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
32
class BzrFetchGraphWalker(object):
34
def __init__(self, repository, mapping):
35
self.repository = repository
36
self.mapping = mapping
38
self.heads = set(repository.all_revision_ids())
42
revid = self.mapping.revision_id_foreign_to_bzr(sha)
45
def remove(self, revid):
48
self.heads.remove(revid)
49
if revid in self.parents:
50
for p in self.parents[revid]:
55
ret = self.heads.pop()
56
ps = self.repository.get_parent_map([ret])[ret]
57
self.parents[ret] = ps
58
self.heads.update([p for p in ps if not p in self.done])
61
return self.mapping.revision_id_bzr_to_foreign(ret)
62
except InvalidRevisionId:
67
def import_git_blob(repo, mapping, path, blob):
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 (
75
from bzrlib.plugins.git.repository import (
82
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
83
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
68
84
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
86
:param texts: VersionedFiles to add to
71
87
:param path: Path in the tree
72
88
:param blob: A git blob
89
:return: Inventory delta for this file
74
91
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):
96
# We just have to hope this is indeed utf-8:
97
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
98
ie.executable = executable
100
# See if this has changed at all
102
base_ie = base_inv[file_id]
107
base_sha = base_ie.text_id
110
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
114
if (base_sha == hexsha and base_ie.executable == ie.executable
115
and base_ie.kind == ie.kind):
116
# If nothing has changed since the base revision, we're done
118
if base_sha == hexsha:
119
ie.text_size = base_ie.text_size
120
ie.text_sha1 = base_ie.text_sha1
121
ie.symlink_target = base_ie.symlink_target
122
ie.revision = base_ie.revision
124
blob = lookup_object(hexsha)
125
if ie.kind == "symlink":
126
ie.symlink_target = blob.data
130
ie.text_size = len(blob.data)
131
ie.text_sha1 = osutils.sha_string(blob.data)
132
# Check what revision we should store
134
for pinv in parent_invs:
135
if pinv.revision_id == base_inv.revision_id:
144
if pie.text_sha1 == ie.text_sha1:
145
# found a revision in one of the parents to use
146
ie.revision = pie.revision
148
parent_keys.append((file_id, pie.revision))
149
if ie.revision is None:
150
# Need to store a new revision
151
ie.revision = revision_id
152
assert file_id is not None
153
assert ie.revision is not None
154
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
155
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
158
if file_id in base_inv:
159
old_path = base_inv.id2path(file_id)
162
return ([(old_path, path, file_id, ie)], shamap)
165
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
166
revision_id, parent_invs, shagitmap, lookup_object):
82
167
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
169
:param texts: VersionedFiles object to add to
85
170
:param path: Path in the tree
86
171
:param tree: A git tree object
87
:param inv: Inventory object
172
:param base_inv: Base inventory against which to return inventory delta
173
:return: Inventory delta for this subtree
89
176
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
177
# We just have to hope this is indeed utf-8:
178
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
182
base_ie = base_inv[file_id]
184
# Newly appeared here
186
ie.revision = revision_id
187
texts.add_lines((file_id, ie.revision), (), [])
188
invdelta.append((None, path, file_id, ie))
190
base_sha = base_ie.text_id
191
# See if this has changed at all
194
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
198
if base_sha == hexsha:
199
# If nothing has changed since the base revision, we're done
201
# Remember for next time
202
existing_children = set()
205
tree = lookup_object(hexsha)
206
for mode, name, child_hexsha in tree.entries():
96
207
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):
208
existing_children.add(basename)
209
child_path = osutils.pathjoin(path, name)
210
if stat.S_ISDIR(mode):
211
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
212
mapping, child_path, child_hexsha, base_inv, file_id,
213
revision_id, parent_invs, shagitmap, lookup_object)
214
invdelta.extend(subinvdelta)
215
child_modes.update(grandchildmodes)
216
shamap.extend(subshamap)
218
fs_mode = stat.S_IMODE(mode)
219
symlink = stat.S_ISLNK(mode)
220
subinvdelta, subshamap = import_git_blob(texts, mapping,
221
child_path, child_hexsha, base_inv, file_id, revision_id,
222
parent_invs, shagitmap, lookup_object,
223
bool(fs_mode & 0111), symlink)
224
invdelta.extend(subinvdelta)
225
shamap.extend(subshamap)
226
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
227
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
228
child_modes[child_path] = mode
229
# Remove any children that have disappeared
230
if base_ie is not None:
231
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
234
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
235
if ie.kind == "directory":
236
deletable.extend(ie.children.values())
237
shamap.append((hexsha, "tree", (file_id, revision_id)))
238
return invdelta, child_modes, shamap
241
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
110
243
"""Import a set of git objects into a bzr repository.
112
245
:param repo: Bazaar repository
113
246
:param mapping: Mapping to use
114
247
:param object_iter: Iterator over Git objects.
249
def lookup_object(sha):
251
return object_iter[sha]
253
return target_git_object_retriever[sha]
116
254
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
260
parent_invs_cache = LRUCache(50)
121
261
# Find and convert commit objects
122
for o in objects.iterkeys():
264
pb.update("finding revisions to fetch", len(graph), None)
266
assert isinstance(head, str)
268
o = lookup_object(head)
123
271
if isinstance(o, Commit):
124
272
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
273
if repo.has_revision(rev.revision_id):
275
root_trees[rev.revision_id] = o.tree
276
revisions[rev.revision_id] = rev
277
graph.append((rev.revision_id, rev.parent_ids))
278
target_git_object_retriever._idmap.add_entry(o.id, "commit",
279
(rev.revision_id, o.tree))
280
heads.extend([p for p in o.parents if p not in checked])
281
elif isinstance(o, Tag):
282
heads.append(o.object[1])
284
trace.warning("Unable to import head object %r" % o)
286
# Order the revisions
126
287
# Create the inventory objects
127
for rev, root_tree in root_trees.iteritems():
288
for i, revid in enumerate(topo_sort(graph)):
290
pb.update("fetching revisions", i, len(graph))
291
rev = revisions[revid]
128
292
# 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
293
# we need to make sure to import the blobs / trees with the right
130
294
# path; this may involve adding them more than once.
132
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)
296
for parent_id in rev.parent_ids:
298
parent_invs.append(parent_invs_cache[parent_id])
300
parent_inv = repo.get_inventory(parent_id)
301
parent_invs.append(parent_inv)
302
parent_invs_cache[parent_id] = parent_inv
303
if parent_invs == []:
304
base_inv = Inventory(root_id=None)
306
base_inv = parent_invs[0]
307
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
308
mapping, "", root_trees[revid], base_inv, None, revid,
309
parent_invs, target_git_object_retriever._idmap, lookup_object)
310
target_git_object_retriever._idmap.add_entries(shamap)
311
if unusual_modes != {}:
312
ret = "unusual modes: \n"
313
for item in unusual_modes.iteritems():
314
ret += "\t%s: %o\n" % item
315
raise AssertionError(ret)
317
basis_id = rev.parent_ids[0]
319
basis_id = NULL_REVISION
320
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
321
inv_delta, rev.revision_id, rev.parent_ids)
322
parent_invs_cache[rev.revision_id] = inv
323
repo.add_revision(rev.revision_id, rev)
324
if "verify" in debug.debug_flags:
325
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
326
for sha1, newobj, path in objs:
327
assert path is not None
328
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
329
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
331
target_git_object_retriever._idmap.commit()
159
334
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
336
_matching_repo_format = GitRepositoryFormat()
164
339
def _get_repo_format_to_test():
168
343
"""See InterRepository.copy_content."""
169
344
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
346
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
348
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
349
mapping=mapping, fetch_spec=fetch_spec)
352
class InterGitNonGitRepository(InterGitRepository):
353
"""Base InterRepository that copies revisions from a Git into a non-Git
356
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
357
mapping=None, fetch_spec=None):
173
358
if mapping is None:
174
359
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)]
360
if revision_id is not None:
361
interesting_heads = [revision_id]
362
elif fetch_spec is not None:
363
interesting_heads = fetch_spec.heads
365
interesting_heads = None
367
def determine_wants(refs):
369
if interesting_heads is None:
370
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
372
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
373
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
186
graph_walker = BzrFetchGraphWalker(self.target, mapping)
187
self.target.lock_write()
374
self.fetch_objects(determine_wants, mapping, pb)
378
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
379
"""InterRepository that copies revisions from a remote Git into a non-Git
382
def fetch_objects(self, determine_wants, mapping, pb=None):
384
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
385
store = BazaarObjectStore(self.target, mapping)
386
self.target.lock_read()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
388
heads = self.target.get_graph().heads(self.target.all_revision_ids())
193
390
self.target.unlock()
391
graph_walker = store.get_graph_walker(
392
[store._lookup_revision_sha1(head) for head in heads])
395
create_pb = pb = ui.ui_factory.nested_progress_bar()
398
def record_determine_wants(heads):
399
wants = determine_wants(heads)
400
recorded_wants.extend(wants)
404
self.target.lock_write()
406
self.target.start_write_group()
408
objects_iter = self.source.fetch_objects(
409
record_determine_wants, graph_walker,
410
store.get_raw, progress)
411
import_git_objects(self.target, mapping, objects_iter,
412
store, 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()
196
457
def is_compatible(source, target):
197
458
"""Be compatible with GitRepository."""
198
459
# FIXME: Also check target uses VersionedFile
199
460
return (isinstance(source, LocalGitRepository) and
200
target.supports_rich_root())
461
target.supports_rich_root() and
462
not isinstance(target, GitRepository))
465
class InterGitGitRepository(InterGitRepository):
466
"""InterRepository that copies between Git repositories."""
468
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
469
mapping=None, fetch_spec=None, branches=None):
471
mapping = self.source.get_mapping()
473
trace.info("git: %s", text)
475
if revision_id is not None:
476
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
477
elif fetch_spec is not None:
478
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
479
if branches is not None:
480
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
481
elif fetch_spec is None and revision_id is None:
482
determine_wants = r.object_store.determine_wants_all
484
determine_wants = lambda x: [y for y in args if not y in r.object_store]
486
graphwalker = r.get_graph_walker()
487
f, commit = r.object_store.add_thin_pack()
489
refs = self.source.fetch_pack(determine_wants, graphwalker,
498
def is_compatible(source, target):
499
"""Be compatible with GitRepository."""
500
return (isinstance(source, GitRepository) and
501
isinstance(target, GitRepository))