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)):
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
return (invdelta, [(ie.file_id, "blob", hexsha, ie.revision)])
157
class SubmodulesRequireSubtrees(BzrError):
158
_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'."""
162
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
163
base_inv, parent_id, revision_id, parent_invs, lookup_object,
165
if base_hexsha == hexsha and base_mode == mode:
167
file_id = mapping.generate_file_id(path)
168
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
169
ie.revision = revision_id
170
if base_hexsha is None:
174
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
175
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
176
invdelta = [(oldpath, path, file_id, ie)]
177
return invdelta, {}, []
180
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
183
for name, mode, hexsha in base_tree.iteritems():
184
if name in existing_children:
186
c_path = posixpath.join(path, name.decode("utf-8"))
187
ret.append((c_path, None, base_inv.path2id(c_path), None))
188
if stat.S_ISDIR(mode):
189
ret.extend(remove_disappeared_children(
190
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
194
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
195
base_inv, base_inv_shamap, parent_id, revision_id, parent_invs,
196
lookup_object, (base_mode, mode), allow_submodules=False):
92
197
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
199
:param texts: VersionedFiles object to add to
95
200
:param path: Path in the tree
96
201
:param tree: A git tree object
97
:param inv: Inventory object
202
:param base_inv: Base inventory against which to return inventory delta
203
:return: Inventory delta for this subtree
205
if base_hexsha == hexsha and base_mode == mode:
206
# If nothing has changed since the base revision, we're done
99
209
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):
210
# We just have to hope this is indeed utf-8:
211
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
212
tree = lookup_object(hexsha)
213
if base_hexsha is None:
215
old_path = None # Newly appeared here
217
base_tree = lookup_object(base_hexsha)
218
old_path = path # Renames aren't supported yet
219
if base_tree is None or type(base_tree) is not Tree:
220
ie.revision = revision_id
221
invdelta.append((old_path, path, ie.file_id, ie))
222
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
223
# Remember for next time
224
existing_children = set()
226
shamap = [(ie.file_id, "tree", hexsha, revision_id)]
227
for child_mode, name, child_hexsha in tree.entries():
228
existing_children.add(name)
229
child_path = posixpath.join(path, name)
230
if type(base_tree) is Tree:
232
child_base_mode, child_base_hexsha = base_tree[name]
234
child_base_hexsha = None
237
child_base_hexsha = None
239
if stat.S_ISDIR(child_mode):
240
subinvdelta, grandchildmodes, subshamap = import_git_tree(
241
texts, mapping, child_path, name,
242
(child_base_hexsha, child_hexsha),
243
base_inv, base_inv_shamap,
244
file_id, revision_id, parent_invs, lookup_object,
245
(child_base_mode, child_mode),
246
allow_submodules=allow_submodules)
247
elif S_ISGITLINK(child_mode): # submodule
248
if not allow_submodules:
249
raise SubmodulesRequireSubtrees()
250
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
251
texts, mapping, child_path, name,
252
(child_base_hexsha, child_hexsha),
253
base_inv, file_id, revision_id, parent_invs, lookup_object,
254
(child_base_mode, child_mode))
256
subinvdelta, subshamap = import_git_blob(texts, mapping,
257
child_path, name, (child_base_hexsha, child_hexsha),
258
base_inv, base_inv_shamap,
260
revision_id, parent_invs, lookup_object,
261
(child_base_mode, child_mode))
263
child_modes.update(grandchildmodes)
264
invdelta.extend(subinvdelta)
265
shamap.extend(subshamap)
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
return invdelta, child_modes, shamap
276
def import_git_commit(repo, mapping, head, lookup_object,
277
target_git_object_retriever, parent_invs_cache):
278
o = lookup_object(head)
279
rev = mapping.import_commit(o)
280
# We have to do this here, since we have to walk the tree and
281
# we need to make sure to import the blobs / trees with the right
282
# path; this may involve adding them more than once.
283
parent_invs = parent_invs_cache.get_inventories(rev.parent_ids)
284
if parent_invs == []:
285
base_inv = Inventory(root_id=None)
286
base_inv_shamap = None # Should never be accessed
290
base_inv = parent_invs[0]
291
base_inv_shamap = target_git_object_retriever._idmap.get_inventory_sha_map(base_inv.revision_id)
292
base_tree = lookup_object(o.parents[0]).tree
293
base_mode = stat.S_IFDIR
294
inv_delta, unusual_modes, entries = import_git_tree(repo.texts,
295
mapping, "", u"", (base_tree, o.tree), base_inv, base_inv_shamap,
296
None, rev.revision_id, parent_invs, lookup_object,
297
(base_mode, stat.S_IFDIR),
298
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
299
target_git_object_retriever._idmap.add_entries(rev.revision_id,
300
rev.parent_ids, head, o.tree, entries)
301
if unusual_modes != {}:
302
for path, mode in unusual_modes.iteritems():
303
warn_unusual_mode(rev.foreign_revid, path, mode)
304
mapping.import_unusual_file_modes(rev, unusual_modes)
306
basis_id = rev.parent_ids[0]
308
basis_id = NULL_REVISION
310
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
311
inv_delta, rev.revision_id, rev.parent_ids,
313
parent_invs_cache.add(rev.revision_id, inv)
314
repo.add_revision(rev.revision_id, rev)
315
if "verify" in debug.debug_flags:
316
new_unusual_modes = mapping.export_unusual_file_modes(rev)
317
if new_unusual_modes != unusual_modes:
318
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
319
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
320
for newsha1, newobj, path in objs:
321
assert path is not None
325
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
326
if oldsha1 != newsha1:
327
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
330
def import_git_objects(repo, mapping, object_iter,
331
target_git_object_retriever, heads, pb=None, limit=None):
125
332
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
334
:param repo: Target Bazaar repository
128
335
:param mapping: Mapping to use
129
336
:param object_iter: Iterator over Git objects.
337
: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)
339
def lookup_object(sha):
341
return object_iter[sha]
343
return target_git_object_retriever[sha]
346
heads = list(set(heads))
347
parent_invs_cache = LRUInventoryCache(repo)
140
348
# Find and convert commit objects
141
for o in objects.itervalues():
351
pb.update("finding revisions to fetch", len(graph), None)
353
assert isinstance(head, str)
355
o = lookup_object(head)
142
358
if isinstance(o, Commit):
143
359
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))
360
if repo.has_revision(rev.revision_id):
362
squash_revision(repo, rev)
363
graph.append((o.id, o.parents))
364
heads.extend([p for p in o.parents if p not in checked])
365
elif isinstance(o, Tag):
366
if o.object[1] not in checked:
367
heads.append(o.object[1])
369
trace.warning("Unable to import head object %r" % o)
147
372
# Order the revisions
148
373
# 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)
375
revision_ids = topo_sort(graph)
377
if limit is not None:
378
revision_ids = revision_ids[:limit]
380
for offset in range(0, len(revision_ids), batch_size):
381
target_git_object_retriever.start_write_group() # FIXME: try/finally
383
repo.start_write_group()
385
for i, head in enumerate(
386
revision_ids[offset:offset+batch_size]):
388
pb.update("fetching revisions", offset+i,
390
import_git_commit(repo, mapping, head, lookup_object,
391
target_git_object_retriever,
395
repo.abort_write_group()
398
hint = repo.commit_write_group()
400
pack_hints.extend(hint)
402
target_git_object_retriever.abort_write_group()
405
target_git_object_retriever.commit_write_group()
406
return pack_hints, last_imported
187
409
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
411
_matching_repo_format = GitRepositoryFormat()
192
414
def _get_repo_format_to_test():
196
418
"""See InterRepository.copy_content."""
197
419
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
421
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
422
mapping=None, fetch_spec=None):
423
self.fetch_refs(revision_id=revision_id, pb=pb,
424
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
427
class InterGitNonGitRepository(InterGitRepository):
428
"""Base InterRepository that copies revisions from a Git into a non-Git
431
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
432
mapping=None, fetch_spec=None):
201
433
if mapping is None:
202
434
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
435
if revision_id is not None:
436
interesting_heads = [revision_id]
437
elif fetch_spec is not None:
438
interesting_heads = fetch_spec.heads
440
interesting_heads = None
442
def determine_wants(refs):
444
if interesting_heads is None:
445
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
447
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
448
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)
449
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
450
if pack_hint is not None and self.target._format.pack_compresses:
451
self.target.pack(hint=pack_hint)
452
if interesting_heads is not None:
453
present_interesting_heads = self.target.has_revisions(interesting_heads)
454
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
455
if missing_interesting_heads:
456
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
460
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
461
def report_git_progress(pb, text):
462
text = text.rstrip("\r\n")
463
g = _GIT_PROGRESS_RE.match(text)
465
(text, pct, current, total) = g.groups()
466
pb.update(text, int(current), int(total))
468
pb.update(text, 0, 0)
471
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
472
"""InterRepository that copies revisions from a remote Git into a non-Git
475
def get_target_heads(self):
476
# FIXME: This should be more efficient
477
all_revs = self.target.all_revision_ids()
478
parent_map = self.target.get_parent_map(all_revs)
480
map(all_parents.update, parent_map.itervalues())
481
return set(all_revs) - all_parents
483
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
485
report_git_progress(pb, text)
486
store = BazaarObjectStore(self.target, mapping)
487
self.target.lock_write()
489
heads = self.get_target_heads()
490
graph_walker = store.get_graph_walker(
491
[store._lookup_revision_sha1(head) for head in heads])
494
def record_determine_wants(heads):
495
wants = determine_wants(heads)
496
recorded_wants.extend(wants)
501
create_pb = pb = ui.ui_factory.nested_progress_bar()
503
objects_iter = self.source.fetch_objects(
504
record_determine_wants, graph_walker,
505
store.get_raw, progress)
506
return import_git_objects(self.target, mapping,
507
objects_iter, store, recorded_wants, pb, limit)
515
def is_compatible(source, target):
516
"""Be compatible with GitRepository."""
517
# FIXME: Also check target uses VersionedFile
518
return (isinstance(source, RemoteGitRepository) and
519
target.supports_rich_root() and
520
not isinstance(target, GitRepository))
523
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
524
"""InterRepository that copies revisions from a local Git into a non-Git
527
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
530
wants = determine_wants(self.source._git.get_refs())
214
533
create_pb = pb = ui.ui_factory.nested_progress_bar()
534
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
536
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()
538
return import_git_objects(self.target, mapping,
539
self.source._git.object_store,
540
target_git_object_retriever, wants, pb, limit)
226
542
self.target.unlock()
added
removed
fetch.py
