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
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
154
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
158
if base_ie is not None:
159
old_path = base_inv.id2path(file_id)
160
if base_ie.kind == "directory":
161
invdelta.extend(remove_disappeared_children(old_path, base_ie.children, []))
164
invdelta.append((old_path, path, file_id, ie))
165
return (invdelta, shamap)
168
class SubmodulesRequireSubtrees(BzrError):
169
_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'."""
173
def import_git_submodule(texts, mapping, path, hexsha, base_inv, base_ie,
174
parent_id, revision_id, parent_invs, shagitmap, lookup_object):
175
file_id = mapping.generate_file_id(path)
176
ie = TreeReference(file_id, urlutils.basename(path.decode("utf-8")),
178
ie.revision = revision_id
183
if (base_ie.kind == ie.kind and
184
base_ie.reference_revision == ie.reference_revision):
185
ie.revision = base_ie.revision
186
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
187
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
188
invdelta = [(oldpath, path, file_id, ie)]
189
return invdelta, {}, {}
192
def remove_disappeared_children(path, base_children, existing_children):
194
deletable = [(osutils.pathjoin(path, k), v) for k,v in base_children.iteritems() if k not in existing_children]
196
(path, ie) = deletable.pop()
197
ret.append((path, None, ie.file_id, None))
198
if ie.kind == "directory":
199
for name, child_ie in ie.children.iteritems():
200
deletable.append((osutils.pathjoin(path, name), child_ie))
204
def import_git_tree(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
205
revision_id, parent_invs, shagitmap, lookup_object, allow_submodules=False):
82
206
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
208
:param texts: VersionedFiles object to add to
85
209
:param path: Path in the tree
86
210
:param tree: A git tree object
87
:param inv: Inventory object
211
:param base_inv: Base inventory against which to return inventory delta
212
:return: Inventory delta for this subtree
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
216
# We just have to hope this is indeed utf-8:
217
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
220
# Newly appeared here
221
ie.revision = revision_id
222
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
223
invdelta.append((None, path, file_id, ie))
225
# See if this has changed at all
227
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
231
if base_sha == hexsha:
232
# If nothing has changed since the base revision, we're done
234
if base_ie.kind != "directory":
235
ie.revision = revision_id
236
texts.insert_record_stream([FulltextContentFactory((ie.file_id, ie.revision), (), None, "")])
237
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
238
if base_ie is not None and base_ie.kind == "directory":
239
base_children = base_ie.children
242
# Remember for next time
243
existing_children = set()
246
tree = lookup_object(hexsha)
247
for mode, name, child_hexsha in tree.entries():
96
248
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):
249
existing_children.add(basename)
250
child_path = osutils.pathjoin(path, name)
251
if stat.S_ISDIR(mode):
252
subinvdelta, grandchildmodes, subshamap = import_git_tree(
253
texts, mapping, child_path, child_hexsha, base_inv,
254
base_children.get(basename), file_id, revision_id,
255
parent_invs, shagitmap, lookup_object,
256
allow_submodules=allow_submodules)
257
invdelta.extend(subinvdelta)
258
child_modes.update(grandchildmodes)
259
shamap.extend(subshamap)
260
elif S_ISGITLINK(mode): # submodule
261
if not allow_submodules:
262
raise SubmodulesRequireSubtrees()
263
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
264
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
265
file_id, revision_id, parent_invs, shagitmap, lookup_object)
266
invdelta.extend(subinvdelta)
267
child_modes.update(grandchildmodes)
268
shamap.extend(subshamap)
270
subinvdelta, subshamap = import_git_blob(texts, mapping,
271
child_path, child_hexsha, base_inv, base_children.get(basename), file_id,
272
revision_id, parent_invs, shagitmap, lookup_object,
273
mode_is_executable(mode), stat.S_ISLNK(mode))
274
invdelta.extend(subinvdelta)
275
shamap.extend(subshamap)
276
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
277
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
278
child_modes[child_path] = mode
279
# Remove any children that have disappeared
280
if base_ie is not None and base_ie.kind == "directory":
281
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
282
base_children, existing_children))
283
shamap.append((hexsha, "tree", (file_id, revision_id)))
284
return invdelta, child_modes, shamap
287
def approx_inv_size(inv):
288
# Very rough estimate, 1k per inventory entry
289
return len(inv) * 1024
292
def import_git_commit(repo, mapping, head, lookup_object,
293
target_git_object_retriever, parent_invs_cache):
294
o = lookup_object(head)
295
rev = mapping.import_commit(o)
296
# We have to do this here, since we have to walk the tree and
297
# we need to make sure to import the blobs / trees with the right
298
# path; this may involve adding them more than once.
300
for parent_id in rev.parent_ids:
302
parent_invs.append(parent_invs_cache[parent_id])
304
parent_inv = repo.get_inventory(parent_id)
305
parent_invs.append(parent_inv)
306
parent_invs_cache[parent_id] = parent_inv
307
if parent_invs == []:
308
base_inv = Inventory(root_id=None)
311
base_inv = parent_invs[0]
312
base_ie = base_inv.root
313
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
314
mapping, "", o.tree, base_inv, base_ie, None, rev.revision_id,
315
parent_invs, target_git_object_retriever._idmap, lookup_object,
316
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
317
target_git_object_retriever._idmap.add_entries(shamap)
318
if unusual_modes != {}:
319
for path, mode in unusual_modes.iteritems():
320
warn_unusual_mode(rev.foreign_revid, path, mode)
321
mapping.import_unusual_file_modes(rev, unusual_modes)
323
basis_id = rev.parent_ids[0]
325
basis_id = NULL_REVISION
327
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
328
inv_delta, rev.revision_id, rev.parent_ids,
330
parent_invs_cache[rev.revision_id] = inv
331
repo.add_revision(rev.revision_id, rev)
332
if "verify" in debug.debug_flags:
333
new_unusual_modes = mapping.export_unusual_file_modes(rev)
334
if new_unusual_modes != unusual_modes:
335
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
336
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
337
for sha1, newobj, path in objs:
338
assert path is not None
339
oldobj = tree_lookup_path(lookup_object, o.tree, path)
341
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
344
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
110
346
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
348
:param repo: Target Bazaar repository
113
349
:param mapping: Mapping to use
114
350
:param object_iter: Iterator over Git objects.
352
target_git_object_retriever._idmap.start_write_group() # FIXME: try/finally
353
def lookup_object(sha):
355
return object_iter[sha]
357
return target_git_object_retriever[sha]
116
358
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
362
parent_invs_cache = lru_cache.LRUSizeCache(compute_size=approx_inv_size,
363
max_size=MAX_INV_CACHE_SIZE)
121
364
# Find and convert commit objects
122
for o in objects.iterkeys():
367
pb.update("finding revisions to fetch", len(graph), None)
369
assert isinstance(head, str)
371
o = lookup_object(head)
373
trace.mutter('missing head %s', head)
123
375
if isinstance(o, Commit):
124
376
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
377
if repo.has_revision(rev.revision_id):
379
squash_revision(repo, rev)
380
graph.append((o.id, o.parents))
381
target_git_object_retriever._idmap.add_entry(o.id, "commit",
382
(rev.revision_id, o.tree))
383
heads.extend([p for p in o.parents if p not in checked])
384
elif isinstance(o, Tag):
385
heads.append(o.object[1])
387
trace.warning("Unable to import head object %r" % o)
390
# Order the revisions
126
391
# 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)
393
revision_ids = topo_sort(graph)
395
for offset in range(0, len(revision_ids), batch_size):
396
repo.start_write_group()
398
for i, head in enumerate(revision_ids[offset:offset+batch_size]):
400
pb.update("fetching revisions", offset+i, len(revision_ids))
401
import_git_commit(repo, mapping, head, lookup_object,
402
target_git_object_retriever,
405
repo.abort_write_group()
408
hint = repo.commit_write_group()
410
pack_hints.extend(hint)
411
target_git_object_retriever._idmap.commit_write_group()
159
415
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
417
_matching_repo_format = GitRepositoryFormat()
164
420
def _get_repo_format_to_test():
168
424
"""See InterRepository.copy_content."""
169
425
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
427
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
429
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
430
mapping=mapping, fetch_spec=fetch_spec)
433
class InterGitNonGitRepository(InterGitRepository):
434
"""Base InterRepository that copies revisions from a Git into a non-Git
437
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
438
mapping=None, fetch_spec=None):
173
439
if mapping is None:
174
440
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)]
441
if revision_id is not None:
442
interesting_heads = [revision_id]
443
elif fetch_spec is not None:
444
interesting_heads = fetch_spec.heads
446
interesting_heads = None
448
def determine_wants(refs):
450
if interesting_heads is None:
451
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
453
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
454
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)
455
pack_hint = self.fetch_objects(determine_wants, mapping, pb)
456
if pack_hint is not None and self.target._format.pack_compresses:
457
self.target.pack(hint=pack_hint)
458
if interesting_heads is not None:
459
present_interesting_heads = self.target.has_revisions(interesting_heads)
460
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
461
if missing_interesting_heads:
462
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
466
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
467
def report_git_progress(pb, text):
468
text = text.rstrip("\r\n")
469
g = _GIT_PROGRESS_RE.match(text)
471
(text, pct, current, total) = g.groups()
472
pb.update(text, int(current), int(total))
474
pb.update(text, 0, 0)
477
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
478
"""InterRepository that copies revisions from a remote Git into a non-Git
481
def get_target_heads(self):
482
# FIXME: This should be more efficient
483
all_revs = self.target.all_revision_ids()
484
parent_map = self.target.get_parent_map(all_revs)
486
map(all_parents.update, parent_map.itervalues())
487
return set(all_revs) - all_parents
489
def fetch_objects(self, determine_wants, mapping, pb=None):
491
report_git_progress(pb, text)
492
store = BazaarObjectStore(self.target, mapping)
187
493
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
495
heads = self.get_target_heads()
496
graph_walker = store.get_graph_walker(
497
[store._lookup_revision_sha1(head) for head in heads])
500
def record_determine_wants(heads):
501
wants = determine_wants(heads)
502
recorded_wants.extend(wants)
507
create_pb = pb = ui.ui_factory.nested_progress_bar()
509
objects_iter = self.source.fetch_objects(
510
record_determine_wants, graph_walker,
511
store.get_raw, progress)
512
return import_git_objects(self.target, mapping,
513
objects_iter, store, recorded_wants, pb)
193
518
self.target.unlock()
added
removed
fetch.py
