/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 breezy/git/tree.py

  • Committer: Jelmer Vernooij
  • Date: 2019-03-05 07:32:38 UTC
  • mto: (7290.1.21 work)
  • mto: This revision was merged to the branch mainline in revision 7311.
  • Revision ID: jelmer@jelmer.uk-20190305073238-zlqn981opwnqsmzi
Add appveyor configuration.

Show diffs side-by-side

added added

removed removed

Lines of Context:
breezy/git/tree.py
17
17
 
18
18
"""Git Trees."""
19
19
 
 
20
from __future__ import absolute_import
 
21
 
20
22
from collections import deque
21
23
import errno
22
24
from io import BytesIO
23
25
import os
24
26
 
25
 
from dulwich.config import (
26
 
    parse_submodules,
27
 
    ConfigFile as GitConfigFile,
28
 
    )
29
 
from dulwich.diff_tree import tree_changes, RenameDetector
30
 
from dulwich.errors import NotTreeError
31
27
from dulwich.index import (
32
28
    blob_from_path_and_stat,
33
29
    cleanup_mode,
34
30
    commit_tree,
35
31
    index_entry_from_stat,
36
 
    Index,
37
32
    )
38
33
from dulwich.object_store import (
39
34
    tree_lookup_path,
64
59
    CURRENT_REVISION,
65
60
    NULL_REVISION,
66
61
    )
 
62
from ..sixish import (
 
63
    text_type,
 
64
    viewitems,
 
65
    )
67
66
 
68
67
from .mapping import (
69
 
    encode_git_path,
70
 
    decode_git_path,
71
68
    mode_is_executable,
72
69
    mode_kind,
 
70
    GitFileIdMap,
73
71
    default_mapping,
74
72
    )
75
 
from .transportgit import (
76
 
    TransportObjectStore,
77
 
    TransportRepo,
78
 
    )
79
 
from ..bzr.inventorytree import InventoryTreeChange
80
73
 
81
74
 
82
75
class GitTreeDirectory(_mod_tree.TreeDirectory):
83
76
 
84
 
    __slots__ = ['file_id', 'name', 'parent_id']
 
77
    __slots__ = ['file_id', 'name', 'parent_id', 'children']
85
78
 
86
79
    def __init__(self, file_id, name, parent_id):
87
80
        self.file_id = file_id
88
81
        self.name = name
89
82
        self.parent_id = parent_id
 
83
        # TODO(jelmer)
 
84
        self.children = {}
90
85
 
91
86
    @property
92
87
    def kind(self):
114
109
 
115
110
class GitTreeFile(_mod_tree.TreeFile):
116
111
 
117
 
    __slots__ = ['file_id', 'name', 'parent_id', 'text_size',
118
 
                 'executable', 'git_sha1']
 
112
    __slots__ = ['file_id', 'name', 'parent_id', 'text_size', 'text_sha1',
 
113
                 'executable']
119
114
 
120
115
    def __init__(self, file_id, name, parent_id, text_size=None,
121
 
                 git_sha1=None, executable=None):
 
116
                 text_sha1=None, executable=None):
122
117
        self.file_id = file_id
123
118
        self.name = name
124
119
        self.parent_id = parent_id
125
120
        self.text_size = text_size
126
 
        self.git_sha1 = git_sha1
 
121
        self.text_sha1 = text_sha1
127
122
        self.executable = executable
128
123
 
129
124
    @property
135
130
                self.file_id == other.file_id and
136
131
                self.name == other.name and
137
132
                self.parent_id == other.parent_id and
138
 
                self.git_sha1 == other.git_sha1 and
 
133
                self.text_sha1 == other.text_sha1 and
139
134
                self.text_size == other.text_size and
140
135
                self.executable == other.executable)
141
136
 
142
137
    def __repr__(self):
143
138
        return ("%s(file_id=%r, name=%r, parent_id=%r, text_size=%r, "
144
 
                "git_sha1=%r, executable=%r)") % (
 
139
                "text_sha1=%r, executable=%r)") % (
145
140
            type(self).__name__, self.file_id, self.name, self.parent_id,
146
 
            self.text_size, self.git_sha1, self.executable)
 
141
            self.text_size, self.text_sha1, self.executable)
147
142
 
148
143
    def copy(self):
149
144
        ret = self.__class__(
150
145
            self.file_id, self.name, self.parent_id)
151
 
        ret.git_sha1 = self.git_sha1
 
146
        ret.text_sha1 = self.text_sha1
152
147
        ret.text_size = self.text_size
153
148
        ret.executable = self.executable
154
149
        return ret
195
190
            self.symlink_target)
196
191
 
197
192
 
198
 
class GitTreeSubmodule(_mod_tree.TreeReference):
 
193
class GitTreeSubmodule(_mod_tree.TreeLink):
199
194
 
200
195
    __slots__ = ['file_id', 'name', 'parent_id', 'reference_revision']
201
196
 
206
201
        self.reference_revision = reference_revision
207
202
 
208
203
    @property
209
 
    def executable(self):
210
 
        return False
211
 
 
212
 
    @property
213
204
    def kind(self):
214
205
        return 'tree-reference'
215
206
 
256
247
    return path
257
248
 
258
249
 
259
 
class GitTree(_mod_tree.Tree):
260
 
 
261
 
    def iter_git_objects(self):
262
 
        """Iterate over all the objects in the tree.
263
 
 
264
 
        :return :Yields tuples with (path, sha, mode)
265
 
        """
266
 
        raise NotImplementedError(self.iter_git_objects)
267
 
 
268
 
    def git_snapshot(self, want_unversioned=False):
269
 
        """Snapshot a tree, and return tree object.
270
 
 
271
 
        :return: Tree sha and set of extras
272
 
        """
273
 
        raise NotImplementedError(self.snapshot)
274
 
 
275
 
    def preview_transform(self, pb=None):
276
 
        from .transform import GitTransformPreview
277
 
        return GitTransformPreview(self, pb=pb)
278
 
 
279
 
    def find_related_paths_across_trees(self, paths, trees=[],
280
 
                                        require_versioned=True):
281
 
        if paths is None:
282
 
            return None
283
 
        if require_versioned:
284
 
            trees = [self] + (trees if trees is not None else [])
285
 
            unversioned = set()
286
 
            for p in paths:
287
 
                for t in trees:
288
 
                    if t.is_versioned(p):
289
 
                        break
290
 
                else:
291
 
                    unversioned.add(p)
292
 
            if unversioned:
293
 
                raise errors.PathsNotVersionedError(unversioned)
294
 
        return filter(self.is_versioned, paths)
295
 
 
296
 
    def _submodule_info(self):
297
 
        if self._submodules is None:
298
 
            try:
299
 
                with self.get_file('.gitmodules') as f:
300
 
                    config = GitConfigFile.from_file(f)
301
 
                    self._submodules = {
302
 
                        path: (url, section)
303
 
                        for path, url, section in parse_submodules(config)}
304
 
            except errors.NoSuchFile:
305
 
                self._submodules = {}
306
 
        return self._submodules
307
 
 
308
 
 
309
 
class GitRevisionTree(revisiontree.RevisionTree, GitTree):
 
250
class GitRevisionTree(revisiontree.RevisionTree):
310
251
    """Revision tree implementation based on Git objects."""
311
252
 
312
253
    def __init__(self, repository, revision_id):
313
254
        self._revision_id = revision_id
314
255
        self._repository = repository
315
 
        self._submodules = None
316
256
        self.store = repository._git.object_store
317
257
        if not isinstance(revision_id, bytes):
318
258
            raise TypeError(revision_id)
321
261
        if revision_id == NULL_REVISION:
322
262
            self.tree = None
323
263
            self.mapping = default_mapping
 
264
            self._fileid_map = GitFileIdMap(
 
265
                {},
 
266
                default_mapping)
324
267
        else:
325
268
            try:
326
269
                commit = self.store[self.commit_id]
327
270
            except KeyError:
328
271
                raise errors.NoSuchRevision(repository, revision_id)
329
272
            self.tree = commit.tree
330
 
 
331
 
    def git_snapshot(self, want_unversioned=False):
332
 
        return self.tree, set()
333
 
 
334
 
    def _get_submodule_repository(self, relpath):
335
 
        if not isinstance(relpath, bytes):
336
 
            raise TypeError(relpath)
337
 
        try:
338
 
            info = self._submodule_info()[relpath]
339
 
        except KeyError:
340
 
            nested_repo_transport = self._repository.controldir.user_transport.clone(
341
 
                decode_git_path(relpath))
342
 
        else:
343
 
            nested_repo_transport = self._repository.controldir.control_transport.clone(
344
 
                posixpath.join('modules', decode_git_path(info[1])))
 
273
            self._fileid_map = self.mapping.get_fileid_map(
 
274
                self.store.__getitem__, self.tree)
 
275
 
 
276
    def _get_nested_repository(self, path):
 
277
        nested_repo_transport = self._repository.user_transport.clone(path)
345
278
        nested_controldir = _mod_controldir.ControlDir.open_from_transport(
346
279
            nested_repo_transport)
347
280
        return nested_controldir.find_repository()
348
281
 
349
 
    def _get_submodule_store(self, relpath):
350
 
        return self._get_submodule_repository(relpath)._git.object_store
351
 
 
352
 
    def get_nested_tree(self, path):
353
 
        encoded_path = encode_git_path(path)
354
 
        nested_repo = self._get_submodule_repository(encoded_path)
355
 
        ref_rev = self.get_reference_revision(path)
356
 
        return nested_repo.revision_tree(ref_rev)
357
 
 
358
282
    def supports_rename_tracking(self):
359
283
        return False
360
284
 
363
287
        if self.commit_id == ZERO_SHA:
364
288
            return NULL_REVISION
365
289
        (unused_path, commit_id) = change_scanner.find_last_change_revision(
366
 
            encode_git_path(path), self.commit_id)
 
290
            path.encode('utf-8'), self.commit_id)
367
291
        return self._repository.lookup_foreign_revision_id(
368
292
            commit_id, self.mapping)
369
293
 
378
302
            raise _mod_tree.FileTimestampUnavailable(path)
379
303
        return rev.timestamp
380
304
 
381
 
    def id2path(self, file_id, recurse='down'):
 
305
    def id2path(self, file_id):
382
306
        try:
383
 
            path = self.mapping.parse_file_id(file_id)
 
307
            path = self._fileid_map.lookup_path(file_id)
384
308
        except ValueError:
385
309
            raise errors.NoSuchId(self, file_id)
386
310
        if self.is_versioned(path):
395
319
            return None
396
320
        if not self.is_versioned(path):
397
321
            return None
398
 
        return self.mapping.generate_file_id(osutils.safe_unicode(path))
 
322
        return self._fileid_map.lookup_file_id(osutils.safe_unicode(path))
399
323
 
400
324
    def all_file_ids(self):
401
325
        raise errors.UnsupportedOperation(self.all_file_ids, self)
410
334
            tree = store[tree_id]
411
335
            for name, mode, hexsha in tree.items():
412
336
                subpath = posixpath.join(path, name)
413
 
                ret.add(decode_git_path(subpath))
 
337
                ret.add(subpath.decode('utf-8'))
414
338
                if stat.S_ISDIR(mode):
415
339
                    todo.append((store, subpath, hexsha))
416
340
        return ret
417
341
 
 
342
    def get_root_id(self):
 
343
        if self.tree is None:
 
344
            return None
 
345
        return self.path2id("")
 
346
 
 
347
    def has_or_had_id(self, file_id):
 
348
        try:
 
349
            self.id2path(file_id)
 
350
        except errors.NoSuchId:
 
351
            return False
 
352
        return True
 
353
 
 
354
    def has_id(self, file_id):
 
355
        try:
 
356
            path = self.id2path(file_id)
 
357
        except errors.NoSuchId:
 
358
            return False
 
359
        return self.has_filename(path)
 
360
 
418
361
    def _lookup_path(self, path):
419
362
        if self.tree is None:
420
363
            raise errors.NoSuchFile(path)
421
 
 
422
 
        encoded_path = encode_git_path(path)
423
 
        parts = encoded_path.split(b'/')
424
 
        hexsha = self.tree
425
 
        store = self.store
426
 
        mode = None
427
 
        for i, p in enumerate(parts):
428
 
            if not p:
429
 
                continue
430
 
            obj = store[hexsha]
431
 
            if not isinstance(obj, Tree):
432
 
                raise NotTreeError(hexsha)
433
 
            try:
434
 
                mode, hexsha = obj[p]
435
 
            except KeyError:
436
 
                raise errors.NoSuchFile(path)
437
 
            if S_ISGITLINK(mode) and i != len(parts) - 1:
438
 
                store = self._get_submodule_store(b'/'.join(parts[:i + 1]))
439
 
                hexsha = store[hexsha].tree
440
 
        return (store, mode, hexsha)
 
364
        try:
 
365
            (mode, hexsha) = tree_lookup_path(
 
366
                self.store.__getitem__, self.tree, path.encode('utf-8'))
 
367
        except KeyError:
 
368
            raise errors.NoSuchFile(self, path)
 
369
        else:
 
370
            return (self.store, mode, hexsha)
441
371
 
442
372
    def is_executable(self, path):
443
373
        (store, mode, hexsha) = self._lookup_path(path)
461
391
        else:
462
392
            return True
463
393
 
464
 
    def list_files(self, include_root=False, from_dir=None, recursive=True,
465
 
                   recurse_nested=False):
 
394
    def list_files(self, include_root=False, from_dir=None, recursive=True):
466
395
        if self.tree is None:
467
396
            return
468
397
        if from_dir is None or from_dir == '.':
472
401
            root_ie = self._get_dir_ie(b"", None)
473
402
        else:
474
403
            parent_path = posixpath.dirname(from_dir)
475
 
            parent_id = self.mapping.generate_file_id(parent_path)
 
404
            parent_id = self._fileid_map.lookup_file_id(parent_path)
476
405
            if mode_kind(mode) == 'directory':
477
 
                root_ie = self._get_dir_ie(encode_git_path(from_dir), parent_id)
 
406
                root_ie = self._get_dir_ie(from_dir.encode("utf-8"), parent_id)
478
407
            else:
479
408
                root_ie = self._get_file_ie(
480
 
                    store, encode_git_path(from_dir),
 
409
                    store, from_dir.encode("utf-8"),
481
410
                    posixpath.basename(from_dir), mode, hexsha)
482
411
        if include_root:
483
412
            yield (from_dir, "V", root_ie.kind, root_ie)
484
413
        todo = []
485
414
        if root_ie.kind == 'directory':
486
 
            todo.append((store, encode_git_path(from_dir),
 
415
            todo.append((store, from_dir.encode("utf-8"),
487
416
                         b"", hexsha, root_ie.file_id))
488
417
        while todo:
489
418
            (store, path, relpath, hexsha, parent_id) = todo.pop()
493
422
                    continue
494
423
                child_path = posixpath.join(path, name)
495
424
                child_relpath = posixpath.join(relpath, name)
496
 
                if S_ISGITLINK(mode) and recurse_nested:
497
 
                    mode = stat.S_IFDIR
498
 
                    store = self._get_submodule_store(child_relpath)
499
 
                    hexsha = store[hexsha].tree
500
425
                if stat.S_ISDIR(mode):
501
426
                    ie = self._get_dir_ie(child_path, parent_id)
502
427
                    if recursive:
506
431
                else:
507
432
                    ie = self._get_file_ie(
508
433
                        store, child_path, name, mode, hexsha, parent_id)
509
 
                yield (decode_git_path(child_relpath), "V", ie.kind, ie)
 
434
                yield (child_relpath.decode('utf-8'), "V", ie.kind, ie)
510
435
 
511
436
    def _get_file_ie(self, store, path, name, mode, hexsha, parent_id):
512
437
        if not isinstance(path, bytes):
514
439
        if not isinstance(name, bytes):
515
440
            raise TypeError(name)
516
441
        kind = mode_kind(mode)
517
 
        path = decode_git_path(path)
518
 
        name = decode_git_path(name)
519
 
        file_id = self.mapping.generate_file_id(path)
 
442
        path = path.decode('utf-8')
 
443
        name = name.decode("utf-8")
 
444
        file_id = self._fileid_map.lookup_file_id(path)
520
445
        ie = entry_factory[kind](file_id, name, parent_id)
521
446
        if kind == 'symlink':
522
 
            ie.symlink_target = decode_git_path(store[hexsha].data)
 
447
            ie.symlink_target = store[hexsha].data.decode('utf-8')
523
448
        elif kind == 'tree-reference':
524
449
            ie.reference_revision = self.mapping.revision_id_foreign_to_bzr(
525
450
                hexsha)
526
451
        else:
527
 
            ie.git_sha1 = hexsha
528
 
            ie.text_size = None
 
452
            data = store[hexsha].data
 
453
            ie.text_sha1 = osutils.sha_string(data)
 
454
            ie.text_size = len(data)
529
455
            ie.executable = mode_is_executable(mode)
530
456
        return ie
531
457
 
532
458
    def _get_dir_ie(self, path, parent_id):
533
 
        path = decode_git_path(path)
534
 
        file_id = self.mapping.generate_file_id(path)
 
459
        path = path.decode('utf-8')
 
460
        file_id = self._fileid_map.lookup_file_id(path)
535
461
        return GitTreeDirectory(file_id, posixpath.basename(path), parent_id)
536
462
 
537
463
    def iter_child_entries(self, path):
540
466
        if mode is not None and not stat.S_ISDIR(mode):
541
467
            return
542
468
 
543
 
        encoded_path = encode_git_path(path)
 
469
        encoded_path = path.encode('utf-8')
544
470
        file_id = self.path2id(path)
545
471
        tree = store[tree_sha]
546
472
        for name, mode, hexsha in tree.iteritems():
553
479
                yield self._get_file_ie(store, child_path, name, mode, hexsha,
554
480
                                        file_id)
555
481
 
556
 
    def iter_entries_by_dir(self, specific_files=None,
557
 
                            recurse_nested=False):
 
482
    def iter_entries_by_dir(self, specific_files=None, yield_parents=False):
558
483
        if self.tree is None:
559
484
            return
 
485
        if yield_parents:
 
486
            # TODO(jelmer): Support yield parents
 
487
            raise NotImplementedError
560
488
        if specific_files is not None:
561
489
            if specific_files in ([""], []):
562
490
                specific_files = None
563
491
            else:
564
 
                specific_files = set([encode_git_path(p)
 
492
                specific_files = set([p.encode('utf-8')
565
493
                                      for p in specific_files])
566
 
        todo = deque([(self.store, b"", self.tree, self.path2id(''))])
 
494
        todo = deque([(self.store, b"", self.tree, self.get_root_id())])
567
495
        if specific_files is None or u"" in specific_files:
568
496
            yield u"", self._get_dir_ie(b"", None)
569
497
        while todo:
574
502
                if self.mapping.is_special_file(name):
575
503
                    continue
576
504
                child_path = posixpath.join(path, name)
577
 
                child_path_decoded = decode_git_path(child_path)
578
 
                if recurse_nested and S_ISGITLINK(mode):
579
 
                    mode = stat.S_IFDIR
580
 
                    store = self._get_submodule_store(child_path)
581
 
                    hexsha = store[hexsha].tree
 
505
                child_path_decoded = child_path.decode('utf-8')
582
506
                if stat.S_ISDIR(mode):
583
507
                    if (specific_files is None or
584
508
                            any([p for p in specific_files if p.startswith(
600
524
        if self.supports_tree_reference():
601
525
            for path, entry in self.iter_entries_by_dir():
602
526
                if entry.kind == 'tree-reference':
603
 
                    yield path
 
527
                    yield path, self.mapping.generate_file_id(b'')
604
528
 
605
529
    def get_revision_id(self):
606
530
        """See RevisionTree.get_revision_id."""
633
557
        """See RevisionTree.get_symlink_target."""
634
558
        (store, mode, hexsha) = self._lookup_path(path)
635
559
        if stat.S_ISLNK(mode):
636
 
            return decode_git_path(store[hexsha].data)
 
560
            return store[hexsha].data.decode('utf-8')
637
561
        else:
638
562
            return None
639
563
 
641
565
        """See RevisionTree.get_symlink_target."""
642
566
        (store, mode, hexsha) = self._lookup_path(path)
643
567
        if S_ISGITLINK(mode):
644
 
            try:
645
 
                nested_repo = self._get_submodule_repository(encode_git_path(path))
646
 
            except errors.NotBranchError:
647
 
                return self.mapping.revision_id_foreign_to_bzr(hexsha)
648
 
            else:
649
 
                return nested_repo.lookup_foreign_revision_id(hexsha)
 
568
            nested_repo = self._get_nested_repository(path)
 
569
            return nested_repo.lookup_foreign_revision_id(hexsha)
650
570
        else:
651
571
            return None
652
572
 
668
588
            return (kind, len(contents), executable,
669
589
                    osutils.sha_string(contents))
670
590
        elif kind == 'symlink':
671
 
            return (kind, None, None, decode_git_path(store[hexsha].data))
 
591
            return (kind, None, None, store[hexsha].data.decode('utf-8'))
672
592
        elif kind == 'tree-reference':
673
 
            nested_repo = self._get_submodule_repository(encode_git_path(path))
 
593
            nested_repo = self._get_nested_repository(path)
674
594
            return (kind, None, None,
675
595
                    nested_repo.lookup_foreign_revision_id(hexsha))
676
596
        else:
677
597
            return (kind, None, None, None)
678
598
 
 
599
    def find_related_paths_across_trees(self, paths, trees=[],
 
600
                                        require_versioned=True):
 
601
        if paths is None:
 
602
            return None
 
603
        if require_versioned:
 
604
            trees = [self] + (trees if trees is not None else [])
 
605
            unversioned = set()
 
606
            for p in paths:
 
607
                for t in trees:
 
608
                    if t.is_versioned(p):
 
609
                        break
 
610
                else:
 
611
                    unversioned.add(p)
 
612
            if unversioned:
 
613
                raise errors.PathsNotVersionedError(unversioned)
 
614
        return filter(self.is_versioned, paths)
 
615
 
679
616
    def _iter_tree_contents(self, include_trees=False):
680
617
        if self.tree is None:
681
618
            return iter([])
709
646
    def walkdirs(self, prefix=u""):
710
647
        (store, mode, hexsha) = self._lookup_path(prefix)
711
648
        todo = deque(
712
 
            [(store, encode_git_path(prefix), hexsha)])
 
649
            [(store, prefix.encode('utf-8'), hexsha, self.path2id(prefix))])
713
650
        while todo:
714
 
            store, path, tree_sha = todo.popleft()
715
 
            path_decoded = decode_git_path(path)
 
651
            store, path, tree_sha, parent_id = todo.popleft()
 
652
            path_decoded = path.decode('utf-8')
716
653
            tree = store[tree_sha]
717
654
            children = []
718
655
            for name, mode, hexsha in tree.iteritems():
719
656
                if self.mapping.is_special_file(name):
720
657
                    continue
721
658
                child_path = posixpath.join(path, name)
 
659
                file_id = self.path2id(child_path.decode('utf-8'))
722
660
                if stat.S_ISDIR(mode):
723
 
                    todo.append((store, child_path, hexsha))
 
661
                    todo.append((store, child_path, hexsha, file_id))
724
662
                children.append(
725
 
                    (decode_git_path(child_path), decode_git_path(name),
 
663
                    (child_path.decode('utf-8'), name.decode('utf-8'),
726
664
                        mode_kind(mode), None,
727
 
                        mode_kind(mode)))
728
 
            yield path_decoded, children
729
 
 
730
 
 
731
 
def tree_delta_from_git_changes(changes, mappings,
732
 
                                specific_files=None,
 
665
                        file_id, mode_kind(mode)))
 
666
            yield (path_decoded, parent_id), children
 
667
 
 
668
 
 
669
def tree_delta_from_git_changes(changes, mapping,
 
670
                                fileid_maps, specific_files=None,
733
671
                                require_versioned=False, include_root=False,
734
 
                                source_extras=None, target_extras=None):
 
672
                                target_extras=None):
735
673
    """Create a TreeDelta from two git trees.
736
674
 
737
675
    source and target are iterators over tuples with:
738
676
        (filename, sha, mode)
739
677
    """
740
 
    (old_mapping, new_mapping) = mappings
 
678
    (old_fileid_map, new_fileid_map) = fileid_maps
741
679
    if target_extras is None:
742
680
        target_extras = set()
743
 
    if source_extras is None:
744
 
        source_extras = set()
745
681
    ret = delta.TreeDelta()
746
682
    added = []
747
 
    for (change_type, old, new) in changes:
748
 
        (oldpath, oldmode, oldsha) = old
749
 
        (newpath, newmode, newsha) = new
 
683
    for (oldpath, newpath), (oldmode, newmode), (oldsha, newsha) in changes:
750
684
        if newpath == b'' and not include_root:
751
685
            continue
752
 
        copied = (change_type == 'copy')
753
 
        if oldpath is not None:
754
 
            oldpath_decoded = decode_git_path(oldpath)
755
 
        else:
 
686
        if oldpath is None:
756
687
            oldpath_decoded = None
757
 
        if newpath is not None:
758
 
            newpath_decoded = decode_git_path(newpath)
759
688
        else:
 
689
            oldpath_decoded = oldpath.decode('utf-8')
 
690
        if newpath is None:
760
691
            newpath_decoded = None
 
692
        else:
 
693
            newpath_decoded = newpath.decode('utf-8')
761
694
        if not (specific_files is None or
762
695
                (oldpath is not None and
763
696
                    osutils.is_inside_or_parent_of_any(
766
699
                    osutils.is_inside_or_parent_of_any(
767
700
                        specific_files, newpath_decoded))):
768
701
            continue
769
 
 
770
 
        if oldpath is None:
771
 
            oldexe = None
772
 
            oldkind = None
773
 
            oldname = None
774
 
            oldparent = None
775
 
            oldversioned = False
776
 
        else:
777
 
            oldversioned = (oldpath not in source_extras)
778
 
            if oldmode:
779
 
                oldexe = mode_is_executable(oldmode)
780
 
                oldkind = mode_kind(oldmode)
781
 
            else:
782
 
                oldexe = False
783
 
                oldkind = None
784
 
            if oldpath == b'':
785
 
                oldparent = None
786
 
                oldname = u''
787
 
            else:
788
 
                (oldparentpath, oldname) = osutils.split(oldpath_decoded)
789
 
                oldparent = old_mapping.generate_file_id(oldparentpath)
790
 
        if newpath is None:
791
 
            newexe = None
792
 
            newkind = None
793
 
            newname = None
794
 
            newparent = None
795
 
            newversioned = False
796
 
        else:
797
 
            newversioned = (newpath not in target_extras)
798
 
            if newmode:
799
 
                newexe = mode_is_executable(newmode)
800
 
                newkind = mode_kind(newmode)
801
 
            else:
802
 
                newexe = False
803
 
                newkind = None
804
 
            if newpath_decoded == u'':
805
 
                newparent = None
806
 
                newname = u''
807
 
            else:
808
 
                newparentpath, newname = osutils.split(newpath_decoded)
809
 
                newparent = new_mapping.generate_file_id(newparentpath)
810
 
        if oldversioned and not copied:
811
 
            fileid = old_mapping.generate_file_id(oldpath_decoded)
812
 
        elif newversioned:
813
 
            fileid = new_mapping.generate_file_id(newpath_decoded)
814
 
        else:
815
 
            fileid = None
816
 
        if old_mapping.is_special_file(oldpath):
 
702
        if mapping.is_special_file(oldpath):
817
703
            oldpath = None
818
 
        if new_mapping.is_special_file(newpath):
 
704
        if mapping.is_special_file(newpath):
819
705
            newpath = None
820
706
        if oldpath is None and newpath is None:
821
707
            continue
822
 
        change = InventoryTreeChange(
823
 
            fileid, (oldpath_decoded, newpath_decoded), (oldsha != newsha),
824
 
            (oldversioned, newversioned),
825
 
            (oldparent, newparent), (oldname, newname),
826
 
            (oldkind, newkind), (oldexe, newexe),
827
 
            copied=copied)
828
 
        if newpath is not None and not newversioned and newkind != 'directory':
829
 
            change.file_id = None
830
 
            ret.unversioned.append(change)
831
 
        elif change_type == 'add':
832
 
            added.append((newpath, newkind))
 
708
        if oldpath is None:
 
709
            added.append((newpath, mode_kind(newmode)))
833
710
        elif newpath is None or newmode == 0:
834
 
            ret.removed.append(change)
835
 
        elif change_type == 'delete':
836
 
            ret.removed.append(change)
837
 
        elif change_type == 'copy':
838
 
            if stat.S_ISDIR(oldmode) and stat.S_ISDIR(newmode):
839
 
                continue
840
 
            ret.copied.append(change)
841
 
        elif change_type == 'rename':
842
 
            if stat.S_ISDIR(oldmode) and stat.S_ISDIR(newmode):
843
 
                continue
844
 
            ret.renamed.append(change)
 
711
            file_id = old_fileid_map.lookup_file_id(oldpath_decoded)
 
712
            ret.removed.append((oldpath_decoded, file_id, mode_kind(oldmode)))
 
713
        elif oldpath != newpath:
 
714
            file_id = old_fileid_map.lookup_file_id(oldpath_decoded)
 
715
            ret.renamed.append(
 
716
                (oldpath_decoded, newpath.decode('utf-8'), file_id,
 
717
                 mode_kind(newmode), (oldsha != newsha),
 
718
                 (oldmode != newmode)))
845
719
        elif mode_kind(oldmode) != mode_kind(newmode):
846
 
            ret.kind_changed.append(change)
 
720
            file_id = new_fileid_map.lookup_file_id(newpath_decoded)
 
721
            ret.kind_changed.append(
 
722
                (newpath_decoded, file_id, mode_kind(oldmode),
 
723
                 mode_kind(newmode)))
847
724
        elif oldsha != newsha or oldmode != newmode:
848
725
            if stat.S_ISDIR(oldmode) and stat.S_ISDIR(newmode):
849
726
                continue
850
 
            ret.modified.append(change)
 
727
            file_id = new_fileid_map.lookup_file_id(newpath_decoded)
 
728
            ret.modified.append(
 
729
                (newpath_decoded, file_id, mode_kind(newmode),
 
730
                 (oldsha != newsha), (oldmode != newmode)))
851
731
        else:
852
 
            ret.unchanged.append(change)
 
732
            file_id = new_fileid_map.lookup_file_id(newpath_decoded)
 
733
            ret.unchanged.append(
 
734
                (newpath_decoded, file_id, mode_kind(newmode)))
853
735
 
854
736
    implicit_dirs = {b''}
855
737
    for path, kind in added:
860
742
    for path, kind in added:
861
743
        if kind == 'directory' and path not in implicit_dirs:
862
744
            continue
863
 
        path_decoded = decode_git_path(path)
864
 
        parent_path, basename = osutils.split(path_decoded)
865
 
        parent_id = new_mapping.generate_file_id(parent_path)
866
 
        file_id = new_mapping.generate_file_id(path_decoded)
867
 
        ret.added.append(
868
 
            InventoryTreeChange(
869
 
                file_id, (None, path_decoded), True,
870
 
                (False, True),
871
 
                (None, parent_id),
872
 
                (None, basename), (None, kind), (None, False)))
 
745
        path_decoded = osutils.normalized_filename(path)[0]
 
746
        if path in target_extras:
 
747
            ret.unversioned.append((path_decoded, None, kind))
 
748
        else:
 
749
            file_id = new_fileid_map.lookup_file_id(path_decoded)
 
750
            ret.added.append((path_decoded, file_id, kind))
873
751
 
874
752
    return ret
875
753
 
876
754
 
877
755
def changes_from_git_changes(changes, mapping, specific_files=None,
878
 
                             include_unchanged=False, source_extras=None,
879
 
                             target_extras=None):
 
756
                             include_unchanged=False, target_extras=None):
880
757
    """Create a iter_changes-like generator from a git stream.
881
758
 
882
759
    source and target are iterators over tuples with:
884
761
    """
885
762
    if target_extras is None:
886
763
        target_extras = set()
887
 
    if source_extras is None:
888
 
        source_extras = set()
889
 
    for (change_type, old, new) in changes:
890
 
        if change_type == 'unchanged' and not include_unchanged:
891
 
            continue
892
 
        (oldpath, oldmode, oldsha) = old
893
 
        (newpath, newmode, newsha) = new
 
764
    for (oldpath, newpath), (oldmode, newmode), (oldsha, newsha) in changes:
894
765
        if oldpath is not None:
895
 
            oldpath_decoded = decode_git_path(oldpath)
 
766
            oldpath_decoded = oldpath.decode('utf-8')
896
767
        else:
897
768
            oldpath_decoded = None
898
769
        if newpath is not None:
899
 
            newpath_decoded = decode_git_path(newpath)
 
770
            newpath_decoded = newpath.decode('utf-8')
900
771
        else:
901
772
            newpath_decoded = None
902
773
        if not (specific_files is None or
911
782
            continue
912
783
        if newpath is not None and mapping.is_special_file(newpath):
913
784
            continue
914
 
        if oldpath is None:
 
785
        if oldpath_decoded is None:
 
786
            fileid = mapping.generate_file_id(newpath_decoded)
915
787
            oldexe = None
916
788
            oldkind = None
917
789
            oldname = None
918
790
            oldparent = None
919
791
            oldversioned = False
920
792
        else:
921
 
            oldversioned = (oldpath not in source_extras)
 
793
            oldversioned = True
922
794
            if oldmode:
923
795
                oldexe = mode_is_executable(oldmode)
924
796
                oldkind = mode_kind(oldmode)
931
803
            else:
932
804
                (oldparentpath, oldname) = osutils.split(oldpath_decoded)
933
805
                oldparent = mapping.generate_file_id(oldparentpath)
934
 
        if newpath is None:
 
806
            fileid = mapping.generate_file_id(oldpath_decoded)
 
807
        if newpath_decoded is None:
935
808
            newexe = None
936
809
            newkind = None
937
810
            newname = None
938
811
            newparent = None
939
812
            newversioned = False
940
813
        else:
941
 
            newversioned = (newpath not in target_extras)
 
814
            newversioned = (newpath_decoded not in target_extras)
942
815
            if newmode:
943
816
                newexe = mode_is_executable(newmode)
944
817
                newkind = mode_kind(newmode)
952
825
                newparentpath, newname = osutils.split(newpath_decoded)
953
826
                newparent = mapping.generate_file_id(newparentpath)
954
827
        if (not include_unchanged and
955
 
                oldkind == 'directory' and newkind == 'directory' and
 
828
            oldkind == 'directory' and newkind == 'directory' and
956
829
                oldpath_decoded == newpath_decoded):
957
830
            continue
958
 
        if oldversioned and change_type != 'copy':
959
 
            fileid = mapping.generate_file_id(oldpath_decoded)
960
 
        elif newversioned:
961
 
            fileid = mapping.generate_file_id(newpath_decoded)
962
 
        else:
963
 
            fileid = None
964
 
        if oldkind == 'directory' and newkind == 'directory':
965
 
            modified = False
966
 
        else:
967
 
            modified = (oldsha != newsha) or (oldmode != newmode)
968
 
        yield InventoryTreeChange(
969
 
            fileid, (oldpath_decoded, newpath_decoded),
970
 
            modified,
971
 
            (oldversioned, newversioned),
972
 
            (oldparent, newparent), (oldname, newname),
973
 
            (oldkind, newkind), (oldexe, newexe),
974
 
            copied=(change_type == 'copy'))
 
831
        yield (fileid, (oldpath_decoded, newpath_decoded), (oldsha != newsha),
 
832
               (oldversioned, newversioned),
 
833
               (oldparent, newparent), (oldname, newname),
 
834
               (oldkind, newkind), (oldexe, newexe))
975
835
 
976
836
 
977
837
class InterGitTrees(_mod_tree.InterTree):
981
841
    _matching_to_tree_format = None
982
842
    _test_mutable_trees_to_test_trees = None
983
843
 
984
 
    def __init__(self, source, target):
985
 
        super(InterGitTrees, self).__init__(source, target)
986
 
        if self.source.store == self.target.store:
987
 
            self.store = self.source.store
988
 
        else:
989
 
            self.store = OverlayObjectStore(
990
 
                [self.source.store, self.target.store])
991
 
        self.rename_detector = RenameDetector(self.store)
992
 
 
993
844
    @classmethod
994
845
    def is_compatible(cls, source, target):
995
 
        return isinstance(source, GitTree) and isinstance(target, GitTree)
 
846
        return (isinstance(source, GitRevisionTree) and
 
847
                isinstance(target, GitRevisionTree))
996
848
 
997
849
    def compare(self, want_unchanged=False, specific_files=None,
998
850
                extra_trees=None, require_versioned=False, include_root=False,
999
851
                want_unversioned=False):
1000
852
        with self.lock_read():
1001
 
            changes, source_extras, target_extras = self._iter_git_changes(
 
853
            changes, target_extras = self._iter_git_changes(
1002
854
                want_unchanged=want_unchanged,
1003
855
                require_versioned=require_versioned,
1004
856
                specific_files=specific_files,
1005
857
                extra_trees=extra_trees,
1006
858
                want_unversioned=want_unversioned)
 
859
            source_fileid_map = self.source._fileid_map
 
860
            target_fileid_map = self.target._fileid_map
1007
861
            return tree_delta_from_git_changes(
1008
 
                changes, (self.source.mapping, self.target.mapping),
 
862
                changes, self.target.mapping,
 
863
                (source_fileid_map, target_fileid_map),
1009
864
                specific_files=specific_files,
1010
 
                include_root=include_root,
1011
 
                source_extras=source_extras, target_extras=target_extras)
 
865
                include_root=include_root, target_extras=target_extras)
1012
866
 
1013
867
    def iter_changes(self, include_unchanged=False, specific_files=None,
1014
868
                     pb=None, extra_trees=[], require_versioned=True,
1015
869
                     want_unversioned=False):
1016
870
        with self.lock_read():
1017
 
            changes, source_extras, target_extras = self._iter_git_changes(
 
871
            changes, target_extras = self._iter_git_changes(
1018
872
                want_unchanged=include_unchanged,
1019
873
                require_versioned=require_versioned,
1020
874
                specific_files=specific_files,
1024
878
                changes, self.target.mapping,
1025
879
                specific_files=specific_files,
1026
880
                include_unchanged=include_unchanged,
1027
 
                source_extras=source_extras,
1028
881
                target_extras=target_extras)
1029
882
 
1030
883
    def _iter_git_changes(self, want_unchanged=False, specific_files=None,
1031
884
                          require_versioned=False, extra_trees=None,
1032
 
                          want_unversioned=False, include_trees=True):
 
885
                          want_unversioned=False):
 
886
        raise NotImplementedError(self._iter_git_changes)
 
887
 
 
888
 
 
889
class InterGitRevisionTrees(InterGitTrees):
 
890
    """InterTree that works between two git revision trees."""
 
891
 
 
892
    _matching_from_tree_format = None
 
893
    _matching_to_tree_format = None
 
894
    _test_mutable_trees_to_test_trees = None
 
895
 
 
896
    @classmethod
 
897
    def is_compatible(cls, source, target):
 
898
        return (isinstance(source, GitRevisionTree) and
 
899
                isinstance(target, GitRevisionTree))
 
900
 
 
901
    def _iter_git_changes(self, want_unchanged=False, specific_files=None,
 
902
                          require_versioned=True, extra_trees=None,
 
903
                          want_unversioned=False):
1033
904
        trees = [self.source]
1034
905
        if extra_trees is not None:
1035
906
            trees.extend(extra_trees)
1037
908
            specific_files = self.target.find_related_paths_across_trees(
1038
909
                specific_files, trees,
1039
910
                require_versioned=require_versioned)
1040
 
        # TODO(jelmer): Restrict to specific_files, for performance reasons.
1041
 
        with self.lock_read():
1042
 
            from_tree_sha, from_extras = self.source.git_snapshot(
1043
 
                want_unversioned=want_unversioned)
1044
 
            to_tree_sha, to_extras = self.target.git_snapshot(
1045
 
                want_unversioned=want_unversioned)
1046
 
            changes = tree_changes(
1047
 
                self.store, from_tree_sha, to_tree_sha,
1048
 
                include_trees=include_trees,
1049
 
                rename_detector=self.rename_detector,
1050
 
                want_unchanged=want_unchanged, change_type_same=True)
1051
 
            return changes, from_extras, to_extras
1052
 
 
1053
 
    def find_target_path(self, path, recurse='none'):
1054
 
        ret = self.find_target_paths([path], recurse=recurse)
1055
 
        return ret[path]
1056
 
 
1057
 
    def find_source_path(self, path, recurse='none'):
1058
 
        ret = self.find_source_paths([path], recurse=recurse)
1059
 
        return ret[path]
1060
 
 
1061
 
    def find_target_paths(self, paths, recurse='none'):
1062
 
        paths = set(paths)
1063
 
        ret = {}
1064
 
        changes = self._iter_git_changes(
1065
 
            specific_files=paths, include_trees=False)[0]
1066
 
        for (change_type, old, new) in changes:
1067
 
            if old[0] is None:
1068
 
                continue
1069
 
            oldpath = decode_git_path(old[0])
1070
 
            if oldpath in paths:
1071
 
                ret[oldpath] = decode_git_path(new[0]) if new[0] else None
1072
 
        for path in paths:
1073
 
            if path not in ret:
1074
 
                if self.source.has_filename(path):
1075
 
                    if self.target.has_filename(path):
1076
 
                        ret[path] = path
1077
 
                    else:
1078
 
                        ret[path] = None
1079
 
                else:
1080
 
                    raise errors.NoSuchFile(path)
1081
 
        return ret
1082
 
 
1083
 
    def find_source_paths(self, paths, recurse='none'):
1084
 
        paths = set(paths)
1085
 
        ret = {}
1086
 
        changes = self._iter_git_changes(
1087
 
            specific_files=paths, include_trees=False)[0]
1088
 
        for (change_type, old, new) in changes:
1089
 
            if new[0] is None:
1090
 
                continue
1091
 
            newpath = decode_git_path(new[0])
1092
 
            if newpath in paths:
1093
 
                ret[newpath] = decode_git_path(old[0]) if old[0] else None
1094
 
        for path in paths:
1095
 
            if path not in ret:
1096
 
                if self.target.has_filename(path):
1097
 
                    if self.source.has_filename(path):
1098
 
                        ret[path] = path
1099
 
                    else:
1100
 
                        ret[path] = None
1101
 
                else:
1102
 
                    raise errors.NoSuchFile(path)
1103
 
        return ret
1104
 
 
1105
 
 
1106
 
_mod_tree.InterTree.register_optimiser(InterGitTrees)
1107
 
 
1108
 
 
1109
 
class MutableGitIndexTree(mutabletree.MutableTree, GitTree):
 
911
 
 
912
        if (self.source._repository._git.object_store !=
 
913
                self.target._repository._git.object_store):
 
914
            store = OverlayObjectStore(
 
915
                [self.source._repository._git.object_store,
 
916
                    self.target._repository._git.object_store])
 
917
        else:
 
918
            store = self.source._repository._git.object_store
 
919
        return store.tree_changes(
 
920
            self.source.tree, self.target.tree, want_unchanged=want_unchanged,
 
921
            include_trees=True, change_type_same=True), set()
 
922
 
 
923
 
 
924
_mod_tree.InterTree.register_optimiser(InterGitRevisionTrees)
 
925
 
 
926
 
 
927
class MutableGitIndexTree(mutabletree.MutableTree):
1110
928
 
1111
929
    def __init__(self):
1112
930
        self._lock_mode = None
1113
931
        self._lock_count = 0
1114
932
        self._versioned_dirs = None
1115
933
        self._index_dirty = False
1116
 
        self._submodules = None
1117
 
 
1118
 
    def git_snapshot(self, want_unversioned=False):
1119
 
        return snapshot_workingtree(self, want_unversioned=want_unversioned)
1120
934
 
1121
935
    def is_versioned(self, path):
1122
936
        with self.lock_read():
1123
 
            path = encode_git_path(path.rstrip('/'))
 
937
            path = path.rstrip('/').encode('utf-8')
1124
938
            (index, subpath) = self._lookup_index(path)
1125
939
            return (subpath in index or self._has_dir(path))
1126
940
 
1137
951
        if self._lock_mode is None:
1138
952
            raise errors.ObjectNotLocked(self)
1139
953
        self._versioned_dirs = set()
1140
 
        for p, sha, mode in self.iter_git_objects():
 
954
        # TODO(jelmer): Browse over all indexes
 
955
        for p, i in self._recurse_index_entries():
1141
956
            self._ensure_versioned_dir(posixpath.dirname(p))
1142
957
 
1143
958
    def _ensure_versioned_dir(self, dirname):
1153
968
        with self.lock_read():
1154
969
            path = path.rstrip('/')
1155
970
            if self.is_versioned(path.rstrip('/')):
1156
 
                return self.mapping.generate_file_id(
 
971
                return self._fileid_map.lookup_file_id(
1157
972
                    osutils.safe_unicode(path))
1158
973
            return None
1159
974
 
1160
 
    def id2path(self, file_id, recurse='down'):
 
975
    def has_id(self, file_id):
 
976
        try:
 
977
            self.id2path(file_id)
 
978
        except errors.NoSuchId:
 
979
            return False
 
980
        else:
 
981
            return True
 
982
 
 
983
    def id2path(self, file_id):
1161
984
        if file_id is None:
1162
985
            return ''
1163
986
        if type(file_id) is not bytes:
1164
987
            raise TypeError(file_id)
1165
988
        with self.lock_read():
1166
989
            try:
1167
 
                path = self.mapping.parse_file_id(file_id)
 
990
                path = self._fileid_map.lookup_path(file_id)
1168
991
            except ValueError:
1169
992
                raise errors.NoSuchId(self, file_id)
1170
993
            if self.is_versioned(path):
1174
997
    def _set_root_id(self, file_id):
1175
998
        raise errors.UnsupportedOperation(self._set_root_id, self)
1176
999
 
 
1000
    def get_root_id(self):
 
1001
        return self.path2id(u"")
 
1002
 
1177
1003
    def _add(self, files, ids, kinds):
1178
1004
        for (path, file_id, kind) in zip(files, ids, kinds):
1179
1005
            if file_id is not None:
1189
1015
    def _lookup_index(self, encoded_path):
1190
1016
        if not isinstance(encoded_path, bytes):
1191
1017
            raise TypeError(encoded_path)
1192
 
        # Common case:
1193
 
        if encoded_path in self.index:
1194
 
            return self.index, encoded_path
1195
 
        # TODO(jelmer): Perhaps have a cache with paths under which some
1196
 
        # submodules exist?
1197
 
        index = self.index
1198
 
        remaining_path = encoded_path
1199
 
        while True:
1200
 
            parts = remaining_path.split(b'/')
1201
 
            for i in range(1, len(parts)):
1202
 
                basepath = b'/'.join(parts[:i])
1203
 
                try:
1204
 
                    (ctime, mtime, dev, ino, mode, uid, gid, size, sha,
1205
 
                     flags) = index[basepath]
1206
 
                except KeyError:
1207
 
                    continue
1208
 
                else:
1209
 
                    if S_ISGITLINK(mode):
1210
 
                        index = self._get_submodule_index(basepath)
1211
 
                        remaining_path = b'/'.join(parts[i:])
1212
 
                        break
1213
 
                    else:
1214
 
                        return index, remaining_path
1215
 
            else:
1216
 
                return index, remaining_path
1217
 
        return index, remaining_path
 
1018
        # TODO(jelmer): Look in other indexes
 
1019
        return self.index, encoded_path
1218
1020
 
1219
1021
    def _index_del_entry(self, index, path):
1220
1022
        del index[path]
1221
1023
        # TODO(jelmer): Keep track of dirty per index
1222
1024
        self._index_dirty = True
1223
1025
 
1224
 
    def _apply_index_changes(self, changes):
1225
 
        for (path, kind, executability, reference_revision,
1226
 
             symlink_target) in changes:
1227
 
            if kind is None or kind == 'directory':
1228
 
                (index, subpath) = self._lookup_index(
1229
 
                    encode_git_path(path))
1230
 
                try:
1231
 
                    self._index_del_entry(index, subpath)
1232
 
                except KeyError:
1233
 
                    pass
1234
 
                else:
1235
 
                    self._versioned_dirs = None
1236
 
            else:
1237
 
                self._index_add_entry(
1238
 
                    path, kind,
1239
 
                    reference_revision=reference_revision,
1240
 
                    symlink_target=symlink_target)
1241
 
        self.flush()
1242
 
 
1243
 
    def _index_add_entry(
1244
 
            self, path, kind, flags=0, reference_revision=None,
1245
 
            symlink_target=None):
 
1026
    def _index_add_entry(self, path, kind, flags=0, reference_revision=None):
1246
1027
        if kind == "directory":
1247
1028
            # Git indexes don't contain directories
1248
1029
            return
1249
 
        elif kind == "file":
 
1030
        if kind == "file":
1250
1031
            blob = Blob()
1251
1032
            try:
1252
1033
                file, stat_val = self.get_file_with_stat(path)
1271
1052
                # old index
1272
1053
                stat_val = os.stat_result(
1273
1054
                    (stat.S_IFLNK, 0, 0, 0, 0, 0, 0, 0, 0, 0))
1274
 
            if symlink_target is None:
1275
 
                symlink_target = self.get_symlink_target(path)
1276
 
            blob.set_raw_string(encode_git_path(symlink_target))
 
1055
            blob.set_raw_string(
 
1056
                self.get_symlink_target(path).encode("utf-8"))
1277
1057
            # Add object to the repository if it didn't exist yet
1278
1058
            if blob.id not in self.store:
1279
1059
                self.store.add_object(blob)
1296
1076
            raise AssertionError("unknown kind '%s'" % kind)
1297
1077
        # Add an entry to the index or update the existing entry
1298
1078
        ensure_normalized_path(path)
1299
 
        encoded_path = encode_git_path(path)
 
1079
        encoded_path = path.encode("utf-8")
1300
1080
        if b'\r' in encoded_path or b'\n' in encoded_path:
1301
1081
            # TODO(jelmer): Why do we need to do this?
1302
1082
            trace.mutter('ignoring path with invalid newline in it: %r', path)
1307
1087
        if self._versioned_dirs is not None:
1308
1088
            self._ensure_versioned_dir(index_path)
1309
1089
 
1310
 
    def iter_git_objects(self):
1311
 
        for p, entry in self._recurse_index_entries():
1312
 
            yield p, entry.sha, entry.mode
1313
 
 
1314
 
    def _recurse_index_entries(self, index=None, basepath=b"",
1315
 
                               recurse_nested=False):
 
1090
    def _recurse_index_entries(self, index=None, basepath=b""):
1316
1091
        # Iterate over all index entries
1317
1092
        with self.lock_read():
1318
1093
            if index is None:
1319
1094
                index = self.index
1320
1095
            for path, value in index.items():
 
1096
                yield (posixpath.join(basepath, path), value)
1321
1097
                (ctime, mtime, dev, ino, mode, uid, gid, size, sha,
1322
1098
                 flags) = value
1323
 
                if S_ISGITLINK(mode) and recurse_nested:
1324
 
                    subindex = self._get_submodule_index(path)
1325
 
                    for entry in self._recurse_index_entries(
1326
 
                            index=subindex, basepath=path,
1327
 
                            recurse_nested=recurse_nested):
1328
 
                        yield entry
1329
 
                else:
1330
 
                    yield (posixpath.join(basepath, path), value)
 
1099
                if S_ISGITLINK(mode):
 
1100
                    pass  # TODO(jelmer): dive into submodule
1331
1101
 
1332
 
    def iter_entries_by_dir(self, specific_files=None,
1333
 
                            recurse_nested=False):
 
1102
    def iter_entries_by_dir(self, specific_files=None, yield_parents=False):
 
1103
        if yield_parents:
 
1104
            raise NotImplementedError(self.iter_entries_by_dir)
1334
1105
        with self.lock_read():
1335
1106
            if specific_files is not None:
1336
1107
                specific_files = set(specific_files)
1341
1112
            if specific_files is None or u"" in specific_files:
1342
1113
                ret[(u"", u"")] = root_ie
1343
1114
            dir_ids = {u"": root_ie.file_id}
1344
 
            for path, value in self._recurse_index_entries(
1345
 
                    recurse_nested=recurse_nested):
 
1115
            for path, value in self._recurse_index_entries():
1346
1116
                if self.mapping.is_special_file(path):
1347
1117
                    continue
1348
 
                path = decode_git_path(path)
 
1118
                path = path.decode("utf-8")
1349
1119
                if specific_files is not None and path not in specific_files:
1350
1120
                    continue
1351
1121
                (parent, name) = posixpath.split(path)
1353
1123
                    file_ie = self._get_file_ie(name, path, value, None)
1354
1124
                except errors.NoSuchFile:
1355
1125
                    continue
1356
 
                if specific_files is None:
 
1126
                if yield_parents or specific_files is None:
1357
1127
                    for (dir_path, dir_ie) in self._add_missing_parent_ids(
1358
1128
                            parent, dir_ids):
1359
1129
                        ret[(posixpath.dirname(dir_path), dir_path)] = dir_ie
1360
1130
                file_ie.parent_id = self.path2id(parent)
1361
1131
                ret[(posixpath.dirname(path), path)] = file_ie
1362
 
            # Special casing for directories
1363
 
            if specific_files:
1364
 
                for path in specific_files:
1365
 
                    key = (posixpath.dirname(path), path)
1366
 
                    if key not in ret and self.is_versioned(path):
1367
 
                        ret[key] = self._get_dir_ie(path, self.path2id(key[0]))
1368
 
            return ((path, ie) for ((_, path), ie) in sorted(ret.items()))
 
1132
            return ((path, ie) for ((_, path), ie) in sorted(viewitems(ret)))
1369
1133
 
1370
1134
    def iter_references(self):
1371
 
        if self.supports_tree_reference():
1372
 
            # TODO(jelmer): Implement a more efficient version of this
1373
 
            for path, entry in self.iter_entries_by_dir():
1374
 
                if entry.kind == 'tree-reference':
1375
 
                    yield path
 
1135
        # TODO(jelmer): Implement a more efficient version of this
 
1136
        for path, entry in self.iter_entries_by_dir():
 
1137
            if entry.kind == 'tree-reference':
 
1138
                yield path, self.mapping.generate_file_id(b'')
1376
1139
 
1377
1140
    def _get_dir_ie(self, path, parent_id):
1378
1141
        file_id = self.path2id(path)
1380
1143
                                posixpath.basename(path).strip("/"), parent_id)
1381
1144
 
1382
1145
    def _get_file_ie(self, name, path, value, parent_id):
1383
 
        if not isinstance(name, str):
 
1146
        if not isinstance(name, text_type):
1384
1147
            raise TypeError(name)
1385
 
        if not isinstance(path, str):
 
1148
        if not isinstance(path, text_type):
1386
1149
            raise TypeError(path)
1387
1150
        if not isinstance(value, tuple) or len(value) != 10:
1388
1151
            raise TypeError(value)
1397
1160
        elif kind == 'tree-reference':
1398
1161
            ie.reference_revision = self.get_reference_revision(path)
1399
1162
        else:
1400
 
            ie.git_sha1 = sha
1401
 
            ie.text_size = size
 
1163
            try:
 
1164
                data = self.get_file_text(path)
 
1165
            except errors.NoSuchFile:
 
1166
                data = None
 
1167
            except IOError as e:
 
1168
                if e.errno != errno.ENOENT:
 
1169
                    raise
 
1170
                data = None
 
1171
            if data is None:
 
1172
                data = self.branch.repository._git.object_store[sha].data
 
1173
            ie.text_sha1 = osutils.sha_string(data)
 
1174
            ie.text_size = len(data)
1402
1175
            ie.executable = bool(stat.S_ISREG(mode) and stat.S_IEXEC & mode)
1403
1176
        return ie
1404
1177
 
1421
1194
    def _unversion_path(self, path):
1422
1195
        if self._lock_mode is None:
1423
1196
            raise errors.ObjectNotLocked(self)
1424
 
        encoded_path = encode_git_path(path)
 
1197
        encoded_path = path.encode("utf-8")
1425
1198
        count = 0
1426
1199
        (index, subpath) = self._lookup_index(encoded_path)
1427
1200
        try:
1454
1227
        for (old_path, new_path, file_id, ie) in delta:
1455
1228
            if old_path is not None:
1456
1229
                (index, old_subpath) = self._lookup_index(
1457
 
                    encode_git_path(old_path))
 
1230
                    old_path.encode('utf-8'))
1458
1231
                if old_subpath in index:
1459
1232
                    self._index_del_entry(index, old_subpath)
1460
1233
                    self._versioned_dirs = None
1480
1253
            return rename_tuples
1481
1254
 
1482
1255
    def rename_one(self, from_rel, to_rel, after=None):
1483
 
        from_path = encode_git_path(from_rel)
 
1256
        from_path = from_rel.encode("utf-8")
1484
1257
        to_rel, can_access = osutils.normalized_filename(to_rel)
1485
1258
        if not can_access:
1486
1259
            raise errors.InvalidNormalization(to_rel)
1487
 
        to_path = encode_git_path(to_rel)
 
1260
        to_path = to_rel.encode("utf-8")
1488
1261
        with self.lock_tree_write():
1489
1262
            if not after:
1490
1263
                # Perhaps it's already moved?
1565
1338
            self._versioned_dirs = None
1566
1339
            self.flush()
1567
1340
 
 
1341
    def find_related_paths_across_trees(self, paths, trees=[],
 
1342
                                        require_versioned=True):
 
1343
        if paths is None:
 
1344
            return None
 
1345
 
 
1346
        if require_versioned:
 
1347
            trees = [self] + (trees if trees is not None else [])
 
1348
            unversioned = set()
 
1349
            for p in paths:
 
1350
                for t in trees:
 
1351
                    if t.is_versioned(p):
 
1352
                        break
 
1353
                else:
 
1354
                    unversioned.add(p)
 
1355
            if unversioned:
 
1356
                raise errors.PathsNotVersionedError(unversioned)
 
1357
 
 
1358
        return filter(self.is_versioned, paths)
 
1359
 
1568
1360
    def path_content_summary(self, path):
1569
1361
        """See Tree.path_content_summary."""
1570
1362
        try:
1590
1382
        else:
1591
1383
            return (kind, None, None, None)
1592
1384
 
1593
 
    def stored_kind(self, relpath):
1594
 
        if relpath == '':
1595
 
            return 'directory'
1596
 
        (index, index_path) = self._lookup_index(encode_git_path(relpath))
1597
 
        if index is None:
1598
 
            return None
1599
 
        try:
1600
 
            mode = index[index_path].mode
1601
 
        except KeyError:
1602
 
            for p in index:
1603
 
                if osutils.is_inside(
1604
 
                        decode_git_path(index_path), decode_git_path(p)):
1605
 
                    return 'directory'
1606
 
            return None
1607
 
        else:
1608
 
            return mode_kind(mode)
1609
 
 
1610
1385
    def kind(self, relpath):
1611
1386
        kind = osutils.file_kind(self.abspath(relpath))
1612
1387
        if kind == 'directory':
1613
 
            if self._directory_is_tree_reference(relpath):
1614
 
                return 'tree-reference'
1615
 
            return 'directory'
 
1388
            (index, index_path) = self._lookup_index(relpath.encode('utf-8'))
 
1389
            if index is None:
 
1390
                return kind
 
1391
            try:
 
1392
                mode = index[index_path].mode
 
1393
            except KeyError:
 
1394
                return kind
 
1395
            else:
 
1396
                if S_ISGITLINK(mode):
 
1397
                    return 'tree-reference'
 
1398
                return 'directory'
1616
1399
        else:
1617
1400
            return kind
1618
1401
 
1619
1402
    def _live_entry(self, relpath):
1620
1403
        raise NotImplementedError(self._live_entry)
1621
1404
 
1622
 
    def transform(self, pb=None):
1623
 
        from .transform import GitTreeTransform
1624
 
        return GitTreeTransform(self, pb=pb)
1625
 
 
1626
 
    def has_changes(self, _from_tree=None):
1627
 
        """Quickly check that the tree contains at least one commitable change.
1628
 
 
1629
 
        :param _from_tree: tree to compare against to find changes (default to
1630
 
            the basis tree and is intended to be used by tests).
1631
 
 
1632
 
        :return: True if a change is found. False otherwise
1633
 
        """
 
1405
 
 
1406
class InterIndexGitTree(InterGitTrees):
 
1407
    """InterTree that works between a Git revision tree and an index."""
 
1408
 
 
1409
    def __init__(self, source, target):
 
1410
        super(InterIndexGitTree, self).__init__(source, target)
 
1411
        self._index = target.index
 
1412
 
 
1413
    @classmethod
 
1414
    def is_compatible(cls, source, target):
 
1415
        return (isinstance(source, GitRevisionTree) and
 
1416
                isinstance(target, MutableGitIndexTree))
 
1417
 
 
1418
    def _iter_git_changes(self, want_unchanged=False, specific_files=None,
 
1419
                          require_versioned=False, extra_trees=None,
 
1420
                          want_unversioned=False):
 
1421
        trees = [self.source]
 
1422
        if extra_trees is not None:
 
1423
            trees.extend(extra_trees)
 
1424
        if specific_files is not None:
 
1425
            specific_files = self.target.find_related_paths_across_trees(
 
1426
                specific_files, trees,
 
1427
                require_versioned=require_versioned)
 
1428
        # TODO(jelmer): Restrict to specific_files, for performance reasons.
1634
1429
        with self.lock_read():
1635
 
            # Check pending merges
1636
 
            if len(self.get_parent_ids()) > 1:
1637
 
                return True
1638
 
            if _from_tree is None:
1639
 
                _from_tree = self.basis_tree()
1640
 
            changes = self.iter_changes(_from_tree)
1641
 
            if self.supports_symlinks():
1642
 
                # Fast path for has_changes.
1643
 
                try:
1644
 
                    change = next(changes)
1645
 
                    if change.path[1] == '':
1646
 
                        next(changes)
1647
 
                    return True
1648
 
                except StopIteration:
1649
 
                    # No changes
1650
 
                    return False
1651
 
            else:
1652
 
                # Slow path for has_changes.
1653
 
                # Handle platforms that do not support symlinks in the
1654
 
                # conditional below. This is slower than the try/except
1655
 
                # approach below that but we don't have a choice as we
1656
 
                # need to be sure that all symlinks are removed from the
1657
 
                # entire changeset. This is because in platforms that
1658
 
                # do not support symlinks, they show up as None in the
1659
 
                # working copy as compared to the repository.
1660
 
                # Also, exclude root as mention in the above fast path.
1661
 
                changes = filter(
1662
 
                    lambda c: c[6][0] != 'symlink' and c[4] != (None, None),
1663
 
                    changes)
1664
 
                try:
1665
 
                    next(iter(changes))
1666
 
                except StopIteration:
1667
 
                    return False
1668
 
                return True
1669
 
 
1670
 
 
1671
 
def snapshot_workingtree(target, want_unversioned=False):
 
1430
            return changes_between_git_tree_and_working_copy(
 
1431
                self.source.store, self.source.tree,
 
1432
                self.target, want_unchanged=want_unchanged,
 
1433
                want_unversioned=want_unversioned)
 
1434
 
 
1435
 
 
1436
_mod_tree.InterTree.register_optimiser(InterIndexGitTree)
 
1437
 
 
1438
 
 
1439
def changes_between_git_tree_and_working_copy(store, from_tree_sha, target,
 
1440
                                              want_unchanged=False,
 
1441
                                              want_unversioned=False):
 
1442
    """Determine the changes between a git tree and a working tree with index.
 
1443
 
 
1444
    """
1672
1445
    extras = set()
1673
1446
    blobs = {}
1674
1447
    # Report dirified directories to commit_tree first, so that they can be
1675
1448
    # replaced with non-empty directories if they have contents.
1676
1449
    dirified = []
1677
 
    trust_executable = target._supports_executable()
1678
1450
    for path, index_entry in target._recurse_index_entries():
1679
1451
        try:
1680
1452
            live_entry = target._live_entry(path)
1682
1454
            if e.errno == errno.ENOENT:
1683
1455
                # Entry was removed; keep it listed, but mark it as gone.
1684
1456
                blobs[path] = (ZERO_SHA, 0)
1685
 
            else:
1686
 
                raise
1687
 
        else:
1688
 
            if live_entry is None:
1689
 
                # Entry was turned into a directory.
1690
 
                # Maybe it's just a submodule that's not checked out?
 
1457
            elif e.errno == errno.EISDIR:
 
1458
                # TODO(jelmer): Only do this if 'path' appears in .gitmodules?
1691
1459
                if S_ISGITLINK(index_entry.mode):
1692
1460
                    blobs[path] = (index_entry.sha, index_entry.mode)
1693
1461
                else:
 
1462
                    # Entry was turned into a directory
1694
1463
                    dirified.append((path, Tree().id, stat.S_IFDIR))
1695
 
                    target.store.add_object(Tree())
 
1464
                    store.add_object(Tree())
1696
1465
            else:
1697
 
                mode = live_entry.mode
1698
 
                if not trust_executable:
1699
 
                    if mode_is_executable(index_entry.mode):
1700
 
                        mode |= 0o111
1701
 
                    else:
1702
 
                        mode &= ~0o111
1703
 
                if live_entry.sha != index_entry.sha:
1704
 
                    rp = decode_git_path(path)
1705
 
                    if stat.S_ISREG(live_entry.mode):
1706
 
                        blob = Blob()
1707
 
                        with target.get_file(rp) as f:
1708
 
                            blob.data = f.read()
1709
 
                    elif stat.S_ISLNK(live_entry.mode):
1710
 
                        blob = Blob()
1711
 
                        blob.data = target.get_symlink_target(rp).encode(osutils._fs_enc)
1712
 
                    else:
1713
 
                        blob = None
1714
 
                    if blob is not None:
1715
 
                        target.store.add_object(blob)
1716
 
                blobs[path] = (live_entry.sha, cleanup_mode(live_entry.mode))
 
1466
                raise
 
1467
        else:
 
1468
            blobs[path] = (live_entry.sha, cleanup_mode(live_entry.mode))
1717
1469
    if want_unversioned:
1718
 
        for extra in target._iter_files_recursive(include_dirs=False):
 
1470
        for e in target.extras():
 
1471
            st = target._lstat(e)
1719
1472
            try:
1720
 
                extra, accessible = osutils.normalized_filename(extra)
 
1473
                np, accessible = osutils.normalized_filename(e)
1721
1474
            except UnicodeDecodeError:
1722
1475
                raise errors.BadFilenameEncoding(
1723
 
                    extra, osutils._fs_enc)
1724
 
            np = encode_git_path(extra)
1725
 
            if np in blobs:
1726
 
                continue
1727
 
            st = target._lstat(extra)
 
1476
                    e, osutils._fs_enc)
1728
1477
            if stat.S_ISDIR(st.st_mode):
1729
1478
                blob = Tree()
1730
 
            elif stat.S_ISREG(st.st_mode) or stat.S_ISLNK(st.st_mode):
 
1479
            else:
1731
1480
                blob = blob_from_path_and_stat(
1732
 
                    target.abspath(extra).encode(osutils._fs_enc), st)
1733
 
            else:
1734
 
                continue
1735
 
            target.store.add_object(blob)
 
1481
                    target.abspath(e).encode(osutils._fs_enc), st)
 
1482
            store.add_object(blob)
 
1483
            np = np.encode('utf-8')
1736
1484
            blobs[np] = (blob.id, cleanup_mode(st.st_mode))
1737
1485
            extras.add(np)
1738
 
    return commit_tree(
1739
 
        target.store, dirified + [(p, s, m) for (p, (s, m)) in blobs.items()]), extras
 
1486
    to_tree_sha = commit_tree(
 
1487
        store, dirified + [(p, s, m) for (p, (s, m)) in blobs.items()])
 
1488
    return store.tree_changes(
 
1489
        from_tree_sha, to_tree_sha, include_trees=True,
 
1490
        want_unchanged=want_unchanged, change_type_same=True), extras