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, ui, urlutils
18
from bzrlib.errors import InvalidRevisionId
19
from bzrlib.inventory import Inventory
20
from bzrlib.repository import InterRepository
21
from bzrlib.trace import info
22
from bzrlib.tsort import topo_sort
24
from bzrlib.plugins.git import git
17
from cStringIO import (
21
from dulwich.objects import (
25
from dulwich.object_store import (
37
from bzrlib.errors import (
42
from bzrlib.inventory import (
48
from bzrlib.lru_cache import (
51
from bzrlib.repository import (
54
from bzrlib.revision import (
57
from bzrlib.tsort import (
60
from bzrlib.versionedfile import (
61
FulltextContentFactory,
64
from bzrlib.plugins.git.mapping import (
66
inventory_to_tree_and_blobs,
69
from bzrlib.plugins.git.object_store import (
72
from bzrlib.plugins.git.remote import (
25
75
from bzrlib.plugins.git.repository import (
30
from bzrlib.plugins.git.remote import RemoteGitRepository
32
from dulwich.objects import Commit
34
from cStringIO import StringIO
37
82
class BzrFetchGraphWalker(object):
83
"""GraphWalker implementation that uses a Bazaar repository."""
39
85
def __init__(self, repository, mapping):
40
86
self.repository = repository
63
112
self.heads.update([p for p in ps if not p in self.done])
65
114
self.done.add(ret)
66
return self.mapping.revision_id_bzr_to_foreign(ret)
115
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
67
116
except InvalidRevisionId:
72
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
121
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
122
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
73
123
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
125
:param texts: VersionedFiles to add to
76
126
:param path: Path in the tree
77
127
:param blob: A git blob
128
:return: Inventory delta for this file
79
130
file_id = mapping.generate_file_id(path)
80
text_revision = inv.revision_id
81
repo.texts.add_lines((file_id, text_revision),
82
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
83
osutils.split_lines(blob.data))
84
ie = inv.add_path(path, "file", file_id)
85
ie.revision = text_revision
86
ie.text_size = len(blob.data)
87
ie.text_sha1 = osutils.sha_string(blob.data)
135
# We just have to hope this is indeed utf-8:
136
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
137
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
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.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
194
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
197
if file_id in base_inv:
198
old_path = base_inv.id2path(file_id)
201
return ([(old_path, path, file_id, ie)], shamap)
204
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
205
revision_id, parent_invs, shagitmap, lookup_object):
92
206
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
208
:param texts: VersionedFiles object to add to
95
209
:param path: Path in the tree
96
210
:param tree: A git tree object
97
:param inv: Inventory object
211
:param base_inv: Base inventory against which to return inventory delta
212
:return: Inventory delta for this subtree
99
215
file_id = mapping.generate_file_id(path)
100
text_revision = inv.revision_id
101
repo.texts.add_lines((file_id, text_revision),
102
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
104
ie = inv.add_path(path, "directory", file_id)
105
ie.revision = text_revision
106
for mode, name, hexsha in tree.entries():
107
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")),
221
base_ie = base_inv[file_id]
223
# Newly appeared here
225
ie.revision = revision_id
226
texts.add_lines((file_id, ie.revision), (), [])
227
invdelta.append((None, path, file_id, ie))
229
base_sha = base_ie.text_id
230
# See if this has changed at all
233
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
237
if base_sha == hexsha:
238
# If nothing has changed since the base revision, we're done
240
# Remember for next time
241
existing_children = set()
244
tree = lookup_object(hexsha)
245
for mode, name, child_hexsha in tree.entries():
108
246
basename = name.decode("utf-8")
112
child_path = urlutils.join(path, name)
114
tree = lookup_object(hexsha)
115
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
116
elif entry_kind == 1:
117
blob = lookup_object(hexsha)
118
fs_mode = mode & 0777
119
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
121
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
124
def import_git_objects(repo, mapping, object_iter, pb=None):
247
existing_children.add(basename)
248
child_path = osutils.pathjoin(path, name)
249
if stat.S_ISDIR(mode):
250
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
251
mapping, child_path, child_hexsha, base_inv, file_id,
252
revision_id, parent_invs, shagitmap, lookup_object)
253
invdelta.extend(subinvdelta)
254
child_modes.update(grandchildmodes)
255
shamap.extend(subshamap)
257
fs_mode = stat.S_IMODE(mode)
258
symlink = stat.S_ISLNK(mode)
259
subinvdelta, subshamap = import_git_blob(texts, mapping,
260
child_path, child_hexsha, base_inv, file_id, revision_id,
261
parent_invs, shagitmap, lookup_object,
262
bool(fs_mode & 0111), symlink)
263
invdelta.extend(subinvdelta)
264
shamap.extend(subshamap)
265
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
266
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
267
child_modes[child_path] = mode
268
# Remove any children that have disappeared
269
if base_ie is not None:
270
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
273
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
274
if ie.kind == "directory":
275
deletable.extend(ie.children.values())
276
shamap.append((hexsha, "tree", (file_id, revision_id)))
277
return invdelta, child_modes, shamap
280
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
282
"""Import a set of git objects into a bzr repository.
127
284
:param repo: Bazaar repository
129
286
:param object_iter: Iterator over Git objects.
131
288
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
294
parent_invs_cache = LRUCache(50)
140
295
# Find and convert commit objects
141
for o in objects.itervalues():
298
pb.update("finding revisions to fetch", len(graph), None)
300
assert isinstance(head, str)
302
o = object_iter[head]
142
305
if isinstance(o, Commit):
143
306
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
307
if repo.has_revision(rev.revision_id):
309
root_trees[rev.revision_id] = o.tree
145
310
revisions[rev.revision_id] = rev
146
311
graph.append((rev.revision_id, rev.parent_ids))
312
target_git_object_retriever._idmap.add_entry(o.id, "commit",
313
(rev.revision_id, o.tree))
314
heads.extend([p for p in o.parents if p not in checked])
315
elif isinstance(o, Tag):
316
heads.append(o.object[1])
318
trace.warning("Unable to import head object %r" % o)
147
320
# Order the revisions
148
321
# Create the inventory objects
149
322
for i, revid in enumerate(topo_sort(graph)):
150
323
if pb is not None:
151
324
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
325
rev = revisions[revid]
154
326
# We have to do this here, since we have to walk the tree and
155
# we need to make sure to import the blobs / trees with the riht
327
# we need to make sure to import the blobs / trees with the right
156
328
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
329
def lookup_object(sha):
162
return reconstruct_git_object(repo, mapping, sha)
163
parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
164
import_git_tree(repo, mapping, "", root_tree, inv, parent_invs, lookup_object)
165
repo.add_revision(rev.revision_id, rev, inv)
168
def reconstruct_git_commit(repo, rev):
169
raise NotImplementedError(self.reconstruct_git_commit)
172
def reconstruct_git_object(repo, mapping, sha):
174
revid = mapping.revision_id_foreign_to_bzr(sha)
176
rev = repo.get_revision(revid)
177
except NoSuchRevision:
180
return reconstruct_git_commit(rev)
184
raise KeyError("No such object %s" % sha)
331
return object_iter[sha]
333
return target_git_object_retriever[sha]
335
for parent_id in rev.parent_ids:
337
parent_invs.append(parent_invs_cache[parent_id])
339
parent_inv = repo.get_inventory(parent_id)
340
parent_invs.append(parent_inv)
341
parent_invs_cache[parent_id] = parent_inv
342
if parent_invs == []:
343
base_inv = Inventory(root_id=None)
345
base_inv = parent_invs[0]
346
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
347
mapping, "", root_trees[revid], base_inv, None, revid,
348
parent_invs, target_git_object_retriever._idmap, lookup_object)
349
target_git_object_retriever._idmap.add_entries(shamap)
350
if unusual_modes != {}:
351
ret = "unusual modes: \n"
352
for item in unusual_modes.iteritems():
353
ret += "\t%s: %o\n" % item
354
raise AssertionError(ret)
356
basis_id = rev.parent_ids[0]
358
basis_id = NULL_REVISION
359
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
360
inv_delta, rev.revision_id, rev.parent_ids)
361
parent_invs_cache[rev.revision_id] = inv
362
repo.add_revision(rev.revision_id, rev)
363
if "verify" in debug.debug_flags:
364
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
365
for sha1, newobj, path in objs:
366
assert path is not None
367
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
368
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
370
target_git_object_retriever._idmap.commit()
187
373
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
375
_matching_repo_format = GitRepositoryFormat()
192
378
def _get_repo_format_to_test():
196
382
"""See InterRepository.copy_content."""
197
383
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
385
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
387
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
388
mapping=mapping, fetch_spec=fetch_spec)
391
class InterGitNonGitRepository(InterGitRepository):
392
"""Base InterRepository that copies revisions from a Git into a non-Git
395
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
396
mapping=None, fetch_spec=None):
201
397
if mapping is None:
202
398
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
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("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
411
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid != NULL_REVISION]
210
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)
211
425
graph_walker = BzrFetchGraphWalker(self.target, mapping)
214
428
create_pb = pb = ui.ui_factory.nested_progress_bar()
216
self.target.lock_write()
218
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
224
self.target.commit_write_group()
232
def is_compatible(source, target):
233
"""Be compatible with GitRepository."""
234
# FIXME: Also check target uses VersionedFile
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()
493
def is_compatible(source, target):
494
"""Be compatible with GitRepository."""
495
# FIXME: Also check target uses VersionedFile
496
return (isinstance(source, LocalGitRepository) and
497
target.supports_rich_root() and
498
not isinstance(target, GitRepository))
501
class InterGitGitRepository(InterGitRepository):
502
"""InterRepository that copies between Git repositories."""
504
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
505
mapping=None, fetch_spec=None, branches=None):
507
mapping = self.source.get_mapping()
509
trace.info("git: %s", text)
511
if revision_id is not None:
512
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
513
elif fetch_spec is not None:
514
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
515
if branches is not None:
516
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
517
elif fetch_spec is None and revision_id is None:
518
determine_wants = r.object_store.determine_wants_all
520
determine_wants = lambda x: [y for y in args if not y in r.object_store]
522
graphwalker = r.get_graph_walker()
523
f, commit = r.object_store.add_thin_pack()
525
refs = self.source.fetch_pack(determine_wants, graphwalker,
534
def is_compatible(source, target):
535
"""Be compatible with GitRepository."""
235
536
return (isinstance(source, GitRepository) and
236
target.supports_rich_root())
537
isinstance(target, GitRepository))