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,
71
from bzrlib.plugins.git.object_store import (
74
from bzrlib.plugins.git.remote import (
25
77
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
class BzrFetchGraphWalker(object):
39
def __init__(self, repository, mapping):
40
self.repository = repository
41
self.mapping = mapping
43
self.heads = set(repository.all_revision_ids())
47
revid = self.mapping.revision_id_foreign_to_bzr(sha)
50
def remove(self, revid):
53
self.heads.remove(revid)
54
if revid in self.parents:
55
for p in self.parents[revid]:
60
ret = self.heads.pop()
61
ps = self.repository.get_parent_map([ret])[ret]
62
self.parents[ret] = ps
63
self.heads.update([p for p in ps if not p in self.done])
66
return self.mapping.revision_id_bzr_to_foreign(ret)
67
except InvalidRevisionId:
72
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
84
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
85
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
73
86
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
88
:param texts: VersionedFiles to add to
76
89
:param path: Path in the tree
77
90
:param blob: A git blob
91
:return: Inventory delta for this file
79
93
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)
98
# We just have to hope this is indeed utf-8:
99
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
100
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
102
# See if this has changed at all
104
base_ie = base_inv[file_id]
109
base_sha = base_ie.text_id
112
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
116
if (base_sha == hexsha and base_ie.executable == ie.executable
117
and base_ie.kind == ie.kind):
118
# If nothing has changed since the base revision, we're done
120
if base_sha == hexsha and base_ie.kind == ie.kind:
121
ie.text_size = base_ie.text_size
122
ie.text_sha1 = base_ie.text_sha1
123
ie.symlink_target = base_ie.symlink_target
124
if ie.executable == base_ie.executable:
125
ie.revision = base_ie.revision
127
blob = lookup_object(hexsha)
129
blob = lookup_object(hexsha)
130
if ie.kind == "symlink":
131
ie.symlink_target = blob.data
135
ie.text_size = len(blob.data)
136
ie.text_sha1 = osutils.sha_string(blob.data)
137
# Check what revision we should store
139
for pinv in parent_invs:
140
if pinv.revision_id == base_inv.revision_id:
149
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable:
150
# found a revision in one of the parents to use
151
ie.revision = pie.revision
153
parent_keys.append((file_id, pie.revision))
154
if ie.revision is None:
155
# Need to store a new revision
156
ie.revision = revision_id
157
assert file_id is not None
158
assert ie.revision is not None
159
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
160
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
163
if file_id in base_inv:
164
old_path = base_inv.id2path(file_id)
167
return ([(old_path, path, file_id, ie)], shamap)
170
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
171
revision_id, parent_invs, shagitmap, lookup_object):
92
172
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
174
:param texts: VersionedFiles object to add to
95
175
:param path: Path in the tree
96
176
:param tree: A git tree object
97
:param inv: Inventory object
177
:param base_inv: Base inventory against which to return inventory delta
178
:return: Inventory delta for this subtree
99
181
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
182
# We just have to hope this is indeed utf-8:
183
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
187
base_ie = base_inv[file_id]
189
# Newly appeared here
191
ie.revision = revision_id
192
texts.add_lines((file_id, ie.revision), (), [])
193
invdelta.append((None, path, file_id, ie))
195
base_sha = base_ie.text_id
196
# See if this has changed at all
199
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
203
if base_sha == hexsha:
204
# If nothing has changed since the base revision, we're done
206
# Remember for next time
207
existing_children = set()
210
tree = lookup_object(hexsha)
211
for mode, name, child_hexsha in tree.entries():
108
212
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):
213
existing_children.add(basename)
214
child_path = osutils.pathjoin(path, name)
215
if stat.S_ISDIR(mode):
216
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
217
mapping, child_path, child_hexsha, base_inv, file_id,
218
revision_id, parent_invs, shagitmap, lookup_object)
219
invdelta.extend(subinvdelta)
220
child_modes.update(grandchildmodes)
221
shamap.extend(subshamap)
223
subinvdelta, subshamap = import_git_blob(texts, mapping,
224
child_path, child_hexsha, base_inv, file_id, revision_id,
225
parent_invs, shagitmap, lookup_object,
226
mode_is_executable(mode), stat.S_ISLNK(mode))
227
invdelta.extend(subinvdelta)
228
shamap.extend(subshamap)
229
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
230
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
231
child_modes[child_path] = mode
232
# Remove any children that have disappeared
233
if base_ie is not None and base_ie.kind == 'directory':
234
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
237
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
238
if ie.kind == "directory":
239
deletable.extend(ie.children.values())
240
shamap.append((hexsha, "tree", (file_id, revision_id)))
241
return invdelta, child_modes, shamap
244
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
246
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
248
:param repo: Target Bazaar repository
128
249
:param mapping: Mapping to use
129
250
:param object_iter: Iterator over Git objects.
252
def lookup_object(sha):
254
return object_iter[sha]
256
return target_git_object_retriever[sha]
131
257
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
263
parent_invs_cache = LRUCache(50)
140
264
# Find and convert commit objects
141
for o in objects.itervalues():
267
pb.update("finding revisions to fetch", len(graph), None)
269
assert isinstance(head, str)
271
o = lookup_object(head)
142
274
if isinstance(o, Commit):
143
275
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
276
if repo.has_revision(rev.revision_id):
278
root_trees[rev.revision_id] = o.tree
145
279
revisions[rev.revision_id] = rev
146
280
graph.append((rev.revision_id, rev.parent_ids))
281
target_git_object_retriever._idmap.add_entry(o.id, "commit",
282
(rev.revision_id, o.tree))
283
heads.extend([p for p in o.parents if p not in checked])
284
elif isinstance(o, Tag):
285
heads.append(o.object[1])
287
trace.warning("Unable to import head object %r" % o)
147
289
# Order the revisions
148
290
# Create the inventory objects
149
291
for i, revid in enumerate(topo_sort(graph)):
150
292
if pb is not None:
151
293
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
294
rev = revisions[revid]
154
295
# 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
296
# we need to make sure to import the blobs / trees with the right
156
297
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
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)
299
for parent_id in rev.parent_ids:
301
parent_invs.append(parent_invs_cache[parent_id])
303
parent_inv = repo.get_inventory(parent_id)
304
parent_invs.append(parent_inv)
305
parent_invs_cache[parent_id] = parent_inv
306
if parent_invs == []:
307
base_inv = Inventory(root_id=None)
309
base_inv = parent_invs[0]
310
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
311
mapping, "", root_trees[revid], base_inv, None, revid,
312
parent_invs, target_git_object_retriever._idmap, lookup_object)
313
target_git_object_retriever._idmap.add_entries(shamap)
314
if unusual_modes != {}:
315
for path, mode in unusual_modes.iteritems():
316
warn_unusual_mode(rev.foreign_revid, path, mode)
318
basis_id = rev.parent_ids[0]
320
basis_id = NULL_REVISION
321
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
322
inv_delta, rev.revision_id, rev.parent_ids)
323
parent_invs_cache[rev.revision_id] = inv
324
repo.add_revision(rev.revision_id, rev)
325
if "verify" in debug.debug_flags:
326
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
327
for sha1, newobj, path in objs:
328
assert path is not None
329
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
330
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
332
target_git_object_retriever._idmap.commit()
187
335
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
337
_matching_repo_format = GitRepositoryFormat()
192
340
def _get_repo_format_to_test():
196
344
"""See InterRepository.copy_content."""
197
345
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
347
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
349
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
350
mapping=mapping, fetch_spec=fetch_spec)
353
class InterGitNonGitRepository(InterGitRepository):
354
"""Base InterRepository that copies revisions from a Git into a non-Git
357
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
358
mapping=None, fetch_spec=None):
201
359
if mapping is None:
202
360
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
361
if revision_id is not None:
362
interesting_heads = [revision_id]
363
elif fetch_spec is not None:
364
interesting_heads = fetch_spec.heads
366
interesting_heads = None
368
def determine_wants(refs):
370
if interesting_heads is None:
371
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
373
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
374
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
211
graph_walker = BzrFetchGraphWalker(self.target, mapping)
375
self.fetch_objects(determine_wants, mapping, pb)
379
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
380
"""InterRepository that copies revisions from a remote Git into a non-Git
383
def fetch_objects(self, determine_wants, mapping, pb=None):
385
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
386
store = BazaarObjectStore(self.target, mapping)
387
self.target.lock_write()
389
heads = self.target.get_graph().heads(self.target.all_revision_ids())
390
graph_walker = store.get_graph_walker(
391
[store._lookup_revision_sha1(head) for head in heads])
394
def record_determine_wants(heads):
395
wants = determine_wants(heads)
396
recorded_wants.extend(wants)
401
create_pb = pb = ui.ui_factory.nested_progress_bar()
403
self.target.start_write_group()
405
objects_iter = self.source.fetch_objects(
406
record_determine_wants, graph_walker,
407
store.get_raw, progress)
408
import_git_objects(self.target, mapping, objects_iter,
409
store, recorded_wants, pb)
411
self.target.commit_write_group()
419
def is_compatible(source, target):
420
"""Be compatible with GitRepository."""
421
# FIXME: Also check target uses VersionedFile
422
return (isinstance(source, RemoteGitRepository) and
423
target.supports_rich_root() and
424
not isinstance(target, GitRepository))
427
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
428
"""InterRepository that copies revisions from a local Git into a non-Git
431
def fetch_objects(self, determine_wants, mapping, pb=None):
432
wants = determine_wants(self.source._git.get_refs())
214
435
create_pb = pb = ui.ui_factory.nested_progress_bar()
436
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
438
self.target.lock_write()
218
440
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
442
import_git_objects(self.target, mapping,
443
self.source._git.object_store,
444
target_git_object_retriever, wants, pb)
224
446
self.target.commit_write_group()