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_ie, parent_id,
209
revision_id, parent_invs, shagitmap, lookup_object, allow_submodules=False):
82
210
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
212
:param texts: VersionedFiles object to add to
85
213
:param path: Path in the tree
86
214
:param tree: A git tree object
87
:param inv: Inventory object
215
:param base_inv: Base inventory against which to return inventory delta
216
:return: Inventory delta for this subtree
89
219
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
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():
96
252
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):
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 sha1, newobj, path in objs:
342
assert path is not None
343
oldobj = tree_lookup_path(lookup_object, o.tree, path)
345
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
348
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
349
heads, pb=None, limit=None):
110
350
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
352
:param repo: Target Bazaar repository
113
353
:param mapping: Mapping to use
114
354
:param object_iter: Iterator over Git objects.
356
def lookup_object(sha):
358
return object_iter[sha]
360
return target_git_object_retriever[sha]
116
361
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
364
heads = list(set(heads))
365
parent_invs_cache = lru_cache.LRUSizeCache(compute_size=approx_inv_size,
366
max_size=MAX_INV_CACHE_SIZE)
367
target_git_object_retriever.start_write_group() # FIXME: try/finally
121
368
# Find and convert commit objects
122
for o in objects.iterkeys():
371
pb.update("finding revisions to fetch", len(graph), None)
373
assert isinstance(head, str)
375
o = object_iter[head]
123
378
if isinstance(o, Commit):
124
379
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
380
if repo.has_revision(rev.revision_id):
382
squash_revision(repo, rev)
383
graph.append((o.id, o.parents))
384
target_git_object_retriever._idmap.add_entry(o.id, "commit",
385
(rev.revision_id, o.tree))
386
heads.extend([p for p in o.parents if p not in checked])
387
elif isinstance(o, Tag):
388
if o.object[1] not in checked:
389
heads.append(o.object[1])
391
trace.warning("Unable to import head object %r" % o)
394
# Order the revisions
126
395
# 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)
397
revision_ids = topo_sort(graph)
399
if limit is not None:
400
revision_ids = revision_ids[:limit]
402
for offset in range(0, len(revision_ids), batch_size):
403
repo.start_write_group()
405
for i, head in enumerate(revision_ids[offset:offset+batch_size]):
407
pb.update("fetching revisions", offset+i, len(revision_ids))
408
import_git_commit(repo, mapping, head, lookup_object,
409
target_git_object_retriever,
413
repo.abort_write_group()
416
hint = repo.commit_write_group()
418
pack_hints.extend(hint)
419
target_git_object_retriever.commit_write_group()
420
return pack_hints, last_imported
159
423
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
425
_matching_repo_format = GitRepositoryFormat()
164
428
def _get_repo_format_to_test():
168
432
"""See InterRepository.copy_content."""
169
433
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
435
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
437
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
438
mapping=mapping, fetch_spec=fetch_spec)
441
class InterGitNonGitRepository(InterGitRepository):
442
"""Base InterRepository that copies revisions from a Git into a non-Git
445
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
446
mapping=None, fetch_spec=None):
173
447
if mapping is None:
174
448
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)]
449
if revision_id is not None:
450
interesting_heads = [revision_id]
451
elif fetch_spec is not None:
452
interesting_heads = fetch_spec.heads
454
interesting_heads = None
456
def determine_wants(refs):
458
if interesting_heads is None:
459
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
461
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
462
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)
463
pack_hint = self.fetch_objects(determine_wants, mapping, pb)[0]
464
if pack_hint is not None and self.target._format.pack_compresses:
465
self.target.pack(hint=pack_hint)
466
if interesting_heads is not None:
467
present_interesting_heads = self.target.has_revisions(interesting_heads)
468
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
469
if missing_interesting_heads:
470
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
474
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
475
def report_git_progress(pb, text):
476
text = text.rstrip("\r\n")
477
g = _GIT_PROGRESS_RE.match(text)
479
(text, pct, current, total) = g.groups()
480
pb.update(text, int(current), int(total))
482
pb.update(text, 0, 0)
485
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
486
"""InterRepository that copies revisions from a remote Git into a non-Git
489
def get_target_heads(self):
490
# FIXME: This should be more efficient
491
all_revs = self.target.all_revision_ids()
492
parent_map = self.target.get_parent_map(all_revs)
494
map(all_parents.update, parent_map.itervalues())
495
return set(all_revs) - all_parents
497
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
499
report_git_progress(pb, text)
500
store = BazaarObjectStore(self.target, mapping)
187
501
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
503
heads = self.get_target_heads()
504
graph_walker = store.get_graph_walker(
505
[store._lookup_revision_sha1(head) for head in heads])
508
def record_determine_wants(heads):
509
wants = determine_wants(heads)
510
recorded_wants.extend(wants)
515
create_pb = pb = ui.ui_factory.nested_progress_bar()
517
objects_iter = self.source.fetch_objects(
518
record_determine_wants, graph_walker,
519
store.get_raw, progress)
520
return import_git_objects(self.target, mapping,
521
objects_iter, store, recorded_wants, pb, limit)
193
526
self.target.unlock()
added
removed
fetch.py
