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, ui, urlutils
18
from bzrlib.errors import InvalidRevisionId
19
from bzrlib.inventory import Inventory
20
from bzrlib.repository import InterRepository
21
from bzrlib.trace import info
22
from bzrlib.tsort import topo_sort
24
from bzrlib.plugins.git import git
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 (
25
73
from bzrlib.plugins.git.repository import (
30
from bzrlib.plugins.git.remote import RemoteGitRepository
32
from dulwich.objects import Commit
34
from cStringIO import StringIO
37
class BzrFetchGraphWalker(object):
39
def __init__(self, repository, mapping):
40
self.repository = repository
41
self.mapping = mapping
43
self.heads = set(repository.all_revision_ids())
47
revid = self.mapping.revision_id_foreign_to_bzr(sha)
50
def remove(self, revid):
53
self.heads.remove(revid)
54
if revid in self.parents:
55
for p in self.parents[revid]:
60
ret = self.heads.pop()
61
ps = self.repository.get_parent_map([ret])[ret]
62
self.parents[ret] = ps
63
self.heads.update([p for p in ps if not p in self.done])
66
return self.mapping.revision_id_bzr_to_foreign(ret)
67
except InvalidRevisionId:
72
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
80
MAX_INV_CACHE_SIZE = 50 * 1024 * 1024
83
def import_git_blob(texts, mapping, path, hexsha, base_inv, base_inv_shamap,
84
base_ie, parent_id, revision_id, parent_invs, lookup_object,
73
86
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
88
:param texts: VersionedFiles to add to
76
89
:param path: Path in the tree
77
90
:param blob: A git blob
91
:return: Inventory delta for this file
79
93
file_id = mapping.generate_file_id(path)
80
text_revision = inv.revision_id
81
repo.texts.add_lines((file_id, text_revision),
82
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
83
osutils.split_lines(blob.data))
84
ie = inv.add_path(path, "file", file_id)
85
ie.revision = text_revision
86
ie.text_size = len(blob.data)
87
ie.text_sha1 = osutils.sha_string(blob.data)
98
# We just have to hope this is indeed utf-8:
99
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
100
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
101
# See if this has changed at all
106
base_sha = base_inv_shamap.lookup_blob(file_id,
107
revision_hint=base_ie.revision)
111
if (base_sha == hexsha and base_ie.executable == ie.executable
112
and base_ie.kind == ie.kind):
113
# If nothing has changed since the base revision, we're done
115
if base_sha == hexsha and base_ie.kind == ie.kind:
116
ie.text_size = base_ie.text_size
117
ie.text_sha1 = base_ie.text_sha1
118
ie.symlink_target = base_ie.symlink_target
119
if ie.executable == base_ie.executable:
120
ie.revision = base_ie.revision
122
blob = lookup_object(hexsha)
124
blob = lookup_object(hexsha)
125
if ie.kind == "symlink":
127
ie.symlink_target = blob.data
131
ie.text_size = len(blob.data)
132
ie.text_sha1 = osutils.sha_string(blob.data)
133
# Check what revision we should store
135
for pinv in parent_invs:
136
if pinv.revision_id == base_inv.revision_id:
145
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
146
# found a revision in one of the parents to use
147
ie.revision = pie.revision
149
parent_keys.append((file_id, pie.revision))
150
if ie.revision is None:
151
# Need to store a new revision
152
ie.revision = revision_id
153
assert file_id is not None
154
assert ie.revision is not None
155
if ie.kind == 'symlink':
159
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, data)])
160
shamap = { ie.file_id: hexsha }
162
if base_ie is not None:
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, []))
168
invdelta.append((old_path, path, file_id, ie))
169
return (invdelta, shamap)
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, 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):
198
deletable = [(osutils.pathjoin(path, k), v) for k,v in base_children.iteritems() if k not in existing_children]
200
(path, ie) = deletable.pop()
201
ret.append((path, None, ie.file_id, None))
202
if ie.kind == "directory":
203
for name, child_ie in ie.children.iteritems():
204
deletable.append((osutils.pathjoin(path, name), child_ie))
208
def import_git_tree(texts, mapping, path, hexsha, base_inv, base_inv_shamap,
209
base_ie, parent_id, revision_id, parent_invs, lookup_object,
210
allow_submodules=False):
92
211
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
213
:param texts: VersionedFiles object to add to
95
214
:param path: Path in the tree
96
215
:param tree: A git tree object
97
:param inv: Inventory object
216
:param base_inv: Base inventory against which to return inventory delta
217
:return: Inventory delta for this subtree
99
220
file_id = mapping.generate_file_id(path)
100
text_revision = inv.revision_id
101
repo.texts.add_lines((file_id, text_revision),
102
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
104
ie = inv.add_path(path, "directory", file_id)
105
ie.revision = text_revision
106
for mode, name, hexsha in tree.entries():
107
entry_kind = (mode & 0700000) / 0100000
221
# We just have to hope this is indeed utf-8:
222
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
225
# Newly appeared here
226
ie.revision = revision_id
227
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
228
invdelta.append((None, path, file_id, ie))
230
# See if this has changed at all
232
base_sha = base_inv_shamap.lookup_tree(file_id)
233
except (KeyError, NotImplementedError):
236
if base_sha == hexsha:
237
# If nothing has changed since the base revision, we're done
239
if base_ie.kind != "directory":
240
ie.revision = revision_id
241
texts.insert_record_stream([FulltextContentFactory((ie.file_id, ie.revision), (), None, "")])
242
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
243
if base_ie is not None and base_ie.kind == "directory":
244
base_children = base_ie.children
247
# Remember for next time
248
existing_children = set()
251
tree = lookup_object(hexsha)
252
for mode, name, child_hexsha in tree.entries():
108
253
basename = name.decode("utf-8")
112
child_path = urlutils.join(path, name)
114
tree = lookup_object(hexsha)
115
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
116
elif entry_kind == 1:
117
blob = lookup_object(hexsha)
118
fs_mode = mode & 0777
119
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
121
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
124
def import_git_objects(repo, mapping, object_iter, pb=None):
254
existing_children.add(basename)
255
child_path = osutils.pathjoin(path, name)
256
if stat.S_ISDIR(mode):
257
subinvdelta, grandchildmodes, subshamap = import_git_tree(
258
texts, mapping, child_path, child_hexsha, base_inv,
259
base_inv_shamap, base_children.get(basename), file_id,
260
revision_id, parent_invs, lookup_object,
261
allow_submodules=allow_submodules)
262
elif S_ISGITLINK(mode): # submodule
263
if not allow_submodules:
264
raise SubmodulesRequireSubtrees()
265
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
266
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
267
file_id, revision_id, parent_invs, lookup_object)
269
subinvdelta, subshamap = import_git_blob(texts, mapping,
270
child_path, child_hexsha, base_inv, base_inv_shamap,
271
base_children.get(basename), file_id,
272
revision_id, parent_invs, lookup_object,
273
mode_is_executable(mode), stat.S_ISLNK(mode))
275
child_modes.update(grandchildmodes)
276
invdelta.extend(subinvdelta)
277
shamap.update(subshamap)
278
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
279
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
280
child_modes[child_path] = mode
281
# Remove any children that have disappeared
282
if base_ie is not None and base_ie.kind == "directory":
283
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
284
base_children, existing_children))
285
shamap[file_id] = hexsha
286
return invdelta, child_modes, shamap
289
def approx_inv_size(inv):
290
# Very rough estimate, 1k per inventory entry
291
return len(inv) * 1024
294
def import_git_commit(repo, mapping, head, lookup_object,
295
target_git_object_retriever, parent_invs_cache):
296
o = lookup_object(head)
297
rev = mapping.import_commit(o)
298
# We have to do this here, since we have to walk the tree and
299
# we need to make sure to import the blobs / trees with the right
300
# path; this may involve adding them more than once.
302
for parent_id in rev.parent_ids:
304
parent_invs.append(parent_invs_cache[parent_id])
306
parent_inv = repo.get_inventory(parent_id)
307
parent_invs.append(parent_inv)
308
parent_invs_cache[parent_id] = parent_inv
309
if parent_invs == []:
310
base_inv = Inventory(root_id=None)
312
base_inv_shamap = None # Should never be accessed
314
base_inv = parent_invs[0]
315
base_ie = base_inv.root
316
base_inv_shamap = target_git_object_retriever._idmap.get_inventory_sha_map(base_inv.revision_id)
317
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
318
mapping, "", o.tree, base_inv, base_inv_shamap, base_ie, None,
319
rev.revision_id, parent_invs, lookup_object,
320
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
322
for (oldpath, newpath, fileid, new_ie) in inv_delta:
324
entries.append((fileid, None, None, None))
326
if new_ie.kind in ("file", "symlink"):
327
entries.append((fileid, "blob", shamap[fileid], new_ie.revision))
328
elif new_ie.kind == "directory":
329
entries.append((fileid, "tree", shamap[fileid], rev.revision_id))
332
target_git_object_retriever._idmap.add_entries(rev.revision_id,
333
rev.parent_ids, head, o.tree, entries)
334
if unusual_modes != {}:
335
for path, mode in unusual_modes.iteritems():
336
warn_unusual_mode(rev.foreign_revid, path, mode)
337
mapping.import_unusual_file_modes(rev, unusual_modes)
339
basis_id = rev.parent_ids[0]
341
basis_id = NULL_REVISION
343
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
344
inv_delta, rev.revision_id, rev.parent_ids,
346
parent_invs_cache[rev.revision_id] = inv
347
repo.add_revision(rev.revision_id, rev)
348
if "verify" in debug.debug_flags:
349
new_unusual_modes = mapping.export_unusual_file_modes(rev)
350
if new_unusual_modes != unusual_modes:
351
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
352
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
353
for newsha1, newobj, path in objs:
354
assert path is not None
358
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
359
if oldsha1 != newsha1:
360
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
363
def import_git_objects(repo, mapping, object_iter,
364
target_git_object_retriever, heads, pb=None, limit=None):
125
365
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
367
:param repo: Target Bazaar repository
128
368
:param mapping: Mapping to use
129
369
:param object_iter: Iterator over Git objects.
370
:return: Tuple with pack hints and last imported revision id
372
def lookup_object(sha):
374
return object_iter[sha]
376
return target_git_object_retriever[sha]
131
377
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
380
heads = list(set(heads))
381
parent_invs_cache = lru_cache.LRUSizeCache(compute_size=approx_inv_size,
382
max_size=MAX_INV_CACHE_SIZE)
383
target_git_object_retriever.start_write_group() # FIXME: try/finally
140
384
# Find and convert commit objects
141
for o in objects.itervalues():
387
pb.update("finding revisions to fetch", len(graph), None)
389
assert isinstance(head, str)
391
o = lookup_object(head)
142
394
if isinstance(o, Commit):
143
395
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
145
revisions[rev.revision_id] = rev
146
graph.append((rev.revision_id, rev.parent_ids))
396
if repo.has_revision(rev.revision_id):
398
squash_revision(repo, rev)
399
graph.append((o.id, o.parents))
400
heads.extend([p for p in o.parents if p not in checked])
401
elif isinstance(o, Tag):
402
if o.object[1] not in checked:
403
heads.append(o.object[1])
405
trace.warning("Unable to import head object %r" % o)
147
408
# Order the revisions
148
409
# Create the inventory objects
149
for i, revid in enumerate(topo_sort(graph)):
151
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
rev = revisions[revid]
154
# We have to do this here, since we have to walk the tree and
155
# we need to make sure to import the blobs / trees with the riht
156
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
def lookup_object(sha):
162
return reconstruct_git_object(repo, mapping, sha)
163
parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
164
import_git_tree(repo, mapping, "", root_tree, inv, parent_invs, lookup_object)
165
repo.add_revision(rev.revision_id, rev, inv)
168
def reconstruct_git_commit(repo, rev):
169
raise NotImplementedError(self.reconstruct_git_commit)
172
def reconstruct_git_object(repo, mapping, sha):
174
revid = mapping.revision_id_foreign_to_bzr(sha)
176
rev = repo.get_revision(revid)
177
except NoSuchRevision:
180
return reconstruct_git_commit(rev)
184
raise KeyError("No such object %s" % sha)
411
revision_ids = topo_sort(graph)
413
if limit is not None:
414
revision_ids = revision_ids[:limit]
416
for offset in range(0, len(revision_ids), batch_size):
417
repo.start_write_group()
419
for i, head in enumerate(revision_ids[offset:offset+batch_size]):
421
pb.update("fetching revisions", offset+i, len(revision_ids))
422
import_git_commit(repo, mapping, head, lookup_object,
423
target_git_object_retriever,
427
repo.abort_write_group()
430
hint = repo.commit_write_group()
432
pack_hints.extend(hint)
433
target_git_object_retriever.commit_write_group()
434
return pack_hints, last_imported
187
437
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
439
_matching_repo_format = GitRepositoryFormat()
192
442
def _get_repo_format_to_test():
196
446
"""See InterRepository.copy_content."""
197
447
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
449
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
450
mapping=None, fetch_spec=None):
451
self.fetch_refs(revision_id=revision_id, pb=pb,
452
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
455
class InterGitNonGitRepository(InterGitRepository):
456
"""Base InterRepository that copies revisions from a Git into a non-Git
459
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
460
mapping=None, fetch_spec=None):
201
461
if mapping is None:
202
462
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
463
if revision_id is not None:
464
interesting_heads = [revision_id]
465
elif fetch_spec is not None:
466
interesting_heads = fetch_spec.heads
468
interesting_heads = None
470
def determine_wants(refs):
472
if interesting_heads is None:
473
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
475
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
476
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
211
graph_walker = BzrFetchGraphWalker(self.target, mapping)
477
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
478
if pack_hint is not None and self.target._format.pack_compresses:
479
self.target.pack(hint=pack_hint)
480
if interesting_heads is not None:
481
present_interesting_heads = self.target.has_revisions(interesting_heads)
482
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
483
if missing_interesting_heads:
484
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
488
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
489
def report_git_progress(pb, text):
490
text = text.rstrip("\r\n")
491
g = _GIT_PROGRESS_RE.match(text)
493
(text, pct, current, total) = g.groups()
494
pb.update(text, int(current), int(total))
496
pb.update(text, 0, 0)
499
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
500
"""InterRepository that copies revisions from a remote Git into a non-Git
503
def get_target_heads(self):
504
# FIXME: This should be more efficient
505
all_revs = self.target.all_revision_ids()
506
parent_map = self.target.get_parent_map(all_revs)
508
map(all_parents.update, parent_map.itervalues())
509
return set(all_revs) - all_parents
511
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
513
report_git_progress(pb, text)
514
store = BazaarObjectStore(self.target, mapping)
515
self.target.lock_write()
517
heads = self.get_target_heads()
518
graph_walker = store.get_graph_walker(
519
[store._lookup_revision_sha1(head) for head in heads])
522
def record_determine_wants(heads):
523
wants = determine_wants(heads)
524
recorded_wants.extend(wants)
529
create_pb = pb = ui.ui_factory.nested_progress_bar()
531
objects_iter = self.source.fetch_objects(
532
record_determine_wants, graph_walker,
533
store.get_raw, progress)
534
return import_git_objects(self.target, mapping,
535
objects_iter, store, recorded_wants, pb, limit)
543
def is_compatible(source, target):
544
"""Be compatible with GitRepository."""
545
# FIXME: Also check target uses VersionedFile
546
return (isinstance(source, RemoteGitRepository) and
547
target.supports_rich_root() and
548
not isinstance(target, GitRepository))
551
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
552
"""InterRepository that copies revisions from a local Git into a non-Git
555
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
558
wants = determine_wants(self.source._git.get_refs())
214
561
create_pb = pb = ui.ui_factory.nested_progress_bar()
562
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
564
self.target.lock_write()
218
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
224
self.target.commit_write_group()
566
return import_git_objects(self.target, mapping,
567
self.source._git.object_store,
568
target_git_object_retriever, wants, pb, limit)
226
570
self.target.unlock()
added
removed
fetch.py
