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 (
64
from bzrlib.plugins.git.converter import (
67
from bzrlib.plugins.git.mapping import (
69
inventory_to_tree_and_blobs,
72
from bzrlib.plugins.git.repository import (
77
from bzrlib.plugins.git.remote import (
32
82
class BzrFetchGraphWalker(object):
83
"""GraphWalker implementation that uses a Bazaar repository."""
34
85
def __init__(self, repository, mapping):
35
86
self.repository = repository
58
112
self.heads.update([p for p in ps if not p in self.done])
60
114
self.done.add(ret)
61
return self.mapping.revision_id_bzr_to_foreign(ret)
115
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
62
116
except InvalidRevisionId:
67
def import_git_blob(repo, mapping, path, blob):
121
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
122
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
68
123
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
125
:param texts: VersionedFiles to add to
71
126
:param path: Path in the tree
72
127
:param blob: A git blob
128
:return: Inventory delta for this file
74
130
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):
135
# We just have to hope this is indeed utf-8:
136
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
137
ie.executable = executable
139
# See if this has changed at all
141
base_ie = base_inv[file_id]
146
base_sha = base_ie.text_id
149
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
153
if (base_sha == hexsha and base_ie.executable == ie.executable
154
and base_ie.kind == ie.kind):
155
# If nothing has changed since the base revision, we're done
157
if base_sha == hexsha:
158
ie.text_size = base_ie.text_size
159
ie.text_sha1 = base_ie.text_sha1
160
ie.symlink_target = base_ie.symlink_target
161
ie.revision = base_ie.revision
163
blob = lookup_object(hexsha)
164
if ie.kind == "symlink":
165
ie.symlink_target = blob.data
169
ie.text_size = len(blob.data)
170
ie.text_sha1 = osutils.sha_string(blob.data)
171
# Check what revision we should store
173
for pinv in parent_invs:
174
if pinv.revision_id == base_inv.revision_id:
183
if pie.text_sha1 == ie.text_sha1:
184
# found a revision in one of the parents to use
185
ie.revision = pie.revision
187
parent_keys.append((file_id, pie.revision))
188
if ie.revision is None:
189
# Need to store a new revision
190
ie.revision = revision_id
191
assert file_id is not None
192
assert ie.revision is not None
193
texts.add_lines((file_id, ie.revision), tuple(parent_keys),
194
osutils.split_lines(blob.data))
195
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
198
if file_id in base_inv:
199
old_path = base_inv.id2path(file_id)
202
return ([(old_path, path, file_id, ie)], shamap)
205
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
206
revision_id, parent_invs, shagitmap, lookup_object):
82
207
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
209
:param texts: VersionedFiles object to add to
85
210
:param path: Path in the tree
86
211
:param tree: A git tree object
87
:param inv: Inventory object
212
:param base_inv: Base inventory against which to return inventory delta
213
:return: Inventory delta for this subtree
89
216
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
217
# We just have to hope this is indeed utf-8:
218
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
222
base_ie = base_inv[file_id]
224
# Newly appeared here
226
ie.revision = revision_id
227
texts.add_lines((file_id, ie.revision), (), [])
228
invdelta.append((None, path, file_id, ie))
230
base_sha = base_ie.text_id
231
# See if this has changed at all
234
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
238
if base_sha == hexsha:
239
# If nothing has changed since the base revision, we're done
241
# Remember for next time
242
existing_children = set()
245
tree = lookup_object(hexsha)
246
for mode, name, child_hexsha in tree.entries():
96
247
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):
248
existing_children.add(basename)
249
child_path = osutils.pathjoin(path, name)
250
if stat.S_ISDIR(mode):
251
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
252
mapping, child_path, child_hexsha, base_inv, file_id,
253
revision_id, parent_invs, shagitmap, lookup_object)
254
invdelta.extend(subinvdelta)
255
child_modes.update(grandchildmodes)
256
shamap.extend(subshamap)
258
fs_mode = stat.S_IMODE(mode)
259
symlink = stat.S_ISLNK(mode)
260
subinvdelta, subshamap = import_git_blob(texts, mapping,
261
child_path, child_hexsha, base_inv, file_id, revision_id,
262
parent_invs, shagitmap, lookup_object,
263
bool(fs_mode & 0111), symlink)
264
invdelta.extend(subinvdelta)
265
shamap.extend(subshamap)
266
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
267
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
268
child_modes[child_path] = mode
269
# Remove any children that have disappeared
270
if base_ie is not None:
271
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
274
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
275
if ie.kind == "directory":
276
deletable.extend(ie.children.values())
277
shamap.append((hexsha, "tree", (file_id, revision_id)))
278
return invdelta, child_modes, shamap
281
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
110
283
"""Import a set of git objects into a bzr repository.
112
285
:param repo: Bazaar repository
114
287
:param object_iter: Iterator over Git objects.
116
289
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
295
parent_invs_cache = LRUCache(50)
121
296
# Find and convert commit objects
122
for o in objects.iterkeys():
299
pb.update("finding revisions to fetch", len(graph), None)
301
assert isinstance(head, str)
303
o = object_iter[head]
123
306
if isinstance(o, Commit):
124
307
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
308
if repo.has_revision(rev.revision_id):
310
root_trees[rev.revision_id] = o.tree
311
revisions[rev.revision_id] = rev
312
graph.append((rev.revision_id, rev.parent_ids))
313
target_git_object_retriever._idmap.add_entry(o.id, "commit",
314
(rev.revision_id, o.tree))
315
heads.extend([p for p in o.parents if p not in checked])
316
elif isinstance(o, Tag):
317
heads.append(o.object[1])
319
trace.warning("Unable to import head object %r" % o)
321
# Order the revisions
126
322
# Create the inventory objects
127
for rev, root_tree in root_trees.iteritems():
323
for i, revid in enumerate(topo_sort(graph)):
325
pb.update("fetching revisions", i, len(graph))
326
rev = revisions[revid]
128
327
# 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
328
# we need to make sure to import the blobs / trees with the right
130
329
# path; this may involve adding them more than once.
132
330
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()
332
return object_iter[sha]
334
return target_git_object_retriever[sha]
336
for parent_id in rev.parent_ids:
338
parent_invs.append(parent_invs_cache[parent_id])
340
parent_inv = repo.get_inventory(parent_id)
341
parent_invs.append(parent_inv)
342
parent_invs_cache[parent_id] = parent_inv
343
if parent_invs == []:
344
base_inv = Inventory(root_id=None)
346
base_inv = parent_invs[0]
347
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
348
mapping, "", root_trees[revid], base_inv, None, revid,
349
parent_invs, target_git_object_retriever._idmap, lookup_object)
350
target_git_object_retriever._idmap.add_entries(shamap)
351
if unusual_modes != {}:
352
ret = "unusual modes: \n"
353
for item in unusual_modes.iteritems():
354
ret += "\t%s: %o\n" % item
355
raise AssertionError(ret)
357
basis_id = rev.parent_ids[0]
359
basis_id = NULL_REVISION
360
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
361
inv_delta, rev.revision_id, rev.parent_ids)
362
parent_invs_cache[rev.revision_id] = inv
363
repo.add_revision(rev.revision_id, rev)
364
if "verify" in debug.debug_flags:
365
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
366
for sha1, newobj, path in objs:
367
assert path is not None
368
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
369
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
371
target_git_object_retriever._idmap.commit()
374
class InterGitNonGitRepository(InterRepository):
375
"""Base InterRepository that copies revisions from a Git into a non-Git
378
_matching_repo_format = GitRepositoryFormat()
164
381
def _get_repo_format_to_test():
168
385
"""See InterRepository.copy_content."""
169
386
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
388
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
390
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
391
mapping=mapping, fetch_spec=fetch_spec)
393
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
394
mapping=None, fetch_spec=None):
173
395
if mapping is None:
174
396
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)]
397
if revision_id is not None:
398
interesting_heads = [revision_id]
399
elif fetch_spec is not None:
400
interesting_heads = fetch_spec.heads
402
interesting_heads = None
404
def determine_wants(refs):
406
if interesting_heads is None:
407
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
409
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid != NULL_REVISION]
185
410
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
411
self.fetch_objects(determine_wants, mapping, pb)
416
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
417
"""InterRepository that copies revisions from a remote Git into a non-Git
420
def fetch_objects(self, determine_wants, mapping, pb=None):
422
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
186
423
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,
426
create_pb = pb = ui.ui_factory.nested_progress_bar()
427
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
430
def record_determine_wants(heads):
431
wants = determine_wants(heads)
432
recorded_wants.extend(wants)
436
self.target.lock_write()
438
self.target.start_write_group()
440
objects_iter = self.source.fetch_objects(
441
record_determine_wants,
443
target_git_object_retriever.get_raw,
445
import_git_objects(self.target, mapping, objects_iter,
446
target_git_object_retriever, recorded_wants, pb)
448
self.target.commit_write_group()
456
def is_compatible(source, target):
457
"""Be compatible with GitRepository."""
458
# FIXME: Also check target uses VersionedFile
459
return (isinstance(source, RemoteGitRepository) and
460
target.supports_rich_root() and
461
not isinstance(target, GitRepository))
464
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
465
"""InterRepository that copies revisions from a remote Git into a non-Git
468
def fetch_objects(self, determine_wants, mapping, pb=None):
469
wants = determine_wants(self.source._git.get_refs())
472
create_pb = pb = ui.ui_factory.nested_progress_bar()
473
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
475
self.target.lock_write()
477
self.target.start_write_group()
479
import_git_objects(self.target, mapping,
480
self.source._git.object_store,
481
target_git_object_retriever, wants, pb)
483
self.target.commit_write_group()
196
491
def is_compatible(source, target):
197
492
"""Be compatible with GitRepository."""
198
493
# FIXME: Also check target uses VersionedFile
199
494
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
495
target.supports_rich_root() and
496
not isinstance(target, GitRepository))
499
class InterGitRepository(InterRepository):
500
"""InterRepository that copies between Git repositories."""
502
_matching_repo_format = GitRepositoryFormat()
505
def _get_repo_format_to_test():
508
def copy_content(self, revision_id=None, pb=None):
509
"""See InterRepository.copy_content."""
510
self.fetch(revision_id, pb, find_ghosts=False)
512
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
513
mapping=None, fetch_spec=None):
515
mapping = self.source.get_mapping()
517
trace.info("git: %s", text)
519
if revision_id is not None:
520
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
521
elif fetch_spec is not None:
522
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
523
if fetch_spec is None and revision_id is None:
524
determine_wants = r.object_store.determine_wants_all
526
determine_wants = lambda x: [y for y in args if not y in r.object_store]
528
graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
529
f, commit = r.object_store.add_thin_pack()
531
self.source.fetch_pack(determine_wants, graphwalker, f.write, progress)
538
def is_compatible(source, target):
539
"""Be compatible with GitRepository."""
540
return (isinstance(source, GitRepository) and
541
isinstance(target, GitRepository))