~ubuntu-branches/debian/lenny/bzr/lenny

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Bazaar Package Importer
  • Author(s): Thomas Viehmann
  • Date: 2008-08-22 20:06:37 UTC
  • mfrom: (3.1.63 intrepid)
  • Revision ID: james.westby@ubuntu.com-20080822200637-kxobfsnjlzojhqra
Tags: 1.5-1.1
* Non-maintainer upload.
* Apply patch from upstream VCS to fix FTBFS in tools/rst2html.py
  with older docutils. Thanks to Olivier Tétard for digging it
  up.
  Closes: #494246.

Show diffs side-by-side

added added

removed removed

Lines of Context:
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  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,
 
36
    patiencediff,
 
37
    textfile,
 
38
    timestamp,
28
39
    )
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
 
40
""")
 
41
 
 
42
from bzrlib.symbol_versioning import (
 
43
        deprecated_function,
 
44
        one_three
 
45
        )
38
46
from bzrlib.trace import mutter, warning
39
47
 
40
48
 
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'):
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:
87
107
        to_file.write(line)
88
108
        if not line.endswith('\n'):
89
109
            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)
 
110
    to_file.write('\n')
99
111
 
100
112
 
101
113
def _spawn_external_diff(diffcmd, capture_errors=True):
102
114
    """Spawn the externall diff process, and return the child handle.
103
115
 
104
116
    :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.
 
117
    :param capture_errors: Capture stderr as well as setting LANG=C
 
118
        and LC_ALL=C. This lets us read and understand the output of diff,
 
119
        and respond to any errors.
108
120
    :return: A Popen object.
109
121
    """
110
122
    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
 
123
        # construct minimal environment
 
124
        env = {}
 
125
        path = os.environ.get('PATH')
 
126
        if path is not None:
 
127
            env['PATH'] = path
 
128
        env['LANGUAGE'] = 'C'   # on win32 only LANGUAGE has effect
 
129
        env['LANG'] = 'C'
 
130
        env['LC_ALL'] = 'C'
117
131
        stderr = subprocess.PIPE
118
132
    else:
119
 
        preexec_fn = None
 
133
        env = None
120
134
        stderr = None
121
135
 
122
136
    try:
124
138
                                stdin=subprocess.PIPE,
125
139
                                stdout=subprocess.PIPE,
126
140
                                stderr=stderr,
127
 
                                preexec_fn=preexec_fn)
 
141
                                env=env)
128
142
    except OSError, e:
129
143
        if e.errno == errno.ENOENT:
130
144
            raise errors.NoDiff(str(e))
212
226
            # Write out the new i18n diff response
213
227
            to_file.write(out+'\n')
214
228
            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,))
 
229
                raise errors.BzrError(
 
230
                               'external diff failed with exit code 2'
 
231
                               ' when run with LANG=C and LC_ALL=C,'
 
232
                               ' but not when run natively: %r' % (diffcmd,))
218
233
 
219
234
            first_line = lang_c_out.split('\n', 1)[0]
220
235
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
221
236
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
222
237
            if m is None:
223
 
                raise BzrError('external diff failed with exit code 2;'
224
 
                               ' command: %r' % (diffcmd,))
 
238
                raise errors.BzrError('external diff failed with exit code 2;'
 
239
                                      ' command: %r' % (diffcmd,))
225
240
            else:
226
241
                # Binary files differ, just return
227
242
                return
236
251
            else:
237
252
                msg = 'exit code %d' % rc
238
253
                
239
 
            raise BzrError('external diff failed with %s; command: %r' 
240
 
                           % (rc, diffcmd))
 
254
            raise errors.BzrError('external diff failed with %s; command: %r' 
 
255
                                  % (rc, diffcmd))
241
256
 
242
257
 
243
258
    finally:
260
275
                        new_abspath, e)
261
276
 
262
277
 
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)
 
278
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url):
 
279
    """Get the trees and specific files to diff given a list of paths.
 
280
 
 
281
    This method works out the trees to be diff'ed and the files of
 
282
    interest within those trees.
 
283
 
 
284
    :param path_list:
 
285
        the list of arguments passed to the diff command
 
286
    :param revision_specs:
 
287
        Zero, one or two RevisionSpecs from the diff command line,
 
288
        saying what revisions to compare.
 
289
    :param old_url:
 
290
        The url of the old branch or tree. If None, the tree to use is
 
291
        taken from the first path, if any, or the current working tree.
 
292
    :param new_url:
 
293
        The url of the new branch or tree. If None, the tree to use is
 
294
        taken from the first path, if any, or the current working tree.
 
295
    :returns:
 
296
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
297
        extra_trees is a sequence of additional trees to search in for
 
298
        file-ids.
 
299
    """
 
300
    # Get the old and new revision specs
 
301
    old_revision_spec = None
 
302
    new_revision_spec = None
 
303
    if revision_specs is not None:
 
304
        if len(revision_specs) > 0:
 
305
            old_revision_spec = revision_specs[0]
 
306
            if old_url is None:
 
307
                old_url = old_revision_spec.get_branch()
 
308
        if len(revision_specs) > 1:
 
309
            new_revision_spec = revision_specs[1]
 
310
            if new_url is None:
 
311
                new_url = new_revision_spec.get_branch()
 
312
 
 
313
    other_paths = []
 
314
    make_paths_wt_relative = True
 
315
    consider_relpath = True
 
316
    if path_list is None or len(path_list) == 0:
 
317
        # If no path is given, the current working tree is used
 
318
        default_location = u'.'
 
319
        consider_relpath = False
 
320
    elif old_url is not None and new_url is not None:
 
321
        other_paths = path_list
 
322
        make_paths_wt_relative = False
 
323
    else:
 
324
        default_location = path_list[0]
 
325
        other_paths = path_list[1:]
 
326
 
 
327
    # Get the old location
 
328
    specific_files = []
 
329
    if old_url is None:
 
330
        old_url = default_location
 
331
    working_tree, branch, relpath = \
 
332
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
 
333
    if consider_relpath and relpath != '':
 
334
        specific_files.append(relpath)
 
335
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
 
336
 
 
337
    # Get the new location
 
338
    if new_url is None:
 
339
        new_url = default_location
 
340
    if new_url != old_url:
 
341
        working_tree, branch, relpath = \
 
342
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
 
343
        if consider_relpath and relpath != '':
 
344
            specific_files.append(relpath)
 
345
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
 
346
        basis_is_default=working_tree is None)
 
347
 
 
348
    # Get the specific files (all files is None, no files is [])
 
349
    if make_paths_wt_relative and working_tree is not None:
 
350
        other_paths = _relative_paths_in_tree(working_tree, other_paths)
 
351
    specific_files.extend(other_paths)
 
352
    if len(specific_files) == 0:
 
353
        specific_files = None
 
354
 
 
355
    # Get extra trees that ought to be searched for file-ids
 
356
    extra_trees = None
 
357
    if working_tree is not None and working_tree not in (old_tree, new_tree):
 
358
        extra_trees = (working_tree,)
 
359
    return old_tree, new_tree, specific_files, extra_trees
 
360
 
 
361
 
 
362
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
 
363
    if branch is None and tree is not None:
 
364
        branch = tree.branch
 
365
    if spec is None or spec.spec is None:
 
366
        if basis_is_default:
 
367
            if tree is not None:
 
368
                return tree.basis_tree()
 
369
            else:
 
370
                return branch.basis_tree()
 
371
        else:
 
372
            return tree
 
373
    if not spec.needs_branch():
 
374
        branch = _mod_branch.Branch.open(spec.get_branch())
 
375
    revision_id = spec.as_revision_id(branch)
 
376
    return branch.repository.revision_tree(revision_id)
 
377
 
 
378
 
 
379
def _relative_paths_in_tree(tree, paths):
 
380
    """Get the relative paths within a working tree.
 
381
 
 
382
    Each path may be either an absolute path or a path relative to the
 
383
    current working directory.
 
384
    """
 
385
    result = []
 
386
    for filename in paths:
 
387
        try:
 
388
            result.append(tree.relpath(osutils.dereference_path(filename)))
 
389
        except errors.PathNotChild:
 
390
            raise errors.BzrCommandError("Files are in different branches")
 
391
    return result
352
392
 
353
393
 
354
394
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
355
395
                    external_diff_options=None,
356
396
                    old_label='a/', new_label='b/',
357
 
                    extra_trees=None):
 
397
                    extra_trees=None,
 
398
                    path_encoding='utf8',
 
399
                    using=None):
358
400
    """Show in text form the changes from one tree to another.
359
401
 
360
 
    to_files
361
 
        If set, include only changes to these files.
 
402
    to_file
 
403
        The output stream.
 
404
 
 
405
    specific_files
 
406
        Include only changes to these files - None for all changes.
362
407
 
363
408
    external_diff_options
364
409
        If set, use an external GNU diff and pass these options.
365
410
 
366
411
    extra_trees
367
412
        If set, more Trees to use for looking up file ids
 
413
 
 
414
    path_encoding
 
415
        If set, the path will be encoded as specified, otherwise is supposed
 
416
        to be utf8
368
417
    """
369
418
    old_tree.lock_read()
370
419
    try:
 
420
        if extra_trees is not None:
 
421
            for tree in extra_trees:
 
422
                tree.lock_read()
371
423
        new_tree.lock_read()
372
424
        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)
 
425
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
426
                                                 path_encoding,
 
427
                                                 external_diff_options,
 
428
                                                 old_label, new_label, using)
 
429
            return differ.show_diff(specific_files, extra_trees)
377
430
        finally:
378
431
            new_tree.unlock()
 
432
            if extra_trees is not None:
 
433
                for tree in extra_trees:
 
434
                    tree.unlock()
379
435
    finally:
380
436
        old_tree.unlock()
381
437
 
382
438
 
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):
386
 
 
387
 
    # GNU Patch uses the epoch date to detect files that are being added
388
 
    # or removed in a diff.
389
 
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
390
 
 
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
439
def _patch_header_date(tree, file_id, path):
457
440
    """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)
 
441
    mtime = tree.get_file_mtime(file_id, path)
 
442
    return timestamp.format_patch_date(mtime)
460
443
 
461
444
 
462
445
def _raise_if_nonexistent(paths, old_tree, new_tree):
478
461
        raise errors.PathsDoNotExist(sorted(s))
479
462
 
480
463
 
 
464
@deprecated_function(one_three)
481
465
def get_prop_change(meta_modified):
482
466
    if meta_modified:
483
467
        return " (properties changed)"
484
468
    else:
485
469
        return  ""
486
470
 
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)
 
471
def get_executable_change(old_is_x, new_is_x):
 
472
    descr = { True:"+x", False:"-x", None:"??" }
 
473
    if old_is_x != new_is_x:
 
474
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
475
    else:
 
476
        return []
 
477
 
 
478
 
 
479
class DiffPath(object):
 
480
    """Base type for command object that compare files"""
 
481
 
 
482
    # The type or contents of the file were unsuitable for diffing
 
483
    CANNOT_DIFF = 'CANNOT_DIFF'
 
484
    # The file has changed in a semantic way
 
485
    CHANGED = 'CHANGED'
 
486
    # The file content may have changed, but there is no semantic change
 
487
    UNCHANGED = 'UNCHANGED'
 
488
 
 
489
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
 
490
        """Constructor.
 
491
 
 
492
        :param old_tree: The tree to show as the old tree in the comparison
 
493
        :param new_tree: The tree to show as new in the comparison
 
494
        :param to_file: The file to write comparison data to
 
495
        :param path_encoding: The character encoding to write paths in
 
496
        """
 
497
        self.old_tree = old_tree
 
498
        self.new_tree = new_tree
 
499
        self.to_file = to_file
 
500
        self.path_encoding = path_encoding
 
501
 
 
502
    def finish(self):
 
503
        pass
 
504
 
 
505
    @classmethod
 
506
    def from_diff_tree(klass, diff_tree):
 
507
        return klass(diff_tree.old_tree, diff_tree.new_tree,
 
508
                     diff_tree.to_file, diff_tree.path_encoding)
 
509
 
 
510
    @staticmethod
 
511
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
 
512
        for file_differ in differs:
 
513
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
514
                                      new_kind)
 
515
            if result is not DiffPath.CANNOT_DIFF:
 
516
                return result
 
517
        else:
 
518
            return DiffPath.CANNOT_DIFF
 
519
 
 
520
 
 
521
class DiffKindChange(object):
 
522
    """Special differ for file kind changes.
 
523
 
 
524
    Represents kind change as deletion + creation.  Uses the other differs
 
525
    to do this.
 
526
    """
 
527
    def __init__(self, differs):
 
528
        self.differs = differs
 
529
 
 
530
    def finish(self):
 
531
        pass
 
532
 
 
533
    @classmethod
 
534
    def from_diff_tree(klass, diff_tree):
 
535
        return klass(diff_tree.differs)
 
536
 
 
537
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
538
        """Perform comparison
 
539
 
 
540
        :param file_id: The file_id of the file to compare
 
541
        :param old_path: Path of the file in the old tree
 
542
        :param new_path: Path of the file in the new tree
 
543
        :param old_kind: Old file-kind of the file
 
544
        :param new_kind: New file-kind of the file
 
545
        """
 
546
        if None in (old_kind, new_kind):
 
547
            return DiffPath.CANNOT_DIFF
 
548
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
549
                                       new_path, old_kind, None)
 
550
        if result is DiffPath.CANNOT_DIFF:
 
551
            return result
 
552
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
553
                                     None, new_kind)
 
554
 
 
555
 
 
556
class DiffDirectory(DiffPath):
 
557
 
 
558
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
559
        """Perform comparison between two directories.  (dummy)
 
560
 
 
561
        """
 
562
        if 'directory' not in (old_kind, new_kind):
 
563
            return self.CANNOT_DIFF
 
564
        if old_kind not in ('directory', None):
 
565
            return self.CANNOT_DIFF
 
566
        if new_kind not in ('directory', None):
 
567
            return self.CANNOT_DIFF
 
568
        return self.CHANGED
 
569
 
 
570
 
 
571
class DiffSymlink(DiffPath):
 
572
 
 
573
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
574
        """Perform comparison between two symlinks
 
575
 
 
576
        :param file_id: The file_id of the file to compare
 
577
        :param old_path: Path of the file in the old tree
 
578
        :param new_path: Path of the file in the new tree
 
579
        :param old_kind: Old file-kind of the file
 
580
        :param new_kind: New file-kind of the file
 
581
        """
 
582
        if 'symlink' not in (old_kind, new_kind):
 
583
            return self.CANNOT_DIFF
 
584
        if old_kind == 'symlink':
 
585
            old_target = self.old_tree.get_symlink_target(file_id)
 
586
        elif old_kind is None:
 
587
            old_target = None
 
588
        else:
 
589
            return self.CANNOT_DIFF
 
590
        if new_kind == 'symlink':
 
591
            new_target = self.new_tree.get_symlink_target(file_id)
 
592
        elif new_kind is None:
 
593
            new_target = None
 
594
        else:
 
595
            return self.CANNOT_DIFF
 
596
        return self.diff_symlink(old_target, new_target)
 
597
 
 
598
    def diff_symlink(self, old_target, new_target):
 
599
        if old_target is None:
 
600
            self.to_file.write('=== target is %r\n' % new_target)
 
601
        elif new_target is None:
 
602
            self.to_file.write('=== target was %r\n' % old_target)
 
603
        else:
 
604
            self.to_file.write('=== target changed %r => %r\n' %
 
605
                              (old_target, new_target))
 
606
        return self.CHANGED
 
607
 
 
608
 
 
609
class DiffText(DiffPath):
 
610
 
 
611
    # GNU Patch uses the epoch date to detect files that are being added
 
612
    # or removed in a diff.
 
613
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
 
614
 
 
615
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
616
                 old_label='', new_label='', text_differ=internal_diff):
 
617
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
618
        self.text_differ = text_differ
 
619
        self.old_label = old_label
 
620
        self.new_label = new_label
 
621
        self.path_encoding = path_encoding
 
622
 
 
623
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
624
        """Compare two files in unified diff format
 
625
 
 
626
        :param file_id: The file_id of the file to compare
 
627
        :param old_path: Path of the file in the old tree
 
628
        :param new_path: Path of the file in the new tree
 
629
        :param old_kind: Old file-kind of the file
 
630
        :param new_kind: New file-kind of the file
 
631
        """
 
632
        if 'file' not in (old_kind, new_kind):
 
633
            return self.CANNOT_DIFF
 
634
        from_file_id = to_file_id = file_id
 
635
        if old_kind == 'file':
 
636
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
 
637
        elif old_kind is None:
 
638
            old_date = self.EPOCH_DATE
 
639
            from_file_id = None
 
640
        else:
 
641
            return self.CANNOT_DIFF
 
642
        if new_kind == 'file':
 
643
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
 
644
        elif new_kind is None:
 
645
            new_date = self.EPOCH_DATE
 
646
            to_file_id = None
 
647
        else:
 
648
            return self.CANNOT_DIFF
 
649
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
650
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
651
        return self.diff_text(from_file_id, to_file_id, from_label, to_label)
 
652
 
 
653
    def diff_text(self, from_file_id, to_file_id, from_label, to_label):
 
654
        """Diff the content of given files in two trees
 
655
 
 
656
        :param from_file_id: The id of the file in the from tree.  If None,
 
657
            the file is not present in the from tree.
 
658
        :param to_file_id: The id of the file in the to tree.  This may refer
 
659
            to a different file from from_file_id.  If None,
 
660
            the file is not present in the to tree.
 
661
        """
 
662
        def _get_text(tree, file_id):
 
663
            if file_id is not None:
 
664
                return tree.get_file(file_id).readlines()
 
665
            else:
 
666
                return []
 
667
        try:
 
668
            from_text = _get_text(self.old_tree, from_file_id)
 
669
            to_text = _get_text(self.new_tree, to_file_id)
 
670
            self.text_differ(from_label, from_text, to_label, to_text,
 
671
                             self.to_file)
 
672
        except errors.BinaryFile:
 
673
            self.to_file.write(
 
674
                  ("Binary files %s and %s differ\n" %
 
675
                  (from_label, to_label)).encode(self.path_encoding))
 
676
        return self.CHANGED
 
677
 
 
678
 
 
679
class DiffFromTool(DiffPath):
 
680
 
 
681
    def __init__(self, command_template, old_tree, new_tree, to_file,
 
682
                 path_encoding='utf-8'):
 
683
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
684
        self.command_template = command_template
 
685
        self._root = tempfile.mkdtemp(prefix='bzr-diff-')
 
686
 
 
687
    @classmethod
 
688
    def from_string(klass, command_string, old_tree, new_tree, to_file,
 
689
                    path_encoding='utf-8'):
 
690
        command_template = commands.shlex_split_unicode(command_string)
 
691
        command_template.extend(['%(old_path)s', '%(new_path)s'])
 
692
        return klass(command_template, old_tree, new_tree, to_file,
 
693
                     path_encoding)
 
694
 
 
695
    @classmethod
 
696
    def make_from_diff_tree(klass, command_string):
 
697
        def from_diff_tree(diff_tree):
 
698
            return klass.from_string(command_string, diff_tree.old_tree,
 
699
                                     diff_tree.new_tree, diff_tree.to_file)
 
700
        return from_diff_tree
 
701
 
 
702
    def _get_command(self, old_path, new_path):
 
703
        my_map = {'old_path': old_path, 'new_path': new_path}
 
704
        return [t % my_map for t in self.command_template]
 
705
 
 
706
    def _execute(self, old_path, new_path):
 
707
        command = self._get_command(old_path, new_path)
 
708
        try:
 
709
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
 
710
                                    cwd=self._root)
 
711
        except OSError, e:
 
712
            if e.errno == errno.ENOENT:
 
713
                raise errors.ExecutableMissing(command[0])
 
714
            else:
 
715
                raise
 
716
        self.to_file.write(proc.stdout.read())
 
717
        return proc.wait()
 
718
 
 
719
    def _try_symlink_root(self, tree, prefix):
 
720
        if not (getattr(tree, 'abspath', None) is not None
 
721
                and osutils.has_symlinks()):
 
722
            return False
 
723
        try:
 
724
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
 
725
        except OSError, e:
 
726
            if e.errno != errno.EEXIST:
 
727
                raise
 
728
        return True
 
729
 
 
730
    def _write_file(self, file_id, tree, prefix, relpath):
 
731
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
732
        if self._try_symlink_root(tree, prefix):
 
733
            return full_path
 
734
        parent_dir = osutils.dirname(full_path)
 
735
        try:
 
736
            os.makedirs(parent_dir)
 
737
        except OSError, e:
 
738
            if e.errno != errno.EEXIST:
 
739
                raise
 
740
        source = tree.get_file(file_id, relpath)
 
741
        try:
 
742
            target = open(full_path, 'wb')
 
743
            try:
 
744
                osutils.pumpfile(source, target)
 
745
            finally:
 
746
                target.close()
 
747
        finally:
 
748
            source.close()
 
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):
 
755
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
756
                                         old_path)
 
757
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
758
                                         new_path)
 
759
        return old_disk_path, new_disk_path
 
760
 
 
761
    def finish(self):
 
762
        osutils.rmtree(self._root)
 
763
 
 
764
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
765
        if (old_kind, new_kind) != ('file', 'file'):
 
766
            return DiffPath.CANNOT_DIFF
 
767
        self._prepare_files(file_id, old_path, new_path)
 
768
        self._execute(osutils.pathjoin('old', old_path),
 
769
                      osutils.pathjoin('new', new_path))
 
770
 
 
771
 
 
772
class DiffTree(object):
 
773
    """Provides textual representations of the difference between two trees.
 
774
 
 
775
    A DiffTree examines two trees and where a file-id has altered
 
776
    between them, generates a textual representation of the difference.
 
777
    DiffTree uses a sequence of DiffPath objects which are each
 
778
    given the opportunity to handle a given altered fileid. The list
 
779
    of DiffPath objects can be extended globally by appending to
 
780
    DiffTree.diff_factories, or for a specific diff operation by
 
781
    supplying the extra_factories option to the appropriate method.
 
782
    """
 
783
 
 
784
    # list of factories that can provide instances of DiffPath objects
 
785
    # may be extended by plugins.
 
786
    diff_factories = [DiffSymlink.from_diff_tree,
 
787
                      DiffDirectory.from_diff_tree]
 
788
 
 
789
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
790
                 diff_text=None, extra_factories=None):
 
791
        """Constructor
 
792
 
 
793
        :param old_tree: Tree to show as old in the comparison
 
794
        :param new_tree: Tree to show as new in the comparison
 
795
        :param to_file: File to write comparision to
 
796
        :param path_encoding: Character encoding to write paths in
 
797
        :param diff_text: DiffPath-type object to use as a last resort for
 
798
            diffing text files.
 
799
        :param extra_factories: Factories of DiffPaths to try before any other
 
800
            DiffPaths"""
 
801
        if diff_text is None:
 
802
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
803
                                 '', '',  internal_diff)
 
804
        self.old_tree = old_tree
 
805
        self.new_tree = new_tree
 
806
        self.to_file = to_file
 
807
        self.path_encoding = path_encoding
 
808
        self.differs = []
 
809
        if extra_factories is not None:
 
810
            self.differs.extend(f(self) for f in extra_factories)
 
811
        self.differs.extend(f(self) for f in self.diff_factories)
 
812
        self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
 
813
 
 
814
    @classmethod
 
815
    def from_trees_options(klass, old_tree, new_tree, to_file,
 
816
                           path_encoding, external_diff_options, old_label,
 
817
                           new_label, using):
 
818
        """Factory for producing a DiffTree.
 
819
 
 
820
        Designed to accept options used by show_diff_trees.
 
821
        :param old_tree: The tree to show as old in the comparison
 
822
        :param new_tree: The tree to show as new in the comparison
 
823
        :param to_file: File to write comparisons to
 
824
        :param path_encoding: Character encoding to use for writing paths
 
825
        :param external_diff_options: If supplied, use the installed diff
 
826
            binary to perform file comparison, using supplied options.
 
827
        :param old_label: Prefix to use for old file labels
 
828
        :param new_label: Prefix to use for new file labels
 
829
        :param using: Commandline to use to invoke an external diff tool
 
830
        """
 
831
        if using is not None:
 
832
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
 
833
        else:
 
834
            extra_factories = []
 
835
        if external_diff_options:
 
836
            opts = external_diff_options.split()
 
837
            def diff_file(olab, olines, nlab, nlines, to_file):
 
838
                external_diff(olab, olines, nlab, nlines, to_file, opts)
 
839
        else:
 
840
            diff_file = internal_diff
 
841
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
842
                             old_label, new_label, diff_file)
 
843
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
 
844
                     extra_factories)
 
845
 
 
846
    def show_diff(self, specific_files, extra_trees=None):
 
847
        """Write tree diff to self.to_file
 
848
 
 
849
        :param sepecific_files: the specific files to compare (recursive)
 
850
        :param extra_trees: extra trees to use for mapping paths to file_ids
 
851
        """
 
852
        try:
 
853
            return self._show_diff(specific_files, extra_trees)
 
854
        finally:
 
855
            for differ in self.differs:
 
856
                differ.finish()
 
857
 
 
858
    def _show_diff(self, specific_files, extra_trees):
 
859
        # TODO: Generation of pseudo-diffs for added/deleted files could
 
860
        # be usefully made into a much faster special case.
 
861
        iterator = self.new_tree.iter_changes(self.old_tree,
 
862
                                               specific_files=specific_files,
 
863
                                               extra_trees=extra_trees,
 
864
                                               require_versioned=True)
 
865
        has_changes = 0
 
866
        def changes_key(change):
 
867
            old_path, new_path = change[1]
 
868
            path = new_path
 
869
            if path is None:
 
870
                path = old_path
 
871
            return path
 
872
        def get_encoded_path(path):
 
873
            if path is not None:
 
874
                return path.encode(self.path_encoding, "replace")
 
875
        for (file_id, paths, changed_content, versioned, parent, name, kind,
 
876
             executable) in sorted(iterator, key=changes_key):
 
877
            if parent == (None, None):
 
878
                continue
 
879
            oldpath, newpath = paths
 
880
            oldpath_encoded = get_encoded_path(paths[0])
 
881
            newpath_encoded = get_encoded_path(paths[1])
 
882
            old_present = (kind[0] is not None and versioned[0])
 
883
            new_present = (kind[1] is not None and versioned[1])
 
884
            renamed = (parent[0], name[0]) != (parent[1], name[1])
 
885
 
 
886
            properties_changed = []
 
887
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
 
888
 
 
889
            if properties_changed:
 
890
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
 
891
            else:
 
892
                prop_str = ""
 
893
 
 
894
            if (old_present, new_present) == (True, False):
 
895
                self.to_file.write("=== removed %s '%s'\n" %
 
896
                                   (kind[0], oldpath_encoded))
 
897
                newpath = oldpath
 
898
            elif (old_present, new_present) == (False, True):
 
899
                self.to_file.write("=== added %s '%s'\n" %
 
900
                                   (kind[1], newpath_encoded))
 
901
                oldpath = newpath
 
902
            elif renamed:
 
903
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
904
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
 
905
            else:
 
906
                # if it was produced by iter_changes, it must be
 
907
                # modified *somehow*, either content or execute bit.
 
908
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
909
                                   newpath_encoded, prop_str))
 
910
            if changed_content:
 
911
                self.diff(file_id, oldpath, newpath)
 
912
                has_changes = 1
 
913
            if renamed:
 
914
                has_changes = 1
 
915
        return has_changes
 
916
 
 
917
    def diff(self, file_id, old_path, new_path):
 
918
        """Perform a diff of a single file
 
919
 
 
920
        :param file_id: file-id of the file
 
921
        :param old_path: The path of the file in the old tree
 
922
        :param new_path: The path of the file in the new tree
 
923
        """
 
924
        try:
 
925
            old_kind = self.old_tree.kind(file_id)
 
926
        except (errors.NoSuchId, errors.NoSuchFile):
 
927
            old_kind = None
 
928
        try:
 
929
            new_kind = self.new_tree.kind(file_id)
 
930
        except (errors.NoSuchId, errors.NoSuchFile):
 
931
            new_kind = None
 
932
 
 
933
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
934
                                       new_path, old_kind, new_kind)
 
935
        if result is DiffPath.CANNOT_DIFF:
 
936
            error_path = new_path
 
937
            if error_path is None:
 
938
                error_path = old_path
 
939
            raise errors.NoDiffFound(error_path)