/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: 2018-06-14 17:59:16 UTC
  • mto: This revision was merged to the branch mainline in revision 7065.
  • Revision ID: jelmer@jelmer.uk-20180614175916-a2e2xh5k533guq1x
Move breezy.plugins.git to breezy.git.

Show diffs side-by-side

added added

removed removed

Lines of Context:
17
17
 
18
18
"""Git Trees."""
19
19
 
20
 
from collections import deque
 
20
from __future__ import absolute_import
 
21
 
21
22
import errno
22
23
from io import BytesIO
23
24
import os
24
25
 
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
26
from dulwich.index import (
32
27
    blob_from_path_and_stat,
33
28
    cleanup_mode,
34
29
    commit_tree,
35
30
    index_entry_from_stat,
36
 
    Index,
37
31
    )
38
32
from dulwich.object_store import (
39
33
    tree_lookup_path,
53
47
    controldir as _mod_controldir,
54
48
    delta,
55
49
    errors,
 
50
    lock,
56
51
    mutabletree,
57
52
    osutils,
58
53
    revisiontree,
66
61
    )
67
62
 
68
63
from .mapping import (
69
 
    encode_git_path,
70
 
    decode_git_path,
71
64
    mode_is_executable,
72
65
    mode_kind,
 
66
    GitFileIdMap,
73
67
    default_mapping,
74
68
    )
75
 
from .transportgit import (
76
 
    TransportObjectStore,
77
 
    TransportRepo,
78
 
    )
79
69
 
80
70
 
81
71
class GitTreeDirectory(_mod_tree.TreeDirectory):
141
131
                self.executable == other.executable)
142
132
 
143
133
    def __repr__(self):
144
 
        return ("%s(file_id=%r, name=%r, parent_id=%r, text_size=%r, "
145
 
                "text_sha1=%r, executable=%r)") % (
 
134
        return "%s(file_id=%r, name=%r, parent_id=%r, text_size=%r, text_sha1=%r, executable=%r)" % (
146
135
            type(self).__name__, self.file_id, self.name, self.parent_id,
147
136
            self.text_size, self.text_sha1, self.executable)
148
137
 
149
138
    def copy(self):
150
139
        ret = self.__class__(
151
 
            self.file_id, self.name, self.parent_id)
 
140
                self.file_id, self.name, self.parent_id)
152
141
        ret.text_sha1 = self.text_sha1
153
142
        ret.text_size = self.text_size
154
143
        ret.executable = self.executable
192
181
 
193
182
    def copy(self):
194
183
        return self.__class__(
195
 
            self.file_id, self.name, self.parent_id,
196
 
            self.symlink_target)
197
 
 
198
 
 
199
 
class GitTreeSubmodule(_mod_tree.TreeReference):
 
184
                self.file_id, self.name, self.parent_id,
 
185
                self.symlink_target)
 
186
 
 
187
 
 
188
class GitTreeSubmodule(_mod_tree.TreeLink):
200
189
 
201
190
    __slots__ = ['file_id', 'name', 'parent_id', 'reference_revision']
202
191
 
207
196
        self.reference_revision = reference_revision
208
197
 
209
198
    @property
210
 
    def executable(self):
211
 
        return False
212
 
 
213
 
    @property
214
199
    def kind(self):
215
200
        return 'tree-reference'
216
201
 
217
202
    def __repr__(self):
218
 
        return ("%s(file_id=%r, name=%r, parent_id=%r, "
219
 
                "reference_revision=%r)") % (
 
203
        return "%s(file_id=%r, name=%r, parent_id=%r, reference_revision=%r)" % (
220
204
            type(self).__name__, self.file_id, self.name, self.parent_id,
221
205
            self.reference_revision)
222
206
 
229
213
 
230
214
    def copy(self):
231
215
        return self.__class__(
232
 
            self.file_id, self.name, self.parent_id,
233
 
            self.reference_revision)
 
216
                self.file_id, self.name, self.parent_id,
 
217
                self.reference_revision)
234
218
 
235
219
 
236
220
entry_factory = {
263
247
    def __init__(self, repository, revision_id):
264
248
        self._revision_id = revision_id
265
249
        self._repository = repository
266
 
        self._submodules = None
267
250
        self.store = repository._git.object_store
268
251
        if not isinstance(revision_id, bytes):
269
252
            raise TypeError(revision_id)
270
 
        self.commit_id, self.mapping = repository.lookup_bzr_revision_id(
271
 
            revision_id)
 
253
        self.commit_id, self.mapping = repository.lookup_bzr_revision_id(revision_id)
272
254
        if revision_id == NULL_REVISION:
273
255
            self.tree = None
274
256
            self.mapping = default_mapping
 
257
            self._fileid_map = GitFileIdMap(
 
258
                {},
 
259
                default_mapping)
275
260
        else:
276
261
            try:
277
262
                commit = self.store[self.commit_id]
278
263
            except KeyError:
279
264
                raise errors.NoSuchRevision(repository, revision_id)
280
265
            self.tree = commit.tree
281
 
 
282
 
    def _submodule_info(self):
283
 
        if self._submodules is None:
284
 
            try:
285
 
                with self.get_file('.gitmodules') as f:
286
 
                    config = GitConfigFile.from_file(f)
287
 
                    self._submodules = {
288
 
                        path: (url, section)
289
 
                        for path, url, section in parse_submodules(config)}
290
 
            except errors.NoSuchFile:
291
 
                self._submodules = {}
292
 
        return self._submodules
293
 
 
294
 
    def _get_submodule_repository(self, relpath):
295
 
        if not isinstance(relpath, bytes):
296
 
            raise TypeError(relpath)
297
 
        try:
298
 
            info = self._submodule_info()[relpath]
299
 
        except KeyError:
300
 
            nested_repo_transport = self._repository.controldir.user_transport.clone(
301
 
                decode_git_path(relpath))
302
 
        else:
303
 
            nested_repo_transport = self._repository.controldir.control_transport.clone(
304
 
                posixpath.join('modules', decode_git_path(info[1])))
305
 
        nested_controldir = _mod_controldir.ControlDir.open_from_transport(
306
 
            nested_repo_transport)
 
266
            self._fileid_map = self.mapping.get_fileid_map(self.store.__getitem__, self.tree)
 
267
 
 
268
    def _get_nested_repository(self, path):
 
269
        nested_repo_transport = self._repository.user_transport.clone(path)
 
270
        nested_controldir = _mod_controldir.ControlDir.open_from_transport(nested_repo_transport)
307
271
        return nested_controldir.find_repository()
308
272
 
309
 
    def _get_submodule_store(self, relpath):
310
 
        return self._get_submodule_repository(relpath)._git.object_store
311
 
 
312
 
    def get_nested_tree(self, path):
313
 
        encoded_path = encode_git_path(path)
314
 
        nested_repo = self._get_submodule_repository(encoded_path)
315
 
        ref_rev = self.get_reference_revision(path)
316
 
        return nested_repo.revision_tree(ref_rev)
317
 
 
318
273
    def supports_rename_tracking(self):
319
274
        return False
320
275
 
321
 
    def get_file_revision(self, path):
 
276
    def get_file_revision(self, path, file_id=None):
322
277
        change_scanner = self._repository._file_change_scanner
323
278
        if self.commit_id == ZERO_SHA:
324
279
            return NULL_REVISION
325
 
        (unused_path, commit_id) = change_scanner.find_last_change_revision(
326
 
            encode_git_path(path), self.commit_id)
327
 
        return self._repository.lookup_foreign_revision_id(
328
 
            commit_id, self.mapping)
 
280
        (path, commit_id) = change_scanner.find_last_change_revision(
 
281
            path.encode('utf-8'), self.commit_id)
 
282
        return self._repository.lookup_foreign_revision_id(commit_id, self.mapping)
329
283
 
330
 
    def get_file_mtime(self, path):
 
284
    def get_file_mtime(self, path, file_id=None):
331
285
        try:
332
 
            revid = self.get_file_revision(path)
 
286
            revid = self.get_file_revision(path, file_id)
333
287
        except KeyError:
334
 
            raise errors.NoSuchFile(path)
 
288
            raise _mod_tree.FileTimestampUnavailable(path)
335
289
        try:
336
290
            rev = self._repository.get_revision(revid)
337
291
        except errors.NoSuchRevision:
338
292
            raise _mod_tree.FileTimestampUnavailable(path)
339
293
        return rev.timestamp
340
294
 
341
 
    def id2path(self, file_id, recurse='down'):
 
295
    def id2path(self, file_id):
342
296
        try:
343
 
            path = self.mapping.parse_file_id(file_id)
 
297
            path = self._fileid_map.lookup_path(file_id)
344
298
        except ValueError:
345
299
            raise errors.NoSuchId(self, file_id)
 
300
        path = path.decode('utf-8')
346
301
        if self.is_versioned(path):
347
302
            return path
348
303
        raise errors.NoSuchId(self, file_id)
353
308
    def path2id(self, path):
354
309
        if self.mapping.is_special_file(path):
355
310
            return None
356
 
        if not self.is_versioned(path):
357
 
            return None
358
 
        return self.mapping.generate_file_id(osutils.safe_unicode(path))
 
311
        return self._fileid_map.lookup_file_id(path.encode('utf-8'))
359
312
 
360
313
    def all_file_ids(self):
361
 
        raise errors.UnsupportedOperation(self.all_file_ids, self)
 
314
        return set(self._fileid_map.all_file_ids())
362
315
 
363
316
    def all_versioned_paths(self):
364
 
        ret = {u''}
365
 
        todo = [(self.store, b'', self.tree)]
 
317
        ret = set()
 
318
        todo = set([(store, '', self.tree)])
366
319
        while todo:
367
320
            (store, path, tree_id) = todo.pop()
368
321
            if tree_id is None:
370
323
            tree = store[tree_id]
371
324
            for name, mode, hexsha in tree.items():
372
325
                subpath = posixpath.join(path, name)
373
 
                ret.add(decode_git_path(subpath))
374
326
                if stat.S_ISDIR(mode):
375
 
                    todo.append((store, subpath, hexsha))
 
327
                    todo.add((store, subpath, hexsha))
 
328
                else:
 
329
                    ret.add(subpath)
376
330
        return ret
377
331
 
 
332
    def get_root_id(self):
 
333
        if self.tree is None:
 
334
            return None
 
335
        return self.path2id("")
 
336
 
 
337
    def has_or_had_id(self, file_id):
 
338
        try:
 
339
            path = self.id2path(file_id)
 
340
        except errors.NoSuchId:
 
341
            return False
 
342
        return True
 
343
 
 
344
    def has_id(self, file_id):
 
345
        try:
 
346
            path = self.id2path(file_id)
 
347
        except errors.NoSuchId:
 
348
            return False
 
349
        return self.has_filename(path)
 
350
 
378
351
    def _lookup_path(self, path):
379
352
        if self.tree is None:
380
353
            raise errors.NoSuchFile(path)
381
 
 
382
 
        encoded_path = encode_git_path(path)
383
 
        parts = encoded_path.split(b'/')
384
 
        hexsha = self.tree
385
 
        store = self.store
386
 
        mode = None
387
 
        for i, p in enumerate(parts):
388
 
            if not p:
389
 
                continue
390
 
            obj = store[hexsha]
391
 
            if not isinstance(obj, Tree):
392
 
                raise NotTreeError(hexsha)
393
 
            try:
394
 
                mode, hexsha = obj[p]
395
 
            except KeyError:
396
 
                raise errors.NoSuchFile(path)
397
 
            if S_ISGITLINK(mode) and i != len(parts) - 1:
398
 
                store = self._get_submodule_store(b'/'.join(parts[:i + 1]))
399
 
                hexsha = store[hexsha].tree
400
 
        return (store, mode, hexsha)
401
 
 
402
 
    def is_executable(self, path):
 
354
        try:
 
355
            (mode, hexsha) = tree_lookup_path(self.store.__getitem__, self.tree,
 
356
                path.encode('utf-8'))
 
357
        except KeyError:
 
358
            raise errors.NoSuchFile(self, path)
 
359
        else:
 
360
            return (self.store, mode, hexsha)
 
361
 
 
362
    def is_executable(self, path, file_id=None):
403
363
        (store, mode, hexsha) = self._lookup_path(path)
404
364
        if mode is None:
405
365
            # the tree root is a directory
406
366
            return False
407
367
        return mode_is_executable(mode)
408
368
 
409
 
    def kind(self, path):
 
369
    def kind(self, path, file_id=None):
410
370
        (store, mode, hexsha) = self._lookup_path(path)
411
371
        if mode is None:
412
372
            # the tree root is a directory
421
381
        else:
422
382
            return True
423
383
 
424
 
    def _submodule_info(self):
425
 
        if self._submodules is None:
426
 
            try:
427
 
                with self.get_file('.gitmodules') as f:
428
 
                    config = GitConfigFile.from_file(f)
429
 
                    self._submodules = {
430
 
                        path: (url, section)
431
 
                        for path, url, section in parse_submodules(config)}
432
 
            except errors.NoSuchFile:
433
 
                self._submodules = {}
434
 
        return self._submodules
435
 
 
436
 
    def list_files(self, include_root=False, from_dir=None, recursive=True,
437
 
                   recurse_nested=False):
 
384
    def list_files(self, include_root=False, from_dir=None, recursive=True):
438
385
        if self.tree is None:
439
386
            return
440
 
        if from_dir is None or from_dir == '.':
 
387
        if from_dir is None:
441
388
            from_dir = u""
442
389
        (store, mode, hexsha) = self._lookup_path(from_dir)
443
 
        if mode is None:  # Root
 
390
        if mode is None: # Root
444
391
            root_ie = self._get_dir_ie(b"", None)
445
392
        else:
446
 
            parent_path = posixpath.dirname(from_dir)
447
 
            parent_id = self.mapping.generate_file_id(parent_path)
 
393
            parent_path = posixpath.dirname(from_dir.encode("utf-8"))
 
394
            parent_id = self._fileid_map.lookup_file_id(parent_path)
448
395
            if mode_kind(mode) == 'directory':
449
 
                root_ie = self._get_dir_ie(encode_git_path(from_dir), parent_id)
 
396
                root_ie = self._get_dir_ie(from_dir.encode("utf-8"), parent_id)
450
397
            else:
451
 
                root_ie = self._get_file_ie(
452
 
                    store, encode_git_path(from_dir),
 
398
                root_ie = self._get_file_ie(store, from_dir.encode("utf-8"),
453
399
                    posixpath.basename(from_dir), mode, hexsha)
454
 
        if include_root:
455
 
            yield (from_dir, "V", root_ie.kind, root_ie)
456
 
        todo = []
 
400
        if from_dir != "" or include_root:
 
401
            yield (from_dir, "V", root_ie.kind, root_ie.file_id, root_ie)
 
402
        todo = set()
457
403
        if root_ie.kind == 'directory':
458
 
            todo.append((store, encode_git_path(from_dir),
459
 
                         b"", hexsha, root_ie.file_id))
 
404
            todo.add((store, from_dir.encode("utf-8"), hexsha, root_ie.file_id))
460
405
        while todo:
461
 
            (store, path, relpath, hexsha, parent_id) = todo.pop()
 
406
            (store, path, hexsha, parent_id) = todo.pop()
462
407
            tree = store[hexsha]
463
408
            for name, mode, hexsha in tree.iteritems():
464
409
                if self.mapping.is_special_file(name):
465
410
                    continue
466
411
                child_path = posixpath.join(path, name)
467
 
                child_relpath = posixpath.join(relpath, name)
468
 
                if S_ISGITLINK(mode) and recurse_nested:
469
 
                    mode = stat.S_IFDIR
470
 
                    store = self._get_submodule_store(child_relpath)
471
 
                    hexsha = store[hexsha].tree
472
412
                if stat.S_ISDIR(mode):
473
413
                    ie = self._get_dir_ie(child_path, parent_id)
474
414
                    if recursive:
475
 
                        todo.append(
476
 
                            (store, child_path, child_relpath, hexsha,
477
 
                             ie.file_id))
 
415
                        todo.add((store, child_path, hexsha, ie.file_id))
478
416
                else:
479
 
                    ie = self._get_file_ie(
480
 
                        store, child_path, name, mode, hexsha, parent_id)
481
 
                yield (decode_git_path(child_relpath), "V", ie.kind, ie)
 
417
                    ie = self._get_file_ie(store, child_path, name, mode, hexsha, parent_id)
 
418
                yield child_path.decode('utf-8'), "V", ie.kind, ie.file_id, ie
482
419
 
483
420
    def _get_file_ie(self, store, path, name, mode, hexsha, parent_id):
484
 
        if not isinstance(path, bytes):
 
421
        if type(path) is not bytes:
485
422
            raise TypeError(path)
486
 
        if not isinstance(name, bytes):
 
423
        if type(name) is not bytes:
487
424
            raise TypeError(name)
488
425
        kind = mode_kind(mode)
489
 
        path = decode_git_path(path)
490
 
        name = decode_git_path(name)
491
 
        file_id = self.mapping.generate_file_id(path)
492
 
        ie = entry_factory[kind](file_id, name, parent_id)
 
426
        file_id = self._fileid_map.lookup_file_id(path)
 
427
        ie = entry_factory[kind](file_id, name.decode("utf-8"), parent_id)
493
428
        if kind == 'symlink':
494
 
            ie.symlink_target = decode_git_path(store[hexsha].data)
 
429
            ie.symlink_target = store[hexsha].data.decode('utf-8')
495
430
        elif kind == 'tree-reference':
496
 
            ie.reference_revision = self.mapping.revision_id_foreign_to_bzr(
497
 
                hexsha)
 
431
            ie.reference_revision = self.mapping.revision_id_foreign_to_bzr(hexsha)
498
432
        else:
499
433
            data = store[hexsha].data
500
434
            ie.text_sha1 = osutils.sha_string(data)
503
437
        return ie
504
438
 
505
439
    def _get_dir_ie(self, path, parent_id):
506
 
        path = decode_git_path(path)
507
 
        file_id = self.mapping.generate_file_id(path)
508
 
        return GitTreeDirectory(file_id, posixpath.basename(path), parent_id)
 
440
        file_id = self._fileid_map.lookup_file_id(path)
 
441
        return GitTreeDirectory(file_id,
 
442
            posixpath.basename(path).decode("utf-8"), parent_id)
509
443
 
510
 
    def iter_child_entries(self, path):
 
444
    def iter_child_entries(self, path, file_id=None):
511
445
        (store, mode, tree_sha) = self._lookup_path(path)
512
446
 
513
 
        if mode is not None and not stat.S_ISDIR(mode):
 
447
        if not stat.S_ISDIR(mode):
514
448
            return
515
449
 
516
 
        encoded_path = encode_git_path(path)
 
450
        encoded_path = path.encode('utf-8')
517
451
        file_id = self.path2id(path)
518
452
        tree = store[tree_sha]
519
453
        for name, mode, hexsha in tree.iteritems():
526
460
                yield self._get_file_ie(store, child_path, name, mode, hexsha,
527
461
                                        file_id)
528
462
 
529
 
    def iter_entries_by_dir(self, specific_files=None,
530
 
                            recurse_nested=False):
 
463
    def iter_entries_by_dir(self, specific_files=None, yield_parents=False):
531
464
        if self.tree is None:
532
465
            return
 
466
        if yield_parents:
 
467
            # TODO(jelmer): Support yield parents
 
468
            raise NotImplementedError
533
469
        if specific_files is not None:
534
470
            if specific_files in ([""], []):
535
471
                specific_files = None
536
472
            else:
537
 
                specific_files = set([encode_git_path(p)
538
 
                                      for p in specific_files])
539
 
        todo = deque([(self.store, b"", self.tree, self.path2id(''))])
540
 
        if specific_files is None or u"" in specific_files:
541
 
            yield u"", self._get_dir_ie(b"", None)
 
473
                specific_files = set([p.encode('utf-8') for p in specific_files])
 
474
        todo = set([(self.store, "", self.tree, None)])
542
475
        while todo:
543
 
            store, path, tree_sha, parent_id = todo.popleft()
 
476
            store, path, tree_sha, parent_id = todo.pop()
 
477
            ie = self._get_dir_ie(path, parent_id)
 
478
            if specific_files is None or path in specific_files:
 
479
                yield path.decode("utf-8"), ie
544
480
            tree = store[tree_sha]
545
 
            extradirs = []
546
481
            for name, mode, hexsha in tree.iteritems():
547
482
                if self.mapping.is_special_file(name):
548
483
                    continue
549
484
                child_path = posixpath.join(path, name)
550
 
                child_path_decoded = decode_git_path(child_path)
551
 
                if recurse_nested and S_ISGITLINK(mode):
552
 
                    mode = stat.S_IFDIR
553
 
                    store = self._get_submodule_store(child_path)
554
 
                    hexsha = store[hexsha].tree
555
485
                if stat.S_ISDIR(mode):
556
486
                    if (specific_files is None or
557
 
                            any([p for p in specific_files if p.startswith(
558
 
                                child_path)])):
559
 
                        extradirs.append(
560
 
                            (store, child_path, hexsha,
561
 
                             self.path2id(child_path_decoded)))
562
 
                if specific_files is None or child_path in specific_files:
563
 
                    if stat.S_ISDIR(mode):
564
 
                        yield (child_path_decoded,
565
 
                               self._get_dir_ie(child_path, parent_id))
566
 
                    else:
567
 
                        yield (child_path_decoded,
568
 
                               self._get_file_ie(store, child_path, name, mode,
569
 
                                                 hexsha, parent_id))
570
 
            todo.extendleft(reversed(extradirs))
571
 
 
572
 
    def iter_references(self):
573
 
        if self.supports_tree_reference():
574
 
            for path, entry in self.iter_entries_by_dir():
575
 
                if entry.kind == 'tree-reference':
576
 
                    yield path
 
487
                        any(filter(lambda p: p.startswith(child_path), specific_files))):
 
488
                        todo.add((store, child_path, hexsha, ie.file_id))
 
489
                elif specific_files is None or child_path in specific_files:
 
490
                    yield (child_path.decode("utf-8"),
 
491
                            self._get_file_ie(store, child_path, name, mode, hexsha,
 
492
                           ie.file_id))
577
493
 
578
494
    def get_revision_id(self):
579
495
        """See RevisionTree.get_revision_id."""
580
496
        return self._revision_id
581
497
 
582
 
    def get_file_sha1(self, path, stat_value=None):
 
498
    def get_file_sha1(self, path, file_id=None, stat_value=None):
583
499
        if self.tree is None:
584
500
            raise errors.NoSuchFile(path)
585
 
        return osutils.sha_string(self.get_file_text(path))
 
501
        return osutils.sha_string(self.get_file_text(path, file_id))
586
502
 
587
 
    def get_file_verifier(self, path, stat_value=None):
 
503
    def get_file_verifier(self, path, file_id=None, stat_value=None):
588
504
        (store, mode, hexsha) = self._lookup_path(path)
589
505
        return ("GIT", hexsha)
590
506
 
591
 
    def get_file_size(self, path):
592
 
        (store, mode, hexsha) = self._lookup_path(path)
593
 
        if stat.S_ISREG(mode):
594
 
            return len(store[hexsha].data)
595
 
        return None
596
 
 
597
 
    def get_file_text(self, path):
 
507
    def get_file_text(self, path, file_id=None):
598
508
        """See RevisionTree.get_file_text."""
599
509
        (store, mode, hexsha) = self._lookup_path(path)
600
510
        if stat.S_ISREG(mode):
602
512
        else:
603
513
            return b""
604
514
 
605
 
    def get_symlink_target(self, path):
 
515
    def get_symlink_target(self, path, file_id=None):
606
516
        """See RevisionTree.get_symlink_target."""
607
517
        (store, mode, hexsha) = self._lookup_path(path)
608
518
        if stat.S_ISLNK(mode):
609
 
            return decode_git_path(store[hexsha].data)
 
519
            return store[hexsha].data.decode('utf-8')
610
520
        else:
611
521
            return None
612
522
 
613
 
    def get_reference_revision(self, path):
 
523
    def get_reference_revision(self, path, file_id=None):
614
524
        """See RevisionTree.get_symlink_target."""
615
525
        (store, mode, hexsha) = self._lookup_path(path)
616
526
        if S_ISGITLINK(mode):
617
 
            try:
618
 
                nested_repo = self._get_submodule_repository(encode_git_path(path))
619
 
            except errors.NotBranchError:
620
 
                return self.mapping.revision_id_foreign_to_bzr(hexsha)
621
 
            else:
622
 
                return nested_repo.lookup_foreign_revision_id(hexsha)
 
527
            nested_repo = self._get_nested_repository(path)
 
528
            return nested_repo.lookup_foreign_revision_id(hexsha)
623
529
        else:
624
530
            return None
625
531
 
638
544
        if kind == 'file':
639
545
            executable = mode_is_executable(mode)
640
546
            contents = store[hexsha].data
641
 
            return (kind, len(contents), executable,
642
 
                    osutils.sha_string(contents))
 
547
            return (kind, len(contents), executable, osutils.sha_string(contents))
643
548
        elif kind == 'symlink':
644
 
            return (kind, None, None, decode_git_path(store[hexsha].data))
 
549
            return (kind, None, None, store[hexsha].data)
645
550
        elif kind == 'tree-reference':
646
 
            nested_repo = self._get_submodule_repository(encode_git_path(path))
 
551
            nested_repo = self._get_nested_repository(path)
647
552
            return (kind, None, None,
648
553
                    nested_repo.lookup_foreign_revision_id(hexsha))
649
554
        else:
650
555
            return (kind, None, None, None)
651
556
 
652
557
    def find_related_paths_across_trees(self, paths, trees=[],
653
 
                                        require_versioned=True):
 
558
            require_versioned=True):
654
559
        if paths is None:
655
560
            return None
656
561
        if require_versioned:
670
575
        if self.tree is None:
671
576
            return iter([])
672
577
        return self.store.iter_tree_contents(
673
 
            self.tree, include_trees=include_trees)
 
578
                self.tree, include_trees=include_trees)
674
579
 
675
 
    def annotate_iter(self, path, default_revision=CURRENT_REVISION):
 
580
    def annotate_iter(self, path, file_id=None,
 
581
                      default_revision=CURRENT_REVISION):
676
582
        """Return an iterator of revision_id, line tuples.
677
583
 
678
584
        For working trees (and mutable trees in general), the special
679
585
        revision_id 'current:' will be used for lines that are new in this
680
586
        tree, e.g. uncommitted changes.
 
587
        :param file_id: The file to produce an annotated version from
681
588
        :param default_revision: For lines that don't match a basis, mark them
682
589
            with this revision id. Not all implementations will make use of
683
590
            this value.
693
600
                           for key, line in annotator.annotate_flat(this_key)]
694
601
            return annotations
695
602
 
696
 
    def _get_rules_searcher(self, default_searcher):
697
 
        return default_searcher
698
 
 
699
 
    def walkdirs(self, prefix=u""):
700
 
        (store, mode, hexsha) = self._lookup_path(prefix)
701
 
        todo = deque(
702
 
            [(store, encode_git_path(prefix), hexsha, self.path2id(prefix))])
703
 
        while todo:
704
 
            store, path, tree_sha, parent_id = todo.popleft()
705
 
            path_decoded = decode_git_path(path)
706
 
            tree = store[tree_sha]
707
 
            children = []
708
 
            for name, mode, hexsha in tree.iteritems():
709
 
                if self.mapping.is_special_file(name):
710
 
                    continue
711
 
                child_path = posixpath.join(path, name)
712
 
                file_id = self.path2id(decode_git_path(child_path))
713
 
                if stat.S_ISDIR(mode):
714
 
                    todo.append((store, child_path, hexsha, file_id))
715
 
                children.append(
716
 
                    (decode_git_path(child_path), decode_git_path(name),
717
 
                        mode_kind(mode), None,
718
 
                        file_id, mode_kind(mode)))
719
 
            yield (path_decoded, parent_id), children
720
 
 
721
 
    def preview_transform(self, pb=None):
722
 
        from .transform import GitTransformPreview
723
 
        return GitTransformPreview(self, pb=pb)
724
 
 
725
 
 
726
 
def tree_delta_from_git_changes(changes, mappings,
727
 
                                specific_files=None,
728
 
                                require_versioned=False, include_root=False,
729
 
                                source_extras=None, target_extras=None):
 
603
 
 
604
def tree_delta_from_git_changes(changes, mapping,
 
605
        fileid_maps, specific_files=None,
 
606
        require_versioned=False, include_root=False,
 
607
        target_extras=None):
730
608
    """Create a TreeDelta from two git trees.
731
609
 
732
610
    source and target are iterators over tuples with:
733
611
        (filename, sha, mode)
734
612
    """
735
 
    (old_mapping, new_mapping) = mappings
 
613
    (old_fileid_map, new_fileid_map) = fileid_maps
736
614
    if target_extras is None:
737
615
        target_extras = set()
738
 
    if source_extras is None:
739
 
        source_extras = set()
740
616
    ret = delta.TreeDelta()
741
 
    added = []
742
 
    for (change_type, old, new) in changes:
743
 
        (oldpath, oldmode, oldsha) = old
744
 
        (newpath, newmode, newsha) = new
745
 
        if newpath == b'' and not include_root:
 
617
    for (oldpath, newpath), (oldmode, newmode), (oldsha, newsha) in changes:
 
618
        if newpath == u'' and not include_root:
746
619
            continue
747
 
        copied = (change_type == 'copy')
748
 
        if oldpath is not None:
749
 
            oldpath_decoded = decode_git_path(oldpath)
750
 
        else:
751
 
            oldpath_decoded = None
752
 
        if newpath is not None:
753
 
            newpath_decoded = decode_git_path(newpath)
754
 
        else:
755
 
            newpath_decoded = None
756
620
        if not (specific_files is None or
757
 
                (oldpath is not None and
758
 
                    osutils.is_inside_or_parent_of_any(
759
 
                        specific_files, oldpath_decoded)) or
760
 
                (newpath is not None and
761
 
                    osutils.is_inside_or_parent_of_any(
762
 
                        specific_files, newpath_decoded))):
 
621
                (oldpath is not None and osutils.is_inside_or_parent_of_any(specific_files, oldpath)) or
 
622
                (newpath is not None and osutils.is_inside_or_parent_of_any(specific_files, newpath))):
763
623
            continue
764
 
 
765
 
        if oldpath is None:
766
 
            oldexe = None
767
 
            oldkind = None
768
 
            oldname = None
769
 
            oldparent = None
770
 
            oldversioned = False
771
 
        else:
772
 
            oldversioned = (oldpath not in source_extras)
773
 
            if oldmode:
774
 
                oldexe = mode_is_executable(oldmode)
775
 
                oldkind = mode_kind(oldmode)
776
 
            else:
777
 
                oldexe = False
778
 
                oldkind = None
779
 
            if oldpath == b'':
780
 
                oldparent = None
781
 
                oldname = u''
782
 
            else:
783
 
                (oldparentpath, oldname) = osutils.split(oldpath_decoded)
784
 
                oldparent = old_mapping.generate_file_id(oldparentpath)
785
 
        if newpath is None:
786
 
            newexe = None
787
 
            newkind = None
788
 
            newname = None
789
 
            newparent = None
790
 
            newversioned = False
791
 
        else:
792
 
            newversioned = (newpath not in target_extras)
793
 
            if newmode:
794
 
                newexe = mode_is_executable(newmode)
795
 
                newkind = mode_kind(newmode)
796
 
            else:
797
 
                newexe = False
798
 
                newkind = None
799
 
            if newpath_decoded == u'':
800
 
                newparent = None
801
 
                newname = u''
802
 
            else:
803
 
                newparentpath, newname = osutils.split(newpath_decoded)
804
 
                newparent = new_mapping.generate_file_id(newparentpath)
805
 
        if oldversioned and not copied:
806
 
            fileid = old_mapping.generate_file_id(oldpath_decoded)
807
 
        elif newversioned:
808
 
            fileid = new_mapping.generate_file_id(newpath_decoded)
809
 
        else:
810
 
            fileid = None
811
 
        if old_mapping.is_special_file(oldpath):
 
624
        if mapping.is_special_file(oldpath):
812
625
            oldpath = None
813
 
        if new_mapping.is_special_file(newpath):
 
626
        if mapping.is_special_file(newpath):
814
627
            newpath = None
815
628
        if oldpath is None and newpath is None:
816
629
            continue
817
 
        change = _mod_tree.TreeChange(
818
 
            fileid, (oldpath_decoded, newpath_decoded), (oldsha != newsha),
819
 
            (oldversioned, newversioned),
820
 
            (oldparent, newparent), (oldname, newname),
821
 
            (oldkind, newkind), (oldexe, newexe),
822
 
            copied=copied)
823
 
        if newpath is not None and not newversioned and newkind != 'directory':
824
 
            change.file_id = None
825
 
            ret.unversioned.append(change)
826
 
        elif change_type == 'add':
827
 
            added.append((newpath, newkind))
 
630
        if oldpath is None:
 
631
            if newpath in target_extras:
 
632
                ret.unversioned.append(
 
633
                    (osutils.normalized_filename(newpath)[0], None, mode_kind(newmode)))
 
634
            else:
 
635
                file_id = new_fileid_map.lookup_file_id(newpath)
 
636
                ret.added.append((newpath.decode('utf-8'), file_id, mode_kind(newmode)))
828
637
        elif newpath is None or newmode == 0:
829
 
            ret.removed.append(change)
830
 
        elif change_type == 'delete':
831
 
            ret.removed.append(change)
832
 
        elif change_type == 'copy':
833
 
            if stat.S_ISDIR(oldmode) and stat.S_ISDIR(newmode):
834
 
                continue
835
 
            ret.copied.append(change)
836
 
        elif change_type == 'rename':
837
 
            if stat.S_ISDIR(oldmode) and stat.S_ISDIR(newmode):
838
 
                continue
839
 
            ret.renamed.append(change)
 
638
            file_id = old_fileid_map.lookup_file_id(oldpath)
 
639
            ret.removed.append((oldpath.decode('utf-8'), file_id, mode_kind(oldmode)))
 
640
        elif oldpath != newpath:
 
641
            file_id = old_fileid_map.lookup_file_id(oldpath)
 
642
            ret.renamed.append(
 
643
                (oldpath.decode('utf-8'), newpath.decode('utf-8'), file_id,
 
644
                mode_kind(newmode), (oldsha != newsha),
 
645
                (oldmode != newmode)))
840
646
        elif mode_kind(oldmode) != mode_kind(newmode):
841
 
            ret.kind_changed.append(change)
 
647
            file_id = new_fileid_map.lookup_file_id(newpath)
 
648
            ret.kind_changed.append(
 
649
                (newpath.decode('utf-8'), file_id, mode_kind(oldmode),
 
650
                mode_kind(newmode)))
842
651
        elif oldsha != newsha or oldmode != newmode:
843
652
            if stat.S_ISDIR(oldmode) and stat.S_ISDIR(newmode):
844
653
                continue
845
 
            ret.modified.append(change)
 
654
            file_id = new_fileid_map.lookup_file_id(newpath)
 
655
            ret.modified.append(
 
656
                (newpath.decode('utf-8'), file_id, mode_kind(newmode),
 
657
                (oldsha != newsha), (oldmode != newmode)))
846
658
        else:
847
 
            ret.unchanged.append(change)
848
 
 
849
 
    implicit_dirs = {b''}
850
 
    for path, kind in added:
851
 
        if kind == 'directory' or path in target_extras:
852
 
            continue
853
 
        implicit_dirs.update(osutils.parent_directories(path))
854
 
 
855
 
    for path, kind in added:
856
 
        if kind == 'directory' and path not in implicit_dirs:
857
 
            continue
858
 
        path_decoded = decode_git_path(path)
859
 
        parent_path, basename = osutils.split(path_decoded)
860
 
        parent_id = new_mapping.generate_file_id(parent_path)
861
 
        file_id = new_mapping.generate_file_id(path_decoded)
862
 
        ret.added.append(
863
 
            _mod_tree.TreeChange(
864
 
                file_id, (None, path_decoded), True,
865
 
                (False, True),
866
 
                (None, parent_id),
867
 
                (None, basename), (None, kind), (None, False)))
 
659
            file_id = new_fileid_map.lookup_file_id(newpath)
 
660
            ret.unchanged.append((newpath.decode('utf-8'), file_id, mode_kind(newmode)))
868
661
 
869
662
    return ret
870
663
 
871
664
 
872
 
def changes_from_git_changes(changes, mapping, specific_files=None,
873
 
                             include_unchanged=False, source_extras=None,
 
665
def changes_from_git_changes(changes, mapping, specific_files=None, include_unchanged=False,
874
666
                             target_extras=None):
875
667
    """Create a iter_changes-like generator from a git stream.
876
668
 
879
671
    """
880
672
    if target_extras is None:
881
673
        target_extras = set()
882
 
    if source_extras is None:
883
 
        source_extras = set()
884
 
    for (change_type, old, new) in changes:
885
 
        if change_type == 'unchanged' and not include_unchanged:
886
 
            continue
887
 
        (oldpath, oldmode, oldsha) = old
888
 
        (newpath, newmode, newsha) = new
889
 
        if oldpath is not None:
890
 
            oldpath_decoded = decode_git_path(oldpath)
891
 
        else:
892
 
            oldpath_decoded = None
893
 
        if newpath is not None:
894
 
            newpath_decoded = decode_git_path(newpath)
895
 
        else:
896
 
            newpath_decoded = None
 
674
    for (oldpath, newpath), (oldmode, newmode), (oldsha, newsha) in changes:
897
675
        if not (specific_files is None or
898
 
                (oldpath_decoded is not None and
899
 
                    osutils.is_inside_or_parent_of_any(
900
 
                        specific_files, oldpath_decoded)) or
901
 
                (newpath_decoded is not None and
902
 
                    osutils.is_inside_or_parent_of_any(
903
 
                        specific_files, newpath_decoded))):
 
676
                (oldpath is not None and osutils.is_inside_or_parent_of_any(specific_files, oldpath)) or
 
677
                (newpath is not None and osutils.is_inside_or_parent_of_any(specific_files, newpath))):
904
678
            continue
 
679
        path = (oldpath, newpath)
905
680
        if oldpath is not None and mapping.is_special_file(oldpath):
906
681
            continue
907
682
        if newpath is not None and mapping.is_special_file(newpath):
908
683
            continue
909
684
        if oldpath is None:
 
685
            fileid = mapping.generate_file_id(newpath)
910
686
            oldexe = None
911
687
            oldkind = None
912
688
            oldname = None
913
689
            oldparent = None
914
690
            oldversioned = False
915
691
        else:
916
 
            oldversioned = (oldpath not in source_extras)
 
692
            oldversioned = True
 
693
            oldpath = oldpath.decode("utf-8")
917
694
            if oldmode:
918
695
                oldexe = mode_is_executable(oldmode)
919
696
                oldkind = mode_kind(oldmode)
920
697
            else:
921
698
                oldexe = False
922
699
                oldkind = None
923
 
            if oldpath_decoded == u'':
 
700
            if oldpath == u'':
924
701
                oldparent = None
925
 
                oldname = u''
 
702
                oldname = ''
926
703
            else:
927
 
                (oldparentpath, oldname) = osutils.split(oldpath_decoded)
 
704
                (oldparentpath, oldname) = osutils.split(oldpath)
928
705
                oldparent = mapping.generate_file_id(oldparentpath)
 
706
            fileid = mapping.generate_file_id(oldpath)
929
707
        if newpath is None:
930
708
            newexe = None
931
709
            newkind = None
940
718
            else:
941
719
                newexe = False
942
720
                newkind = None
943
 
            if newpath_decoded == u'':
 
721
            newpath = newpath.decode("utf-8")
 
722
            if newpath == u'':
944
723
                newparent = None
945
724
                newname = u''
946
725
            else:
947
 
                newparentpath, newname = osutils.split(newpath_decoded)
 
726
                newparentpath, newname = osutils.split(newpath)
948
727
                newparent = mapping.generate_file_id(newparentpath)
949
728
        if (not include_unchanged and
950
 
                oldkind == 'directory' and newkind == 'directory' and
951
 
                oldpath_decoded == newpath_decoded):
 
729
            oldkind == 'directory' and newkind == 'directory' and
 
730
            oldpath == newpath):
952
731
            continue
953
 
        if oldversioned and change_type != 'copy':
954
 
            fileid = mapping.generate_file_id(oldpath_decoded)
955
 
        elif newversioned:
956
 
            fileid = mapping.generate_file_id(newpath_decoded)
957
 
        else:
958
 
            fileid = None
959
 
        yield _mod_tree.TreeChange(
960
 
            fileid, (oldpath_decoded, newpath_decoded), (oldsha != newsha),
961
 
            (oldversioned, newversioned),
962
 
            (oldparent, newparent), (oldname, newname),
963
 
            (oldkind, newkind), (oldexe, newexe),
964
 
            copied=(change_type == 'copy'))
 
732
        yield (fileid, (oldpath, newpath), (oldsha != newsha),
 
733
             (oldversioned, newversioned),
 
734
             (oldparent, newparent), (oldname, newname),
 
735
             (oldkind, newkind), (oldexe, newexe))
965
736
 
966
737
 
967
738
class InterGitTrees(_mod_tree.InterTree):
980
751
                extra_trees=None, require_versioned=False, include_root=False,
981
752
                want_unversioned=False):
982
753
        with self.lock_read():
983
 
            changes, source_extras, target_extras = self._iter_git_changes(
984
 
                want_unchanged=want_unchanged,
985
 
                require_versioned=require_versioned,
986
 
                specific_files=specific_files,
987
 
                extra_trees=extra_trees,
988
 
                want_unversioned=want_unversioned)
989
 
            return tree_delta_from_git_changes(
990
 
                changes, (self.source.mapping, self.target.mapping),
991
 
                specific_files=specific_files,
992
 
                include_root=include_root,
993
 
                source_extras=source_extras, target_extras=target_extras)
 
754
            changes, target_extras = self._iter_git_changes(
 
755
                    want_unchanged=want_unchanged,
 
756
                    require_versioned=require_versioned,
 
757
                    specific_files=specific_files,
 
758
                    extra_trees=extra_trees,
 
759
                    want_unversioned=want_unversioned)
 
760
            source_fileid_map = self.source._fileid_map
 
761
            target_fileid_map = self.target._fileid_map
 
762
            return tree_delta_from_git_changes(changes, self.target.mapping,
 
763
                (source_fileid_map, target_fileid_map),
 
764
                specific_files=specific_files, include_root=include_root,
 
765
                target_extras=target_extras)
994
766
 
995
767
    def iter_changes(self, include_unchanged=False, specific_files=None,
996
768
                     pb=None, extra_trees=[], require_versioned=True,
997
769
                     want_unversioned=False):
998
770
        with self.lock_read():
999
 
            changes, source_extras, target_extras = self._iter_git_changes(
1000
 
                want_unchanged=include_unchanged,
1001
 
                require_versioned=require_versioned,
1002
 
                specific_files=specific_files,
1003
 
                extra_trees=extra_trees,
1004
 
                want_unversioned=want_unversioned)
 
771
            changes, target_extras = self._iter_git_changes(
 
772
                    want_unchanged=include_unchanged,
 
773
                    require_versioned=require_versioned,
 
774
                    specific_files=specific_files,
 
775
                    extra_trees=extra_trees,
 
776
                    want_unversioned=want_unversioned)
1005
777
            return changes_from_git_changes(
1006
 
                changes, self.target.mapping,
1007
 
                specific_files=specific_files,
1008
 
                include_unchanged=include_unchanged,
1009
 
                source_extras=source_extras,
1010
 
                target_extras=target_extras)
 
778
                    changes, self.target.mapping,
 
779
                    specific_files=specific_files,
 
780
                    include_unchanged=include_unchanged,
 
781
                    target_extras=target_extras)
1011
782
 
1012
783
    def _iter_git_changes(self, want_unchanged=False, specific_files=None,
1013
 
                          require_versioned=False, extra_trees=None,
1014
 
                          want_unversioned=False, include_trees=True):
 
784
            require_versioned=False, extra_trees=None,
 
785
            want_unversioned=False):
1015
786
        raise NotImplementedError(self._iter_git_changes)
1016
787
 
1017
 
    def find_target_path(self, path, recurse='none'):
1018
 
        ret = self.find_target_paths([path], recurse=recurse)
1019
 
        return ret[path]
1020
 
 
1021
 
    def find_source_path(self, path, recurse='none'):
1022
 
        ret = self.find_source_paths([path], recurse=recurse)
1023
 
        return ret[path]
1024
 
 
1025
 
    def find_target_paths(self, paths, recurse='none'):
1026
 
        paths = set(paths)
1027
 
        ret = {}
1028
 
        changes = self._iter_git_changes(
1029
 
            specific_files=paths, include_trees=False)[0]
1030
 
        for (change_type, old, new) in changes:
1031
 
            if old[0] is None:
1032
 
                continue
1033
 
            oldpath = decode_git_path(old[0])
1034
 
            if oldpath in paths:
1035
 
                ret[oldpath] = decode_git_path(new[0]) if new[0] else None
1036
 
        for path in paths:
1037
 
            if path not in ret:
1038
 
                if self.source.has_filename(path):
1039
 
                    if self.target.has_filename(path):
1040
 
                        ret[path] = path
1041
 
                    else:
1042
 
                        ret[path] = None
1043
 
                else:
1044
 
                    raise errors.NoSuchFile(path)
1045
 
        return ret
1046
 
 
1047
 
    def find_source_paths(self, paths, recurse='none'):
1048
 
        paths = set(paths)
1049
 
        ret = {}
1050
 
        changes = self._iter_git_changes(
1051
 
            specific_files=paths, include_trees=False)[0]
1052
 
        for (change_type, old, new) in changes:
1053
 
            if new[0] is None:
1054
 
                continue
1055
 
            newpath = decode_git_path(new[0])
1056
 
            if newpath in paths:
1057
 
                ret[newpath] = decode_git_path(old[0]) if old[0] else None
1058
 
        for path in paths:
1059
 
            if path not in ret:
1060
 
                if self.target.has_filename(path):
1061
 
                    if self.source.has_filename(path):
1062
 
                        ret[path] = path
1063
 
                    else:
1064
 
                        ret[path] = None
1065
 
                else:
1066
 
                    raise errors.NoSuchFile(path)
1067
 
        return ret
1068
 
 
1069
788
 
1070
789
class InterGitRevisionTrees(InterGitTrees):
1071
790
    """InterTree that works between two git revision trees."""
1080
799
                isinstance(target, GitRevisionTree))
1081
800
 
1082
801
    def _iter_git_changes(self, want_unchanged=False, specific_files=None,
1083
 
                          require_versioned=True, extra_trees=None,
1084
 
                          want_unversioned=False, include_trees=True):
 
802
            require_versioned=True, extra_trees=None,
 
803
            want_unversioned=False):
1085
804
        trees = [self.source]
1086
805
        if extra_trees is not None:
1087
806
            trees.extend(extra_trees)
1088
807
        if specific_files is not None:
1089
808
            specific_files = self.target.find_related_paths_across_trees(
1090
 
                specific_files, trees,
1091
 
                require_versioned=require_versioned)
 
809
                    specific_files, trees,
 
810
                    require_versioned=require_versioned)
1092
811
 
1093
 
        if (self.source._repository._git.object_store !=
1094
 
                self.target._repository._git.object_store):
1095
 
            store = OverlayObjectStore(
1096
 
                [self.source._repository._git.object_store,
1097
 
                    self.target._repository._git.object_store])
 
812
        if self.source._repository._git.object_store != self.target._repository._git.object_store:
 
813
            store = OverlayObjectStore([self.source._repository._git.object_store,
 
814
                                        self.target._repository._git.object_store])
1098
815
        else:
1099
816
            store = self.source._repository._git.object_store
1100
 
        rename_detector = RenameDetector(store)
1101
 
        changes = tree_changes(
1102
 
            store, self.source.tree, self.target.tree,
1103
 
            want_unchanged=want_unchanged, include_trees=include_trees,
1104
 
            change_type_same=True, rename_detector=rename_detector)
1105
 
        return changes, set(), set()
 
817
        return self.source._repository._git.object_store.tree_changes(
 
818
            self.source.tree, self.target.tree, want_unchanged=want_unchanged,
 
819
            include_trees=True, change_type_same=True), set()
1106
820
 
1107
821
 
1108
822
_mod_tree.InterTree.register_optimiser(InterGitRevisionTrees)
1115
829
        self._lock_count = 0
1116
830
        self._versioned_dirs = None
1117
831
        self._index_dirty = False
1118
 
        self._submodules = None
1119
832
 
1120
833
    def is_versioned(self, path):
1121
834
        with self.lock_read():
1122
 
            path = encode_git_path(path.rstrip('/'))
 
835
            path = path.rstrip('/').encode('utf-8')
1123
836
            (index, subpath) = self._lookup_index(path)
1124
837
            return (subpath in index or self._has_dir(path))
1125
838
 
1126
839
    def _has_dir(self, path):
1127
 
        if not isinstance(path, bytes):
1128
 
            raise TypeError(path)
1129
 
        if path == b"":
 
840
        if path == "":
1130
841
            return True
1131
842
        if self._versioned_dirs is None:
1132
843
            self._load_dirs()
1136
847
        if self._lock_mode is None:
1137
848
            raise errors.ObjectNotLocked(self)
1138
849
        self._versioned_dirs = set()
 
850
        # TODO(jelmer): Browse over all indexes
1139
851
        for p, i in self._recurse_index_entries():
1140
852
            self._ensure_versioned_dir(posixpath.dirname(p))
1141
853
 
1142
854
    def _ensure_versioned_dir(self, dirname):
1143
 
        if not isinstance(dirname, bytes):
1144
 
            raise TypeError(dirname)
1145
855
        if dirname in self._versioned_dirs:
1146
856
            return
1147
 
        if dirname != b"":
 
857
        if dirname != "":
1148
858
            self._ensure_versioned_dir(posixpath.dirname(dirname))
1149
859
        self._versioned_dirs.add(dirname)
1150
860
 
1152
862
        with self.lock_read():
1153
863
            path = path.rstrip('/')
1154
864
            if self.is_versioned(path.rstrip('/')):
1155
 
                return self.mapping.generate_file_id(
1156
 
                    osutils.safe_unicode(path))
 
865
                return self._fileid_map.lookup_file_id(path.encode("utf-8"))
1157
866
            return None
1158
867
 
1159
 
    def id2path(self, file_id, recurse='down'):
 
868
    def has_id(self, file_id):
 
869
        try:
 
870
            self.id2path(file_id)
 
871
        except errors.NoSuchId:
 
872
            return False
 
873
        else:
 
874
            return True
 
875
 
 
876
    def id2path(self, file_id):
1160
877
        if file_id is None:
1161
878
            return ''
1162
879
        if type(file_id) is not bytes:
1163
880
            raise TypeError(file_id)
1164
881
        with self.lock_read():
1165
882
            try:
1166
 
                path = self.mapping.parse_file_id(file_id)
 
883
                path = self._fileid_map.lookup_path(file_id)
1167
884
            except ValueError:
1168
885
                raise errors.NoSuchId(self, file_id)
 
886
            path = path.decode('utf-8')
1169
887
            if self.is_versioned(path):
1170
888
                return path
1171
889
            raise errors.NoSuchId(self, file_id)
1172
890
 
1173
891
    def _set_root_id(self, file_id):
1174
 
        raise errors.UnsupportedOperation(self._set_root_id, self)
 
892
        self._fileid_map.set_file_id("", file_id)
 
893
 
 
894
    def get_root_id(self):
 
895
        return self.path2id("")
1175
896
 
1176
897
    def _add(self, files, ids, kinds):
1177
898
        for (path, file_id, kind) in zip(files, ids, kinds):
1185
906
    def _read_submodule_head(self, path):
1186
907
        raise NotImplementedError(self._read_submodule_head)
1187
908
 
1188
 
    def _submodule_info(self):
1189
 
        if self._submodules is None:
1190
 
            try:
1191
 
                with self.get_file('.gitmodules') as f:
1192
 
                    config = GitConfigFile.from_file(f)
1193
 
                    self._submodules = {
1194
 
                        path: (url, section)
1195
 
                        for path, url, section in parse_submodules(config)}
1196
 
            except errors.NoSuchFile:
1197
 
                self._submodules = {}
1198
 
        return self._submodules
1199
 
 
1200
909
    def _lookup_index(self, encoded_path):
1201
910
        if not isinstance(encoded_path, bytes):
1202
911
            raise TypeError(encoded_path)
1203
 
        # Common case:
1204
 
        if encoded_path in self.index:
1205
 
            return self.index, encoded_path
1206
 
        # TODO(jelmer): Perhaps have a cache with paths under which some
1207
 
        # submodules exist?
1208
 
        index = self.index
1209
 
        remaining_path = encoded_path
1210
 
        while True:
1211
 
            parts = remaining_path.split(b'/')
1212
 
            for i in range(1, len(parts)):
1213
 
                basepath = b'/'.join(parts[:i])
1214
 
                try:
1215
 
                    (ctime, mtime, dev, ino, mode, uid, gid, size, sha,
1216
 
                     flags) = index[basepath]
1217
 
                except KeyError:
1218
 
                    continue
1219
 
                else:
1220
 
                    if S_ISGITLINK(mode):
1221
 
                        index = self._get_submodule_index(basepath)
1222
 
                        remaining_path = b'/'.join(parts[i:])
1223
 
                        break
1224
 
                    else:
1225
 
                        return index, remaining_path
1226
 
            else:
1227
 
                return index, remaining_path
1228
 
        return index, remaining_path
 
912
        # TODO(jelmer): Look in other indexes
 
913
        return self.index, encoded_path
1229
914
 
1230
915
    def _index_del_entry(self, index, path):
1231
916
        del index[path]
1241
926
            try:
1242
927
                file, stat_val = self.get_file_with_stat(path)
1243
928
            except (errors.NoSuchFile, IOError):
1244
 
                # TODO: Rather than come up with something here, use the old
1245
 
                # index
 
929
                # TODO: Rather than come up with something here, use the old index
1246
930
                file = BytesIO()
1247
931
                stat_val = os.stat_result(
1248
932
                    (stat.S_IFREG | 0o644, 0, 0, 0, 0, 0, 0, 0, 0, 0))
1249
 
            with file:
1250
 
                blob.set_raw_string(file.read())
 
933
            blob.set_raw_string(file.read())
1251
934
            # Add object to the repository if it didn't exist yet
1252
 
            if blob.id not in self.store:
 
935
            if not blob.id in self.store:
1253
936
                self.store.add_object(blob)
1254
937
            hexsha = blob.id
1255
938
        elif kind == "symlink":
1261
944
                # old index
1262
945
                stat_val = os.stat_result(
1263
946
                    (stat.S_IFLNK, 0, 0, 0, 0, 0, 0, 0, 0, 0))
1264
 
            blob.set_raw_string(encode_git_path(self.get_symlink_target(path)))
 
947
            blob.set_raw_string(
 
948
                self.get_symlink_target(path).encode("utf-8"))
1265
949
            # Add object to the repository if it didn't exist yet
1266
 
            if blob.id not in self.store:
 
950
            if not blob.id in self.store:
1267
951
                self.store.add_object(blob)
1268
952
            hexsha = blob.id
1269
953
        elif kind == "tree-reference":
1270
954
            if reference_revision is not None:
1271
 
                hexsha = self.branch.lookup_bzr_revision_id(
1272
 
                    reference_revision)[0]
 
955
                hexsha = self.branch.lookup_bzr_revision_id(reference_revision)[0]
1273
956
            else:
1274
957
                hexsha = self._read_submodule_head(path)
1275
958
                if hexsha is None:
1284
967
            raise AssertionError("unknown kind '%s'" % kind)
1285
968
        # Add an entry to the index or update the existing entry
1286
969
        ensure_normalized_path(path)
1287
 
        encoded_path = encode_git_path(path)
 
970
        encoded_path = path.encode("utf-8")
1288
971
        if b'\r' in encoded_path or b'\n' in encoded_path:
1289
972
            # TODO(jelmer): Why do we need to do this?
1290
973
            trace.mutter('ignoring path with invalid newline in it: %r', path)
1295
978
        if self._versioned_dirs is not None:
1296
979
            self._ensure_versioned_dir(index_path)
1297
980
 
1298
 
    def _recurse_index_entries(self, index=None, basepath=b"",
1299
 
                               recurse_nested=False):
 
981
    def _recurse_index_entries(self, index=None, basepath=""):
1300
982
        # Iterate over all index entries
1301
983
        with self.lock_read():
1302
984
            if index is None:
1303
985
                index = self.index
1304
 
            for path, value in index.items():
1305
 
                (ctime, mtime, dev, ino, mode, uid, gid, size, sha,
1306
 
                 flags) = value
1307
 
                if S_ISGITLINK(mode) and recurse_nested:
1308
 
                    subindex = self._get_submodule_index(path)
1309
 
                    for entry in self._recurse_index_entries(
1310
 
                            index=subindex, basepath=path,
1311
 
                            recurse_nested=recurse_nested):
1312
 
                        yield entry
1313
 
                else:
1314
 
                    yield (posixpath.join(basepath, path), value)
1315
 
 
1316
 
    def iter_entries_by_dir(self, specific_files=None,
1317
 
                            recurse_nested=False):
 
986
            for path, value in index.iteritems():
 
987
                yield (posixpath.join(basepath, path), value)
 
988
                (ctime, mtime, dev, ino, mode, uid, gid, size, sha, flags) = value
 
989
                if S_ISGITLINK(mode):
 
990
                    pass # TODO(jelmer): dive into submodule
 
991
 
 
992
 
 
993
    def iter_entries_by_dir(self, specific_files=None, yield_parents=False):
 
994
        if yield_parents:
 
995
            raise NotImplementedError(self.iter_entries_by_dir)
1318
996
        with self.lock_read():
1319
997
            if specific_files is not None:
1320
998
                specific_files = set(specific_files)
1323
1001
            root_ie = self._get_dir_ie(u"", None)
1324
1002
            ret = {}
1325
1003
            if specific_files is None or u"" in specific_files:
1326
 
                ret[(u"", u"")] = root_ie
 
1004
                ret[(None, u"")] = root_ie
1327
1005
            dir_ids = {u"": root_ie.file_id}
1328
 
            for path, value in self._recurse_index_entries(
1329
 
                    recurse_nested=recurse_nested):
 
1006
            for path, value in self._recurse_index_entries():
1330
1007
                if self.mapping.is_special_file(path):
1331
1008
                    continue
1332
 
                path = decode_git_path(path)
1333
 
                if specific_files is not None and path not in specific_files:
 
1009
                path = path.decode("utf-8")
 
1010
                if specific_files is not None and not path in specific_files:
1334
1011
                    continue
1335
1012
                (parent, name) = posixpath.split(path)
1336
1013
                try:
1337
1014
                    file_ie = self._get_file_ie(name, path, value, None)
1338
1015
                except errors.NoSuchFile:
1339
1016
                    continue
1340
 
                if specific_files is None:
1341
 
                    for (dir_path, dir_ie) in self._add_missing_parent_ids(
1342
 
                            parent, dir_ids):
 
1017
                if yield_parents or specific_files is None:
 
1018
                    for (dir_path, dir_ie) in self._add_missing_parent_ids(parent,
 
1019
                            dir_ids):
1343
1020
                        ret[(posixpath.dirname(dir_path), dir_path)] = dir_ie
1344
1021
                file_ie.parent_id = self.path2id(parent)
1345
1022
                ret[(posixpath.dirname(path), path)] = file_ie
1346
 
            # Special casing for directories
1347
 
            if specific_files:
1348
 
                for path in specific_files:
1349
 
                    key = (posixpath.dirname(path), path)
1350
 
                    if key not in ret and self.is_versioned(path):
1351
 
                        ret[key] = self._get_dir_ie(path, self.path2id(key[0]))
1352
1023
            return ((path, ie) for ((_, path), ie) in sorted(ret.items()))
1353
1024
 
1354
1025
    def iter_references(self):
1355
 
        if self.supports_tree_reference():
1356
 
            # TODO(jelmer): Implement a more efficient version of this
1357
 
            for path, entry in self.iter_entries_by_dir():
1358
 
                if entry.kind == 'tree-reference':
1359
 
                    yield path
 
1026
        # TODO(jelmer): Implement a more efficient version of this
 
1027
        for path, entry in self.iter_entries_by_dir():
 
1028
            if entry.kind == 'tree-reference':
 
1029
                yield path, self.mapping.generate_file_id(b'')
1360
1030
 
1361
1031
    def _get_dir_ie(self, path, parent_id):
1362
1032
        file_id = self.path2id(path)
1363
1033
        return GitTreeDirectory(file_id,
1364
 
                                posixpath.basename(path).strip("/"), parent_id)
 
1034
            posixpath.basename(path).strip("/"), parent_id)
1365
1035
 
1366
1036
    def _get_file_ie(self, name, path, value, parent_id):
1367
 
        if not isinstance(name, str):
 
1037
        if type(name) is not unicode:
1368
1038
            raise TypeError(name)
1369
 
        if not isinstance(path, str):
 
1039
        if type(path) is not unicode:
1370
1040
            raise TypeError(path)
1371
1041
        if not isinstance(value, tuple) or len(value) != 10:
1372
1042
            raise TypeError(value)
1373
1043
        (ctime, mtime, dev, ino, mode, uid, gid, size, sha, flags) = value
1374
1044
        file_id = self.path2id(path)
1375
 
        if not isinstance(file_id, bytes):
1376
 
            raise TypeError(file_id)
 
1045
        if type(file_id) != str:
 
1046
            raise AssertionError
1377
1047
        kind = mode_kind(mode)
1378
1048
        ie = entry_factory[kind](file_id, name, parent_id)
1379
1049
        if kind == 'symlink':
1380
 
            ie.symlink_target = self.get_symlink_target(path)
 
1050
            ie.symlink_target = self.get_symlink_target(path, file_id)
1381
1051
        elif kind == 'tree-reference':
1382
 
            ie.reference_revision = self.get_reference_revision(path)
 
1052
            ie.reference_revision = self.get_reference_revision(path, file_id)
1383
1053
        else:
1384
1054
            try:
1385
 
                data = self.get_file_text(path)
 
1055
                data = self.get_file_text(path, file_id)
1386
1056
            except errors.NoSuchFile:
1387
1057
                data = None
1388
1058
            except IOError as e:
1415
1085
    def _unversion_path(self, path):
1416
1086
        if self._lock_mode is None:
1417
1087
            raise errors.ObjectNotLocked(self)
1418
 
        encoded_path = encode_git_path(path)
 
1088
        encoded_path = path.encode("utf-8")
1419
1089
        count = 0
1420
1090
        (index, subpath) = self._lookup_index(encoded_path)
1421
1091
        try:
1424
1094
            # A directory, perhaps?
1425
1095
            # TODO(jelmer): Deletes that involve submodules?
1426
1096
            for p in list(index):
1427
 
                if p.startswith(subpath + b"/"):
 
1097
                if p.startswith(subpath+b"/"):
1428
1098
                    count += 1
1429
1099
                    self._index_del_entry(index, p)
1430
1100
        else:
1432
1102
        self._versioned_dirs = None
1433
1103
        return count
1434
1104
 
1435
 
    def unversion(self, paths):
 
1105
    def unversion(self, paths, file_ids=None):
1436
1106
        with self.lock_tree_write():
1437
1107
            for path in paths:
1438
1108
                if self._unversion_path(path) == 0:
1447
1117
        # TODO(jelmer): This shouldn't be called, it's inventory specific.
1448
1118
        for (old_path, new_path, file_id, ie) in delta:
1449
1119
            if old_path is not None:
1450
 
                (index, old_subpath) = self._lookup_index(
1451
 
                    encode_git_path(old_path))
 
1120
                (index, old_subpath) = self._lookup_index(old_path.encode('utf-8'))
1452
1121
                if old_subpath in index:
1453
1122
                    self._index_del_entry(index, old_subpath)
1454
1123
                    self._versioned_dirs = None
1463
1132
            to_abs = self.abspath(to_dir)
1464
1133
            if not os.path.isdir(to_abs):
1465
1134
                raise errors.BzrMoveFailedError('', to_dir,
1466
 
                                                errors.NotADirectory(to_abs))
 
1135
                    errors.NotADirectory(to_abs))
1467
1136
 
1468
1137
            for from_rel in from_paths:
1469
1138
                from_tail = os.path.split(from_rel)[-1]
1474
1143
            return rename_tuples
1475
1144
 
1476
1145
    def rename_one(self, from_rel, to_rel, after=None):
1477
 
        from_path = encode_git_path(from_rel)
 
1146
        from_path = from_rel.encode("utf-8")
1478
1147
        to_rel, can_access = osutils.normalized_filename(to_rel)
1479
1148
        if not can_access:
1480
1149
            raise errors.InvalidNormalization(to_rel)
1481
 
        to_path = encode_git_path(to_rel)
 
1150
        to_path = to_rel.encode("utf-8")
1482
1151
        with self.lock_tree_write():
1483
1152
            if not after:
1484
1153
                # Perhaps it's already moved?
1488
1157
                    not self.is_versioned(to_rel))
1489
1158
            if after:
1490
1159
                if not self.has_filename(to_rel):
1491
 
                    raise errors.BzrMoveFailedError(
1492
 
                        from_rel, to_rel, errors.NoSuchFile(to_rel))
 
1160
                    raise errors.BzrMoveFailedError(from_rel, to_rel,
 
1161
                        errors.NoSuchFile(to_rel))
1493
1162
                if self.basis_tree().is_versioned(to_rel):
1494
 
                    raise errors.BzrMoveFailedError(
1495
 
                        from_rel, to_rel, errors.AlreadyVersionedError(to_rel))
 
1163
                    raise errors.BzrMoveFailedError(from_rel, to_rel,
 
1164
                        errors.AlreadyVersionedError(to_rel))
1496
1165
 
1497
1166
                kind = self.kind(to_rel)
1498
1167
            else:
1505
1174
                    exc_type = errors.BzrMoveFailedError
1506
1175
                if self.is_versioned(to_rel):
1507
1176
                    raise exc_type(from_rel, to_rel,
1508
 
                                   errors.AlreadyVersionedError(to_rel))
 
1177
                        errors.AlreadyVersionedError(to_rel))
1509
1178
                if not self.has_filename(from_rel):
1510
 
                    raise errors.BzrMoveFailedError(
1511
 
                        from_rel, to_rel, errors.NoSuchFile(from_rel))
 
1179
                    raise errors.BzrMoveFailedError(from_rel, to_rel,
 
1180
                        errors.NoSuchFile(from_rel))
1512
1181
                kind = self.kind(from_rel)
1513
1182
                if not self.is_versioned(from_rel) and kind != 'directory':
1514
1183
                    raise exc_type(from_rel, to_rel,
1515
 
                                   errors.NotVersionedError(from_rel))
 
1184
                        errors.NotVersionedError(from_rel))
1516
1185
                if self.has_filename(to_rel):
1517
1186
                    raise errors.RenameFailedFilesExist(
1518
1187
                        from_rel, to_rel, errors.FileExists(to_rel))
1523
1192
                (index, from_subpath) = self._lookup_index(from_path)
1524
1193
                if from_subpath not in index:
1525
1194
                    # It's not a file
1526
 
                    raise errors.BzrMoveFailedError(
1527
 
                        from_rel, to_rel,
 
1195
                    raise errors.BzrMoveFailedError(from_rel, to_rel,
1528
1196
                        errors.NotVersionedError(path=from_rel))
1529
1197
 
1530
1198
            if not after:
1532
1200
                    self._rename_one(from_rel, to_rel)
1533
1201
                except OSError as e:
1534
1202
                    if e.errno == errno.ENOENT:
1535
 
                        raise errors.BzrMoveFailedError(
1536
 
                            from_rel, to_rel, errors.NoSuchFile(to_rel))
 
1203
                        raise errors.BzrMoveFailedError(from_rel, to_rel,
 
1204
                            errors.NoSuchFile(to_rel))
1537
1205
                    raise
1538
1206
            if kind != 'directory':
1539
1207
                (index, from_index_path) = self._lookup_index(from_path)
1543
1211
                    pass
1544
1212
                self._index_add_entry(to_rel, kind)
1545
1213
            else:
1546
 
                todo = [(p, i) for (p, i) in self._recurse_index_entries()
1547
 
                        if p.startswith(from_path + b'/')]
 
1214
                todo = [(p, i) for (p, i) in self._recurse_index_entries() if p.startswith(from_path+'/')]
1548
1215
                for child_path, child_value in todo:
1549
1216
                    (child_to_index, child_to_index_path) = self._lookup_index(
1550
 
                        posixpath.join(to_path, posixpath.relpath(child_path, from_path)))
 
1217
                            posixpath.join(to_path, posixpath.relpath(child_path, from_path)))
1551
1218
                    child_to_index[child_to_index_path] = child_value
1552
1219
                    # TODO(jelmer): Mark individual index as dirty
1553
1220
                    self._index_dirty = True
1554
 
                    (child_from_index, child_from_index_path) = self._lookup_index(
1555
 
                        child_path)
1556
 
                    self._index_del_entry(
1557
 
                        child_from_index, child_from_index_path)
 
1221
                    (child_from_index, child_from_index_path) = self._lookup_index(child_path)
 
1222
                    self._index_del_entry(child_from_index, child_from_index_path)
1558
1223
 
1559
1224
            self._versioned_dirs = None
1560
1225
            self.flush()
1561
1226
 
1562
1227
    def find_related_paths_across_trees(self, paths, trees=[],
1563
 
                                        require_versioned=True):
 
1228
            require_versioned=True):
1564
1229
        if paths is None:
1565
1230
            return None
1566
1231
 
1603
1268
        else:
1604
1269
            return (kind, None, None, None)
1605
1270
 
1606
 
    def stored_kind(self, relpath):
1607
 
        (index, index_path) = self._lookup_index(encode_git_path(relpath))
1608
 
        if index is None:
1609
 
            return kind
1610
 
        try:
1611
 
            mode = index[index_path].mode
1612
 
        except KeyError:
1613
 
            return kind
1614
 
        else:
1615
 
            if S_ISGITLINK(mode):
1616
 
                return 'tree-reference'
1617
 
            return 'directory'
1618
 
 
1619
 
    def kind(self, relpath):
 
1271
    def kind(self, relpath, file_id=None):
1620
1272
        kind = osutils.file_kind(self.abspath(relpath))
1621
1273
        if kind == 'directory':
1622
 
            if self._directory_is_tree_reference(relpath):
1623
 
                return 'tree-reference'
1624
 
            return 'directory'
 
1274
            (index, index_path) = self._lookup_index(relpath.encode('utf-8'))
 
1275
            try:
 
1276
                mode = index[index_path].mode
 
1277
            except KeyError:
 
1278
                return kind
 
1279
            else:
 
1280
                if S_ISGITLINK(mode):
 
1281
                    return 'tree-reference'
 
1282
                return 'directory'
1625
1283
        else:
1626
1284
            return kind
1627
1285
 
1628
1286
    def _live_entry(self, relpath):
1629
1287
        raise NotImplementedError(self._live_entry)
1630
1288
 
1631
 
    def transform(self, pb=None):
1632
 
        from .transform import GitTreeTransform
1633
 
        return GitTreeTransform(self, pb=pb)
1634
 
 
1635
 
    def preview_transform(self, pb=None):
1636
 
        from .transform import GitTransformPreview
1637
 
        return GitTransformPreview(self, pb=pb)
1638
 
 
1639
 
 
1640
 
class InterToIndexGitTree(InterGitTrees):
 
1289
 
 
1290
class InterIndexGitTree(InterGitTrees):
1641
1291
    """InterTree that works between a Git revision tree and an index."""
1642
1292
 
1643
1293
    def __init__(self, source, target):
1644
 
        super(InterToIndexGitTree, self).__init__(source, target)
1645
 
        if self.source.store == self.target.store:
1646
 
            self.store = self.source.store
1647
 
        else:
1648
 
            self.store = OverlayObjectStore(
1649
 
                [self.source.store, self.target.store])
1650
 
        self.rename_detector = RenameDetector(self.store)
 
1294
        super(InterIndexGitTree, self).__init__(source, target)
 
1295
        self._index = target.index
1651
1296
 
1652
1297
    @classmethod
1653
1298
    def is_compatible(cls, source, target):
1655
1300
                isinstance(target, MutableGitIndexTree))
1656
1301
 
1657
1302
    def _iter_git_changes(self, want_unchanged=False, specific_files=None,
1658
 
                          require_versioned=False, extra_trees=None,
1659
 
                          want_unversioned=False, include_trees=True):
 
1303
            require_versioned=False, extra_trees=None,
 
1304
            want_unversioned=False):
1660
1305
        trees = [self.source]
1661
1306
        if extra_trees is not None:
1662
1307
            trees.extend(extra_trees)
1663
1308
        if specific_files is not None:
1664
1309
            specific_files = self.target.find_related_paths_across_trees(
1665
 
                specific_files, trees,
1666
 
                require_versioned=require_versioned)
 
1310
                    specific_files, trees,
 
1311
                    require_versioned=require_versioned)
1667
1312
        # TODO(jelmer): Restrict to specific_files, for performance reasons.
1668
1313
        with self.lock_read():
1669
 
            changes, target_extras = changes_between_git_tree_and_working_copy(
 
1314
            return changes_between_git_tree_and_working_copy(
1670
1315
                self.source.store, self.source.tree,
1671
1316
                self.target, want_unchanged=want_unchanged,
1672
 
                want_unversioned=want_unversioned,
1673
 
                rename_detector=self.rename_detector,
1674
 
                include_trees=include_trees)
1675
 
            return changes, set(), target_extras
1676
 
 
1677
 
 
1678
 
_mod_tree.InterTree.register_optimiser(InterToIndexGitTree)
1679
 
 
1680
 
 
1681
 
class InterFromIndexGitTree(InterGitTrees):
1682
 
    """InterTree that works between a Git revision tree and an index."""
1683
 
 
1684
 
    def __init__(self, source, target):
1685
 
        super(InterFromIndexGitTree, self).__init__(source, target)
1686
 
        if self.source.store == self.target.store:
1687
 
            self.store = self.source.store
1688
 
        else:
1689
 
            self.store = OverlayObjectStore(
1690
 
                [self.source.store, self.target.store])
1691
 
        self.rename_detector = RenameDetector(self.store)
1692
 
 
1693
 
    @classmethod
1694
 
    def is_compatible(cls, source, target):
1695
 
        return (isinstance(target, GitRevisionTree) and
1696
 
                isinstance(source, MutableGitIndexTree))
1697
 
 
1698
 
    def _iter_git_changes(self, want_unchanged=False, specific_files=None,
1699
 
                          require_versioned=False, extra_trees=None,
1700
 
                          want_unversioned=False, include_trees=True):
1701
 
        trees = [self.source]
1702
 
        if extra_trees is not None:
1703
 
            trees.extend(extra_trees)
1704
 
        if specific_files is not None:
1705
 
            specific_files = self.target.find_related_paths_across_trees(
1706
 
                specific_files, trees,
1707
 
                require_versioned=require_versioned)
1708
 
        # TODO(jelmer): Restrict to specific_files, for performance reasons.
1709
 
        with self.lock_read():
1710
 
            from_tree_sha, extras = snapshot_workingtree(self.source, want_unversioned=want_unversioned)
1711
 
            return tree_changes(
1712
 
                self.store, from_tree_sha, self.target.tree,
1713
 
                include_trees=include_trees,
1714
 
                rename_detector=self.rename_detector,
1715
 
                want_unchanged=want_unchanged, change_type_same=True), extras
1716
 
 
1717
 
 
1718
 
_mod_tree.InterTree.register_optimiser(InterFromIndexGitTree)
1719
 
 
1720
 
 
1721
 
class InterIndexGitTree(InterGitTrees):
1722
 
    """InterTree that works between a Git revision tree and an index."""
1723
 
 
1724
 
    def __init__(self, source, target):
1725
 
        super(InterIndexGitTree, self).__init__(source, target)
1726
 
        if self.source.store == self.target.store:
1727
 
            self.store = self.source.store
1728
 
        else:
1729
 
            self.store = OverlayObjectStore(
1730
 
                [self.source.store, self.target.store])
1731
 
        self.rename_detector = RenameDetector(self.store)
1732
 
 
1733
 
    @classmethod
1734
 
    def is_compatible(cls, source, target):
1735
 
        return (isinstance(target, MutableGitIndexTree) and
1736
 
                isinstance(source, MutableGitIndexTree))
1737
 
 
1738
 
    def _iter_git_changes(self, want_unchanged=False, specific_files=None,
1739
 
                          require_versioned=False, extra_trees=None,
1740
 
                          want_unversioned=False, include_trees=True):
1741
 
        trees = [self.source]
1742
 
        if extra_trees is not None:
1743
 
            trees.extend(extra_trees)
1744
 
        if specific_files is not None:
1745
 
            specific_files = self.target.find_related_paths_across_trees(
1746
 
                specific_files, trees,
1747
 
                require_versioned=require_versioned)
1748
 
        # TODO(jelmer): Restrict to specific_files, for performance reasons.
1749
 
        with self.lock_read():
1750
 
            from_tree_sha, from_extras = snapshot_workingtree(
1751
 
                self.source, want_unversioned=want_unversioned)
1752
 
            to_tree_sha, to_extras = snapshot_workingtree(
1753
 
                self.target, want_unversioned=want_unversioned)
1754
 
            changes = tree_changes(
1755
 
                self.store, from_tree_sha, to_tree_sha,
1756
 
                include_trees=include_trees,
1757
 
                rename_detector=self.rename_detector,
1758
 
                want_unchanged=want_unchanged, change_type_same=True)
1759
 
            return changes, from_extras, to_extras
 
1317
                want_unversioned=want_unversioned)
1760
1318
 
1761
1319
 
1762
1320
_mod_tree.InterTree.register_optimiser(InterIndexGitTree)
1763
1321
 
1764
1322
 
1765
 
def snapshot_workingtree(target, want_unversioned=False):
 
1323
def changes_between_git_tree_and_working_copy(store, from_tree_sha, target,
 
1324
        want_unchanged=False, want_unversioned=False):
 
1325
    """Determine the changes between a git tree and a working tree with index.
 
1326
 
 
1327
    """
1766
1328
    extras = set()
1767
1329
    blobs = {}
1768
1330
    # Report dirified directories to commit_tree first, so that they can be
1769
1331
    # replaced with non-empty directories if they have contents.
1770
1332
    dirified = []
1771
 
    trust_executable = target._supports_executable()
1772
1333
    for path, index_entry in target._recurse_index_entries():
1773
1334
        try:
1774
1335
            live_entry = target._live_entry(path)
1776
1337
            if e.errno == errno.ENOENT:
1777
1338
                # Entry was removed; keep it listed, but mark it as gone.
1778
1339
                blobs[path] = (ZERO_SHA, 0)
 
1340
            elif e.errno == errno.EISDIR:
 
1341
                # Entry was turned into a directory
 
1342
                dirified.append((path, Tree().id, stat.S_IFDIR))
 
1343
                store.add_object(Tree())
1779
1344
            else:
1780
1345
                raise
1781
1346
        else:
1782
 
            if live_entry is None:
1783
 
                # Entry was turned into a directory.
1784
 
                # Maybe it's just a submodule that's not checked out?
1785
 
                if S_ISGITLINK(index_entry.mode):
1786
 
                    blobs[path] = (index_entry.sha, index_entry.mode)
1787
 
                else:
1788
 
                    dirified.append((path, Tree().id, stat.S_IFDIR))
1789
 
                    target.store.add_object(Tree())
1790
 
            else:
1791
 
                mode = live_entry.mode
1792
 
                if not trust_executable:
1793
 
                    if mode_is_executable(index_entry.mode):
1794
 
                        mode |= 0o111
1795
 
                    else:
1796
 
                        mode &= ~0o111
1797
 
                if live_entry.sha != index_entry.sha:
1798
 
                    rp = decode_git_path(path)
1799
 
                    if stat.S_ISREG(live_entry.mode):
1800
 
                        blob = Blob()
1801
 
                        with target.get_file(rp) as f:
1802
 
                            blob.data = f.read()
1803
 
                    elif stat.S_ISLNK(live_entry.mode):
1804
 
                        blob = Blob()
1805
 
                        blob.data = target.get_symlink_target(rp).encode(osutils._fs_enc)
1806
 
                    else:
1807
 
                        blob = None
1808
 
                    if blob is not None:
1809
 
                        target.store.add_object(blob)
1810
 
                blobs[path] = (live_entry.sha, cleanup_mode(live_entry.mode))
 
1347
            blobs[path] = (live_entry.sha, cleanup_mode(live_entry.mode))
1811
1348
    if want_unversioned:
1812
 
        for e in target._iter_files_recursive(include_dirs=False):
 
1349
        for e in target.extras():
 
1350
            st = target._lstat(e)
1813
1351
            try:
1814
 
                e, accessible = osutils.normalized_filename(e)
 
1352
                np, accessible = osutils.normalized_filename(e)
1815
1353
            except UnicodeDecodeError:
1816
1354
                raise errors.BadFilenameEncoding(
1817
1355
                    e, osutils._fs_enc)
1818
 
            np = encode_git_path(e)
1819
 
            if np in blobs:
1820
 
                continue
1821
 
            st = target._lstat(e)
1822
1356
            if stat.S_ISDIR(st.st_mode):
1823
1357
                blob = Tree()
1824
 
            elif stat.S_ISREG(st.st_mode) or stat.S_ISLNK(st.st_mode):
1825
 
                blob = blob_from_path_and_stat(
1826
 
                    target.abspath(e).encode(osutils._fs_enc), st)
1827
1358
            else:
1828
 
                continue
1829
 
            target.store.add_object(blob)
 
1359
                blob = blob_from_path_and_stat(target.abspath(e).encode(osutils._fs_enc), st)
 
1360
            store.add_object(blob)
 
1361
            np = np.encode('utf-8')
1830
1362
            blobs[np] = (blob.id, cleanup_mode(st.st_mode))
1831
1363
            extras.add(np)
1832
 
    return commit_tree(
1833
 
        target.store, dirified + [(p, s, m) for (p, (s, m)) in blobs.items()]), extras
1834
 
 
1835
 
 
1836
 
def changes_between_git_tree_and_working_copy(source_store, from_tree_sha, target,
1837
 
                                              want_unchanged=False,
1838
 
                                              want_unversioned=False,
1839
 
                                              rename_detector=None,
1840
 
                                              include_trees=True):
1841
 
    """Determine the changes between a git tree and a working tree with index.
1842
 
 
1843
 
    """
1844
 
    to_tree_sha, extras = snapshot_workingtree(target, want_unversioned=want_unversioned)
1845
 
    store = OverlayObjectStore([source_store, target.store])
1846
 
    return tree_changes(
1847
 
        store, from_tree_sha, to_tree_sha, include_trees=include_trees,
1848
 
        rename_detector=rename_detector,
 
1364
    to_tree_sha = commit_tree(store, dirified + [(p, s, m) for (p, (s, m)) in blobs.items()])
 
1365
    return store.tree_changes(
 
1366
        from_tree_sha, to_tree_sha, include_trees=True,
1849
1367
        want_unchanged=want_unchanged, change_type_same=True), extras