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.revisiontree import (
59
from bzrlib.tsort import (
62
from bzrlib.versionedfile import (
63
ChunkedContentFactory,
66
from bzrlib.plugins.git.mapping import (
73
from bzrlib.plugins.git.object_store import (
78
from bzrlib.plugins.git.remote import (
81
from bzrlib.plugins.git.repository import (
88
def import_git_blob(texts, mapping, path, name, (base_hexsha, hexsha),
89
base_inv, parent_id, revision_id,
90
parent_invs, lookup_object, (base_mode, mode), store_updater,
68
92
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
94
:param texts: VersionedFiles to add to
71
95
:param path: Path in the tree
72
96
:param blob: A git blob
97
: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):
99
if mapping.is_control_file(path):
101
if base_hexsha == hexsha and base_mode == mode:
102
# If nothing has changed since the base revision, we're done
104
file_id = lookup_file_id(path)
105
if stat.S_ISLNK(mode):
109
ie = cls(file_id, name.decode("utf-8"), parent_id)
110
ie.executable = mode_is_executable(mode)
111
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
112
base_ie = base_inv[base_inv.path2id(path)]
113
ie.text_size = base_ie.text_size
114
ie.text_sha1 = base_ie.text_sha1
115
ie.symlink_target = base_ie.symlink_target
116
if ie.executable == base_ie.executable:
117
ie.revision = base_ie.revision
119
blob = lookup_object(hexsha)
121
blob = lookup_object(hexsha)
122
if ie.kind == "symlink":
124
ie.symlink_target = blob.data
128
ie.text_size = sum(imap(len, blob.chunked))
129
ie.text_sha1 = osutils.sha_strings(blob.chunked)
130
# Check what revision we should store
132
for pinv in parent_invs:
137
if (pie.text_sha1 == ie.text_sha1 and
138
pie.executable == ie.executable and
139
pie.symlink_target == ie.symlink_target):
140
# found a revision in one of the parents to use
141
ie.revision = pie.revision
143
parent_key = (file_id, pie.revision)
144
if not parent_key in parent_keys:
145
parent_keys.append(parent_key)
146
if ie.revision is None:
147
# Need to store a new revision
148
ie.revision = revision_id
149
assert ie.revision is not None
150
if ie.kind == 'symlink':
153
chunks = blob.chunked
154
texts.insert_record_stream([
155
ChunkedContentFactory((file_id, ie.revision),
156
tuple(parent_keys), ie.text_sha1, chunks)])
158
if base_hexsha is not None:
159
old_path = path # Renames are not supported yet
160
if stat.S_ISDIR(base_mode):
161
invdelta.extend(remove_disappeared_children(base_inv, old_path,
162
lookup_object(base_hexsha), [], lookup_object))
165
invdelta.append((old_path, path, file_id, ie))
166
if base_hexsha != hexsha:
167
store_updater.add_object(blob, ie)
171
class SubmodulesRequireSubtrees(BzrError):
172
_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'."""
176
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
177
base_inv, parent_id, revision_id, parent_invs, lookup_object,
178
(base_mode, mode), store_updater, lookup_file_id):
179
if base_hexsha == hexsha and base_mode == mode:
181
file_id = lookup_file_id(path)
182
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
183
ie.revision = revision_id
184
if base_hexsha is None:
188
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
189
texts.insert_record_stream([
190
ChunkedContentFactory((file_id, ie.revision), (), None, [])])
191
invdelta = [(oldpath, path, file_id, ie)]
195
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
198
for name, mode, hexsha in base_tree.iteritems():
199
if name in existing_children:
201
c_path = posixpath.join(path, name.decode("utf-8"))
202
ret.append((c_path, None, base_inv.path2id(c_path), None))
203
if stat.S_ISDIR(mode):
204
ret.extend(remove_disappeared_children(
205
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
209
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
210
base_inv, parent_id, revision_id, parent_invs,
211
lookup_object, (base_mode, mode), store_updater,
212
lookup_file_id, allow_submodules=False):
82
213
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
215
:param texts: VersionedFiles object to add to
85
216
:param path: Path in the tree
86
217
:param tree: A git tree object
87
:param inv: Inventory object
218
:param base_inv: Base inventory against which to return inventory delta
219
:return: Inventory delta for this subtree
89
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):
221
if base_hexsha == hexsha and base_mode == mode:
222
# If nothing has changed since the base revision, we're done
225
file_id = lookup_file_id(path)
226
# We just have to hope this is indeed utf-8:
227
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
228
tree = lookup_object(hexsha)
229
if base_hexsha is None:
231
old_path = None # Newly appeared here
233
base_tree = lookup_object(base_hexsha)
234
old_path = path # Renames aren't supported yet
235
if base_tree is None or type(base_tree) is not Tree:
236
ie.revision = revision_id
237
invdelta.append((old_path, path, ie.file_id, ie))
238
texts.insert_record_stream([
239
ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
240
# Remember for next time
241
existing_children = set()
243
for child_mode, name, child_hexsha in tree.entries():
244
existing_children.add(name)
245
child_path = posixpath.join(path, name)
246
if type(base_tree) is Tree:
248
child_base_mode, child_base_hexsha = base_tree[name]
250
child_base_hexsha = None
253
child_base_hexsha = None
255
if stat.S_ISDIR(child_mode):
256
subinvdelta, grandchildmodes = import_git_tree(texts, mapping,
257
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
258
file_id, revision_id, parent_invs, lookup_object,
259
(child_base_mode, child_mode), store_updater, lookup_file_id,
260
allow_submodules=allow_submodules)
261
elif S_ISGITLINK(child_mode): # submodule
262
if not allow_submodules:
263
raise SubmodulesRequireSubtrees()
264
subinvdelta, grandchildmodes = import_git_submodule(texts, mapping,
265
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
266
file_id, revision_id, parent_invs, lookup_object,
267
(child_base_mode, child_mode), store_updater, lookup_file_id)
269
subinvdelta = import_git_blob(texts, mapping, child_path, name,
270
(child_base_hexsha, child_hexsha), base_inv, file_id,
271
revision_id, parent_invs, lookup_object,
272
(child_base_mode, child_mode), store_updater, lookup_file_id)
274
child_modes.update(grandchildmodes)
275
invdelta.extend(subinvdelta)
276
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
277
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
278
child_modes[child_path] = child_mode
279
# Remove any children that have disappeared
280
if base_tree is not None and type(base_tree) is Tree:
281
invdelta.extend(remove_disappeared_children(base_inv, old_path,
282
base_tree, existing_children, lookup_object))
283
store_updater.add_object(tree, ie)
284
return invdelta, child_modes
287
def verify_commit_reconstruction(target_git_object_retriever, lookup_object,
288
o, rev, ret_tree, parent_trees, mapping, unusual_modes):
289
new_unusual_modes = mapping.export_unusual_file_modes(rev)
290
if new_unusual_modes != unusual_modes:
291
raise AssertionError("unusual modes don't match: %r != %r" % (
292
unusual_modes, new_unusual_modes))
293
# Verify that we can reconstruct the commit properly
294
rec_o = target_git_object_retriever._reconstruct_commit(rev, o.tree)
296
raise AssertionError("Reconstructed commit differs: %r != %r" % (
300
for path, obj, ie in _tree_to_objects(ret_tree, parent_trees,
301
target_git_object_retriever._cache.idmap, unusual_modes, mapping.BZR_DUMMY_FILE):
302
old_obj_id = tree_lookup_path(lookup_object, o.tree, path)[1]
304
if obj.id != old_obj_id:
305
diff.append((path, lookup_object(old_obj_id), obj))
306
for (path, old_obj, new_obj) in diff:
307
while (old_obj.type_name == "tree" and
308
new_obj.type_name == "tree" and
309
sorted(old_obj) == sorted(new_obj)):
311
if old_obj[name][0] != new_obj[name][0]:
312
raise AssertionError("Modes for %s differ: %o != %o" %
313
(path, old_obj[name][0], new_obj[name][0]))
314
if old_obj[name][1] != new_obj[name][1]:
315
# Found a differing child, delve deeper
316
path = posixpath.join(path, name)
317
old_obj = lookup_object(old_obj[name][1])
318
new_obj = new_objs[path]
320
raise AssertionError("objects differ for %s: %r != %r" % (path,
324
def import_git_commit(repo, mapping, head, lookup_object,
325
target_git_object_retriever, trees_cache):
326
o = lookup_object(head)
327
rev = mapping.import_commit(o)
328
# We have to do this here, since we have to walk the tree and
329
# we need to make sure to import the blobs / trees with the right
330
# path; this may involve adding them more than once.
331
parent_trees = trees_cache.revision_trees(rev.parent_ids)
332
if parent_trees == []:
333
base_inv = Inventory(root_id=None)
337
base_inv = parent_trees[0].inventory
338
base_tree = lookup_object(o.parents[0]).tree
339
base_mode = stat.S_IFDIR
340
store_updater = target_git_object_retriever._get_updater(rev)
341
store_updater.add_object(o, None)
342
lookup_file_id = mapping.get_fileid_map(lookup_object, o.tree).lookup_file_id
343
inv_delta, unusual_modes = import_git_tree(repo.texts,
344
mapping, "", u"", (base_tree, o.tree), base_inv,
345
None, rev.revision_id, [p.inventory for p in parent_trees],
346
lookup_object, (base_mode, stat.S_IFDIR), store_updater,
348
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
349
store_updater.finish()
350
if unusual_modes != {}:
351
for path, mode in unusual_modes.iteritems():
352
warn_unusual_mode(rev.foreign_revid, path, mode)
353
mapping.import_unusual_file_modes(rev, unusual_modes)
355
basis_id = rev.parent_ids[0]
357
basis_id = NULL_REVISION
359
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
360
inv_delta, rev.revision_id, rev.parent_ids,
362
ret_tree = RevisionTree(repo, inv, rev.revision_id)
363
trees_cache.add(ret_tree)
364
repo.add_revision(rev.revision_id, rev)
365
if "verify" in debug.debug_flags:
366
verify_commit_reconstruction(target_git_object_retriever,
367
lookup_object, o, rev, ret_tree, parent_trees, mapping,
371
def import_git_objects(repo, mapping, object_iter,
372
target_git_object_retriever, heads, pb=None, limit=None):
110
373
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
375
:param repo: Target Bazaar repository
113
376
:param mapping: Mapping to use
114
377
:param object_iter: Iterator over Git objects.
378
: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:
380
def lookup_object(sha):
382
return object_iter[sha]
384
return target_git_object_retriever[sha]
387
heads = list(set(heads))
388
trees_cache = LRUTreeCache(repo)
121
389
# Find and convert commit objects
122
for o in objects.iterkeys():
392
pb.update("finding revisions to fetch", len(graph), None)
394
assert isinstance(head, str)
396
o = lookup_object(head)
123
399
if isinstance(o, Commit):
124
400
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
401
if repo.has_revision(rev.revision_id):
403
squash_revision(repo, rev)
404
graph.append((o.id, o.parents))
405
heads.extend([p for p in o.parents if p not in checked])
406
elif isinstance(o, Tag):
407
if o.object[1] not in checked:
408
heads.append(o.object[1])
410
trace.warning("Unable to import head object %r" % o)
413
# Order the revisions
126
414
# 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)
416
revision_ids = topo_sort(graph)
418
if limit is not None:
419
revision_ids = revision_ids[:limit]
421
for offset in range(0, len(revision_ids), batch_size):
422
target_git_object_retriever.start_write_group()
424
repo.start_write_group()
426
for i, head in enumerate(
427
revision_ids[offset:offset+batch_size]):
429
pb.update("fetching revisions", offset+i,
431
import_git_commit(repo, mapping, head, lookup_object,
432
target_git_object_retriever, trees_cache)
435
repo.abort_write_group()
438
hint = repo.commit_write_group()
440
pack_hints.extend(hint)
442
target_git_object_retriever.abort_write_group()
445
target_git_object_retriever.commit_write_group()
446
return pack_hints, last_imported
159
449
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
451
_matching_repo_format = GitRepositoryFormat()
164
454
def _get_repo_format_to_test():
168
458
"""See InterRepository.copy_content."""
169
459
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
461
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
462
mapping=None, fetch_spec=None):
463
self.fetch_refs(revision_id=revision_id, pb=pb,
464
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
467
class InterGitNonGitRepository(InterGitRepository):
468
"""Base InterRepository that copies revisions from a Git into a non-Git
471
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
472
mapping=None, fetch_spec=None):
173
473
if mapping is None:
174
474
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)]
475
if revision_id is not None:
476
interesting_heads = [revision_id]
477
elif fetch_spec is not None:
478
interesting_heads = fetch_spec.heads
480
interesting_heads = None
482
def determine_wants(refs):
484
if interesting_heads is None:
485
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
487
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
488
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)
489
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
490
if pack_hint is not None and self.target._format.pack_compresses:
491
self.target.pack(hint=pack_hint)
492
if interesting_heads is not None:
493
present_interesting_heads = self.target.has_revisions(interesting_heads)
494
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
495
if missing_interesting_heads:
496
raise AssertionError("Missing interesting heads: %r" %
497
missing_interesting_heads)
501
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
502
def report_git_progress(pb, text):
503
text = text.rstrip("\r\n")
504
g = _GIT_PROGRESS_RE.match(text)
506
(text, pct, current, total) = g.groups()
507
pb.update(text, int(current), int(total))
509
pb.update(text, 0, 0)
512
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
513
"""InterRepository that copies revisions from a remote Git into a non-Git
516
def get_target_heads(self):
517
# FIXME: This should be more efficient
518
all_revs = self.target.all_revision_ids()
519
parent_map = self.target.get_parent_map(all_revs)
521
map(all_parents.update, parent_map.itervalues())
522
return set(all_revs) - all_parents
524
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
526
report_git_progress(pb, text)
527
store = BazaarObjectStore(self.target, mapping)
187
528
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
530
heads = self.get_target_heads()
531
graph_walker = store.get_graph_walker(
532
[store._lookup_revision_sha1(head) for head in heads])
535
def record_determine_wants(heads):
536
wants = determine_wants(heads)
537
recorded_wants.extend(wants)
542
create_pb = pb = ui.ui_factory.nested_progress_bar()
544
objects_iter = self.source.fetch_objects(record_determine_wants,
545
graph_walker, store.get_raw, progress)
546
return import_git_objects(self.target, mapping,
547
objects_iter, store, recorded_wants, pb, limit)
193
552
self.target.unlock()
196
555
def is_compatible(source, target):
197
556
"""Be compatible with GitRepository."""
198
# FIXME: Also check target uses VersionedFile
199
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
557
return (isinstance(source, RemoteGitRepository) and
558
target.supports_rich_root() and
559
not isinstance(target, GitRepository) and
560
target.texts is not None)
563
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
564
"""InterRepository that copies revisions from a local Git into a non-Git
567
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
570
wants = determine_wants(self.source._git.get_refs())
573
create_pb = pb = ui.ui_factory.nested_progress_bar()
574
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
576
self.target.lock_write()
578
return import_git_objects(self.target, mapping,
579
self.source._git.object_store,
580
target_git_object_retriever, wants, pb, limit)
588
def is_compatible(source, target):
589
"""Be compatible with GitRepository."""
590
return (isinstance(source, LocalGitRepository) and
591
target.supports_rich_root() and
592
not isinstance(target, GitRepository) and
593
target.texts is not None)
596
class InterGitGitRepository(InterGitRepository):
597
"""InterRepository that copies between Git repositories."""
599
def fetch_objects(self, determine_wants, mapping, pb=None):
601
trace.note("git: %s", text)
602
graphwalker = self.target._git.get_graph_walker()
603
if (isinstance(self.source, LocalGitRepository) and
604
isinstance(self.target, LocalGitRepository)):
605
return self.source._git.fetch(self.target._git, determine_wants,
607
elif (isinstance(self.source, LocalGitRepository) and
608
isinstance(self.target, RemoteGitRepository)):
609
raise NotImplementedError
610
elif (isinstance(self.source, RemoteGitRepository) and
611
isinstance(self.target, LocalGitRepository)):
612
f, commit = self.target._git.object_store.add_thin_pack()
614
refs = self.source._git.fetch_pack(determine_wants,
615
graphwalker, f.write, progress)
624
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
625
mapping=None, fetch_spec=None, branches=None):
627
mapping = self.source.get_mapping()
629
if revision_id is not None:
630
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
631
elif fetch_spec is not None:
632
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
633
if branches is not None:
634
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
635
elif fetch_spec is None and revision_id is None:
636
determine_wants = r.object_store.determine_wants_all
638
determine_wants = lambda x: [y for y in args if not y in r.object_store]
639
return self.fetch_objects(determine_wants, mapping)[0]
643
def is_compatible(source, target):
644
"""Be compatible with GitRepository."""
645
return (isinstance(source, GitRepository) and
646
isinstance(target, GitRepository))
added
removed
fetch.py
