/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: 2009-05-12 03:50:39 UTC
  • mto: This revision was merged to the branch mainline in revision 4593.
  • Revision ID: robertc@robertcollins.net-20090512035039-6x0pahpjpkdnm9zb
Note another possible error.

Show diffs side-by-side

added added

removed removed

Lines of Context:
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
 
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
16
 
17
 
import errno
 
17
import difflib
18
18
import os
19
19
import re
 
20
import shutil
 
21
import sys
 
22
 
 
23
from bzrlib.lazy_import import lazy_import
 
24
lazy_import(globals(), """
 
25
import errno
20
26
import subprocess
21
 
import sys
22
27
import tempfile
23
28
import time
24
29
 
25
30
from bzrlib import (
 
31
    branch as _mod_branch,
 
32
    bzrdir,
 
33
    commands,
26
34
    errors,
27
35
    osutils,
28
 
    )
29
 
# compatability - plugins import compare_trees from diff!!!
30
 
# deprecated as of 0.10
31
 
from bzrlib.delta import compare_trees
32
 
from bzrlib.errors import BzrError
33
 
from bzrlib.patiencediff import unified_diff
34
 
import bzrlib.patiencediff
35
 
from bzrlib.symbol_versioning import (deprecated_function,
36
 
        zero_eight)
37
 
from bzrlib.textfile import check_text_lines
38
 
from bzrlib.trace import mutter, warning
 
36
    patiencediff,
 
37
    textfile,
 
38
    timestamp,
 
39
    views,
 
40
    )
 
41
""")
 
42
 
 
43
from bzrlib.symbol_versioning import (
 
44
    deprecated_function,
 
45
    )
 
46
from bzrlib.trace import mutter, note, warning
39
47
 
40
48
 
41
49
# TODO: Rather than building a changeset object, we should probably
42
50
# invoke callbacks on an object.  That object can either accumulate a
43
51
# list, write them out directly, etc etc.
44
52
 
 
53
 
 
54
class _PrematchedMatcher(difflib.SequenceMatcher):
 
55
    """Allow SequenceMatcher operations to use predetermined blocks"""
 
56
 
 
57
    def __init__(self, matching_blocks):
 
58
        difflib.SequenceMatcher(self, None, None)
 
59
        self.matching_blocks = matching_blocks
 
60
        self.opcodes = None
 
61
 
 
62
 
45
63
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
46
64
                  allow_binary=False, sequence_matcher=None,
47
65
                  path_encoding='utf8'):
60
78
    # both sequences are empty.
61
79
    if not oldlines and not newlines:
62
80
        return
63
 
    
 
81
 
64
82
    if allow_binary is False:
65
 
        check_text_lines(oldlines)
66
 
        check_text_lines(newlines)
 
83
        textfile.check_text_lines(oldlines)
 
84
        textfile.check_text_lines(newlines)
67
85
 
68
86
    if sequence_matcher is None:
69
 
        sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
70
 
    ud = unified_diff(oldlines, newlines,
 
87
        sequence_matcher = patiencediff.PatienceSequenceMatcher
 
88
    ud = patiencediff.unified_diff(oldlines, newlines,
71
89
                      fromfile=old_filename.encode(path_encoding),
72
90
                      tofile=new_filename.encode(path_encoding),
73
91
                      sequencematcher=sequence_matcher)
74
92
 
75
93
    ud = list(ud)
 
94
    if len(ud) == 0: # Identical contents, nothing to do
 
95
        return
76
96
    # work-around for difflib being too smart for its own good
77
97
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
78
98
    if not oldlines:
79
99
        ud[2] = ud[2].replace('-1,0', '-0,0')
80
100
    elif not newlines:
81
101
        ud[2] = ud[2].replace('+1,0', '+0,0')
82
 
    # work around for difflib emitting random spaces after the label
83
 
    ud[0] = ud[0][:-2] + '\n'
84
 
    ud[1] = ud[1][:-2] + '\n'
85
102
 
86
103
    for line in ud:
87
104
        to_file.write(line)
88
105
        if not line.endswith('\n'):
89
106
            to_file.write("\n\\ No newline at end of file\n")
90
 
    print >>to_file
91
 
 
92
 
 
93
 
def _set_lang_C():
94
 
    """Set the env var LANG=C"""
95
 
    osutils.set_or_unset_env('LANG', 'C')
96
 
    osutils.set_or_unset_env('LC_ALL', None)
97
 
    osutils.set_or_unset_env('LC_CTYPE', None)
98
 
    osutils.set_or_unset_env('LANGUAGE', None)
 
107
    to_file.write('\n')
99
108
 
100
109
 
101
110
def _spawn_external_diff(diffcmd, capture_errors=True):
102
111
    """Spawn the externall diff process, and return the child handle.
103
112
 
104
113
    :param diffcmd: The command list to spawn
105
 
    :param capture_errors: Capture stderr as well as setting LANG=C.
106
 
        This lets us read and understand the output of diff, and respond 
107
 
        to any errors.
 
114
    :param capture_errors: Capture stderr as well as setting LANG=C
 
115
        and LC_ALL=C. This lets us read and understand the output of diff,
 
116
        and respond to any errors.
108
117
    :return: A Popen object.
109
118
    """
110
119
    if capture_errors:
111
 
        if sys.platform == 'win32':
112
 
            # Win32 doesn't support preexec_fn, but that is
113
 
            # okay, because it doesn't support LANG either.
114
 
            preexec_fn = None
115
 
        else:
116
 
            preexec_fn = _set_lang_C
 
120
        # construct minimal environment
 
121
        env = {}
 
122
        path = os.environ.get('PATH')
 
123
        if path is not None:
 
124
            env['PATH'] = path
 
125
        env['LANGUAGE'] = 'C'   # on win32 only LANGUAGE has effect
 
126
        env['LANG'] = 'C'
 
127
        env['LC_ALL'] = 'C'
117
128
        stderr = subprocess.PIPE
118
129
    else:
119
 
        preexec_fn = None
 
130
        env = None
120
131
        stderr = None
121
132
 
122
133
    try:
124
135
                                stdin=subprocess.PIPE,
125
136
                                stdout=subprocess.PIPE,
126
137
                                stderr=stderr,
127
 
                                preexec_fn=preexec_fn)
 
138
                                env=env)
128
139
    except OSError, e:
129
140
        if e.errno == errno.ENOENT:
130
141
            raise errors.NoDiff(str(e))
188
199
            break
189
200
        else:
190
201
            diffcmd.append('-u')
191
 
                  
 
202
 
192
203
        if diff_opts:
193
204
            diffcmd.extend(diff_opts)
194
205
 
195
206
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
196
207
        out,err = pipe.communicate()
197
208
        rc = pipe.returncode
198
 
        
 
209
 
199
210
        # internal_diff() adds a trailing newline, add one here for consistency
200
211
        out += '\n'
201
212
        if rc == 2:
212
223
            # Write out the new i18n diff response
213
224
            to_file.write(out+'\n')
214
225
            if pipe.returncode != 2:
215
 
                raise BzrError('external diff failed with exit code 2'
216
 
                               ' when run with LANG=C, but not when run'
217
 
                               ' natively: %r' % (diffcmd,))
 
226
                raise errors.BzrError(
 
227
                               'external diff failed with exit code 2'
 
228
                               ' when run with LANG=C and LC_ALL=C,'
 
229
                               ' but not when run natively: %r' % (diffcmd,))
218
230
 
219
231
            first_line = lang_c_out.split('\n', 1)[0]
220
232
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
221
233
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
222
234
            if m is None:
223
 
                raise BzrError('external diff failed with exit code 2;'
224
 
                               ' command: %r' % (diffcmd,))
 
235
                raise errors.BzrError('external diff failed with exit code 2;'
 
236
                                      ' command: %r' % (diffcmd,))
225
237
            else:
226
238
                # Binary files differ, just return
227
239
                return
235
247
                msg = 'signal %d' % (-rc)
236
248
            else:
237
249
                msg = 'exit code %d' % rc
238
 
                
239
 
            raise BzrError('external diff failed with %s; command: %r' 
240
 
                           % (rc, diffcmd))
 
250
 
 
251
            raise errors.BzrError('external diff failed with %s; command: %r'
 
252
                                  % (rc, diffcmd))
241
253
 
242
254
 
243
255
    finally:
260
272
                        new_abspath, e)
261
273
 
262
274
 
263
 
@deprecated_function(zero_eight)
264
 
def show_diff(b, from_spec, specific_files, external_diff_options=None,
265
 
              revision2=None, output=None, b2=None):
266
 
    """Shortcut for showing the diff to the working tree.
267
 
 
268
 
    Please use show_diff_trees instead.
269
 
 
270
 
    b
271
 
        Branch.
272
 
 
273
 
    revision
274
 
        None for 'basis tree', or otherwise the old revision to compare against.
275
 
    
276
 
    The more general form is show_diff_trees(), where the caller
277
 
    supplies any two trees.
278
 
    """
279
 
    if output is None:
280
 
        output = sys.stdout
281
 
 
282
 
    if from_spec is None:
283
 
        old_tree = b.bzrdir.open_workingtree()
284
 
        if b2 is None:
285
 
            old_tree = old_tree = old_tree.basis_tree()
286
 
    else:
287
 
        old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
288
 
 
289
 
    if revision2 is None:
290
 
        if b2 is None:
291
 
            new_tree = b.bzrdir.open_workingtree()
292
 
        else:
293
 
            new_tree = b2.bzrdir.open_workingtree()
294
 
    else:
295
 
        new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
296
 
 
297
 
    return show_diff_trees(old_tree, new_tree, output, specific_files,
298
 
                           external_diff_options)
299
 
 
300
 
 
301
 
def diff_cmd_helper(tree, specific_files, external_diff_options, 
302
 
                    old_revision_spec=None, new_revision_spec=None,
303
 
                    old_label='a/', new_label='b/'):
304
 
    """Helper for cmd_diff.
305
 
 
306
 
   tree 
307
 
        A WorkingTree
308
 
 
309
 
    specific_files
310
 
        The specific files to compare, or None
311
 
 
312
 
    external_diff_options
313
 
        If non-None, run an external diff, and pass it these options
314
 
 
315
 
    old_revision_spec
316
 
        If None, use basis tree as old revision, otherwise use the tree for
317
 
        the specified revision. 
318
 
 
319
 
    new_revision_spec
320
 
        If None, use working tree as new revision, otherwise use the tree for
321
 
        the specified revision.
322
 
    
323
 
    The more general form is show_diff_trees(), where the caller
324
 
    supplies any two trees.
325
 
    """
326
 
    def spec_tree(spec):
327
 
        if tree:
328
 
            revision = spec.in_store(tree.branch)
329
 
        else:
330
 
            revision = spec.in_store(None)
331
 
        revision_id = revision.rev_id
332
 
        branch = revision.branch
333
 
        return branch.repository.revision_tree(revision_id)
334
 
    if old_revision_spec is None:
335
 
        old_tree = tree.basis_tree()
336
 
    else:
337
 
        old_tree = spec_tree(old_revision_spec)
338
 
 
339
 
    if new_revision_spec is None:
340
 
        new_tree = tree
341
 
    else:
342
 
        new_tree = spec_tree(new_revision_spec)
343
 
    if new_tree is not tree:
344
 
        extra_trees = (tree,)
345
 
    else:
346
 
        extra_trees = None
347
 
 
348
 
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
349
 
                           external_diff_options,
350
 
                           old_label=old_label, new_label=new_label,
351
 
                           extra_trees=extra_trees)
 
275
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url,
 
276
    apply_view=True):
 
277
    """Get the trees and specific files to diff given a list of paths.
 
278
 
 
279
    This method works out the trees to be diff'ed and the files of
 
280
    interest within those trees.
 
281
 
 
282
    :param path_list:
 
283
        the list of arguments passed to the diff command
 
284
    :param revision_specs:
 
285
        Zero, one or two RevisionSpecs from the diff command line,
 
286
        saying what revisions to compare.
 
287
    :param old_url:
 
288
        The url of the old branch or tree. If None, the tree to use is
 
289
        taken from the first path, if any, or the current working tree.
 
290
    :param new_url:
 
291
        The url of the new branch or tree. If None, the tree to use is
 
292
        taken from the first path, if any, or the current working tree.
 
293
    :param apply_view:
 
294
        if True and a view is set, apply the view or check that the paths
 
295
        are within it
 
296
    :returns:
 
297
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
298
        extra_trees is a sequence of additional trees to search in for
 
299
        file-ids.
 
300
    """
 
301
    # Get the old and new revision specs
 
302
    old_revision_spec = None
 
303
    new_revision_spec = None
 
304
    if revision_specs is not None:
 
305
        if len(revision_specs) > 0:
 
306
            old_revision_spec = revision_specs[0]
 
307
            if old_url is None:
 
308
                old_url = old_revision_spec.get_branch()
 
309
        if len(revision_specs) > 1:
 
310
            new_revision_spec = revision_specs[1]
 
311
            if new_url is None:
 
312
                new_url = new_revision_spec.get_branch()
 
313
 
 
314
    other_paths = []
 
315
    make_paths_wt_relative = True
 
316
    consider_relpath = True
 
317
    if path_list is None or len(path_list) == 0:
 
318
        # If no path is given, the current working tree is used
 
319
        default_location = u'.'
 
320
        consider_relpath = False
 
321
    elif old_url is not None and new_url is not None:
 
322
        other_paths = path_list
 
323
        make_paths_wt_relative = False
 
324
    else:
 
325
        default_location = path_list[0]
 
326
        other_paths = path_list[1:]
 
327
 
 
328
    # Get the old location
 
329
    specific_files = []
 
330
    if old_url is None:
 
331
        old_url = default_location
 
332
    working_tree, branch, relpath = \
 
333
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
 
334
    if consider_relpath and relpath != '':
 
335
        if working_tree is not None and apply_view:
 
336
            views.check_path_in_view(working_tree, relpath)
 
337
        specific_files.append(relpath)
 
338
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
 
339
 
 
340
    # Get the new location
 
341
    if new_url is None:
 
342
        new_url = default_location
 
343
    if new_url != old_url:
 
344
        working_tree, branch, relpath = \
 
345
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
 
346
        if consider_relpath and relpath != '':
 
347
            if working_tree is not None and apply_view:
 
348
                views.check_path_in_view(working_tree, relpath)
 
349
            specific_files.append(relpath)
 
350
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
 
351
        basis_is_default=working_tree is None)
 
352
 
 
353
    # Get the specific files (all files is None, no files is [])
 
354
    if make_paths_wt_relative and working_tree is not None:
 
355
        try:
 
356
            from bzrlib.builtins import safe_relpath_files
 
357
            other_paths = safe_relpath_files(working_tree, other_paths,
 
358
            apply_view=apply_view)
 
359
        except errors.FileInWrongBranch:
 
360
            raise errors.BzrCommandError("Files are in different branches")
 
361
    specific_files.extend(other_paths)
 
362
    if len(specific_files) == 0:
 
363
        specific_files = None
 
364
        if (working_tree is not None and working_tree.supports_views()
 
365
            and apply_view):
 
366
            view_files = working_tree.views.lookup_view()
 
367
            if view_files:
 
368
                specific_files = view_files
 
369
                view_str = views.view_display_str(view_files)
 
370
                note("*** Ignoring files outside view. View is %s" % view_str)
 
371
 
 
372
    # Get extra trees that ought to be searched for file-ids
 
373
    extra_trees = None
 
374
    if working_tree is not None and working_tree not in (old_tree, new_tree):
 
375
        extra_trees = (working_tree,)
 
376
    return old_tree, new_tree, specific_files, extra_trees
 
377
 
 
378
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
 
379
    if branch is None and tree is not None:
 
380
        branch = tree.branch
 
381
    if spec is None or spec.spec is None:
 
382
        if basis_is_default:
 
383
            if tree is not None:
 
384
                return tree.basis_tree()
 
385
            else:
 
386
                return branch.basis_tree()
 
387
        else:
 
388
            return tree
 
389
    return spec.as_tree(branch)
352
390
 
353
391
 
354
392
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
355
393
                    external_diff_options=None,
356
394
                    old_label='a/', new_label='b/',
357
 
                    extra_trees=None):
 
395
                    extra_trees=None,
 
396
                    path_encoding='utf8',
 
397
                    using=None):
358
398
    """Show in text form the changes from one tree to another.
359
399
 
360
 
    to_files
361
 
        If set, include only changes to these files.
 
400
    to_file
 
401
        The output stream.
 
402
 
 
403
    specific_files
 
404
        Include only changes to these files - None for all changes.
362
405
 
363
406
    external_diff_options
364
407
        If set, use an external GNU diff and pass these options.
365
408
 
366
409
    extra_trees
367
410
        If set, more Trees to use for looking up file ids
 
411
 
 
412
    path_encoding
 
413
        If set, the path will be encoded as specified, otherwise is supposed
 
414
        to be utf8
368
415
    """
369
416
    old_tree.lock_read()
370
417
    try:
 
418
        if extra_trees is not None:
 
419
            for tree in extra_trees:
 
420
                tree.lock_read()
371
421
        new_tree.lock_read()
372
422
        try:
373
 
            return _show_diff_trees(old_tree, new_tree, to_file,
374
 
                                    specific_files, external_diff_options,
375
 
                                    old_label=old_label, new_label=new_label,
376
 
                                    extra_trees=extra_trees)
 
423
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
424
                                                 path_encoding,
 
425
                                                 external_diff_options,
 
426
                                                 old_label, new_label, using)
 
427
            return differ.show_diff(specific_files, extra_trees)
377
428
        finally:
378
429
            new_tree.unlock()
 
430
            if extra_trees is not None:
 
431
                for tree in extra_trees:
 
432
                    tree.unlock()
379
433
    finally:
380
434
        old_tree.unlock()
381
435
 
382
436
 
383
 
def _show_diff_trees(old_tree, new_tree, to_file,
384
 
                     specific_files, external_diff_options, 
385
 
                     old_label='a/', new_label='b/', extra_trees=None):
 
437
def _patch_header_date(tree, file_id, path):
 
438
    """Returns a timestamp suitable for use in a patch header."""
 
439
    mtime = tree.get_file_mtime(file_id, path)
 
440
    return timestamp.format_patch_date(mtime)
 
441
 
 
442
 
 
443
def get_executable_change(old_is_x, new_is_x):
 
444
    descr = { True:"+x", False:"-x", None:"??" }
 
445
    if old_is_x != new_is_x:
 
446
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
447
    else:
 
448
        return []
 
449
 
 
450
 
 
451
class DiffPath(object):
 
452
    """Base type for command object that compare files"""
 
453
 
 
454
    # The type or contents of the file were unsuitable for diffing
 
455
    CANNOT_DIFF = 'CANNOT_DIFF'
 
456
    # The file has changed in a semantic way
 
457
    CHANGED = 'CHANGED'
 
458
    # The file content may have changed, but there is no semantic change
 
459
    UNCHANGED = 'UNCHANGED'
 
460
 
 
461
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
 
462
        """Constructor.
 
463
 
 
464
        :param old_tree: The tree to show as the old tree in the comparison
 
465
        :param new_tree: The tree to show as new in the comparison
 
466
        :param to_file: The file to write comparison data to
 
467
        :param path_encoding: The character encoding to write paths in
 
468
        """
 
469
        self.old_tree = old_tree
 
470
        self.new_tree = new_tree
 
471
        self.to_file = to_file
 
472
        self.path_encoding = path_encoding
 
473
 
 
474
    def finish(self):
 
475
        pass
 
476
 
 
477
    @classmethod
 
478
    def from_diff_tree(klass, diff_tree):
 
479
        return klass(diff_tree.old_tree, diff_tree.new_tree,
 
480
                     diff_tree.to_file, diff_tree.path_encoding)
 
481
 
 
482
    @staticmethod
 
483
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
 
484
        for file_differ in differs:
 
485
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
486
                                      new_kind)
 
487
            if result is not DiffPath.CANNOT_DIFF:
 
488
                return result
 
489
        else:
 
490
            return DiffPath.CANNOT_DIFF
 
491
 
 
492
 
 
493
class DiffKindChange(object):
 
494
    """Special differ for file kind changes.
 
495
 
 
496
    Represents kind change as deletion + creation.  Uses the other differs
 
497
    to do this.
 
498
    """
 
499
    def __init__(self, differs):
 
500
        self.differs = differs
 
501
 
 
502
    def finish(self):
 
503
        pass
 
504
 
 
505
    @classmethod
 
506
    def from_diff_tree(klass, diff_tree):
 
507
        return klass(diff_tree.differs)
 
508
 
 
509
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
510
        """Perform comparison
 
511
 
 
512
        :param file_id: The file_id of the file to compare
 
513
        :param old_path: Path of the file in the old tree
 
514
        :param new_path: Path of the file in the new tree
 
515
        :param old_kind: Old file-kind of the file
 
516
        :param new_kind: New file-kind of the file
 
517
        """
 
518
        if None in (old_kind, new_kind):
 
519
            return DiffPath.CANNOT_DIFF
 
520
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
521
                                       new_path, old_kind, None)
 
522
        if result is DiffPath.CANNOT_DIFF:
 
523
            return result
 
524
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
525
                                     None, new_kind)
 
526
 
 
527
 
 
528
class DiffDirectory(DiffPath):
 
529
 
 
530
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
531
        """Perform comparison between two directories.  (dummy)
 
532
 
 
533
        """
 
534
        if 'directory' not in (old_kind, new_kind):
 
535
            return self.CANNOT_DIFF
 
536
        if old_kind not in ('directory', None):
 
537
            return self.CANNOT_DIFF
 
538
        if new_kind not in ('directory', None):
 
539
            return self.CANNOT_DIFF
 
540
        return self.CHANGED
 
541
 
 
542
 
 
543
class DiffSymlink(DiffPath):
 
544
 
 
545
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
546
        """Perform comparison between two symlinks
 
547
 
 
548
        :param file_id: The file_id of the file to compare
 
549
        :param old_path: Path of the file in the old tree
 
550
        :param new_path: Path of the file in the new tree
 
551
        :param old_kind: Old file-kind of the file
 
552
        :param new_kind: New file-kind of the file
 
553
        """
 
554
        if 'symlink' not in (old_kind, new_kind):
 
555
            return self.CANNOT_DIFF
 
556
        if old_kind == 'symlink':
 
557
            old_target = self.old_tree.get_symlink_target(file_id)
 
558
        elif old_kind is None:
 
559
            old_target = None
 
560
        else:
 
561
            return self.CANNOT_DIFF
 
562
        if new_kind == 'symlink':
 
563
            new_target = self.new_tree.get_symlink_target(file_id)
 
564
        elif new_kind is None:
 
565
            new_target = None
 
566
        else:
 
567
            return self.CANNOT_DIFF
 
568
        return self.diff_symlink(old_target, new_target)
 
569
 
 
570
    def diff_symlink(self, old_target, new_target):
 
571
        if old_target is None:
 
572
            self.to_file.write('=== target is %r\n' % new_target)
 
573
        elif new_target is None:
 
574
            self.to_file.write('=== target was %r\n' % old_target)
 
575
        else:
 
576
            self.to_file.write('=== target changed %r => %r\n' %
 
577
                              (old_target, new_target))
 
578
        return self.CHANGED
 
579
 
 
580
 
 
581
class DiffText(DiffPath):
386
582
 
387
583
    # GNU Patch uses the epoch date to detect files that are being added
388
584
    # or removed in a diff.
389
585
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
390
586
 
391
 
    # TODO: Generation of pseudo-diffs for added/deleted files could
392
 
    # be usefully made into a much faster special case.
393
 
 
394
 
    if external_diff_options:
395
 
        assert isinstance(external_diff_options, basestring)
396
 
        opts = external_diff_options.split()
397
 
        def diff_file(olab, olines, nlab, nlines, to_file):
398
 
            external_diff(olab, olines, nlab, nlines, to_file, opts)
399
 
    else:
400
 
        diff_file = internal_diff
401
 
    
402
 
    delta = new_tree.changes_from(old_tree,
403
 
        specific_files=specific_files,
404
 
        extra_trees=extra_trees, require_versioned=True)
405
 
 
406
 
    has_changes = 0
407
 
    for path, file_id, kind in delta.removed:
408
 
        has_changes = 1
409
 
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
410
 
        old_name = '%s%s\t%s' % (old_label, path,
411
 
                                 _patch_header_date(old_tree, file_id, path))
412
 
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
413
 
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
414
 
                                         new_name, None, None, to_file)
415
 
    for path, file_id, kind in delta.added:
416
 
        has_changes = 1
417
 
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
418
 
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
419
 
        new_name = '%s%s\t%s' % (new_label, path,
420
 
                                 _patch_header_date(new_tree, file_id, path))
421
 
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
422
 
                                         old_name, None, None, to_file, 
423
 
                                         reverse=True)
424
 
    for (old_path, new_path, file_id, kind,
425
 
         text_modified, meta_modified) in delta.renamed:
426
 
        has_changes = 1
427
 
        prop_str = get_prop_change(meta_modified)
428
 
        print >>to_file, '=== renamed %s %r => %r%s' % (
429
 
                    kind, old_path.encode('utf8'),
430
 
                    new_path.encode('utf8'), prop_str)
431
 
        old_name = '%s%s\t%s' % (old_label, old_path,
432
 
                                 _patch_header_date(old_tree, file_id,
433
 
                                                    old_path))
434
 
        new_name = '%s%s\t%s' % (new_label, new_path,
435
 
                                 _patch_header_date(new_tree, file_id,
436
 
                                                    new_path))
437
 
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
438
 
                                    new_name, new_tree,
439
 
                                    text_modified, kind, to_file, diff_file)
440
 
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
441
 
        has_changes = 1
442
 
        prop_str = get_prop_change(meta_modified)
443
 
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
444
 
        old_name = '%s%s\t%s' % (old_label, path,
445
 
                                 _patch_header_date(old_tree, file_id, path))
446
 
        new_name = '%s%s\t%s' % (new_label, path,
447
 
                                 _patch_header_date(new_tree, file_id, path))
448
 
        if text_modified:
449
 
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
450
 
                                        new_name, new_tree,
451
 
                                        True, kind, to_file, diff_file)
452
 
 
453
 
    return has_changes
454
 
 
455
 
 
456
 
def _patch_header_date(tree, file_id, path):
457
 
    """Returns a timestamp suitable for use in a patch header."""
458
 
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
459
 
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
460
 
 
461
 
 
462
 
def _raise_if_nonexistent(paths, old_tree, new_tree):
463
 
    """Complain if paths are not in either inventory or tree.
464
 
 
465
 
    It's OK with the files exist in either tree's inventory, or 
466
 
    if they exist in the tree but are not versioned.
467
 
    
468
 
    This can be used by operations such as bzr status that can accept
469
 
    unknown or ignored files.
 
587
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
588
                 old_label='', new_label='', text_differ=internal_diff):
 
589
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
590
        self.text_differ = text_differ
 
591
        self.old_label = old_label
 
592
        self.new_label = new_label
 
593
        self.path_encoding = path_encoding
 
594
 
 
595
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
596
        """Compare two files in unified diff format
 
597
 
 
598
        :param file_id: The file_id of the file to compare
 
599
        :param old_path: Path of the file in the old tree
 
600
        :param new_path: Path of the file in the new tree
 
601
        :param old_kind: Old file-kind of the file
 
602
        :param new_kind: New file-kind of the file
 
603
        """
 
604
        if 'file' not in (old_kind, new_kind):
 
605
            return self.CANNOT_DIFF
 
606
        from_file_id = to_file_id = file_id
 
607
        if old_kind == 'file':
 
608
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
 
609
        elif old_kind is None:
 
610
            old_date = self.EPOCH_DATE
 
611
            from_file_id = None
 
612
        else:
 
613
            return self.CANNOT_DIFF
 
614
        if new_kind == 'file':
 
615
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
 
616
        elif new_kind is None:
 
617
            new_date = self.EPOCH_DATE
 
618
            to_file_id = None
 
619
        else:
 
620
            return self.CANNOT_DIFF
 
621
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
622
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
623
        return self.diff_text(from_file_id, to_file_id, from_label, to_label)
 
624
 
 
625
    def diff_text(self, from_file_id, to_file_id, from_label, to_label):
 
626
        """Diff the content of given files in two trees
 
627
 
 
628
        :param from_file_id: The id of the file in the from tree.  If None,
 
629
            the file is not present in the from tree.
 
630
        :param to_file_id: The id of the file in the to tree.  This may refer
 
631
            to a different file from from_file_id.  If None,
 
632
            the file is not present in the to tree.
 
633
        """
 
634
        def _get_text(tree, file_id):
 
635
            if file_id is not None:
 
636
                return tree.get_file(file_id).readlines()
 
637
            else:
 
638
                return []
 
639
        try:
 
640
            from_text = _get_text(self.old_tree, from_file_id)
 
641
            to_text = _get_text(self.new_tree, to_file_id)
 
642
            self.text_differ(from_label, from_text, to_label, to_text,
 
643
                             self.to_file)
 
644
        except errors.BinaryFile:
 
645
            self.to_file.write(
 
646
                  ("Binary files %s and %s differ\n" %
 
647
                  (from_label, to_label)).encode(self.path_encoding))
 
648
        return self.CHANGED
 
649
 
 
650
 
 
651
class DiffFromTool(DiffPath):
 
652
 
 
653
    def __init__(self, command_template, old_tree, new_tree, to_file,
 
654
                 path_encoding='utf-8'):
 
655
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
656
        self.command_template = command_template
 
657
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
 
658
 
 
659
    @classmethod
 
660
    def from_string(klass, command_string, old_tree, new_tree, to_file,
 
661
                    path_encoding='utf-8'):
 
662
        command_template = commands.shlex_split_unicode(command_string)
 
663
        command_template.extend(['%(old_path)s', '%(new_path)s'])
 
664
        return klass(command_template, old_tree, new_tree, to_file,
 
665
                     path_encoding)
 
666
 
 
667
    @classmethod
 
668
    def make_from_diff_tree(klass, command_string):
 
669
        def from_diff_tree(diff_tree):
 
670
            return klass.from_string(command_string, diff_tree.old_tree,
 
671
                                     diff_tree.new_tree, diff_tree.to_file)
 
672
        return from_diff_tree
 
673
 
 
674
    def _get_command(self, old_path, new_path):
 
675
        my_map = {'old_path': old_path, 'new_path': new_path}
 
676
        return [t % my_map for t in self.command_template]
 
677
 
 
678
    def _execute(self, old_path, new_path):
 
679
        command = self._get_command(old_path, new_path)
 
680
        try:
 
681
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
 
682
                                    cwd=self._root)
 
683
        except OSError, e:
 
684
            if e.errno == errno.ENOENT:
 
685
                raise errors.ExecutableMissing(command[0])
 
686
            else:
 
687
                raise
 
688
        self.to_file.write(proc.stdout.read())
 
689
        return proc.wait()
 
690
 
 
691
    def _try_symlink_root(self, tree, prefix):
 
692
        if (getattr(tree, 'abspath', None) is None
 
693
            or not osutils.host_os_dereferences_symlinks()):
 
694
            return False
 
695
        try:
 
696
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
 
697
        except OSError, e:
 
698
            if e.errno != errno.EEXIST:
 
699
                raise
 
700
        return True
 
701
 
 
702
    def _write_file(self, file_id, tree, prefix, relpath):
 
703
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
704
        if self._try_symlink_root(tree, prefix):
 
705
            return full_path
 
706
        parent_dir = osutils.dirname(full_path)
 
707
        try:
 
708
            os.makedirs(parent_dir)
 
709
        except OSError, e:
 
710
            if e.errno != errno.EEXIST:
 
711
                raise
 
712
        source = tree.get_file(file_id, relpath)
 
713
        try:
 
714
            target = open(full_path, 'wb')
 
715
            try:
 
716
                osutils.pumpfile(source, target)
 
717
            finally:
 
718
                target.close()
 
719
        finally:
 
720
            source.close()
 
721
        osutils.make_readonly(full_path)
 
722
        mtime = tree.get_file_mtime(file_id)
 
723
        os.utime(full_path, (mtime, mtime))
 
724
        return full_path
 
725
 
 
726
    def _prepare_files(self, file_id, old_path, new_path):
 
727
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
728
                                         old_path)
 
729
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
730
                                         new_path)
 
731
        return old_disk_path, new_disk_path
 
732
 
 
733
    def finish(self):
 
734
        osutils.rmtree(self._root)
 
735
 
 
736
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
737
        if (old_kind, new_kind) != ('file', 'file'):
 
738
            return DiffPath.CANNOT_DIFF
 
739
        self._prepare_files(file_id, old_path, new_path)
 
740
        self._execute(osutils.pathjoin('old', old_path),
 
741
                      osutils.pathjoin('new', new_path))
 
742
 
 
743
 
 
744
class DiffTree(object):
 
745
    """Provides textual representations of the difference between two trees.
 
746
 
 
747
    A DiffTree examines two trees and where a file-id has altered
 
748
    between them, generates a textual representation of the difference.
 
749
    DiffTree uses a sequence of DiffPath objects which are each
 
750
    given the opportunity to handle a given altered fileid. The list
 
751
    of DiffPath objects can be extended globally by appending to
 
752
    DiffTree.diff_factories, or for a specific diff operation by
 
753
    supplying the extra_factories option to the appropriate method.
470
754
    """
471
 
    mutter("check paths: %r", paths)
472
 
    if not paths:
473
 
        return
474
 
    s = old_tree.filter_unversioned_files(paths)
475
 
    s = new_tree.filter_unversioned_files(s)
476
 
    s = [path for path in s if not new_tree.has_filename(path)]
477
 
    if s:
478
 
        raise errors.PathsDoNotExist(sorted(s))
479
 
 
480
 
 
481
 
def get_prop_change(meta_modified):
482
 
    if meta_modified:
483
 
        return " (properties changed)"
484
 
    else:
485
 
        return  ""
486
 
 
487
 
 
488
 
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
489
 
                                new_path, new_tree, text_modified,
490
 
                                kind, to_file, diff_file):
491
 
    if text_modified:
492
 
        new_entry = new_tree.inventory[file_id]
493
 
        old_tree.inventory[file_id].diff(diff_file,
494
 
                                         old_path, old_tree,
495
 
                                         new_path, new_entry, 
496
 
                                         new_tree, to_file)
 
755
 
 
756
    # list of factories that can provide instances of DiffPath objects
 
757
    # may be extended by plugins.
 
758
    diff_factories = [DiffSymlink.from_diff_tree,
 
759
                      DiffDirectory.from_diff_tree]
 
760
 
 
761
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
762
                 diff_text=None, extra_factories=None):
 
763
        """Constructor
 
764
 
 
765
        :param old_tree: Tree to show as old in the comparison
 
766
        :param new_tree: Tree to show as new in the comparison
 
767
        :param to_file: File to write comparision to
 
768
        :param path_encoding: Character encoding to write paths in
 
769
        :param diff_text: DiffPath-type object to use as a last resort for
 
770
            diffing text files.
 
771
        :param extra_factories: Factories of DiffPaths to try before any other
 
772
            DiffPaths"""
 
773
        if diff_text is None:
 
774
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
775
                                 '', '',  internal_diff)
 
776
        self.old_tree = old_tree
 
777
        self.new_tree = new_tree
 
778
        self.to_file = to_file
 
779
        self.path_encoding = path_encoding
 
780
        self.differs = []
 
781
        if extra_factories is not None:
 
782
            self.differs.extend(f(self) for f in extra_factories)
 
783
        self.differs.extend(f(self) for f in self.diff_factories)
 
784
        self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
 
785
 
 
786
    @classmethod
 
787
    def from_trees_options(klass, old_tree, new_tree, to_file,
 
788
                           path_encoding, external_diff_options, old_label,
 
789
                           new_label, using):
 
790
        """Factory for producing a DiffTree.
 
791
 
 
792
        Designed to accept options used by show_diff_trees.
 
793
        :param old_tree: The tree to show as old in the comparison
 
794
        :param new_tree: The tree to show as new in the comparison
 
795
        :param to_file: File to write comparisons to
 
796
        :param path_encoding: Character encoding to use for writing paths
 
797
        :param external_diff_options: If supplied, use the installed diff
 
798
            binary to perform file comparison, using supplied options.
 
799
        :param old_label: Prefix to use for old file labels
 
800
        :param new_label: Prefix to use for new file labels
 
801
        :param using: Commandline to use to invoke an external diff tool
 
802
        """
 
803
        if using is not None:
 
804
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
 
805
        else:
 
806
            extra_factories = []
 
807
        if external_diff_options:
 
808
            opts = external_diff_options.split()
 
809
            def diff_file(olab, olines, nlab, nlines, to_file):
 
810
                external_diff(olab, olines, nlab, nlines, to_file, opts)
 
811
        else:
 
812
            diff_file = internal_diff
 
813
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
814
                             old_label, new_label, diff_file)
 
815
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
 
816
                     extra_factories)
 
817
 
 
818
    def show_diff(self, specific_files, extra_trees=None):
 
819
        """Write tree diff to self.to_file
 
820
 
 
821
        :param sepecific_files: the specific files to compare (recursive)
 
822
        :param extra_trees: extra trees to use for mapping paths to file_ids
 
823
        """
 
824
        try:
 
825
            return self._show_diff(specific_files, extra_trees)
 
826
        finally:
 
827
            for differ in self.differs:
 
828
                differ.finish()
 
829
 
 
830
    def _show_diff(self, specific_files, extra_trees):
 
831
        # TODO: Generation of pseudo-diffs for added/deleted files could
 
832
        # be usefully made into a much faster special case.
 
833
        iterator = self.new_tree.iter_changes(self.old_tree,
 
834
                                               specific_files=specific_files,
 
835
                                               extra_trees=extra_trees,
 
836
                                               require_versioned=True)
 
837
        has_changes = 0
 
838
        def changes_key(change):
 
839
            old_path, new_path = change[1]
 
840
            path = new_path
 
841
            if path is None:
 
842
                path = old_path
 
843
            return path
 
844
        def get_encoded_path(path):
 
845
            if path is not None:
 
846
                return path.encode(self.path_encoding, "replace")
 
847
        for (file_id, paths, changed_content, versioned, parent, name, kind,
 
848
             executable) in sorted(iterator, key=changes_key):
 
849
            # The root does not get diffed, and items with no known kind (that
 
850
            # is, missing) in both trees are skipped as well.
 
851
            if parent == (None, None) or kind == (None, None):
 
852
                continue
 
853
            oldpath, newpath = paths
 
854
            oldpath_encoded = get_encoded_path(paths[0])
 
855
            newpath_encoded = get_encoded_path(paths[1])
 
856
            old_present = (kind[0] is not None and versioned[0])
 
857
            new_present = (kind[1] is not None and versioned[1])
 
858
            renamed = (parent[0], name[0]) != (parent[1], name[1])
 
859
 
 
860
            properties_changed = []
 
861
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
 
862
 
 
863
            if properties_changed:
 
864
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
 
865
            else:
 
866
                prop_str = ""
 
867
 
 
868
            if (old_present, new_present) == (True, False):
 
869
                self.to_file.write("=== removed %s '%s'\n" %
 
870
                                   (kind[0], oldpath_encoded))
 
871
                newpath = oldpath
 
872
            elif (old_present, new_present) == (False, True):
 
873
                self.to_file.write("=== added %s '%s'\n" %
 
874
                                   (kind[1], newpath_encoded))
 
875
                oldpath = newpath
 
876
            elif renamed:
 
877
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
878
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
 
879
            else:
 
880
                # if it was produced by iter_changes, it must be
 
881
                # modified *somehow*, either content or execute bit.
 
882
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
883
                                   newpath_encoded, prop_str))
 
884
            if changed_content:
 
885
                self.diff(file_id, oldpath, newpath)
 
886
                has_changes = 1
 
887
            if renamed:
 
888
                has_changes = 1
 
889
        return has_changes
 
890
 
 
891
    def diff(self, file_id, old_path, new_path):
 
892
        """Perform a diff of a single file
 
893
 
 
894
        :param file_id: file-id of the file
 
895
        :param old_path: The path of the file in the old tree
 
896
        :param new_path: The path of the file in the new tree
 
897
        """
 
898
        try:
 
899
            old_kind = self.old_tree.kind(file_id)
 
900
        except (errors.NoSuchId, errors.NoSuchFile):
 
901
            old_kind = None
 
902
        try:
 
903
            new_kind = self.new_tree.kind(file_id)
 
904
        except (errors.NoSuchId, errors.NoSuchFile):
 
905
            new_kind = None
 
906
 
 
907
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
908
                                       new_path, old_kind, new_kind)
 
909
        if result is DiffPath.CANNOT_DIFF:
 
910
            error_path = new_path
 
911
            if error_path is None:
 
912
                error_path = old_path
 
913
            raise errors.NoDiffFound(error_path)