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
ChunkedContentFactory,
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, base_ie.revision)
107
if (base_sha == hexsha and base_ie.executable == ie.executable
108
and base_ie.kind == ie.kind):
109
# If nothing has changed since the base revision, we're done
111
if base_sha == hexsha and base_ie.kind == ie.kind:
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 = len(blob.data)
128
ie.text_sha1 = osutils.sha_string(blob.data)
129
# Check what revision we should store
131
for pinv in parent_invs:
132
if pinv.revision_id == base_inv.revision_id:
141
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
142
# found a revision in one of the parents to use
143
ie.revision = pie.revision
145
parent_keys.append((file_id, pie.revision))
146
if ie.revision is None:
147
# Need to store a new revision
148
ie.revision = revision_id
149
assert file_id is not None
150
assert ie.revision is not None
151
if ie.kind == 'symlink':
155
chunks = blob.chunked
156
except AttributeError: # older version of dulwich
158
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, chunks)])
159
shamap = { ie.file_id: hexsha }
161
if base_ie is not None:
162
old_path = base_inv.id2path(file_id)
163
if base_ie.kind == "directory":
164
invdelta.extend(remove_disappeared_children(old_path, base_ie.children, []))
167
invdelta.append((old_path, path, file_id, ie))
168
return (invdelta, shamap)
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, hexsha, base_inv, base_ie,
177
parent_id, revision_id, parent_invs, lookup_object):
178
file_id = mapping.generate_file_id(path)
179
ie = TreeReference(file_id, urlutils.basename(path.decode("utf-8")),
181
ie.revision = revision_id
186
if (base_ie.kind == ie.kind and
187
base_ie.reference_revision == ie.reference_revision):
188
ie.revision = base_ie.revision
189
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
190
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
191
invdelta = [(oldpath, path, file_id, ie)]
192
return invdelta, {}, {}
195
def remove_disappeared_children(path, base_children, existing_children):
197
deletable = [(osutils.pathjoin(path, k), v) for k,v in base_children.iteritems() if k not in existing_children]
199
(path, ie) = deletable.pop()
200
ret.append((path, None, ie.file_id, None))
201
if ie.kind == "directory":
202
for name, child_ie in ie.children.iteritems():
203
deletable.append((osutils.pathjoin(path, name), child_ie))
207
def import_git_tree(texts, mapping, path, hexsha, base_inv, base_inv_shamap,
208
base_ie, parent_id, revision_id, parent_invs, lookup_object,
209
allow_submodules=False):
82
210
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
212
:param texts: VersionedFiles object to add to
85
213
:param path: Path in the tree
86
214
:param tree: A git tree object
87
:param inv: Inventory object
215
:param base_inv: Base inventory against which to return inventory delta
216
:return: Inventory delta for this subtree
89
219
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
220
# We just have to hope this is indeed utf-8:
221
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
224
# Newly appeared here
225
ie.revision = revision_id
226
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
227
invdelta.append((None, path, file_id, ie))
229
# See if this has changed at all
231
base_sha = base_inv_shamap.lookup_tree(file_id)
232
except (KeyError, NotImplementedError):
235
if base_sha == hexsha:
236
# If nothing has changed since the base revision, we're done
238
if base_ie.kind != "directory":
239
ie.revision = revision_id
240
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
241
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
242
if base_ie is not None and base_ie.kind == "directory":
243
base_children = base_ie.children
246
# Remember for next time
247
existing_children = set()
250
tree = lookup_object(hexsha)
251
for mode, name, child_hexsha in tree.entries():
96
252
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):
253
existing_children.add(basename)
254
child_path = osutils.pathjoin(path, name)
255
if stat.S_ISDIR(mode):
256
subinvdelta, grandchildmodes, subshamap = import_git_tree(
257
texts, mapping, child_path, child_hexsha, base_inv,
258
base_inv_shamap, base_children.get(basename), file_id,
259
revision_id, parent_invs, lookup_object,
260
allow_submodules=allow_submodules)
261
elif S_ISGITLINK(mode): # submodule
262
if not allow_submodules:
263
raise SubmodulesRequireSubtrees()
264
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
265
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
266
file_id, revision_id, parent_invs, lookup_object)
268
subinvdelta, subshamap = import_git_blob(texts, mapping,
269
child_path, child_hexsha, base_inv, base_inv_shamap,
270
base_children.get(basename), file_id,
271
revision_id, parent_invs, lookup_object,
272
mode_is_executable(mode), stat.S_ISLNK(mode))
274
child_modes.update(grandchildmodes)
275
invdelta.extend(subinvdelta)
276
shamap.update(subshamap)
277
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
278
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
279
child_modes[child_path] = mode
280
# Remove any children that have disappeared
281
if base_ie is not None and base_ie.kind == "directory":
282
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
283
base_children, existing_children))
284
shamap[file_id] = hexsha
285
return invdelta, child_modes, shamap
288
def import_git_commit(repo, mapping, head, lookup_object,
289
target_git_object_retriever, parent_invs_cache):
290
o = lookup_object(head)
291
rev = mapping.import_commit(o)
292
# We have to do this here, since we have to walk the tree and
293
# we need to make sure to import the blobs / trees with the right
294
# path; this may involve adding them more than once.
295
parent_invs = parent_invs_cache.get_inventories(rev.parent_ids)
296
if parent_invs == []:
297
base_inv = Inventory(root_id=None)
299
base_inv_shamap = None # Should never be accessed
301
base_inv = parent_invs[0]
302
base_ie = base_inv.root
303
base_inv_shamap = target_git_object_retriever._idmap.get_inventory_sha_map(base_inv.revision_id)
304
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
305
mapping, "", o.tree, base_inv, base_inv_shamap, base_ie, None,
306
rev.revision_id, parent_invs, lookup_object,
307
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
309
for (oldpath, newpath, fileid, new_ie) in inv_delta:
311
entries.append((fileid, None, None, None))
313
if new_ie.kind in ("file", "symlink"):
314
entries.append((fileid, "blob", shamap[fileid], new_ie.revision))
315
elif new_ie.kind == "directory":
316
entries.append((fileid, "tree", shamap[fileid], rev.revision_id))
319
target_git_object_retriever._idmap.add_entries(rev.revision_id,
320
rev.parent_ids, head, o.tree, entries)
321
if unusual_modes != {}:
322
for path, mode in unusual_modes.iteritems():
323
warn_unusual_mode(rev.foreign_revid, path, mode)
324
mapping.import_unusual_file_modes(rev, unusual_modes)
326
basis_id = rev.parent_ids[0]
328
basis_id = NULL_REVISION
330
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
331
inv_delta, rev.revision_id, rev.parent_ids,
333
parent_invs_cache.add(rev.revision_id, inv)
334
repo.add_revision(rev.revision_id, rev)
335
if "verify" in debug.debug_flags:
336
new_unusual_modes = mapping.export_unusual_file_modes(rev)
337
if new_unusual_modes != unusual_modes:
338
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
339
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
340
for newsha1, newobj, path in objs:
341
assert path is not None
345
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
346
if oldsha1 != newsha1:
347
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
350
def import_git_objects(repo, mapping, object_iter,
351
target_git_object_retriever, heads, pb=None, limit=None):
110
352
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
354
:param repo: Target Bazaar repository
113
355
:param mapping: Mapping to use
114
356
:param object_iter: Iterator over Git objects.
357
:return: Tuple with pack hints and last imported revision id
359
def lookup_object(sha):
361
return object_iter[sha]
363
return target_git_object_retriever[sha]
116
364
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
367
heads = list(set(heads))
368
parent_invs_cache = LRUInventoryCache(repo)
369
target_git_object_retriever.start_write_group() # FIXME: try/finally
121
370
# Find and convert commit objects
122
for o in objects.iterkeys():
373
pb.update("finding revisions to fetch", len(graph), None)
375
assert isinstance(head, str)
377
o = lookup_object(head)
123
380
if isinstance(o, Commit):
124
381
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
382
if repo.has_revision(rev.revision_id):
384
squash_revision(repo, rev)
385
graph.append((o.id, o.parents))
386
heads.extend([p for p in o.parents if p not in checked])
387
elif isinstance(o, Tag):
388
if o.object[1] not in checked:
389
heads.append(o.object[1])
391
trace.warning("Unable to import head object %r" % o)
394
# Order the revisions
126
395
# 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)
397
revision_ids = topo_sort(graph)
399
if limit is not None:
400
revision_ids = revision_ids[:limit]
402
for offset in range(0, len(revision_ids), batch_size):
403
repo.start_write_group()
405
for i, head in enumerate(revision_ids[offset:offset+batch_size]):
407
pb.update("fetching revisions", offset+i, len(revision_ids))
408
import_git_commit(repo, mapping, head, lookup_object,
409
target_git_object_retriever,
413
repo.abort_write_group()
416
hint = repo.commit_write_group()
418
pack_hints.extend(hint)
419
target_git_object_retriever.commit_write_group()
420
return pack_hints, last_imported
159
423
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
425
_matching_repo_format = GitRepositoryFormat()
164
428
def _get_repo_format_to_test():
168
432
"""See InterRepository.copy_content."""
169
433
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
435
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
436
mapping=None, fetch_spec=None):
437
self.fetch_refs(revision_id=revision_id, pb=pb,
438
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
441
class InterGitNonGitRepository(InterGitRepository):
442
"""Base InterRepository that copies revisions from a Git into a non-Git
445
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
446
mapping=None, fetch_spec=None):
173
447
if mapping is None:
174
448
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)]
449
if revision_id is not None:
450
interesting_heads = [revision_id]
451
elif fetch_spec is not None:
452
interesting_heads = fetch_spec.heads
454
interesting_heads = None
456
def determine_wants(refs):
458
if interesting_heads is None:
459
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
461
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
462
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)
463
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
464
if pack_hint is not None and self.target._format.pack_compresses:
465
self.target.pack(hint=pack_hint)
466
if interesting_heads is not None:
467
present_interesting_heads = self.target.has_revisions(interesting_heads)
468
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
469
if missing_interesting_heads:
470
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
474
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
475
def report_git_progress(pb, text):
476
text = text.rstrip("\r\n")
477
g = _GIT_PROGRESS_RE.match(text)
479
(text, pct, current, total) = g.groups()
480
pb.update(text, int(current), int(total))
482
pb.update(text, 0, 0)
485
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
486
"""InterRepository that copies revisions from a remote Git into a non-Git
489
def get_target_heads(self):
490
# FIXME: This should be more efficient
491
all_revs = self.target.all_revision_ids()
492
parent_map = self.target.get_parent_map(all_revs)
494
map(all_parents.update, parent_map.itervalues())
495
return set(all_revs) - all_parents
497
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
499
report_git_progress(pb, text)
500
store = BazaarObjectStore(self.target, mapping)
187
501
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
503
heads = self.get_target_heads()
504
graph_walker = store.get_graph_walker(
505
[store._lookup_revision_sha1(head) for head in heads])
508
def record_determine_wants(heads):
509
wants = determine_wants(heads)
510
recorded_wants.extend(wants)
515
create_pb = pb = ui.ui_factory.nested_progress_bar()
517
objects_iter = self.source.fetch_objects(
518
record_determine_wants, graph_walker,
519
store.get_raw, progress)
520
return import_git_objects(self.target, mapping,
521
objects_iter, store, recorded_wants, pb, limit)
193
526
self.target.unlock()
196
529
def is_compatible(source, target):
197
530
"""Be compatible with GitRepository."""
198
531
# FIXME: Also check target uses VersionedFile
199
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
532
return (isinstance(source, RemoteGitRepository) and
533
target.supports_rich_root() and
534
not isinstance(target, GitRepository))
537
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
538
"""InterRepository that copies revisions from a local Git into a non-Git
541
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
544
wants = determine_wants(self.source._git.get_refs())
547
create_pb = pb = ui.ui_factory.nested_progress_bar()
548
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
550
self.target.lock_write()
552
return import_git_objects(self.target, mapping,
553
self.source._git.object_store,
554
target_git_object_retriever, wants, pb, limit)
562
def is_compatible(source, target):
563
"""Be compatible with GitRepository."""
564
# FIXME: Also check target uses VersionedFile
565
return (isinstance(source, LocalGitRepository) and
566
target.supports_rich_root() and
567
not isinstance(target, GitRepository))
570
class InterGitGitRepository(InterGitRepository):
571
"""InterRepository that copies between Git repositories."""
573
def fetch_objects(self, determine_wants, mapping, pb=None):
575
trace.note("git: %s", text)
576
graphwalker = self.target._git.get_graph_walker()
577
if (isinstance(self.source, LocalGitRepository) and
578
isinstance(self.target, LocalGitRepository)):
579
return self.source._git.fetch(self.target._git, determine_wants,
581
elif (isinstance(self.source, LocalGitRepository) and
582
isinstance(self.target, RemoteGitRepository)):
583
raise NotImplementedError
584
elif (isinstance(self.source, RemoteGitRepository) and
585
isinstance(self.target, LocalGitRepository)):
586
f, commit = self.target._git.object_store.add_thin_pack()
588
refs = self.source._git.fetch_pack(determine_wants,
589
graphwalker, f.write, progress)
598
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
599
mapping=None, fetch_spec=None, branches=None):
601
mapping = self.source.get_mapping()
603
if revision_id is not None:
604
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
605
elif fetch_spec is not None:
606
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
607
if branches is not None:
608
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
609
elif fetch_spec is None and revision_id is None:
610
determine_wants = r.object_store.determine_wants_all
612
determine_wants = lambda x: [y for y in args if not y in r.object_store]
613
return self.fetch_objects(determine_wants, mapping)[0]
617
def is_compatible(source, target):
618
"""Be compatible with GitRepository."""
619
return (isinstance(source, GitRepository) and
620
isinstance(target, GitRepository))
added
removed
fetch.py
