156
150
ie.revision = revision_id
157
151
assert file_id is not None
158
152
assert ie.revision is not None
159
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
153
if ie.kind == 'symlink':
157
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, data)])
160
158
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
163
if file_id in base_inv:
162
if base_ie is not None:
164
163
old_path = base_inv.id2path(file_id)
164
if base_ie.kind == "directory":
165
invdelta.extend(remove_disappeared_children(old_path, base_ie.children, []))
167
invdelta = [(old_path, path, file_id, ie)]
168
invdelta.extend(remove_disappeared_children(base_inv, base_ie, []))
168
invdelta.append((old_path, path, file_id, ie))
169
169
return (invdelta, shamap)
172
def import_git_submodule(texts, mapping, path, hexsha, base_inv, parent_id,
173
revision_id, parent_invs, shagitmap, lookup_object):
174
raise NotImplementedError(import_git_submodule)
177
def remove_disappeared_children(base_inv, base_ie, existing_children):
178
if base_ie is None or base_ie.kind != 'directory':
172
class SubmodulesRequireSubtrees(BzrError):
173
_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'."""
177
def import_git_submodule(texts, mapping, path, hexsha, base_inv, base_ie,
178
parent_id, revision_id, parent_invs, shagitmap, lookup_object):
179
file_id = mapping.generate_file_id(path)
180
ie = TreeReference(file_id, urlutils.basename(path.decode("utf-8")),
182
ie.revision = revision_id
187
if (base_ie.kind == ie.kind and
188
base_ie.reference_revision == ie.reference_revision):
189
ie.revision = base_ie.revision
190
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
191
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
192
invdelta = [(oldpath, path, file_id, ie)]
193
return invdelta, {}, {}
196
def remove_disappeared_children(path, base_children, existing_children):
181
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
198
deletable = [(osutils.pathjoin(path, k), v) for k,v in base_children.iteritems() if k not in existing_children]
184
ret.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
200
(path, ie) = deletable.pop()
201
ret.append((path, None, ie.file_id, None))
185
202
if ie.kind == "directory":
186
deletable.extend(ie.children.values())
203
for name, child_ie in ie.children.iteritems():
204
deletable.append((osutils.pathjoin(path, name), child_ie))
190
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
191
revision_id, parent_invs, shagitmap, lookup_object):
208
def import_git_tree(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
209
revision_id, parent_invs, shagitmap, lookup_object, allow_submodules=False):
192
210
"""Import a git tree object into a bzr repository.
194
212
:param texts: VersionedFiles object to add to
201
219
file_id = mapping.generate_file_id(path)
202
220
# We just have to hope this is indeed utf-8:
203
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
221
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
206
base_ie = base_inv[file_id]
208
224
# Newly appeared here
210
225
ie.revision = revision_id
211
texts.add_lines((file_id, ie.revision), (), [])
226
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
212
227
invdelta.append((None, path, file_id, ie))
214
229
# See if this has changed at all
235
254
child_path = osutils.pathjoin(path, name)
236
255
if stat.S_ISDIR(mode):
237
256
subinvdelta, grandchildmodes, subshamap = import_git_tree(
238
texts, mapping, child_path, child_hexsha, base_inv,
239
file_id, revision_id, parent_invs, shagitmap, lookup_object)
257
texts, mapping, child_path, child_hexsha, base_inv,
258
base_children.get(basename), file_id, revision_id,
259
parent_invs, shagitmap, lookup_object,
260
allow_submodules=allow_submodules)
240
261
invdelta.extend(subinvdelta)
241
262
child_modes.update(grandchildmodes)
242
263
shamap.extend(subshamap)
243
264
elif S_ISGITLINK(mode): # submodule
265
if not allow_submodules:
266
raise SubmodulesRequireSubtrees()
244
267
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
245
texts, mapping, child_path, child_hexsha, base_inv,
268
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
246
269
file_id, revision_id, parent_invs, shagitmap, lookup_object)
247
270
invdelta.extend(subinvdelta)
248
271
child_modes.update(grandchildmodes)
249
272
shamap.extend(subshamap)
251
subinvdelta, subshamap = import_git_blob(texts, mapping,
252
child_path, child_hexsha, base_inv, file_id, revision_id,
253
parent_invs, shagitmap, lookup_object,
274
subinvdelta, subshamap = import_git_blob(texts, mapping,
275
child_path, child_hexsha, base_inv, base_children.get(basename), file_id,
276
revision_id, parent_invs, shagitmap, lookup_object,
254
277
mode_is_executable(mode), stat.S_ISLNK(mode))
255
278
invdelta.extend(subinvdelta)
256
279
shamap.extend(subshamap)
258
281
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
259
282
child_modes[child_path] = mode
260
283
# Remove any children that have disappeared
261
invdelta.extend(remove_disappeared_children(base_inv, base_ie, existing_children))
284
if base_ie is not None and base_ie.kind == "directory":
285
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
286
base_children, existing_children))
262
287
shamap.append((hexsha, "tree", (file_id, revision_id)))
263
288
return invdelta, child_modes, shamap
266
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
291
def approx_inv_size(inv):
292
# Very rough estimate, 1k per inventory entry
293
return len(inv) * 1024
296
def import_git_commit(repo, mapping, head, lookup_object,
297
target_git_object_retriever, parent_invs_cache):
298
o = lookup_object(head)
299
rev = mapping.import_commit(o)
300
# We have to do this here, since we have to walk the tree and
301
# we need to make sure to import the blobs / trees with the right
302
# path; this may involve adding them more than once.
304
for parent_id in rev.parent_ids:
306
parent_invs.append(parent_invs_cache[parent_id])
308
parent_inv = repo.get_inventory(parent_id)
309
parent_invs.append(parent_inv)
310
parent_invs_cache[parent_id] = parent_inv
311
if parent_invs == []:
312
base_inv = Inventory(root_id=None)
315
base_inv = parent_invs[0]
316
base_ie = base_inv.root
317
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
318
mapping, "", o.tree, base_inv, base_ie, None, rev.revision_id,
319
parent_invs, target_git_object_retriever._idmap, lookup_object,
320
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
321
target_git_object_retriever._idmap.add_entries(shamap)
322
if unusual_modes != {}:
323
for path, mode in unusual_modes.iteritems():
324
warn_unusual_mode(rev.foreign_revid, path, mode)
325
mapping.import_unusual_file_modes(rev, unusual_modes)
327
basis_id = rev.parent_ids[0]
329
basis_id = NULL_REVISION
331
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
332
inv_delta, rev.revision_id, rev.parent_ids,
334
parent_invs_cache[rev.revision_id] = inv
335
repo.add_revision(rev.revision_id, rev)
336
if "verify" in debug.debug_flags:
337
new_unusual_modes = mapping.export_unusual_file_modes(rev)
338
if new_unusual_modes != unusual_modes:
339
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
340
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
341
for sha1, newobj, path in objs:
342
assert path is not None
343
oldobj = tree_lookup_path(lookup_object, o.tree, path)
345
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
348
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
268
350
"""Import a set of git objects into a bzr repository.
293
375
o = lookup_object(head)
377
trace.mutter('missing head %s', head)
296
379
if isinstance(o, Commit):
297
380
rev = mapping.import_commit(o)
298
381
if repo.has_revision(rev.revision_id):
300
383
squash_revision(repo, rev)
301
root_trees[rev.revision_id] = o.tree
302
revisions[rev.revision_id] = rev
303
graph.append((rev.revision_id, rev.parent_ids))
304
target_git_object_retriever._idmap.add_entry(o.id, "commit",
384
graph.append((o.id, o.parents))
385
target_git_object_retriever._idmap.add_entry(o.id, "commit",
305
386
(rev.revision_id, o.tree))
306
387
heads.extend([p for p in o.parents if p not in checked])
307
388
elif isinstance(o, Tag):
308
389
heads.append(o.object[1])
310
391
trace.warning("Unable to import head object %r" % o)
312
394
# Order the revisions
313
395
# Create the inventory objects
314
for i, revid in enumerate(topo_sort(graph)):
316
pb.update("fetching revisions", i, len(graph))
317
rev = revisions[revid]
318
# We have to do this here, since we have to walk the tree and
319
# we need to make sure to import the blobs / trees with the right
320
# path; this may involve adding them more than once.
322
for parent_id in rev.parent_ids:
324
parent_invs.append(parent_invs_cache[parent_id])
326
parent_inv = repo.get_inventory(parent_id)
327
parent_invs.append(parent_inv)
328
parent_invs_cache[parent_id] = parent_inv
329
if parent_invs == []:
330
base_inv = Inventory(root_id=None)
397
revision_ids = topo_sort(graph)
399
for offset in range(0, len(revision_ids), batch_size):
400
repo.start_write_group()
402
for i, head in enumerate(revision_ids[offset:offset+batch_size]):
404
pb.update("fetching revisions", offset+i, len(revision_ids))
405
import_git_commit(repo, mapping, head, lookup_object,
406
target_git_object_retriever,
409
repo.abort_write_group()
332
base_inv = parent_invs[0]
333
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
334
mapping, "", root_trees[revid], base_inv, None, revid,
335
parent_invs, target_git_object_retriever._idmap, lookup_object)
336
target_git_object_retriever._idmap.add_entries(shamap)
337
if unusual_modes != {}:
338
for path, mode in unusual_modes.iteritems():
339
warn_unusual_mode(rev.foreign_revid, path, mode)
340
mapping.import_unusual_file_modes(rev, unusual_modes)
342
basis_id = rev.parent_ids[0]
344
basis_id = NULL_REVISION
345
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
346
inv_delta, rev.revision_id, rev.parent_ids)
347
parent_invs_cache[rev.revision_id] = inv
348
repo.add_revision(rev.revision_id, rev)
349
if "verify" in debug.debug_flags:
350
new_unusual_modes = mapping.export_unusual_file_modes(rev)
351
if new_unusual_modes != unusual_modes:
352
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
353
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
354
for sha1, newobj, path in objs:
355
assert path is not None
356
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
358
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
360
target_git_object_retriever._idmap.commit()
412
hint = repo.commit_write_group()
414
pack_hints.extend(hint)
415
target_git_object_retriever._idmap.commit_write_group()
363
419
class InterGitRepository(InterRepository):
401
457
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
402
458
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
403
self.fetch_objects(determine_wants, mapping, pb)
459
pack_hint = self.fetch_objects(determine_wants, mapping, pb)
460
if pack_hint is not None and self.target._format.pack_compresses:
461
self.target.pack(hint=pack_hint)
462
if interesting_heads is not None:
463
present_interesting_heads = self.target.has_revisions(interesting_heads)
464
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
465
if missing_interesting_heads:
466
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
404
467
return self._refs
470
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
471
def report_git_progress(pb, text):
472
text = text.rstrip("\r\n")
473
g = _GIT_PROGRESS_RE.match(text)
475
(text, pct, current, total) = g.groups()
476
pb.update(text, int(current), int(total))
478
pb.update(text, 0, 0)
407
481
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
408
"""InterRepository that copies revisions from a remote Git into a non-Git
482
"""InterRepository that copies revisions from a remote Git into a non-Git
485
def get_target_heads(self):
486
# FIXME: This should be more efficient
487
all_revs = self.target.all_revision_ids()
488
parent_map = self.target.get_parent_map(all_revs)
490
map(all_parents.update, parent_map.itervalues())
491
return set(all_revs) - all_parents
411
493
def fetch_objects(self, determine_wants, mapping, pb=None):
412
494
def progress(text):
413
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
495
report_git_progress(pb, text)
414
496
store = BazaarObjectStore(self.target, mapping)
415
497
self.target.lock_write()
417
heads = self.target.get_graph().heads(self.target.all_revision_ids())
499
heads = self.get_target_heads()
418
500
graph_walker = store.get_graph_walker(
419
501
[store._lookup_revision_sha1(head) for head in heads])
420
502
recorded_wants = []
423
505
wants = determine_wants(heads)
424
506
recorded_wants.extend(wants)
429
511
create_pb = pb = ui.ui_factory.nested_progress_bar()
431
self.target.start_write_group()
433
objects_iter = self.source.fetch_objects(
434
record_determine_wants, graph_walker,
435
store.get_raw, progress)
436
import_git_objects(self.target, mapping, objects_iter,
437
store, recorded_wants, pb)
439
self.target.commit_write_group()
513
objects_iter = self.source.fetch_objects(
514
record_determine_wants, graph_walker,
515
store.get_raw, progress)
516
return import_git_objects(self.target, mapping,
517
objects_iter, store, recorded_wants, pb)
442
520
create_pb.finished()
447
525
def is_compatible(source, target):
448
526
"""Be compatible with GitRepository."""
449
527
# FIXME: Also check target uses VersionedFile
450
return (isinstance(source, RemoteGitRepository) and
528
return (isinstance(source, RemoteGitRepository) and
451
529
target.supports_rich_root() and
452
530
not isinstance(target, GitRepository))
455
533
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
456
"""InterRepository that copies revisions from a local Git into a non-Git
534
"""InterRepository that copies revisions from a local Git into a non-Git
459
537
def fetch_objects(self, determine_wants, mapping, pb=None):
490
564
class InterGitGitRepository(InterGitRepository):
491
565
"""InterRepository that copies between Git repositories."""
493
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
567
def fetch_objects(self, determine_wants, mapping, pb=None):
569
trace.note("git: %s", text)
570
graphwalker = self.target._git.get_graph_walker()
571
if (isinstance(self.source, LocalGitRepository) and
572
isinstance(self.target, LocalGitRepository)):
573
return self.source._git.fetch(self.target._git, determine_wants,
575
elif (isinstance(self.source, LocalGitRepository) and
576
isinstance(self.target, RemoteGitRepository)):
577
raise NotImplementedError
578
elif (isinstance(self.source, RemoteGitRepository) and
579
isinstance(self.target, LocalGitRepository)):
580
f, commit = self.target._git.object_store.add_thin_pack()
582
refs = self.source._git.fetch_pack(determine_wants,
583
graphwalker, f.write, progress)
592
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
494
593
mapping=None, fetch_spec=None, branches=None):
495
594
if mapping is None:
496
595
mapping = self.source.get_mapping()
498
trace.info("git: %s", text)
499
596
r = self.target._git
500
597
if revision_id is not None:
501
598
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
507
604
determine_wants = r.object_store.determine_wants_all
509
606
determine_wants = lambda x: [y for y in args if not y in r.object_store]
607
return self.fetch_objects(determine_wants, mapping)
511
graphwalker = r.get_graph_walker()
512
f, commit = r.object_store.add_thin_pack()
514
refs = self.source.fetch_pack(determine_wants, graphwalker,
523
611
def is_compatible(source, target):
524
612
"""Be compatible with GitRepository."""
525
return (isinstance(source, GitRepository) and
613
return (isinstance(source, GitRepository) and
526
614
isinstance(target, GitRepository))