
325
 
 
326
def get_trees_and_branches_to_diff_locked(
 
327
    path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
 
328
    """Get the trees and specific files to diff given a list of paths.
 
329
 
 
330
    This method works out the trees to be diff'ed and the files of
 
331
    interest within those trees.
 
332
 
 
333
    :param path_list:
 
334
        the list of arguments passed to the diff command
 
335
    :param revision_specs:
 
336
        Zero, one or two RevisionSpecs from the diff command line,
 
337
        saying what revisions to compare.
 
338
    :param old_url:
 
339
        The url of the old branch or tree. If None, the tree to use is
 
340
        taken from the first path, if any, or the current working tree.
 
341
    :param new_url:
 
342
        The url of the new branch or tree. If None, the tree to use is
 
343
        taken from the first path, if any, or the current working tree.
 
344
    :param add_cleanup:
 
345
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
 
346
        will register cleanups that must be run to unlock the trees, etc.
 
347
    :param apply_view:
 
348
        if True and a view is set, apply the view or check that the paths
 
349
        are within it
 
350
    :returns:
 
351
        a tuple of (old_tree, new_tree, old_branch, new_branch,
 
352
        specific_files, extra_trees) where extra_trees is a sequence of
 
353
        additional trees to search in for file-ids.  The trees and branches
 
354
        will be read-locked until the cleanups registered via the add_cleanup
 
355
        param are run.
 
356
    """
 
357
    # Get the old and new revision specs
 
358
    old_revision_spec = None
 
359
    new_revision_spec = None
348
360
    if revision_specs is not None:
349
 
        assert (old_revision_spec is None
350
 
                and new_revision_spec is None)
351
361
        if len(revision_specs) > 0:
352
362
            old_revision_spec = revision_specs[0]
 
363
            if old_url is None:
 
364
                old_url = old_revision_spec.get_branch()
353
365
        if len(revision_specs) > 1:
354
366
            new_revision_spec = revision_specs[1]
355
 
 
356
 
    if old_revision_spec is None:
357
 
        old_tree = tree.basis_tree()
358
 
    else:
359
 
        old_tree = spec_tree(old_revision_spec)
360
 
 
361
 
    if (new_revision_spec is None
362
 
        or new_revision_spec.spec is None):
363
 
        new_tree = tree
364
 
    else:
365
 
        new_tree = spec_tree(new_revision_spec)
366
 
 
367
 
    if new_tree is not tree:
368
 
        extra_trees = (tree,)
369
 
    else:
370
 
        extra_trees = None
371
 
 
372
 
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
373
 
                           external_diff_options,
374
 
                           old_label=old_label, new_label=new_label,
375
 
                           extra_trees=extra_trees)
 
367
            if new_url is None:
 
368
                new_url = new_revision_spec.get_branch()
 
369
 
 
370
    other_paths = []
 
371
    make_paths_wt_relative = True
 
372
    consider_relpath = True
 
373
    if path_list is None or len(path_list) == 0:
 
374
        # If no path is given, the current working tree is used
 
375
        default_location = u'.'
 
376
        consider_relpath = False
 
377
    elif old_url is not None and new_url is not None:
 
378
        other_paths = path_list
 
379
        make_paths_wt_relative = False
 
380
    else:
 
381
        default_location = path_list[0]
 
382
        other_paths = path_list[1:]
 
383
 
 
384
    def lock_tree_or_branch(wt, br):
 
385
        if wt is not None:
 
386
            wt.lock_read()
 
387
            add_cleanup(wt.unlock)
 
388
        elif br is not None:
 
389
            br.lock_read()
 
390
            add_cleanup(br.unlock)
 
391
 
 
392
    # Get the old location
 
393
    specific_files = []
 
394
    if old_url is None:
 
395
        old_url = default_location
 
396
    working_tree, branch, relpath = \
 
397
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
 
398
    lock_tree_or_branch(working_tree, branch)
 
399
    if consider_relpath and relpath != '':
 
400
        if working_tree is not None and apply_view:
 
401
            views.check_path_in_view(working_tree, relpath)
 
402
        specific_files.append(relpath)
 
403
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
 
404
    old_branch = branch
 
405
 
 
406
    # Get the new location
 
407
    if new_url is None:
 
408
        new_url = default_location
 
409
    if new_url != old_url:
 
410
        working_tree, branch, relpath = \
 
411
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
 
412
        lock_tree_or_branch(working_tree, branch)
 
413
        if consider_relpath and relpath != '':
 
414
            if working_tree is not None and apply_view:
 
415
                views.check_path_in_view(working_tree, relpath)
 
416
            specific_files.append(relpath)
 
417
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
 
418
        basis_is_default=working_tree is None)
 
419
    new_branch = branch
 
420
 
 
421
    # Get the specific files (all files is None, no files is [])
 
422
    if make_paths_wt_relative and working_tree is not None:
 
423
        try:
 
424
            from bzrlib.builtins import safe_relpath_files
 
425
            other_paths = safe_relpath_files(working_tree, other_paths,
 
426
            apply_view=apply_view)
 
427
        except errors.FileInWrongBranch:
 
428
            raise errors.BzrCommandError("Files are in different branches")
 
429
    specific_files.extend(other_paths)
 
430
    if len(specific_files) == 0:
 
431
        specific_files = None
 
432
        if (working_tree is not None and working_tree.supports_views()
 
433
            and apply_view):
 
434
            view_files = working_tree.views.lookup_view()
 
435
            if view_files:
 
436
                specific_files = view_files
 
437
                view_str = views.view_display_str(view_files)
 
438
                note("*** Ignoring files outside view. View is %s" % view_str)
 
439
 
 
440
    # Get extra trees that ought to be searched for file-ids
 
441
    extra_trees = None
 
442
    if working_tree is not None and working_tree not in (old_tree, new_tree):
 
443
        extra_trees = (working_tree,)
 
444
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
 
445
 
 
446
 
 
447
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
 
448
    if branch is None and tree is not None:
 
449
        branch = tree.branch
 
450
    if spec is None or spec.spec is None:
 
451
        if basis_is_default:
 
452
            if tree is not None:
 
453
                return tree.basis_tree()
 
454
            else:
 
455
                return branch.basis_tree()
 
456
        else:
 
457
            return tree
 
458
    return spec.as_tree(branch)
376
459
 
377
460
 
378
461
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
379
462
                    external_diff_options=None,
380
463
                    old_label='a/', new_label='b/',
381
 
                    extra_trees=None):
 
464
                    extra_trees=None,
 
465
                    path_encoding='utf8',
 
466
                    using=None,
 
467
                    format_cls=None):
382
468
    """Show in text form the changes from one tree to another.
383
469
 
384
 
    to_files
385
 
        If set, include only changes to these files.
386
 
 
387
 
    external_diff_options
388
 
        If set, use an external GNU diff and pass these options.
389
 
 
390
 
    extra_trees
391
 
        If set, more Trees to use for looking up file ids
 
470
    :param to_file: The output stream.
 
471
    :param specific_files:Include only changes to these files - None for all
 
472
        changes.
 
473
    :param external_diff_options: If set, use an external GNU diff and pass 
 
474
        these options.
 
475
    :param extra_trees: If set, more Trees to use for looking up file ids
 
476
    :param path_encoding: If set, the path will be encoded as specified, 
 
477
        otherwise is supposed to be utf8
 
478
    :param format_cls: Formatter class (DiffTree subclass)
392
479
    """
 
480
    if format_cls is None:
 
481
        format_cls = DiffTree
393
482
    old_tree.lock_read()
394
483
    try:
395
484
        if extra_trees is not None:
397
486
                tree.lock_read()
398
487
        new_tree.lock_read()
399
488
        try:
400
 
            return _show_diff_trees(old_tree, new_tree, to_file,
401
 
                                    specific_files, external_diff_options,
402
 
                                    old_label=old_label, new_label=new_label,
403
 
                                    extra_trees=extra_trees)
 
489
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
 
490
                                                   path_encoding,
 
491
                                                   external_diff_options,
 
492
                                                   old_label, new_label, using)
 
493
            return differ.show_diff(specific_files, extra_trees)
404
494
        finally:
405
495
            new_tree.unlock()
406
496
            if extra_trees is not None:
410
500
        old_tree.unlock()
411
501
 
412
502
 
413
 
def _show_diff_trees(old_tree, new_tree, to_file,
414
 
                     specific_files, external_diff_options, 
415
 
                     old_label='a/', new_label='b/', extra_trees=None):
 
503
def _patch_header_date(tree, file_id, path):
 
504
    """Returns a timestamp suitable for use in a patch header."""
 
505
    try:
 
506
        mtime = tree.get_file_mtime(file_id, path)
 
507
    except errors.FileTimestampUnavailable:
 
508
        mtime = 0
 
509
    return timestamp.format_patch_date(mtime)
 
510
 
 
511
 
 
512
def get_executable_change(old_is_x, new_is_x):
 
513
    descr = { True:"+x", False:"-x", None:"??" }
 
514
    if old_is_x != new_is_x:
 
515
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
516
    else:
 
517
        return []
 
518
 
 
519
 
 
520
class DiffPath(object):
 
521
    """Base type for command object that compare files"""
 
522
 
 
523
    # The type or contents of the file were unsuitable for diffing
 
524
    CANNOT_DIFF = 'CANNOT_DIFF'
 
525
    # The file has changed in a semantic way
 
526
    CHANGED = 'CHANGED'
 
527
    # The file content may have changed, but there is no semantic change
 
528
    UNCHANGED = 'UNCHANGED'
 
529
 
 
530
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
 
531
        """Constructor.
 
532
 
 
533
        :param old_tree: The tree to show as the old tree in the comparison
 
534
        :param new_tree: The tree to show as new in the comparison
 
535
        :param to_file: The file to write comparison data to
 
536
        :param path_encoding: The character encoding to write paths in
 
537
        """
 
538
        self.old_tree = old_tree
 
539
        self.new_tree = new_tree
 
540
        self.to_file = to_file
 
541
        self.path_encoding = path_encoding
 
542
 
 
543
    def finish(self):
 
544
        pass
 
545
 
 
546
    @classmethod
 
547
    def from_diff_tree(klass, diff_tree):
 
548
        return klass(diff_tree.old_tree, diff_tree.new_tree,
 
549
                     diff_tree.to_file, diff_tree.path_encoding)
 
550
 
 
551
    @staticmethod
 
552
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
 
553
        for file_differ in differs:
 
554
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
555
                                      new_kind)
 
556
            if result is not DiffPath.CANNOT_DIFF:
 
557
                return result
 
558
        else:
 
559
            return DiffPath.CANNOT_DIFF
 
560
 
 
561
 
 
562
class DiffKindChange(object):
 
563
    """Special differ for file kind changes.
 
564
 
 
565
    Represents kind change as deletion + creation.  Uses the other differs
 
566
    to do this.
 
567
    """
 
568
    def __init__(self, differs):
 
569
        self.differs = differs
 
570
 
 
571
    def finish(self):
 
572
        pass
 
573
 
 
574
    @classmethod
 
575
    def from_diff_tree(klass, diff_tree):
 
576
        return klass(diff_tree.differs)
 
577
 
 
578
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
579
        """Perform comparison
 
580
 
 
581
        :param file_id: The file_id of the file to compare
 
582
        :param old_path: Path of the file in the old tree
 
583
        :param new_path: Path of the file in the new tree
 
584
        :param old_kind: Old file-kind of the file
 
585
        :param new_kind: New file-kind of the file
 
586
        """
 
587
        if None in (old_kind, new_kind):
 
588
            return DiffPath.CANNOT_DIFF
 
589
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
590
                                       new_path, old_kind, None)
 
591
        if result is DiffPath.CANNOT_DIFF:
 
592
            return result
 
593
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
594
                                     None, new_kind)
 
595
 
 
596
 
 
597
class DiffDirectory(DiffPath):
 
598
 
 
599
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
600
        """Perform comparison between two directories.  (dummy)
 
601
 
 
602
        """
 
603
        if 'directory' not in (old_kind, new_kind):
 
604
            return self.CANNOT_DIFF
 
605
        if old_kind not in ('directory', None):
 
606
            return self.CANNOT_DIFF
 
607
        if new_kind not in ('directory', None):
 
608
            return self.CANNOT_DIFF
 
609
        return self.CHANGED
 
610
 
 
611
 
 
612
class DiffSymlink(DiffPath):
 
613
 
 
614
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
615
        """Perform comparison between two symlinks
 
616
 
 
617
        :param file_id: The file_id of the file to compare
 
618
        :param old_path: Path of the file in the old tree
 
619
        :param new_path: Path of the file in the new tree
 
620
        :param old_kind: Old file-kind of the file
 
621
        :param new_kind: New file-kind of the file
 
622
        """
 
623
        if 'symlink' not in (old_kind, new_kind):
 
624
            return self.CANNOT_DIFF
 
625
        if old_kind == 'symlink':
 
626
            old_target = self.old_tree.get_symlink_target(file_id)
 
627
        elif old_kind is None:
 
628
            old_target = None
 
629
        else:
 
630
            return self.CANNOT_DIFF
 
631
        if new_kind == 'symlink':
 
632
            new_target = self.new_tree.get_symlink_target(file_id)
 
633
        elif new_kind is None:
 
634
            new_target = None
 
635
        else:
 
636
            return self.CANNOT_DIFF
 
637
        return self.diff_symlink(old_target, new_target)
 
638
 
 
639
    def diff_symlink(self, old_target, new_target):
 
640
        if old_target is None:
 
641
            self.to_file.write('=== target is %r\n' % new_target)
 
642
        elif new_target is None:
 
643
            self.to_file.write('=== target was %r\n' % old_target)
 
644
        else:
 
645
            self.to_file.write('=== target changed %r => %r\n' %
 
646
                              (old_target, new_target))
 
647
        return self.CHANGED
 
648
 
 
649
 
 
650
class DiffText(DiffPath):
416
651
 
417
652
    # GNU Patch uses the epoch date to detect files that are being added
418
653
    # or removed in a diff.
419
654
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
420
655
 
421
 
    # TODO: Generation of pseudo-diffs for added/deleted files could
422
 
    # be usefully made into a much faster special case.
423
 
 
424
 
    if external_diff_options:
425
 
        assert isinstance(external_diff_options, basestring)
426
 
        opts = external_diff_options.split()
427
 
        def diff_file(olab, olines, nlab, nlines, to_file):
428
 
            external_diff(olab, olines, nlab, nlines, to_file, opts)
429
 
    else:
430
 
        diff_file = internal_diff
431
 
    
432
 
    delta = new_tree.changes_from(old_tree,
433
 
        specific_files=specific_files,
434
 
        extra_trees=extra_trees, require_versioned=True)
435
 
 
436
 
    has_changes = 0
437
 
    for path, file_id, kind in delta.removed:
438
 
        has_changes = 1
439
 
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
440
 
        old_name = '%s%s\t%s' % (old_label, path,
441
 
                                 _patch_header_date(old_tree, file_id, path))
442
 
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
443
 
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
444
 
                                         new_name, None, None, to_file)
445
 
    for path, file_id, kind in delta.added:
446
 
        has_changes = 1
447
 
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
448
 
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
449
 
        new_name = '%s%s\t%s' % (new_label, path,
450
 
                                 _patch_header_date(new_tree, file_id, path))
451
 
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
452
 
                                         old_name, None, None, to_file, 
453
 
                                         reverse=True)
454
 
    for (old_path, new_path, file_id, kind,
455
 
         text_modified, meta_modified) in delta.renamed:
456
 
        has_changes = 1
457
 
        prop_str = get_prop_change(meta_modified)
458
 
        print >>to_file, '=== renamed %s %r => %r%s' % (
459
 
                    kind, old_path.encode('utf8'),
460
 
                    new_path.encode('utf8'), prop_str)
461
 
        old_name = '%s%s\t%s' % (old_label, old_path,
462
 
                                 _patch_header_date(old_tree, file_id,
463
 
                                                    old_path))
464
 
        new_name = '%s%s\t%s' % (new_label, new_path,
465
 
                                 _patch_header_date(new_tree, file_id,
466
 
                                                    new_path))
467
 
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
468
 
                                    new_name, new_tree,
469
 
                                    text_modified, kind, to_file, diff_file)
470
 
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
471
 
        has_changes = 1
472
 
        prop_str = get_prop_change(meta_modified)
473
 
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
474
 
        old_name = '%s%s\t%s' % (old_label, path,
475
 
                                 _patch_header_date(old_tree, file_id, path))
476
 
        new_name = '%s%s\t%s' % (new_label, path,
477
 
                                 _patch_header_date(new_tree, file_id, path))
478
 
        if text_modified:
479
 
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
480
 
                                        new_name, new_tree,
481
 
                                        True, kind, to_file, diff_file)
482
 
 
483
 
    return has_changes
484
 
 
485
 
 
486
 
def _patch_header_date(tree, file_id, path):
487
 
    """Returns a timestamp suitable for use in a patch header."""
488
 
    return timestamp.format_patch_date(tree.get_file_mtime(file_id, path))
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.
 
656
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
657
                 old_label='', new_label='', text_differ=internal_diff):
 
658
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
659
        self.text_differ = text_differ
 
660
        self.old_label = old_label
 
661
        self.new_label = new_label
 
662
        self.path_encoding = path_encoding
 
663
 
 
664
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
665
        """Compare two files in unified diff format
 
666
 
 
667
        :param file_id: The file_id of the file to compare
 
668
        :param old_path: Path of the file in the old tree
 
669
        :param new_path: Path of the file in the new tree
 
670
        :param old_kind: Old file-kind of the file
 
671
        :param new_kind: New file-kind of the file
 
672
        """
 
673
        if 'file' not in (old_kind, new_kind):
 
674
            return self.CANNOT_DIFF
 
675
        from_file_id = to_file_id = file_id
 
676
        if old_kind == 'file':
 
677
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
 
678
        elif old_kind is None:
 
679
            old_date = self.EPOCH_DATE
 
680
            from_file_id = None
 
681
        else:
 
682
            return self.CANNOT_DIFF
 
683
        if new_kind == 'file':
 
684
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
 
685
        elif new_kind is None:
 
686
            new_date = self.EPOCH_DATE
 
687
            to_file_id = None
 
688
        else:
 
689
            return self.CANNOT_DIFF
 
690
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
691
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
692
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
 
693
            old_path, new_path)
 
694
 
 
695
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
 
696
        from_path=None, to_path=None):
 
697
        """Diff the content of given files in two trees
 
698
 
 
699
        :param from_file_id: The id of the file in the from tree.  If None,
 
700
            the file is not present in the from tree.
 
701
        :param to_file_id: The id of the file in the to tree.  This may refer
 
702
            to a different file from from_file_id.  If None,
 
703
            the file is not present in the to tree.
 
704
        :param from_path: The path in the from tree or None if unknown.
 
705
        :param to_path: The path in the to tree or None if unknown.
 
706
        """
 
707
        def _get_text(tree, file_id, path):
 
708
            if file_id is not None:
 
709
                return tree.get_file(file_id, path).readlines()
 
710
            else:
 
711
                return []
 
712
        try:
 
713
            from_text = _get_text(self.old_tree, from_file_id, from_path)
 
714
            to_text = _get_text(self.new_tree, to_file_id, to_path)
 
715
            self.text_differ(from_label, from_text, to_label, to_text,
 
716
                             self.to_file)
 
717
        except errors.BinaryFile:
 
718
            self.to_file.write(
 
719
                  ("Binary files %s and %s differ\n" %
 
720
                  (from_label, to_label)).encode(self.path_encoding))
 
721
        return self.CHANGED
 
722
 
 
723
 
 
724
class DiffFromTool(DiffPath):
 
725
 
 
726
    def __init__(self, command_template, old_tree, new_tree, to_file,
 
727
                 path_encoding='utf-8'):
 
728
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
729
        self.command_template = command_template
 
730
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
 
731
 
 
732
    @classmethod
 
733
    def from_string(klass, command_string, old_tree, new_tree, to_file,
 
734
                    path_encoding='utf-8'):
 
735
        command_template = cmdline.split(command_string)
 
736
        if '@' not in command_string:
 
737
            command_template.extend(['@old_path', '@new_path'])
 
738
        return klass(command_template, old_tree, new_tree, to_file,
 
739
                     path_encoding)
 
740
 
 
741
    @classmethod
 
742
    def make_from_diff_tree(klass, command_string):
 
743
        def from_diff_tree(diff_tree):
 
744
            return klass.from_string(command_string, diff_tree.old_tree,
 
745
                                     diff_tree.new_tree, diff_tree.to_file)
 
746
        return from_diff_tree
 
747
 
 
748
    def _get_command(self, old_path, new_path):
 
749
        my_map = {'old_path': old_path, 'new_path': new_path}
 
750
        return [AtTemplate(t).substitute(my_map) for t in
 
751
                self.command_template]
 
752
 
 
753
    def _execute(self, old_path, new_path):
 
754
        command = self._get_command(old_path, new_path)
 
755
        try:
 
756
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
 
757
                                    cwd=self._root)
 
758
        except OSError, e:
 
759
            if e.errno == errno.ENOENT:
 
760
                raise errors.ExecutableMissing(command[0])
 
761
            else:
 
762
                raise
 
763
        self.to_file.write(proc.stdout.read())
 
764
        return proc.wait()
 
765
 
 
766
    def _try_symlink_root(self, tree, prefix):
 
767
        if (getattr(tree, 'abspath', None) is None
 
768
            or not osutils.host_os_dereferences_symlinks()):
 
769
            return False
 
770
        try:
 
771
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
 
772
        except OSError, e:
 
773
            if e.errno != errno.EEXIST:
 
774
                raise
 
775
        return True
 
776
 
 
777
    def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
 
778
                    allow_write=False):
 
779
        if not force_temp and isinstance(tree, WorkingTree):
 
780
            return tree.abspath(tree.id2path(file_id))
 
781
        
 
782
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
783
        if not force_temp and self._try_symlink_root(tree, prefix):
 
784
            return full_path
 
785
        parent_dir = osutils.dirname(full_path)
 
786
        try:
 
787
            os.makedirs(parent_dir)
 
788
        except OSError, e:
 
789
            if e.errno != errno.EEXIST:
 
790
                raise
 
791
        source = tree.get_file(file_id, relpath)
 
792
        try:
 
793
            target = open(full_path, 'wb')
 
794
            try:
 
795
                osutils.pumpfile(source, target)
 
796
            finally:
 
797
                target.close()
 
798
        finally:
 
799
            source.close()
 
800
        try:
 
801
            mtime = tree.get_file_mtime(file_id)
 
802
        except errors.FileTimestampUnavailable:
 
803
            pass
 
804
        else:
 
805
            os.utime(full_path, (mtime, mtime))
 
806
        if not allow_write:
 
807
            osutils.make_readonly(full_path)
 
808
        return full_path
 
809
 
 
810
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
 
811
                       allow_write_new=False):
 
812
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
813
                                         old_path, force_temp)
 
814
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
815
                                         new_path, force_temp,
 
816
                                         allow_write=allow_write_new)
 
817
        return old_disk_path, new_disk_path
 
818
 
 
819
    def finish(self):
 
820
        try:
 
821
            osutils.rmtree(self._root)
 
822
        except OSError, e:
 
823
            if e.errno != errno.ENOENT:
 
824
                mutter("The temporary directory \"%s\" was not "
 
825
                        "cleanly removed: %s." % (self._root, e))
 
826
 
 
827
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
828
        if (old_kind, new_kind) != ('file', 'file'):
 
829
            return DiffPath.CANNOT_DIFF
 
830
        (old_disk_path, new_disk_path) = self._prepare_files(
 
831
                                                file_id, old_path, new_path)
 
832
        self._execute(old_disk_path, new_disk_path)
 
833
 
 
834
    def edit_file(self, file_id):
 
835
        """Use this tool to edit a file.
 
836
 
 
837
        A temporary copy will be edited, and the new contents will be
 
838
        returned.
 
839
 
 
840
        :param file_id: The id of the file to edit.
 
841
        :return: The new contents of the file.
 
842
        """
 
843
        old_path = self.old_tree.id2path(file_id)
 
844
        new_path = self.new_tree.id2path(file_id)
 
845
        new_abs_path = self._prepare_files(file_id, old_path, new_path,
 
846
                                           allow_write_new=True,
 
847
                                           force_temp=True)[1]
 
848
        command = self._get_command(osutils.pathjoin('old', old_path),
 
849
                                    osutils.pathjoin('new', new_path))
 
850
        subprocess.call(command, cwd=self._root)
 
851
        new_file = open(new_abs_path, 'r')
 
852
        try:
 
853
            return new_file.read()
 
854
        finally:
 
855
            new_file.close()
 
856
 
 
857
 
 
858
class DiffTree(object):
 
859
    """Provides textual representations of the difference between two trees.
 
860
 
 
861
    A DiffTree examines two trees and where a file-id has altered
 
862
    between them, generates a textual representation of the difference.
 
863
    DiffTree uses a sequence of DiffPath objects which are each
 
864
    given the opportunity to handle a given altered fileid. The list
 
865
    of DiffPath objects can be extended globally by appending to
 
866
    DiffTree.diff_factories, or for a specific diff operation by
 
867
    supplying the extra_factories option to the appropriate method.
499
868
    """
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)
 
869
 
 
870
    # list of factories that can provide instances of DiffPath objects
 
871
    # may be extended by plugins.
 
872
    diff_factories = [DiffSymlink.from_diff_tree,
 
873
                      DiffDirectory.from_diff_tree]
 
874
 
 
875
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
876
                 diff_text=None, extra_factories=None):
 
877
        """Constructor
 
878
 
 
879
        :param old_tree: Tree to show as old in the comparison
 
880
        :param new_tree: Tree to show as new in the comparison
 
881
        :param to_file: File to write comparision to
 
882
        :param path_encoding: Character encoding to write paths in
 
883
        :param diff_text: DiffPath-type object to use as a last resort for
 
884
            diffing text files.
 
885
        :param extra_factories: Factories of DiffPaths to try before any other
 
886
            DiffPaths"""
 
887
        if diff_text is None:
 
888
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
889
                                 '', '',  internal_diff)
 
890
        self.old_tree = old_tree
 
891
        self.new_tree = new_tree
 
892
        self.to_file = to_file
 
893
        self.path_encoding = path_encoding
 
894
        self.differs = []
 
895
        if extra_factories is not None:
 
896
            self.differs.extend(f(self) for f in extra_factories)
 
897
        self.differs.extend(f(self) for f in self.diff_factories)
 
898
        self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
 
899
 
 
900
    @classmethod
 
901
    def from_trees_options(klass, old_tree, new_tree, to_file,
 
902
                           path_encoding, external_diff_options, old_label,
 
903
                           new_label, using):
 
904
        """Factory for producing a DiffTree.
 
905
 
 
906
        Designed to accept options used by show_diff_trees.
 
907
        :param old_tree: The tree to show as old in the comparison
 
908
        :param new_tree: The tree to show as new in the comparison
 
909
        :param to_file: File to write comparisons to
 
910
        :param path_encoding: Character encoding to use for writing paths
 
911
        :param external_diff_options: If supplied, use the installed diff
 
912
            binary to perform file comparison, using supplied options.
 
913
        :param old_label: Prefix to use for old file labels
 
914
        :param new_label: Prefix to use for new file labels
 
915
        :param using: Commandline to use to invoke an external diff tool
 
916
        """
 
917
        if using is not None:
 
918
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
 
919
        else:
 
920
            extra_factories = []
 
921
        if external_diff_options:
 
922
            opts = external_diff_options.split()
 
923
            def diff_file(olab, olines, nlab, nlines, to_file):
 
924
                external_diff(olab, olines, nlab, nlines, to_file, opts)
 
925
        else:
 
926
            diff_file = internal_diff
 
927
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
928
                             old_label, new_label, diff_file)
 
929
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
 
930
                     extra_factories)
 
931
 
 
932
    def show_diff(self, specific_files, extra_trees=None):
 
933
        """Write tree diff to self.to_file
 
934
 
 
935
        :param specific_files: the specific files to compare (recursive)
 
936
        :param extra_trees: extra trees to use for mapping paths to file_ids
 
937
        """
 
938
        try:
 
939
            return self._show_diff(specific_files, extra_trees)
 
940
        finally:
 
941
            for differ in self.differs:
 
942
                differ.finish()
 
943
 
 
944
    def _show_diff(self, specific_files, extra_trees):
 
945
        # TODO: Generation of pseudo-diffs for added/deleted files could
 
946
        # be usefully made into a much faster special case.
 
947
        iterator = self.new_tree.iter_changes(self.old_tree,
 
948
                                               specific_files=specific_files,
 
949
                                               extra_trees=extra_trees,
 
950
                                               require_versioned=True)
 
951
        has_changes = 0
 
952
        def changes_key(change):
 
953
            old_path, new_path = change[1]
 
954
            path = new_path
 
955
            if path is None:
 
956
                path = old_path
 
957
            return path
 
958
        def get_encoded_path(path):
 
959
            if path is not None:
 
960
                return path.encode(self.path_encoding, "replace")
 
961
        for (file_id, paths, changed_content, versioned, parent, name, kind,
 
962
             executable) in sorted(iterator, key=changes_key):
 
963
            # The root does not get diffed, and items with no known kind (that
 
964
            # is, missing) in both trees are skipped as well.
 
965
            if parent == (None, None) or kind == (None, None):
 
966
                continue
 
967
            oldpath, newpath = paths
 
968
            oldpath_encoded = get_encoded_path(paths[0])
 
969
            newpath_encoded = get_encoded_path(paths[1])
 
970
            old_present = (kind[0] is not None and versioned[0])
 
971
            new_present = (kind[1] is not None and versioned[1])
 
972
            renamed = (parent[0], name[0]) != (parent[1], name[1])
 
973
 
 
974
            properties_changed = []
 
975
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
 
976
 
 
977
            if properties_changed:
 
978
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
 
979
            else:
 
980
                prop_str = ""
 
981
 
 
982
            if (old_present, new_present) == (True, False):
 
983
                self.to_file.write("=== removed %s '%s'\n" %
 
984
                                   (kind[0], oldpath_encoded))
 
985
                newpath = oldpath
 
986
            elif (old_present, new_present) == (False, True):
 
987
                self.to_file.write("=== added %s '%s'\n" %
 
988
                                   (kind[1], newpath_encoded))
 
989
                oldpath = newpath
 
990
            elif renamed:
 
991
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
992
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
 
993
            else:
 
994
                # if it was produced by iter_changes, it must be
 
995
                # modified *somehow*, either content or execute bit.
 
996
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
997
                                   newpath_encoded, prop_str))
 
998
            if changed_content:
 
999
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
 
1000
                has_changes = 1
 
1001
            if renamed:
 
1002
                has_changes = 1
 
1003
        return has_changes
 
1004
 
 
1005
    def diff(self, file_id, old_path, new_path):
 
1006
        """Perform a diff of a single file
 
1007
 
 
1008
        :param file_id: file-id of the file
 
1009
        :param old_path: The path of the file in the old tree
 
1010
        :param new_path: The path of the file in the new tree
 
1011
        """
 
1012
        try:
 
1013
            old_kind = self.old_tree.kind(file_id)
 
1014
        except (errors.NoSuchId, errors.NoSuchFile):
 
1015
            old_kind = None
 
1016
        try:
 
1017
            new_kind = self.new_tree.kind(file_id)
 
1018
        except (errors.NoSuchId, errors.NoSuchFile):
 
1019
            new_kind = None
 
1020
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
 
1021
 
 
1022
 
 
1023
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
1024
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
1025
                                       new_path, old_kind, new_kind)
 
1026
        if result is DiffPath.CANNOT_DIFF:
 
1027
            error_path = new_path
 
1028
            if error_path is None:
 
1029
                error_path = old_path
 
1030
            raise errors.NoDiffFound(error_path)
 
1031
 
 
1032
 
 
1033
format_registry = Registry()
 
1034
format_registry.register('default', DiffTree)