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 (
73
from bzrlib.plugins.git.object_store import (
78
from bzrlib.plugins.git.remote import (
25
81
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):
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,
73
92
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
94
:param texts: VersionedFiles to add to
76
95
:param path: Path in the tree
77
96
:param blob: A git blob
97
: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):
99
if base_hexsha == hexsha and base_mode == mode:
100
# If nothing has changed since the base revision, we're done
102
file_id = lookup_file_id(path)
103
if stat.S_ISLNK(mode):
107
ie = cls(file_id, name.decode("utf-8"), parent_id)
108
ie.executable = mode_is_executable(mode)
109
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
110
base_ie = base_inv[base_inv.path2id(path)]
111
ie.text_size = base_ie.text_size
112
ie.text_sha1 = base_ie.text_sha1
113
ie.symlink_target = base_ie.symlink_target
114
if ie.executable == base_ie.executable:
115
ie.revision = base_ie.revision
117
blob = lookup_object(hexsha)
119
blob = lookup_object(hexsha)
120
if ie.kind == "symlink":
122
ie.symlink_target = blob.data
126
ie.text_size = sum(imap(len, blob.chunked))
127
ie.text_sha1 = osutils.sha_strings(blob.chunked)
128
# Check what revision we should store
130
for pinv in parent_invs[1:]:
135
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
136
# found a revision in one of the parents to use
137
ie.revision = pie.revision
139
parent_keys.append((file_id, pie.revision))
140
if ie.revision is None:
141
# Need to store a new revision
142
ie.revision = revision_id
143
assert ie.revision is not None
144
if ie.kind == 'symlink':
147
chunks = blob.chunked
148
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, chunks)])
150
if base_hexsha is not None:
151
old_path = path # Renames are not supported yet
152
if stat.S_ISDIR(base_mode):
153
invdelta.extend(remove_disappeared_children(base_inv, old_path,
154
lookup_object(base_hexsha), [], lookup_object))
157
invdelta.append((old_path, path, file_id, ie))
158
if base_hexsha != hexsha:
159
store_updater.add_object(blob, ie)
163
class SubmodulesRequireSubtrees(BzrError):
164
_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'."""
168
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
169
base_inv, parent_id, revision_id, parent_invs, lookup_object,
170
(base_mode, mode), store_updater, lookup_file_id):
171
if base_hexsha == hexsha and base_mode == mode:
173
file_id = lookup_file_id(path)
174
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
175
ie.revision = revision_id
176
if base_hexsha is None:
180
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
181
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
182
invdelta = [(oldpath, path, file_id, ie)]
186
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
189
for name, mode, hexsha in base_tree.iteritems():
190
if name in existing_children:
192
c_path = posixpath.join(path, name.decode("utf-8"))
193
ret.append((c_path, None, base_inv.path2id(c_path), None))
194
if stat.S_ISDIR(mode):
195
ret.extend(remove_disappeared_children(
196
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
200
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
201
base_inv, parent_id, revision_id, parent_invs,
202
lookup_object, (base_mode, mode), store_updater,
203
lookup_file_id, allow_submodules=False):
92
204
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
206
:param texts: VersionedFiles object to add to
95
207
:param path: Path in the tree
96
208
:param tree: A git tree object
97
:param inv: Inventory object
209
:param base_inv: Base inventory against which to return inventory delta
210
: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):
212
if base_hexsha == hexsha and base_mode == mode:
213
# If nothing has changed since the base revision, we're done
216
file_id = lookup_file_id(path)
217
# We just have to hope this is indeed utf-8:
218
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
219
tree = lookup_object(hexsha)
220
if base_hexsha is None:
222
old_path = None # Newly appeared here
224
base_tree = lookup_object(base_hexsha)
225
old_path = path # Renames aren't supported yet
226
if base_tree is None or type(base_tree) is not Tree:
227
ie.revision = revision_id
228
invdelta.append((old_path, path, ie.file_id, ie))
229
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
230
# Remember for next time
231
existing_children = set()
233
for child_mode, name, child_hexsha in tree.entries():
234
existing_children.add(name)
235
child_path = posixpath.join(path, name)
236
if type(base_tree) is Tree:
238
child_base_mode, child_base_hexsha = base_tree[name]
240
child_base_hexsha = None
243
child_base_hexsha = None
245
if stat.S_ISDIR(child_mode):
246
subinvdelta, grandchildmodes = import_git_tree(
247
texts, mapping, child_path, name,
248
(child_base_hexsha, child_hexsha),
249
base_inv, file_id, revision_id, parent_invs, lookup_object,
250
(child_base_mode, child_mode), store_updater,
251
lookup_file_id, allow_submodules=allow_submodules)
252
elif S_ISGITLINK(child_mode): # submodule
253
if not allow_submodules:
254
raise SubmodulesRequireSubtrees()
255
subinvdelta, grandchildmodes = import_git_submodule(
256
texts, mapping, child_path, name,
257
(child_base_hexsha, child_hexsha),
258
base_inv, file_id, revision_id, parent_invs, lookup_object,
259
(child_base_mode, child_mode), store_updater,
262
subinvdelta = import_git_blob(texts, mapping,
263
child_path, name, (child_base_hexsha, child_hexsha),
264
base_inv, file_id, revision_id, parent_invs, lookup_object,
265
(child_base_mode, child_mode), store_updater,
268
child_modes.update(grandchildmodes)
269
invdelta.extend(subinvdelta)
270
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
271
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
272
child_modes[child_path] = child_mode
273
# Remove any children that have disappeared
274
if base_tree is not None and type(base_tree) is Tree:
275
invdelta.extend(remove_disappeared_children(base_inv, old_path,
276
base_tree, existing_children, lookup_object))
277
store_updater.add_object(tree, ie)
278
return invdelta, child_modes
281
def verify_commit_reconstruction(target_git_object_retriever, lookup_object,
282
o, rev, ret_tree, parent_trees, mapping, unusual_modes):
283
new_unusual_modes = mapping.export_unusual_file_modes(rev)
284
if new_unusual_modes != unusual_modes:
285
raise AssertionError("unusual modes don't match: %r != %r" % (
286
unusual_modes, new_unusual_modes))
287
# Verify that we can reconstruct the commit properly
288
rec_o = target_git_object_retriever._reconstruct_commit(rev, o.tree)
290
raise AssertionError("Reconstructed commit differs: %r != %r" % (
294
for path, obj, ie in _tree_to_objects(ret_tree, parent_trees,
295
target_git_object_retriever._cache.idmap, unusual_modes):
296
old_obj_id = tree_lookup_path(lookup_object, o.tree, path)[1]
298
if obj.id != old_obj_id:
299
diff.append((path, lookup_object(old_obj_id), obj))
300
for (path, old_obj, new_obj) in diff:
301
while (old_obj.type_name == "tree" and
302
new_obj.type_name == "tree" and
303
sorted(old_obj) == sorted(new_obj)):
305
if old_obj[name][0] != new_obj[name][0]:
306
raise AssertionError("Modes for %s differ: %o != %o" % (path, old_obj[name][0], new_obj[name][0]))
307
if old_obj[name][1] != new_obj[name][1]:
308
# Found a differing child, delve deeper
309
path = posixpath.join(path, name)
310
old_obj = lookup_object(old_obj[name][1])
311
new_obj = new_objs[path]
313
raise AssertionError("objects differ for %s: %r != %r" % (path,
317
def import_git_commit(repo, mapping, head, lookup_object,
318
target_git_object_retriever, trees_cache):
319
o = lookup_object(head)
320
rev = mapping.import_commit(o)
321
# We have to do this here, since we have to walk the tree and
322
# we need to make sure to import the blobs / trees with the right
323
# path; this may involve adding them more than once.
324
parent_trees = trees_cache.revision_trees(rev.parent_ids)
325
if parent_trees == []:
326
base_inv = Inventory(root_id=None)
330
base_inv = parent_trees[0].inventory
331
base_tree = lookup_object(o.parents[0]).tree
332
base_mode = stat.S_IFDIR
333
store_updater = target_git_object_retriever._get_updater(rev)
334
store_updater.add_object(o, None)
335
inv_delta, unusual_modes = import_git_tree(repo.texts,
336
mapping, "", u"", (base_tree, o.tree), base_inv,
337
None, rev.revision_id, [p.inventory for p in parent_trees],
338
lookup_object, (base_mode, stat.S_IFDIR), store_updater,
339
mapping.generate_file_id,
340
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
341
store_updater.finish()
342
if unusual_modes != {}:
343
for path, mode in unusual_modes.iteritems():
344
warn_unusual_mode(rev.foreign_revid, path, mode)
345
mapping.import_unusual_file_modes(rev, unusual_modes)
347
basis_id = rev.parent_ids[0]
349
basis_id = NULL_REVISION
351
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
352
inv_delta, rev.revision_id, rev.parent_ids,
354
ret_tree = RevisionTree(repo, inv, rev.revision_id)
355
trees_cache.add(ret_tree)
356
repo.add_revision(rev.revision_id, rev)
357
if "verify" in debug.debug_flags:
358
verify_commit_reconstruction(target_git_object_retriever,
359
lookup_object, o, rev, ret_tree, parent_trees, mapping,
363
def import_git_objects(repo, mapping, object_iter,
364
target_git_object_retriever, heads, pb=None, limit=None):
125
365
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
367
:param repo: Target Bazaar repository
128
368
:param mapping: Mapping to use
129
369
:param object_iter: Iterator over Git objects.
370
: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)
372
def lookup_object(sha):
374
return object_iter[sha]
376
return target_git_object_retriever[sha]
379
heads = list(set(heads))
380
trees_cache = LRUTreeCache(repo)
140
381
# Find and convert commit objects
141
for o in objects.itervalues():
384
pb.update("finding revisions to fetch", len(graph), None)
386
assert isinstance(head, str)
388
o = lookup_object(head)
142
391
if isinstance(o, Commit):
143
392
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))
393
if repo.has_revision(rev.revision_id):
395
squash_revision(repo, rev)
396
graph.append((o.id, o.parents))
397
heads.extend([p for p in o.parents if p not in checked])
398
elif isinstance(o, Tag):
399
if o.object[1] not in checked:
400
heads.append(o.object[1])
402
trace.warning("Unable to import head object %r" % o)
147
405
# Order the revisions
148
406
# 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)
408
revision_ids = topo_sort(graph)
410
if limit is not None:
411
revision_ids = revision_ids[:limit]
413
for offset in range(0, len(revision_ids), batch_size):
414
target_git_object_retriever.start_write_group()
416
repo.start_write_group()
418
for i, head in enumerate(
419
revision_ids[offset:offset+batch_size]):
421
pb.update("fetching revisions", offset+i,
423
import_git_commit(repo, mapping, head, lookup_object,
424
target_git_object_retriever,
428
repo.abort_write_group()
431
hint = repo.commit_write_group()
433
pack_hints.extend(hint)
435
target_git_object_retriever.abort_write_group()
438
target_git_object_retriever.commit_write_group()
439
return pack_hints, last_imported
187
442
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
444
_matching_repo_format = GitRepositoryFormat()
192
447
def _get_repo_format_to_test():
196
451
"""See InterRepository.copy_content."""
197
452
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
454
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
455
mapping=None, fetch_spec=None):
456
self.fetch_refs(revision_id=revision_id, pb=pb,
457
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
460
class InterGitNonGitRepository(InterGitRepository):
461
"""Base InterRepository that copies revisions from a Git into a non-Git
464
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
465
mapping=None, fetch_spec=None):
201
466
if mapping is None:
202
467
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
468
if revision_id is not None:
469
interesting_heads = [revision_id]
470
elif fetch_spec is not None:
471
interesting_heads = fetch_spec.heads
473
interesting_heads = None
475
def determine_wants(refs):
477
if interesting_heads is None:
478
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
480
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
481
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)
482
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
483
if pack_hint is not None and self.target._format.pack_compresses:
484
self.target.pack(hint=pack_hint)
485
if interesting_heads is not None:
486
present_interesting_heads = self.target.has_revisions(interesting_heads)
487
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
488
if missing_interesting_heads:
489
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
493
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
494
def report_git_progress(pb, text):
495
text = text.rstrip("\r\n")
496
g = _GIT_PROGRESS_RE.match(text)
498
(text, pct, current, total) = g.groups()
499
pb.update(text, int(current), int(total))
501
pb.update(text, 0, 0)
504
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
505
"""InterRepository that copies revisions from a remote Git into a non-Git
508
def get_target_heads(self):
509
# FIXME: This should be more efficient
510
all_revs = self.target.all_revision_ids()
511
parent_map = self.target.get_parent_map(all_revs)
513
map(all_parents.update, parent_map.itervalues())
514
return set(all_revs) - all_parents
516
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
518
report_git_progress(pb, text)
519
store = BazaarObjectStore(self.target, mapping)
520
self.target.lock_write()
522
heads = self.get_target_heads()
523
graph_walker = store.get_graph_walker(
524
[store._lookup_revision_sha1(head) for head in heads])
527
def record_determine_wants(heads):
528
wants = determine_wants(heads)
529
recorded_wants.extend(wants)
534
create_pb = pb = ui.ui_factory.nested_progress_bar()
536
objects_iter = self.source.fetch_objects(
537
record_determine_wants, graph_walker,
538
store.get_raw, progress)
539
return import_git_objects(self.target, mapping,
540
objects_iter, store, recorded_wants, pb, limit)
548
def is_compatible(source, target):
549
"""Be compatible with GitRepository."""
550
# FIXME: Also check target uses VersionedFile
551
return (isinstance(source, RemoteGitRepository) and
552
target.supports_rich_root() and
553
not isinstance(target, GitRepository))
556
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
557
"""InterRepository that copies revisions from a local Git into a non-Git
560
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
563
wants = determine_wants(self.source._git.get_refs())
214
566
create_pb = pb = ui.ui_factory.nested_progress_bar()
567
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
569
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()
571
return import_git_objects(self.target, mapping,
572
self.source._git.object_store,
573
target_git_object_retriever, wants, pb, limit)
226
575
self.target.unlock()
added
removed
fetch.py
