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 (
36
from bzrlib.errors import (
40
from bzrlib.inventory import (
46
from bzrlib.lru_cache import (
49
from bzrlib.repository import (
52
from bzrlib.revision import (
55
from bzrlib.tsort import (
59
from bzrlib.plugins.git.converter import (
62
from bzrlib.plugins.git.mapping 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 import git
65
25
from bzrlib.plugins.git.repository import (
70
from bzrlib.plugins.git.remote import (
30
from bzrlib.plugins.git.remote import RemoteGitRepository
32
from dulwich.client import SimpleFetchGraphWalker
33
from dulwich.objects import Commit
35
from cStringIO import StringIO
75
38
class BzrFetchGraphWalker(object):
76
"""GraphWalker implementation that uses a Bazaar repository."""
78
40
def __init__(self, repository, mapping):
79
41
self.repository = repository
105
64
self.heads.update([p for p in ps if not p in self.done])
107
66
self.done.add(ret)
108
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
67
return self.mapping.revision_id_bzr_to_foreign(ret)
109
68
except InvalidRevisionId:
114
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
115
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
73
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
116
74
"""Import a git blob object into a bzr repository.
118
:param texts: VersionedFiles to add to
76
:param repo: bzr repository
119
77
:param path: Path in the tree
120
78
:param blob: A git blob
121
:return: Inventory delta for this file
123
80
file_id = mapping.generate_file_id(path)
128
# We just have to hope this is indeed utf-8:
129
ie = cls(file_id, urlutils.basename(path).decode("utf-8"),
81
text_revision = inv.revision_id
82
repo.texts.add_lines((file_id, text_revision),
83
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
84
osutils.split_lines(blob.data))
85
ie = inv.add_path(path, "file", file_id)
86
ie.revision = text_revision
87
ie.text_size = len(blob.data)
88
ie.text_sha1 = osutils.sha_string(blob.data)
131
89
ie.executable = executable
132
# See if this has changed at all
134
base_sha = shagitmap.lookup_blob(file_id, base_inv.revision_id)
138
if (base_sha == hexsha and base_inv[file_id].executable == ie.executable
139
and base_inv[file_id].kind == ie.kind):
140
# If nothing has changed since the base revision, we're done
142
if base_sha == hexsha:
143
ie.text_size = base_inv[file_id].text_size
144
ie.text_sha1 = base_inv[file_id].text_sha1
145
ie.symlink_target = base_inv[file_id].symlink_target
146
ie.revision = base_inv[file_id].revision
148
blob = lookup_object(hexsha)
149
if ie.kind == "symlink":
150
ie.symlink_target = blob.data
154
ie.text_size = len(blob.data)
155
ie.text_sha1 = osutils.sha_string(blob.data)
156
# Check what revision we should store
158
for pinv in parent_invs:
159
if not file_id in pinv:
161
if pinv[file_id].text_sha1 == ie.text_sha1:
162
# found a revision in one of the parents to use
163
ie.revision = pinv[file_id].revision
165
parent_keys.append((file_id, pinv[file_id].revision))
166
if ie.revision is None:
167
# Need to store a new revision
168
ie.revision = revision_id
169
assert file_id is not None
170
assert ie.revision is not None
171
texts.add_lines((file_id, ie.revision), parent_keys,
172
osutils.split_lines(blob.data))
173
if "verify" in debug.debug_flags:
174
assert text_to_blob(blob.data).id == hexsha
175
shagitmap.add_entry(hexsha, "blob", (ie.file_id, ie.revision))
176
if file_id in base_inv:
177
old_path = base_inv.id2path(file_id)
180
return [(old_path, path, file_id, ie)]
183
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
184
revision_id, parent_invs, shagitmap, lookup_object):
92
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
185
93
"""Import a git tree object into a bzr repository.
187
:param texts: VersionedFiles object to add to
95
:param repo: A Bzr repository object
188
96
:param path: Path in the tree
189
97
:param tree: A git tree object
190
:param base_inv: Base inventory against which to return inventory delta
191
:return: Inventory delta for this subtree
98
:param inv: Inventory object
194
100
file_id = mapping.generate_file_id(path)
195
# We just have to hope this is indeed utf-8:
196
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
198
if not file_id in base_inv:
199
# Newly appeared here
200
ie.revision = revision_id
201
texts.add_lines((file_id, ie.revision), [], [])
202
ret.append((None, path, file_id, ie))
204
# See if this has changed at all
206
base_sha = shagitmap.lookup_tree(path, base_inv.revision_id)
210
if base_sha == hexsha:
211
# If nothing has changed since the base revision, we're done
213
# Remember for next time
214
existing_children = set()
215
if "verify" in debug.debug_flags:
218
shagitmap.add_entry(hexsha, "tree", (file_id, revision_id))
219
tree = lookup_object(hexsha)
101
text_revision = inv.revision_id
102
repo.texts.add_lines((file_id, text_revision),
103
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
105
ie = inv.add_path(path, "directory", file_id)
106
ie.revision = text_revision
220
107
for mode, name, hexsha in tree.entries():
221
108
entry_kind = (mode & 0700000) / 0100000
222
109
basename = name.decode("utf-8")
223
existing_children.add(basename)
225
111
child_path = name
227
113
child_path = urlutils.join(path, name)
228
114
if entry_kind == 0:
229
ret.extend(import_git_tree(texts, mapping, child_path, hexsha, base_inv,
230
file_id, revision_id, parent_invs, shagitmap, lookup_object))
115
tree = lookup_object(hexsha)
116
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
231
117
elif entry_kind == 1:
118
blob = lookup_object(hexsha)
232
119
fs_mode = mode & 0777
233
file_kind = (mode & 070000) / 010000
234
if file_kind == 0: # regular file
239
raise AssertionError("Unknown file kind, mode=%r" % (mode,))
240
ret.extend(import_git_blob(texts, mapping, child_path, hexsha, base_inv,
241
file_id, revision_id, parent_invs, shagitmap, lookup_object,
242
bool(fs_mode & 0111), symlink))
120
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
244
raise AssertionError("Unknown object kind, perms=%r." % (mode,))
245
# Remove any children that have disappeared
246
if file_id in base_inv:
247
deletable = [v for k,v in base_inv[file_id].children.iteritems() if k not in existing_children]
250
ret.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
251
if ie.kind == "directory":
252
deletable.extend(ie.children.values())
256
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
122
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
125
def import_git_objects(repo, mapping, object_iter, pb=None):
258
126
"""Import a set of git objects into a bzr repository.
260
128
:param repo: Bazaar repository
262
130
:param object_iter: Iterator over Git objects.
264
132
# TODO: a more (memory-)efficient implementation of this
134
for i, o in enumerate(object_iter):
136
pb.update("fetching objects", i)
270
parent_invs_cache = LRUCache(50)
271
141
# 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]
142
for o in objects.itervalues():
281
143
if isinstance(o, Commit):
282
144
rev = mapping.import_commit(o)
283
if repo.has_revision(rev.revision_id):
285
root_trees[rev.revision_id] = o.tree
145
root_trees[rev.revision_id] = objects[o.tree]
286
146
revisions[rev.revision_id] = rev
287
147
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
148
# Order the revisions
297
149
# Create the inventory objects
298
150
for i, revid in enumerate(topo_sort(graph)):
299
151
if pb is not None:
300
152
pb.update("fetching revisions", i, len(graph))
153
root_tree = root_trees[revid]
301
154
rev = revisions[revid]
302
155
# 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
156
# we need to make sure to import the blobs / trees with the riht
304
157
# path; this may involve adding them more than once.
159
inv.revision_id = rev.revision_id
305
160
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 = import_git_tree(repo.texts, mapping, "",
323
root_trees[revid], base_inv, None, revid, parent_invs,
324
target_git_object_retriever._idmap, lookup_object)
326
basis_id = rev.parent_ids[0]
328
basis_id = NULL_REVISION
329
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
330
inv_delta, rev.revision_id, rev.parent_ids)
331
parent_invs_cache[rev.revision_id] = inv
332
repo.add_revision(rev.revision_id, rev)
333
target_git_object_retriever._idmap.commit()
163
return reconstruct_git_object(repo, mapping, sha)
164
parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
165
import_git_tree(repo, mapping, "", root_tree, inv, parent_invs,
167
repo.add_revision(rev.revision_id, rev, inv)
170
def reconstruct_git_commit(repo, rev):
171
raise NotImplementedError(self.reconstruct_git_commit)
174
def reconstruct_git_object(repo, mapping, sha):
176
revid = mapping.revision_id_foreign_to_bzr(sha)
178
rev = repo.get_revision(revid)
179
except NoSuchRevision:
182
return reconstruct_git_commit(rev)
186
raise KeyError("No such object %s" % sha)
336
189
class InterGitNonGitRepository(InterRepository):
337
"""Base InterRepository that copies revisions from a Git into a non-Git
340
_matching_repo_format = GitRepositoryFormat()
191
_matching_repo_format = GitFormat()
343
194
def _get_repo_format_to_test():
347
198
"""See InterRepository.copy_content."""
348
199
self.fetch(revision_id, pb, find_ghosts=False)
350
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
352
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
353
mapping=mapping, fetch_spec=fetch_spec)
355
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
356
mapping=None, fetch_spec=None):
201
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
357
203
if mapping is None:
358
204
mapping = self.source.get_mapping()
359
if revision_id is not None:
360
interesting_heads = [revision_id]
361
elif fetch_spec is not None:
362
interesting_heads = fetch_spec.heads
364
interesting_heads = None
366
def determine_wants(refs):
368
if interesting_heads is None:
369
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
206
pb.note("git: %s", text)
207
def determine_wants(heads):
208
if revision_id is None:
371
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads]
211
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
372
212
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
373
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
213
graph_walker = BzrFetchGraphWalker(self.target, mapping)
388
216
create_pb = pb = ui.ui_factory.nested_progress_bar()
389
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
392
def record_determine_wants(heads):
393
wants = determine_wants(heads)
394
recorded_wants.extend(wants)
398
self.target.lock_write()
400
self.target.start_write_group()
402
objects_iter = self.source.fetch_objects(
403
record_determine_wants,
405
target_git_object_retriever.get_raw,
407
import_git_objects(self.target, mapping, objects_iter,
408
target_git_object_retriever, recorded_wants, pb)
410
self.target.commit_write_group()
418
def is_compatible(source, target):
419
"""Be compatible with GitRepository."""
420
# FIXME: Also check target uses VersionedFile
421
return (isinstance(source, RemoteGitRepository) and
422
target.supports_rich_root() and
423
not isinstance(target, GitRepository))
426
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
427
"""InterRepository that copies revisions from a remote Git into a non-Git
430
def fetch_objects(self, determine_wants, mapping, pb=None):
431
wants = determine_wants(self.source._git.get_refs())
434
create_pb = pb = ui.ui_factory.nested_progress_bar()
435
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
437
self.target.lock_write()
439
self.target.start_write_group()
441
import_git_objects(self.target, mapping,
442
self.source._git.object_store,
443
target_git_object_retriever, wants, pb)
445
self.target.commit_write_group()
453
def is_compatible(source, target):
454
"""Be compatible with GitRepository."""
455
# FIXME: Also check target uses VersionedFile
456
return (isinstance(source, LocalGitRepository) and
218
self.target.lock_write()
220
self.target.start_write_group()
222
import_git_objects(self.target, mapping,
223
iter(self.source.fetch_objects(determine_wants, graph_walker,
226
self.target.commit_write_group()
234
def is_compatible(source, target):
235
"""Be compatible with GitRepository."""
236
# FIXME: Also check target uses VersionedFile
237
return (isinstance(source, GitRepository) and
457
238
target.supports_rich_root() and
458
239
not isinstance(target, GitRepository))
461
242
class InterGitRepository(InterRepository):
462
"""InterRepository that copies between Git repositories."""
464
_matching_repo_format = GitRepositoryFormat()
244
_matching_repo_format = GitFormat()
467
247
def _get_repo_format_to_test():
added
removed
fetch.py
