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 (
28
from dulwich.object_store import (
40
from bzrlib.errors import (
44
from bzrlib.inventory import (
50
from bzrlib.lru_cache import (
53
from bzrlib.repository import (
56
from bzrlib.revision import (
59
from bzrlib.tsort import (
63
from bzrlib.plugins.git.converter import (
66
from bzrlib.plugins.git.mapping import (
70
inventory_to_tree_and_blobs,
25
73
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
78
from bzrlib.plugins.git.remote import (
37
83
class BzrFetchGraphWalker(object):
84
"""GraphWalker implementation that uses a Bazaar repository."""
39
86
def __init__(self, repository, mapping):
40
87
self.repository = repository
63
113
self.heads.update([p for p in ps if not p in self.done])
65
115
self.done.add(ret)
66
return self.mapping.revision_id_bzr_to_foreign(ret)
116
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
67
117
except InvalidRevisionId:
72
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
122
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
123
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
73
124
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
126
:param texts: VersionedFiles to add to
76
127
:param path: Path in the tree
77
128
:param blob: A git blob
129
:return: Inventory delta for this file
79
131
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)
136
# We just have to hope this is indeed utf-8:
137
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
138
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_sha = shagitmap.lookup_blob(file_id, base_inv.revision_id)
145
if (base_sha == hexsha and base_inv[file_id].executable == ie.executable
146
and base_inv[file_id].kind == ie.kind):
147
# If nothing has changed since the base revision, we're done
149
if base_sha == hexsha:
150
ie.text_size = base_inv[file_id].text_size
151
ie.text_sha1 = base_inv[file_id].text_sha1
152
ie.symlink_target = base_inv[file_id].symlink_target
153
ie.revision = base_inv[file_id].revision
155
blob = lookup_object(hexsha)
156
if ie.kind == "symlink":
157
ie.symlink_target = blob.data
161
ie.text_size = len(blob.data)
162
ie.text_sha1 = osutils.sha_string(blob.data)
163
# Check what revision we should store
165
for pinv in parent_invs:
166
if not file_id in pinv:
168
if pinv[file_id].text_sha1 == ie.text_sha1:
169
# found a revision in one of the parents to use
170
ie.revision = pinv[file_id].revision
172
parent_keys.append((file_id, pinv[file_id].revision))
173
if ie.revision is None:
174
# Need to store a new revision
175
ie.revision = revision_id
176
assert file_id is not None
177
assert ie.revision is not None
178
texts.add_lines((file_id, ie.revision), parent_keys,
179
osutils.split_lines(blob.data))
180
if "verify" in debug.debug_flags:
181
assert text_to_blob(blob.data).id == hexsha
182
shagitmap.add_entry(hexsha, "blob", (ie.file_id, ie.revision))
183
if file_id in base_inv:
184
old_path = base_inv.id2path(file_id)
187
return [(old_path, path, file_id, ie)]
190
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
191
revision_id, parent_invs, shagitmap, lookup_object):
92
192
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
194
:param texts: VersionedFiles object to add to
95
195
:param path: Path in the tree
96
196
:param tree: A git tree object
97
:param inv: Inventory object
197
:param base_inv: Base inventory against which to return inventory delta
198
:return: Inventory delta for this subtree
99
201
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
202
# We just have to hope this is indeed utf-8:
203
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
205
if not file_id in base_inv:
206
# Newly appeared here
207
ie.revision = revision_id
208
texts.add_lines((file_id, ie.revision), [], [])
209
invdelta.append((None, path, file_id, ie))
211
# See if this has changed at all
213
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
217
if base_sha == hexsha:
218
# If nothing has changed since the base revision, we're done
220
# Remember for next time
221
existing_children = set()
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,
231
child_path, hexsha, base_inv, file_id, revision_id,
232
parent_invs, shagitmap, lookup_object)
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,
243
DEFAULT_SYMLINK_MODE, DEFAULT_FILE_MODE|0111):
244
child_modes[child_path] = mode
245
# Remove any children that have disappeared
246
if file_id in base_inv:
247
deletable = [v for k,v in base_inv[file_id].children.iteritems() if k not in existing_children]
250
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
251
if ie.kind == "directory":
252
deletable.extend(ie.children.values())
253
return invdelta, child_modes
256
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
258
"""Import a set of git objects into a bzr repository.
127
260
:param repo: Bazaar repository
129
262
:param object_iter: Iterator over Git objects.
131
264
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
270
parent_invs_cache = LRUCache(50)
140
271
# Find and convert commit objects
141
for o in objects.itervalues():
274
pb.update("finding revisions to fetch", len(graph), None)
276
assert isinstance(head, str)
278
o = object_iter[head]
142
281
if isinstance(o, Commit):
143
282
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
283
if repo.has_revision(rev.revision_id):
285
root_trees[rev.revision_id] = o.tree
145
286
revisions[rev.revision_id] = rev
146
287
graph.append((rev.revision_id, rev.parent_ids))
288
target_git_object_retriever._idmap.add_entry(o.sha().hexdigest(),
289
"commit", (rev.revision_id, o.tree))
290
heads.extend([p for p in o.parents if p not in checked])
291
elif isinstance(o, Tag):
292
heads.append(o.object[1])
294
trace.warning("Unable to import head object %r" % o)
147
296
# Order the revisions
148
297
# Create the inventory objects
149
298
for i, revid in enumerate(topo_sort(graph)):
150
299
if pb is not None:
151
300
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
301
rev = revisions[revid]
154
302
# 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
303
# we need to make sure to import the blobs / trees with the right
156
304
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
305
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()
307
return object_iter[sha]
309
return target_git_object_retriever[sha]
311
for parent_id in rev.parent_ids:
313
parent_invs.append(parent_invs_cache[parent_id])
315
parent_inv = repo.get_inventory(parent_id)
316
parent_invs.append(parent_inv)
317
parent_invs_cache[parent_id] = parent_inv
318
if parent_invs == []:
319
base_inv = Inventory(root_id=None)
321
base_inv = parent_invs[0]
322
inv_delta, unusual_modes = import_git_tree(repo.texts, mapping, "",
323
root_trees[revid], base_inv, None, revid, parent_invs,
324
target_git_object_retriever._idmap, lookup_object)
325
if unusual_modes != {}:
326
ret = "unusual modes: \n"
327
for item in unusual_modes.iteritems():
328
ret += "\t%s: %o\n" % item
329
raise AssertionError(ret)
331
basis_id = rev.parent_ids[0]
333
basis_id = NULL_REVISION
334
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
335
inv_delta, rev.revision_id, rev.parent_ids)
336
parent_invs_cache[rev.revision_id] = inv
337
repo.add_revision(rev.revision_id, rev)
338
if "verify" in debug.debug_flags:
339
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
340
for sha1, newobj, path in objs:
341
assert path is not None
342
oldobj = tree_lookup_path(object_iter, root_trees[revid], path)
343
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
345
target_git_object_retriever._idmap.commit()
348
class InterGitNonGitRepository(InterRepository):
349
"""Base InterRepository that copies revisions from a Git into a non-Git
352
_matching_repo_format = GitRepositoryFormat()
192
355
def _get_repo_format_to_test():
196
359
"""See InterRepository.copy_content."""
197
360
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
362
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
364
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
365
mapping=mapping, fetch_spec=fetch_spec)
367
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
368
mapping=None, fetch_spec=None):
201
369
if mapping is None:
202
370
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
371
if revision_id is not None:
372
interesting_heads = [revision_id]
373
elif fetch_spec is not None:
374
interesting_heads = fetch_spec.heads
376
interesting_heads = None
378
def determine_wants(refs):
380
if interesting_heads is None:
381
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
383
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid != NULL_REVISION]
210
384
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
385
self.fetch_objects(determine_wants, mapping, pb)
390
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
391
"""InterRepository that copies revisions from a remote Git into a non-Git
394
def fetch_objects(self, determine_wants, mapping, pb=None):
396
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
211
397
graph_walker = BzrFetchGraphWalker(self.target, mapping)
214
400
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
401
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
404
def record_determine_wants(heads):
405
wants = determine_wants(heads)
406
recorded_wants.extend(wants)
410
self.target.lock_write()
412
self.target.start_write_group()
414
objects_iter = self.source.fetch_objects(
415
record_determine_wants,
417
target_git_object_retriever.get_raw,
419
import_git_objects(self.target, mapping, objects_iter,
420
target_git_object_retriever, recorded_wants, pb)
422
self.target.commit_write_group()
430
def is_compatible(source, target):
431
"""Be compatible with GitRepository."""
432
# FIXME: Also check target uses VersionedFile
433
return (isinstance(source, RemoteGitRepository) and
434
target.supports_rich_root() and
435
not isinstance(target, GitRepository))
438
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
439
"""InterRepository that copies revisions from a remote Git into a non-Git
442
def fetch_objects(self, determine_wants, mapping, pb=None):
443
wants = determine_wants(self.source._git.get_refs())
446
create_pb = pb = ui.ui_factory.nested_progress_bar()
447
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
449
self.target.lock_write()
451
self.target.start_write_group()
453
import_git_objects(self.target, mapping,
454
self.source._git.object_store,
455
target_git_object_retriever, wants, pb)
457
self.target.commit_write_group()
465
def is_compatible(source, target):
466
"""Be compatible with GitRepository."""
467
# FIXME: Also check target uses VersionedFile
468
return (isinstance(source, LocalGitRepository) and
469
target.supports_rich_root() and
470
not isinstance(target, GitRepository))
473
class InterGitRepository(InterRepository):
474
"""InterRepository that copies between Git repositories."""
476
_matching_repo_format = GitRepositoryFormat()
479
def _get_repo_format_to_test():
482
def copy_content(self, revision_id=None, pb=None):
483
"""See InterRepository.copy_content."""
484
self.fetch(revision_id, pb, find_ghosts=False)
486
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
487
mapping=None, fetch_spec=None):
489
mapping = self.source.get_mapping()
491
trace.info("git: %s", text)
493
if revision_id is not None:
494
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
495
elif fetch_spec is not None:
496
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
497
if fetch_spec is None and revision_id is None:
498
determine_wants = r.object_store.determine_wants_all
500
determine_wants = lambda x: [y for y in args if not y in r.object_store]
502
graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
503
f, commit = r.object_store.add_thin_pack()
505
self.source.fetch_pack(determine_wants, graphwalker, f.write, progress)
512
def is_compatible(source, target):
513
"""Be compatible with GitRepository."""
235
514
return (isinstance(source, GitRepository) and
236
target.supports_rich_root())
515
isinstance(target, GitRepository))