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.lru_cache import (
49
from bzrlib.repository import (
52
from bzrlib.revision import (
55
from bzrlib.tsort import (
58
from bzrlib.versionedfile import (
59
FulltextContentFactory,
62
from bzrlib.plugins.git.mapping import (
64
inventory_to_tree_and_blobs,
69
from bzrlib.plugins.git.object_store import (
72
from bzrlib.plugins.git.remote import (
75
from bzrlib.plugins.git.repository import (
82
def import_git_blob(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
83
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
68
84
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
86
:param texts: VersionedFiles to add to
71
87
:param path: Path in the tree
72
88
:param blob: A git blob
89
:return: Inventory delta for this file
74
91
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):
96
# We just have to hope this is indeed utf-8:
97
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
98
ie.executable = executable
99
# See if this has changed at all
104
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
108
if (base_sha == hexsha and base_ie.executable == ie.executable
109
and base_ie.kind == ie.kind):
110
# If nothing has changed since the base revision, we're done
112
if base_sha == hexsha and base_ie.kind == ie.kind:
113
ie.text_size = base_ie.text_size
114
ie.text_sha1 = base_ie.text_sha1
115
ie.symlink_target = base_ie.symlink_target
116
if ie.executable == base_ie.executable:
117
ie.revision = base_ie.revision
119
blob = lookup_object(hexsha)
121
blob = lookup_object(hexsha)
122
if ie.kind == "symlink":
124
ie.symlink_target = blob.data
128
ie.text_size = len(blob.data)
129
ie.text_sha1 = osutils.sha_string(blob.data)
130
# Check what revision we should store
132
for pinv in parent_invs:
133
if pinv.revision_id == base_inv.revision_id:
142
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
143
# found a revision in one of the parents to use
144
ie.revision = pie.revision
146
parent_keys.append((file_id, pie.revision))
147
if ie.revision is None:
148
# Need to store a new revision
149
ie.revision = revision_id
150
assert file_id is not None
151
assert ie.revision is not None
152
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
153
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
157
if base_ie is not None:
158
old_path = base_inv.id2path(file_id)
159
if base_ie.kind == "directory":
160
invdelta.extend(remove_disappeared_children(old_path, base_ie.children, []))
163
invdelta.append((old_path, path, file_id, ie))
164
return (invdelta, shamap)
167
class SubmodulesRequireSubtrees(BzrError):
168
_fmt = """The repository you are fetching from contains submodules. Please run 'bzr upgrade --development-subtree'."""
172
def import_git_submodule(texts, mapping, path, hexsha, base_inv, base_ie,
173
parent_id, revision_id, parent_invs, shagitmap, lookup_object):
174
file_id = mapping.generate_file_id(path)
175
ie = TreeReference(file_id, urlutils.basename(path.decode("utf-8")),
177
ie.revision = revision_id
182
if base_ie.kind == ie.kind and base_ie.reference_revision == ie.reference_revision:
183
ie.revision = base_ie.revision
184
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
185
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
186
invdelta = [(oldpath, path, file_id, ie)]
187
return invdelta, {}, {}
190
def remove_disappeared_children(path, base_children, existing_children):
192
deletable = [(osutils.pathjoin(path, k), v) for k,v in base_children.iteritems() if k not in existing_children]
194
(path, ie) = deletable.pop()
195
ret.append((path, None, ie.file_id, None))
196
if ie.kind == "directory":
197
for name, child_ie in ie.children.iteritems():
198
deletable.append((osutils.pathjoin(path, name), child_ie))
202
def import_git_tree(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
203
revision_id, parent_invs, shagitmap, lookup_object):
82
204
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
206
:param texts: VersionedFiles object to add to
85
207
:param path: Path in the tree
86
208
:param tree: A git tree object
87
:param inv: Inventory object
209
:param base_inv: Base inventory against which to return inventory delta
210
:return: Inventory delta for this subtree
89
213
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
214
# We just have to hope this is indeed utf-8:
215
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
218
# Newly appeared here
219
ie.revision = revision_id
220
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
221
invdelta.append((None, path, file_id, ie))
223
# See if this has changed at all
225
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
229
if base_sha == hexsha:
230
# If nothing has changed since the base revision, we're done
232
if base_ie.kind != "directory":
233
ie.revision = revision_id
234
texts.insert_record_stream([FulltextContentFactory((ie.file_id, ie.revision), (), None, "")])
235
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
236
if base_ie is not None and base_ie.kind == "directory":
237
base_children = base_ie.children
240
# Remember for next time
241
existing_children = set()
244
tree = lookup_object(hexsha)
245
for mode, name, child_hexsha in tree.entries():
96
246
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):
247
existing_children.add(basename)
248
child_path = osutils.pathjoin(path, name)
249
if stat.S_ISDIR(mode):
250
subinvdelta, grandchildmodes, subshamap = import_git_tree(
251
texts, mapping, child_path, child_hexsha, base_inv,
252
base_children.get(basename), file_id, revision_id, parent_invs, shagitmap,
254
invdelta.extend(subinvdelta)
255
child_modes.update(grandchildmodes)
256
shamap.extend(subshamap)
257
elif S_ISGITLINK(mode): # submodule
258
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
259
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
260
file_id, revision_id, parent_invs, shagitmap, lookup_object)
261
invdelta.extend(subinvdelta)
262
child_modes.update(grandchildmodes)
263
shamap.extend(subshamap)
265
subinvdelta, subshamap = import_git_blob(texts, mapping,
266
child_path, child_hexsha, base_inv, base_children.get(basename), file_id,
267
revision_id, parent_invs, shagitmap, lookup_object,
268
mode_is_executable(mode), stat.S_ISLNK(mode))
269
invdelta.extend(subinvdelta)
270
shamap.extend(subshamap)
271
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
272
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
273
child_modes[child_path] = mode
274
# Remove any children that have disappeared
275
if base_ie is not None and base_ie.kind == "directory":
276
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
277
base_children, existing_children))
278
shamap.append((hexsha, "tree", (file_id, revision_id)))
279
return invdelta, child_modes, shamap
282
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
110
284
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
286
:param repo: Target Bazaar repository
113
287
:param mapping: Mapping to use
114
288
:param object_iter: Iterator over Git objects.
290
def lookup_object(sha):
292
return object_iter[sha]
294
return target_git_object_retriever[sha]
116
295
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
301
parent_invs_cache = LRUCache(50)
121
302
# Find and convert commit objects
122
for o in objects.iterkeys():
305
pb.update("finding revisions to fetch", len(graph), None)
307
assert isinstance(head, str)
309
o = lookup_object(head)
311
trace.mutter('missing head %s', head)
123
313
if isinstance(o, Commit):
124
314
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
315
if repo.has_revision(rev.revision_id):
317
squash_revision(repo, rev)
318
root_trees[rev.revision_id] = o.tree
319
revisions[rev.revision_id] = rev
320
graph.append((rev.revision_id, rev.parent_ids))
321
target_git_object_retriever._idmap.add_entry(o.id, "commit",
322
(rev.revision_id, o.tree))
323
heads.extend([p for p in o.parents if p not in checked])
324
elif isinstance(o, Tag):
325
heads.append(o.object[1])
327
trace.warning("Unable to import head object %r" % o)
329
# Order the revisions
126
330
# 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
331
for i, revid in enumerate(topo_sort(graph)):
333
pb.update("fetching revisions", i, len(graph))
334
rev = revisions[revid]
335
# We have to do this here, since we have to walk the tree and
336
# we need to make sure to import the blobs / trees with the right
130
337
# 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)
339
for parent_id in rev.parent_ids:
341
parent_invs.append(parent_invs_cache[parent_id])
343
parent_inv = repo.get_inventory(parent_id)
344
parent_invs.append(parent_inv)
345
parent_invs_cache[parent_id] = parent_inv
346
if parent_invs == []:
347
base_inv = Inventory(root_id=None)
350
base_inv = parent_invs[0]
351
base_ie = base_inv.root
352
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
353
mapping, "", root_trees[revid], base_inv, base_ie, None, revid,
354
parent_invs, target_git_object_retriever._idmap, lookup_object)
355
target_git_object_retriever._idmap.add_entries(shamap)
356
if unusual_modes != {}:
357
for path, mode in unusual_modes.iteritems():
358
warn_unusual_mode(rev.foreign_revid, path, mode)
359
mapping.import_unusual_file_modes(rev, unusual_modes)
361
basis_id = rev.parent_ids[0]
363
basis_id = NULL_REVISION
365
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
366
inv_delta, rev.revision_id, rev.parent_ids,
368
parent_invs_cache[rev.revision_id] = inv
369
repo.add_revision(rev.revision_id, rev)
370
if "verify" in debug.debug_flags:
371
new_unusual_modes = mapping.export_unusual_file_modes(rev)
372
if new_unusual_modes != unusual_modes:
373
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
374
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
375
for sha1, newobj, path in objs:
376
assert path is not None
377
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
379
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
381
target_git_object_retriever._idmap.commit()
159
384
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
386
_matching_repo_format = GitRepositoryFormat()
164
389
def _get_repo_format_to_test():
168
393
"""See InterRepository.copy_content."""
169
394
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
396
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
398
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
399
mapping=mapping, fetch_spec=fetch_spec)
402
class InterGitNonGitRepository(InterGitRepository):
403
"""Base InterRepository that copies revisions from a Git into a non-Git
406
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
407
mapping=None, fetch_spec=None):
173
408
if mapping is None:
174
409
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)]
410
if revision_id is not None:
411
interesting_heads = [revision_id]
412
elif fetch_spec is not None:
413
interesting_heads = fetch_spec.heads
415
interesting_heads = None
417
def determine_wants(refs):
419
if interesting_heads is None:
420
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
422
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
423
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)
424
pack_hint = self.fetch_objects(determine_wants, mapping, pb)
425
if pack_hint is not None and self.target._format.pack_compresses:
426
self.target.pack(hint=pack_hint)
427
if interesting_heads is not None:
428
present_interesting_heads = self.target.has_revisions(interesting_heads)
429
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
430
if missing_interesting_heads:
431
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
435
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
436
def report_git_progress(pb, text):
437
text = text.rstrip("\r\n")
438
g = _GIT_PROGRESS_RE.match(text)
440
(text, pct, current, total) = g.groups()
441
pb.update(text, int(current), int(total))
443
pb.update(text, 0, 0)
446
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
447
"""InterRepository that copies revisions from a remote Git into a non-Git
450
def get_target_heads(self):
451
# FIXME: This should be more efficient
452
all_revs = self.target.all_revision_ids()
453
parent_map = self.target.get_parent_map(all_revs)
455
map(all_parents.update, parent_map.itervalues())
456
return set(all_revs) - all_parents
458
def fetch_objects(self, determine_wants, mapping, pb=None):
460
report_git_progress(pb, text)
461
store = BazaarObjectStore(self.target, mapping)
187
462
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
464
heads = self.get_target_heads()
465
graph_walker = store.get_graph_walker(
466
[store._lookup_revision_sha1(head) for head in heads])
469
def record_determine_wants(heads):
470
wants = determine_wants(heads)
471
recorded_wants.extend(wants)
476
create_pb = pb = ui.ui_factory.nested_progress_bar()
478
self.target.start_write_group()
480
objects_iter = self.source.fetch_objects(
481
record_determine_wants, graph_walker,
482
store.get_raw, progress)
483
import_git_objects(self.target, mapping, objects_iter,
484
store, recorded_wants, pb)
486
pack_hint = self.target.commit_write_group()
193
492
self.target.unlock()
196
495
def is_compatible(source, target):
197
496
"""Be compatible with GitRepository."""
198
497
# FIXME: Also check target uses VersionedFile
199
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
498
return (isinstance(source, RemoteGitRepository) and
499
target.supports_rich_root() and
500
not isinstance(target, GitRepository))
503
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
504
"""InterRepository that copies revisions from a local Git into a non-Git
507
def fetch_objects(self, determine_wants, mapping, pb=None):
508
wants = determine_wants(self.source._git.get_refs())
511
create_pb = pb = ui.ui_factory.nested_progress_bar()
512
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
514
self.target.lock_write()
516
self.target.start_write_group()
518
import_git_objects(self.target, mapping,
519
self.source._git.object_store,
520
target_git_object_retriever, wants, pb)
522
pack_hint = self.target.commit_write_group()
531
def is_compatible(source, target):
532
"""Be compatible with GitRepository."""
533
# FIXME: Also check target uses VersionedFile
534
return (isinstance(source, LocalGitRepository) and
535
target.supports_rich_root() and
536
not isinstance(target, GitRepository))
539
class InterGitGitRepository(InterGitRepository):
540
"""InterRepository that copies between Git repositories."""
542
def fetch_objects(self, determine_wants, mapping, pb=None):
544
trace.note("git: %s", text)
545
graphwalker = self.target._git.get_graph_walker()
546
if isinstance(self.source, LocalGitRepository) and isinstance(self.target, LocalGitRepository):
547
return self.source._git.fetch(self.target._git, determine_wants,
549
elif isinstance(self.source, LocalGitRepository) and isinstance(self.target, RemoteGitRepository):
550
raise NotImplementedError
551
elif isinstance(self.source, RemoteGitRepository) and isinstance(self.target, LocalGitRepository):
552
f, commit = self.target._git.object_store.add_thin_pack()
554
refs = self.source._git.fetch_pack(determine_wants, graphwalker,
564
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
565
mapping=None, fetch_spec=None, branches=None):
567
mapping = self.source.get_mapping()
569
if revision_id is not None:
570
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
571
elif fetch_spec is not None:
572
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
573
if branches is not None:
574
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
575
elif fetch_spec is None and revision_id is None:
576
determine_wants = r.object_store.determine_wants_all
578
determine_wants = lambda x: [y for y in args if not y in r.object_store]
579
return self.fetch_objects(determine_wants, mapping)
583
def is_compatible(source, target):
584
"""Be compatible with GitRepository."""
585
return (isinstance(source, GitRepository) and
586
isinstance(target, GitRepository))
added
removed
fetch.py
