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, ui, urlutils
18
from bzrlib.errors import InvalidRevisionId
19
from bzrlib.inventory import Inventory
20
from bzrlib.repository import InterRepository
21
from bzrlib.trace import info
22
from bzrlib.tsort import topo_sort
24
from bzrlib.plugins.git import git
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 (
74
from bzrlib.plugins.git.object_store import (
79
from bzrlib.plugins.git.remote import (
25
82
from bzrlib.plugins.git.repository import (
30
from bzrlib.plugins.git.remote import RemoteGitRepository
32
from dulwich.objects import Commit
34
from cStringIO import StringIO
37
class BzrFetchGraphWalker(object):
39
def __init__(self, repository, mapping):
40
self.repository = repository
41
self.mapping = mapping
43
self.heads = set(repository.all_revision_ids())
47
revid = self.mapping.revision_id_foreign_to_bzr(sha)
50
def remove(self, revid):
53
self.heads.remove(revid)
54
if revid in self.parents:
55
for p in self.parents[revid]:
60
ret = self.heads.pop()
61
ps = self.repository.get_parent_map([ret])[ret]
62
self.parents[ret] = ps
63
self.heads.update([p for p in ps if not p in self.done])
66
return self.mapping.revision_id_bzr_to_foreign(ret)
67
except InvalidRevisionId:
72
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
89
def import_git_blob(texts, mapping, path, name, (base_hexsha, hexsha),
90
base_inv, parent_id, revision_id,
91
parent_invs, lookup_object, (base_mode, mode), store_updater,
73
93
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
95
:param texts: VersionedFiles to add to
76
96
:param path: Path in the tree
77
97
:param blob: A git blob
98
:return: Inventory delta for this file
79
file_id = mapping.generate_file_id(path)
80
text_revision = inv.revision_id
81
repo.texts.add_lines((file_id, text_revision),
82
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
83
osutils.split_lines(blob.data))
84
ie = inv.add_path(path, "file", file_id)
85
ie.revision = text_revision
86
ie.text_size = len(blob.data)
87
ie.text_sha1 = osutils.sha_string(blob.data)
88
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
100
if mapping.is_control_file(path):
102
if base_hexsha == hexsha and base_mode == mode:
103
# If nothing has changed since the base revision, we're done
105
file_id = lookup_file_id(path)
106
if stat.S_ISLNK(mode):
110
ie = cls(file_id, name.decode("utf-8"), parent_id)
111
ie.executable = mode_is_executable(mode)
112
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
113
base_ie = base_inv[base_inv.path2id(path)]
114
ie.text_size = base_ie.text_size
115
ie.text_sha1 = base_ie.text_sha1
116
ie.symlink_target = base_ie.symlink_target
117
if ie.executable == base_ie.executable:
118
ie.revision = base_ie.revision
120
blob = lookup_object(hexsha)
122
blob = lookup_object(hexsha)
123
if ie.kind == "symlink":
125
ie.symlink_target = blob.data
129
ie.text_size = sum(imap(len, blob.chunked))
130
ie.text_sha1 = osutils.sha_strings(blob.chunked)
131
# Check what revision we should store
133
for pinv in parent_invs:
138
if (pie.text_sha1 == ie.text_sha1 and
139
pie.executable == ie.executable and
140
pie.symlink_target == ie.symlink_target):
141
# found a revision in one of the parents to use
142
ie.revision = pie.revision
144
parent_key = (file_id, pie.revision)
145
if not parent_key in parent_keys:
146
parent_keys.append(parent_key)
147
if ie.revision is None:
148
# Need to store a new revision
149
ie.revision = revision_id
150
assert ie.revision is not None
151
if ie.kind == 'symlink':
154
chunks = blob.chunked
155
texts.insert_record_stream([
156
ChunkedContentFactory((file_id, ie.revision),
157
tuple(parent_keys), ie.text_sha1, chunks)])
159
if base_hexsha is not None:
160
old_path = path # Renames are not supported yet
161
if stat.S_ISDIR(base_mode):
162
invdelta.extend(remove_disappeared_children(base_inv, old_path,
163
lookup_object(base_hexsha), [], lookup_object))
166
invdelta.append((old_path, path, file_id, ie))
167
if base_hexsha != hexsha:
168
store_updater.add_object(blob, ie)
172
class SubmodulesRequireSubtrees(BzrError):
173
_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'."""
177
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
178
base_inv, parent_id, revision_id, parent_invs, lookup_object,
179
(base_mode, mode), store_updater, lookup_file_id):
180
if base_hexsha == hexsha and base_mode == mode:
182
file_id = lookup_file_id(path)
183
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
184
ie.revision = revision_id
185
if base_hexsha is None:
189
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
190
texts.insert_record_stream([
191
ChunkedContentFactory((file_id, ie.revision), (), None, [])])
192
invdelta = [(oldpath, path, file_id, ie)]
196
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
199
for name, mode, hexsha in base_tree.iteritems():
200
if name in existing_children:
202
c_path = posixpath.join(path, name.decode("utf-8"))
203
ret.append((c_path, None, base_inv.path2id(c_path), None))
204
if stat.S_ISDIR(mode):
205
ret.extend(remove_disappeared_children(
206
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
210
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
211
base_inv, parent_id, revision_id, parent_invs,
212
lookup_object, (base_mode, mode), store_updater,
213
lookup_file_id, allow_submodules=False):
92
214
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
216
:param texts: VersionedFiles object to add to
95
217
:param path: Path in the tree
96
218
:param tree: A git tree object
97
:param inv: Inventory object
219
:param base_inv: Base inventory against which to return inventory delta
220
:return: Inventory delta for this subtree
99
file_id = mapping.generate_file_id(path)
100
text_revision = inv.revision_id
101
repo.texts.add_lines((file_id, text_revision),
102
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
104
ie = inv.add_path(path, "directory", file_id)
105
ie.revision = text_revision
106
for mode, name, hexsha in tree.entries():
107
entry_kind = (mode & 0700000) / 0100000
108
basename = name.decode("utf-8")
112
child_path = urlutils.join(path, name)
114
tree = lookup_object(hexsha)
115
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
116
elif entry_kind == 1:
117
blob = lookup_object(hexsha)
118
fs_mode = mode & 0777
119
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
121
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
124
def import_git_objects(repo, mapping, object_iter, pb=None):
222
if base_hexsha == hexsha and base_mode == mode:
223
# If nothing has changed since the base revision, we're done
226
file_id = lookup_file_id(path)
227
# We just have to hope this is indeed utf-8:
228
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
229
tree = lookup_object(hexsha)
230
if base_hexsha is None:
232
old_path = None # Newly appeared here
234
base_tree = lookup_object(base_hexsha)
235
old_path = path # Renames aren't supported yet
236
if base_tree is None or type(base_tree) is not Tree:
237
ie.revision = revision_id
238
invdelta.append((old_path, path, ie.file_id, ie))
239
texts.insert_record_stream([
240
ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
241
# Remember for next time
242
existing_children = set()
244
for child_mode, name, child_hexsha in tree.entries():
245
existing_children.add(name)
246
child_path = posixpath.join(path, name)
247
if type(base_tree) is Tree:
249
child_base_mode, child_base_hexsha = base_tree[name]
251
child_base_hexsha = None
254
child_base_hexsha = None
256
if stat.S_ISDIR(child_mode):
257
subinvdelta, grandchildmodes = import_git_tree(texts, mapping,
258
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
259
file_id, revision_id, parent_invs, lookup_object,
260
(child_base_mode, child_mode), store_updater, lookup_file_id,
261
allow_submodules=allow_submodules)
262
elif S_ISGITLINK(child_mode): # submodule
263
if not allow_submodules:
264
raise SubmodulesRequireSubtrees()
265
subinvdelta, grandchildmodes = import_git_submodule(texts, mapping,
266
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
267
file_id, revision_id, parent_invs, lookup_object,
268
(child_base_mode, child_mode), store_updater, lookup_file_id)
270
subinvdelta = import_git_blob(texts, mapping, child_path, name,
271
(child_base_hexsha, child_hexsha), base_inv, file_id,
272
revision_id, parent_invs, lookup_object,
273
(child_base_mode, child_mode), store_updater, lookup_file_id)
275
child_modes.update(grandchildmodes)
276
invdelta.extend(subinvdelta)
277
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
278
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
279
child_modes[child_path] = child_mode
280
# Remove any children that have disappeared
281
if base_tree is not None and type(base_tree) is Tree:
282
invdelta.extend(remove_disappeared_children(base_inv, old_path,
283
base_tree, existing_children, lookup_object))
284
store_updater.add_object(tree, ie)
285
return invdelta, child_modes
288
def verify_commit_reconstruction(target_git_object_retriever, lookup_object,
289
o, rev, ret_tree, parent_trees, mapping, unusual_modes):
290
new_unusual_modes = mapping.export_unusual_file_modes(rev)
291
if new_unusual_modes != unusual_modes:
292
raise AssertionError("unusual modes don't match: %r != %r" % (
293
unusual_modes, new_unusual_modes))
294
# Verify that we can reconstruct the commit properly
295
rec_o = target_git_object_retriever._reconstruct_commit(rev, o.tree)
297
raise AssertionError("Reconstructed commit differs: %r != %r" % (
301
for path, obj, ie in _tree_to_objects(ret_tree, parent_trees,
302
target_git_object_retriever._cache.idmap, unusual_modes, mapping.BZR_DUMMY_FILE):
303
old_obj_id = tree_lookup_path(lookup_object, o.tree, path)[1]
305
if obj.id != old_obj_id:
306
diff.append((path, lookup_object(old_obj_id), obj))
307
for (path, old_obj, new_obj) in diff:
308
while (old_obj.type_name == "tree" and
309
new_obj.type_name == "tree" and
310
sorted(old_obj) == sorted(new_obj)):
312
if old_obj[name][0] != new_obj[name][0]:
313
raise AssertionError("Modes for %s differ: %o != %o" %
314
(path, old_obj[name][0], new_obj[name][0]))
315
if old_obj[name][1] != new_obj[name][1]:
316
# Found a differing child, delve deeper
317
path = posixpath.join(path, name)
318
old_obj = lookup_object(old_obj[name][1])
319
new_obj = new_objs[path]
321
raise AssertionError("objects differ for %s: %r != %r" % (path,
325
def import_git_commit(repo, mapping, head, lookup_object,
326
target_git_object_retriever, trees_cache):
327
o = lookup_object(head)
328
rev = mapping.import_commit(o)
329
# We have to do this here, since we have to walk the tree and
330
# we need to make sure to import the blobs / trees with the right
331
# path; this may involve adding them more than once.
332
parent_trees = trees_cache.revision_trees(rev.parent_ids)
333
if parent_trees == []:
334
base_inv = Inventory(root_id=None)
338
base_inv = parent_trees[0].inventory
339
base_tree = lookup_object(o.parents[0]).tree
340
base_mode = stat.S_IFDIR
341
store_updater = target_git_object_retriever._get_updater(rev)
342
store_updater.add_object(o, None)
343
root_tree = lookup_object(o.tree)
345
file_id_map_sha = root_tree[mapping.BZR_FILE_IDS_FILE][1]
349
file_ids = mapping.import_fileid_map(lookup_object(file_id_map_sha))
350
lookup_file_id = GitFileIdMap(file_ids, mapping).lookup_file_id
351
inv_delta, unusual_modes = import_git_tree(repo.texts,
352
mapping, "", u"", (base_tree, o.tree), base_inv,
353
None, rev.revision_id, [p.inventory for p in parent_trees],
354
lookup_object, (base_mode, stat.S_IFDIR), store_updater,
356
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
357
store_updater.finish()
358
if unusual_modes != {}:
359
for path, mode in unusual_modes.iteritems():
360
warn_unusual_mode(rev.foreign_revid, path, mode)
361
mapping.import_unusual_file_modes(rev, unusual_modes)
363
basis_id = rev.parent_ids[0]
365
basis_id = NULL_REVISION
367
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
368
inv_delta, rev.revision_id, rev.parent_ids,
370
ret_tree = RevisionTree(repo, inv, rev.revision_id)
371
trees_cache.add(ret_tree)
372
repo.add_revision(rev.revision_id, rev)
373
if "verify" in debug.debug_flags:
374
verify_commit_reconstruction(target_git_object_retriever,
375
lookup_object, o, rev, ret_tree, parent_trees, mapping,
379
def import_git_objects(repo, mapping, object_iter,
380
target_git_object_retriever, heads, pb=None, limit=None):
125
381
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
383
:param repo: Target Bazaar repository
128
384
:param mapping: Mapping to use
129
385
:param object_iter: Iterator over Git objects.
386
:return: Tuple with pack hints and last imported revision id
131
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
388
def lookup_object(sha):
390
return object_iter[sha]
392
return target_git_object_retriever[sha]
395
heads = list(set(heads))
396
trees_cache = LRUTreeCache(repo)
140
397
# Find and convert commit objects
141
for o in objects.itervalues():
400
pb.update("finding revisions to fetch", len(graph), None)
402
assert isinstance(head, str)
404
o = lookup_object(head)
142
407
if isinstance(o, Commit):
143
408
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
145
revisions[rev.revision_id] = rev
146
graph.append((rev.revision_id, rev.parent_ids))
409
if repo.has_revision(rev.revision_id):
411
squash_revision(repo, rev)
412
graph.append((o.id, o.parents))
413
heads.extend([p for p in o.parents if p not in checked])
414
elif isinstance(o, Tag):
415
if o.object[1] not in checked:
416
heads.append(o.object[1])
418
trace.warning("Unable to import head object %r" % o)
147
421
# Order the revisions
148
422
# Create the inventory objects
149
for i, revid in enumerate(topo_sort(graph)):
151
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
rev = revisions[revid]
154
# We have to do this here, since we have to walk the tree and
155
# we need to make sure to import the blobs / trees with the riht
156
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
def lookup_object(sha):
162
return reconstruct_git_object(repo, mapping, sha)
163
parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
164
import_git_tree(repo, mapping, "", root_tree, inv, parent_invs, lookup_object)
165
repo.add_revision(rev.revision_id, rev, inv)
168
def reconstruct_git_commit(repo, rev):
169
raise NotImplementedError(self.reconstruct_git_commit)
172
def reconstruct_git_object(repo, mapping, sha):
174
revid = mapping.revision_id_foreign_to_bzr(sha)
176
rev = repo.get_revision(revid)
177
except NoSuchRevision:
180
return reconstruct_git_commit(rev)
184
raise KeyError("No such object %s" % sha)
424
revision_ids = topo_sort(graph)
426
if limit is not None:
427
revision_ids = revision_ids[:limit]
429
for offset in range(0, len(revision_ids), batch_size):
430
target_git_object_retriever.start_write_group()
432
repo.start_write_group()
434
for i, head in enumerate(
435
revision_ids[offset:offset+batch_size]):
437
pb.update("fetching revisions", offset+i,
439
import_git_commit(repo, mapping, head, lookup_object,
440
target_git_object_retriever, trees_cache)
443
repo.abort_write_group()
446
hint = repo.commit_write_group()
448
pack_hints.extend(hint)
450
target_git_object_retriever.abort_write_group()
453
target_git_object_retriever.commit_write_group()
454
return pack_hints, last_imported
187
457
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
459
_matching_repo_format = GitRepositoryFormat()
192
462
def _get_repo_format_to_test():
196
466
"""See InterRepository.copy_content."""
197
467
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
469
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
470
mapping=None, fetch_spec=None):
471
self.fetch_refs(revision_id=revision_id, pb=pb,
472
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
475
class InterGitNonGitRepository(InterGitRepository):
476
"""Base InterRepository that copies revisions from a Git into a non-Git
479
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
480
mapping=None, fetch_spec=None):
201
481
if mapping is None:
202
482
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
483
if revision_id is not None:
484
interesting_heads = [revision_id]
485
elif fetch_spec is not None:
486
interesting_heads = fetch_spec.heads
488
interesting_heads = None
490
def determine_wants(refs):
492
if interesting_heads is None:
493
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
495
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
496
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
211
graph_walker = BzrFetchGraphWalker(self.target, mapping)
497
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
498
if pack_hint is not None and self.target._format.pack_compresses:
499
self.target.pack(hint=pack_hint)
500
if interesting_heads is not None:
501
present_interesting_heads = self.target.has_revisions(interesting_heads)
502
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
503
if missing_interesting_heads:
504
raise AssertionError("Missing interesting heads: %r" %
505
missing_interesting_heads)
509
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
510
def report_git_progress(pb, text):
511
text = text.rstrip("\r\n")
512
g = _GIT_PROGRESS_RE.match(text)
514
(text, pct, current, total) = g.groups()
515
pb.update(text, int(current), int(total))
517
pb.update(text, 0, 0)
520
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
521
"""InterRepository that copies revisions from a remote Git into a non-Git
524
def get_target_heads(self):
525
# FIXME: This should be more efficient
526
all_revs = self.target.all_revision_ids()
527
parent_map = self.target.get_parent_map(all_revs)
529
map(all_parents.update, parent_map.itervalues())
530
return set(all_revs) - all_parents
532
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
534
report_git_progress(pb, text)
535
store = BazaarObjectStore(self.target, mapping)
536
self.target.lock_write()
538
heads = self.get_target_heads()
539
graph_walker = store.get_graph_walker(
540
[store._lookup_revision_sha1(head) for head in heads])
543
def record_determine_wants(heads):
544
wants = determine_wants(heads)
545
recorded_wants.extend(wants)
550
create_pb = pb = ui.ui_factory.nested_progress_bar()
552
objects_iter = self.source.fetch_objects(record_determine_wants,
553
graph_walker, store.get_raw, progress)
554
return import_git_objects(self.target, mapping,
555
objects_iter, store, recorded_wants, pb, limit)
563
def is_compatible(source, target):
564
"""Be compatible with GitRepository."""
565
return (isinstance(source, RemoteGitRepository) and
566
target.supports_rich_root() and
567
not isinstance(target, GitRepository) and
568
target.texts is not None)
571
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
572
"""InterRepository that copies revisions from a local Git into a non-Git
575
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
578
wants = determine_wants(self.source._git.get_refs())
214
581
create_pb = pb = ui.ui_factory.nested_progress_bar()
582
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
584
self.target.lock_write()
218
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
224
self.target.commit_write_group()
586
return import_git_objects(self.target, mapping,
587
self.source._git.object_store,
588
target_git_object_retriever, wants, pb, limit)
226
590
self.target.unlock()
added
removed
fetch.py
