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
92
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
94
:param texts: VersionedFiles to add to
76
95
:param path: Path in the tree
77
96
:param blob: A git blob
97
: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):
99
if base_hexsha == hexsha and base_mode == mode:
100
# If nothing has changed since the base revision, we're done
102
file_id = lookup_file_id(path)
103
if stat.S_ISLNK(mode):
107
ie = cls(file_id, name.decode("utf-8"), parent_id)
108
ie.executable = mode_is_executable(mode)
109
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
110
base_ie = base_inv[base_inv.path2id(path)]
111
ie.text_size = base_ie.text_size
112
ie.text_sha1 = base_ie.text_sha1
113
ie.symlink_target = base_ie.symlink_target
114
if ie.executable == base_ie.executable:
115
ie.revision = base_ie.revision
117
blob = lookup_object(hexsha)
119
blob = lookup_object(hexsha)
120
if ie.kind == "symlink":
122
ie.symlink_target = blob.data
126
ie.text_size = sum(imap(len, blob.chunked))
127
ie.text_sha1 = osutils.sha_strings(blob.chunked)
128
# Check what revision we should store
130
for pinv in parent_invs:
135
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
136
# found a revision in one of the parents to use
137
ie.revision = pie.revision
139
parent_key = (file_id, pie.revision)
140
if not parent_key in parent_keys:
141
parent_keys.append(parent_key)
142
if ie.revision is None:
143
# Need to store a new revision
144
ie.revision = revision_id
145
assert ie.revision is not None
146
if ie.kind == 'symlink':
149
chunks = blob.chunked
150
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, chunks)])
152
if base_hexsha is not None:
153
old_path = path # Renames are not supported yet
154
if stat.S_ISDIR(base_mode):
155
invdelta.extend(remove_disappeared_children(base_inv, old_path,
156
lookup_object(base_hexsha), [], lookup_object))
159
invdelta.append((old_path, path, file_id, ie))
160
if base_hexsha != hexsha:
161
store_updater.add_object(blob, ie)
165
class SubmodulesRequireSubtrees(BzrError):
166
_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'."""
170
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
171
base_inv, parent_id, revision_id, parent_invs, lookup_object,
172
(base_mode, mode), store_updater, lookup_file_id):
173
if base_hexsha == hexsha and base_mode == mode:
175
file_id = lookup_file_id(path)
176
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
177
ie.revision = revision_id
178
if base_hexsha is None:
182
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
183
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
184
invdelta = [(oldpath, path, file_id, ie)]
188
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
191
for name, mode, hexsha in base_tree.iteritems():
192
if name in existing_children:
194
c_path = posixpath.join(path, name.decode("utf-8"))
195
ret.append((c_path, None, base_inv.path2id(c_path), None))
196
if stat.S_ISDIR(mode):
197
ret.extend(remove_disappeared_children(
198
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
202
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
203
base_inv, parent_id, revision_id, parent_invs,
204
lookup_object, (base_mode, mode), store_updater,
205
lookup_file_id, allow_submodules=False):
92
206
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
208
:param texts: VersionedFiles object to add to
95
209
:param path: Path in the tree
96
210
:param tree: A git tree object
97
:param inv: Inventory object
211
:param base_inv: Base inventory against which to return inventory delta
212
: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):
214
if base_hexsha == hexsha and base_mode == mode:
215
# If nothing has changed since the base revision, we're done
218
file_id = lookup_file_id(path)
219
# We just have to hope this is indeed utf-8:
220
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
221
tree = lookup_object(hexsha)
222
if base_hexsha is None:
224
old_path = None # Newly appeared here
226
base_tree = lookup_object(base_hexsha)
227
old_path = path # Renames aren't supported yet
228
if base_tree is None or type(base_tree) is not Tree:
229
ie.revision = revision_id
230
invdelta.append((old_path, path, ie.file_id, ie))
231
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
232
# Remember for next time
233
existing_children = set()
235
for child_mode, name, child_hexsha in tree.entries():
236
existing_children.add(name)
237
child_path = posixpath.join(path, name)
238
if type(base_tree) is Tree:
240
child_base_mode, child_base_hexsha = base_tree[name]
242
child_base_hexsha = None
245
child_base_hexsha = None
247
if stat.S_ISDIR(child_mode):
248
subinvdelta, grandchildmodes = import_git_tree(
249
texts, mapping, child_path, name,
250
(child_base_hexsha, child_hexsha),
251
base_inv, file_id, revision_id, parent_invs, lookup_object,
252
(child_base_mode, child_mode), store_updater,
253
lookup_file_id, allow_submodules=allow_submodules)
254
elif S_ISGITLINK(child_mode): # submodule
255
if not allow_submodules:
256
raise SubmodulesRequireSubtrees()
257
subinvdelta, grandchildmodes = import_git_submodule(
258
texts, mapping, child_path, name,
259
(child_base_hexsha, child_hexsha),
260
base_inv, file_id, revision_id, parent_invs, lookup_object,
261
(child_base_mode, child_mode), store_updater,
264
subinvdelta = import_git_blob(texts, mapping,
265
child_path, name, (child_base_hexsha, child_hexsha),
266
base_inv, file_id, revision_id, parent_invs, lookup_object,
267
(child_base_mode, child_mode), store_updater,
270
child_modes.update(grandchildmodes)
271
invdelta.extend(subinvdelta)
272
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
273
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
274
child_modes[child_path] = child_mode
275
# Remove any children that have disappeared
276
if base_tree is not None and type(base_tree) is Tree:
277
invdelta.extend(remove_disappeared_children(base_inv, old_path,
278
base_tree, existing_children, lookup_object))
279
store_updater.add_object(tree, ie)
280
return invdelta, child_modes
283
def verify_commit_reconstruction(target_git_object_retriever, lookup_object,
284
o, rev, ret_tree, parent_trees, mapping, unusual_modes):
285
new_unusual_modes = mapping.export_unusual_file_modes(rev)
286
if new_unusual_modes != unusual_modes:
287
raise AssertionError("unusual modes don't match: %r != %r" % (
288
unusual_modes, new_unusual_modes))
289
# Verify that we can reconstruct the commit properly
290
rec_o = target_git_object_retriever._reconstruct_commit(rev, o.tree)
292
raise AssertionError("Reconstructed commit differs: %r != %r" % (
296
for path, obj, ie in _tree_to_objects(ret_tree, parent_trees,
297
target_git_object_retriever._cache.idmap, unusual_modes):
298
old_obj_id = tree_lookup_path(lookup_object, o.tree, path)[1]
300
if obj.id != old_obj_id:
301
diff.append((path, lookup_object(old_obj_id), obj))
302
for (path, old_obj, new_obj) in diff:
303
while (old_obj.type_name == "tree" and
304
new_obj.type_name == "tree" and
305
sorted(old_obj) == sorted(new_obj)):
307
if old_obj[name][0] != new_obj[name][0]:
308
raise AssertionError("Modes for %s differ: %o != %o" % (path, old_obj[name][0], new_obj[name][0]))
309
if old_obj[name][1] != new_obj[name][1]:
310
# Found a differing child, delve deeper
311
path = posixpath.join(path, name)
312
old_obj = lookup_object(old_obj[name][1])
313
new_obj = new_objs[path]
315
raise AssertionError("objects differ for %s: %r != %r" % (path,
319
def import_git_commit(repo, mapping, head, lookup_object,
320
target_git_object_retriever, trees_cache):
321
o = lookup_object(head)
322
rev = mapping.import_commit(o)
323
# We have to do this here, since we have to walk the tree and
324
# we need to make sure to import the blobs / trees with the right
325
# path; this may involve adding them more than once.
326
parent_trees = trees_cache.revision_trees(rev.parent_ids)
327
if parent_trees == []:
328
base_inv = Inventory(root_id=None)
332
base_inv = parent_trees[0].inventory
333
base_tree = lookup_object(o.parents[0]).tree
334
base_mode = stat.S_IFDIR
335
store_updater = target_git_object_retriever._get_updater(rev)
336
store_updater.add_object(o, None)
337
inv_delta, unusual_modes = import_git_tree(repo.texts,
338
mapping, "", u"", (base_tree, o.tree), base_inv,
339
None, rev.revision_id, [p.inventory for p in parent_trees],
340
lookup_object, (base_mode, stat.S_IFDIR), store_updater,
341
mapping.generate_file_id,
342
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
343
store_updater.finish()
344
if unusual_modes != {}:
345
for path, mode in unusual_modes.iteritems():
346
warn_unusual_mode(rev.foreign_revid, path, mode)
347
mapping.import_unusual_file_modes(rev, unusual_modes)
349
basis_id = rev.parent_ids[0]
351
basis_id = NULL_REVISION
353
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
354
inv_delta, rev.revision_id, rev.parent_ids,
356
ret_tree = RevisionTree(repo, inv, rev.revision_id)
357
trees_cache.add(ret_tree)
358
repo.add_revision(rev.revision_id, rev)
359
if "verify" in debug.debug_flags:
360
verify_commit_reconstruction(target_git_object_retriever,
361
lookup_object, o, rev, ret_tree, parent_trees, mapping,
365
def import_git_objects(repo, mapping, object_iter,
366
target_git_object_retriever, heads, pb=None, limit=None):
125
367
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
369
:param repo: Target Bazaar repository
128
370
:param mapping: Mapping to use
129
371
:param object_iter: Iterator over Git objects.
372
: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)
374
def lookup_object(sha):
376
return object_iter[sha]
378
return target_git_object_retriever[sha]
381
heads = list(set(heads))
382
trees_cache = LRUTreeCache(repo)
140
383
# Find and convert commit objects
141
for o in objects.itervalues():
386
pb.update("finding revisions to fetch", len(graph), None)
388
assert isinstance(head, str)
390
o = lookup_object(head)
142
393
if isinstance(o, Commit):
143
394
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))
395
if repo.has_revision(rev.revision_id):
397
squash_revision(repo, rev)
398
graph.append((o.id, o.parents))
399
heads.extend([p for p in o.parents if p not in checked])
400
elif isinstance(o, Tag):
401
if o.object[1] not in checked:
402
heads.append(o.object[1])
404
trace.warning("Unable to import head object %r" % o)
147
407
# Order the revisions
148
408
# 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)
410
revision_ids = topo_sort(graph)
412
if limit is not None:
413
revision_ids = revision_ids[:limit]
415
for offset in range(0, len(revision_ids), batch_size):
416
target_git_object_retriever.start_write_group()
418
repo.start_write_group()
420
for i, head in enumerate(
421
revision_ids[offset:offset+batch_size]):
423
pb.update("fetching revisions", offset+i,
425
import_git_commit(repo, mapping, head, lookup_object,
426
target_git_object_retriever,
430
repo.abort_write_group()
433
hint = repo.commit_write_group()
435
pack_hints.extend(hint)
437
target_git_object_retriever.abort_write_group()
440
target_git_object_retriever.commit_write_group()
441
return pack_hints, last_imported
187
444
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
446
_matching_repo_format = GitRepositoryFormat()
192
449
def _get_repo_format_to_test():
196
453
"""See InterRepository.copy_content."""
197
454
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
456
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
457
mapping=None, fetch_spec=None):
458
self.fetch_refs(revision_id=revision_id, pb=pb,
459
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
462
class InterGitNonGitRepository(InterGitRepository):
463
"""Base InterRepository that copies revisions from a Git into a non-Git
466
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
467
mapping=None, fetch_spec=None):
201
468
if mapping is None:
202
469
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
470
if revision_id is not None:
471
interesting_heads = [revision_id]
472
elif fetch_spec is not None:
473
interesting_heads = fetch_spec.heads
475
interesting_heads = None
477
def determine_wants(refs):
479
if interesting_heads is None:
480
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
482
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
483
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)
484
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
485
if pack_hint is not None and self.target._format.pack_compresses:
486
self.target.pack(hint=pack_hint)
487
if interesting_heads is not None:
488
present_interesting_heads = self.target.has_revisions(interesting_heads)
489
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
490
if missing_interesting_heads:
491
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
495
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
496
def report_git_progress(pb, text):
497
text = text.rstrip("\r\n")
498
g = _GIT_PROGRESS_RE.match(text)
500
(text, pct, current, total) = g.groups()
501
pb.update(text, int(current), int(total))
503
pb.update(text, 0, 0)
506
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
507
"""InterRepository that copies revisions from a remote Git into a non-Git
510
def get_target_heads(self):
511
# FIXME: This should be more efficient
512
all_revs = self.target.all_revision_ids()
513
parent_map = self.target.get_parent_map(all_revs)
515
map(all_parents.update, parent_map.itervalues())
516
return set(all_revs) - all_parents
518
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
520
report_git_progress(pb, text)
521
store = BazaarObjectStore(self.target, mapping)
522
self.target.lock_write()
524
heads = self.get_target_heads()
525
graph_walker = store.get_graph_walker(
526
[store._lookup_revision_sha1(head) for head in heads])
529
def record_determine_wants(heads):
530
wants = determine_wants(heads)
531
recorded_wants.extend(wants)
536
create_pb = pb = ui.ui_factory.nested_progress_bar()
538
objects_iter = self.source.fetch_objects(
539
record_determine_wants, graph_walker,
540
store.get_raw, progress)
541
return import_git_objects(self.target, mapping,
542
objects_iter, store, recorded_wants, pb, limit)
550
def is_compatible(source, target):
551
"""Be compatible with GitRepository."""
552
return (isinstance(source, RemoteGitRepository) and
553
target.supports_rich_root() and
554
not isinstance(target, GitRepository) and
555
target.texts is not None)
558
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
559
"""InterRepository that copies revisions from a local Git into a non-Git
562
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
565
wants = determine_wants(self.source._git.get_refs())
214
568
create_pb = pb = ui.ui_factory.nested_progress_bar()
569
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
571
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()
573
return import_git_objects(self.target, mapping,
574
self.source._git.object_store,
575
target_git_object_retriever, wants, pb, limit)
226
577
self.target.unlock()
added
removed
fetch.py
