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.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 (
25
78
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):
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):
73
88
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
90
:param texts: VersionedFiles to add to
76
91
:param path: Path in the tree
77
92
:param blob: A git blob
93
: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):
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):
92
199
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
201
:param texts: VersionedFiles object to add to
95
202
:param path: Path in the tree
96
203
:param tree: A git tree object
97
: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
99
211
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
# 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)
290
base_inv = parent_invs[0]
291
base_tree = lookup_object(o.parents[0]).tree
292
base_mode = stat.S_IFDIR
293
store_updater = target_git_object_retriever._get_updater(rev)
294
store_updater.add_object(o, None)
295
inv_delta, unusual_modes = import_git_tree(repo.texts,
296
mapping, "", u"", (base_tree, o.tree), base_inv,
297
target_git_object_retriever._cache.idmap,
298
None, rev.revision_id, parent_invs, lookup_object,
299
(base_mode, stat.S_IFDIR), store_updater,
300
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
301
store_updater.finish()
302
if unusual_modes != {}:
303
for path, mode in unusual_modes.iteritems():
304
warn_unusual_mode(rev.foreign_revid, path, mode)
305
mapping.import_unusual_file_modes(rev, unusual_modes)
307
basis_id = rev.parent_ids[0]
309
basis_id = NULL_REVISION
311
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
312
inv_delta, rev.revision_id, rev.parent_ids,
314
parent_invs_cache.add(rev.revision_id, inv)
315
repo.add_revision(rev.revision_id, rev)
316
if "verify" in debug.debug_flags:
317
new_unusual_modes = mapping.export_unusual_file_modes(rev)
318
if new_unusual_modes != unusual_modes:
319
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
320
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
321
for newsha1, newobj, path in objs:
322
assert path is not None
326
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
327
if oldsha1 != newsha1:
328
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
331
def import_git_objects(repo, mapping, object_iter,
332
target_git_object_retriever, heads, pb=None, limit=None):
125
333
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
335
:param repo: Target Bazaar repository
128
336
:param mapping: Mapping to use
129
337
:param object_iter: Iterator over Git objects.
338
: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)
340
def lookup_object(sha):
342
return object_iter[sha]
344
return target_git_object_retriever[sha]
347
heads = list(set(heads))
348
parent_invs_cache = LRUInventoryCache(repo)
140
349
# Find and convert commit objects
141
for o in objects.itervalues():
352
pb.update("finding revisions to fetch", len(graph), None)
354
assert isinstance(head, str)
356
o = lookup_object(head)
142
359
if isinstance(o, Commit):
143
360
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))
361
if repo.has_revision(rev.revision_id):
363
squash_revision(repo, rev)
364
graph.append((o.id, o.parents))
365
heads.extend([p for p in o.parents if p not in checked])
366
elif isinstance(o, Tag):
367
if o.object[1] not in checked:
368
heads.append(o.object[1])
370
trace.warning("Unable to import head object %r" % o)
147
373
# Order the revisions
148
374
# 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)
376
revision_ids = topo_sort(graph)
378
if limit is not None:
379
revision_ids = revision_ids[:limit]
381
for offset in range(0, len(revision_ids), batch_size):
382
target_git_object_retriever.start_write_group()
384
repo.start_write_group()
386
for i, head in enumerate(
387
revision_ids[offset:offset+batch_size]):
389
pb.update("fetching revisions", offset+i,
391
import_git_commit(repo, mapping, head, lookup_object,
392
target_git_object_retriever,
396
repo.abort_write_group()
399
hint = repo.commit_write_group()
401
pack_hints.extend(hint)
403
target_git_object_retriever.abort_write_group()
406
target_git_object_retriever.commit_write_group()
407
return pack_hints, last_imported
187
410
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
412
_matching_repo_format = GitRepositoryFormat()
192
415
def _get_repo_format_to_test():
196
419
"""See InterRepository.copy_content."""
197
420
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
422
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
423
mapping=None, fetch_spec=None):
424
self.fetch_refs(revision_id=revision_id, pb=pb,
425
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
428
class InterGitNonGitRepository(InterGitRepository):
429
"""Base InterRepository that copies revisions from a Git into a non-Git
432
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
433
mapping=None, fetch_spec=None):
201
434
if mapping is None:
202
435
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
436
if revision_id is not None:
437
interesting_heads = [revision_id]
438
elif fetch_spec is not None:
439
interesting_heads = fetch_spec.heads
441
interesting_heads = None
443
def determine_wants(refs):
445
if interesting_heads is None:
446
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
448
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
449
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)
450
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
451
if pack_hint is not None and self.target._format.pack_compresses:
452
self.target.pack(hint=pack_hint)
453
if interesting_heads is not None:
454
present_interesting_heads = self.target.has_revisions(interesting_heads)
455
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
456
if missing_interesting_heads:
457
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
461
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
462
def report_git_progress(pb, text):
463
text = text.rstrip("\r\n")
464
g = _GIT_PROGRESS_RE.match(text)
466
(text, pct, current, total) = g.groups()
467
pb.update(text, int(current), int(total))
469
pb.update(text, 0, 0)
472
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
473
"""InterRepository that copies revisions from a remote Git into a non-Git
476
def get_target_heads(self):
477
# FIXME: This should be more efficient
478
all_revs = self.target.all_revision_ids()
479
parent_map = self.target.get_parent_map(all_revs)
481
map(all_parents.update, parent_map.itervalues())
482
return set(all_revs) - all_parents
484
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
486
report_git_progress(pb, text)
487
store = BazaarObjectStore(self.target, mapping)
488
self.target.lock_write()
490
heads = self.get_target_heads()
491
graph_walker = store.get_graph_walker(
492
[store._lookup_revision_sha1(head) for head in heads])
495
def record_determine_wants(heads):
496
wants = determine_wants(heads)
497
recorded_wants.extend(wants)
502
create_pb = pb = ui.ui_factory.nested_progress_bar()
504
objects_iter = self.source.fetch_objects(
505
record_determine_wants, graph_walker,
506
store.get_raw, progress)
507
return import_git_objects(self.target, mapping,
508
objects_iter, store, recorded_wants, pb, limit)
516
def is_compatible(source, target):
517
"""Be compatible with GitRepository."""
518
# FIXME: Also check target uses VersionedFile
519
return (isinstance(source, RemoteGitRepository) and
520
target.supports_rich_root() and
521
not isinstance(target, GitRepository))
524
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
525
"""InterRepository that copies revisions from a local Git into a non-Git
528
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
531
wants = determine_wants(self.source._git.get_refs())
214
534
create_pb = pb = ui.ui_factory.nested_progress_bar()
535
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
537
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()
539
return import_git_objects(self.target, mapping,
540
self.source._git.object_store,
541
target_git_object_retriever, wants, pb, limit)
226
543
self.target.unlock()
added
removed
fetch.py
