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
91
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
93
:param texts: VersionedFiles to add to
76
94
:param path: Path in the tree
77
95
:param blob: A git blob
96
: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):
98
if base_hexsha == hexsha and base_mode == mode:
99
# If nothing has changed since the base revision, we're done
101
file_id = mapping.generate_file_id(path)
102
if stat.S_ISLNK(mode):
106
ie = cls(file_id, name.decode("utf-8"), parent_id)
107
ie.executable = mode_is_executable(mode)
108
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
109
base_ie = base_inv[base_inv.path2id(path)]
110
ie.text_size = base_ie.text_size
111
ie.text_sha1 = base_ie.text_sha1
112
ie.symlink_target = base_ie.symlink_target
113
if ie.executable == base_ie.executable:
114
ie.revision = base_ie.revision
116
blob = lookup_object(hexsha)
118
blob = lookup_object(hexsha)
119
if ie.kind == "symlink":
121
ie.symlink_target = blob.data
125
ie.text_size = sum(imap(len, blob.chunked))
126
ie.text_sha1 = osutils.sha_strings(blob.chunked)
127
# Check what revision we should store
129
for pinv in parent_invs[1:]:
134
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
135
# found a revision in one of the parents to use
136
ie.revision = pie.revision
138
parent_keys.append((file_id, pie.revision))
139
if ie.revision is None:
140
# Need to store a new revision
141
ie.revision = revision_id
142
assert ie.revision is not None
143
if ie.kind == 'symlink':
146
chunks = blob.chunked
147
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, chunks)])
149
if base_hexsha is not None:
150
old_path = path # Renames are not supported yet
151
if stat.S_ISDIR(base_mode):
152
invdelta.extend(remove_disappeared_children(base_inv, old_path,
153
lookup_object(base_hexsha), [], lookup_object))
156
invdelta.append((old_path, path, file_id, ie))
157
if base_hexsha != hexsha:
158
store_updater.add_object(blob, ie)
162
class SubmodulesRequireSubtrees(BzrError):
163
_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'."""
167
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
168
base_inv, parent_id, revision_id, parent_invs, lookup_object,
169
(base_mode, mode), store_updater):
170
if base_hexsha == hexsha and base_mode == mode:
172
file_id = mapping.generate_file_id(path)
173
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
174
ie.revision = revision_id
175
if base_hexsha is None:
179
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
180
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
181
invdelta = [(oldpath, path, file_id, ie)]
185
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
188
for name, mode, hexsha in base_tree.iteritems():
189
if name in existing_children:
191
c_path = posixpath.join(path, name.decode("utf-8"))
192
ret.append((c_path, None, base_inv.path2id(c_path), None))
193
if stat.S_ISDIR(mode):
194
ret.extend(remove_disappeared_children(
195
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
199
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
200
base_inv, parent_id, revision_id, parent_invs,
201
lookup_object, (base_mode, mode), store_updater, allow_submodules=False):
92
202
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
204
:param texts: VersionedFiles object to add to
95
205
:param path: Path in the tree
96
206
:param tree: A git tree object
97
:param inv: Inventory object
207
:param base_inv: Base inventory against which to return inventory delta
208
:return: Inventory delta for this subtree
210
if base_hexsha == hexsha and base_mode == mode:
211
# If nothing has changed since the base revision, we're done
99
214
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):
215
# We just have to hope this is indeed utf-8:
216
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
217
tree = lookup_object(hexsha)
218
if base_hexsha is None:
220
old_path = None # Newly appeared here
222
base_tree = lookup_object(base_hexsha)
223
old_path = path # Renames aren't supported yet
224
if base_tree is None or type(base_tree) is not Tree:
225
ie.revision = revision_id
226
invdelta.append((old_path, path, ie.file_id, ie))
227
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
228
# Remember for next time
229
existing_children = set()
231
for child_mode, name, child_hexsha in tree.entries():
232
existing_children.add(name)
233
child_path = posixpath.join(path, name)
234
if type(base_tree) is Tree:
236
child_base_mode, child_base_hexsha = base_tree[name]
238
child_base_hexsha = None
241
child_base_hexsha = None
243
if stat.S_ISDIR(child_mode):
244
subinvdelta, grandchildmodes = import_git_tree(
245
texts, mapping, child_path, name,
246
(child_base_hexsha, child_hexsha),
247
base_inv, file_id, revision_id, parent_invs, lookup_object,
248
(child_base_mode, child_mode), store_updater,
249
allow_submodules=allow_submodules)
250
elif S_ISGITLINK(child_mode): # submodule
251
if not allow_submodules:
252
raise SubmodulesRequireSubtrees()
253
subinvdelta, grandchildmodes = import_git_submodule(
254
texts, mapping, child_path, name,
255
(child_base_hexsha, child_hexsha),
256
base_inv, file_id, revision_id, parent_invs, lookup_object,
257
(child_base_mode, child_mode), store_updater)
259
subinvdelta = import_git_blob(texts, mapping,
260
child_path, name, (child_base_hexsha, child_hexsha),
261
base_inv, file_id, 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 verify_commit_reconstruction(target_git_object_retriever, lookup_object,
278
o, rev, ret_tree, parent_trees, mapping, unusual_modes):
279
new_unusual_modes = mapping.export_unusual_file_modes(rev)
280
if new_unusual_modes != unusual_modes:
281
raise AssertionError("unusual modes don't match: %r != %r" % (
282
unusual_modes, new_unusual_modes))
283
# Verify that we can reconstruct the commit properly
284
rec_o = target_git_object_retriever._reconstruct_commit(rev, o.tree)
286
raise AssertionError("Reconstructed commit differs: %r != %r" % (
290
for path, obj, ie in _tree_to_objects(ret_tree, parent_trees,
291
target_git_object_retriever._cache.idmap, unusual_modes):
292
old_obj_id = tree_lookup_path(lookup_object, o.tree, path)[1]
294
if obj.id != old_obj_id:
295
diff.append((path, lookup_object(old_obj_id), obj))
296
for (path, old_obj, new_obj) in diff:
297
while (old_obj.type_name == "tree" and
298
new_obj.type_name == "tree" and
299
sorted(old_obj) == sorted(new_obj)):
301
if old_obj[name][0] != new_obj[name][0]:
302
raise AssertionError("Modes for %s differ: %o != %o" % (path, old_obj[name][0], new_obj[name][0]))
303
if old_obj[name][1] != new_obj[name][1]:
304
# Found a differing child, delve deeper
305
path = posixpath.join(path, name)
306
old_obj = lookup_object(old_obj[name][1])
307
new_obj = new_objs[path]
309
raise AssertionError("objects differ for %s: %r != %r" % (path,
313
def import_git_commit(repo, mapping, head, lookup_object,
314
target_git_object_retriever, trees_cache):
315
o = lookup_object(head)
316
rev = mapping.import_commit(o)
317
# We have to do this here, since we have to walk the tree and
318
# we need to make sure to import the blobs / trees with the right
319
# path; this may involve adding them more than once.
320
parent_trees = trees_cache.revision_trees(rev.parent_ids)
321
if parent_trees == []:
322
base_inv = Inventory(root_id=None)
326
base_inv = parent_trees[0].inventory
327
base_tree = lookup_object(o.parents[0]).tree
328
base_mode = stat.S_IFDIR
329
store_updater = target_git_object_retriever._get_updater(rev)
330
store_updater.add_object(o, None)
331
inv_delta, unusual_modes = import_git_tree(repo.texts,
332
mapping, "", u"", (base_tree, o.tree), base_inv,
333
None, rev.revision_id, [p.inventory for p in parent_trees],
334
lookup_object, (base_mode, stat.S_IFDIR), store_updater,
335
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
336
store_updater.finish()
337
if unusual_modes != {}:
338
for path, mode in unusual_modes.iteritems():
339
warn_unusual_mode(rev.foreign_revid, path, mode)
340
mapping.import_unusual_file_modes(rev, unusual_modes)
342
basis_id = rev.parent_ids[0]
344
basis_id = NULL_REVISION
346
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
347
inv_delta, rev.revision_id, rev.parent_ids,
349
ret_tree = RevisionTree(repo, inv, rev.revision_id)
350
trees_cache.add(ret_tree)
351
repo.add_revision(rev.revision_id, rev)
352
if "verify" in debug.debug_flags:
353
verify_commit_reconstruction(target_git_object_retriever,
354
lookup_object, o, rev, ret_tree, parent_trees, mapping,
358
def import_git_objects(repo, mapping, object_iter,
359
target_git_object_retriever, heads, pb=None, limit=None):
125
360
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
362
:param repo: Target Bazaar repository
128
363
:param mapping: Mapping to use
129
364
:param object_iter: Iterator over Git objects.
365
: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)
367
def lookup_object(sha):
369
return object_iter[sha]
371
return target_git_object_retriever[sha]
374
heads = list(set(heads))
375
trees_cache = LRUTreeCache(repo)
140
376
# Find and convert commit objects
141
for o in objects.itervalues():
379
pb.update("finding revisions to fetch", len(graph), None)
381
assert isinstance(head, str)
383
o = lookup_object(head)
142
386
if isinstance(o, Commit):
143
387
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))
388
if repo.has_revision(rev.revision_id):
390
squash_revision(repo, rev)
391
graph.append((o.id, o.parents))
392
heads.extend([p for p in o.parents if p not in checked])
393
elif isinstance(o, Tag):
394
if o.object[1] not in checked:
395
heads.append(o.object[1])
397
trace.warning("Unable to import head object %r" % o)
147
400
# Order the revisions
148
401
# 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)
403
revision_ids = topo_sort(graph)
405
if limit is not None:
406
revision_ids = revision_ids[:limit]
408
for offset in range(0, len(revision_ids), batch_size):
409
target_git_object_retriever.start_write_group()
411
repo.start_write_group()
413
for i, head in enumerate(
414
revision_ids[offset:offset+batch_size]):
416
pb.update("fetching revisions", offset+i,
418
import_git_commit(repo, mapping, head, lookup_object,
419
target_git_object_retriever,
423
repo.abort_write_group()
426
hint = repo.commit_write_group()
428
pack_hints.extend(hint)
430
target_git_object_retriever.abort_write_group()
433
target_git_object_retriever.commit_write_group()
434
return pack_hints, last_imported
187
437
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
439
_matching_repo_format = GitRepositoryFormat()
192
442
def _get_repo_format_to_test():
196
446
"""See InterRepository.copy_content."""
197
447
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
449
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
450
mapping=None, fetch_spec=None):
451
self.fetch_refs(revision_id=revision_id, pb=pb,
452
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
455
class InterGitNonGitRepository(InterGitRepository):
456
"""Base InterRepository that copies revisions from a Git into a non-Git
459
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
460
mapping=None, fetch_spec=None):
201
461
if mapping is None:
202
462
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
463
if revision_id is not None:
464
interesting_heads = [revision_id]
465
elif fetch_spec is not None:
466
interesting_heads = fetch_spec.heads
468
interesting_heads = None
470
def determine_wants(refs):
472
if interesting_heads is None:
473
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
475
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
476
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)
477
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
478
if pack_hint is not None and self.target._format.pack_compresses:
479
self.target.pack(hint=pack_hint)
480
if interesting_heads is not None:
481
present_interesting_heads = self.target.has_revisions(interesting_heads)
482
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
483
if missing_interesting_heads:
484
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
488
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
489
def report_git_progress(pb, text):
490
text = text.rstrip("\r\n")
491
g = _GIT_PROGRESS_RE.match(text)
493
(text, pct, current, total) = g.groups()
494
pb.update(text, int(current), int(total))
496
pb.update(text, 0, 0)
499
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
500
"""InterRepository that copies revisions from a remote Git into a non-Git
503
def get_target_heads(self):
504
# FIXME: This should be more efficient
505
all_revs = self.target.all_revision_ids()
506
parent_map = self.target.get_parent_map(all_revs)
508
map(all_parents.update, parent_map.itervalues())
509
return set(all_revs) - all_parents
511
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
513
report_git_progress(pb, text)
514
store = BazaarObjectStore(self.target, mapping)
515
self.target.lock_write()
517
heads = self.get_target_heads()
518
graph_walker = store.get_graph_walker(
519
[store._lookup_revision_sha1(head) for head in heads])
522
def record_determine_wants(heads):
523
wants = determine_wants(heads)
524
recorded_wants.extend(wants)
529
create_pb = pb = ui.ui_factory.nested_progress_bar()
531
objects_iter = self.source.fetch_objects(
532
record_determine_wants, graph_walker,
533
store.get_raw, progress)
534
return import_git_objects(self.target, mapping,
535
objects_iter, store, recorded_wants, pb, limit)
543
def is_compatible(source, target):
544
"""Be compatible with GitRepository."""
545
# FIXME: Also check target uses VersionedFile
546
return (isinstance(source, RemoteGitRepository) and
547
target.supports_rich_root() and
548
not isinstance(target, GitRepository))
551
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
552
"""InterRepository that copies revisions from a local Git into a non-Git
555
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
558
wants = determine_wants(self.source._git.get_refs())
214
561
create_pb = pb = ui.ui_factory.nested_progress_bar()
562
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
564
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()
566
return import_git_objects(self.target, mapping,
567
self.source._git.object_store,
568
target_git_object_retriever, wants, pb, limit)
226
570
self.target.unlock()
added
removed
fetch.py
