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
ie.executable = mode_is_executable(mode)
110
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
111
base_ie = base_inv[base_inv.path2id(path)]
112
ie.text_size = base_ie.text_size
113
ie.text_sha1 = base_ie.text_sha1
114
ie.symlink_target = base_ie.symlink_target
115
if ie.executable == base_ie.executable:
116
ie.revision = base_ie.revision
118
blob = lookup_object(hexsha)
120
blob = lookup_object(hexsha)
121
if ie.kind == "symlink":
123
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 # 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
invdelta.append((old_path, path, file_id, ie))
165
if base_hexsha != hexsha:
166
store_updater.add_object(blob, ie, path)
170
class SubmodulesRequireSubtrees(BzrError):
171
_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'."""
175
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
176
base_inv, parent_id, revision_id, parent_invs, lookup_object,
177
(base_mode, mode), store_updater, lookup_file_id):
178
if base_hexsha == hexsha and base_mode == mode:
180
file_id = lookup_file_id(path)
181
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
182
ie.revision = revision_id
183
if base_hexsha is None:
187
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
188
texts.insert_record_stream([
189
ChunkedContentFactory((file_id, ie.revision), (), None, [])])
190
invdelta = [(oldpath, path, file_id, ie)]
194
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
196
"""Generate an inventory delta for removed children.
198
:param base_inv: Base inventory against which to generate the
200
:param path: Path to process
201
:param base_tree: Git Tree base object
202
:param existing_children: Children that still exist
203
:param lookup_object: Lookup a git object by its SHA1
204
:return: Inventory delta, as list
207
for name, mode, hexsha in base_tree.iteritems():
208
if name in existing_children:
210
c_path = posixpath.join(path, name.decode("utf-8"))
211
file_id = base_inv.path2id(c_path)
212
assert file_id is not None
213
ret.append((c_path, None, file_id, None))
214
if stat.S_ISDIR(mode):
215
ret.extend(remove_disappeared_children(
216
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
220
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
221
base_inv, parent_id, revision_id, parent_invs,
222
lookup_object, (base_mode, mode), store_updater,
223
lookup_file_id, allow_submodules=False):
92
224
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
226
:param texts: VersionedFiles object to add to
95
227
:param path: Path in the tree
96
228
:param tree: A git tree object
97
:param inv: Inventory object
229
:param base_inv: Base inventory against which to return inventory delta
230
: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):
232
if base_hexsha == hexsha and base_mode == mode:
233
# If nothing has changed since the base revision, we're done
236
file_id = lookup_file_id(path)
237
# We just have to hope this is indeed utf-8:
238
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
239
tree = lookup_object(hexsha)
240
if base_hexsha is None:
242
old_path = None # Newly appeared here
244
base_tree = lookup_object(base_hexsha)
245
old_path = path # Renames aren't supported yet
246
if base_tree is None or type(base_tree) is not Tree:
247
ie.revision = revision_id
248
invdelta.append((old_path, path, ie.file_id, ie))
249
texts.insert_record_stream([
250
ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
251
# Remember for next time
252
existing_children = set()
254
for child_mode, name, child_hexsha in tree.entries():
255
existing_children.add(name)
256
child_path = posixpath.join(path, name)
257
if type(base_tree) is Tree:
259
child_base_mode, child_base_hexsha = base_tree[name]
261
child_base_hexsha = None
264
child_base_hexsha = None
266
if stat.S_ISDIR(child_mode):
267
subinvdelta, grandchildmodes = import_git_tree(texts, mapping,
268
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
269
file_id, revision_id, parent_invs, lookup_object,
270
(child_base_mode, child_mode), store_updater, lookup_file_id,
271
allow_submodules=allow_submodules)
272
elif S_ISGITLINK(child_mode): # submodule
273
if not allow_submodules:
274
raise SubmodulesRequireSubtrees()
275
subinvdelta, grandchildmodes = import_git_submodule(texts, mapping,
276
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
277
file_id, revision_id, parent_invs, lookup_object,
278
(child_base_mode, child_mode), store_updater, lookup_file_id)
280
subinvdelta = import_git_blob(texts, mapping, child_path, name,
281
(child_base_hexsha, child_hexsha), base_inv, file_id,
282
revision_id, parent_invs, lookup_object,
283
(child_base_mode, child_mode), store_updater, lookup_file_id)
285
child_modes.update(grandchildmodes)
286
invdelta.extend(subinvdelta)
287
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
288
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
289
child_modes[child_path] = child_mode
290
# Remove any children that have disappeared
291
if base_tree is not None and type(base_tree) is Tree:
292
invdelta.extend(remove_disappeared_children(base_inv, old_path,
293
base_tree, existing_children, lookup_object))
294
store_updater.add_object(tree, ie, path)
295
return invdelta, child_modes
298
def verify_commit_reconstruction(target_git_object_retriever, lookup_object,
299
o, rev, ret_tree, parent_trees, mapping, unusual_modes):
300
new_unusual_modes = mapping.export_unusual_file_modes(rev)
301
if new_unusual_modes != unusual_modes:
302
raise AssertionError("unusual modes don't match: %r != %r" % (
303
unusual_modes, new_unusual_modes))
304
# Verify that we can reconstruct the commit properly
305
rec_o = target_git_object_retriever._reconstruct_commit(rev, o.tree, True)
307
raise AssertionError("Reconstructed commit differs: %r != %r" % (
311
for path, obj, ie in _tree_to_objects(ret_tree, parent_trees,
312
target_git_object_retriever._cache.idmap, unusual_modes, mapping.BZR_DUMMY_FILE):
313
old_obj_id = tree_lookup_path(lookup_object, o.tree, path)[1]
315
if obj.id != old_obj_id:
316
diff.append((path, lookup_object(old_obj_id), obj))
317
for (path, old_obj, new_obj) in diff:
318
while (old_obj.type_name == "tree" and
319
new_obj.type_name == "tree" and
320
sorted(old_obj) == sorted(new_obj)):
322
if old_obj[name][0] != new_obj[name][0]:
323
raise AssertionError("Modes for %s differ: %o != %o" %
324
(path, old_obj[name][0], new_obj[name][0]))
325
if old_obj[name][1] != new_obj[name][1]:
326
# Found a differing child, delve deeper
327
path = posixpath.join(path, name)
328
old_obj = lookup_object(old_obj[name][1])
329
new_obj = new_objs[path]
331
raise AssertionError("objects differ for %s: %r != %r" % (path,
335
def import_git_commit(repo, mapping, head, lookup_object,
336
target_git_object_retriever, trees_cache):
337
o = lookup_object(head)
338
rev = mapping.import_commit(o,
339
lambda x: target_git_object_retriever.lookup_git_sha(x)[1][0])
340
# We have to do this here, since we have to walk the tree and
341
# we need to make sure to import the blobs / trees with the right
342
# path; this may involve adding them more than once.
343
parent_trees = trees_cache.revision_trees(rev.parent_ids)
344
if parent_trees == []:
345
base_inv = Inventory(root_id=None)
349
base_inv = parent_trees[0].inventory
350
base_tree = lookup_object(o.parents[0]).tree
351
base_mode = stat.S_IFDIR
352
store_updater = target_git_object_retriever._get_updater(rev)
353
store_updater.add_object(o, None, None)
354
lookup_file_id = mapping.get_fileid_map(lookup_object, o.tree).lookup_file_id
355
inv_delta, unusual_modes = import_git_tree(repo.texts,
356
mapping, "", u"", (base_tree, o.tree), base_inv,
357
None, rev.revision_id, [p.inventory for p in parent_trees],
358
lookup_object, (base_mode, stat.S_IFDIR), store_updater,
360
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
361
store_updater.finish()
362
if unusual_modes != {}:
363
for path, mode in unusual_modes.iteritems():
364
warn_unusual_mode(rev.foreign_revid, path, mode)
365
mapping.import_unusual_file_modes(rev, unusual_modes)
367
basis_id = rev.parent_ids[0]
369
basis_id = NULL_REVISION
371
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
372
inv_delta, rev.revision_id, rev.parent_ids,
374
ret_tree = RevisionTree(repo, inv, rev.revision_id)
375
trees_cache.add(ret_tree)
376
repo.add_revision(rev.revision_id, rev)
377
if "verify" in debug.debug_flags:
378
verify_commit_reconstruction(target_git_object_retriever,
379
lookup_object, o, rev, ret_tree, parent_trees, mapping,
383
def import_git_objects(repo, mapping, object_iter,
384
target_git_object_retriever, heads, pb=None, limit=None):
125
385
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
387
:param repo: Target Bazaar repository
128
388
:param mapping: Mapping to use
129
389
:param object_iter: Iterator over Git objects.
390
: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)
392
def lookup_object(sha):
394
return object_iter[sha]
396
return target_git_object_retriever[sha]
399
heads = list(set(heads))
400
trees_cache = LRUTreeCache(repo)
140
401
# Find and convert commit objects
141
for o in objects.itervalues():
404
pb.update("finding revisions to fetch", len(graph), None)
406
assert isinstance(head, str)
408
o = lookup_object(head)
142
411
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))
412
rev = mapping.import_commit(o, lambda x: None)
413
if repo.has_revision(rev.revision_id):
415
graph.append((o.id, o.parents))
416
heads.extend([p for p in o.parents if p not in checked])
417
elif isinstance(o, Tag):
418
if o.object[1] not in checked:
419
heads.append(o.object[1])
421
trace.warning("Unable to import head object %r" % o)
147
424
# Order the revisions
148
425
# 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)
427
revision_ids = topo_sort(graph)
429
if limit is not None:
430
revision_ids = revision_ids[:limit]
432
for offset in range(0, len(revision_ids), batch_size):
433
target_git_object_retriever.start_write_group()
435
repo.start_write_group()
437
for i, head in enumerate(
438
revision_ids[offset:offset+batch_size]):
440
pb.update("fetching revisions", offset+i,
442
import_git_commit(repo, mapping, head, lookup_object,
443
target_git_object_retriever, trees_cache)
446
repo.abort_write_group()
449
hint = repo.commit_write_group()
451
pack_hints.extend(hint)
453
target_git_object_retriever.abort_write_group()
456
target_git_object_retriever.commit_write_group()
457
return pack_hints, last_imported
187
460
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
462
_matching_repo_format = GitRepositoryFormat()
192
465
def _get_repo_format_to_test():
196
469
"""See InterRepository.copy_content."""
197
470
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
473
class InterGitNonGitRepository(InterGitRepository):
474
"""Base InterRepository that copies revisions from a Git into a non-Git
477
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
478
mapping=None, fetch_spec=None):
201
479
if mapping is None:
202
480
mapping = self.source.get_mapping()
481
if revision_id is not None:
482
interesting_heads = [revision_id]
483
elif fetch_spec is not None:
484
interesting_heads = fetch_spec.heads
486
interesting_heads = None
488
def determine_wants(refs):
490
if interesting_heads is None:
491
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
493
ret = [self.source.lookup_bzr_revision_id(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
494
return [rev for rev in ret if not self.target.has_revision(self.source.lookup_foreign_revision_id(rev))]
495
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
496
if pack_hint is not None and self.target._format.pack_compresses:
497
self.target.pack(hint=pack_hint)
501
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
502
def report_git_progress(pb, text):
503
text = text.rstrip("\r\n")
504
g = _GIT_PROGRESS_RE.match(text)
506
(text, pct, current, total) = g.groups()
507
pb.update(text, int(current), int(total))
509
pb.update(text, 0, 0)
512
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
513
"""InterRepository that copies revisions from a remote Git into a non-Git
516
def get_target_heads(self):
517
# FIXME: This should be more efficient
518
all_revs = self.target.all_revision_ids()
519
parent_map = self.target.get_parent_map(all_revs)
521
map(all_parents.update, parent_map.itervalues())
522
return set(all_revs) - all_parents
524
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
203
525
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)
526
report_git_progress(pb, text)
527
store = BazaarObjectStore(self.target, mapping)
528
self.target.lock_write()
530
heads = self.get_target_heads()
531
graph_walker = store.get_graph_walker(
532
[store._lookup_revision_sha1(head) for head in heads])
535
def record_determine_wants(heads):
536
wants = determine_wants(heads)
537
recorded_wants.extend(wants)
542
create_pb = pb = ui.ui_factory.nested_progress_bar()
544
objects_iter = self.source.fetch_objects(record_determine_wants,
545
graph_walker, store.get_raw, progress)
546
return import_git_objects(self.target, mapping,
547
objects_iter, store, recorded_wants, pb, limit)
555
def is_compatible(source, target):
556
"""Be compatible with GitRepository."""
557
return (isinstance(source, RemoteGitRepository) and
558
target.supports_rich_root() and
559
not isinstance(target, GitRepository) and
560
target.texts is not None)
563
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
564
"""InterRepository that copies revisions from a local Git into a non-Git
567
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
570
wants = determine_wants(self.source._git.get_refs())
214
573
create_pb = pb = ui.ui_factory.nested_progress_bar()
574
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
576
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()
578
return import_git_objects(self.target, mapping,
579
self.source._git.object_store,
580
target_git_object_retriever, wants, pb, limit)
226
582
self.target.unlock()
added
removed
fetch.py
