77
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
121
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
122
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
78
123
"""Import a git blob object into a bzr repository.
80
:param repo: bzr repository
125
:param texts: VersionedFiles to add to
81
126
:param path: Path in the tree
82
127
:param blob: A git blob
128
:return: Inventory delta for this file
84
130
file_id = mapping.generate_file_id(path)
85
text_revision = inv.revision_id
86
repo.texts.add_lines((file_id, text_revision),
87
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
88
osutils.split_lines(blob.data))
89
ie = inv.add_path(path, "file", file_id)
90
ie.revision = text_revision
91
ie.text_size = len(blob.data)
92
ie.text_sha1 = osutils.sha_string(blob.data)
135
# We just have to hope this is indeed utf-8:
136
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
93
137
ie.executable = executable
96
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
139
# See if this has changed at all
141
base_ie = base_inv[file_id]
146
base_sha = base_ie.text_id
149
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
153
if (base_sha == hexsha and base_ie.executable == ie.executable
154
and base_ie.kind == ie.kind):
155
# If nothing has changed since the base revision, we're done
157
if base_sha == hexsha:
158
ie.text_size = base_ie.text_size
159
ie.text_sha1 = base_ie.text_sha1
160
ie.symlink_target = base_ie.symlink_target
161
ie.revision = base_ie.revision
163
blob = lookup_object(hexsha)
164
if ie.kind == "symlink":
165
ie.symlink_target = blob.data
169
ie.text_size = len(blob.data)
170
ie.text_sha1 = osutils.sha_string(blob.data)
171
# Check what revision we should store
173
for pinv in parent_invs:
174
if pinv.revision_id == base_inv.revision_id:
183
if pie.text_sha1 == ie.text_sha1:
184
# found a revision in one of the parents to use
185
ie.revision = pie.revision
187
parent_keys.append((file_id, pie.revision))
188
if ie.revision is None:
189
# Need to store a new revision
190
ie.revision = revision_id
191
assert file_id is not None
192
assert ie.revision is not None
193
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
194
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
197
if file_id in base_inv:
198
old_path = base_inv.id2path(file_id)
201
return ([(old_path, path, file_id, ie)], shamap)
204
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
205
revision_id, parent_invs, shagitmap, lookup_object):
97
206
"""Import a git tree object into a bzr repository.
99
:param repo: A Bzr repository object
208
:param texts: VersionedFiles object to add to
100
209
:param path: Path in the tree
101
210
:param tree: A git tree object
102
:param inv: Inventory object
211
:param base_inv: Base inventory against which to return inventory delta
212
:return: Inventory delta for this subtree
104
215
file_id = mapping.generate_file_id(path)
105
text_revision = inv.revision_id
106
repo.texts.add_lines((file_id, text_revision),
107
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
109
ie = inv.add_path(path, "directory", file_id)
110
ie.revision = text_revision
111
for mode, name, hexsha in tree.entries():
112
entry_kind = (mode & 0700000) / 0100000
216
# We just have to hope this is indeed utf-8:
217
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
221
base_ie = base_inv[file_id]
223
# Newly appeared here
225
ie.revision = revision_id
226
texts.add_lines((file_id, ie.revision), (), [])
227
invdelta.append((None, path, file_id, ie))
229
base_sha = base_ie.text_id
230
# See if this has changed at all
233
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
237
if base_sha == hexsha:
238
# If nothing has changed since the base revision, we're done
240
# Remember for next time
241
existing_children = set()
244
tree = lookup_object(hexsha)
245
for mode, name, child_hexsha in tree.entries():
113
246
basename = name.decode("utf-8")
117
child_path = urlutils.join(path, name)
119
tree = lookup_object(hexsha)
120
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
121
elif entry_kind == 1:
122
blob = lookup_object(hexsha)
123
fs_mode = mode & 0777
124
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
126
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
129
def import_git_objects(repo, mapping, num_objects, object_iter, pb=None):
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(texts,
251
mapping, child_path, child_hexsha, base_inv, file_id,
252
revision_id, parent_invs, shagitmap, lookup_object)
253
invdelta.extend(subinvdelta)
254
child_modes.update(grandchildmodes)
255
shamap.extend(subshamap)
257
fs_mode = stat.S_IMODE(mode)
258
symlink = stat.S_ISLNK(mode)
259
subinvdelta, subshamap = import_git_blob(texts, mapping,
260
child_path, child_hexsha, base_inv, file_id, revision_id,
261
parent_invs, shagitmap, lookup_object,
262
bool(fs_mode & 0111), symlink)
263
invdelta.extend(subinvdelta)
264
shamap.extend(subshamap)
265
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
266
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
267
child_modes[child_path] = mode
268
# Remove any children that have disappeared
269
if base_ie is not None:
270
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
273
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
274
if ie.kind == "directory":
275
deletable.extend(ie.children.values())
276
shamap.append((hexsha, "tree", (file_id, revision_id)))
277
return invdelta, child_modes, shamap
280
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
130
282
"""Import a set of git objects into a bzr repository.
132
284
:param repo: Bazaar repository
133
285
:param mapping: Mapping to use
134
:param num_objects: Number of objects.
135
286
:param object_iter: Iterator over Git objects.
137
288
# TODO: a more (memory-)efficient implementation of this
139
for i, (o, _) in enumerate(object_iter):
141
pb.update("fetching objects", i, num_objects)
294
parent_invs_cache = LRUCache(50)
146
295
# Find and convert commit objects
147
for o in objects.itervalues():
298
pb.update("finding revisions to fetch", len(graph), None)
300
assert isinstance(head, str)
302
o = object_iter[head]
148
305
if isinstance(o, Commit):
149
306
rev = mapping.import_commit(o)
150
root_trees[rev.revision_id] = objects[o.tree]
307
if repo.has_revision(rev.revision_id):
309
root_trees[rev.revision_id] = o.tree
151
310
revisions[rev.revision_id] = rev
152
311
graph.append((rev.revision_id, rev.parent_ids))
312
target_git_object_retriever._idmap.add_entry(o.id, "commit",
313
(rev.revision_id, o.tree))
314
heads.extend([p for p in o.parents if p not in checked])
315
elif isinstance(o, Tag):
316
heads.append(o.object[1])
318
trace.warning("Unable to import head object %r" % o)
153
320
# Order the revisions
154
321
# Create the inventory objects
155
322
for i, revid in enumerate(topo_sort(graph)):
156
323
if pb is not None:
157
324
pb.update("fetching revisions", i, len(graph))
158
root_tree = root_trees[revid]
159
325
rev = revisions[revid]
160
326
# We have to do this here, since we have to walk the tree and
161
# we need to make sure to import the blobs / trees with the riht
327
# we need to make sure to import the blobs / trees with the right
162
328
# path; this may involve adding them more than once.
164
inv.revision_id = rev.revision_id
165
329
def lookup_object(sha):
168
return reconstruct_git_object(repo, mapping, sha)
169
parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
170
import_git_tree(repo, mapping, "", root_tree, inv, parent_invs,
172
repo.add_revision(rev.revision_id, rev, inv)
175
def reconstruct_git_commit(repo, rev):
176
raise NotImplementedError(self.reconstruct_git_commit)
179
def reconstruct_git_object(repo, mapping, sha):
181
revid = mapping.revision_id_foreign_to_bzr(sha)
183
rev = repo.get_revision(revid)
184
except NoSuchRevision:
187
return reconstruct_git_commit(rev)
191
raise KeyError("No such object %s" % sha)
194
class InterGitNonGitRepository(InterRepository):
196
_matching_repo_format = GitFormat()
331
return object_iter[sha]
333
return target_git_object_retriever[sha]
335
for parent_id in rev.parent_ids:
337
parent_invs.append(parent_invs_cache[parent_id])
339
parent_inv = repo.get_inventory(parent_id)
340
parent_invs.append(parent_inv)
341
parent_invs_cache[parent_id] = parent_inv
342
if parent_invs == []:
343
base_inv = Inventory(root_id=None)
345
base_inv = parent_invs[0]
346
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
347
mapping, "", root_trees[revid], base_inv, None, revid,
348
parent_invs, target_git_object_retriever._idmap, lookup_object)
349
target_git_object_retriever._idmap.add_entries(shamap)
350
if unusual_modes != {}:
351
ret = "unusual modes: \n"
352
for item in unusual_modes.iteritems():
353
ret += "\t%s: %o\n" % item
354
raise AssertionError(ret)
356
basis_id = rev.parent_ids[0]
358
basis_id = NULL_REVISION
359
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
360
inv_delta, rev.revision_id, rev.parent_ids)
361
parent_invs_cache[rev.revision_id] = inv
362
repo.add_revision(rev.revision_id, rev)
363
if "verify" in debug.debug_flags:
364
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
365
for sha1, newobj, path in objs:
366
assert path is not None
367
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
368
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
370
target_git_object_retriever._idmap.commit()
373
class InterGitRepository(InterRepository):
375
_matching_repo_format = GitRepositoryFormat()
199
378
def _get_repo_format_to_test():
203
382
"""See InterRepository.copy_content."""
204
383
self.fetch(revision_id, pb, find_ghosts=False)
206
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
385
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
387
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
388
mapping=mapping, fetch_spec=fetch_spec)
391
class InterGitNonGitRepository(InterGitRepository):
392
"""Base InterRepository that copies revisions from a Git into a non-Git
395
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
396
mapping=None, fetch_spec=None):
208
397
if mapping is None:
209
398
mapping = self.source.get_mapping()
211
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
212
def determine_wants(heads):
213
if revision_id is None:
399
if revision_id is not None:
400
interesting_heads = [revision_id]
401
elif fetch_spec is not None:
402
interesting_heads = fetch_spec.heads
404
interesting_heads = None
406
def determine_wants(refs):
408
if interesting_heads is None:
409
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
216
ret = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
411
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid != NULL_REVISION]
217
412
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
413
self.fetch_objects(determine_wants, mapping, pb)
418
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
419
"""InterRepository that copies revisions from a remote Git into a non-Git
422
def fetch_objects(self, determine_wants, mapping, pb=None):
424
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
218
425
graph_walker = BzrFetchGraphWalker(self.target, mapping)
221
428
create_pb = pb = ui.ui_factory.nested_progress_bar()
223
self.target.lock_write()
225
self.target.start_write_group()
227
(num_objects, objects_iter) = \
228
self.source.fetch_objects(determine_wants,
229
graph_walker, progress)
230
import_git_objects(self.target, mapping, num_objects,
233
self.target.commit_write_group()
241
def is_compatible(source, target):
242
"""Be compatible with GitRepository."""
243
# FIXME: Also check target uses VersionedFile
244
return (isinstance(source, GitRepository) and
245
target.supports_rich_root() and
246
not isinstance(target, GitRepository))
249
class InterGitRepository(InterRepository):
251
_matching_repo_format = GitFormat()
254
def _get_repo_format_to_test():
257
def copy_content(self, revision_id=None, pb=None):
258
"""See InterRepository.copy_content."""
259
self.fetch(revision_id, pb, find_ghosts=False)
261
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
429
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
432
def record_determine_wants(heads):
433
wants = determine_wants(heads)
434
recorded_wants.extend(wants)
438
self.target.lock_write()
440
self.target.start_write_group()
442
objects_iter = self.source.fetch_objects(
443
record_determine_wants,
445
target_git_object_retriever.get_raw,
447
import_git_objects(self.target, mapping, objects_iter,
448
target_git_object_retriever, recorded_wants, pb)
450
self.target.commit_write_group()
458
def is_compatible(source, target):
459
"""Be compatible with GitRepository."""
460
# FIXME: Also check target uses VersionedFile
461
return (isinstance(source, RemoteGitRepository) and
462
target.supports_rich_root() and
463
not isinstance(target, GitRepository))
466
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
467
"""InterRepository that copies revisions from a remote Git into a non-Git
470
def fetch_objects(self, determine_wants, mapping, pb=None):
471
wants = determine_wants(self.source._git.get_refs())
474
create_pb = pb = ui.ui_factory.nested_progress_bar()
475
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
477
self.target.lock_write()
479
self.target.start_write_group()
481
import_git_objects(self.target, mapping,
482
self.source._git.object_store,
483
target_git_object_retriever, wants, pb)
485
self.target.commit_write_group()
493
def is_compatible(source, target):
494
"""Be compatible with GitRepository."""
495
# FIXME: Also check target uses VersionedFile
496
return (isinstance(source, LocalGitRepository) and
497
target.supports_rich_root() and
498
not isinstance(target, GitRepository))
501
class InterGitGitRepository(InterGitRepository):
502
"""InterRepository that copies between Git repositories."""
504
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
505
mapping=None, fetch_spec=None, branches=None):
263
506
if mapping is None:
264
507
mapping = self.source.get_mapping()
265
508
def progress(text):
266
info("git: %s", text)
509
trace.info("git: %s", text)
267
510
r = self.target._git
268
if revision_id is None:
269
determine_wants = lambda x: [y for y in x.values() if not y in r.object_store]
511
if revision_id is not None:
271
512
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
513
elif fetch_spec is not None:
514
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
515
if branches is not None:
516
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
517
elif fetch_spec is None and revision_id is None:
518
determine_wants = r.object_store.determine_wants_all
272
520
determine_wants = lambda x: [y for y in args if not y in r.object_store]
274
graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
275
f, commit = r.object_store.add_pack()
522
graphwalker = r.get_graph_walker()
523
f, commit = r.object_store.add_thin_pack()
277
self.source._git.fetch_pack(path, determine_wants, graphwalker, f.write, progress)
525
refs = self.source.fetch_pack(determine_wants, graphwalker,