14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
from bzrlib import osutils
18
from bzrlib.errors import InvalidRevisionId
19
from bzrlib.inventory import Inventory
20
from bzrlib.repository import InterRepository
21
from bzrlib.trace import info
23
from bzrlib.plugins.git import git
24
from bzrlib.plugins.git.repository import LocalGitRepository, GitRepository, GitFormat
25
from bzrlib.plugins.git.remote import RemoteGitRepository
27
from dulwich.objects import Commit
29
from cStringIO import StringIO
32
class BzrFetchGraphWalker(object):
34
def __init__(self, repository, mapping):
35
self.repository = repository
36
self.mapping = mapping
38
self.heads = set(repository.all_revision_ids())
42
revid = self.mapping.revision_id_foreign_to_bzr(sha)
45
def remove(self, revid):
48
self.heads.remove(revid)
49
if revid in self.parents:
50
for p in self.parents[revid]:
55
ret = self.heads.pop()
56
ps = self.repository.get_parent_map([ret])[ret]
57
self.parents[ret] = ps
58
self.heads.update([p for p in ps if not p in self.done])
61
return self.mapping.revision_id_bzr_to_foreign(ret)
62
except InvalidRevisionId:
67
def import_git_blob(repo, mapping, path, blob):
17
from dulwich.objects import (
23
from dulwich.object_store import (
26
from itertools import (
39
from bzrlib.errors import (
43
from bzrlib.inventory import (
50
from bzrlib.repository import (
53
from bzrlib.revision import (
56
from bzrlib.tsort import (
59
from bzrlib.versionedfile import (
60
ChunkedContentFactory,
63
from bzrlib.plugins.git.mapping import (
65
inventory_to_tree_and_blobs,
71
from bzrlib.plugins.git.object_store import (
75
from bzrlib.plugins.git.remote import (
78
from bzrlib.plugins.git.repository import (
85
def import_git_blob(texts, mapping, path, name, (base_hexsha, hexsha),
86
base_inv, base_inv_shamap, parent_id, revision_id,
87
parent_invs, lookup_object, (base_mode, mode), store_updater):
68
88
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
90
:param texts: VersionedFiles to add to
71
91
:param path: Path in the tree
72
92
:param blob: A git blob
93
:return: Inventory delta for this file
74
file_id = mapping.generate_file_id(path)
75
repo.texts.add_lines((file_id, blob.id),
77
osutils.split_lines(blob.data))
78
inv.add_path(path, "file", file_id)
81
def import_git_tree(repo, mapping, path, tree, inv, lookup_object):
95
if base_hexsha == hexsha and base_mode == mode:
96
# If nothing has changed since the base revision, we're done
98
file_id = mapping.generate_file_id(path)
99
if stat.S_ISLNK(mode):
103
ie = cls(file_id, name.decode("utf-8"), parent_id)
104
ie.executable = mode_is_executable(mode)
105
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
106
base_ie = base_inv[base_inv.path2id(path)]
107
ie.text_size = base_ie.text_size
108
ie.text_sha1 = base_ie.text_sha1
109
ie.symlink_target = base_ie.symlink_target
110
if ie.executable == base_ie.executable:
111
ie.revision = base_ie.revision
113
blob = lookup_object(hexsha)
115
blob = lookup_object(hexsha)
116
if ie.kind == "symlink":
118
ie.symlink_target = blob.data
122
ie.text_size = sum(imap(len, blob.chunked))
123
ie.text_sha1 = osutils.sha_strings(blob.chunked)
124
# Check what revision we should store
126
for pinv in parent_invs[1:]:
131
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
132
# found a revision in one of the parents to use
133
ie.revision = pie.revision
135
parent_keys.append((file_id, pie.revision))
136
if ie.revision is None:
137
# Need to store a new revision
138
ie.revision = revision_id
139
assert ie.revision is not None
140
if ie.kind == 'symlink':
143
chunks = blob.chunked
144
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, chunks)])
146
if base_hexsha is not None:
147
old_path = path # Renames are not supported yet
148
if stat.S_ISDIR(base_mode):
149
invdelta.extend(remove_disappeared_children(base_inv, old_path,
150
lookup_object(base_hexsha), [], lookup_object))
153
invdelta.append((old_path, path, file_id, ie))
154
if base_hexsha != hexsha:
155
store_updater.add_object(blob, ie)
159
class SubmodulesRequireSubtrees(BzrError):
160
_fmt = """The repository you are fetching from contains submodules. To continue, upgrade your Bazaar repository to a format that supports nested trees, such as 'development-subtree'."""
164
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
165
base_inv, parent_id, revision_id, parent_invs, lookup_object,
166
(base_mode, mode), store_updater):
167
if base_hexsha == hexsha and base_mode == mode:
169
file_id = mapping.generate_file_id(path)
170
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
171
ie.revision = revision_id
172
if base_hexsha is None:
176
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
177
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
178
invdelta = [(oldpath, path, file_id, ie)]
182
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
185
for name, mode, hexsha in base_tree.iteritems():
186
if name in existing_children:
188
c_path = posixpath.join(path, name.decode("utf-8"))
189
ret.append((c_path, None, base_inv.path2id(c_path), None))
190
if stat.S_ISDIR(mode):
191
ret.extend(remove_disappeared_children(
192
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
196
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
197
base_inv, base_inv_shamap, parent_id, revision_id, parent_invs,
198
lookup_object, (base_mode, mode), store_updater, allow_submodules=False):
82
199
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
201
:param texts: VersionedFiles object to add to
85
202
:param path: Path in the tree
86
203
:param tree: A git tree object
87
:param inv: Inventory object
204
:param base_inv: Base inventory against which to return inventory delta
205
:return: Inventory delta for this subtree
207
if base_hexsha == hexsha and base_mode == mode:
208
# If nothing has changed since the base revision, we're done
89
211
file_id = mapping.generate_file_id(path)
90
repo.texts.add_lines((file_id, tree.id),
93
inv.add_path(path, "directory", file_id)
94
for mode, name, hexsha in tree.entries():
95
entry_kind = (mode & 0700000) / 0100000
96
basename = name.decode("utf-8")
100
child_path = urlutils.join(path, name)
102
import_git_tree(repo, mapping, child_path, lookup_object, inv)
103
elif entry_kind == 1:
104
import_git_blob(repo, mapping, child_path, lookup_object, inv)
106
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
109
def import_git_objects(repo, mapping, object_iter):
212
# We just have to hope this is indeed utf-8:
213
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
214
tree = lookup_object(hexsha)
215
if base_hexsha is None:
217
old_path = None # Newly appeared here
219
base_tree = lookup_object(base_hexsha)
220
old_path = path # Renames aren't supported yet
221
if base_tree is None or type(base_tree) is not Tree:
222
ie.revision = revision_id
223
invdelta.append((old_path, path, ie.file_id, ie))
224
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
225
# Remember for next time
226
existing_children = set()
228
for child_mode, name, child_hexsha in tree.entries():
229
existing_children.add(name)
230
child_path = posixpath.join(path, name)
231
if type(base_tree) is Tree:
233
child_base_mode, child_base_hexsha = base_tree[name]
235
child_base_hexsha = None
238
child_base_hexsha = None
240
if stat.S_ISDIR(child_mode):
241
subinvdelta, grandchildmodes = import_git_tree(
242
texts, mapping, child_path, name,
243
(child_base_hexsha, child_hexsha),
244
base_inv, base_inv_shamap,
245
file_id, revision_id, parent_invs, lookup_object,
246
(child_base_mode, child_mode), store_updater,
247
allow_submodules=allow_submodules)
248
elif S_ISGITLINK(child_mode): # submodule
249
if not allow_submodules:
250
raise SubmodulesRequireSubtrees()
251
subinvdelta, grandchildmodes = import_git_submodule(
252
texts, mapping, child_path, name,
253
(child_base_hexsha, child_hexsha),
254
base_inv, file_id, revision_id, parent_invs, lookup_object,
255
(child_base_mode, child_mode), store_updater)
257
subinvdelta = import_git_blob(texts, mapping,
258
child_path, name, (child_base_hexsha, child_hexsha),
259
base_inv, base_inv_shamap,
261
revision_id, parent_invs, lookup_object,
262
(child_base_mode, child_mode), store_updater)
264
child_modes.update(grandchildmodes)
265
invdelta.extend(subinvdelta)
266
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
267
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
268
child_modes[child_path] = child_mode
269
# Remove any children that have disappeared
270
if base_tree is not None and type(base_tree) is Tree:
271
invdelta.extend(remove_disappeared_children(base_inv, old_path,
272
base_tree, existing_children, lookup_object))
273
store_updater.add_object(tree, ie)
274
return invdelta, child_modes
277
def import_git_commit(repo, mapping, head, lookup_object,
278
target_git_object_retriever, parent_invs_cache):
279
o = lookup_object(head)
280
rev = mapping.import_commit(o)
281
# We have to do this here, since we have to walk the tree and
282
# we need to make sure to import the blobs / trees with the right
283
# path; this may involve adding them more than once.
284
parent_invs = parent_invs_cache.get_inventories(rev.parent_ids)
285
if parent_invs == []:
286
base_inv = Inventory(root_id=None)
287
base_inv_shamap = None # Should never be accessed
291
base_inv = parent_invs[0]
292
base_inv_shamap = target_git_object_retriever._idmap.get_inventory_sha_map(base_inv.revision_id)
293
base_tree = lookup_object(o.parents[0]).tree
294
base_mode = stat.S_IFDIR
295
store_updater = target_git_object_retriever._get_updater(rev)
296
store_updater.add_object(o, None)
297
inv_delta, unusual_modes = import_git_tree(repo.texts,
298
mapping, "", u"", (base_tree, o.tree), base_inv, base_inv_shamap,
299
None, rev.revision_id, parent_invs, lookup_object,
300
(base_mode, stat.S_IFDIR), store_updater,
301
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
302
store_updater.finish()
303
if unusual_modes != {}:
304
for path, mode in unusual_modes.iteritems():
305
warn_unusual_mode(rev.foreign_revid, path, mode)
306
mapping.import_unusual_file_modes(rev, unusual_modes)
308
basis_id = rev.parent_ids[0]
310
basis_id = NULL_REVISION
312
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
313
inv_delta, rev.revision_id, rev.parent_ids,
315
parent_invs_cache.add(rev.revision_id, inv)
316
repo.add_revision(rev.revision_id, rev)
317
if "verify" in debug.debug_flags:
318
new_unusual_modes = mapping.export_unusual_file_modes(rev)
319
if new_unusual_modes != unusual_modes:
320
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
321
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
322
for newsha1, newobj, path in objs:
323
assert path is not None
327
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
328
if oldsha1 != newsha1:
329
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
332
def import_git_objects(repo, mapping, object_iter,
333
target_git_object_retriever, heads, pb=None, limit=None):
110
334
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
336
:param repo: Target Bazaar repository
113
337
:param mapping: Mapping to use
114
338
:param object_iter: Iterator over Git objects.
339
:return: Tuple with pack hints and last imported revision id
116
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
341
def lookup_object(sha):
343
return object_iter[sha]
345
return target_git_object_retriever[sha]
348
heads = list(set(heads))
349
parent_invs_cache = LRUInventoryCache(repo)
121
350
# Find and convert commit objects
122
for o in objects.iterkeys():
353
pb.update("finding revisions to fetch", len(graph), None)
355
assert isinstance(head, str)
357
o = lookup_object(head)
123
360
if isinstance(o, Commit):
124
361
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
362
if repo.has_revision(rev.revision_id):
364
squash_revision(repo, rev)
365
graph.append((o.id, o.parents))
366
heads.extend([p for p in o.parents if p not in checked])
367
elif isinstance(o, Tag):
368
if o.object[1] not in checked:
369
heads.append(o.object[1])
371
trace.warning("Unable to import head object %r" % o)
374
# Order the revisions
126
375
# Create the inventory objects
127
for rev, root_tree in root_trees.iteritems():
128
# We have to do this here, since we have to walk the tree and
129
# we need to make sure to import the blobs / trees with the riht
130
# path; this may involve adding them more than once.
132
def lookup_object(sha):
135
return reconstruct_git_object(repo, mapping, sha)
136
import_git_tree(repo, mapping, "", tree, inv, lookup_object)
137
repo.add_revision(rev.revision_id, rev, inv)
140
def reconstruct_git_commit(repo, rev):
141
raise NotImplementedError(self.reconstruct_git_commit)
144
def reconstruct_git_object(repo, mapping, sha):
146
revid = mapping.revision_id_foreign_to_bzr(sha)
148
rev = repo.get_revision(revid)
149
except NoSuchRevision:
152
return reconstruct_git_commit(rev)
156
raise KeyError("No such object %s" % sha)
377
revision_ids = topo_sort(graph)
379
if limit is not None:
380
revision_ids = revision_ids[:limit]
382
for offset in range(0, len(revision_ids), batch_size):
383
target_git_object_retriever.start_write_group() # FIXME: try/finally
385
repo.start_write_group()
387
for i, head in enumerate(
388
revision_ids[offset:offset+batch_size]):
390
pb.update("fetching revisions", offset+i,
392
import_git_commit(repo, mapping, head, lookup_object,
393
target_git_object_retriever,
397
repo.abort_write_group()
400
hint = repo.commit_write_group()
402
pack_hints.extend(hint)
404
target_git_object_retriever.abort_write_group()
407
target_git_object_retriever.commit_write_group()
408
return pack_hints, last_imported
159
411
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
413
_matching_repo_format = GitRepositoryFormat()
164
416
def _get_repo_format_to_test():
168
420
"""See InterRepository.copy_content."""
169
421
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
423
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
424
mapping=None, fetch_spec=None):
425
self.fetch_refs(revision_id=revision_id, pb=pb,
426
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
429
class InterGitNonGitRepository(InterGitRepository):
430
"""Base InterRepository that copies revisions from a Git into a non-Git
433
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
434
mapping=None, fetch_spec=None):
173
435
if mapping is None:
174
436
mapping = self.source.get_mapping()
177
pb.note("git: %s" % text)
179
info("git: %s" % text)
180
def determine_wants(heads):
181
if revision_id is None:
184
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
437
if revision_id is not None:
438
interesting_heads = [revision_id]
439
elif fetch_spec is not None:
440
interesting_heads = fetch_spec.heads
442
interesting_heads = None
444
def determine_wants(refs):
446
if interesting_heads is None:
447
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
449
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
450
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
186
graph_walker = BzrFetchGraphWalker(self.target, mapping)
451
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
452
if pack_hint is not None and self.target._format.pack_compresses:
453
self.target.pack(hint=pack_hint)
454
if interesting_heads is not None:
455
present_interesting_heads = self.target.has_revisions(interesting_heads)
456
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
457
if missing_interesting_heads:
458
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
462
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
463
def report_git_progress(pb, text):
464
text = text.rstrip("\r\n")
465
g = _GIT_PROGRESS_RE.match(text)
467
(text, pct, current, total) = g.groups()
468
pb.update(text, int(current), int(total))
470
pb.update(text, 0, 0)
473
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
474
"""InterRepository that copies revisions from a remote Git into a non-Git
477
def get_target_heads(self):
478
# FIXME: This should be more efficient
479
all_revs = self.target.all_revision_ids()
480
parent_map = self.target.get_parent_map(all_revs)
482
map(all_parents.update, parent_map.itervalues())
483
return set(all_revs) - all_parents
485
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
487
report_git_progress(pb, text)
488
store = BazaarObjectStore(self.target, mapping)
187
489
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
491
heads = self.get_target_heads()
492
graph_walker = store.get_graph_walker(
493
[store._lookup_revision_sha1(head) for head in heads])
496
def record_determine_wants(heads):
497
wants = determine_wants(heads)
498
recorded_wants.extend(wants)
503
create_pb = pb = ui.ui_factory.nested_progress_bar()
505
objects_iter = self.source.fetch_objects(
506
record_determine_wants, graph_walker,
507
store.get_raw, progress)
508
return import_git_objects(self.target, mapping,
509
objects_iter, store, recorded_wants, pb, limit)
193
514
self.target.unlock()
196
517
def is_compatible(source, target):
197
518
"""Be compatible with GitRepository."""
198
519
# FIXME: Also check target uses VersionedFile
199
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
520
return (isinstance(source, RemoteGitRepository) and
521
target.supports_rich_root() and
522
not isinstance(target, GitRepository))
525
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
526
"""InterRepository that copies revisions from a local Git into a non-Git
529
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
532
wants = determine_wants(self.source._git.get_refs())
535
create_pb = pb = ui.ui_factory.nested_progress_bar()
536
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
538
self.target.lock_write()
540
return import_git_objects(self.target, mapping,
541
self.source._git.object_store,
542
target_git_object_retriever, wants, pb, limit)
550
def is_compatible(source, target):
551
"""Be compatible with GitRepository."""
552
# FIXME: Also check target uses VersionedFile
553
return (isinstance(source, LocalGitRepository) and
554
target.supports_rich_root() and
555
not isinstance(target, GitRepository))
558
class InterGitGitRepository(InterGitRepository):
559
"""InterRepository that copies between Git repositories."""
561
def fetch_objects(self, determine_wants, mapping, pb=None):
563
trace.note("git: %s", text)
564
graphwalker = self.target._git.get_graph_walker()
565
if (isinstance(self.source, LocalGitRepository) and
566
isinstance(self.target, LocalGitRepository)):
567
return self.source._git.fetch(self.target._git, determine_wants,
569
elif (isinstance(self.source, LocalGitRepository) and
570
isinstance(self.target, RemoteGitRepository)):
571
raise NotImplementedError
572
elif (isinstance(self.source, RemoteGitRepository) and
573
isinstance(self.target, LocalGitRepository)):
574
f, commit = self.target._git.object_store.add_thin_pack()
576
refs = self.source._git.fetch_pack(determine_wants,
577
graphwalker, f.write, progress)
586
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
587
mapping=None, fetch_spec=None, branches=None):
589
mapping = self.source.get_mapping()
591
if revision_id is not None:
592
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
593
elif fetch_spec is not None:
594
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
595
if branches is not None:
596
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
597
elif fetch_spec is None and revision_id is None:
598
determine_wants = r.object_store.determine_wants_all
600
determine_wants = lambda x: [y for y in args if not y in r.object_store]
601
return self.fetch_objects(determine_wants, mapping)[0]
605
def is_compatible(source, target):
606
"""Be compatible with GitRepository."""
607
return (isinstance(source, GitRepository) and
608
isinstance(target, GitRepository))
added
removed
fetch.py
