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 (
23
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
ChunkedContentFactory,
60
from bzrlib.plugins.git.mapping import (
62
inventory_to_tree_and_blobs,
68
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, name, (base_hexsha, hexsha),
83
base_inv, base_inv_shamap, parent_id, revision_id,
84
parent_invs, lookup_object, (base_mode, mode)):
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
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):
92
if base_hexsha == hexsha and base_mode == mode:
93
# If nothing has changed since the base revision, we're done
95
file_id = mapping.generate_file_id(path)
96
if stat.S_ISLNK(mode):
100
ie = cls(file_id, name.decode("utf-8"), parent_id)
101
ie.executable = mode_is_executable(mode)
102
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
103
base_ie = base_inv[base_inv.path2id(path)]
104
ie.text_size = base_ie.text_size
105
ie.text_sha1 = base_ie.text_sha1
106
ie.symlink_target = base_ie.symlink_target
107
if ie.executable == base_ie.executable:
108
ie.revision = base_ie.revision
110
blob = lookup_object(hexsha)
112
blob = lookup_object(hexsha)
113
if ie.kind == "symlink":
115
ie.symlink_target = blob.data
119
ie.text_size = len(blob.data)
120
ie.text_sha1 = osutils.sha_string(blob.data)
121
# Check what revision we should store
123
for pinv in parent_invs[1:]:
124
if pinv.revision_id == base_inv.revision_id:
133
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
134
# found a revision in one of the parents to use
135
ie.revision = pie.revision
137
parent_keys.append((file_id, pie.revision))
138
if ie.revision is None:
139
# Need to store a new revision
140
ie.revision = revision_id
141
assert ie.revision is not None
142
if ie.kind == 'symlink':
146
chunks = blob.chunked
147
except AttributeError: # older version of dulwich
149
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, chunks)])
150
shamap = { ie.file_id: hexsha }
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
return (invdelta, shamap)
163
class SubmodulesRequireSubtrees(BzrError):
164
_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'."""
168
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
169
base_inv, parent_id, revision_id, parent_invs, lookup_object,
171
if base_hexsha == hexsha and base_mode == mode:
173
file_id = mapping.generate_file_id(path)
174
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
175
ie.revision = revision_id
176
if base_hexsha is None:
180
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
181
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
182
invdelta = [(oldpath, path, file_id, ie)]
183
return invdelta, {}, {}
186
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
189
for name, mode, hexsha in base_tree.iteritems():
190
if name in existing_children:
192
c_path = posixpath.join(path, name.decode("utf-8"))
193
ret.append((c_path, None, base_inv.path2id(c_path), None))
194
if stat.S_ISDIR(mode):
195
ret.extend(remove_disappeared_children(
196
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
200
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
201
base_inv, base_inv_shamap, parent_id, revision_id, parent_invs,
202
lookup_object, (base_mode, mode), allow_submodules=False):
82
203
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
205
:param texts: VersionedFiles object to add to
85
206
:param path: Path in the tree
86
207
:param tree: A git tree object
87
:param inv: Inventory object
208
:param base_inv: Base inventory against which to return inventory delta
209
:return: Inventory delta for this subtree
211
if base_hexsha == hexsha and base_mode == mode:
212
# If nothing has changed since the base revision, we're done
89
215
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
96
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):
216
# We just have to hope this is indeed utf-8:
217
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
218
tree = lookup_object(hexsha)
219
if base_hexsha is None:
221
old_path = None # Newly appeared here
223
base_tree = lookup_object(base_hexsha)
224
old_path = path # Renames aren't supported yet
225
if base_tree is None or type(base_tree) is not Tree:
226
ie.revision = revision_id
227
invdelta.append((old_path, path, ie.file_id, ie))
228
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
229
# Remember for next time
230
existing_children = set()
233
for child_mode, name, child_hexsha in tree.entries():
234
existing_children.add(name)
235
child_path = posixpath.join(path, name)
236
if type(base_tree) is Tree:
238
child_base_mode, child_base_hexsha = base_tree[name]
240
child_base_hexsha = None
243
child_base_hexsha = None
245
if stat.S_ISDIR(child_mode):
246
subinvdelta, grandchildmodes, subshamap = import_git_tree(
247
texts, mapping, child_path, name,
248
(child_base_hexsha, child_hexsha),
249
base_inv, base_inv_shamap,
250
file_id, revision_id, parent_invs, lookup_object,
251
(child_base_mode, child_mode),
252
allow_submodules=allow_submodules)
253
elif S_ISGITLINK(child_mode): # submodule
254
if not allow_submodules:
255
raise SubmodulesRequireSubtrees()
256
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
257
texts, mapping, child_path, name,
258
(child_base_hexsha, child_hexsha),
259
base_inv, file_id, revision_id, parent_invs, lookup_object,
260
(child_base_mode, child_mode))
262
subinvdelta, subshamap = import_git_blob(texts, mapping,
263
child_path, name, (child_base_hexsha, child_hexsha),
264
base_inv, base_inv_shamap,
266
revision_id, parent_invs, lookup_object,
267
(child_base_mode, child_mode))
269
child_modes.update(grandchildmodes)
270
invdelta.extend(subinvdelta)
271
shamap.update(subshamap)
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
shamap[file_id] = hexsha
280
return invdelta, child_modes, shamap
283
def import_git_commit(repo, mapping, head, lookup_object,
284
target_git_object_retriever, parent_invs_cache):
285
o = lookup_object(head)
286
rev = mapping.import_commit(o)
287
# We have to do this here, since we have to walk the tree and
288
# we need to make sure to import the blobs / trees with the right
289
# path; this may involve adding them more than once.
290
parent_invs = parent_invs_cache.get_inventories(rev.parent_ids)
291
if parent_invs == []:
292
base_inv = Inventory(root_id=None)
293
base_inv_shamap = None # Should never be accessed
297
base_inv = parent_invs[0]
298
base_inv_shamap = target_git_object_retriever._idmap.get_inventory_sha_map(base_inv.revision_id)
299
base_tree = lookup_object(o.parents[0]).tree
300
base_mode = stat.S_IFDIR
301
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
302
mapping, "", u"", (base_tree, o.tree), base_inv, base_inv_shamap,
303
None, rev.revision_id, parent_invs, lookup_object,
304
(base_mode, stat.S_IFDIR),
305
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
307
for (oldpath, newpath, fileid, new_ie) in inv_delta:
309
entries.append((fileid, None, None, None))
311
if new_ie.kind in ("file", "symlink"):
312
entries.append((fileid, "blob", shamap[fileid], new_ie.revision))
313
elif new_ie.kind == "directory":
314
entries.append((fileid, "tree", shamap[fileid], rev.revision_id))
317
target_git_object_retriever._idmap.add_entries(rev.revision_id,
318
rev.parent_ids, head, o.tree, entries)
319
if unusual_modes != {}:
320
for path, mode in unusual_modes.iteritems():
321
warn_unusual_mode(rev.foreign_revid, path, mode)
322
mapping.import_unusual_file_modes(rev, unusual_modes)
324
basis_id = rev.parent_ids[0]
326
basis_id = NULL_REVISION
328
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
329
inv_delta, rev.revision_id, rev.parent_ids,
331
parent_invs_cache.add(rev.revision_id, inv)
332
repo.add_revision(rev.revision_id, rev)
333
if "verify" in debug.debug_flags:
334
new_unusual_modes = mapping.export_unusual_file_modes(rev)
335
if new_unusual_modes != unusual_modes:
336
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
337
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
338
for newsha1, newobj, path in objs:
339
assert path is not None
343
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
344
if oldsha1 != newsha1:
345
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
348
def import_git_objects(repo, mapping, object_iter,
349
target_git_object_retriever, heads, pb=None, limit=None):
110
350
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
352
:param repo: Target Bazaar repository
113
353
:param mapping: Mapping to use
114
354
:param object_iter: Iterator over Git objects.
355
:return: Tuple with pack hints and last imported revision id
116
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
357
def lookup_object(sha):
359
return object_iter[sha]
361
return target_git_object_retriever[sha]
364
heads = list(set(heads))
365
parent_invs_cache = LRUInventoryCache(repo)
121
366
# Find and convert commit objects
122
for o in objects.iterkeys():
369
pb.update("finding revisions to fetch", len(graph), None)
371
assert isinstance(head, str)
373
o = lookup_object(head)
123
376
if isinstance(o, Commit):
124
377
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
378
if repo.has_revision(rev.revision_id):
380
squash_revision(repo, rev)
381
graph.append((o.id, o.parents))
382
heads.extend([p for p in o.parents if p not in checked])
383
elif isinstance(o, Tag):
384
if o.object[1] not in checked:
385
heads.append(o.object[1])
387
trace.warning("Unable to import head object %r" % o)
390
# Order the revisions
126
391
# 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)
393
revision_ids = topo_sort(graph)
395
if limit is not None:
396
revision_ids = revision_ids[:limit]
398
for offset in range(0, len(revision_ids), batch_size):
399
target_git_object_retriever.start_write_group() # FIXME: try/finally
401
repo.start_write_group()
403
for i, head in enumerate(
404
revision_ids[offset:offset+batch_size]):
406
pb.update("fetching revisions", offset+i,
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)
420
target_git_object_retriever.abort_write_group()
423
target_git_object_retriever.commit_write_group()
424
return pack_hints, last_imported
159
427
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
429
_matching_repo_format = GitRepositoryFormat()
164
432
def _get_repo_format_to_test():
168
436
"""See InterRepository.copy_content."""
169
437
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
439
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
440
mapping=None, fetch_spec=None):
441
self.fetch_refs(revision_id=revision_id, pb=pb,
442
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
445
class InterGitNonGitRepository(InterGitRepository):
446
"""Base InterRepository that copies revisions from a Git into a non-Git
449
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
450
mapping=None, fetch_spec=None):
173
451
if mapping is None:
174
452
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)]
453
if revision_id is not None:
454
interesting_heads = [revision_id]
455
elif fetch_spec is not None:
456
interesting_heads = fetch_spec.heads
458
interesting_heads = None
460
def determine_wants(refs):
462
if interesting_heads is None:
463
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
465
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
466
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)
467
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
468
if pack_hint is not None and self.target._format.pack_compresses:
469
self.target.pack(hint=pack_hint)
470
if interesting_heads is not None:
471
present_interesting_heads = self.target.has_revisions(interesting_heads)
472
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
473
if missing_interesting_heads:
474
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
478
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
479
def report_git_progress(pb, text):
480
text = text.rstrip("\r\n")
481
g = _GIT_PROGRESS_RE.match(text)
483
(text, pct, current, total) = g.groups()
484
pb.update(text, int(current), int(total))
486
pb.update(text, 0, 0)
489
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
490
"""InterRepository that copies revisions from a remote Git into a non-Git
493
def get_target_heads(self):
494
# FIXME: This should be more efficient
495
all_revs = self.target.all_revision_ids()
496
parent_map = self.target.get_parent_map(all_revs)
498
map(all_parents.update, parent_map.itervalues())
499
return set(all_revs) - all_parents
501
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
503
report_git_progress(pb, text)
504
store = BazaarObjectStore(self.target, mapping)
187
505
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
507
heads = self.get_target_heads()
508
graph_walker = store.get_graph_walker(
509
[store._lookup_revision_sha1(head) for head in heads])
512
def record_determine_wants(heads):
513
wants = determine_wants(heads)
514
recorded_wants.extend(wants)
519
create_pb = pb = ui.ui_factory.nested_progress_bar()
521
objects_iter = self.source.fetch_objects(
522
record_determine_wants, graph_walker,
523
store.get_raw, progress)
524
return import_git_objects(self.target, mapping,
525
objects_iter, store, recorded_wants, pb, limit)
193
530
self.target.unlock()
added
removed
fetch.py
