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 (
35
from bzrlib.errors import (
39
from bzrlib.inventory import (
45
from bzrlib.lru_cache import (
48
from bzrlib.repository import (
51
from bzrlib.revision import (
54
from bzrlib.tsort import (
58
from bzrlib.plugins.git.converter import (
25
61
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
66
from bzrlib.plugins.git.remote import (
37
71
class BzrFetchGraphWalker(object):
72
"""GraphWalker implementation that uses a Bazaar repository."""
39
74
def __init__(self, repository, mapping):
40
75
self.repository = repository
63
101
self.heads.update([p for p in ps if not p in self.done])
65
103
self.done.add(ret)
66
return self.mapping.revision_id_bzr_to_foreign(ret)
104
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
67
105
except InvalidRevisionId:
72
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
110
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
111
revision_id, parent_invs, shagitmap, lookup_object, executable):
73
112
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
114
:param texts: VersionedFiles to add to
76
115
:param path: Path in the tree
77
116
:param blob: A git blob
117
:return: Inventory delta for this file
79
119
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)
120
# See if this has changed at all
122
base_sha = shagitmap.lookup_blob(file_id, base_inv.revision_id)
126
if base_sha == hexsha and base_inv[file_id].executable == executable:
127
# If nothing has changed since the base revision, we're done
129
# We just have to hope this is indeed utf-8:
130
ie = InventoryFile(file_id, urlutils.basename(path).decode("utf-8"),
132
if base_sha == hexsha:
133
ie.text_size = base_inv[file_id].text_size
134
ie.text_sha1 = base_inv[file_id].text_sha1
135
ie.revision = base_inv[file_id].revision
137
blob = lookup_object(hexsha)
138
ie.text_size = len(blob.data)
139
ie.text_sha1 = osutils.sha_string(blob.data)
88
140
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
141
# Check what revision we should store
143
for pinv in parent_invs:
144
if not file_id in pinv:
146
if pinv[file_id].text_sha1 == ie.text_sha1:
147
# found a revision in one of the parents to use
148
ie.revision = pinv[file_id].revision
150
parent_keys.append((file_id, pinv[file_id].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.add_lines((file_id, ie.revision), parent_keys,
157
osutils.split_lines(blob.data))
158
shagitmap.add_entry(hexsha, "blob", (ie.file_id, ie.revision))
159
if file_id in base_inv:
160
old_path = base_inv.id2path(file_id)
163
return [(old_path, path, file_id, ie)]
166
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
167
revision_id, parent_invs, shagitmap, lookup_object):
92
168
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
170
:param texts: VersionedFiles object to add to
95
171
:param path: Path in the tree
96
172
:param tree: A git tree object
97
:param inv: Inventory object
173
:param base_inv: Base inventory against which to return inventory delta
174
:return: Inventory delta for this subtree
99
177
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
178
# We just have to hope this is indeed utf-8:
179
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
181
if not file_id in base_inv:
182
# Newly appeared here
183
ie.revision = revision_id
184
texts.add_lines((file_id, ie.revision), [], [])
185
ret.append((None, path, file_id, ie))
187
# See if this has changed at all
189
base_sha = shagitmap.lookup_tree(path, base_inv.revision_id)
193
if base_sha == hexsha:
194
# If nothing has changed since the base revision, we're done
196
# Remember for next time
197
existing_children = set()
198
shagitmap.add_entry(hexsha, "tree", (file_id, revision_id))
199
tree = lookup_object(hexsha)
106
200
for mode, name, hexsha in tree.entries():
107
201
entry_kind = (mode & 0700000) / 0100000
108
202
basename = name.decode("utf-8")
203
existing_children.add(basename)
110
205
child_path = name
112
207
child_path = urlutils.join(path, name)
113
208
if entry_kind == 0:
114
tree = lookup_object(hexsha)
115
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
209
ret.extend(import_git_tree(texts, mapping, child_path, hexsha, base_inv,
210
file_id, revision_id, parent_invs, shagitmap, lookup_object))
116
211
elif entry_kind == 1:
117
blob = lookup_object(hexsha)
118
212
fs_mode = mode & 0777
119
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
213
ret.extend(import_git_blob(texts, mapping, child_path, hexsha, base_inv,
214
file_id, revision_id, parent_invs, shagitmap, lookup_object,
215
bool(fs_mode & 0111)))
121
217
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
124
def import_git_objects(repo, mapping, object_iter, pb=None):
218
# Remove any children that have disappeared
219
if file_id in base_inv:
220
deletable = [v for k,v in base_inv[file_id].children.iteritems() if k not in existing_children]
223
ret.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
224
if ie.kind == "directory":
225
deletable.extend(ie.children.values())
229
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
231
"""Import a set of git objects into a bzr repository.
127
233
:param repo: Bazaar repository
129
235
:param object_iter: Iterator over Git objects.
131
237
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
243
parent_invs_cache = LRUCache(50)
140
244
# Find and convert commit objects
141
for o in objects.itervalues():
247
pb.update("finding revisions to fetch", len(graph), None)
249
assert isinstance(head, str)
251
o = object_iter[head]
142
254
if isinstance(o, Commit):
143
255
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
256
if repo.has_revision(rev.revision_id):
258
root_trees[rev.revision_id] = o.tree
145
259
revisions[rev.revision_id] = rev
146
260
graph.append((rev.revision_id, rev.parent_ids))
261
target_git_object_retriever._idmap.add_entry(o.sha().hexdigest(),
262
"commit", (rev.revision_id, o._tree))
263
heads.extend([p for p in o.parents if p not in checked])
264
elif isinstance(o, Tag):
265
heads.append(o.object[1])
267
trace.warning("Unable to import head object %r" % o)
147
269
# Order the revisions
148
270
# Create the inventory objects
149
271
for i, revid in enumerate(topo_sort(graph)):
150
272
if pb is not None:
151
273
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
274
rev = revisions[revid]
154
275
# 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
276
# we need to make sure to import the blobs / trees with the right
156
277
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
278
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()
280
return object_iter[sha]
282
return target_git_object_retriever[sha]
284
for parent_id in rev.parent_ids:
286
parent_invs.append(parent_invs_cache[parent_id])
288
parent_inv = repo.get_inventory(parent_id)
289
parent_invs.append(parent_inv)
290
parent_invs_cache[parent_id] = parent_inv
291
if parent_invs == []:
292
base_inv = Inventory(root_id=None)
294
base_inv = parent_invs[0]
295
inv_delta = import_git_tree(repo.texts, mapping, "",
296
root_trees[revid], base_inv, None, revid, parent_invs,
297
target_git_object_retriever._idmap, lookup_object)
299
basis_id = rev.parent_ids[0]
301
basis_id = NULL_REVISION
302
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
303
inv_delta, rev.revision_id, rev.parent_ids)
304
parent_invs_cache[rev.revision_id] = inv
305
repo.add_revision(rev.revision_id, rev)
306
target_git_object_retriever._idmap.commit()
309
class InterGitNonGitRepository(InterRepository):
310
"""Base InterRepository that copies revisions from a Git into a non-Git
313
_matching_repo_format = GitRepositoryFormat()
192
316
def _get_repo_format_to_test():
196
320
"""See InterRepository.copy_content."""
197
321
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
323
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
325
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
326
mapping=mapping, fetch_spec=fetch_spec)
328
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
329
mapping=None, fetch_spec=None):
201
330
if mapping is None:
202
331
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
332
if revision_id is not None:
333
interesting_heads = [revision_id]
334
elif fetch_spec is not None:
335
interesting_heads = fetch_spec.heads
337
interesting_heads = None
339
def determine_wants(refs):
341
if interesting_heads is None:
342
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
344
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads]
210
345
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
346
self.fetch_objects(determine_wants, mapping, pb)
351
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
352
"""InterRepository that copies revisions from a remote Git into a non-Git
355
def fetch_objects(self, determine_wants, mapping, pb=None):
357
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
211
358
graph_walker = BzrFetchGraphWalker(self.target, mapping)
214
361
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
362
target_git_object_retriever = GitObjectConverter(self.target, mapping)
365
def record_determine_wants(heads):
366
wants = determine_wants(heads)
367
recorded_wants.extend(wants)
371
self.target.lock_write()
373
self.target.start_write_group()
375
objects_iter = self.source.fetch_objects(
376
record_determine_wants,
378
target_git_object_retriever.get_raw,
380
import_git_objects(self.target, mapping, objects_iter,
381
target_git_object_retriever, recorded_wants, pb)
383
self.target.commit_write_group()
391
def is_compatible(source, target):
392
"""Be compatible with GitRepository."""
393
# FIXME: Also check target uses VersionedFile
394
return (isinstance(source, RemoteGitRepository) and
395
target.supports_rich_root() and
396
not isinstance(target, GitRepository))
399
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
400
"""InterRepository that copies revisions from a remote Git into a non-Git
403
def fetch_objects(self, determine_wants, mapping, pb=None):
404
wants = determine_wants(self.source._git.get_refs())
407
create_pb = pb = ui.ui_factory.nested_progress_bar()
408
target_git_object_retriever = GitObjectConverter(self.target, mapping)
410
self.target.lock_write()
412
self.target.start_write_group()
414
import_git_objects(self.target, mapping,
415
self.source._git.object_store,
416
target_git_object_retriever, wants, pb)
418
self.target.commit_write_group()
426
def is_compatible(source, target):
427
"""Be compatible with GitRepository."""
428
# FIXME: Also check target uses VersionedFile
429
return (isinstance(source, LocalGitRepository) and
430
target.supports_rich_root() and
431
not isinstance(target, GitRepository))
434
class InterGitRepository(InterRepository):
435
"""InterRepository that copies between Git repositories."""
437
_matching_repo_format = GitRepositoryFormat()
440
def _get_repo_format_to_test():
443
def copy_content(self, revision_id=None, pb=None):
444
"""See InterRepository.copy_content."""
445
self.fetch(revision_id, pb, find_ghosts=False)
447
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
448
mapping=None, fetch_spec=None):
450
mapping = self.source.get_mapping()
452
trace.info("git: %s", text)
454
if revision_id is not None:
455
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
456
elif fetch_spec is not None:
457
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
458
if fetch_spec is None and revision_id is None:
459
determine_wants = r.object_store.determine_wants_all
461
determine_wants = lambda x: [y for y in args if not y in r.object_store]
463
graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
464
f, commit = r.object_store.add_pack()
466
self.source._git.fetch_pack(path, determine_wants, graphwalker, f.write, progress)
474
def is_compatible(source, target):
475
"""Be compatible with GitRepository."""
235
476
return (isinstance(source, GitRepository) and
236
target.supports_rich_root())
477
isinstance(target, GitRepository))