/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: Neil Martinsen-Burrell
  • Date: 2009-12-07 21:51:01 UTC
  • mto: This revision was merged to the branch mainline in revision 4910.
  • Revision ID: nmb@wartburg.edu-20091207215101-lv1fyi2blzer4h5j
tweaks based on JAMs review

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 string
 
22
import sys
 
23
 
 
24
from bzrlib.lazy_import import lazy_import
 
25
lazy_import(globals(), """
 
26
import errno
20
27
import subprocess
21
 
import sys
22
28
import tempfile
23
29
import time
24
30
 
25
31
from bzrlib import (
 
32
    branch as _mod_branch,
 
33
    bzrdir,
 
34
    commands,
26
35
    errors,
27
36
    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
 
37
    patiencediff,
 
38
    textfile,
 
39
    timestamp,
 
40
    views,
 
41
    )
 
42
 
 
43
from bzrlib.workingtree import WorkingTree
 
44
""")
 
45
 
 
46
from bzrlib.symbol_versioning import (
 
47
    deprecated_function,
 
48
    )
 
49
from bzrlib.trace import mutter, note, warning
 
50
 
 
51
 
 
52
class AtTemplate(string.Template):
 
53
    """Templating class that uses @ instead of $."""
 
54
 
 
55
    delimiter = '@'
39
56
 
40
57
 
41
58
# TODO: Rather than building a changeset object, we should probably
42
59
# invoke callbacks on an object.  That object can either accumulate a
43
60
# list, write them out directly, etc etc.
44
61
 
 
62
 
 
63
class _PrematchedMatcher(difflib.SequenceMatcher):
 
64
    """Allow SequenceMatcher operations to use predetermined blocks"""
 
65
 
 
66
    def __init__(self, matching_blocks):
 
67
        difflib.SequenceMatcher(self, None, None)
 
68
        self.matching_blocks = matching_blocks
 
69
        self.opcodes = None
 
70
 
 
71
 
45
72
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
46
73
                  allow_binary=False, sequence_matcher=None,
47
74
                  path_encoding='utf8'):
60
87
    # both sequences are empty.
61
88
    if not oldlines and not newlines:
62
89
        return
63
 
    
 
90
 
64
91
    if allow_binary is False:
65
 
        check_text_lines(oldlines)
66
 
        check_text_lines(newlines)
 
92
        textfile.check_text_lines(oldlines)
 
93
        textfile.check_text_lines(newlines)
67
94
 
68
95
    if sequence_matcher is None:
69
 
        sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
70
 
    ud = unified_diff(oldlines, newlines,
 
96
        sequence_matcher = patiencediff.PatienceSequenceMatcher
 
97
    ud = patiencediff.unified_diff(oldlines, newlines,
71
98
                      fromfile=old_filename.encode(path_encoding),
72
99
                      tofile=new_filename.encode(path_encoding),
73
100
                      sequencematcher=sequence_matcher)
74
101
 
75
102
    ud = list(ud)
 
103
    if len(ud) == 0: # Identical contents, nothing to do
 
104
        return
76
105
    # work-around for difflib being too smart for its own good
77
106
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
78
107
    if not oldlines:
79
108
        ud[2] = ud[2].replace('-1,0', '-0,0')
80
109
    elif not newlines:
81
110
        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
111
 
86
112
    for line in ud:
87
113
        to_file.write(line)
88
114
        if not line.endswith('\n'):
89
115
            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)
 
116
    to_file.write('\n')
99
117
 
100
118
 
101
119
def _spawn_external_diff(diffcmd, capture_errors=True):
102
120
    """Spawn the externall diff process, and return the child handle.
103
121
 
104
122
    :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.
 
123
    :param capture_errors: Capture stderr as well as setting LANG=C
 
124
        and LC_ALL=C. This lets us read and understand the output of diff,
 
125
        and respond to any errors.
108
126
    :return: A Popen object.
109
127
    """
110
128
    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
 
129
        # construct minimal environment
 
130
        env = {}
 
131
        path = os.environ.get('PATH')
 
132
        if path is not None:
 
133
            env['PATH'] = path
 
134
        env['LANGUAGE'] = 'C'   # on win32 only LANGUAGE has effect
 
135
        env['LANG'] = 'C'
 
136
        env['LC_ALL'] = 'C'
117
137
        stderr = subprocess.PIPE
118
138
    else:
119
 
        preexec_fn = None
 
139
        env = None
120
140
        stderr = None
121
141
 
122
142
    try:
124
144
                                stdin=subprocess.PIPE,
125
145
                                stdout=subprocess.PIPE,
126
146
                                stderr=stderr,
127
 
                                preexec_fn=preexec_fn)
 
147
                                env=env)
128
148
    except OSError, e:
129
149
        if e.errno == errno.ENOENT:
130
150
            raise errors.NoDiff(str(e))
160
180
 
161
181
        if not diff_opts:
162
182
            diff_opts = []
 
183
        if sys.platform == 'win32':
 
184
            # Popen doesn't do the proper encoding for external commands
 
185
            # Since we are dealing with an ANSI api, use mbcs encoding
 
186
            old_filename = old_filename.encode('mbcs')
 
187
            new_filename = new_filename.encode('mbcs')
163
188
        diffcmd = ['diff',
164
189
                   '--label', old_filename,
165
190
                   old_abspath,
188
213
            break
189
214
        else:
190
215
            diffcmd.append('-u')
191
 
                  
 
216
 
192
217
        if diff_opts:
193
218
            diffcmd.extend(diff_opts)
194
219
 
195
220
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
196
221
        out,err = pipe.communicate()
197
222
        rc = pipe.returncode
198
 
        
 
223
 
199
224
        # internal_diff() adds a trailing newline, add one here for consistency
200
225
        out += '\n'
201
226
        if rc == 2:
212
237
            # Write out the new i18n diff response
213
238
            to_file.write(out+'\n')
214
239
            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,))
 
240
                raise errors.BzrError(
 
241
                               'external diff failed with exit code 2'
 
242
                               ' when run with LANG=C and LC_ALL=C,'
 
243
                               ' but not when run natively: %r' % (diffcmd,))
218
244
 
219
245
            first_line = lang_c_out.split('\n', 1)[0]
220
246
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
221
247
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
222
248
            if m is None:
223
 
                raise BzrError('external diff failed with exit code 2;'
224
 
                               ' command: %r' % (diffcmd,))
 
249
                raise errors.BzrError('external diff failed with exit code 2;'
 
250
                                      ' command: %r' % (diffcmd,))
225
251
            else:
226
252
                # Binary files differ, just return
227
253
                return
235
261
                msg = 'signal %d' % (-rc)
236
262
            else:
237
263
                msg = 'exit code %d' % rc
238
 
                
239
 
            raise BzrError('external diff failed with %s; command: %r' 
240
 
                           % (rc, diffcmd))
 
264
 
 
265
            raise errors.BzrError('external diff failed with %s; command: %r'
 
266
                                  % (rc, diffcmd))
241
267
 
242
268
 
243
269
    finally:
260
286
                        new_abspath, e)
261
287
 
262
288
 
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)
 
289
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
 
290
                                   apply_view=True):
 
291
    """Get the trees and specific files to diff given a list of paths.
 
292
 
 
293
    This method works out the trees to be diff'ed and the files of
 
294
    interest within those trees.
 
295
 
 
296
    :param path_list:
 
297
        the list of arguments passed to the diff command
 
298
    :param revision_specs:
 
299
        Zero, one or two RevisionSpecs from the diff command line,
 
300
        saying what revisions to compare.
 
301
    :param old_url:
 
302
        The url of the old branch or tree. If None, the tree to use is
 
303
        taken from the first path, if any, or the current working tree.
 
304
    :param new_url:
 
305
        The url of the new branch or tree. If None, the tree to use is
 
306
        taken from the first path, if any, or the current working tree.
 
307
    :param apply_view:
 
308
        if True and a view is set, apply the view or check that the paths
 
309
        are within it
 
310
    :returns:
 
311
        a tuple of (old_tree, new_tree, old_branch, new_branch,
 
312
        specific_files, extra_trees) where extra_trees is a sequence of
 
313
        additional trees to search in for file-ids.
 
314
    """
 
315
    # Get the old and new revision specs
 
316
    old_revision_spec = None
 
317
    new_revision_spec = None
 
318
    if revision_specs is not None:
 
319
        if len(revision_specs) > 0:
 
320
            old_revision_spec = revision_specs[0]
 
321
            if old_url is None:
 
322
                old_url = old_revision_spec.get_branch()
 
323
        if len(revision_specs) > 1:
 
324
            new_revision_spec = revision_specs[1]
 
325
            if new_url is None:
 
326
                new_url = new_revision_spec.get_branch()
 
327
 
 
328
    other_paths = []
 
329
    make_paths_wt_relative = True
 
330
    consider_relpath = True
 
331
    if path_list is None or len(path_list) == 0:
 
332
        # If no path is given, the current working tree is used
 
333
        default_location = u'.'
 
334
        consider_relpath = False
 
335
    elif old_url is not None and new_url is not None:
 
336
        other_paths = path_list
 
337
        make_paths_wt_relative = False
 
338
    else:
 
339
        default_location = path_list[0]
 
340
        other_paths = path_list[1:]
 
341
 
 
342
    # Get the old location
 
343
    specific_files = []
 
344
    if old_url is None:
 
345
        old_url = default_location
 
346
    working_tree, branch, relpath = \
 
347
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
 
348
    if consider_relpath and relpath != '':
 
349
        if working_tree is not None and apply_view:
 
350
            views.check_path_in_view(working_tree, relpath)
 
351
        specific_files.append(relpath)
 
352
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
 
353
    old_branch = branch
 
354
 
 
355
    # Get the new location
 
356
    if new_url is None:
 
357
        new_url = default_location
 
358
    if new_url != old_url:
 
359
        working_tree, branch, relpath = \
 
360
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
 
361
        if consider_relpath and relpath != '':
 
362
            if working_tree is not None and apply_view:
 
363
                views.check_path_in_view(working_tree, relpath)
 
364
            specific_files.append(relpath)
 
365
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
 
366
        basis_is_default=working_tree is None)
 
367
    new_branch = branch
 
368
 
 
369
    # Get the specific files (all files is None, no files is [])
 
370
    if make_paths_wt_relative and working_tree is not None:
 
371
        try:
 
372
            from bzrlib.builtins import safe_relpath_files
 
373
            other_paths = safe_relpath_files(working_tree, other_paths,
 
374
            apply_view=apply_view)
 
375
        except errors.FileInWrongBranch:
 
376
            raise errors.BzrCommandError("Files are in different branches")
 
377
    specific_files.extend(other_paths)
 
378
    if len(specific_files) == 0:
 
379
        specific_files = None
 
380
        if (working_tree is not None and working_tree.supports_views()
 
381
            and apply_view):
 
382
            view_files = working_tree.views.lookup_view()
 
383
            if view_files:
 
384
                specific_files = view_files
 
385
                view_str = views.view_display_str(view_files)
 
386
                note("*** Ignoring files outside view. View is %s" % view_str)
 
387
 
 
388
    # Get extra trees that ought to be searched for file-ids
 
389
    extra_trees = None
 
390
    if working_tree is not None and working_tree not in (old_tree, new_tree):
 
391
        extra_trees = (working_tree,)
 
392
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
 
393
 
 
394
 
 
395
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
 
396
    if branch is None and tree is not None:
 
397
        branch = tree.branch
 
398
    if spec is None or spec.spec is None:
 
399
        if basis_is_default:
 
400
            if tree is not None:
 
401
                return tree.basis_tree()
 
402
            else:
 
403
                return branch.basis_tree()
 
404
        else:
 
405
            return tree
 
406
    return spec.as_tree(branch)
352
407
 
353
408
 
354
409
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
355
410
                    external_diff_options=None,
356
411
                    old_label='a/', new_label='b/',
357
 
                    extra_trees=None):
 
412
                    extra_trees=None,
 
413
                    path_encoding='utf8',
 
414
                    using=None):
358
415
    """Show in text form the changes from one tree to another.
359
416
 
360
 
    to_files
361
 
        If set, include only changes to these files.
 
417
    to_file
 
418
        The output stream.
 
419
 
 
420
    specific_files
 
421
        Include only changes to these files - None for all changes.
362
422
 
363
423
    external_diff_options
364
424
        If set, use an external GNU diff and pass these options.
365
425
 
366
426
    extra_trees
367
427
        If set, more Trees to use for looking up file ids
 
428
 
 
429
    path_encoding
 
430
        If set, the path will be encoded as specified, otherwise is supposed
 
431
        to be utf8
368
432
    """
369
433
    old_tree.lock_read()
370
434
    try:
 
435
        if extra_trees is not None:
 
436
            for tree in extra_trees:
 
437
                tree.lock_read()
371
438
        new_tree.lock_read()
372
439
        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)
 
440
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
441
                                                 path_encoding,
 
442
                                                 external_diff_options,
 
443
                                                 old_label, new_label, using)
 
444
            return differ.show_diff(specific_files, extra_trees)
377
445
        finally:
378
446
            new_tree.unlock()
 
447
            if extra_trees is not None:
 
448
                for tree in extra_trees:
 
449
                    tree.unlock()
379
450
    finally:
380
451
        old_tree.unlock()
381
452
 
382
453
 
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):
 
454
def _patch_header_date(tree, file_id, path):
 
455
    """Returns a timestamp suitable for use in a patch header."""
 
456
    mtime = tree.get_file_mtime(file_id, path)
 
457
    return timestamp.format_patch_date(mtime)
 
458
 
 
459
 
 
460
def get_executable_change(old_is_x, new_is_x):
 
461
    descr = { True:"+x", False:"-x", None:"??" }
 
462
    if old_is_x != new_is_x:
 
463
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
464
    else:
 
465
        return []
 
466
 
 
467
 
 
468
class DiffPath(object):
 
469
    """Base type for command object that compare files"""
 
470
 
 
471
    # The type or contents of the file were unsuitable for diffing
 
472
    CANNOT_DIFF = 'CANNOT_DIFF'
 
473
    # The file has changed in a semantic way
 
474
    CHANGED = 'CHANGED'
 
475
    # The file content may have changed, but there is no semantic change
 
476
    UNCHANGED = 'UNCHANGED'
 
477
 
 
478
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
 
479
        """Constructor.
 
480
 
 
481
        :param old_tree: The tree to show as the old tree in the comparison
 
482
        :param new_tree: The tree to show as new in the comparison
 
483
        :param to_file: The file to write comparison data to
 
484
        :param path_encoding: The character encoding to write paths in
 
485
        """
 
486
        self.old_tree = old_tree
 
487
        self.new_tree = new_tree
 
488
        self.to_file = to_file
 
489
        self.path_encoding = path_encoding
 
490
 
 
491
    def finish(self):
 
492
        pass
 
493
 
 
494
    @classmethod
 
495
    def from_diff_tree(klass, diff_tree):
 
496
        return klass(diff_tree.old_tree, diff_tree.new_tree,
 
497
                     diff_tree.to_file, diff_tree.path_encoding)
 
498
 
 
499
    @staticmethod
 
500
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
 
501
        for file_differ in differs:
 
502
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
503
                                      new_kind)
 
504
            if result is not DiffPath.CANNOT_DIFF:
 
505
                return result
 
506
        else:
 
507
            return DiffPath.CANNOT_DIFF
 
508
 
 
509
 
 
510
class DiffKindChange(object):
 
511
    """Special differ for file kind changes.
 
512
 
 
513
    Represents kind change as deletion + creation.  Uses the other differs
 
514
    to do this.
 
515
    """
 
516
    def __init__(self, differs):
 
517
        self.differs = differs
 
518
 
 
519
    def finish(self):
 
520
        pass
 
521
 
 
522
    @classmethod
 
523
    def from_diff_tree(klass, diff_tree):
 
524
        return klass(diff_tree.differs)
 
525
 
 
526
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
527
        """Perform comparison
 
528
 
 
529
        :param file_id: The file_id of the file to compare
 
530
        :param old_path: Path of the file in the old tree
 
531
        :param new_path: Path of the file in the new tree
 
532
        :param old_kind: Old file-kind of the file
 
533
        :param new_kind: New file-kind of the file
 
534
        """
 
535
        if None in (old_kind, new_kind):
 
536
            return DiffPath.CANNOT_DIFF
 
537
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
538
                                       new_path, old_kind, None)
 
539
        if result is DiffPath.CANNOT_DIFF:
 
540
            return result
 
541
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
542
                                     None, new_kind)
 
543
 
 
544
 
 
545
class DiffDirectory(DiffPath):
 
546
 
 
547
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
548
        """Perform comparison between two directories.  (dummy)
 
549
 
 
550
        """
 
551
        if 'directory' not in (old_kind, new_kind):
 
552
            return self.CANNOT_DIFF
 
553
        if old_kind not in ('directory', None):
 
554
            return self.CANNOT_DIFF
 
555
        if new_kind not in ('directory', None):
 
556
            return self.CANNOT_DIFF
 
557
        return self.CHANGED
 
558
 
 
559
 
 
560
class DiffSymlink(DiffPath):
 
561
 
 
562
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
563
        """Perform comparison between two symlinks
 
564
 
 
565
        :param file_id: The file_id of the file to compare
 
566
        :param old_path: Path of the file in the old tree
 
567
        :param new_path: Path of the file in the new tree
 
568
        :param old_kind: Old file-kind of the file
 
569
        :param new_kind: New file-kind of the file
 
570
        """
 
571
        if 'symlink' not in (old_kind, new_kind):
 
572
            return self.CANNOT_DIFF
 
573
        if old_kind == 'symlink':
 
574
            old_target = self.old_tree.get_symlink_target(file_id)
 
575
        elif old_kind is None:
 
576
            old_target = None
 
577
        else:
 
578
            return self.CANNOT_DIFF
 
579
        if new_kind == 'symlink':
 
580
            new_target = self.new_tree.get_symlink_target(file_id)
 
581
        elif new_kind is None:
 
582
            new_target = None
 
583
        else:
 
584
            return self.CANNOT_DIFF
 
585
        return self.diff_symlink(old_target, new_target)
 
586
 
 
587
    def diff_symlink(self, old_target, new_target):
 
588
        if old_target is None:
 
589
            self.to_file.write('=== target is %r\n' % new_target)
 
590
        elif new_target is None:
 
591
            self.to_file.write('=== target was %r\n' % old_target)
 
592
        else:
 
593
            self.to_file.write('=== target changed %r => %r\n' %
 
594
                              (old_target, new_target))
 
595
        return self.CHANGED
 
596
 
 
597
 
 
598
class DiffText(DiffPath):
386
599
 
387
600
    # GNU Patch uses the epoch date to detect files that are being added
388
601
    # or removed in a diff.
389
602
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
390
603
 
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.
 
604
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
605
                 old_label='', new_label='', text_differ=internal_diff):
 
606
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
607
        self.text_differ = text_differ
 
608
        self.old_label = old_label
 
609
        self.new_label = new_label
 
610
        self.path_encoding = path_encoding
 
611
 
 
612
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
613
        """Compare two files in unified diff format
 
614
 
 
615
        :param file_id: The file_id of the file to compare
 
616
        :param old_path: Path of the file in the old tree
 
617
        :param new_path: Path of the file in the new tree
 
618
        :param old_kind: Old file-kind of the file
 
619
        :param new_kind: New file-kind of the file
 
620
        """
 
621
        if 'file' not in (old_kind, new_kind):
 
622
            return self.CANNOT_DIFF
 
623
        from_file_id = to_file_id = file_id
 
624
        if old_kind == 'file':
 
625
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
 
626
        elif old_kind is None:
 
627
            old_date = self.EPOCH_DATE
 
628
            from_file_id = None
 
629
        else:
 
630
            return self.CANNOT_DIFF
 
631
        if new_kind == 'file':
 
632
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
 
633
        elif new_kind is None:
 
634
            new_date = self.EPOCH_DATE
 
635
            to_file_id = None
 
636
        else:
 
637
            return self.CANNOT_DIFF
 
638
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
639
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
640
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
 
641
            old_path, new_path)
 
642
 
 
643
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
 
644
        from_path=None, to_path=None):
 
645
        """Diff the content of given files in two trees
 
646
 
 
647
        :param from_file_id: The id of the file in the from tree.  If None,
 
648
            the file is not present in the from tree.
 
649
        :param to_file_id: The id of the file in the to tree.  This may refer
 
650
            to a different file from from_file_id.  If None,
 
651
            the file is not present in the to tree.
 
652
        :param from_path: The path in the from tree or None if unknown.
 
653
        :param to_path: The path in the to tree or None if unknown.
 
654
        """
 
655
        def _get_text(tree, file_id, path):
 
656
            if file_id is not None:
 
657
                return tree.get_file(file_id, path).readlines()
 
658
            else:
 
659
                return []
 
660
        try:
 
661
            from_text = _get_text(self.old_tree, from_file_id, from_path)
 
662
            to_text = _get_text(self.new_tree, to_file_id, to_path)
 
663
            self.text_differ(from_label, from_text, to_label, to_text,
 
664
                             self.to_file)
 
665
        except errors.BinaryFile:
 
666
            self.to_file.write(
 
667
                  ("Binary files %s and %s differ\n" %
 
668
                  (from_label, to_label)).encode(self.path_encoding))
 
669
        return self.CHANGED
 
670
 
 
671
 
 
672
class DiffFromTool(DiffPath):
 
673
 
 
674
    def __init__(self, command_template, old_tree, new_tree, to_file,
 
675
                 path_encoding='utf-8'):
 
676
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
677
        self.command_template = command_template
 
678
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
 
679
 
 
680
    @classmethod
 
681
    def from_string(klass, command_string, old_tree, new_tree, to_file,
 
682
                    path_encoding='utf-8'):
 
683
        command_template = commands.shlex_split_unicode(command_string)
 
684
        if '@' not in command_string:
 
685
            command_template.extend(['@old_path', '@new_path'])
 
686
        return klass(command_template, old_tree, new_tree, to_file,
 
687
                     path_encoding)
 
688
 
 
689
    @classmethod
 
690
    def make_from_diff_tree(klass, command_string):
 
691
        def from_diff_tree(diff_tree):
 
692
            return klass.from_string(command_string, diff_tree.old_tree,
 
693
                                     diff_tree.new_tree, diff_tree.to_file)
 
694
        return from_diff_tree
 
695
 
 
696
    def _get_command(self, old_path, new_path):
 
697
        my_map = {'old_path': old_path, 'new_path': new_path}
 
698
        return [AtTemplate(t).substitute(my_map) for t in
 
699
                self.command_template]
 
700
 
 
701
    def _execute(self, old_path, new_path):
 
702
        command = self._get_command(old_path, new_path)
 
703
        try:
 
704
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
 
705
                                    cwd=self._root)
 
706
        except OSError, e:
 
707
            if e.errno == errno.ENOENT:
 
708
                raise errors.ExecutableMissing(command[0])
 
709
            else:
 
710
                raise
 
711
        self.to_file.write(proc.stdout.read())
 
712
        return proc.wait()
 
713
 
 
714
    def _try_symlink_root(self, tree, prefix):
 
715
        if (getattr(tree, 'abspath', None) is None
 
716
            or not osutils.host_os_dereferences_symlinks()):
 
717
            return False
 
718
        try:
 
719
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
 
720
        except OSError, e:
 
721
            if e.errno != errno.EEXIST:
 
722
                raise
 
723
        return True
 
724
 
 
725
    def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
 
726
                    allow_write=False):
 
727
        if not force_temp and isinstance(tree, WorkingTree):
 
728
            return tree.abspath(tree.id2path(file_id))
 
729
        
 
730
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
731
        if not force_temp and self._try_symlink_root(tree, prefix):
 
732
            return full_path
 
733
        parent_dir = osutils.dirname(full_path)
 
734
        try:
 
735
            os.makedirs(parent_dir)
 
736
        except OSError, e:
 
737
            if e.errno != errno.EEXIST:
 
738
                raise
 
739
        source = tree.get_file(file_id, relpath)
 
740
        try:
 
741
            target = open(full_path, 'wb')
 
742
            try:
 
743
                osutils.pumpfile(source, target)
 
744
            finally:
 
745
                target.close()
 
746
        finally:
 
747
            source.close()
 
748
        if not allow_write:
 
749
            osutils.make_readonly(full_path)
 
750
        mtime = tree.get_file_mtime(file_id)
 
751
        os.utime(full_path, (mtime, mtime))
 
752
        return full_path
 
753
 
 
754
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
 
755
                       allow_write_new=False):
 
756
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
757
                                         old_path, force_temp)
 
758
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
759
                                         new_path, force_temp,
 
760
                                         allow_write=allow_write_new)
 
761
        return old_disk_path, new_disk_path
 
762
 
 
763
    def finish(self):
 
764
        try:
 
765
            osutils.rmtree(self._root)
 
766
        except OSError, e:
 
767
            if e.errno != errno.ENOENT:
 
768
                mutter("The temporary directory \"%s\" was not "
 
769
                        "cleanly removed: %s." % (self._root, e))
 
770
 
 
771
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
772
        if (old_kind, new_kind) != ('file', 'file'):
 
773
            return DiffPath.CANNOT_DIFF
 
774
        (old_disk_path, new_disk_path) = self._prepare_files(
 
775
                                                file_id, old_path, new_path)
 
776
        self._execute(old_disk_path, new_disk_path)
 
777
 
 
778
    def edit_file(self, file_id):
 
779
        """Use this tool to edit a file.
 
780
 
 
781
        A temporary copy will be edited, and the new contents will be
 
782
        returned.
 
783
 
 
784
        :param file_id: The id of the file to edit.
 
785
        :return: The new contents of the file.
 
786
        """
 
787
        old_path = self.old_tree.id2path(file_id)
 
788
        new_path = self.new_tree.id2path(file_id)
 
789
        new_abs_path = self._prepare_files(file_id, old_path, new_path,
 
790
                                           allow_write_new=True,
 
791
                                           force_temp=True)[1]
 
792
        command = self._get_command(osutils.pathjoin('old', old_path),
 
793
                                    osutils.pathjoin('new', new_path))
 
794
        subprocess.call(command, cwd=self._root)
 
795
        new_file = open(new_abs_path, 'r')
 
796
        try:
 
797
            return new_file.read()
 
798
        finally:
 
799
            new_file.close()
 
800
 
 
801
 
 
802
class DiffTree(object):
 
803
    """Provides textual representations of the difference between two trees.
 
804
 
 
805
    A DiffTree examines two trees and where a file-id has altered
 
806
    between them, generates a textual representation of the difference.
 
807
    DiffTree uses a sequence of DiffPath objects which are each
 
808
    given the opportunity to handle a given altered fileid. The list
 
809
    of DiffPath objects can be extended globally by appending to
 
810
    DiffTree.diff_factories, or for a specific diff operation by
 
811
    supplying the extra_factories option to the appropriate method.
470
812
    """
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)
 
813
 
 
814
    # list of factories that can provide instances of DiffPath objects
 
815
    # may be extended by plugins.
 
816
    diff_factories = [DiffSymlink.from_diff_tree,
 
817
                      DiffDirectory.from_diff_tree]
 
818
 
 
819
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
820
                 diff_text=None, extra_factories=None):
 
821
        """Constructor
 
822
 
 
823
        :param old_tree: Tree to show as old in the comparison
 
824
        :param new_tree: Tree to show as new in the comparison
 
825
        :param to_file: File to write comparision to
 
826
        :param path_encoding: Character encoding to write paths in
 
827
        :param diff_text: DiffPath-type object to use as a last resort for
 
828
            diffing text files.
 
829
        :param extra_factories: Factories of DiffPaths to try before any other
 
830
            DiffPaths"""
 
831
        if diff_text is None:
 
832
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
833
                                 '', '',  internal_diff)
 
834
        self.old_tree = old_tree
 
835
        self.new_tree = new_tree
 
836
        self.to_file = to_file
 
837
        self.path_encoding = path_encoding
 
838
        self.differs = []
 
839
        if extra_factories is not None:
 
840
            self.differs.extend(f(self) for f in extra_factories)
 
841
        self.differs.extend(f(self) for f in self.diff_factories)
 
842
        self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
 
843
 
 
844
    @classmethod
 
845
    def from_trees_options(klass, old_tree, new_tree, to_file,
 
846
                           path_encoding, external_diff_options, old_label,
 
847
                           new_label, using):
 
848
        """Factory for producing a DiffTree.
 
849
 
 
850
        Designed to accept options used by show_diff_trees.
 
851
        :param old_tree: The tree to show as old in the comparison
 
852
        :param new_tree: The tree to show as new in the comparison
 
853
        :param to_file: File to write comparisons to
 
854
        :param path_encoding: Character encoding to use for writing paths
 
855
        :param external_diff_options: If supplied, use the installed diff
 
856
            binary to perform file comparison, using supplied options.
 
857
        :param old_label: Prefix to use for old file labels
 
858
        :param new_label: Prefix to use for new file labels
 
859
        :param using: Commandline to use to invoke an external diff tool
 
860
        """
 
861
        if using is not None:
 
862
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
 
863
        else:
 
864
            extra_factories = []
 
865
        if external_diff_options:
 
866
            opts = external_diff_options.split()
 
867
            def diff_file(olab, olines, nlab, nlines, to_file):
 
868
                external_diff(olab, olines, nlab, nlines, to_file, opts)
 
869
        else:
 
870
            diff_file = internal_diff
 
871
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
872
                             old_label, new_label, diff_file)
 
873
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
 
874
                     extra_factories)
 
875
 
 
876
    def show_diff(self, specific_files, extra_trees=None):
 
877
        """Write tree diff to self.to_file
 
878
 
 
879
        :param sepecific_files: the specific files to compare (recursive)
 
880
        :param extra_trees: extra trees to use for mapping paths to file_ids
 
881
        """
 
882
        try:
 
883
            return self._show_diff(specific_files, extra_trees)
 
884
        finally:
 
885
            for differ in self.differs:
 
886
                differ.finish()
 
887
 
 
888
    def _show_diff(self, specific_files, extra_trees):
 
889
        # TODO: Generation of pseudo-diffs for added/deleted files could
 
890
        # be usefully made into a much faster special case.
 
891
        iterator = self.new_tree.iter_changes(self.old_tree,
 
892
                                               specific_files=specific_files,
 
893
                                               extra_trees=extra_trees,
 
894
                                               require_versioned=True)
 
895
        has_changes = 0
 
896
        def changes_key(change):
 
897
            old_path, new_path = change[1]
 
898
            path = new_path
 
899
            if path is None:
 
900
                path = old_path
 
901
            return path
 
902
        def get_encoded_path(path):
 
903
            if path is not None:
 
904
                return path.encode(self.path_encoding, "replace")
 
905
        for (file_id, paths, changed_content, versioned, parent, name, kind,
 
906
             executable) in sorted(iterator, key=changes_key):
 
907
            # The root does not get diffed, and items with no known kind (that
 
908
            # is, missing) in both trees are skipped as well.
 
909
            if parent == (None, None) or kind == (None, None):
 
910
                continue
 
911
            oldpath, newpath = paths
 
912
            oldpath_encoded = get_encoded_path(paths[0])
 
913
            newpath_encoded = get_encoded_path(paths[1])
 
914
            old_present = (kind[0] is not None and versioned[0])
 
915
            new_present = (kind[1] is not None and versioned[1])
 
916
            renamed = (parent[0], name[0]) != (parent[1], name[1])
 
917
 
 
918
            properties_changed = []
 
919
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
 
920
 
 
921
            if properties_changed:
 
922
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
 
923
            else:
 
924
                prop_str = ""
 
925
 
 
926
            if (old_present, new_present) == (True, False):
 
927
                self.to_file.write("=== removed %s '%s'\n" %
 
928
                                   (kind[0], oldpath_encoded))
 
929
                newpath = oldpath
 
930
            elif (old_present, new_present) == (False, True):
 
931
                self.to_file.write("=== added %s '%s'\n" %
 
932
                                   (kind[1], newpath_encoded))
 
933
                oldpath = newpath
 
934
            elif renamed:
 
935
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
936
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
 
937
            else:
 
938
                # if it was produced by iter_changes, it must be
 
939
                # modified *somehow*, either content or execute bit.
 
940
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
941
                                   newpath_encoded, prop_str))
 
942
            if changed_content:
 
943
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
 
944
                has_changes = 1
 
945
            if renamed:
 
946
                has_changes = 1
 
947
        return has_changes
 
948
 
 
949
    def diff(self, file_id, old_path, new_path):
 
950
        """Perform a diff of a single file
 
951
 
 
952
        :param file_id: file-id of the file
 
953
        :param old_path: The path of the file in the old tree
 
954
        :param new_path: The path of the file in the new tree
 
955
        """
 
956
        try:
 
957
            old_kind = self.old_tree.kind(file_id)
 
958
        except (errors.NoSuchId, errors.NoSuchFile):
 
959
            old_kind = None
 
960
        try:
 
961
            new_kind = self.new_tree.kind(file_id)
 
962
        except (errors.NoSuchId, errors.NoSuchFile):
 
963
            new_kind = None
 
964
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
 
965
 
 
966
 
 
967
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
968
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
969
                                       new_path, old_kind, new_kind)
 
970
        if result is DiffPath.CANNOT_DIFF:
 
971
            error_path = new_path
 
972
            if error_path is None:
 
973
                error_path = old_path
 
974
            raise errors.NoDiffFound(error_path)