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 (
22
from dulwich.object_store import (
35
from bzrlib.errors import (
39
from bzrlib.inventory import (
46
from bzrlib.repository import (
49
from bzrlib.revision import (
52
from bzrlib.tsort import (
55
from bzrlib.versionedfile import (
56
FulltextContentFactory,
59
from bzrlib.plugins.git.mapping import (
61
inventory_to_tree_and_blobs,
66
from bzrlib.plugins.git.object_store import (
70
from bzrlib.plugins.git.remote import (
73
from bzrlib.plugins.git.repository import (
80
def import_git_blob(texts, mapping, path, hexsha, base_inv, base_inv_shamap,
81
base_ie, parent_id, revision_id, parent_invs, lookup_object,
68
83
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
85
:param texts: VersionedFiles to add to
71
86
:param path: Path in the tree
72
87
:param blob: A git blob
88
:return: Inventory delta for this file
74
90
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):
95
# We just have to hope this is indeed utf-8:
96
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
97
ie.executable = executable
98
# See if this has changed at all
103
base_sha = base_inv_shamap.lookup_blob(file_id,
104
revision_hint=base_ie.revision)
108
if (base_sha == hexsha and base_ie.executable == ie.executable
109
and base_ie.kind == ie.kind):
110
# If nothing has changed since the base revision, we're done
112
if base_sha == hexsha and base_ie.kind == ie.kind:
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 = len(blob.data)
129
ie.text_sha1 = osutils.sha_string(blob.data)
130
# Check what revision we should store
132
for pinv in parent_invs:
133
if pinv.revision_id == base_inv.revision_id:
142
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
143
# found a revision in one of the parents to use
144
ie.revision = pie.revision
146
parent_keys.append((file_id, pie.revision))
147
if ie.revision is None:
148
# Need to store a new revision
149
ie.revision = revision_id
150
assert file_id is not None
151
assert ie.revision is not None
152
if ie.kind == 'symlink':
156
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, data)])
157
shamap = { ie.file_id: hexsha }
159
if base_ie is not None:
160
old_path = base_inv.id2path(file_id)
161
if base_ie.kind == "directory":
162
invdelta.extend(remove_disappeared_children(old_path, base_ie.children, []))
165
invdelta.append((old_path, path, file_id, ie))
166
return (invdelta, shamap)
169
class SubmodulesRequireSubtrees(BzrError):
170
_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'."""
174
def import_git_submodule(texts, mapping, path, hexsha, base_inv, base_ie,
175
parent_id, revision_id, parent_invs, lookup_object):
176
file_id = mapping.generate_file_id(path)
177
ie = TreeReference(file_id, urlutils.basename(path.decode("utf-8")),
179
ie.revision = revision_id
184
if (base_ie.kind == ie.kind and
185
base_ie.reference_revision == ie.reference_revision):
186
ie.revision = base_ie.revision
187
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
188
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
189
invdelta = [(oldpath, path, file_id, ie)]
190
return invdelta, {}, {}
193
def remove_disappeared_children(path, base_children, existing_children):
195
deletable = [(osutils.pathjoin(path, k), v) for k,v in base_children.iteritems() if k not in existing_children]
197
(path, ie) = deletable.pop()
198
ret.append((path, None, ie.file_id, None))
199
if ie.kind == "directory":
200
for name, child_ie in ie.children.iteritems():
201
deletable.append((osutils.pathjoin(path, name), child_ie))
205
def import_git_tree(texts, mapping, path, hexsha, base_inv, base_inv_shamap,
206
base_ie, parent_id, revision_id, parent_invs, lookup_object,
207
allow_submodules=False):
82
208
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
210
:param texts: VersionedFiles object to add to
85
211
:param path: Path in the tree
86
212
:param tree: A git tree object
87
:param inv: Inventory object
213
:param base_inv: Base inventory against which to return inventory delta
214
:return: Inventory delta for this subtree
89
217
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
218
# We just have to hope this is indeed utf-8:
219
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
222
# Newly appeared here
223
ie.revision = revision_id
224
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
225
invdelta.append((None, path, file_id, ie))
227
# See if this has changed at all
229
base_sha = base_inv_shamap.lookup_tree(file_id)
230
except (KeyError, NotImplementedError):
233
if base_sha == hexsha:
234
# If nothing has changed since the base revision, we're done
236
if base_ie.kind != "directory":
237
ie.revision = revision_id
238
texts.insert_record_stream([FulltextContentFactory((ie.file_id, ie.revision), (), None, "")])
239
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
240
if base_ie is not None and base_ie.kind == "directory":
241
base_children = base_ie.children
244
# Remember for next time
245
existing_children = set()
248
tree = lookup_object(hexsha)
249
for mode, name, child_hexsha in tree.entries():
96
250
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):
251
existing_children.add(basename)
252
child_path = osutils.pathjoin(path, name)
253
if stat.S_ISDIR(mode):
254
subinvdelta, grandchildmodes, subshamap = import_git_tree(
255
texts, mapping, child_path, child_hexsha, base_inv,
256
base_inv_shamap, base_children.get(basename), file_id,
257
revision_id, parent_invs, lookup_object,
258
allow_submodules=allow_submodules)
259
elif S_ISGITLINK(mode): # submodule
260
if not allow_submodules:
261
raise SubmodulesRequireSubtrees()
262
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
263
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
264
file_id, revision_id, parent_invs, lookup_object)
266
subinvdelta, subshamap = import_git_blob(texts, mapping,
267
child_path, child_hexsha, base_inv, base_inv_shamap,
268
base_children.get(basename), file_id,
269
revision_id, parent_invs, lookup_object,
270
mode_is_executable(mode), stat.S_ISLNK(mode))
272
child_modes.update(grandchildmodes)
273
invdelta.extend(subinvdelta)
274
shamap.update(subshamap)
275
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
276
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
277
child_modes[child_path] = mode
278
# Remove any children that have disappeared
279
if base_ie is not None and base_ie.kind == "directory":
280
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
281
base_children, existing_children))
282
shamap[file_id] = hexsha
283
return invdelta, child_modes, shamap
286
def import_git_commit(repo, mapping, head, lookup_object,
287
target_git_object_retriever, parent_invs_cache):
288
o = lookup_object(head)
289
rev = mapping.import_commit(o)
290
# We have to do this here, since we have to walk the tree and
291
# we need to make sure to import the blobs / trees with the right
292
# path; this may involve adding them more than once.
293
parent_invs = parent_invs_cache.get_inventories(rev.parent_ids)
294
if parent_invs == []:
295
base_inv = Inventory(root_id=None)
297
base_inv_shamap = None # Should never be accessed
299
base_inv = parent_invs[0]
300
base_ie = base_inv.root
301
base_inv_shamap = target_git_object_retriever._idmap.get_inventory_sha_map(base_inv.revision_id)
302
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
303
mapping, "", o.tree, base_inv, base_inv_shamap, base_ie, None,
304
rev.revision_id, parent_invs, lookup_object,
305
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
307
for (oldpath, newpath, fileid, new_ie) in inv_delta:
309
entries.append((fileid, None, None, None))
311
if new_ie.kind in ("file", "symlink"):
312
entries.append((fileid, "blob", shamap[fileid], new_ie.revision))
313
elif new_ie.kind == "directory":
314
entries.append((fileid, "tree", shamap[fileid], rev.revision_id))
317
target_git_object_retriever._idmap.add_entries(rev.revision_id,
318
rev.parent_ids, head, o.tree, entries)
319
if unusual_modes != {}:
320
for path, mode in unusual_modes.iteritems():
321
warn_unusual_mode(rev.foreign_revid, path, mode)
322
mapping.import_unusual_file_modes(rev, unusual_modes)
324
basis_id = rev.parent_ids[0]
326
basis_id = NULL_REVISION
328
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
329
inv_delta, rev.revision_id, rev.parent_ids,
331
parent_invs_cache.add(rev.revision_id, inv)
332
repo.add_revision(rev.revision_id, rev)
333
if "verify" in debug.debug_flags:
334
new_unusual_modes = mapping.export_unusual_file_modes(rev)
335
if new_unusual_modes != unusual_modes:
336
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
337
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
338
for newsha1, newobj, path in objs:
339
assert path is not None
343
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
344
if oldsha1 != newsha1:
345
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
348
def import_git_objects(repo, mapping, object_iter,
349
target_git_object_retriever, heads, pb=None, limit=None):
110
350
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
352
:param repo: Target Bazaar repository
113
353
:param mapping: Mapping to use
114
354
:param object_iter: Iterator over Git objects.
355
:return: Tuple with pack hints and last imported revision id
357
def lookup_object(sha):
359
return object_iter[sha]
361
return target_git_object_retriever[sha]
116
362
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
365
heads = list(set(heads))
366
parent_invs_cache = LRUInventoryCache(repo)
367
target_git_object_retriever.start_write_group() # FIXME: try/finally
121
368
# Find and convert commit objects
122
for o in objects.iterkeys():
371
pb.update("finding revisions to fetch", len(graph), None)
373
assert isinstance(head, str)
375
o = lookup_object(head)
123
378
if isinstance(o, Commit):
124
379
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
380
if repo.has_revision(rev.revision_id):
382
squash_revision(repo, rev)
383
graph.append((o.id, o.parents))
384
heads.extend([p for p in o.parents if p not in checked])
385
elif isinstance(o, Tag):
386
if o.object[1] not in checked:
387
heads.append(o.object[1])
389
trace.warning("Unable to import head object %r" % o)
392
# Order the revisions
126
393
# 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)
395
revision_ids = topo_sort(graph)
397
if limit is not None:
398
revision_ids = revision_ids[:limit]
400
for offset in range(0, len(revision_ids), batch_size):
401
repo.start_write_group()
403
for i, head in enumerate(revision_ids[offset:offset+batch_size]):
405
pb.update("fetching revisions", offset+i, len(revision_ids))
406
import_git_commit(repo, mapping, head, lookup_object,
407
target_git_object_retriever,
411
repo.abort_write_group()
414
hint = repo.commit_write_group()
416
pack_hints.extend(hint)
417
target_git_object_retriever.commit_write_group()
418
return pack_hints, last_imported
159
421
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
423
_matching_repo_format = GitRepositoryFormat()
164
426
def _get_repo_format_to_test():
168
430
"""See InterRepository.copy_content."""
169
431
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
433
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
434
mapping=None, fetch_spec=None):
435
self.fetch_refs(revision_id=revision_id, pb=pb,
436
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
439
class InterGitNonGitRepository(InterGitRepository):
440
"""Base InterRepository that copies revisions from a Git into a non-Git
443
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
444
mapping=None, fetch_spec=None):
173
445
if mapping is None:
174
446
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)]
447
if revision_id is not None:
448
interesting_heads = [revision_id]
449
elif fetch_spec is not None:
450
interesting_heads = fetch_spec.heads
452
interesting_heads = None
454
def determine_wants(refs):
456
if interesting_heads is None:
457
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
459
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
460
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)
461
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
462
if pack_hint is not None and self.target._format.pack_compresses:
463
self.target.pack(hint=pack_hint)
464
if interesting_heads is not None:
465
present_interesting_heads = self.target.has_revisions(interesting_heads)
466
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
467
if missing_interesting_heads:
468
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
472
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
473
def report_git_progress(pb, text):
474
text = text.rstrip("\r\n")
475
g = _GIT_PROGRESS_RE.match(text)
477
(text, pct, current, total) = g.groups()
478
pb.update(text, int(current), int(total))
480
pb.update(text, 0, 0)
483
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
484
"""InterRepository that copies revisions from a remote Git into a non-Git
487
def get_target_heads(self):
488
# FIXME: This should be more efficient
489
all_revs = self.target.all_revision_ids()
490
parent_map = self.target.get_parent_map(all_revs)
492
map(all_parents.update, parent_map.itervalues())
493
return set(all_revs) - all_parents
495
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
497
report_git_progress(pb, text)
498
store = BazaarObjectStore(self.target, mapping)
187
499
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
501
heads = self.get_target_heads()
502
graph_walker = store.get_graph_walker(
503
[store._lookup_revision_sha1(head) for head in heads])
506
def record_determine_wants(heads):
507
wants = determine_wants(heads)
508
recorded_wants.extend(wants)
513
create_pb = pb = ui.ui_factory.nested_progress_bar()
515
objects_iter = self.source.fetch_objects(
516
record_determine_wants, graph_walker,
517
store.get_raw, progress)
518
return import_git_objects(self.target, mapping,
519
objects_iter, store, recorded_wants, pb, limit)
193
524
self.target.unlock()
196
527
def is_compatible(source, target):
197
528
"""Be compatible with GitRepository."""
198
529
# FIXME: Also check target uses VersionedFile
199
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
530
return (isinstance(source, RemoteGitRepository) and
531
target.supports_rich_root() and
532
not isinstance(target, GitRepository))
535
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
536
"""InterRepository that copies revisions from a local Git into a non-Git
539
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
542
wants = determine_wants(self.source._git.get_refs())
545
create_pb = pb = ui.ui_factory.nested_progress_bar()
546
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
548
self.target.lock_write()
550
return import_git_objects(self.target, mapping,
551
self.source._git.object_store,
552
target_git_object_retriever, wants, pb, limit)
560
def is_compatible(source, target):
561
"""Be compatible with GitRepository."""
562
# FIXME: Also check target uses VersionedFile
563
return (isinstance(source, LocalGitRepository) and
564
target.supports_rich_root() and
565
not isinstance(target, GitRepository))
568
class InterGitGitRepository(InterGitRepository):
569
"""InterRepository that copies between Git repositories."""
571
def fetch_objects(self, determine_wants, mapping, pb=None):
573
trace.note("git: %s", text)
574
graphwalker = self.target._git.get_graph_walker()
575
if (isinstance(self.source, LocalGitRepository) and
576
isinstance(self.target, LocalGitRepository)):
577
return self.source._git.fetch(self.target._git, determine_wants,
579
elif (isinstance(self.source, LocalGitRepository) and
580
isinstance(self.target, RemoteGitRepository)):
581
raise NotImplementedError
582
elif (isinstance(self.source, RemoteGitRepository) and
583
isinstance(self.target, LocalGitRepository)):
584
f, commit = self.target._git.object_store.add_thin_pack()
586
refs = self.source._git.fetch_pack(determine_wants,
587
graphwalker, f.write, progress)
596
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
597
mapping=None, fetch_spec=None, branches=None):
599
mapping = self.source.get_mapping()
601
if revision_id is not None:
602
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
603
elif fetch_spec is not None:
604
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
605
if branches is not None:
606
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
607
elif fetch_spec is None and revision_id is None:
608
determine_wants = r.object_store.determine_wants_all
610
determine_wants = lambda x: [y for y in args if not y in r.object_store]
611
return self.fetch_objects(determine_wants, mapping)[0]
615
def is_compatible(source, target):
616
"""Be compatible with GitRepository."""
617
return (isinstance(source, GitRepository) and
618
isinstance(target, GitRepository))
added
removed
fetch.py
