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.client import (
22
SimpleFetchGraphWalker,
24
from dulwich.objects import (
37
from bzrlib.errors import (
41
from bzrlib.inventory import (
47
from bzrlib.lru_cache import (
50
from bzrlib.repository import (
53
from bzrlib.revision import (
56
from bzrlib.tsort import (
60
from bzrlib.plugins.git.converter import (
63
from bzrlib.plugins.git.mapping import (
25
69
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
74
from bzrlib.plugins.git.remote import (
37
79
class BzrFetchGraphWalker(object):
80
"""GraphWalker implementation that uses a Bazaar repository."""
39
82
def __init__(self, repository, mapping):
40
83
self.repository = repository
63
109
self.heads.update([p for p in ps if not p in self.done])
65
111
self.done.add(ret)
66
return self.mapping.revision_id_bzr_to_foreign(ret)
112
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
67
113
except InvalidRevisionId:
72
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
118
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
119
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
73
120
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
122
:param texts: VersionedFiles to add to
76
123
:param path: Path in the tree
77
124
:param blob: A git blob
125
:return: Inventory delta for this file
79
127
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)
132
# We just have to hope this is indeed utf-8:
133
ie = cls(file_id, urlutils.basename(path).decode("utf-8"),
88
135
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
136
# See if this has changed at all
138
base_sha = shagitmap.lookup_blob(file_id, base_inv.revision_id)
142
if (base_sha == hexsha and base_inv[file_id].executable == ie.executable
143
and base_inv[file_id].kind == ie.kind):
144
# If nothing has changed since the base revision, we're done
146
if base_sha == hexsha:
147
ie.text_size = base_inv[file_id].text_size
148
ie.text_sha1 = base_inv[file_id].text_sha1
149
ie.symlink_target = base_inv[file_id].symlink_target
150
ie.revision = base_inv[file_id].revision
152
blob = lookup_object(hexsha)
153
if ie.kind == "symlink":
154
ie.symlink_target = blob.data
158
ie.text_size = len(blob.data)
159
ie.text_sha1 = osutils.sha_string(blob.data)
160
# Check what revision we should store
162
for pinv in parent_invs:
163
if not file_id in pinv:
165
if pinv[file_id].text_sha1 == ie.text_sha1:
166
# found a revision in one of the parents to use
167
ie.revision = pinv[file_id].revision
169
parent_keys.append((file_id, pinv[file_id].revision))
170
if ie.revision is None:
171
# Need to store a new revision
172
ie.revision = revision_id
173
assert file_id is not None
174
assert ie.revision is not None
175
texts.add_lines((file_id, ie.revision), parent_keys,
176
osutils.split_lines(blob.data))
177
if "verify" in debug.debug_flags:
178
assert text_to_blob(blob.data).id == hexsha
179
shagitmap.add_entry(hexsha, "blob", (ie.file_id, ie.revision))
180
if file_id in base_inv:
181
old_path = base_inv.id2path(file_id)
184
return [(old_path, path, file_id, ie)]
187
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
188
revision_id, parent_invs, shagitmap, lookup_object):
92
189
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
191
:param texts: VersionedFiles object to add to
95
192
:param path: Path in the tree
96
193
:param tree: A git tree object
97
:param inv: Inventory object
194
:param base_inv: Base inventory against which to return inventory delta
195
:return: Inventory delta for this subtree
99
198
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
199
# We just have to hope this is indeed utf-8:
200
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
202
if not file_id in base_inv:
203
# Newly appeared here
204
ie.revision = revision_id
205
texts.add_lines((file_id, ie.revision), [], [])
206
invdelta.append((None, path, file_id, ie))
208
# See if this has changed at all
210
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
214
if base_sha == hexsha:
215
# If nothing has changed since the base revision, we're done
217
# Remember for next time
218
existing_children = set()
219
if "verify" in debug.debug_flags:
222
shagitmap.add_entry(hexsha, "tree", (file_id, revision_id))
224
tree = lookup_object(hexsha)
106
225
for mode, name, hexsha in tree.entries():
107
entry_kind = (mode & 0700000) / 0100000
108
226
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):
227
existing_children.add(basename)
228
child_path = osutils.pathjoin(path, name)
229
if stat.S_ISDIR(mode):
230
subinvdelta, grandchildmodes = import_git_tree(texts, mapping, child_path, hexsha,
231
base_inv, file_id, revision_id, parent_invs, shagitmap,
233
invdelta.extend(subinvdelta)
234
child_modes.update(grandchildmodes)
236
fs_mode = stat.S_IMODE(mode)
237
symlink = stat.S_ISLNK(mode)
238
subinvdelta = import_git_blob(texts, mapping, child_path, hexsha,
239
base_inv, file_id, revision_id, parent_invs, shagitmap,
240
lookup_object, bool(fs_mode & 0111), symlink)
241
invdelta.extend(subinvdelta)
242
if mode not in (DEFAULT_TREE_MODE, DEFAULT_FILE_MODE, DEFAULT_SYMLINK_MODE, DEFAULT_FILE_MODE|0111):
243
child_modes[child_path] = mode
244
# Remove any children that have disappeared
245
if file_id in base_inv:
246
deletable = [v for k,v in base_inv[file_id].children.iteritems() if k not in existing_children]
249
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
250
if ie.kind == "directory":
251
deletable.extend(ie.children.values())
252
return invdelta, child_modes
255
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
257
"""Import a set of git objects into a bzr repository.
127
259
:param repo: Bazaar repository
129
261
:param object_iter: Iterator over Git objects.
131
263
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
269
parent_invs_cache = LRUCache(50)
140
270
# Find and convert commit objects
141
for o in objects.itervalues():
273
pb.update("finding revisions to fetch", len(graph), None)
275
assert isinstance(head, str)
277
o = object_iter[head]
142
280
if isinstance(o, Commit):
143
281
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
282
if repo.has_revision(rev.revision_id):
284
root_trees[rev.revision_id] = o.tree
145
285
revisions[rev.revision_id] = rev
146
286
graph.append((rev.revision_id, rev.parent_ids))
287
target_git_object_retriever._idmap.add_entry(o.sha().hexdigest(),
288
"commit", (rev.revision_id, o._tree))
289
heads.extend([p for p in o.parents if p not in checked])
290
elif isinstance(o, Tag):
291
heads.append(o.object[1])
293
trace.warning("Unable to import head object %r" % o)
147
295
# Order the revisions
148
296
# Create the inventory objects
149
297
for i, revid in enumerate(topo_sort(graph)):
150
298
if pb is not None:
151
299
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
300
rev = revisions[revid]
154
301
# 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
302
# we need to make sure to import the blobs / trees with the right
156
303
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
304
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)
187
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
306
return object_iter[sha]
308
return target_git_object_retriever[sha]
310
for parent_id in rev.parent_ids:
312
parent_invs.append(parent_invs_cache[parent_id])
314
parent_inv = repo.get_inventory(parent_id)
315
parent_invs.append(parent_inv)
316
parent_invs_cache[parent_id] = parent_inv
317
if parent_invs == []:
318
base_inv = Inventory(root_id=None)
320
base_inv = parent_invs[0]
321
inv_delta, unusual_modes = import_git_tree(repo.texts, mapping, "",
322
root_trees[revid], base_inv, None, revid, parent_invs,
323
target_git_object_retriever._idmap, lookup_object)
324
if unusual_modes != {}:
325
ret = "unusual modes: \n"
326
for item in unusual_modes.iteritems():
327
ret += "\t%s: %o\n" % item
328
raise AssertionError(ret)
330
basis_id = rev.parent_ids[0]
332
basis_id = NULL_REVISION
333
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
334
inv_delta, rev.revision_id, rev.parent_ids)
335
parent_invs_cache[rev.revision_id] = inv
336
repo.add_revision(rev.revision_id, rev)
337
target_git_object_retriever._idmap.commit()
340
class InterGitNonGitRepository(InterRepository):
341
"""Base InterRepository that copies revisions from a Git into a non-Git
344
_matching_repo_format = GitRepositoryFormat()
192
347
def _get_repo_format_to_test():
196
351
"""See InterRepository.copy_content."""
197
352
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
354
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
356
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
357
mapping=mapping, fetch_spec=fetch_spec)
359
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
360
mapping=None, fetch_spec=None):
201
361
if mapping is None:
202
362
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
363
if revision_id is not None:
364
interesting_heads = [revision_id]
365
elif fetch_spec is not None:
366
interesting_heads = fetch_spec.heads
368
interesting_heads = None
370
def determine_wants(refs):
372
if interesting_heads is None:
373
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
375
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid != NULL_REVISION]
210
376
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
377
self.fetch_objects(determine_wants, mapping, pb)
382
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
383
"""InterRepository that copies revisions from a remote Git into a non-Git
386
def fetch_objects(self, determine_wants, mapping, pb=None):
388
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
211
389
graph_walker = BzrFetchGraphWalker(self.target, mapping)
214
392
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
393
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
396
def record_determine_wants(heads):
397
wants = determine_wants(heads)
398
recorded_wants.extend(wants)
402
self.target.lock_write()
404
self.target.start_write_group()
406
objects_iter = self.source.fetch_objects(
407
record_determine_wants,
409
target_git_object_retriever.get_raw,
411
import_git_objects(self.target, mapping, objects_iter,
412
target_git_object_retriever, recorded_wants, pb)
414
self.target.commit_write_group()
422
def is_compatible(source, target):
423
"""Be compatible with GitRepository."""
424
# FIXME: Also check target uses VersionedFile
425
return (isinstance(source, RemoteGitRepository) and
426
target.supports_rich_root() and
427
not isinstance(target, GitRepository))
430
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
431
"""InterRepository that copies revisions from a remote Git into a non-Git
434
def fetch_objects(self, determine_wants, mapping, pb=None):
435
wants = determine_wants(self.source._git.get_refs())
438
create_pb = pb = ui.ui_factory.nested_progress_bar()
439
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
441
self.target.lock_write()
443
self.target.start_write_group()
445
import_git_objects(self.target, mapping,
446
self.source._git.object_store,
447
target_git_object_retriever, wants, pb)
449
self.target.commit_write_group()
457
def is_compatible(source, target):
458
"""Be compatible with GitRepository."""
459
# FIXME: Also check target uses VersionedFile
460
return (isinstance(source, LocalGitRepository) and
461
target.supports_rich_root() and
462
not isinstance(target, GitRepository))
465
class InterGitRepository(InterRepository):
466
"""InterRepository that copies between Git repositories."""
468
_matching_repo_format = GitRepositoryFormat()
471
def _get_repo_format_to_test():
474
def copy_content(self, revision_id=None, pb=None):
475
"""See InterRepository.copy_content."""
476
self.fetch(revision_id, pb, find_ghosts=False)
478
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
479
mapping=None, fetch_spec=None):
481
mapping = self.source.get_mapping()
483
trace.info("git: %s", text)
485
if revision_id is not None:
486
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
487
elif fetch_spec is not None:
488
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
489
if fetch_spec is None and revision_id is None:
490
determine_wants = r.object_store.determine_wants_all
492
determine_wants = lambda x: [y for y in args if not y in r.object_store]
494
graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
495
f, commit = r.object_store.add_thin_pack()
497
self.source.fetch_pack(determine_wants, graphwalker, f.write, progress)
504
def is_compatible(source, target):
505
"""Be compatible with GitRepository."""
235
506
return (isinstance(source, GitRepository) and
236
target.supports_rich_root())
507
isinstance(target, GitRepository))