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 cStringIO import (
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.repository import (
29
from bzrlib.plugins.git.remote import RemoteGitRepository
20
31
import dulwich as git
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 (
77
from bzrlib.plugins.git.repository import (
84
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
85
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
32
from dulwich.client import SimpleFetchGraphWalker
33
from dulwich.objects import Commit
35
from cStringIO import StringIO
38
class BzrFetchGraphWalker(object):
39
"""GraphWalker implementation that uses a Bazaar repository."""
41
def __init__(self, repository, mapping):
42
self.repository = repository
43
self.mapping = mapping
45
self.heads = set(repository.all_revision_ids())
49
return iter(self.next, None)
52
revid = self.mapping.revision_id_foreign_to_bzr(sha)
55
def remove(self, revid):
57
if revid in self.heads:
58
self.heads.remove(revid)
59
if revid in self.parents:
60
for p in self.parents[revid]:
65
ret = self.heads.pop()
66
ps = self.repository.get_parent_map([ret])[ret]
67
self.parents[ret] = ps
68
self.heads.update([p for p in ps if not p in self.done])
71
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
72
except InvalidRevisionId:
77
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
86
78
"""Import a git blob object into a bzr repository.
88
:param texts: VersionedFiles to add to
80
:param repo: bzr repository
89
81
:param path: Path in the tree
90
82
:param blob: A git blob
91
:return: Inventory delta for this file
93
84
file_id = mapping.generate_file_id(path)
98
# We just have to hope this is indeed utf-8:
99
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
85
text_revision = inv.revision_id
86
repo.texts.add_lines((file_id, text_revision),
87
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
88
osutils.split_lines(blob.data))
89
ie = inv.add_path(path, "file", file_id)
90
ie.revision = text_revision
91
ie.text_size = len(blob.data)
92
ie.text_sha1 = osutils.sha_string(blob.data)
100
93
ie.executable = executable
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
ie.revision = base_ie.revision
126
blob = lookup_object(hexsha)
127
if ie.kind == "symlink":
128
ie.symlink_target = blob.data
132
ie.text_size = len(blob.data)
133
ie.text_sha1 = osutils.sha_string(blob.data)
134
# Check what revision we should store
136
for pinv in parent_invs:
137
if pinv.revision_id == base_inv.revision_id:
146
if pie.text_sha1 == ie.text_sha1:
147
# found a revision in one of the parents to use
148
ie.revision = pie.revision
150
parent_keys.append((file_id, pie.revision))
151
if ie.revision is None:
152
# Need to store a new revision
153
ie.revision = revision_id
154
assert file_id is not None
155
assert ie.revision is not None
156
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
157
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
160
if file_id in base_inv:
161
old_path = base_inv.id2path(file_id)
164
return ([(old_path, path, file_id, ie)], shamap)
167
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
168
revision_id, parent_invs, shagitmap, lookup_object):
96
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
169
97
"""Import a git tree object into a bzr repository.
171
:param texts: VersionedFiles object to add to
99
:param repo: A Bzr repository object
172
100
:param path: Path in the tree
173
101
:param tree: A git tree object
174
:param base_inv: Base inventory against which to return inventory delta
175
:return: Inventory delta for this subtree
102
:param inv: Inventory object
178
104
file_id = mapping.generate_file_id(path)
179
# We just have to hope this is indeed utf-8:
180
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
184
base_ie = base_inv[file_id]
186
# Newly appeared here
188
ie.revision = revision_id
189
texts.add_lines((file_id, ie.revision), (), [])
190
invdelta.append((None, path, file_id, ie))
192
base_sha = base_ie.text_id
193
# See if this has changed at all
196
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
200
if base_sha == hexsha:
201
# If nothing has changed since the base revision, we're done
203
# Remember for next time
204
existing_children = set()
207
tree = lookup_object(hexsha)
208
for mode, name, child_hexsha in tree.entries():
105
text_revision = inv.revision_id
106
repo.texts.add_lines((file_id, text_revision),
107
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
109
ie = inv.add_path(path, "directory", file_id)
110
ie.revision = text_revision
111
for mode, name, hexsha in tree.entries():
112
entry_kind = (mode & 0700000) / 0100000
209
113
basename = name.decode("utf-8")
210
existing_children.add(basename)
211
child_path = osutils.pathjoin(path, name)
212
if stat.S_ISDIR(mode):
213
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
214
mapping, child_path, child_hexsha, base_inv, file_id,
215
revision_id, parent_invs, shagitmap, lookup_object)
216
invdelta.extend(subinvdelta)
217
child_modes.update(grandchildmodes)
218
shamap.extend(subshamap)
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
mode_is_executable(mode), stat.S_ISLNK(mode))
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 and base_ie.kind == 'directory':
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,
117
child_path = urlutils.join(path, name)
119
tree = lookup_object(hexsha)
120
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
121
elif entry_kind == 1:
122
blob = lookup_object(hexsha)
123
fs_mode = mode & 0777
124
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
126
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
129
def import_git_objects(repo, mapping, num_objects, object_iter, pb=None):
243
130
"""Import a set of git objects into a bzr repository.
245
:param repo: Target Bazaar repository
132
:param repo: Bazaar repository
246
133
:param mapping: Mapping to use
134
:param num_objects: Number of objects.
247
135
:param object_iter: Iterator over Git objects.
249
def lookup_object(sha):
251
return object_iter[sha]
253
return target_git_object_retriever[sha]
254
137
# TODO: a more (memory-)efficient implementation of this
139
for i, (o, _) in enumerate(object_iter):
141
pb.update("fetching objects", i, num_objects)
260
parent_invs_cache = LRUCache(50)
261
146
# Find and convert commit objects
264
pb.update("finding revisions to fetch", len(graph), None)
266
assert isinstance(head, str)
268
o = lookup_object(head)
147
for o in objects.itervalues():
271
148
if isinstance(o, Commit):
272
149
rev = mapping.import_commit(o)
273
if repo.has_revision(rev.revision_id):
275
root_trees[rev.revision_id] = o.tree
150
root_trees[rev.revision_id] = objects[o.tree]
276
151
revisions[rev.revision_id] = rev
277
152
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
153
# Order the revisions
287
154
# Create the inventory objects
288
155
for i, revid in enumerate(topo_sort(graph)):
289
156
if pb is not None:
290
157
pb.update("fetching revisions", i, len(graph))
158
root_tree = root_trees[revid]
291
159
rev = revisions[revid]
292
160
# We have to do this here, since we have to walk the tree and
293
# we need to make sure to import the blobs / trees with the right
161
# we need to make sure to import the blobs / trees with the riht
294
162
# path; this may involve adding them more than once.
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
for path, mode in unusual_modes.iteritems():
313
warn_unusual_mode(rev.foreign_revid, path, mode)
315
basis_id = rev.parent_ids[0]
317
basis_id = NULL_REVISION
318
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
319
inv_delta, rev.revision_id, rev.parent_ids)
320
parent_invs_cache[rev.revision_id] = inv
321
repo.add_revision(rev.revision_id, rev)
322
if "verify" in debug.debug_flags:
323
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
324
for sha1, newobj, path in objs:
325
assert path is not None
326
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
327
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
329
target_git_object_retriever._idmap.commit()
332
class InterGitRepository(InterRepository):
334
_matching_repo_format = GitRepositoryFormat()
164
inv.revision_id = rev.revision_id
165
def lookup_object(sha):
168
return reconstruct_git_object(repo, mapping, sha)
169
parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
170
import_git_tree(repo, mapping, "", root_tree, inv, parent_invs,
172
repo.add_revision(rev.revision_id, rev, inv)
175
def reconstruct_git_commit(repo, rev):
176
raise NotImplementedError(self.reconstruct_git_commit)
179
def reconstruct_git_object(repo, mapping, sha):
181
revid = mapping.revision_id_foreign_to_bzr(sha)
183
rev = repo.get_revision(revid)
184
except NoSuchRevision:
187
return reconstruct_git_commit(rev)
191
raise KeyError("No such object %s" % sha)
194
class InterGitNonGitRepository(InterRepository):
196
_matching_repo_format = GitFormat()
337
199
def _get_repo_format_to_test():
341
203
"""See InterRepository.copy_content."""
342
204
self.fetch(revision_id, pb, find_ghosts=False)
344
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
346
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
347
mapping=mapping, fetch_spec=fetch_spec)
350
class InterGitNonGitRepository(InterGitRepository):
351
"""Base InterRepository that copies revisions from a Git into a non-Git
354
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
355
mapping=None, fetch_spec=None):
206
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
356
208
if mapping is None:
357
209
mapping = self.source.get_mapping()
358
if revision_id is not None:
359
interesting_heads = [revision_id]
360
elif fetch_spec is not None:
361
interesting_heads = fetch_spec.heads
363
interesting_heads = None
365
def determine_wants(refs):
367
if interesting_heads is None:
368
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
211
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
212
def determine_wants(heads):
213
if revision_id is None:
370
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
216
ret = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
371
217
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
372
self.fetch_objects(determine_wants, mapping, pb)
376
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
377
"""InterRepository that copies revisions from a remote Git into a non-Git
380
def fetch_objects(self, determine_wants, mapping, pb=None):
382
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
383
store = BazaarObjectStore(self.target, mapping)
384
self.target.lock_write()
386
heads = self.target.get_graph().heads(self.target.all_revision_ids())
387
graph_walker = store.get_graph_walker(
388
[store._lookup_revision_sha1(head) for head in heads])
391
def record_determine_wants(heads):
392
wants = determine_wants(heads)
393
recorded_wants.extend(wants)
398
create_pb = pb = ui.ui_factory.nested_progress_bar()
400
self.target.start_write_group()
402
objects_iter = self.source.fetch_objects(
403
record_determine_wants, graph_walker,
404
store.get_raw, progress)
405
import_git_objects(self.target, mapping, objects_iter,
406
store, recorded_wants, pb)
408
self.target.commit_write_group()
416
def is_compatible(source, target):
417
"""Be compatible with GitRepository."""
418
# FIXME: Also check target uses VersionedFile
419
return (isinstance(source, RemoteGitRepository) and
420
target.supports_rich_root() and
421
not isinstance(target, GitRepository))
424
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
425
"""InterRepository that copies revisions from a local Git into a non-Git
428
def fetch_objects(self, determine_wants, mapping, pb=None):
429
wants = determine_wants(self.source._git.get_refs())
218
graph_walker = BzrFetchGraphWalker(self.target, mapping)
432
221
create_pb = pb = ui.ui_factory.nested_progress_bar()
433
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
435
223
self.target.lock_write()
437
225
self.target.start_write_group()
439
import_git_objects(self.target, mapping,
440
self.source._git.object_store,
441
target_git_object_retriever, wants, pb)
227
(num_objects, objects_iter) = \
228
self.source.fetch_objects(determine_wants,
229
graph_walker, progress)
230
import_git_objects(self.target, mapping, num_objects,
443
233
self.target.commit_write_group()