/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: John Arbash Meinel
  • Date: 2009-10-21 20:53:21 UTC
  • mto: This revision was merged to the branch mainline in revision 4771.
  • Revision ID: john@arbash-meinel.com-20091021205321-j62yud7d5lfx2za1
Revert previous change.

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))
160
171
 
161
172
        if not diff_opts:
162
173
            diff_opts = []
 
174
        if sys.platform == 'win32':
 
175
            # Popen doesn't do the proper encoding for external commands
 
176
            # Since we are dealing with an ANSI api, use mbcs encoding
 
177
            old_filename = old_filename.encode('mbcs')
 
178
            new_filename = new_filename.encode('mbcs')
163
179
        diffcmd = ['diff',
164
180
                   '--label', old_filename,
165
181
                   old_abspath,
188
204
            break
189
205
        else:
190
206
            diffcmd.append('-u')
191
 
                  
 
207
 
192
208
        if diff_opts:
193
209
            diffcmd.extend(diff_opts)
194
210
 
195
211
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
196
212
        out,err = pipe.communicate()
197
213
        rc = pipe.returncode
198
 
        
 
214
 
199
215
        # internal_diff() adds a trailing newline, add one here for consistency
200
216
        out += '\n'
201
217
        if rc == 2:
212
228
            # Write out the new i18n diff response
213
229
            to_file.write(out+'\n')
214
230
            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,))
 
231
                raise errors.BzrError(
 
232
                               'external diff failed with exit code 2'
 
233
                               ' when run with LANG=C and LC_ALL=C,'
 
234
                               ' but not when run natively: %r' % (diffcmd,))
218
235
 
219
236
            first_line = lang_c_out.split('\n', 1)[0]
220
237
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
221
238
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
222
239
            if m is None:
223
 
                raise BzrError('external diff failed with exit code 2;'
224
 
                               ' command: %r' % (diffcmd,))
 
240
                raise errors.BzrError('external diff failed with exit code 2;'
 
241
                                      ' command: %r' % (diffcmd,))
225
242
            else:
226
243
                # Binary files differ, just return
227
244
                return
235
252
                msg = 'signal %d' % (-rc)
236
253
            else:
237
254
                msg = 'exit code %d' % rc
238
 
                
239
 
            raise BzrError('external diff failed with %s; command: %r' 
240
 
                           % (rc, diffcmd))
 
255
 
 
256
            raise errors.BzrError('external diff failed with %s; command: %r'
 
257
                                  % (rc, diffcmd))
241
258
 
242
259
 
243
260
    finally:
260
277
                        new_abspath, e)
261
278
 
262
279
 
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)
 
280
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
 
281
                                   apply_view=True):
 
282
    """Get the trees and specific files to diff given a list of paths.
 
283
 
 
284
    This method works out the trees to be diff'ed and the files of
 
285
    interest within those trees.
 
286
 
 
287
    :param path_list:
 
288
        the list of arguments passed to the diff command
 
289
    :param revision_specs:
 
290
        Zero, one or two RevisionSpecs from the diff command line,
 
291
        saying what revisions to compare.
 
292
    :param old_url:
 
293
        The url of the old branch or tree. If None, the tree to use is
 
294
        taken from the first path, if any, or the current working tree.
 
295
    :param new_url:
 
296
        The url of the new branch or tree. If None, the tree to use is
 
297
        taken from the first path, if any, or the current working tree.
 
298
    :param apply_view:
 
299
        if True and a view is set, apply the view or check that the paths
 
300
        are within it
 
301
    :returns:
 
302
        a tuple of (old_tree, new_tree, old_branch, new_branch,
 
303
        specific_files, extra_trees) where extra_trees is a sequence of
 
304
        additional trees to search in for file-ids.
 
305
    """
 
306
    # Get the old and new revision specs
 
307
    old_revision_spec = None
 
308
    new_revision_spec = None
 
309
    if revision_specs is not None:
 
310
        if len(revision_specs) > 0:
 
311
            old_revision_spec = revision_specs[0]
 
312
            if old_url is None:
 
313
                old_url = old_revision_spec.get_branch()
 
314
        if len(revision_specs) > 1:
 
315
            new_revision_spec = revision_specs[1]
 
316
            if new_url is None:
 
317
                new_url = new_revision_spec.get_branch()
 
318
 
 
319
    other_paths = []
 
320
    make_paths_wt_relative = True
 
321
    consider_relpath = True
 
322
    if path_list is None or len(path_list) == 0:
 
323
        # If no path is given, the current working tree is used
 
324
        default_location = u'.'
 
325
        consider_relpath = False
 
326
    elif old_url is not None and new_url is not None:
 
327
        other_paths = path_list
 
328
        make_paths_wt_relative = False
 
329
    else:
 
330
        default_location = path_list[0]
 
331
        other_paths = path_list[1:]
 
332
 
 
333
    # Get the old location
 
334
    specific_files = []
 
335
    if old_url is None:
 
336
        old_url = default_location
 
337
    working_tree, branch, relpath = \
 
338
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
 
339
    if consider_relpath and relpath != '':
 
340
        if working_tree is not None and apply_view:
 
341
            views.check_path_in_view(working_tree, relpath)
 
342
        specific_files.append(relpath)
 
343
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
 
344
    old_branch = branch
 
345
 
 
346
    # Get the new location
 
347
    if new_url is None:
 
348
        new_url = default_location
 
349
    if new_url != old_url:
 
350
        working_tree, branch, relpath = \
 
351
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
 
352
        if consider_relpath and relpath != '':
 
353
            if working_tree is not None and apply_view:
 
354
                views.check_path_in_view(working_tree, relpath)
 
355
            specific_files.append(relpath)
 
356
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
 
357
        basis_is_default=working_tree is None)
 
358
    new_branch = branch
 
359
 
 
360
    # Get the specific files (all files is None, no files is [])
 
361
    if make_paths_wt_relative and working_tree is not None:
 
362
        try:
 
363
            from bzrlib.builtins import safe_relpath_files
 
364
            other_paths = safe_relpath_files(working_tree, other_paths,
 
365
            apply_view=apply_view)
 
366
        except errors.FileInWrongBranch:
 
367
            raise errors.BzrCommandError("Files are in different branches")
 
368
    specific_files.extend(other_paths)
 
369
    if len(specific_files) == 0:
 
370
        specific_files = None
 
371
        if (working_tree is not None and working_tree.supports_views()
 
372
            and apply_view):
 
373
            view_files = working_tree.views.lookup_view()
 
374
            if view_files:
 
375
                specific_files = view_files
 
376
                view_str = views.view_display_str(view_files)
 
377
                note("*** Ignoring files outside view. View is %s" % view_str)
 
378
 
 
379
    # Get extra trees that ought to be searched for file-ids
 
380
    extra_trees = None
 
381
    if working_tree is not None and working_tree not in (old_tree, new_tree):
 
382
        extra_trees = (working_tree,)
 
383
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
 
384
 
 
385
 
 
386
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
 
387
    if branch is None and tree is not None:
 
388
        branch = tree.branch
 
389
    if spec is None or spec.spec is None:
 
390
        if basis_is_default:
 
391
            if tree is not None:
 
392
                return tree.basis_tree()
 
393
            else:
 
394
                return branch.basis_tree()
 
395
        else:
 
396
            return tree
 
397
    return spec.as_tree(branch)
352
398
 
353
399
 
354
400
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
355
401
                    external_diff_options=None,
356
402
                    old_label='a/', new_label='b/',
357
 
                    extra_trees=None):
 
403
                    extra_trees=None,
 
404
                    path_encoding='utf8',
 
405
                    using=None):
358
406
    """Show in text form the changes from one tree to another.
359
407
 
360
 
    to_files
361
 
        If set, include only changes to these files.
 
408
    to_file
 
409
        The output stream.
 
410
 
 
411
    specific_files
 
412
        Include only changes to these files - None for all changes.
362
413
 
363
414
    external_diff_options
364
415
        If set, use an external GNU diff and pass these options.
365
416
 
366
417
    extra_trees
367
418
        If set, more Trees to use for looking up file ids
 
419
 
 
420
    path_encoding
 
421
        If set, the path will be encoded as specified, otherwise is supposed
 
422
        to be utf8
368
423
    """
369
424
    old_tree.lock_read()
370
425
    try:
 
426
        if extra_trees is not None:
 
427
            for tree in extra_trees:
 
428
                tree.lock_read()
371
429
        new_tree.lock_read()
372
430
        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)
 
431
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
432
                                                 path_encoding,
 
433
                                                 external_diff_options,
 
434
                                                 old_label, new_label, using)
 
435
            return differ.show_diff(specific_files, extra_trees)
377
436
        finally:
378
437
            new_tree.unlock()
 
438
            if extra_trees is not None:
 
439
                for tree in extra_trees:
 
440
                    tree.unlock()
379
441
    finally:
380
442
        old_tree.unlock()
381
443
 
382
444
 
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):
 
445
def _patch_header_date(tree, file_id, path):
 
446
    """Returns a timestamp suitable for use in a patch header."""
 
447
    mtime = tree.get_file_mtime(file_id, path)
 
448
    return timestamp.format_patch_date(mtime)
 
449
 
 
450
 
 
451
def get_executable_change(old_is_x, new_is_x):
 
452
    descr = { True:"+x", False:"-x", None:"??" }
 
453
    if old_is_x != new_is_x:
 
454
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
455
    else:
 
456
        return []
 
457
 
 
458
 
 
459
class DiffPath(object):
 
460
    """Base type for command object that compare files"""
 
461
 
 
462
    # The type or contents of the file were unsuitable for diffing
 
463
    CANNOT_DIFF = 'CANNOT_DIFF'
 
464
    # The file has changed in a semantic way
 
465
    CHANGED = 'CHANGED'
 
466
    # The file content may have changed, but there is no semantic change
 
467
    UNCHANGED = 'UNCHANGED'
 
468
 
 
469
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
 
470
        """Constructor.
 
471
 
 
472
        :param old_tree: The tree to show as the old tree in the comparison
 
473
        :param new_tree: The tree to show as new in the comparison
 
474
        :param to_file: The file to write comparison data to
 
475
        :param path_encoding: The character encoding to write paths in
 
476
        """
 
477
        self.old_tree = old_tree
 
478
        self.new_tree = new_tree
 
479
        self.to_file = to_file
 
480
        self.path_encoding = path_encoding
 
481
 
 
482
    def finish(self):
 
483
        pass
 
484
 
 
485
    @classmethod
 
486
    def from_diff_tree(klass, diff_tree):
 
487
        return klass(diff_tree.old_tree, diff_tree.new_tree,
 
488
                     diff_tree.to_file, diff_tree.path_encoding)
 
489
 
 
490
    @staticmethod
 
491
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
 
492
        for file_differ in differs:
 
493
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
494
                                      new_kind)
 
495
            if result is not DiffPath.CANNOT_DIFF:
 
496
                return result
 
497
        else:
 
498
            return DiffPath.CANNOT_DIFF
 
499
 
 
500
 
 
501
class DiffKindChange(object):
 
502
    """Special differ for file kind changes.
 
503
 
 
504
    Represents kind change as deletion + creation.  Uses the other differs
 
505
    to do this.
 
506
    """
 
507
    def __init__(self, differs):
 
508
        self.differs = differs
 
509
 
 
510
    def finish(self):
 
511
        pass
 
512
 
 
513
    @classmethod
 
514
    def from_diff_tree(klass, diff_tree):
 
515
        return klass(diff_tree.differs)
 
516
 
 
517
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
518
        """Perform comparison
 
519
 
 
520
        :param file_id: The file_id of the file to compare
 
521
        :param old_path: Path of the file in the old tree
 
522
        :param new_path: Path of the file in the new tree
 
523
        :param old_kind: Old file-kind of the file
 
524
        :param new_kind: New file-kind of the file
 
525
        """
 
526
        if None in (old_kind, new_kind):
 
527
            return DiffPath.CANNOT_DIFF
 
528
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
529
                                       new_path, old_kind, None)
 
530
        if result is DiffPath.CANNOT_DIFF:
 
531
            return result
 
532
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
533
                                     None, new_kind)
 
534
 
 
535
 
 
536
class DiffDirectory(DiffPath):
 
537
 
 
538
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
539
        """Perform comparison between two directories.  (dummy)
 
540
 
 
541
        """
 
542
        if 'directory' not in (old_kind, new_kind):
 
543
            return self.CANNOT_DIFF
 
544
        if old_kind not in ('directory', None):
 
545
            return self.CANNOT_DIFF
 
546
        if new_kind not in ('directory', None):
 
547
            return self.CANNOT_DIFF
 
548
        return self.CHANGED
 
549
 
 
550
 
 
551
class DiffSymlink(DiffPath):
 
552
 
 
553
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
554
        """Perform comparison between two symlinks
 
555
 
 
556
        :param file_id: The file_id of the file to compare
 
557
        :param old_path: Path of the file in the old tree
 
558
        :param new_path: Path of the file in the new tree
 
559
        :param old_kind: Old file-kind of the file
 
560
        :param new_kind: New file-kind of the file
 
561
        """
 
562
        if 'symlink' not in (old_kind, new_kind):
 
563
            return self.CANNOT_DIFF
 
564
        if old_kind == 'symlink':
 
565
            old_target = self.old_tree.get_symlink_target(file_id)
 
566
        elif old_kind is None:
 
567
            old_target = None
 
568
        else:
 
569
            return self.CANNOT_DIFF
 
570
        if new_kind == 'symlink':
 
571
            new_target = self.new_tree.get_symlink_target(file_id)
 
572
        elif new_kind is None:
 
573
            new_target = None
 
574
        else:
 
575
            return self.CANNOT_DIFF
 
576
        return self.diff_symlink(old_target, new_target)
 
577
 
 
578
    def diff_symlink(self, old_target, new_target):
 
579
        if old_target is None:
 
580
            self.to_file.write('=== target is %r\n' % new_target)
 
581
        elif new_target is None:
 
582
            self.to_file.write('=== target was %r\n' % old_target)
 
583
        else:
 
584
            self.to_file.write('=== target changed %r => %r\n' %
 
585
                              (old_target, new_target))
 
586
        return self.CHANGED
 
587
 
 
588
 
 
589
class DiffText(DiffPath):
386
590
 
387
591
    # GNU Patch uses the epoch date to detect files that are being added
388
592
    # or removed in a diff.
389
593
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
390
594
 
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.
 
595
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
596
                 old_label='', new_label='', text_differ=internal_diff):
 
597
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
598
        self.text_differ = text_differ
 
599
        self.old_label = old_label
 
600
        self.new_label = new_label
 
601
        self.path_encoding = path_encoding
 
602
 
 
603
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
604
        """Compare two files in unified diff format
 
605
 
 
606
        :param file_id: The file_id of the file to compare
 
607
        :param old_path: Path of the file in the old tree
 
608
        :param new_path: Path of the file in the new tree
 
609
        :param old_kind: Old file-kind of the file
 
610
        :param new_kind: New file-kind of the file
 
611
        """
 
612
        if 'file' not in (old_kind, new_kind):
 
613
            return self.CANNOT_DIFF
 
614
        from_file_id = to_file_id = file_id
 
615
        if old_kind == 'file':
 
616
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
 
617
        elif old_kind is None:
 
618
            old_date = self.EPOCH_DATE
 
619
            from_file_id = None
 
620
        else:
 
621
            return self.CANNOT_DIFF
 
622
        if new_kind == 'file':
 
623
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
 
624
        elif new_kind is None:
 
625
            new_date = self.EPOCH_DATE
 
626
            to_file_id = None
 
627
        else:
 
628
            return self.CANNOT_DIFF
 
629
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
630
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
631
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
 
632
            old_path, new_path)
 
633
 
 
634
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
 
635
        from_path=None, to_path=None):
 
636
        """Diff the content of given files in two trees
 
637
 
 
638
        :param from_file_id: The id of the file in the from tree.  If None,
 
639
            the file is not present in the from tree.
 
640
        :param to_file_id: The id of the file in the to tree.  This may refer
 
641
            to a different file from from_file_id.  If None,
 
642
            the file is not present in the to tree.
 
643
        :param from_path: The path in the from tree or None if unknown.
 
644
        :param to_path: The path in the to tree or None if unknown.
 
645
        """
 
646
        def _get_text(tree, file_id, path):
 
647
            if file_id is not None:
 
648
                return tree.get_file(file_id, path).readlines()
 
649
            else:
 
650
                return []
 
651
        try:
 
652
            from_text = _get_text(self.old_tree, from_file_id, from_path)
 
653
            to_text = _get_text(self.new_tree, to_file_id, to_path)
 
654
            self.text_differ(from_label, from_text, to_label, to_text,
 
655
                             self.to_file)
 
656
        except errors.BinaryFile:
 
657
            self.to_file.write(
 
658
                  ("Binary files %s and %s differ\n" %
 
659
                  (from_label, to_label)).encode(self.path_encoding))
 
660
        return self.CHANGED
 
661
 
 
662
 
 
663
class DiffFromTool(DiffPath):
 
664
 
 
665
    def __init__(self, command_template, old_tree, new_tree, to_file,
 
666
                 path_encoding='utf-8'):
 
667
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
668
        self.command_template = command_template
 
669
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
 
670
 
 
671
    @classmethod
 
672
    def from_string(klass, command_string, old_tree, new_tree, to_file,
 
673
                    path_encoding='utf-8'):
 
674
        command_template = commands.shlex_split_unicode(command_string)
 
675
        command_template.extend(['%(old_path)s', '%(new_path)s'])
 
676
        return klass(command_template, old_tree, new_tree, to_file,
 
677
                     path_encoding)
 
678
 
 
679
    @classmethod
 
680
    def make_from_diff_tree(klass, command_string):
 
681
        def from_diff_tree(diff_tree):
 
682
            return klass.from_string(command_string, diff_tree.old_tree,
 
683
                                     diff_tree.new_tree, diff_tree.to_file)
 
684
        return from_diff_tree
 
685
 
 
686
    def _get_command(self, old_path, new_path):
 
687
        my_map = {'old_path': old_path, 'new_path': new_path}
 
688
        return [t % my_map for t in self.command_template]
 
689
 
 
690
    def _execute(self, old_path, new_path):
 
691
        command = self._get_command(old_path, new_path)
 
692
        try:
 
693
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
 
694
                                    cwd=self._root)
 
695
        except OSError, e:
 
696
            if e.errno == errno.ENOENT:
 
697
                raise errors.ExecutableMissing(command[0])
 
698
            else:
 
699
                raise
 
700
        self.to_file.write(proc.stdout.read())
 
701
        return proc.wait()
 
702
 
 
703
    def _try_symlink_root(self, tree, prefix):
 
704
        if (getattr(tree, 'abspath', None) is None
 
705
            or not osutils.host_os_dereferences_symlinks()):
 
706
            return False
 
707
        try:
 
708
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
 
709
        except OSError, e:
 
710
            if e.errno != errno.EEXIST:
 
711
                raise
 
712
        return True
 
713
 
 
714
    def _write_file(self, file_id, tree, prefix, relpath):
 
715
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
716
        if self._try_symlink_root(tree, prefix):
 
717
            return full_path
 
718
        parent_dir = osutils.dirname(full_path)
 
719
        try:
 
720
            os.makedirs(parent_dir)
 
721
        except OSError, e:
 
722
            if e.errno != errno.EEXIST:
 
723
                raise
 
724
        source = tree.get_file(file_id, relpath)
 
725
        try:
 
726
            target = open(full_path, 'wb')
 
727
            try:
 
728
                osutils.pumpfile(source, target)
 
729
            finally:
 
730
                target.close()
 
731
        finally:
 
732
            source.close()
 
733
        osutils.make_readonly(full_path)
 
734
        mtime = tree.get_file_mtime(file_id)
 
735
        os.utime(full_path, (mtime, mtime))
 
736
        return full_path
 
737
 
 
738
    def _prepare_files(self, file_id, old_path, new_path):
 
739
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
740
                                         old_path)
 
741
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
742
                                         new_path)
 
743
        return old_disk_path, new_disk_path
 
744
 
 
745
    def finish(self):
 
746
        try:
 
747
            osutils.rmtree(self._root)
 
748
        except OSError, e:
 
749
            if e.errno != errno.ENOENT:
 
750
                mutter("The temporary directory \"%s\" was not "
 
751
                        "cleanly removed: %s." % (self._root, e))
 
752
 
 
753
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
754
        if (old_kind, new_kind) != ('file', 'file'):
 
755
            return DiffPath.CANNOT_DIFF
 
756
        self._prepare_files(file_id, old_path, new_path)
 
757
        self._execute(osutils.pathjoin('old', old_path),
 
758
                      osutils.pathjoin('new', new_path))
 
759
 
 
760
 
 
761
class DiffTree(object):
 
762
    """Provides textual representations of the difference between two trees.
 
763
 
 
764
    A DiffTree examines two trees and where a file-id has altered
 
765
    between them, generates a textual representation of the difference.
 
766
    DiffTree uses a sequence of DiffPath objects which are each
 
767
    given the opportunity to handle a given altered fileid. The list
 
768
    of DiffPath objects can be extended globally by appending to
 
769
    DiffTree.diff_factories, or for a specific diff operation by
 
770
    supplying the extra_factories option to the appropriate method.
470
771
    """
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)
 
772
 
 
773
    # list of factories that can provide instances of DiffPath objects
 
774
    # may be extended by plugins.
 
775
    diff_factories = [DiffSymlink.from_diff_tree,
 
776
                      DiffDirectory.from_diff_tree]
 
777
 
 
778
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
779
                 diff_text=None, extra_factories=None):
 
780
        """Constructor
 
781
 
 
782
        :param old_tree: Tree to show as old in the comparison
 
783
        :param new_tree: Tree to show as new in the comparison
 
784
        :param to_file: File to write comparision to
 
785
        :param path_encoding: Character encoding to write paths in
 
786
        :param diff_text: DiffPath-type object to use as a last resort for
 
787
            diffing text files.
 
788
        :param extra_factories: Factories of DiffPaths to try before any other
 
789
            DiffPaths"""
 
790
        if diff_text is None:
 
791
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
792
                                 '', '',  internal_diff)
 
793
        self.old_tree = old_tree
 
794
        self.new_tree = new_tree
 
795
        self.to_file = to_file
 
796
        self.path_encoding = path_encoding
 
797
        self.differs = []
 
798
        if extra_factories is not None:
 
799
            self.differs.extend(f(self) for f in extra_factories)
 
800
        self.differs.extend(f(self) for f in self.diff_factories)
 
801
        self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
 
802
 
 
803
    @classmethod
 
804
    def from_trees_options(klass, old_tree, new_tree, to_file,
 
805
                           path_encoding, external_diff_options, old_label,
 
806
                           new_label, using):
 
807
        """Factory for producing a DiffTree.
 
808
 
 
809
        Designed to accept options used by show_diff_trees.
 
810
        :param old_tree: The tree to show as old in the comparison
 
811
        :param new_tree: The tree to show as new in the comparison
 
812
        :param to_file: File to write comparisons to
 
813
        :param path_encoding: Character encoding to use for writing paths
 
814
        :param external_diff_options: If supplied, use the installed diff
 
815
            binary to perform file comparison, using supplied options.
 
816
        :param old_label: Prefix to use for old file labels
 
817
        :param new_label: Prefix to use for new file labels
 
818
        :param using: Commandline to use to invoke an external diff tool
 
819
        """
 
820
        if using is not None:
 
821
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
 
822
        else:
 
823
            extra_factories = []
 
824
        if external_diff_options:
 
825
            opts = external_diff_options.split()
 
826
            def diff_file(olab, olines, nlab, nlines, to_file):
 
827
                external_diff(olab, olines, nlab, nlines, to_file, opts)
 
828
        else:
 
829
            diff_file = internal_diff
 
830
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
831
                             old_label, new_label, diff_file)
 
832
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
 
833
                     extra_factories)
 
834
 
 
835
    def show_diff(self, specific_files, extra_trees=None):
 
836
        """Write tree diff to self.to_file
 
837
 
 
838
        :param sepecific_files: the specific files to compare (recursive)
 
839
        :param extra_trees: extra trees to use for mapping paths to file_ids
 
840
        """
 
841
        try:
 
842
            return self._show_diff(specific_files, extra_trees)
 
843
        finally:
 
844
            for differ in self.differs:
 
845
                differ.finish()
 
846
 
 
847
    def _show_diff(self, specific_files, extra_trees):
 
848
        # TODO: Generation of pseudo-diffs for added/deleted files could
 
849
        # be usefully made into a much faster special case.
 
850
        iterator = self.new_tree.iter_changes(self.old_tree,
 
851
                                               specific_files=specific_files,
 
852
                                               extra_trees=extra_trees,
 
853
                                               require_versioned=True)
 
854
        has_changes = 0
 
855
        def changes_key(change):
 
856
            old_path, new_path = change[1]
 
857
            path = new_path
 
858
            if path is None:
 
859
                path = old_path
 
860
            return path
 
861
        def get_encoded_path(path):
 
862
            if path is not None:
 
863
                return path.encode(self.path_encoding, "replace")
 
864
        for (file_id, paths, changed_content, versioned, parent, name, kind,
 
865
             executable) in sorted(iterator, key=changes_key):
 
866
            # The root does not get diffed, and items with no known kind (that
 
867
            # is, missing) in both trees are skipped as well.
 
868
            if parent == (None, None) or kind == (None, None):
 
869
                continue
 
870
            oldpath, newpath = paths
 
871
            oldpath_encoded = get_encoded_path(paths[0])
 
872
            newpath_encoded = get_encoded_path(paths[1])
 
873
            old_present = (kind[0] is not None and versioned[0])
 
874
            new_present = (kind[1] is not None and versioned[1])
 
875
            renamed = (parent[0], name[0]) != (parent[1], name[1])
 
876
 
 
877
            properties_changed = []
 
878
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
 
879
 
 
880
            if properties_changed:
 
881
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
 
882
            else:
 
883
                prop_str = ""
 
884
 
 
885
            if (old_present, new_present) == (True, False):
 
886
                self.to_file.write("=== removed %s '%s'\n" %
 
887
                                   (kind[0], oldpath_encoded))
 
888
                newpath = oldpath
 
889
            elif (old_present, new_present) == (False, True):
 
890
                self.to_file.write("=== added %s '%s'\n" %
 
891
                                   (kind[1], newpath_encoded))
 
892
                oldpath = newpath
 
893
            elif renamed:
 
894
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
895
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
 
896
            else:
 
897
                # if it was produced by iter_changes, it must be
 
898
                # modified *somehow*, either content or execute bit.
 
899
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
900
                                   newpath_encoded, prop_str))
 
901
            if changed_content:
 
902
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
 
903
                has_changes = 1
 
904
            if renamed:
 
905
                has_changes = 1
 
906
        return has_changes
 
907
 
 
908
    def diff(self, file_id, old_path, new_path):
 
909
        """Perform a diff of a single file
 
910
 
 
911
        :param file_id: file-id of the file
 
912
        :param old_path: The path of the file in the old tree
 
913
        :param new_path: The path of the file in the new tree
 
914
        """
 
915
        try:
 
916
            old_kind = self.old_tree.kind(file_id)
 
917
        except (errors.NoSuchId, errors.NoSuchFile):
 
918
            old_kind = None
 
919
        try:
 
920
            new_kind = self.new_tree.kind(file_id)
 
921
        except (errors.NoSuchId, errors.NoSuchFile):
 
922
            new_kind = None
 
923
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
 
924
 
 
925
 
 
926
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
927
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
928
                                       new_path, old_kind, new_kind)
 
929
        if result is DiffPath.CANNOT_DIFF:
 
930
            error_path = new_path
 
931
            if error_path is None:
 
932
                error_path = old_path
 
933
            raise errors.NoDiffFound(error_path)