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
17
from cStringIO import (
21
from dulwich.objects import (
26
from dulwich.object_store import (
38
from bzrlib.errors import (
43
from bzrlib.inventory import (
49
from bzrlib.lru_cache import (
52
from bzrlib.repository import (
55
from bzrlib.revision import (
58
from bzrlib.tsort import (
61
from bzrlib.versionedfile import (
62
FulltextContentFactory,
65
from bzrlib.plugins.git.mapping import (
67
inventory_to_tree_and_blobs,
73
from bzrlib.plugins.git.object_store import (
76
from bzrlib.plugins.git.remote import (
24
79
from bzrlib.plugins.git.repository import (
29
from bzrlib.plugins.git.remote import RemoteGitRepository
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
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
87
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
78
88
"""Import a git blob object into a bzr repository.
80
:param repo: bzr repository
90
:param texts: VersionedFiles to add to
81
91
:param path: Path in the tree
82
92
:param blob: A git blob
93
:return: Inventory delta for this file
84
95
file_id = mapping.generate_file_id(path)
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
# We just have to hope this is indeed utf-8:
101
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
93
102
ie.executable = executable
96
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
103
# See if this has changed at all
105
base_ie = base_inv[file_id]
111
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
115
if (base_sha == hexsha and base_ie.executable == ie.executable
116
and base_ie.kind == ie.kind):
117
# If nothing has changed since the base revision, we're done
119
if base_sha == hexsha and base_ie.kind == ie.kind:
120
ie.text_size = base_ie.text_size
121
ie.text_sha1 = base_ie.text_sha1
122
ie.symlink_target = base_ie.symlink_target
123
if ie.executable == base_ie.executable:
124
ie.revision = base_ie.revision
126
blob = lookup_object(hexsha)
128
blob = lookup_object(hexsha)
129
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 and pie.symlink_target == ie.symlink_target:
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
invdelta = [(old_path, path, file_id, ie)]
168
invdelta.extend(remove_disappeared_children(base_inv, base_ie, []))
169
return (invdelta, shamap)
172
def import_git_submodule(texts, mapping, path, hexsha, base_inv, parent_id,
173
revision_id, parent_invs, shagitmap, lookup_object):
174
raise NotImplementedError(import_git_submodule)
177
def remove_disappeared_children(base_inv, base_ie, existing_children):
178
if base_ie is None or base_ie.kind != 'directory':
181
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
184
ret.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
185
if ie.kind == "directory":
186
deletable.extend(ie.children.values())
190
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
191
revision_id, parent_invs, shagitmap, lookup_object):
97
192
"""Import a git tree object into a bzr repository.
99
:param repo: A Bzr repository object
194
:param texts: VersionedFiles object to add to
100
195
:param path: Path in the tree
101
196
:param tree: A git tree object
102
:param inv: Inventory object
197
:param base_inv: Base inventory against which to return inventory delta
198
:return: Inventory delta for this subtree
104
201
file_id = mapping.generate_file_id(path)
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
202
# We just have to hope this is indeed utf-8:
203
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
206
base_ie = base_inv[file_id]
208
# Newly appeared here
210
ie.revision = revision_id
211
texts.add_lines((file_id, ie.revision), (), [])
212
invdelta.append((None, path, file_id, ie))
214
# See if this has changed at all
216
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
220
if base_sha == hexsha:
221
# If nothing has changed since the base revision, we're done
223
if base_ie.kind != "directory":
224
ie.revision = revision_id
225
texts.add_lines((ie.file_id, ie.revision), (), [])
226
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
227
# Remember for next time
228
existing_children = set()
231
tree = lookup_object(hexsha)
232
for mode, name, child_hexsha in tree.entries():
113
233
basename = name.decode("utf-8")
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):
234
existing_children.add(basename)
235
child_path = osutils.pathjoin(path, name)
236
if stat.S_ISDIR(mode):
237
subinvdelta, grandchildmodes, subshamap = import_git_tree(
238
texts, mapping, child_path, child_hexsha, base_inv,
239
file_id, revision_id, parent_invs, shagitmap, lookup_object)
240
invdelta.extend(subinvdelta)
241
child_modes.update(grandchildmodes)
242
shamap.extend(subshamap)
243
elif S_ISGITLINK(mode): # submodule
244
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
245
texts, mapping, child_path, child_hexsha, base_inv,
246
file_id, revision_id, parent_invs, shagitmap, lookup_object)
247
invdelta.extend(subinvdelta)
248
child_modes.update(grandchildmodes)
249
shamap.extend(subshamap)
251
subinvdelta, subshamap = import_git_blob(texts, mapping,
252
child_path, child_hexsha, base_inv, file_id, revision_id,
253
parent_invs, shagitmap, lookup_object,
254
mode_is_executable(mode), stat.S_ISLNK(mode))
255
invdelta.extend(subinvdelta)
256
shamap.extend(subshamap)
257
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
258
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
259
child_modes[child_path] = mode
260
# Remove any children that have disappeared
261
invdelta.extend(remove_disappeared_children(base_inv, base_ie, existing_children))
262
shamap.append((hexsha, "tree", (file_id, revision_id)))
263
return invdelta, child_modes, shamap
266
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
130
268
"""Import a set of git objects into a bzr repository.
132
:param repo: Bazaar repository
270
:param repo: Target Bazaar repository
133
271
:param mapping: Mapping to use
134
:param num_objects: Number of objects.
135
272
:param object_iter: Iterator over Git objects.
274
def lookup_object(sha):
276
return object_iter[sha]
278
return target_git_object_retriever[sha]
137
279
# TODO: a more (memory-)efficient implementation of this
139
for i, (o, _) in enumerate(object_iter):
141
pb.update("fetching objects", i, num_objects)
285
parent_invs_cache = LRUCache(50)
146
286
# Find and convert commit objects
147
for o in objects.itervalues():
289
pb.update("finding revisions to fetch", len(graph), None)
291
assert isinstance(head, str)
293
o = lookup_object(head)
148
296
if isinstance(o, Commit):
149
297
rev = mapping.import_commit(o)
150
root_trees[rev.revision_id] = objects[o.tree]
298
if repo.has_revision(rev.revision_id):
300
squash_revision(repo, rev)
301
root_trees[rev.revision_id] = o.tree
151
302
revisions[rev.revision_id] = rev
152
303
graph.append((rev.revision_id, rev.parent_ids))
304
target_git_object_retriever._idmap.add_entry(o.id, "commit",
305
(rev.revision_id, o.tree))
306
heads.extend([p for p in o.parents if p not in checked])
307
elif isinstance(o, Tag):
308
heads.append(o.object[1])
310
trace.warning("Unable to import head object %r" % o)
153
312
# Order the revisions
154
313
# Create the inventory objects
155
314
for i, revid in enumerate(topo_sort(graph)):
156
315
if pb is not None:
157
316
pb.update("fetching revisions", i, len(graph))
158
root_tree = root_trees[revid]
159
317
rev = revisions[revid]
160
318
# We have to do this here, since we have to walk the tree and
161
# we need to make sure to import the blobs / trees with the riht
319
# we need to make sure to import the blobs / trees with the right
162
320
# path; this may involve adding them more than once.
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()
322
for parent_id in rev.parent_ids:
324
parent_invs.append(parent_invs_cache[parent_id])
326
parent_inv = repo.get_inventory(parent_id)
327
parent_invs.append(parent_inv)
328
parent_invs_cache[parent_id] = parent_inv
329
if parent_invs == []:
330
base_inv = Inventory(root_id=None)
332
base_inv = parent_invs[0]
333
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
334
mapping, "", root_trees[revid], base_inv, None, revid,
335
parent_invs, target_git_object_retriever._idmap, lookup_object)
336
target_git_object_retriever._idmap.add_entries(shamap)
337
if unusual_modes != {}:
338
for path, mode in unusual_modes.iteritems():
339
warn_unusual_mode(rev.foreign_revid, path, mode)
340
mapping.import_unusual_file_modes(rev, unusual_modes)
342
basis_id = rev.parent_ids[0]
344
basis_id = NULL_REVISION
345
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
346
inv_delta, rev.revision_id, rev.parent_ids)
347
parent_invs_cache[rev.revision_id] = inv
348
repo.add_revision(rev.revision_id, rev)
349
if "verify" in debug.debug_flags:
350
new_unusual_modes = mapping.export_unusual_file_modes(rev)
351
if new_unusual_modes != unusual_modes:
352
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
353
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
354
for sha1, newobj, path in objs:
355
assert path is not None
356
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
358
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
360
target_git_object_retriever._idmap.commit()
363
class InterGitRepository(InterRepository):
365
_matching_repo_format = GitRepositoryFormat()
199
368
def _get_repo_format_to_test():
203
372
"""See InterRepository.copy_content."""
204
373
self.fetch(revision_id, pb, find_ghosts=False)
206
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
375
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
377
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
378
mapping=mapping, fetch_spec=fetch_spec)
381
class InterGitNonGitRepository(InterGitRepository):
382
"""Base InterRepository that copies revisions from a Git into a non-Git
385
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
386
mapping=None, fetch_spec=None):
208
387
if mapping is None:
209
388
mapping = self.source.get_mapping()
211
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
212
def determine_wants(heads):
213
if revision_id is None:
389
if revision_id is not None:
390
interesting_heads = [revision_id]
391
elif fetch_spec is not None:
392
interesting_heads = fetch_spec.heads
394
interesting_heads = None
396
def determine_wants(refs):
398
if interesting_heads is None:
399
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
216
ret = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
401
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
217
402
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
218
graph_walker = BzrFetchGraphWalker(self.target, mapping)
403
self.fetch_objects(determine_wants, mapping, pb)
407
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
408
"""InterRepository that copies revisions from a remote Git into a non-Git
411
def fetch_objects(self, determine_wants, mapping, pb=None):
413
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
414
store = BazaarObjectStore(self.target, mapping)
415
self.target.lock_write()
417
heads = self.target.get_graph().heads(self.target.all_revision_ids())
418
graph_walker = store.get_graph_walker(
419
[store._lookup_revision_sha1(head) for head in heads])
422
def record_determine_wants(heads):
423
wants = determine_wants(heads)
424
recorded_wants.extend(wants)
429
create_pb = pb = ui.ui_factory.nested_progress_bar()
431
self.target.start_write_group()
433
objects_iter = self.source.fetch_objects(
434
record_determine_wants, graph_walker,
435
store.get_raw, progress)
436
import_git_objects(self.target, mapping, objects_iter,
437
store, recorded_wants, pb)
439
self.target.commit_write_group()
447
def is_compatible(source, target):
448
"""Be compatible with GitRepository."""
449
# FIXME: Also check target uses VersionedFile
450
return (isinstance(source, RemoteGitRepository) and
451
target.supports_rich_root() and
452
not isinstance(target, GitRepository))
455
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
456
"""InterRepository that copies revisions from a local Git into a non-Git
459
def fetch_objects(self, determine_wants, mapping, pb=None):
460
wants = determine_wants(self.source._git.get_refs())
221
463
create_pb = pb = ui.ui_factory.nested_progress_bar()
464
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
223
466
self.target.lock_write()
225
468
self.target.start_write_group()
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,
470
import_git_objects(self.target, mapping,
471
self.source._git.object_store,
472
target_git_object_retriever, wants, pb)
233
474
self.target.commit_write_group()