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 (
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 (
68
inventory_to_tree_and_blobs,
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
71
25
from bzrlib.plugins.git.repository import (
76
from bzrlib.plugins.git.remote import (
30
from bzrlib.plugins.git.remote import RemoteGitRepository
32
from dulwich.objects import Commit
34
from cStringIO import StringIO
81
37
class BzrFetchGraphWalker(object):
82
"""GraphWalker implementation that uses a Bazaar repository."""
84
39
def __init__(self, repository, mapping):
85
40
self.repository = repository
111
63
self.heads.update([p for p in ps if not p in self.done])
113
65
self.done.add(ret)
114
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
66
return self.mapping.revision_id_bzr_to_foreign(ret)
115
67
except InvalidRevisionId:
120
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
121
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
72
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
122
73
"""Import a git blob object into a bzr repository.
124
:param texts: VersionedFiles to add to
75
:param repo: bzr repository
125
76
:param path: Path in the tree
126
77
:param blob: A git blob
127
:return: Inventory delta for this file
129
79
file_id = mapping.generate_file_id(path)
134
# We just have to hope this is indeed utf-8:
135
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
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
88
ie.executable = executable
137
# See if this has changed at all
139
base_sha = shagitmap.lookup_blob(file_id, base_inv.revision_id)
143
if (base_sha == hexsha and base_inv[file_id].executable == ie.executable
144
and base_inv[file_id].kind == ie.kind):
145
# If nothing has changed since the base revision, we're done
147
if base_sha == hexsha:
148
ie.text_size = base_inv[file_id].text_size
149
ie.text_sha1 = base_inv[file_id].text_sha1
150
ie.symlink_target = base_inv[file_id].symlink_target
151
ie.revision = base_inv[file_id].revision
153
blob = lookup_object(hexsha)
154
if ie.kind == "symlink":
155
ie.symlink_target = blob.data
159
ie.text_size = len(blob.data)
160
ie.text_sha1 = osutils.sha_string(blob.data)
161
# Check what revision we should store
163
for pinv in parent_invs:
164
if not file_id in pinv:
166
if pinv[file_id].text_sha1 == ie.text_sha1:
167
# found a revision in one of the parents to use
168
ie.revision = pinv[file_id].revision
170
parent_keys.append((file_id, pinv[file_id].revision))
171
if ie.revision is None:
172
# Need to store a new revision
173
ie.revision = revision_id
174
assert file_id is not None
175
assert ie.revision is not None
176
texts.add_lines((file_id, ie.revision), parent_keys,
177
osutils.split_lines(blob.data))
178
if file_id in base_inv:
179
old_path = base_inv.id2path(file_id)
182
return ([(old_path, path, file_id, ie)],
183
[(hexsha, "blob", (ie.file_id, ie.revision))])
186
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
187
revision_id, parent_invs, shagitmap, lookup_object):
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
188
92
"""Import a git tree object into a bzr repository.
190
:param texts: VersionedFiles object to add to
94
:param repo: A Bzr repository object
191
95
:param path: Path in the tree
192
96
:param tree: A git tree object
193
:param base_inv: Base inventory against which to return inventory delta
194
:return: Inventory delta for this subtree
97
:param inv: Inventory object
197
99
file_id = mapping.generate_file_id(path)
198
# We just have to hope this is indeed utf-8:
199
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
201
if not file_id in base_inv:
202
# Newly appeared here
203
ie.revision = revision_id
204
texts.add_lines((file_id, ie.revision), [], [])
205
invdelta.append((None, path, file_id, ie))
207
# See if this has changed at all
209
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
213
if base_sha == hexsha:
214
# If nothing has changed since the base revision, we're done
216
# Remember for next time
217
existing_children = set()
220
tree = lookup_object(hexsha)
221
for mode, name, child_hexsha in tree.entries():
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
222
108
basename = name.decode("utf-8")
223
existing_children.add(basename)
224
child_path = osutils.pathjoin(path, name)
225
if stat.S_ISDIR(mode):
226
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
227
mapping, child_path, child_hexsha, base_inv, file_id,
228
revision_id, parent_invs, shagitmap, lookup_object)
229
invdelta.extend(subinvdelta)
230
child_modes.update(grandchildmodes)
231
shamap.extend(subshamap)
233
fs_mode = stat.S_IMODE(mode)
234
symlink = stat.S_ISLNK(mode)
235
subinvdelta, subshamap = import_git_blob(texts, mapping,
236
child_path, child_hexsha, base_inv, file_id, revision_id,
237
parent_invs, shagitmap, lookup_object,
238
bool(fs_mode & 0111), symlink)
239
invdelta.extend(subinvdelta)
240
shamap.extend(subshamap)
241
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
242
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
243
child_modes[child_path] = mode
244
# Remove any children that have disappeared
245
if file_id in base_inv:
246
deletable = [v for k,v in base_inv[file_id].children.iteritems() if k not in existing_children]
249
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
250
if ie.kind == "directory":
251
deletable.extend(ie.children.values())
252
shamap.append((hexsha, "tree", (file_id, revision_id)))
253
return invdelta, child_modes, shamap
256
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
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):
258
125
"""Import a set of git objects into a bzr repository.
260
127
:param repo: Bazaar repository
262
129
:param object_iter: Iterator over Git objects.
264
131
# 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)
271
140
# Find and convert commit objects
274
pb.update("finding revisions to fetch", len(graph), None)
276
assert isinstance(head, str)
278
o = object_iter[head]
141
for o in objects.itervalues():
281
142
if isinstance(o, Commit):
282
143
rev = mapping.import_commit(o)
283
if repo.has_revision(rev.revision_id):
285
root_trees[rev.revision_id] = o.tree
144
root_trees[rev.revision_id] = objects[o.tree]
286
145
revisions[rev.revision_id] = rev
287
146
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)
296
147
# Order the revisions
297
148
# Create the inventory objects
298
149
for i, revid in enumerate(topo_sort(graph)):
299
150
if pb is not None:
300
151
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
301
153
rev = revisions[revid]
302
154
# We have to do this here, since we have to walk the tree and
303
# we need to make sure to import the blobs / trees with the right
155
# we need to make sure to import the blobs / trees with the riht
304
156
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
305
159
def lookup_object(sha):
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, shamap = import_git_tree(repo.texts,
323
mapping, "", root_trees[revid], base_inv, None, revid,
324
parent_invs, target_git_object_retriever._idmap, lookup_object)
325
target_git_object_retriever._idmap.add_entries(shamap)
326
if unusual_modes != {}:
327
ret = "unusual modes: \n"
328
for item in unusual_modes.iteritems():
329
ret += "\t%s: %o\n" % item
330
raise AssertionError(ret)
332
basis_id = rev.parent_ids[0]
334
basis_id = NULL_REVISION
335
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
336
inv_delta, rev.revision_id, rev.parent_ids)
337
parent_invs_cache[rev.revision_id] = inv
338
repo.add_revision(rev.revision_id, rev)
339
if "verify" in debug.debug_flags:
340
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
341
for sha1, newobj, path in objs:
342
assert path is not None
343
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
344
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
346
target_git_object_retriever._idmap.commit()
349
class InterGitNonGitRepository(InterRepository):
350
"""Base InterRepository that copies revisions from a Git into a non-Git
353
_matching_repo_format = GitRepositoryFormat()
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()
356
192
def _get_repo_format_to_test():
360
196
"""See InterRepository.copy_content."""
361
197
self.fetch(revision_id, pb, find_ghosts=False)
363
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
365
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
366
mapping=mapping, fetch_spec=fetch_spec)
368
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
369
mapping=None, fetch_spec=None):
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
370
201
if mapping is None:
371
202
mapping = self.source.get_mapping()
372
if revision_id is not None:
373
interesting_heads = [revision_id]
374
elif fetch_spec is not None:
375
interesting_heads = fetch_spec.heads
377
interesting_heads = None
379
def determine_wants(refs):
381
if interesting_heads is None:
382
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
384
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid != NULL_REVISION]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
385
210
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
386
self.fetch_objects(determine_wants, mapping, pb)
391
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
392
"""InterRepository that copies revisions from a remote Git into a non-Git
395
def fetch_objects(self, determine_wants, mapping, pb=None):
397
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
398
211
graph_walker = BzrFetchGraphWalker(self.target, mapping)
401
214
create_pb = pb = ui.ui_factory.nested_progress_bar()
402
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
405
def record_determine_wants(heads):
406
wants = determine_wants(heads)
407
recorded_wants.extend(wants)
411
self.target.lock_write()
413
self.target.start_write_group()
415
objects_iter = self.source.fetch_objects(
416
record_determine_wants,
418
target_git_object_retriever.get_raw,
420
import_git_objects(self.target, mapping, objects_iter,
421
target_git_object_retriever, recorded_wants, pb)
423
self.target.commit_write_group()
431
def is_compatible(source, target):
432
"""Be compatible with GitRepository."""
433
# FIXME: Also check target uses VersionedFile
434
return (isinstance(source, RemoteGitRepository) and
435
target.supports_rich_root() and
436
not isinstance(target, GitRepository))
439
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
440
"""InterRepository that copies revisions from a remote Git into a non-Git
443
def fetch_objects(self, determine_wants, mapping, pb=None):
444
wants = determine_wants(self.source._git.get_refs())
447
create_pb = pb = ui.ui_factory.nested_progress_bar()
448
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
450
self.target.lock_write()
452
self.target.start_write_group()
454
import_git_objects(self.target, mapping,
455
self.source._git.object_store,
456
target_git_object_retriever, wants, pb)
458
self.target.commit_write_group()
466
def is_compatible(source, target):
467
"""Be compatible with GitRepository."""
468
# FIXME: Also check target uses VersionedFile
469
return (isinstance(source, LocalGitRepository) and
470
target.supports_rich_root() and
471
not isinstance(target, GitRepository))
474
class InterGitRepository(InterRepository):
475
"""InterRepository that copies between Git repositories."""
477
_matching_repo_format = GitRepositoryFormat()
480
def _get_repo_format_to_test():
483
def copy_content(self, revision_id=None, pb=None):
484
"""See InterRepository.copy_content."""
485
self.fetch(revision_id, pb, find_ghosts=False)
487
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
488
mapping=None, fetch_spec=None):
490
mapping = self.source.get_mapping()
492
trace.info("git: %s", text)
494
if revision_id is not None:
495
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
496
elif fetch_spec is not None:
497
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
498
if fetch_spec is None and revision_id is None:
499
determine_wants = r.object_store.determine_wants_all
501
determine_wants = lambda x: [y for y in args if not y in r.object_store]
503
graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
504
f, commit = r.object_store.add_thin_pack()
506
self.source.fetch_pack(determine_wants, graphwalker, f.write, progress)
513
def is_compatible(source, target):
514
"""Be compatible with GitRepository."""
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
515
235
return (isinstance(source, GitRepository) and
516
isinstance(target, GitRepository))
236
target.supports_rich_root())
added
removed
fetch.py
