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 (
72
from bzrlib.plugins.git.object_store import (
77
from bzrlib.plugins.git.remote import (
25
80
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):
87
def import_git_blob(texts, mapping, path, name, (base_hexsha, hexsha),
88
base_inv, parent_id, revision_id,
89
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
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):
96
:return: Inventory delta for this file
98
if mapping.is_control_file(path):
100
if base_hexsha == hexsha and base_mode == mode:
101
# If nothing has changed since the base revision, we're done
103
file_id = lookup_file_id(path)
104
if stat.S_ISLNK(mode):
108
ie = cls(file_id, name.decode("utf-8"), parent_id)
109
if ie.kind == "file":
110
ie.executable = mode_is_executable(mode)
111
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
112
base_ie = base_inv[base_inv.path2id(path)]
113
ie.text_size = base_ie.text_size
114
ie.text_sha1 = base_ie.text_sha1
115
if ie.kind == "symlink":
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
127
ie.text_size = sum(imap(len, blob.chunked))
128
ie.text_sha1 = osutils.sha_strings(blob.chunked)
129
# Check what revision we should store
131
for pinv in parent_invs:
136
if (pie.text_sha1 == ie.text_sha1 and
137
pie.executable == ie.executable and
138
pie.symlink_target == ie.symlink_target):
139
# found a revision in one of the parents to use
140
ie.revision = pie.revision
142
parent_key = (file_id, pie.revision)
143
if not parent_key in parent_keys:
144
parent_keys.append(parent_key)
145
if ie.revision is None:
146
# Need to store a new revision
147
ie.revision = revision_id
148
assert ie.revision is not None
149
if ie.kind == 'symlink':
152
chunks = blob.chunked
153
texts.insert_record_stream([
154
ChunkedContentFactory((file_id, ie.revision),
155
tuple(parent_keys), ie.text_sha1, chunks)])
157
if base_hexsha is not None:
158
old_path = path.decode("utf-8") # Renames are not supported yet
159
if stat.S_ISDIR(base_mode):
160
invdelta.extend(remove_disappeared_children(base_inv, old_path,
161
lookup_object(base_hexsha), [], lookup_object))
164
new_path = path.decode("utf-8")
165
invdelta.append((old_path, new_path, file_id, ie))
166
if base_hexsha != hexsha:
167
store_updater.add_object(blob, ie, path)
171
class SubmodulesRequireSubtrees(BzrError):
172
_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'."""
176
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
177
base_inv, parent_id, revision_id, parent_invs, lookup_object,
178
(base_mode, mode), store_updater, lookup_file_id):
179
if base_hexsha == hexsha and base_mode == mode:
181
file_id = lookup_file_id(path)
182
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
183
ie.revision = revision_id
184
if base_hexsha is None:
188
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
189
texts.insert_record_stream([
190
ChunkedContentFactory((file_id, ie.revision), (), None, [])])
191
invdelta = [(oldpath, path, file_id, ie)]
195
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
197
"""Generate an inventory delta for removed children.
199
:param base_inv: Base inventory against which to generate the
201
:param path: Path to process (unicode)
202
:param base_tree: Git Tree base object
203
:param existing_children: Children that still exist
204
:param lookup_object: Lookup a git object by its SHA1
205
:return: Inventory delta, as list
207
assert type(path) is unicode
209
for name, mode, hexsha in base_tree.iteritems():
210
if name in existing_children:
212
c_path = posixpath.join(path, name.decode("utf-8"))
213
file_id = base_inv.path2id(c_path)
214
assert file_id is not None
215
ret.append((c_path, None, file_id, None))
216
if stat.S_ISDIR(mode):
217
ret.extend(remove_disappeared_children(
218
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
222
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
223
base_inv, parent_id, revision_id, parent_invs,
224
lookup_object, (base_mode, mode), store_updater,
225
lookup_file_id, allow_submodules=False):
92
226
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
95
:param path: Path in the tree
228
:param texts: VersionedFiles object to add to
229
:param path: Path in the tree (str)
230
:param name: Name of the tree (str)
96
231
:param tree: A git tree object
97
:param inv: Inventory object
232
:param base_inv: Base inventory against which to return inventory delta
233
: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):
235
assert type(path) is str
236
assert type(name) is str
237
if base_hexsha == hexsha and base_mode == mode:
238
# If nothing has changed since the base revision, we're done
241
file_id = lookup_file_id(path)
242
# We just have to hope this is indeed utf-8:
243
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
244
tree = lookup_object(hexsha)
245
if base_hexsha is None:
247
old_path = None # Newly appeared here
249
base_tree = lookup_object(base_hexsha)
250
old_path = path.decode("utf-8") # Renames aren't supported yet
251
new_path = path.decode("utf-8")
252
if base_tree is None or type(base_tree) is not Tree:
253
ie.revision = revision_id
254
invdelta.append((old_path, new_path, ie.file_id, ie))
255
texts.insert_record_stream([
256
ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
257
# Remember for next time
258
existing_children = set()
260
for child_mode, name, child_hexsha in tree.entries():
261
existing_children.add(name)
262
child_path = posixpath.join(path, name)
263
if type(base_tree) is Tree:
265
child_base_mode, child_base_hexsha = base_tree[name]
267
child_base_hexsha = None
270
child_base_hexsha = None
272
if stat.S_ISDIR(child_mode):
273
subinvdelta, grandchildmodes = import_git_tree(texts, mapping,
274
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
275
file_id, revision_id, parent_invs, lookup_object,
276
(child_base_mode, child_mode), store_updater, lookup_file_id,
277
allow_submodules=allow_submodules)
278
elif S_ISGITLINK(child_mode): # submodule
279
if not allow_submodules:
280
raise SubmodulesRequireSubtrees()
281
subinvdelta, grandchildmodes = import_git_submodule(texts, mapping,
282
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
283
file_id, revision_id, parent_invs, lookup_object,
284
(child_base_mode, child_mode), store_updater, lookup_file_id)
286
subinvdelta = import_git_blob(texts, mapping, child_path, name,
287
(child_base_hexsha, child_hexsha), base_inv, file_id,
288
revision_id, parent_invs, lookup_object,
289
(child_base_mode, child_mode), store_updater, lookup_file_id)
291
child_modes.update(grandchildmodes)
292
invdelta.extend(subinvdelta)
293
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
294
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
295
child_modes[child_path] = child_mode
296
# Remove any children that have disappeared
297
if base_tree is not None and type(base_tree) is Tree:
298
invdelta.extend(remove_disappeared_children(base_inv, old_path,
299
base_tree, existing_children, lookup_object))
300
store_updater.add_object(tree, ie, path)
301
return invdelta, child_modes
304
def verify_commit_reconstruction(target_git_object_retriever, lookup_object,
305
o, rev, ret_tree, parent_trees, mapping, unusual_modes):
306
new_unusual_modes = mapping.export_unusual_file_modes(rev)
307
if new_unusual_modes != unusual_modes:
308
raise AssertionError("unusual modes don't match: %r != %r" % (
309
unusual_modes, new_unusual_modes))
310
# Verify that we can reconstruct the commit properly
311
rec_o = target_git_object_retriever._reconstruct_commit(rev, o.tree, True)
313
raise AssertionError("Reconstructed commit differs: %r != %r" % (
317
for path, obj, ie in _tree_to_objects(ret_tree, parent_trees,
318
target_git_object_retriever._cache.idmap, unusual_modes, mapping.BZR_DUMMY_FILE):
319
old_obj_id = tree_lookup_path(lookup_object, o.tree, path)[1]
321
if obj.id != old_obj_id:
322
diff.append((path, lookup_object(old_obj_id), obj))
323
for (path, old_obj, new_obj) in diff:
324
while (old_obj.type_name == "tree" and
325
new_obj.type_name == "tree" and
326
sorted(old_obj) == sorted(new_obj)):
328
if old_obj[name][0] != new_obj[name][0]:
329
raise AssertionError("Modes for %s differ: %o != %o" %
330
(path, old_obj[name][0], new_obj[name][0]))
331
if old_obj[name][1] != new_obj[name][1]:
332
# Found a differing child, delve deeper
333
path = posixpath.join(path, name)
334
old_obj = lookup_object(old_obj[name][1])
335
new_obj = new_objs[path]
337
raise AssertionError("objects differ for %s: %r != %r" % (path,
341
def import_git_commit(repo, mapping, head, lookup_object,
342
target_git_object_retriever, trees_cache):
343
o = lookup_object(head)
344
rev = mapping.import_commit(o,
345
lambda x: target_git_object_retriever.lookup_git_sha(x)[1][0])
346
# We have to do this here, since we have to walk the tree and
347
# we need to make sure to import the blobs / trees with the right
348
# path; this may involve adding them more than once.
349
parent_trees = trees_cache.revision_trees(rev.parent_ids)
350
if parent_trees == []:
351
base_inv = Inventory(root_id=None)
355
base_inv = parent_trees[0].inventory
356
base_tree = lookup_object(o.parents[0]).tree
357
base_mode = stat.S_IFDIR
358
store_updater = target_git_object_retriever._get_updater(rev)
359
store_updater.add_object(o, None, None)
360
fileid_map = mapping.get_fileid_map(lookup_object, o.tree)
361
inv_delta, unusual_modes = import_git_tree(repo.texts,
362
mapping, "", "", (base_tree, o.tree), base_inv,
363
None, rev.revision_id, [p.inventory for p in parent_trees],
364
lookup_object, (base_mode, stat.S_IFDIR), store_updater,
365
fileid_map.lookup_file_id,
366
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
367
store_updater.finish()
368
if unusual_modes != {}:
369
for path, mode in unusual_modes.iteritems():
370
warn_unusual_mode(rev.foreign_revid, path, mode)
371
mapping.import_unusual_file_modes(rev, unusual_modes)
373
basis_id = rev.parent_ids[0]
375
basis_id = NULL_REVISION
377
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
378
inv_delta, rev.revision_id, rev.parent_ids, base_inv)
379
ret_tree = RevisionTree(repo, inv, rev.revision_id)
380
trees_cache.add(ret_tree)
381
repo.add_revision(rev.revision_id, rev)
382
if "verify" in debug.debug_flags:
383
verify_commit_reconstruction(target_git_object_retriever,
384
lookup_object, o, rev, ret_tree, parent_trees, mapping,
388
def import_git_objects(repo, mapping, object_iter,
389
target_git_object_retriever, heads, pb=None, limit=None):
125
390
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
392
:param repo: Target Bazaar repository
128
393
:param mapping: Mapping to use
129
394
:param object_iter: Iterator over Git objects.
395
: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)
397
def lookup_object(sha):
399
return object_iter[sha]
401
return target_git_object_retriever[sha]
404
heads = list(set(heads))
405
trees_cache = LRUTreeCache(repo)
140
406
# Find and convert commit objects
141
for o in objects.itervalues():
409
pb.update("finding revisions to fetch", len(graph), None)
411
assert isinstance(head, str)
413
o = lookup_object(head)
142
416
if isinstance(o, Commit):
143
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))
417
rev = mapping.import_commit(o, lambda x: None)
418
if repo.has_revision(rev.revision_id):
420
graph.append((o.id, o.parents))
421
heads.extend([p for p in o.parents if p not in checked])
422
elif isinstance(o, Tag):
423
if o.object[1] not in checked:
424
heads.append(o.object[1])
426
trace.warning("Unable to import head object %r" % o)
147
429
# Order the revisions
148
430
# 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)
432
revision_ids = topo_sort(graph)
434
if limit is not None:
435
revision_ids = revision_ids[:limit]
437
for offset in range(0, len(revision_ids), batch_size):
438
target_git_object_retriever.start_write_group()
440
repo.start_write_group()
442
for i, head in enumerate(
443
revision_ids[offset:offset+batch_size]):
445
pb.update("fetching revisions", offset+i,
447
import_git_commit(repo, mapping, head, lookup_object,
448
target_git_object_retriever, trees_cache)
451
repo.abort_write_group()
454
hint = repo.commit_write_group()
456
pack_hints.extend(hint)
458
target_git_object_retriever.abort_write_group()
461
target_git_object_retriever.commit_write_group()
462
return pack_hints, last_imported
187
465
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
467
_matching_repo_format = GitRepositoryFormat()
192
470
def _get_repo_format_to_test():
196
474
"""See InterRepository.copy_content."""
197
475
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
478
class InterGitNonGitRepository(InterGitRepository):
479
"""Base InterRepository that copies revisions from a Git into a non-Git
482
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
483
"""Fetch objects from a remote server.
485
:param determine_wants: determine_wants callback
486
:param mapping: BzrGitMapping to use
487
:param pb: Optional progress bar
488
:param limit: Maximum number of commits to import.
489
:return: Tuple with pack hint, last imported revision id and remote refs
491
raise NotImplementedError(self.fetch_objects)
493
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
494
mapping=None, fetch_spec=None):
201
495
if mapping is None:
202
496
mapping = self.source.get_mapping()
497
if revision_id is not None:
498
interesting_heads = [revision_id]
499
elif fetch_spec is not None:
500
interesting_heads = fetch_spec.heads
502
interesting_heads = None
503
def determine_wants(refs):
504
if interesting_heads is None:
505
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
507
ret = [self.source.lookup_bzr_revision_id(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
508
return [rev for rev in ret if not self.target.has_revision(self.source.lookup_foreign_revision_id(rev))]
509
(pack_hint, _, remote_refs) = self.fetch_objects(determine_wants, mapping, pb)
510
if pack_hint is not None and self.target._format.pack_compresses:
511
self.target.pack(hint=pack_hint)
515
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
516
def report_git_progress(pb, text):
517
text = text.rstrip("\r\n")
518
g = _GIT_PROGRESS_RE.match(text)
520
(text, pct, current, total) = g.groups()
521
pb.update(text, int(current), int(total))
523
pb.update(text, 0, 0)
526
class DetermineWantsRecorder(object):
528
def __init__(self, actual):
531
self.remote_refs = {}
533
def __call__(self, refs):
534
self.remote_refs = refs
535
self.wants = self.actual(refs)
539
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
540
"""InterRepository that copies revisions from a remote Git into a non-Git
543
def get_target_heads(self):
544
# FIXME: This should be more efficient
545
all_revs = self.target.all_revision_ids()
546
parent_map = self.target.get_parent_map(all_revs)
548
map(all_parents.update, parent_map.itervalues())
549
return set(all_revs) - all_parents
551
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
552
"""See `InterGitNonGitRepository`."""
203
553
def progress(text):
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
210
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)
554
report_git_progress(pb, text)
555
store = BazaarObjectStore(self.target, mapping)
556
self.target.lock_write()
558
heads = self.get_target_heads()
559
graph_walker = store.get_graph_walker(
560
[store._lookup_revision_sha1(head) for head in heads])
561
wants_recorder = DetermineWantsRecorder(determine_wants)
565
create_pb = pb = ui.ui_factory.nested_progress_bar()
567
objects_iter = self.source.fetch_objects(
568
wants_recorder, graph_walker, store.get_raw,
570
(pack_hint, last_rev) = import_git_objects(self.target, mapping,
571
objects_iter, store, wants_recorder.wants, pb, limit)
572
return (pack_hint, last_rev, wants_recorder.remote_refs)
580
def is_compatible(source, target):
581
"""Be compatible with GitRepository."""
582
return (isinstance(source, RemoteGitRepository) and
583
target.supports_rich_root() and
584
not isinstance(target, GitRepository) and
585
target.texts is not None)
588
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
589
"""InterRepository that copies revisions from a local Git into a non-Git
592
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
593
"""See `InterGitNonGitRepository`."""
594
remote_refs = self.source._git.get_refs()
595
wants = determine_wants(remote_refs)
214
598
create_pb = pb = ui.ui_factory.nested_progress_bar()
599
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
601
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()
603
(pack_hint, last_rev) = import_git_objects(self.target, mapping,
604
self.source._git.object_store,
605
target_git_object_retriever, wants, pb, limit)
606
return (pack_hint, last_rev, remote_refs)
226
608
self.target.unlock()
added
removed
fetch.py
