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_ie, parent_id,
84
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
73
85
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
87
:param texts: VersionedFiles to add to
76
88
:param path: Path in the tree
77
89
:param blob: A git blob
90
:return: Inventory delta for this file
79
92
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)
97
# We just have to hope this is indeed utf-8:
98
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
99
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
100
# See if this has changed at all
105
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
109
if (base_sha == hexsha and base_ie.executable == ie.executable
110
and base_ie.kind == ie.kind):
111
# If nothing has changed since the base revision, we're done
113
if base_sha == hexsha and base_ie.kind == ie.kind:
114
ie.text_size = base_ie.text_size
115
ie.text_sha1 = base_ie.text_sha1
116
ie.symlink_target = base_ie.symlink_target
117
if ie.executable == base_ie.executable:
118
ie.revision = base_ie.revision
120
blob = lookup_object(hexsha)
122
blob = lookup_object(hexsha)
123
if ie.kind == "symlink":
125
ie.symlink_target = blob.data
129
ie.text_size = len(blob.data)
130
ie.text_sha1 = osutils.sha_string(blob.data)
131
# Check what revision we should store
133
for pinv in parent_invs:
134
if pinv.revision_id == base_inv.revision_id:
143
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
144
# found a revision in one of the parents to use
145
ie.revision = pie.revision
147
parent_keys.append((file_id, pie.revision))
148
if ie.revision is None:
149
# Need to store a new revision
150
ie.revision = revision_id
151
assert file_id is not None
152
assert ie.revision is not None
153
if ie.kind == 'symlink':
157
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, data)])
158
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
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, 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):
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_ie, parent_id,
209
revision_id, parent_invs, shagitmap, lookup_object, allow_submodules=False):
92
210
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
212
:param texts: VersionedFiles object to add to
95
213
:param path: Path in the tree
96
214
:param tree: A git tree object
97
:param inv: Inventory object
215
:param base_inv: Base inventory against which to return inventory delta
216
:return: Inventory delta for this subtree
99
219
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
220
# We just have to hope this is indeed utf-8:
221
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
224
# Newly appeared here
225
ie.revision = revision_id
226
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
227
invdelta.append((None, path, file_id, ie))
229
# See if this has changed at all
231
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
235
if base_sha == hexsha:
236
# If nothing has changed since the base revision, we're done
238
if base_ie.kind != "directory":
239
ie.revision = revision_id
240
texts.insert_record_stream([FulltextContentFactory((ie.file_id, ie.revision), (), None, "")])
241
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
242
if base_ie is not None and base_ie.kind == "directory":
243
base_children = base_ie.children
246
# Remember for next time
247
existing_children = set()
250
tree = lookup_object(hexsha)
251
for mode, name, child_hexsha in tree.entries():
108
252
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):
253
existing_children.add(basename)
254
child_path = osutils.pathjoin(path, name)
255
if stat.S_ISDIR(mode):
256
subinvdelta, grandchildmodes, subshamap = import_git_tree(
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)
261
invdelta.extend(subinvdelta)
262
child_modes.update(grandchildmodes)
263
shamap.extend(subshamap)
264
elif S_ISGITLINK(mode): # submodule
265
if not allow_submodules:
266
raise SubmodulesRequireSubtrees()
267
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
268
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
269
file_id, revision_id, parent_invs, shagitmap, lookup_object)
270
invdelta.extend(subinvdelta)
271
child_modes.update(grandchildmodes)
272
shamap.extend(subshamap)
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,
277
mode_is_executable(mode), stat.S_ISLNK(mode))
278
invdelta.extend(subinvdelta)
279
shamap.extend(subshamap)
280
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
281
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
282
child_modes[child_path] = mode
283
# Remove any children that have disappeared
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))
287
shamap.append((hexsha, "tree", (file_id, revision_id)))
288
return invdelta, child_modes, shamap
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 newsha1, newobj, path in objs:
342
assert path is not None
346
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
347
if oldsha1 != newsha1:
348
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
351
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
352
heads, pb=None, limit=None):
125
353
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
355
:param repo: Target Bazaar repository
128
356
:param mapping: Mapping to use
129
357
:param object_iter: Iterator over Git objects.
359
def lookup_object(sha):
361
return object_iter[sha]
363
return target_git_object_retriever[sha]
131
364
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
367
heads = list(set(heads))
368
parent_invs_cache = lru_cache.LRUSizeCache(compute_size=approx_inv_size,
369
max_size=MAX_INV_CACHE_SIZE)
370
target_git_object_retriever.start_write_group() # FIXME: try/finally
140
371
# Find and convert commit objects
141
for o in objects.itervalues():
374
pb.update("finding revisions to fetch", len(graph), None)
376
assert isinstance(head, str)
378
o = object_iter[head]
142
381
if isinstance(o, Commit):
143
382
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))
383
if repo.has_revision(rev.revision_id):
385
squash_revision(repo, rev)
386
graph.append((o.id, o.parents))
387
target_git_object_retriever._idmap.add_entry(o.id, "commit",
388
(rev.revision_id, o.tree))
389
heads.extend([p for p in o.parents if p not in checked])
390
elif isinstance(o, Tag):
391
if o.object[1] not in checked:
392
heads.append(o.object[1])
394
trace.warning("Unable to import head object %r" % o)
147
397
# Order the revisions
148
398
# 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)
400
revision_ids = topo_sort(graph)
402
if limit is not None:
403
revision_ids = revision_ids[:limit]
405
for offset in range(0, len(revision_ids), batch_size):
406
repo.start_write_group()
408
for i, head in enumerate(revision_ids[offset:offset+batch_size]):
410
pb.update("fetching revisions", offset+i, len(revision_ids))
411
import_git_commit(repo, mapping, head, lookup_object,
412
target_git_object_retriever,
416
repo.abort_write_group()
419
hint = repo.commit_write_group()
421
pack_hints.extend(hint)
422
target_git_object_retriever.commit_write_group()
423
return pack_hints, last_imported
187
426
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
428
_matching_repo_format = GitRepositoryFormat()
192
431
def _get_repo_format_to_test():
196
435
"""See InterRepository.copy_content."""
197
436
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
438
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
440
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
441
mapping=mapping, fetch_spec=fetch_spec)
444
class InterGitNonGitRepository(InterGitRepository):
445
"""Base InterRepository that copies revisions from a Git into a non-Git
448
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
449
mapping=None, fetch_spec=None):
201
450
if mapping is None:
202
451
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
452
if revision_id is not None:
453
interesting_heads = [revision_id]
454
elif fetch_spec is not None:
455
interesting_heads = fetch_spec.heads
457
interesting_heads = None
459
def determine_wants(refs):
461
if interesting_heads is None:
462
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
464
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
465
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)
466
pack_hint = self.fetch_objects(determine_wants, mapping, pb)[0]
467
if pack_hint is not None and self.target._format.pack_compresses:
468
self.target.pack(hint=pack_hint)
469
if interesting_heads is not None:
470
present_interesting_heads = self.target.has_revisions(interesting_heads)
471
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
472
if missing_interesting_heads:
473
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
477
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
478
def report_git_progress(pb, text):
479
text = text.rstrip("\r\n")
480
g = _GIT_PROGRESS_RE.match(text)
482
(text, pct, current, total) = g.groups()
483
pb.update(text, int(current), int(total))
485
pb.update(text, 0, 0)
488
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
489
"""InterRepository that copies revisions from a remote Git into a non-Git
492
def get_target_heads(self):
493
# FIXME: This should be more efficient
494
all_revs = self.target.all_revision_ids()
495
parent_map = self.target.get_parent_map(all_revs)
497
map(all_parents.update, parent_map.itervalues())
498
return set(all_revs) - all_parents
500
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
502
report_git_progress(pb, text)
503
store = BazaarObjectStore(self.target, mapping)
504
self.target.lock_write()
506
heads = self.get_target_heads()
507
graph_walker = store.get_graph_walker(
508
[store._lookup_revision_sha1(head) for head in heads])
511
def record_determine_wants(heads):
512
wants = determine_wants(heads)
513
recorded_wants.extend(wants)
518
create_pb = pb = ui.ui_factory.nested_progress_bar()
520
objects_iter = self.source.fetch_objects(
521
record_determine_wants, graph_walker,
522
store.get_raw, progress)
523
return import_git_objects(self.target, mapping,
524
objects_iter, store, recorded_wants, pb, limit)
532
def is_compatible(source, target):
533
"""Be compatible with GitRepository."""
534
# FIXME: Also check target uses VersionedFile
535
return (isinstance(source, RemoteGitRepository) and
536
target.supports_rich_root() and
537
not isinstance(target, GitRepository))
540
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
541
"""InterRepository that copies revisions from a local Git into a non-Git
544
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
545
wants = determine_wants(self.source._git.get_refs())
214
548
create_pb = pb = ui.ui_factory.nested_progress_bar()
549
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
551
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()
553
return import_git_objects(self.target, mapping,
554
self.source._git.object_store, target_git_object_retriever,
226
557
self.target.unlock()
added
removed
fetch.py
