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 (
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 (
73
from bzrlib.plugins.git.repository import (
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):
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
92
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):
97
# We just have to hope this is indeed utf-8:
98
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
99
ie.executable = executable
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_inv_trees,
209
base_ie, parent_id, revision_id, parent_invs, shagitmap, lookup_object,
210
allow_submodules=False):
82
211
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
213
:param texts: VersionedFiles object to add to
85
214
:param path: Path in the tree
86
215
:param tree: A git tree object
87
:param inv: Inventory object
216
:param base_inv: Base inventory against which to return inventory delta
217
:return: Inventory delta for this subtree
89
220
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
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_trees[file_id]
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():
96
253
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):
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_trees, base_children.get(basename), file_id,
260
revision_id, parent_invs, shagitmap, lookup_object,
261
allow_submodules=allow_submodules)
262
invdelta.extend(subinvdelta)
263
child_modes.update(grandchildmodes)
264
shamap.extend(subshamap)
265
elif S_ISGITLINK(mode): # submodule
266
if not allow_submodules:
267
raise SubmodulesRequireSubtrees()
268
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
269
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
270
file_id, revision_id, parent_invs, shagitmap, lookup_object)
271
invdelta.extend(subinvdelta)
272
child_modes.update(grandchildmodes)
273
shamap.extend(subshamap)
275
subinvdelta, subshamap = import_git_blob(texts, mapping,
276
child_path, child_hexsha, base_inv, base_children.get(basename), file_id,
277
revision_id, parent_invs, shagitmap, lookup_object,
278
mode_is_executable(mode), stat.S_ISLNK(mode))
279
invdelta.extend(subinvdelta)
280
shamap.extend(subshamap)
281
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
282
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
283
child_modes[child_path] = mode
284
# Remove any children that have disappeared
285
if base_ie is not None and base_ie.kind == "directory":
286
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
287
base_children, existing_children))
288
shamap.append((hexsha, "tree", (file_id, revision_id)))
289
return invdelta, child_modes, shamap
292
def approx_inv_size(inv):
293
# Very rough estimate, 1k per inventory entry
294
return len(inv) * 1024
297
def import_git_commit(repo, mapping, head, lookup_object,
298
target_git_object_retriever, parent_invs_cache,
299
parent_inv_trees_cache):
300
o = lookup_object(head)
301
rev = mapping.import_commit(o)
302
# We have to do this here, since we have to walk the tree and
303
# we need to make sure to import the blobs / trees with the right
304
# path; this may involve adding them more than once.
306
for parent_id in rev.parent_ids:
308
parent_invs.append(parent_invs_cache[parent_id])
310
parent_inv = repo.get_inventory(parent_id)
311
parent_invs.append(parent_inv)
312
parent_invs_cache[parent_id] = parent_inv
313
if parent_invs == []:
314
base_inv = Inventory(root_id=None)
318
base_inv = parent_invs[0]
319
base_ie = base_inv.root
320
base_inv_trees = parent_inv_trees_cache.get(rev.parent_ids[0], {})
321
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
322
mapping, "", o.tree, base_inv, base_inv_trees, base_ie, None,
323
rev.revision_id, parent_invs, target_git_object_retriever._idmap,
325
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
326
parent_inv_trees_cache[rev.revision_id] = dict(base_inv_trees.iteritems())
327
parent_inv_trees_cache[rev.revision_id].update(dict(
328
((key[0], hexsha) for (hexsha, kind, key) in shamap if kind == "tree")))
329
target_git_object_retriever._idmap.add_entries(shamap)
330
if unusual_modes != {}:
331
for path, mode in unusual_modes.iteritems():
332
warn_unusual_mode(rev.foreign_revid, path, mode)
333
mapping.import_unusual_file_modes(rev, unusual_modes)
335
basis_id = rev.parent_ids[0]
337
basis_id = NULL_REVISION
339
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
340
inv_delta, rev.revision_id, rev.parent_ids,
342
parent_invs_cache[rev.revision_id] = inv
343
repo.add_revision(rev.revision_id, rev)
344
if "verify" in debug.debug_flags:
345
new_unusual_modes = mapping.export_unusual_file_modes(rev)
346
if new_unusual_modes != unusual_modes:
347
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
348
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
349
for newsha1, newobj, path in objs:
350
assert path is not None
354
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
355
if oldsha1 != newsha1:
356
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
359
def import_git_objects(repo, mapping, object_iter,
360
target_git_object_retriever, heads, pb=None, limit=None):
110
361
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
363
:param repo: Target Bazaar repository
113
364
:param mapping: Mapping to use
114
365
:param object_iter: Iterator over Git objects.
366
:return: Tuple with pack hints and last imported revision id
368
def lookup_object(sha):
370
return object_iter[sha]
372
return target_git_object_retriever[sha]
116
373
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
376
heads = list(set(heads))
377
parent_invs_cache = lru_cache.LRUSizeCache(compute_size=approx_inv_size,
378
max_size=MAX_INV_CACHE_SIZE)
379
parent_inv_trees_cache = lru_cache.LRUSizeCache(
380
compute_size=approx_inv_size, max_size=MAX_INV_CACHE_SIZE)
381
target_git_object_retriever.start_write_group() # FIXME: try/finally
121
382
# Find and convert commit objects
122
for o in objects.iterkeys():
385
pb.update("finding revisions to fetch", len(graph), None)
387
assert isinstance(head, str)
389
o = lookup_object(head)
123
392
if isinstance(o, Commit):
124
393
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
394
if repo.has_revision(rev.revision_id):
396
squash_revision(repo, rev)
397
graph.append((o.id, o.parents))
398
target_git_object_retriever._idmap.add_entry(o.id, "commit",
399
(rev.revision_id, o.tree))
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)
408
# Order the revisions
126
409
# 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)
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,
424
parent_invs_cache, parent_inv_trees_cache)
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
159
437
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
439
_matching_repo_format = GitRepositoryFormat()
164
442
def _get_repo_format_to_test():
168
446
"""See InterRepository.copy_content."""
169
447
self.fetch(revision_id, pb, find_ghosts=False)
171
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):
173
461
if mapping is None:
174
462
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)]
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("^{}")]
475
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
476
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)
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)
187
515
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
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)
193
540
self.target.unlock()
added
removed
fetch.py
