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 (
36
from bzrlib.errors import (
40
from bzrlib.inventory import (
47
from bzrlib.repository import (
50
from bzrlib.revision import (
53
from bzrlib.tsort import (
56
from bzrlib.versionedfile import (
57
FulltextContentFactory,
60
from bzrlib.plugins.git.mapping import (
62
inventory_to_tree_and_blobs,
67
from bzrlib.plugins.git.object_store import (
70
from bzrlib.plugins.git.remote import (
73
from bzrlib.plugins.git.repository import (
80
MAX_INV_CACHE_SIZE = 50 * 1024 * 1024
83
def import_git_blob(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
84
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
68
85
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
87
:param texts: VersionedFiles to add to
71
88
:param path: Path in the tree
72
89
:param blob: A git blob
90
:return: Inventory delta for this file
74
92
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):
97
# We just have to hope this is indeed utf-8:
98
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
99
ie.executable = executable
100
# See if this has changed at all
105
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
109
if (base_sha == hexsha and base_ie.executable == ie.executable
110
and base_ie.kind == ie.kind):
111
# If nothing has changed since the base revision, we're done
113
if base_sha == hexsha and base_ie.kind == ie.kind:
114
ie.text_size = base_ie.text_size
115
ie.text_sha1 = base_ie.text_sha1
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
129
ie.text_size = len(blob.data)
130
ie.text_sha1 = osutils.sha_string(blob.data)
131
# Check what revision we should store
133
for pinv in parent_invs:
134
if pinv.revision_id == base_inv.revision_id:
143
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
144
# found a revision in one of the parents to use
145
ie.revision = pie.revision
147
parent_keys.append((file_id, pie.revision))
148
if ie.revision is None:
149
# Need to store a new revision
150
ie.revision = revision_id
151
assert file_id is not None
152
assert ie.revision is not None
153
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
154
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
158
if base_ie is not None:
159
old_path = base_inv.id2path(file_id)
160
if base_ie.kind == "directory":
161
invdelta.extend(remove_disappeared_children(old_path, base_ie.children, []))
164
invdelta.append((old_path, path, file_id, ie))
165
return (invdelta, shamap)
168
class SubmodulesRequireSubtrees(BzrError):
169
_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'."""
173
def import_git_submodule(texts, mapping, path, hexsha, base_inv, base_ie,
174
parent_id, revision_id, parent_invs, shagitmap, lookup_object):
175
file_id = mapping.generate_file_id(path)
176
ie = TreeReference(file_id, urlutils.basename(path.decode("utf-8")),
178
ie.revision = revision_id
183
if base_ie.kind == ie.kind and base_ie.reference_revision == ie.reference_revision:
184
ie.revision = base_ie.revision
185
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
186
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
187
invdelta = [(oldpath, path, file_id, ie)]
188
return invdelta, {}, {}
191
def remove_disappeared_children(path, base_children, existing_children):
193
deletable = [(osutils.pathjoin(path, k), v) for k,v in base_children.iteritems() if k not in existing_children]
195
(path, ie) = deletable.pop()
196
ret.append((path, None, ie.file_id, None))
197
if ie.kind == "directory":
198
for name, child_ie in ie.children.iteritems():
199
deletable.append((osutils.pathjoin(path, name), child_ie))
203
def import_git_tree(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
204
revision_id, parent_invs, shagitmap, lookup_object, allow_submodules=False):
82
205
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
207
:param texts: VersionedFiles object to add to
85
208
:param path: Path in the tree
86
209
:param tree: A git tree object
87
:param inv: Inventory object
210
:param base_inv: Base inventory against which to return inventory delta
211
:return: Inventory delta for this subtree
89
214
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
215
# We just have to hope this is indeed utf-8:
216
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
219
# Newly appeared here
220
ie.revision = revision_id
221
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
222
invdelta.append((None, path, file_id, ie))
224
# See if this has changed at all
226
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
230
if base_sha == hexsha:
231
# If nothing has changed since the base revision, we're done
233
if base_ie.kind != "directory":
234
ie.revision = revision_id
235
texts.insert_record_stream([FulltextContentFactory((ie.file_id, ie.revision), (), None, "")])
236
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
237
if base_ie is not None and base_ie.kind == "directory":
238
base_children = base_ie.children
241
# Remember for next time
242
existing_children = set()
245
tree = lookup_object(hexsha)
246
for mode, name, child_hexsha in tree.entries():
96
247
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):
248
existing_children.add(basename)
249
child_path = osutils.pathjoin(path, name)
250
if stat.S_ISDIR(mode):
251
subinvdelta, grandchildmodes, subshamap = import_git_tree(
252
texts, mapping, child_path, child_hexsha, base_inv,
253
base_children.get(basename), file_id, revision_id,
254
parent_invs, shagitmap, lookup_object,
255
allow_submodules=allow_submodules)
256
invdelta.extend(subinvdelta)
257
child_modes.update(grandchildmodes)
258
shamap.extend(subshamap)
259
elif S_ISGITLINK(mode): # submodule
260
if not allow_submodules:
261
raise SubmodulesRequireSubtrees()
262
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
263
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
264
file_id, revision_id, parent_invs, shagitmap, lookup_object)
265
invdelta.extend(subinvdelta)
266
child_modes.update(grandchildmodes)
267
shamap.extend(subshamap)
269
subinvdelta, subshamap = import_git_blob(texts, mapping,
270
child_path, child_hexsha, base_inv, base_children.get(basename), file_id,
271
revision_id, parent_invs, shagitmap, lookup_object,
272
mode_is_executable(mode), stat.S_ISLNK(mode))
273
invdelta.extend(subinvdelta)
274
shamap.extend(subshamap)
275
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
276
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
277
child_modes[child_path] = mode
278
# Remove any children that have disappeared
279
if base_ie is not None and base_ie.kind == "directory":
280
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
281
base_children, existing_children))
282
shamap.append((hexsha, "tree", (file_id, revision_id)))
283
return invdelta, child_modes, shamap
286
def approx_inv_size(inv):
287
# Very rough estimate, 1k per inventory entry
288
return len(inv) * 1024
291
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
110
293
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
295
:param repo: Target Bazaar repository
113
296
:param mapping: Mapping to use
114
297
:param object_iter: Iterator over Git objects.
299
def lookup_object(sha):
301
return object_iter[sha]
303
return target_git_object_retriever[sha]
116
304
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
308
parent_invs_cache = lru_cache.LRUSizeCache(compute_size=approx_inv_size,
309
max_size=MAX_INV_CACHE_SIZE)
121
310
# Find and convert commit objects
122
for o in objects.iterkeys():
313
pb.update("finding revisions to fetch", len(graph), None)
315
assert isinstance(head, str)
317
o = lookup_object(head)
319
trace.mutter('missing head %s', head)
123
321
if isinstance(o, Commit):
124
322
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
323
if repo.has_revision(rev.revision_id):
325
squash_revision(repo, rev)
326
graph.append((o.id, o.parents))
327
target_git_object_retriever._idmap.add_entry(o.id, "commit",
328
(rev.revision_id, o.tree))
329
heads.extend([p for p in o.parents if p not in checked])
330
elif isinstance(o, Tag):
331
heads.append(o.object[1])
333
trace.warning("Unable to import head object %r" % o)
336
# Order the revisions
126
337
# 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
338
for i, head in enumerate(topo_sort(graph)):
340
pb.update("fetching revisions", i, len(graph))
341
o = lookup_object(head)
342
rev = mapping.import_commit(o)
343
# We have to do this here, since we have to walk the tree and
344
# we need to make sure to import the blobs / trees with the right
130
345
# 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)
347
for parent_id in rev.parent_ids:
349
parent_invs.append(parent_invs_cache[parent_id])
351
parent_inv = repo.get_inventory(parent_id)
352
parent_invs.append(parent_inv)
353
parent_invs_cache[parent_id] = parent_inv
354
if parent_invs == []:
355
base_inv = Inventory(root_id=None)
358
base_inv = parent_invs[0]
359
base_ie = base_inv.root
360
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
361
mapping, "", o.tree, base_inv, base_ie, None, rev.revision_id,
362
parent_invs, target_git_object_retriever._idmap, lookup_object,
363
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
364
target_git_object_retriever._idmap.add_entries(shamap)
365
if unusual_modes != {}:
366
for path, mode in unusual_modes.iteritems():
367
warn_unusual_mode(rev.foreign_revid, path, mode)
368
mapping.import_unusual_file_modes(rev, unusual_modes)
370
basis_id = rev.parent_ids[0]
372
basis_id = NULL_REVISION
374
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
375
inv_delta, rev.revision_id, rev.parent_ids,
377
parent_invs_cache[rev.revision_id] = inv
378
repo.add_revision(rev.revision_id, rev)
379
if "verify" in debug.debug_flags:
380
new_unusual_modes = mapping.export_unusual_file_modes(rev)
381
if new_unusual_modes != unusual_modes:
382
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
383
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
384
for sha1, newobj, path in objs:
385
assert path is not None
386
oldobj = tree_lookup_path(lookup_object, o.tree, path)
388
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
390
target_git_object_retriever._idmap.commit()
159
393
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
395
_matching_repo_format = GitRepositoryFormat()
164
398
def _get_repo_format_to_test():
168
402
"""See InterRepository.copy_content."""
169
403
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
405
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
407
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
408
mapping=mapping, fetch_spec=fetch_spec)
411
class InterGitNonGitRepository(InterGitRepository):
412
"""Base InterRepository that copies revisions from a Git into a non-Git
415
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
416
mapping=None, fetch_spec=None):
173
417
if mapping is None:
174
418
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)]
419
if revision_id is not None:
420
interesting_heads = [revision_id]
421
elif fetch_spec is not None:
422
interesting_heads = fetch_spec.heads
424
interesting_heads = None
426
def determine_wants(refs):
428
if interesting_heads is None:
429
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
431
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
432
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)
433
pack_hint = self.fetch_objects(determine_wants, mapping, pb)
434
if pack_hint is not None and self.target._format.pack_compresses:
435
self.target.pack(hint=pack_hint)
436
if interesting_heads is not None:
437
present_interesting_heads = self.target.has_revisions(interesting_heads)
438
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
439
if missing_interesting_heads:
440
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
444
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
445
def report_git_progress(pb, text):
446
text = text.rstrip("\r\n")
447
g = _GIT_PROGRESS_RE.match(text)
449
(text, pct, current, total) = g.groups()
450
pb.update(text, int(current), int(total))
452
pb.update(text, 0, 0)
455
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
456
"""InterRepository that copies revisions from a remote Git into a non-Git
459
def get_target_heads(self):
460
# FIXME: This should be more efficient
461
all_revs = self.target.all_revision_ids()
462
parent_map = self.target.get_parent_map(all_revs)
464
map(all_parents.update, parent_map.itervalues())
465
return set(all_revs) - all_parents
467
def fetch_objects(self, determine_wants, mapping, pb=None):
469
report_git_progress(pb, text)
470
store = BazaarObjectStore(self.target, mapping)
187
471
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
473
heads = self.get_target_heads()
474
graph_walker = store.get_graph_walker(
475
[store._lookup_revision_sha1(head) for head in heads])
478
def record_determine_wants(heads):
479
wants = determine_wants(heads)
480
recorded_wants.extend(wants)
485
create_pb = pb = ui.ui_factory.nested_progress_bar()
487
self.target.start_write_group()
489
objects_iter = self.source.fetch_objects(
490
record_determine_wants, graph_walker,
491
store.get_raw, progress)
492
import_git_objects(self.target, mapping, objects_iter,
493
store, recorded_wants, pb)
495
pack_hint = self.target.commit_write_group()
193
501
self.target.unlock()
196
504
def is_compatible(source, target):
197
505
"""Be compatible with GitRepository."""
198
506
# FIXME: Also check target uses VersionedFile
199
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
507
return (isinstance(source, RemoteGitRepository) and
508
target.supports_rich_root() and
509
not isinstance(target, GitRepository))
512
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
513
"""InterRepository that copies revisions from a local Git into a non-Git
516
def fetch_objects(self, determine_wants, mapping, pb=None):
517
wants = determine_wants(self.source._git.get_refs())
520
create_pb = pb = ui.ui_factory.nested_progress_bar()
521
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
523
self.target.lock_write()
525
self.target.start_write_group()
527
import_git_objects(self.target, mapping,
528
self.source._git.object_store,
529
target_git_object_retriever, wants, pb)
531
pack_hint = self.target.commit_write_group()
540
def is_compatible(source, target):
541
"""Be compatible with GitRepository."""
542
# FIXME: Also check target uses VersionedFile
543
return (isinstance(source, LocalGitRepository) and
544
target.supports_rich_root() and
545
not isinstance(target, GitRepository))
548
class InterGitGitRepository(InterGitRepository):
549
"""InterRepository that copies between Git repositories."""
551
def fetch_objects(self, determine_wants, mapping, pb=None):
553
trace.note("git: %s", text)
554
graphwalker = self.target._git.get_graph_walker()
555
if isinstance(self.source, LocalGitRepository) and isinstance(self.target, LocalGitRepository):
556
return self.source._git.fetch(self.target._git, determine_wants,
558
elif isinstance(self.source, LocalGitRepository) and isinstance(self.target, RemoteGitRepository):
559
raise NotImplementedError
560
elif isinstance(self.source, RemoteGitRepository) and isinstance(self.target, LocalGitRepository):
561
f, commit = self.target._git.object_store.add_thin_pack()
563
refs = self.source._git.fetch_pack(determine_wants, graphwalker,
573
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
574
mapping=None, fetch_spec=None, branches=None):
576
mapping = self.source.get_mapping()
578
if revision_id is not None:
579
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
580
elif fetch_spec is not None:
581
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
582
if branches is not None:
583
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
584
elif fetch_spec is None and revision_id is None:
585
determine_wants = r.object_store.determine_wants_all
587
determine_wants = lambda x: [y for y in args if not y in r.object_store]
588
return self.fetch_objects(determine_wants, mapping)
592
def is_compatible(source, target):
593
"""Be compatible with GitRepository."""
594
return (isinstance(source, GitRepository) and
595
isinstance(target, GitRepository))
added
removed
fetch.py
