~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Robert Collins
  • Date: 2007-10-23 22:14:32 UTC
  • mto: (2592.6.3 repository)
  • mto: This revision was merged to the branch mainline in revision 2967.
  • Revision ID: robertc@robertcollins.net-20071023221432-j8zndh1oiegql3cu
* Commit updates the state of the working tree via a delta rather than
  supplying entirely new basis trees. For commit of a single specified file
  this reduces the wall clock time for commit by roughly a 30%.
  (Robert Collins, Martin Pool)

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2005-2011 Canonical Ltd.
 
1
# Copyright (C) 2004, 2005, 2006 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., 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
17
import difflib
18
18
import os
19
19
import re
20
 
import string
21
20
import sys
22
21
 
23
22
from bzrlib.lazy_import import lazy_import
25
24
import errno
26
25
import subprocess
27
26
import tempfile
 
27
import time
28
28
 
29
29
from bzrlib import (
30
 
    bzrdir,
31
 
    cmdline,
32
 
    cleanup,
33
30
    errors,
34
31
    osutils,
35
32
    patiencediff,
36
33
    textfile,
37
34
    timestamp,
38
 
    views,
39
35
    )
40
 
 
41
 
from bzrlib.workingtree import WorkingTree
42
36
""")
43
37
 
44
 
from bzrlib.registry import (
45
 
    Registry,
46
 
    )
47
38
from bzrlib.symbol_versioning import (
48
 
    deprecated_function,
49
 
    deprecated_in,
50
 
    )
51
 
from bzrlib.trace import mutter, note, warning
52
 
 
53
 
 
54
 
class AtTemplate(string.Template):
55
 
    """Templating class that uses @ instead of $."""
56
 
 
57
 
    delimiter = '@'
 
39
        deprecated_function,
 
40
        )
 
41
from bzrlib.trace import mutter, warning
58
42
 
59
43
 
60
44
# TODO: Rather than building a changeset object, we should probably
89
73
    # both sequences are empty.
90
74
    if not oldlines and not newlines:
91
75
        return
92
 
 
 
76
    
93
77
    if allow_binary is False:
94
78
        textfile.check_text_lines(oldlines)
95
79
        textfile.check_text_lines(newlines)
97
81
    if sequence_matcher is None:
98
82
        sequence_matcher = patiencediff.PatienceSequenceMatcher
99
83
    ud = patiencediff.unified_diff(oldlines, newlines,
100
 
                      fromfile=old_filename.encode(path_encoding, 'replace'),
101
 
                      tofile=new_filename.encode(path_encoding, 'replace'),
 
84
                      fromfile=old_filename.encode(path_encoding),
 
85
                      tofile=new_filename.encode(path_encoding),
102
86
                      sequencematcher=sequence_matcher)
103
87
 
104
88
    ud = list(ud)
105
 
    if len(ud) == 0: # Identical contents, nothing to do
106
 
        return
107
89
    # work-around for difflib being too smart for its own good
108
90
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
109
91
    if not oldlines:
110
92
        ud[2] = ud[2].replace('-1,0', '-0,0')
111
93
    elif not newlines:
112
94
        ud[2] = ud[2].replace('+1,0', '+0,0')
 
95
    # work around for difflib emitting random spaces after the label
 
96
    ud[0] = ud[0][:-2] + '\n'
 
97
    ud[1] = ud[1][:-2] + '\n'
113
98
 
114
99
    for line in ud:
115
100
        to_file.write(line)
182
167
 
183
168
        if not diff_opts:
184
169
            diff_opts = []
185
 
        if sys.platform == 'win32':
186
 
            # Popen doesn't do the proper encoding for external commands
187
 
            # Since we are dealing with an ANSI api, use mbcs encoding
188
 
            old_filename = old_filename.encode('mbcs')
189
 
            new_filename = new_filename.encode('mbcs')
190
170
        diffcmd = ['diff',
191
171
                   '--label', old_filename,
192
172
                   old_abspath,
215
195
            break
216
196
        else:
217
197
            diffcmd.append('-u')
218
 
 
 
198
                  
219
199
        if diff_opts:
220
200
            diffcmd.extend(diff_opts)
221
201
 
222
202
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
223
203
        out,err = pipe.communicate()
224
204
        rc = pipe.returncode
225
 
 
 
205
        
226
206
        # internal_diff() adds a trailing newline, add one here for consistency
227
207
        out += '\n'
228
208
        if rc == 2:
263
243
                msg = 'signal %d' % (-rc)
264
244
            else:
265
245
                msg = 'exit code %d' % rc
266
 
 
267
 
            raise errors.BzrError('external diff failed with %s; command: %r'
 
246
                
 
247
            raise errors.BzrError('external diff failed with %s; command: %r' 
268
248
                                  % (rc, diffcmd))
269
249
 
270
250
 
288
268
                        new_abspath, e)
289
269
 
290
270
 
291
 
@deprecated_function(deprecated_in((2, 2, 0)))
292
 
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
293
 
                                   apply_view=True):
294
 
    """Get the trees and specific files to diff given a list of paths.
295
 
 
296
 
    This method works out the trees to be diff'ed and the files of
297
 
    interest within those trees.
298
 
 
299
 
    :param path_list:
300
 
        the list of arguments passed to the diff command
301
 
    :param revision_specs:
302
 
        Zero, one or two RevisionSpecs from the diff command line,
303
 
        saying what revisions to compare.
304
 
    :param old_url:
305
 
        The url of the old branch or tree. If None, the tree to use is
306
 
        taken from the first path, if any, or the current working tree.
307
 
    :param new_url:
308
 
        The url of the new branch or tree. If None, the tree to use is
309
 
        taken from the first path, if any, or the current working tree.
310
 
    :param apply_view:
311
 
        if True and a view is set, apply the view or check that the paths
312
 
        are within it
313
 
    :returns:
314
 
        a tuple of (old_tree, new_tree, old_branch, new_branch,
315
 
        specific_files, extra_trees) where extra_trees is a sequence of
316
 
        additional trees to search in for file-ids.  The trees and branches
317
 
        are not locked.
318
 
    """
319
 
    op = cleanup.OperationWithCleanups(get_trees_and_branches_to_diff_locked)
320
 
    return op.run_simple(path_list, revision_specs, old_url, new_url,
321
 
            op.add_cleanup, apply_view=apply_view)
 
271
def diff_cmd_helper(tree, specific_files, external_diff_options, 
 
272
                    old_revision_spec=None, new_revision_spec=None,
 
273
                    revision_specs=None,
 
274
                    old_label='a/', new_label='b/'):
 
275
    """Helper for cmd_diff.
 
276
 
 
277
    :param tree:
 
278
        A WorkingTree
 
279
 
 
280
    :param specific_files:
 
281
        The specific files to compare, or None
 
282
 
 
283
    :param external_diff_options:
 
284
        If non-None, run an external diff, and pass it these options
 
285
 
 
286
    :param old_revision_spec:
 
287
        If None, use basis tree as old revision, otherwise use the tree for
 
288
        the specified revision. 
 
289
 
 
290
    :param new_revision_spec:
 
291
        If None, use working tree as new revision, otherwise use the tree for
 
292
        the specified revision.
322
293
    
323
 
 
324
 
def get_trees_and_branches_to_diff_locked(
325
 
    path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
326
 
    """Get the trees and specific files to diff given a list of paths.
327
 
 
328
 
    This method works out the trees to be diff'ed and the files of
329
 
    interest within those trees.
330
 
 
331
 
    :param path_list:
332
 
        the list of arguments passed to the diff command
333
 
    :param revision_specs:
334
 
        Zero, one or two RevisionSpecs from the diff command line,
335
 
        saying what revisions to compare.
336
 
    :param old_url:
337
 
        The url of the old branch or tree. If None, the tree to use is
338
 
        taken from the first path, if any, or the current working tree.
339
 
    :param new_url:
340
 
        The url of the new branch or tree. If None, the tree to use is
341
 
        taken from the first path, if any, or the current working tree.
342
 
    :param add_cleanup:
343
 
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
344
 
        will register cleanups that must be run to unlock the trees, etc.
345
 
    :param apply_view:
346
 
        if True and a view is set, apply the view or check that the paths
347
 
        are within it
348
 
    :returns:
349
 
        a tuple of (old_tree, new_tree, old_branch, new_branch,
350
 
        specific_files, extra_trees) where extra_trees is a sequence of
351
 
        additional trees to search in for file-ids.  The trees and branches
352
 
        will be read-locked until the cleanups registered via the add_cleanup
353
 
        param are run.
 
294
    :param revision_specs: 
 
295
        Zero, one or two RevisionSpecs from the command line, saying what revisions 
 
296
        to compare.  This can be passed as an alternative to the old_revision_spec 
 
297
        and new_revision_spec parameters.
 
298
 
 
299
    The more general form is show_diff_trees(), where the caller
 
300
    supplies any two trees.
354
301
    """
355
 
    # Get the old and new revision specs
356
 
    old_revision_spec = None
357
 
    new_revision_spec = None
 
302
 
 
303
    # TODO: perhaps remove the old parameters old_revision_spec and
 
304
    # new_revision_spec, since this is only really for use from cmd_diff and
 
305
    # it now always passes through a sequence of revision_specs -- mbp
 
306
    # 20061221
 
307
 
 
308
    def spec_tree(spec):
 
309
        if tree:
 
310
            revision = spec.in_store(tree.branch)
 
311
        else:
 
312
            revision = spec.in_store(None)
 
313
        revision_id = revision.rev_id
 
314
        branch = revision.branch
 
315
        return branch.repository.revision_tree(revision_id)
 
316
 
358
317
    if revision_specs is not None:
 
318
        assert (old_revision_spec is None
 
319
                and new_revision_spec is None)
359
320
        if len(revision_specs) > 0:
360
321
            old_revision_spec = revision_specs[0]
361
 
            if old_url is None:
362
 
                old_url = old_revision_spec.get_branch()
363
322
        if len(revision_specs) > 1:
364
323
            new_revision_spec = revision_specs[1]
365
 
            if new_url is None:
366
 
                new_url = new_revision_spec.get_branch()
367
 
 
368
 
    other_paths = []
369
 
    make_paths_wt_relative = True
370
 
    consider_relpath = True
371
 
    if path_list is None or len(path_list) == 0:
372
 
        # If no path is given, the current working tree is used
373
 
        default_location = u'.'
374
 
        consider_relpath = False
375
 
    elif old_url is not None and new_url is not None:
376
 
        other_paths = path_list
377
 
        make_paths_wt_relative = False
378
 
    else:
379
 
        default_location = path_list[0]
380
 
        other_paths = path_list[1:]
381
 
 
382
 
    def lock_tree_or_branch(wt, br):
383
 
        if wt is not None:
384
 
            wt.lock_read()
385
 
            add_cleanup(wt.unlock)
386
 
        elif br is not None:
387
 
            br.lock_read()
388
 
            add_cleanup(br.unlock)
389
 
 
390
 
    # Get the old location
391
 
    specific_files = []
392
 
    if old_url is None:
393
 
        old_url = default_location
394
 
    working_tree, branch, relpath = \
395
 
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
396
 
    lock_tree_or_branch(working_tree, branch)
397
 
    if consider_relpath and relpath != '':
398
 
        if working_tree is not None and apply_view:
399
 
            views.check_path_in_view(working_tree, relpath)
400
 
        specific_files.append(relpath)
401
 
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
402
 
    old_branch = branch
403
 
 
404
 
    # Get the new location
405
 
    if new_url is None:
406
 
        new_url = default_location
407
 
    if new_url != old_url:
408
 
        working_tree, branch, relpath = \
409
 
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
410
 
        lock_tree_or_branch(working_tree, branch)
411
 
        if consider_relpath and relpath != '':
412
 
            if working_tree is not None and apply_view:
413
 
                views.check_path_in_view(working_tree, relpath)
414
 
            specific_files.append(relpath)
415
 
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
416
 
        basis_is_default=working_tree is None)
417
 
    new_branch = branch
418
 
 
419
 
    # Get the specific files (all files is None, no files is [])
420
 
    if make_paths_wt_relative and working_tree is not None:
421
 
        other_paths = working_tree.safe_relpath_files(
422
 
            other_paths,
423
 
            apply_view=apply_view)
424
 
    specific_files.extend(other_paths)
425
 
    if len(specific_files) == 0:
426
 
        specific_files = None
427
 
        if (working_tree is not None and working_tree.supports_views()
428
 
            and apply_view):
429
 
            view_files = working_tree.views.lookup_view()
430
 
            if view_files:
431
 
                specific_files = view_files
432
 
                view_str = views.view_display_str(view_files)
433
 
                note("*** Ignoring files outside view. View is %s" % view_str)
434
 
 
435
 
    # Get extra trees that ought to be searched for file-ids
436
 
    extra_trees = None
437
 
    if working_tree is not None and working_tree not in (old_tree, new_tree):
438
 
        extra_trees = (working_tree,)
439
 
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
440
 
 
441
 
 
442
 
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
443
 
    if branch is None and tree is not None:
444
 
        branch = tree.branch
445
 
    if spec is None or spec.spec is None:
446
 
        if basis_is_default:
447
 
            if tree is not None:
448
 
                return tree.basis_tree()
449
 
            else:
450
 
                return branch.basis_tree()
451
 
        else:
452
 
            return tree
453
 
    return spec.as_tree(branch)
 
324
 
 
325
    if old_revision_spec is None:
 
326
        old_tree = tree.basis_tree()
 
327
    else:
 
328
        old_tree = spec_tree(old_revision_spec)
 
329
 
 
330
    if (new_revision_spec is None
 
331
        or new_revision_spec.spec is None):
 
332
        new_tree = tree
 
333
    else:
 
334
        new_tree = spec_tree(new_revision_spec)
 
335
 
 
336
    if new_tree is not tree:
 
337
        extra_trees = (tree,)
 
338
    else:
 
339
        extra_trees = None
 
340
 
 
341
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
 
342
                           external_diff_options,
 
343
                           old_label=old_label, new_label=new_label,
 
344
                           extra_trees=extra_trees)
454
345
 
455
346
 
456
347
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
457
348
                    external_diff_options=None,
458
349
                    old_label='a/', new_label='b/',
459
350
                    extra_trees=None,
460
 
                    path_encoding='utf8',
461
 
                    using=None,
462
 
                    format_cls=None):
 
351
                    path_encoding='utf8'):
463
352
    """Show in text form the changes from one tree to another.
464
353
 
465
 
    :param to_file: The output stream.
466
 
    :param specific_files:Include only changes to these files - None for all
467
 
        changes.
468
 
    :param external_diff_options: If set, use an external GNU diff and pass 
469
 
        these options.
470
 
    :param extra_trees: If set, more Trees to use for looking up file ids
471
 
    :param path_encoding: If set, the path will be encoded as specified, 
472
 
        otherwise is supposed to be utf8
473
 
    :param format_cls: Formatter class (DiffTree subclass)
 
354
    to_files
 
355
        If set, include only changes to these files.
 
356
 
 
357
    external_diff_options
 
358
        If set, use an external GNU diff and pass these options.
 
359
 
 
360
    extra_trees
 
361
        If set, more Trees to use for looking up file ids
 
362
 
 
363
    path_encoding
 
364
        If set, the path will be encoded as specified, otherwise is supposed
 
365
        to be utf8
474
366
    """
475
 
    if format_cls is None:
476
 
        format_cls = DiffTree
477
367
    old_tree.lock_read()
478
368
    try:
479
369
        if extra_trees is not None:
481
371
                tree.lock_read()
482
372
        new_tree.lock_read()
483
373
        try:
484
 
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
485
 
                                                   path_encoding,
486
 
                                                   external_diff_options,
487
 
                                                   old_label, new_label, using)
488
 
            return differ.show_diff(specific_files, extra_trees)
 
374
            return _show_diff_trees(old_tree, new_tree, to_file,
 
375
                                    specific_files, external_diff_options,
 
376
                                    old_label=old_label, new_label=new_label,
 
377
                                    extra_trees=extra_trees,
 
378
                                    path_encoding=path_encoding)
489
379
        finally:
490
380
            new_tree.unlock()
491
381
            if extra_trees is not None:
495
385
        old_tree.unlock()
496
386
 
497
387
 
 
388
def _show_diff_trees(old_tree, new_tree, to_file,
 
389
                     specific_files, external_diff_options, path_encoding,
 
390
                     old_label='a/', new_label='b/', extra_trees=None):
 
391
 
 
392
    # GNU Patch uses the epoch date to detect files that are being added
 
393
    # or removed in a diff.
 
394
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
 
395
 
 
396
    # TODO: Generation of pseudo-diffs for added/deleted files could
 
397
    # be usefully made into a much faster special case.
 
398
 
 
399
    if external_diff_options:
 
400
        assert isinstance(external_diff_options, basestring)
 
401
        opts = external_diff_options.split()
 
402
        def diff_file(olab, olines, nlab, nlines, to_file):
 
403
            external_diff(olab, olines, nlab, nlines, to_file, opts)
 
404
    else:
 
405
        diff_file = internal_diff
 
406
    
 
407
    delta = new_tree.changes_from(old_tree,
 
408
        specific_files=specific_files,
 
409
        extra_trees=extra_trees, require_versioned=True)
 
410
 
 
411
    has_changes = 0
 
412
    for path, file_id, kind in delta.removed:
 
413
        has_changes = 1
 
414
        path_encoded = path.encode(path_encoding, "replace")
 
415
        to_file.write("=== removed %s '%s'\n" % (kind, path_encoded))
 
416
        old_name = '%s%s\t%s' % (old_label, path,
 
417
                                 _patch_header_date(old_tree, file_id, path))
 
418
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
419
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
420
                                         new_name, None, None, to_file)
 
421
    for path, file_id, kind in delta.added:
 
422
        has_changes = 1
 
423
        path_encoded = path.encode(path_encoding, "replace")
 
424
        to_file.write("=== added %s '%s'\n" % (kind, path_encoded))
 
425
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
426
        new_name = '%s%s\t%s' % (new_label, path,
 
427
                                 _patch_header_date(new_tree, file_id, path))
 
428
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
429
                                         old_name, None, None, to_file, 
 
430
                                         reverse=True)
 
431
    for (old_path, new_path, file_id, kind,
 
432
         text_modified, meta_modified) in delta.renamed:
 
433
        has_changes = 1
 
434
        prop_str = get_prop_change(meta_modified)
 
435
        oldpath_encoded = old_path.encode(path_encoding, "replace")
 
436
        newpath_encoded = new_path.encode(path_encoding, "replace")
 
437
        to_file.write("=== renamed %s '%s' => '%s'%s\n" % (kind,
 
438
                            oldpath_encoded, newpath_encoded, prop_str))
 
439
        old_name = '%s%s\t%s' % (old_label, old_path,
 
440
                                 _patch_header_date(old_tree, file_id,
 
441
                                                    old_path))
 
442
        new_name = '%s%s\t%s' % (new_label, new_path,
 
443
                                 _patch_header_date(new_tree, file_id,
 
444
                                                    new_path))
 
445
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
446
                                    new_name, new_tree,
 
447
                                    text_modified, kind, to_file, diff_file)
 
448
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
 
449
        has_changes = 1
 
450
        prop_str = get_prop_change(meta_modified)
 
451
        path_encoded = path.encode(path_encoding, "replace")
 
452
        to_file.write("=== modified %s '%s'%s\n" % (kind,
 
453
                            path_encoded, prop_str))
 
454
        # The file may be in a different location in the old tree (because
 
455
        # the containing dir was renamed, but the file itself was not)
 
456
        old_path = old_tree.id2path(file_id)
 
457
        old_name = '%s%s\t%s' % (old_label, old_path,
 
458
                                 _patch_header_date(old_tree, file_id, old_path))
 
459
        new_name = '%s%s\t%s' % (new_label, path,
 
460
                                 _patch_header_date(new_tree, file_id, path))
 
461
        if text_modified:
 
462
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
463
                                        new_name, new_tree,
 
464
                                        True, kind, to_file, diff_file)
 
465
 
 
466
    return has_changes
 
467
 
 
468
 
498
469
def _patch_header_date(tree, file_id, path):
499
470
    """Returns a timestamp suitable for use in a patch header."""
500
 
    try:
501
 
        mtime = tree.get_file_mtime(file_id, path)
502
 
    except errors.FileTimestampUnavailable:
503
 
        mtime = 0
 
471
    mtime = tree.get_file_mtime(file_id, path)
 
472
    assert mtime is not None, \
 
473
        "got an mtime of None for file-id %s, path %s in tree %s" % (
 
474
                file_id, path, tree)
504
475
    return timestamp.format_patch_date(mtime)
505
476
 
506
477
 
507
 
def get_executable_change(old_is_x, new_is_x):
508
 
    descr = { True:"+x", False:"-x", None:"??" }
509
 
    if old_is_x != new_is_x:
510
 
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
478
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
479
    """Complain if paths are not in either inventory or tree.
 
480
 
 
481
    It's OK with the files exist in either tree's inventory, or 
 
482
    if they exist in the tree but are not versioned.
 
483
    
 
484
    This can be used by operations such as bzr status that can accept
 
485
    unknown or ignored files.
 
486
    """
 
487
    mutter("check paths: %r", paths)
 
488
    if not paths:
 
489
        return
 
490
    s = old_tree.filter_unversioned_files(paths)
 
491
    s = new_tree.filter_unversioned_files(s)
 
492
    s = [path for path in s if not new_tree.has_filename(path)]
 
493
    if s:
 
494
        raise errors.PathsDoNotExist(sorted(s))
 
495
 
 
496
 
 
497
def get_prop_change(meta_modified):
 
498
    if meta_modified:
 
499
        return " (properties changed)"
511
500
    else:
512
 
        return []
513
 
 
514
 
 
515
 
class DiffPath(object):
516
 
    """Base type for command object that compare files"""
517
 
 
518
 
    # The type or contents of the file were unsuitable for diffing
519
 
    CANNOT_DIFF = 'CANNOT_DIFF'
520
 
    # The file has changed in a semantic way
521
 
    CHANGED = 'CHANGED'
522
 
    # The file content may have changed, but there is no semantic change
523
 
    UNCHANGED = 'UNCHANGED'
524
 
 
525
 
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
526
 
        """Constructor.
527
 
 
528
 
        :param old_tree: The tree to show as the old tree in the comparison
529
 
        :param new_tree: The tree to show as new in the comparison
530
 
        :param to_file: The file to write comparison data to
531
 
        :param path_encoding: The character encoding to write paths in
532
 
        """
533
 
        self.old_tree = old_tree
534
 
        self.new_tree = new_tree
535
 
        self.to_file = to_file
536
 
        self.path_encoding = path_encoding
537
 
 
538
 
    def finish(self):
539
 
        pass
540
 
 
541
 
    @classmethod
542
 
    def from_diff_tree(klass, diff_tree):
543
 
        return klass(diff_tree.old_tree, diff_tree.new_tree,
544
 
                     diff_tree.to_file, diff_tree.path_encoding)
545
 
 
546
 
    @staticmethod
547
 
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
548
 
        for file_differ in differs:
549
 
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
550
 
                                      new_kind)
551
 
            if result is not DiffPath.CANNOT_DIFF:
552
 
                return result
553
 
        else:
554
 
            return DiffPath.CANNOT_DIFF
555
 
 
556
 
 
557
 
class DiffKindChange(object):
558
 
    """Special differ for file kind changes.
559
 
 
560
 
    Represents kind change as deletion + creation.  Uses the other differs
561
 
    to do this.
562
 
    """
563
 
    def __init__(self, differs):
564
 
        self.differs = differs
565
 
 
566
 
    def finish(self):
567
 
        pass
568
 
 
569
 
    @classmethod
570
 
    def from_diff_tree(klass, diff_tree):
571
 
        return klass(diff_tree.differs)
572
 
 
573
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
574
 
        """Perform comparison
575
 
 
576
 
        :param file_id: The file_id of the file to compare
577
 
        :param old_path: Path of the file in the old tree
578
 
        :param new_path: Path of the file in the new tree
579
 
        :param old_kind: Old file-kind of the file
580
 
        :param new_kind: New file-kind of the file
581
 
        """
582
 
        if None in (old_kind, new_kind):
583
 
            return DiffPath.CANNOT_DIFF
584
 
        result = DiffPath._diff_many(self.differs, file_id, old_path,
585
 
                                       new_path, old_kind, None)
586
 
        if result is DiffPath.CANNOT_DIFF:
587
 
            return result
588
 
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
589
 
                                     None, new_kind)
590
 
 
591
 
 
592
 
class DiffDirectory(DiffPath):
593
 
 
594
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
595
 
        """Perform comparison between two directories.  (dummy)
596
 
 
597
 
        """
598
 
        if 'directory' not in (old_kind, new_kind):
599
 
            return self.CANNOT_DIFF
600
 
        if old_kind not in ('directory', None):
601
 
            return self.CANNOT_DIFF
602
 
        if new_kind not in ('directory', None):
603
 
            return self.CANNOT_DIFF
604
 
        return self.CHANGED
605
 
 
606
 
 
607
 
class DiffSymlink(DiffPath):
608
 
 
609
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
610
 
        """Perform comparison between two symlinks
611
 
 
612
 
        :param file_id: The file_id of the file to compare
613
 
        :param old_path: Path of the file in the old tree
614
 
        :param new_path: Path of the file in the new tree
615
 
        :param old_kind: Old file-kind of the file
616
 
        :param new_kind: New file-kind of the file
617
 
        """
618
 
        if 'symlink' not in (old_kind, new_kind):
619
 
            return self.CANNOT_DIFF
620
 
        if old_kind == 'symlink':
621
 
            old_target = self.old_tree.get_symlink_target(file_id)
622
 
        elif old_kind is None:
623
 
            old_target = None
624
 
        else:
625
 
            return self.CANNOT_DIFF
626
 
        if new_kind == 'symlink':
627
 
            new_target = self.new_tree.get_symlink_target(file_id)
628
 
        elif new_kind is None:
629
 
            new_target = None
630
 
        else:
631
 
            return self.CANNOT_DIFF
632
 
        return self.diff_symlink(old_target, new_target)
633
 
 
634
 
    def diff_symlink(self, old_target, new_target):
635
 
        if old_target is None:
636
 
            self.to_file.write('=== target is %r\n' % new_target)
637
 
        elif new_target is None:
638
 
            self.to_file.write('=== target was %r\n' % old_target)
639
 
        else:
640
 
            self.to_file.write('=== target changed %r => %r\n' %
641
 
                              (old_target, new_target))
642
 
        return self.CHANGED
643
 
 
644
 
 
645
 
class DiffText(DiffPath):
646
 
 
647
 
    # GNU Patch uses the epoch date to detect files that are being added
648
 
    # or removed in a diff.
649
 
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
650
 
 
651
 
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
652
 
                 old_label='', new_label='', text_differ=internal_diff):
653
 
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
654
 
        self.text_differ = text_differ
655
 
        self.old_label = old_label
656
 
        self.new_label = new_label
657
 
        self.path_encoding = path_encoding
658
 
 
659
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
660
 
        """Compare two files in unified diff format
661
 
 
662
 
        :param file_id: The file_id of the file to compare
663
 
        :param old_path: Path of the file in the old tree
664
 
        :param new_path: Path of the file in the new tree
665
 
        :param old_kind: Old file-kind of the file
666
 
        :param new_kind: New file-kind of the file
667
 
        """
668
 
        if 'file' not in (old_kind, new_kind):
669
 
            return self.CANNOT_DIFF
670
 
        from_file_id = to_file_id = file_id
671
 
        if old_kind == 'file':
672
 
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
673
 
        elif old_kind is None:
674
 
            old_date = self.EPOCH_DATE
675
 
            from_file_id = None
676
 
        else:
677
 
            return self.CANNOT_DIFF
678
 
        if new_kind == 'file':
679
 
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
680
 
        elif new_kind is None:
681
 
            new_date = self.EPOCH_DATE
682
 
            to_file_id = None
683
 
        else:
684
 
            return self.CANNOT_DIFF
685
 
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
686
 
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
687
 
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
688
 
            old_path, new_path)
689
 
 
690
 
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
691
 
        from_path=None, to_path=None):
692
 
        """Diff the content of given files in two trees
693
 
 
694
 
        :param from_file_id: The id of the file in the from tree.  If None,
695
 
            the file is not present in the from tree.
696
 
        :param to_file_id: The id of the file in the to tree.  This may refer
697
 
            to a different file from from_file_id.  If None,
698
 
            the file is not present in the to tree.
699
 
        :param from_path: The path in the from tree or None if unknown.
700
 
        :param to_path: The path in the to tree or None if unknown.
701
 
        """
702
 
        def _get_text(tree, file_id, path):
703
 
            if file_id is not None:
704
 
                return tree.get_file_lines(file_id, path)
705
 
            else:
706
 
                return []
707
 
        try:
708
 
            from_text = _get_text(self.old_tree, from_file_id, from_path)
709
 
            to_text = _get_text(self.new_tree, to_file_id, to_path)
710
 
            self.text_differ(from_label, from_text, to_label, to_text,
711
 
                             self.to_file, path_encoding=self.path_encoding)
712
 
        except errors.BinaryFile:
713
 
            self.to_file.write(
714
 
                  ("Binary files %s and %s differ\n" %
715
 
                  (from_label, to_label)).encode(self.path_encoding,'replace'))
716
 
        return self.CHANGED
717
 
 
718
 
 
719
 
class DiffFromTool(DiffPath):
720
 
 
721
 
    def __init__(self, command_template, old_tree, new_tree, to_file,
722
 
                 path_encoding='utf-8'):
723
 
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
724
 
        self.command_template = command_template
725
 
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
726
 
 
727
 
    @classmethod
728
 
    def from_string(klass, command_string, old_tree, new_tree, to_file,
729
 
                    path_encoding='utf-8'):
730
 
        command_template = cmdline.split(command_string)
731
 
        if '@' not in command_string:
732
 
            command_template.extend(['@old_path', '@new_path'])
733
 
        return klass(command_template, old_tree, new_tree, to_file,
734
 
                     path_encoding)
735
 
 
736
 
    @classmethod
737
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
738
 
        def from_diff_tree(diff_tree):
739
 
            full_command_string = [command_string]
740
 
            if external_diff_options is not None:
741
 
                full_command_string += ' ' + external_diff_options
742
 
            return klass.from_string(full_command_string, diff_tree.old_tree,
743
 
                                     diff_tree.new_tree, diff_tree.to_file)
744
 
        return from_diff_tree
745
 
 
746
 
    def _get_command(self, old_path, new_path):
747
 
        my_map = {'old_path': old_path, 'new_path': new_path}
748
 
        return [AtTemplate(t).substitute(my_map) for t in
749
 
                self.command_template]
750
 
 
751
 
    def _execute(self, old_path, new_path):
752
 
        command = self._get_command(old_path, new_path)
753
 
        try:
754
 
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
755
 
                                    cwd=self._root)
756
 
        except OSError, e:
757
 
            if e.errno == errno.ENOENT:
758
 
                raise errors.ExecutableMissing(command[0])
759
 
            else:
760
 
                raise
761
 
        self.to_file.write(proc.stdout.read())
762
 
        return proc.wait()
763
 
 
764
 
    def _try_symlink_root(self, tree, prefix):
765
 
        if (getattr(tree, 'abspath', None) is None
766
 
            or not osutils.host_os_dereferences_symlinks()):
767
 
            return False
768
 
        try:
769
 
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
770
 
        except OSError, e:
771
 
            if e.errno != errno.EEXIST:
772
 
                raise
773
 
        return True
774
 
 
775
 
    def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
776
 
                    allow_write=False):
777
 
        if not force_temp and isinstance(tree, WorkingTree):
778
 
            return tree.abspath(tree.id2path(file_id))
779
 
        
780
 
        full_path = osutils.pathjoin(self._root, prefix, relpath)
781
 
        if not force_temp and self._try_symlink_root(tree, prefix):
782
 
            return full_path
783
 
        parent_dir = osutils.dirname(full_path)
784
 
        try:
785
 
            os.makedirs(parent_dir)
786
 
        except OSError, e:
787
 
            if e.errno != errno.EEXIST:
788
 
                raise
789
 
        source = tree.get_file(file_id, relpath)
790
 
        try:
791
 
            target = open(full_path, 'wb')
792
 
            try:
793
 
                osutils.pumpfile(source, target)
794
 
            finally:
795
 
                target.close()
796
 
        finally:
797
 
            source.close()
798
 
        try:
799
 
            mtime = tree.get_file_mtime(file_id)
800
 
        except errors.FileTimestampUnavailable:
801
 
            pass
802
 
        else:
803
 
            os.utime(full_path, (mtime, mtime))
804
 
        if not allow_write:
805
 
            osutils.make_readonly(full_path)
806
 
        return full_path
807
 
 
808
 
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
809
 
                       allow_write_new=False):
810
 
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
811
 
                                         old_path, force_temp)
812
 
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
813
 
                                         new_path, force_temp,
814
 
                                         allow_write=allow_write_new)
815
 
        return old_disk_path, new_disk_path
816
 
 
817
 
    def finish(self):
818
 
        try:
819
 
            osutils.rmtree(self._root)
820
 
        except OSError, e:
821
 
            if e.errno != errno.ENOENT:
822
 
                mutter("The temporary directory \"%s\" was not "
823
 
                        "cleanly removed: %s." % (self._root, e))
824
 
 
825
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
826
 
        if (old_kind, new_kind) != ('file', 'file'):
827
 
            return DiffPath.CANNOT_DIFF
828
 
        (old_disk_path, new_disk_path) = self._prepare_files(
829
 
                                                file_id, old_path, new_path)
830
 
        self._execute(old_disk_path, new_disk_path)
831
 
 
832
 
    def edit_file(self, file_id):
833
 
        """Use this tool to edit a file.
834
 
 
835
 
        A temporary copy will be edited, and the new contents will be
836
 
        returned.
837
 
 
838
 
        :param file_id: The id of the file to edit.
839
 
        :return: The new contents of the file.
840
 
        """
841
 
        old_path = self.old_tree.id2path(file_id)
842
 
        new_path = self.new_tree.id2path(file_id)
843
 
        new_abs_path = self._prepare_files(file_id, old_path, new_path,
844
 
                                           allow_write_new=True,
845
 
                                           force_temp=True)[1]
846
 
        command = self._get_command(osutils.pathjoin('old', old_path),
847
 
                                    osutils.pathjoin('new', new_path))
848
 
        subprocess.call(command, cwd=self._root)
849
 
        new_file = open(new_abs_path, 'r')
850
 
        try:
851
 
            return new_file.read()
852
 
        finally:
853
 
            new_file.close()
854
 
 
855
 
 
856
 
class DiffTree(object):
857
 
    """Provides textual representations of the difference between two trees.
858
 
 
859
 
    A DiffTree examines two trees and where a file-id has altered
860
 
    between them, generates a textual representation of the difference.
861
 
    DiffTree uses a sequence of DiffPath objects which are each
862
 
    given the opportunity to handle a given altered fileid. The list
863
 
    of DiffPath objects can be extended globally by appending to
864
 
    DiffTree.diff_factories, or for a specific diff operation by
865
 
    supplying the extra_factories option to the appropriate method.
866
 
    """
867
 
 
868
 
    # list of factories that can provide instances of DiffPath objects
869
 
    # may be extended by plugins.
870
 
    diff_factories = [DiffSymlink.from_diff_tree,
871
 
                      DiffDirectory.from_diff_tree]
872
 
 
873
 
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
874
 
                 diff_text=None, extra_factories=None):
875
 
        """Constructor
876
 
 
877
 
        :param old_tree: Tree to show as old in the comparison
878
 
        :param new_tree: Tree to show as new in the comparison
879
 
        :param to_file: File to write comparision to
880
 
        :param path_encoding: Character encoding to write paths in
881
 
        :param diff_text: DiffPath-type object to use as a last resort for
882
 
            diffing text files.
883
 
        :param extra_factories: Factories of DiffPaths to try before any other
884
 
            DiffPaths"""
885
 
        if diff_text is None:
886
 
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
887
 
                                 '', '',  internal_diff)
888
 
        self.old_tree = old_tree
889
 
        self.new_tree = new_tree
890
 
        self.to_file = to_file
891
 
        self.path_encoding = path_encoding
892
 
        self.differs = []
893
 
        if extra_factories is not None:
894
 
            self.differs.extend(f(self) for f in extra_factories)
895
 
        self.differs.extend(f(self) for f in self.diff_factories)
896
 
        self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
897
 
 
898
 
    @classmethod
899
 
    def from_trees_options(klass, old_tree, new_tree, to_file,
900
 
                           path_encoding, external_diff_options, old_label,
901
 
                           new_label, using):
902
 
        """Factory for producing a DiffTree.
903
 
 
904
 
        Designed to accept options used by show_diff_trees.
905
 
        :param old_tree: The tree to show as old in the comparison
906
 
        :param new_tree: The tree to show as new in the comparison
907
 
        :param to_file: File to write comparisons to
908
 
        :param path_encoding: Character encoding to use for writing paths
909
 
        :param external_diff_options: If supplied, use the installed diff
910
 
            binary to perform file comparison, using supplied options.
911
 
        :param old_label: Prefix to use for old file labels
912
 
        :param new_label: Prefix to use for new file labels
913
 
        :param using: Commandline to use to invoke an external diff tool
914
 
        """
915
 
        if using is not None:
916
 
            extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
917
 
        else:
918
 
            extra_factories = []
919
 
        if external_diff_options:
920
 
            opts = external_diff_options.split()
921
 
            def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None):
922
 
                """:param path_encoding: not used but required
923
 
                        to match the signature of internal_diff.
924
 
                """
925
 
                external_diff(olab, olines, nlab, nlines, to_file, opts)
926
 
        else:
927
 
            diff_file = internal_diff
928
 
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
929
 
                             old_label, new_label, diff_file)
930
 
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
931
 
                     extra_factories)
932
 
 
933
 
    def show_diff(self, specific_files, extra_trees=None):
934
 
        """Write tree diff to self.to_file
935
 
 
936
 
        :param specific_files: the specific files to compare (recursive)
937
 
        :param extra_trees: extra trees to use for mapping paths to file_ids
938
 
        """
939
 
        try:
940
 
            return self._show_diff(specific_files, extra_trees)
941
 
        finally:
942
 
            for differ in self.differs:
943
 
                differ.finish()
944
 
 
945
 
    def _show_diff(self, specific_files, extra_trees):
946
 
        # TODO: Generation of pseudo-diffs for added/deleted files could
947
 
        # be usefully made into a much faster special case.
948
 
        iterator = self.new_tree.iter_changes(self.old_tree,
949
 
                                               specific_files=specific_files,
950
 
                                               extra_trees=extra_trees,
951
 
                                               require_versioned=True)
952
 
        has_changes = 0
953
 
        def changes_key(change):
954
 
            old_path, new_path = change[1]
955
 
            path = new_path
956
 
            if path is None:
957
 
                path = old_path
958
 
            return path
959
 
        def get_encoded_path(path):
960
 
            if path is not None:
961
 
                return path.encode(self.path_encoding, "replace")
962
 
        for (file_id, paths, changed_content, versioned, parent, name, kind,
963
 
             executable) in sorted(iterator, key=changes_key):
964
 
            # The root does not get diffed, and items with no known kind (that
965
 
            # is, missing) in both trees are skipped as well.
966
 
            if parent == (None, None) or kind == (None, None):
967
 
                continue
968
 
            oldpath, newpath = paths
969
 
            oldpath_encoded = get_encoded_path(paths[0])
970
 
            newpath_encoded = get_encoded_path(paths[1])
971
 
            old_present = (kind[0] is not None and versioned[0])
972
 
            new_present = (kind[1] is not None and versioned[1])
973
 
            renamed = (parent[0], name[0]) != (parent[1], name[1])
974
 
 
975
 
            properties_changed = []
976
 
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
977
 
 
978
 
            if properties_changed:
979
 
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
980
 
            else:
981
 
                prop_str = ""
982
 
 
983
 
            if (old_present, new_present) == (True, False):
984
 
                self.to_file.write("=== removed %s '%s'\n" %
985
 
                                   (kind[0], oldpath_encoded))
986
 
                newpath = oldpath
987
 
            elif (old_present, new_present) == (False, True):
988
 
                self.to_file.write("=== added %s '%s'\n" %
989
 
                                   (kind[1], newpath_encoded))
990
 
                oldpath = newpath
991
 
            elif renamed:
992
 
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
993
 
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
994
 
            else:
995
 
                # if it was produced by iter_changes, it must be
996
 
                # modified *somehow*, either content or execute bit.
997
 
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
998
 
                                   newpath_encoded, prop_str))
999
 
            if changed_content:
1000
 
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1001
 
                has_changes = 1
1002
 
            if renamed:
1003
 
                has_changes = 1
1004
 
        return has_changes
1005
 
 
1006
 
    def diff(self, file_id, old_path, new_path):
1007
 
        """Perform a diff of a single file
1008
 
 
1009
 
        :param file_id: file-id of the file
1010
 
        :param old_path: The path of the file in the old tree
1011
 
        :param new_path: The path of the file in the new tree
1012
 
        """
1013
 
        try:
1014
 
            old_kind = self.old_tree.kind(file_id)
1015
 
        except (errors.NoSuchId, errors.NoSuchFile):
1016
 
            old_kind = None
1017
 
        try:
1018
 
            new_kind = self.new_tree.kind(file_id)
1019
 
        except (errors.NoSuchId, errors.NoSuchFile):
1020
 
            new_kind = None
1021
 
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
1022
 
 
1023
 
 
1024
 
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
1025
 
        result = DiffPath._diff_many(self.differs, file_id, old_path,
1026
 
                                       new_path, old_kind, new_kind)
1027
 
        if result is DiffPath.CANNOT_DIFF:
1028
 
            error_path = new_path
1029
 
            if error_path is None:
1030
 
                error_path = old_path
1031
 
            raise errors.NoDiffFound(error_path)
1032
 
 
1033
 
 
1034
 
format_registry = Registry()
1035
 
format_registry.register('default', DiffTree)
 
501
        return  ""
 
502
 
 
503
 
 
504
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
505
                                new_path, new_tree, text_modified,
 
506
                                kind, to_file, diff_file):
 
507
    if text_modified:
 
508
        new_entry = new_tree.inventory[file_id]
 
509
        old_tree.inventory[file_id].diff(diff_file,
 
510
                                         old_path, old_tree,
 
511
                                         new_path, new_entry, 
 
512
                                         new_tree, to_file)