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 file_id is not None
142
assert ie.revision is not None
143
if ie.kind == 'symlink':
147
chunks = blob.chunked
148
except AttributeError: # older version of dulwich
150
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, chunks)])
151
shamap = { ie.file_id: hexsha }
153
if base_hexsha is not None:
154
old_path = path # Renames are not supported yet
155
if stat.S_ISDIR(base_mode):
156
invdelta.extend(remove_disappeared_children(base_inv, old_path, 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:
222
base_tree = lookup_object(base_hexsha)
223
if base_tree is None:
224
# Newly appeared here
225
ie.revision = revision_id
226
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
228
invdelta.append((None, path, file_id, ie))
229
elif type(base_tree) is not Tree:
230
ie.revision = revision_id
231
old_path = path # Renames aren't supported yet
232
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
233
invdelta.append((old_path, path, ie.file_id, ie))
235
old_path = path # Renames aren't supported yet
236
# Remember for next time
237
existing_children = set()
240
for child_mode, name, child_hexsha in tree.entries():
241
existing_children.add(name)
242
child_path = posixpath.join(path, name)
243
if type(base_tree) is Tree:
245
child_base_mode, child_base_hexsha = base_tree[name]
247
child_base_hexsha = None
250
child_base_hexsha = None
252
if stat.S_ISDIR(child_mode):
253
subinvdelta, grandchildmodes, subshamap = import_git_tree(
254
texts, mapping, child_path, name,
255
(child_base_hexsha, child_hexsha),
256
base_inv, base_inv_shamap,
257
file_id, revision_id, parent_invs, lookup_object,
258
(child_base_mode, child_mode),
259
allow_submodules=allow_submodules)
260
elif S_ISGITLINK(child_mode): # submodule
261
if not allow_submodules:
262
raise SubmodulesRequireSubtrees()
263
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
264
texts, mapping, child_path, name,
265
(child_base_hexsha, child_hexsha),
266
base_inv, file_id, revision_id, parent_invs, lookup_object,
267
(child_base_mode, child_mode))
269
subinvdelta, subshamap = import_git_blob(texts, mapping,
270
child_path, name, (child_base_hexsha, child_hexsha),
271
base_inv, base_inv_shamap,
273
revision_id, parent_invs, lookup_object,
274
(child_base_mode, child_mode))
276
child_modes.update(grandchildmodes)
277
invdelta.extend(subinvdelta)
278
shamap.update(subshamap)
279
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
280
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
281
child_modes[child_path] = child_mode
282
# Remove any children that have disappeared
283
if base_tree is not None and type(base_tree) is Tree:
284
invdelta.extend(remove_disappeared_children(base_inv, old_path,
285
base_tree, existing_children, lookup_object))
286
shamap[file_id] = hexsha
287
return invdelta, child_modes, shamap
290
def import_git_commit(repo, mapping, head, lookup_object,
291
target_git_object_retriever, parent_invs_cache):
292
o = lookup_object(head)
293
rev = mapping.import_commit(o)
294
# We have to do this here, since we have to walk the tree and
295
# we need to make sure to import the blobs / trees with the right
296
# path; this may involve adding them more than once.
297
parent_invs = parent_invs_cache.get_inventories(rev.parent_ids)
298
if parent_invs == []:
299
base_inv = Inventory(root_id=None)
300
base_inv_shamap = None # Should never be accessed
304
base_inv = parent_invs[0]
305
base_inv_shamap = target_git_object_retriever._idmap.get_inventory_sha_map(base_inv.revision_id)
306
base_tree = lookup_object(o.parents[0]).tree
307
base_mode = stat.S_IFDIR
308
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
309
mapping, "", u"", (base_tree, o.tree), base_inv, base_inv_shamap,
310
None, rev.revision_id, parent_invs, lookup_object,
311
(base_mode, stat.S_IFDIR),
312
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
314
for (oldpath, newpath, fileid, new_ie) in inv_delta:
316
entries.append((fileid, None, None, None))
318
if new_ie.kind in ("file", "symlink"):
319
entries.append((fileid, "blob", shamap[fileid], new_ie.revision))
320
elif new_ie.kind == "directory":
321
entries.append((fileid, "tree", shamap[fileid], rev.revision_id))
324
target_git_object_retriever._idmap.add_entries(rev.revision_id,
325
rev.parent_ids, head, o.tree, entries)
326
if unusual_modes != {}:
327
for path, mode in unusual_modes.iteritems():
328
warn_unusual_mode(rev.foreign_revid, path, mode)
329
mapping.import_unusual_file_modes(rev, unusual_modes)
331
basis_id = rev.parent_ids[0]
333
basis_id = NULL_REVISION
335
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
336
inv_delta, rev.revision_id, rev.parent_ids,
338
parent_invs_cache.add(rev.revision_id, inv)
339
repo.add_revision(rev.revision_id, rev)
340
if "verify" in debug.debug_flags:
341
new_unusual_modes = mapping.export_unusual_file_modes(rev)
342
if new_unusual_modes != unusual_modes:
343
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
344
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
345
for newsha1, newobj, path in objs:
346
assert path is not None
350
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
351
if oldsha1 != newsha1:
352
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
355
def import_git_objects(repo, mapping, object_iter,
356
target_git_object_retriever, heads, pb=None, limit=None):
110
357
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
359
:param repo: Target Bazaar repository
113
360
:param mapping: Mapping to use
114
361
:param object_iter: Iterator over Git objects.
362
:return: Tuple with pack hints and last imported revision id
364
def lookup_object(sha):
366
return object_iter[sha]
368
return target_git_object_retriever[sha]
116
369
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
372
heads = list(set(heads))
373
parent_invs_cache = LRUInventoryCache(repo)
374
target_git_object_retriever.start_write_group() # FIXME: try/finally
121
375
# Find and convert commit objects
122
for o in objects.iterkeys():
378
pb.update("finding revisions to fetch", len(graph), None)
380
assert isinstance(head, str)
382
o = lookup_object(head)
123
385
if isinstance(o, Commit):
124
386
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
387
if repo.has_revision(rev.revision_id):
389
squash_revision(repo, rev)
390
graph.append((o.id, o.parents))
391
heads.extend([p for p in o.parents if p not in checked])
392
elif isinstance(o, Tag):
393
if o.object[1] not in checked:
394
heads.append(o.object[1])
396
trace.warning("Unable to import head object %r" % o)
399
# Order the revisions
126
400
# 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)
402
revision_ids = topo_sort(graph)
404
if limit is not None:
405
revision_ids = revision_ids[:limit]
407
for offset in range(0, len(revision_ids), batch_size):
408
repo.start_write_group()
410
for i, head in enumerate(revision_ids[offset:offset+batch_size]):
412
pb.update("fetching revisions", offset+i, len(revision_ids))
413
import_git_commit(repo, mapping, head, lookup_object,
414
target_git_object_retriever,
418
repo.abort_write_group()
421
hint = repo.commit_write_group()
423
pack_hints.extend(hint)
424
target_git_object_retriever.commit_write_group()
425
return pack_hints, last_imported
159
428
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
430
_matching_repo_format = GitRepositoryFormat()
164
433
def _get_repo_format_to_test():
168
437
"""See InterRepository.copy_content."""
169
438
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
440
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
441
mapping=None, fetch_spec=None):
442
self.fetch_refs(revision_id=revision_id, pb=pb,
443
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
446
class InterGitNonGitRepository(InterGitRepository):
447
"""Base InterRepository that copies revisions from a Git into a non-Git
450
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
451
mapping=None, fetch_spec=None):
173
452
if mapping is None:
174
453
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)]
454
if revision_id is not None:
455
interesting_heads = [revision_id]
456
elif fetch_spec is not None:
457
interesting_heads = fetch_spec.heads
459
interesting_heads = None
461
def determine_wants(refs):
463
if interesting_heads is None:
464
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
466
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
467
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)
468
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
469
if pack_hint is not None and self.target._format.pack_compresses:
470
self.target.pack(hint=pack_hint)
471
if interesting_heads is not None:
472
present_interesting_heads = self.target.has_revisions(interesting_heads)
473
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
474
if missing_interesting_heads:
475
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
479
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
480
def report_git_progress(pb, text):
481
text = text.rstrip("\r\n")
482
g = _GIT_PROGRESS_RE.match(text)
484
(text, pct, current, total) = g.groups()
485
pb.update(text, int(current), int(total))
487
pb.update(text, 0, 0)
490
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
491
"""InterRepository that copies revisions from a remote Git into a non-Git
494
def get_target_heads(self):
495
# FIXME: This should be more efficient
496
all_revs = self.target.all_revision_ids()
497
parent_map = self.target.get_parent_map(all_revs)
499
map(all_parents.update, parent_map.itervalues())
500
return set(all_revs) - all_parents
502
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
504
report_git_progress(pb, text)
505
store = BazaarObjectStore(self.target, mapping)
187
506
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
508
heads = self.get_target_heads()
509
graph_walker = store.get_graph_walker(
510
[store._lookup_revision_sha1(head) for head in heads])
513
def record_determine_wants(heads):
514
wants = determine_wants(heads)
515
recorded_wants.extend(wants)
520
create_pb = pb = ui.ui_factory.nested_progress_bar()
522
objects_iter = self.source.fetch_objects(
523
record_determine_wants, graph_walker,
524
store.get_raw, progress)
525
return import_git_objects(self.target, mapping,
526
objects_iter, store, recorded_wants, pb, limit)
193
531
self.target.unlock()
added
removed
fetch.py
