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
17
from cStringIO import (
21
from dulwich.client import (
22
SimpleFetchGraphWalker,
24
from dulwich.objects import (
28
from dulwich.object_store import (
40
from bzrlib.errors import (
45
from bzrlib.inventory import (
51
from bzrlib.lru_cache import (
54
from bzrlib.repository import (
57
from bzrlib.revision import (
60
from bzrlib.tsort import (
63
from bzrlib.versionedfile import (
64
FulltextContentFactory,
67
from bzrlib.plugins.git.converter import (
70
from bzrlib.plugins.git.mapping import (
72
inventory_to_tree_and_blobs,
75
from bzrlib.plugins.git.remote import (
78
from bzrlib.plugins.git.repository import (
32
85
class BzrFetchGraphWalker(object):
86
"""GraphWalker implementation that uses a Bazaar repository."""
34
88
def __init__(self, repository, mapping):
35
89
self.repository = repository
58
115
self.heads.update([p for p in ps if not p in self.done])
60
117
self.done.add(ret)
61
return self.mapping.revision_id_bzr_to_foreign(ret)
118
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
62
119
except InvalidRevisionId:
67
def import_git_blob(repo, mapping, path, blob):
124
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
125
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
68
126
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
128
:param texts: VersionedFiles to add to
71
129
:param path: Path in the tree
72
130
:param blob: A git blob
131
:return: Inventory delta for this file
74
133
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):
138
# We just have to hope this is indeed utf-8:
139
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
140
ie.executable = executable
142
# See if this has changed at all
144
base_ie = base_inv[file_id]
149
base_sha = base_ie.text_id
152
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
156
if (base_sha == hexsha and base_ie.executable == ie.executable
157
and base_ie.kind == ie.kind):
158
# If nothing has changed since the base revision, we're done
160
if base_sha == hexsha:
161
ie.text_size = base_ie.text_size
162
ie.text_sha1 = base_ie.text_sha1
163
ie.symlink_target = base_ie.symlink_target
164
ie.revision = base_ie.revision
166
blob = lookup_object(hexsha)
167
if ie.kind == "symlink":
168
ie.symlink_target = blob.data
172
ie.text_size = len(blob.data)
173
ie.text_sha1 = osutils.sha_string(blob.data)
174
# Check what revision we should store
176
for pinv in parent_invs:
177
if pinv.revision_id == base_inv.revision_id:
186
if pie.text_sha1 == ie.text_sha1:
187
# found a revision in one of the parents to use
188
ie.revision = pie.revision
190
parent_keys.append((file_id, pie.revision))
191
if ie.revision is None:
192
# Need to store a new revision
193
ie.revision = revision_id
194
assert file_id is not None
195
assert ie.revision is not None
196
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
197
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
200
if file_id in base_inv:
201
old_path = base_inv.id2path(file_id)
204
return ([(old_path, path, file_id, ie)], shamap)
207
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
208
revision_id, parent_invs, shagitmap, lookup_object):
82
209
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
211
:param texts: VersionedFiles object to add to
85
212
:param path: Path in the tree
86
213
:param tree: A git tree object
87
:param inv: Inventory object
214
:param base_inv: Base inventory against which to return inventory delta
215
:return: Inventory delta for this subtree
89
218
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
219
# We just have to hope this is indeed utf-8:
220
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
224
base_ie = base_inv[file_id]
226
# Newly appeared here
228
ie.revision = revision_id
229
texts.add_lines((file_id, ie.revision), (), [])
230
invdelta.append((None, path, file_id, ie))
232
base_sha = base_ie.text_id
233
# See if this has changed at all
236
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
240
if base_sha == hexsha:
241
# If nothing has changed since the base revision, we're done
243
# Remember for next time
244
existing_children = set()
247
tree = lookup_object(hexsha)
248
for mode, name, child_hexsha in tree.entries():
96
249
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):
250
existing_children.add(basename)
251
child_path = osutils.pathjoin(path, name)
252
if stat.S_ISDIR(mode):
253
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
254
mapping, child_path, child_hexsha, base_inv, file_id,
255
revision_id, parent_invs, shagitmap, lookup_object)
256
invdelta.extend(subinvdelta)
257
child_modes.update(grandchildmodes)
258
shamap.extend(subshamap)
260
fs_mode = stat.S_IMODE(mode)
261
symlink = stat.S_ISLNK(mode)
262
subinvdelta, subshamap = import_git_blob(texts, mapping,
263
child_path, child_hexsha, base_inv, file_id, revision_id,
264
parent_invs, shagitmap, lookup_object,
265
bool(fs_mode & 0111), symlink)
266
invdelta.extend(subinvdelta)
267
shamap.extend(subshamap)
268
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
269
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
270
child_modes[child_path] = mode
271
# Remove any children that have disappeared
272
if base_ie is not None:
273
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
276
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
277
if ie.kind == "directory":
278
deletable.extend(ie.children.values())
279
shamap.append((hexsha, "tree", (file_id, revision_id)))
280
return invdelta, child_modes, shamap
283
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
110
285
"""Import a set of git objects into a bzr repository.
112
287
:param repo: Bazaar repository
114
289
:param object_iter: Iterator over Git objects.
116
291
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
297
parent_invs_cache = LRUCache(50)
121
298
# Find and convert commit objects
122
for o in objects.iterkeys():
301
pb.update("finding revisions to fetch", len(graph), None)
303
assert isinstance(head, str)
305
o = object_iter[head]
123
308
if isinstance(o, Commit):
124
309
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
310
if repo.has_revision(rev.revision_id):
312
root_trees[rev.revision_id] = o.tree
313
revisions[rev.revision_id] = rev
314
graph.append((rev.revision_id, rev.parent_ids))
315
target_git_object_retriever._idmap.add_entry(o.id, "commit",
316
(rev.revision_id, o.tree))
317
heads.extend([p for p in o.parents if p not in checked])
318
elif isinstance(o, Tag):
319
heads.append(o.object[1])
321
trace.warning("Unable to import head object %r" % o)
323
# Order the revisions
126
324
# Create the inventory objects
127
for rev, root_tree in root_trees.iteritems():
325
for i, revid in enumerate(topo_sort(graph)):
327
pb.update("fetching revisions", i, len(graph))
328
rev = revisions[revid]
128
329
# 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
330
# we need to make sure to import the blobs / trees with the right
130
331
# path; this may involve adding them more than once.
132
332
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)
159
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
334
return object_iter[sha]
336
return target_git_object_retriever[sha]
338
for parent_id in rev.parent_ids:
340
parent_invs.append(parent_invs_cache[parent_id])
342
parent_inv = repo.get_inventory(parent_id)
343
parent_invs.append(parent_inv)
344
parent_invs_cache[parent_id] = parent_inv
345
if parent_invs == []:
346
base_inv = Inventory(root_id=None)
348
base_inv = parent_invs[0]
349
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
350
mapping, "", root_trees[revid], base_inv, None, revid,
351
parent_invs, target_git_object_retriever._idmap, lookup_object)
352
target_git_object_retriever._idmap.add_entries(shamap)
353
if unusual_modes != {}:
354
ret = "unusual modes: \n"
355
for item in unusual_modes.iteritems():
356
ret += "\t%s: %o\n" % item
357
raise AssertionError(ret)
359
basis_id = rev.parent_ids[0]
361
basis_id = NULL_REVISION
362
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
363
inv_delta, rev.revision_id, rev.parent_ids)
364
parent_invs_cache[rev.revision_id] = inv
365
repo.add_revision(rev.revision_id, rev)
366
if "verify" in debug.debug_flags:
367
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
368
for sha1, newobj, path in objs:
369
assert path is not None
370
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
371
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
373
target_git_object_retriever._idmap.commit()
376
class InterGitNonGitRepository(InterRepository):
377
"""Base InterRepository that copies revisions from a Git into a non-Git
380
_matching_repo_format = GitRepositoryFormat()
164
383
def _get_repo_format_to_test():
168
387
"""See InterRepository.copy_content."""
169
388
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
390
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
392
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
393
mapping=mapping, fetch_spec=fetch_spec)
395
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
396
mapping=None, fetch_spec=None):
173
397
if mapping is None:
174
398
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)]
399
if revision_id is not None:
400
interesting_heads = [revision_id]
401
elif fetch_spec is not None:
402
interesting_heads = fetch_spec.heads
404
interesting_heads = None
406
def determine_wants(refs):
408
if interesting_heads is None:
409
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
411
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid != NULL_REVISION]
185
412
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
413
self.fetch_objects(determine_wants, mapping, pb)
418
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
419
"""InterRepository that copies revisions from a remote Git into a non-Git
422
def fetch_objects(self, determine_wants, mapping, pb=None):
424
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
186
425
graph_walker = BzrFetchGraphWalker(self.target, mapping)
187
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
428
create_pb = pb = ui.ui_factory.nested_progress_bar()
429
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
432
def record_determine_wants(heads):
433
wants = determine_wants(heads)
434
recorded_wants.extend(wants)
438
self.target.lock_write()
440
self.target.start_write_group()
442
objects_iter = self.source.fetch_objects(
443
record_determine_wants,
445
target_git_object_retriever.get_raw,
447
import_git_objects(self.target, mapping, objects_iter,
448
target_git_object_retriever, recorded_wants, pb)
450
self.target.commit_write_group()
458
def is_compatible(source, target):
459
"""Be compatible with GitRepository."""
460
# FIXME: Also check target uses VersionedFile
461
return (isinstance(source, RemoteGitRepository) and
462
target.supports_rich_root() and
463
not isinstance(target, GitRepository))
466
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
467
"""InterRepository that copies revisions from a remote Git into a non-Git
470
def fetch_objects(self, determine_wants, mapping, pb=None):
471
wants = determine_wants(self.source._git.get_refs())
474
create_pb = pb = ui.ui_factory.nested_progress_bar()
475
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
477
self.target.lock_write()
479
self.target.start_write_group()
481
import_git_objects(self.target, mapping,
482
self.source._git.object_store,
483
target_git_object_retriever, wants, pb)
485
self.target.commit_write_group()
196
493
def is_compatible(source, target):
197
494
"""Be compatible with GitRepository."""
198
495
# FIXME: Also check target uses VersionedFile
199
496
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
497
target.supports_rich_root() and
498
not isinstance(target, GitRepository))
501
class InterGitRepository(InterRepository):
502
"""InterRepository that copies between Git repositories."""
504
_matching_repo_format = GitRepositoryFormat()
507
def _get_repo_format_to_test():
510
def copy_content(self, revision_id=None, pb=None):
511
"""See InterRepository.copy_content."""
512
self.fetch(revision_id, pb, find_ghosts=False)
514
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
515
mapping=None, fetch_spec=None):
517
mapping = self.source.get_mapping()
519
trace.info("git: %s", text)
521
if revision_id is not None:
522
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
523
elif fetch_spec is not None:
524
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
525
if fetch_spec is None and revision_id is None:
526
determine_wants = r.object_store.determine_wants_all
528
determine_wants = lambda x: [y for y in args if not y in r.object_store]
530
graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
531
f, commit = r.object_store.add_thin_pack()
533
self.source.fetch_pack(determine_wants, graphwalker, f.write, progress)
540
def is_compatible(source, target):
541
"""Be compatible with GitRepository."""
542
return (isinstance(source, GitRepository) and
543
isinstance(target, GitRepository))