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.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 (
25
79
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
37
class BzrFetchGraphWalker(object):
39
def __init__(self, repository, mapping):
40
self.repository = repository
41
self.mapping = mapping
43
self.heads = set(repository.all_revision_ids())
47
revid = self.mapping.revision_id_foreign_to_bzr(sha)
50
def remove(self, revid):
53
self.heads.remove(revid)
54
if revid in self.parents:
55
for p in self.parents[revid]:
60
ret = self.heads.pop()
61
ps = self.repository.get_parent_map([ret])[ret]
62
self.parents[ret] = ps
63
self.heads.update([p for p in ps if not p in self.done])
66
return self.mapping.revision_id_bzr_to_foreign(ret)
67
except InvalidRevisionId:
72
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):
73
88
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
90
:param texts: VersionedFiles to add to
76
91
:param path: Path in the tree
77
92
:param blob: A git blob
93
:return: Inventory delta for this file
79
95
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)
100
# We just have to hope this is indeed utf-8:
101
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
102
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
104
# See if this has changed at all
106
base_ie = base_inv[file_id]
111
base_sha = base_ie.text_id
114
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
118
if (base_sha == hexsha and base_ie.executable == ie.executable
119
and base_ie.kind == ie.kind):
120
# If nothing has changed since the base revision, we're done
122
if base_sha == hexsha and base_ie.kind == ie.kind:
123
ie.text_size = base_ie.text_size
124
ie.text_sha1 = base_ie.text_sha1
125
ie.symlink_target = base_ie.symlink_target
126
if ie.executable == base_ie.executable:
127
ie.revision = base_ie.revision
129
blob = lookup_object(hexsha)
131
blob = lookup_object(hexsha)
132
if ie.kind == "symlink":
133
ie.symlink_target = blob.data
137
ie.text_size = len(blob.data)
138
ie.text_sha1 = osutils.sha_string(blob.data)
139
# Check what revision we should store
141
for pinv in parent_invs:
142
if pinv.revision_id == base_inv.revision_id:
151
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable:
152
# found a revision in one of the parents to use
153
ie.revision = pie.revision
155
parent_keys.append((file_id, pie.revision))
156
if ie.revision is None:
157
# Need to store a new revision
158
ie.revision = revision_id
159
assert file_id is not None
160
assert ie.revision is not None
161
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
162
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
165
if file_id in base_inv:
166
old_path = base_inv.id2path(file_id)
169
return ([(old_path, path, file_id, ie)], 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 import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
178
revision_id, parent_invs, shagitmap, lookup_object):
92
179
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
181
:param texts: VersionedFiles object to add to
95
182
:param path: Path in the tree
96
183
:param tree: A git tree object
97
:param inv: Inventory object
184
:param base_inv: Base inventory against which to return inventory delta
185
:return: Inventory delta for this subtree
99
188
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
106
for mode, name, hexsha in tree.entries():
107
entry_kind = (mode & 0700000) / 0100000
189
# We just have to hope this is indeed utf-8:
190
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
194
base_ie = base_inv[file_id]
196
# Newly appeared here
198
ie.revision = revision_id
199
texts.add_lines((file_id, ie.revision), (), [])
200
invdelta.append((None, path, file_id, ie))
202
base_sha = base_ie.text_id
203
# See if this has changed at all
206
base_sha = shagitmap.lookup_tree(file_id, 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()
217
tree = lookup_object(hexsha)
218
for mode, name, child_hexsha in tree.entries():
108
219
basename = name.decode("utf-8")
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):
220
existing_children.add(basename)
221
child_path = osutils.pathjoin(path, name)
222
if stat.S_ISDIR(mode):
223
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
224
mapping, child_path, child_hexsha, base_inv, file_id,
225
revision_id, parent_invs, shagitmap, lookup_object)
226
invdelta.extend(subinvdelta)
227
child_modes.update(grandchildmodes)
228
shamap.extend(subshamap)
229
elif S_ISGITLINK(mode): # submodule
230
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
231
texts, mapping, child_path, child_hexsha, base_inv,
232
file_id, revision_id, parent_invs, shagitmap, lookup_object)
233
invdelta.extend(subinvdelta)
234
child_modes.update(grandchildmodes)
235
shamap.extend(subshamap)
237
subinvdelta, subshamap = import_git_blob(texts, mapping,
238
child_path, child_hexsha, base_inv, file_id, revision_id,
239
parent_invs, shagitmap, lookup_object,
240
mode_is_executable(mode), stat.S_ISLNK(mode))
241
invdelta.extend(subinvdelta)
242
shamap.extend(subshamap)
243
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
244
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
245
child_modes[child_path] = mode
246
# Remove any children that have disappeared
247
if base_ie is not None and base_ie.kind == 'directory':
248
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
251
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
252
if ie.kind == "directory":
253
deletable.extend(ie.children.values())
254
shamap.append((hexsha, "tree", (file_id, revision_id)))
255
return invdelta, child_modes, shamap
258
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
260
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
262
:param repo: Target Bazaar repository
128
263
:param mapping: Mapping to use
129
264
:param object_iter: Iterator over Git objects.
266
def lookup_object(sha):
268
return object_iter[sha]
270
return target_git_object_retriever[sha]
131
271
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
277
parent_invs_cache = LRUCache(50)
140
278
# Find and convert commit objects
141
for o in objects.itervalues():
281
pb.update("finding revisions to fetch", len(graph), None)
283
assert isinstance(head, str)
285
o = lookup_object(head)
142
288
if isinstance(o, Commit):
143
289
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
290
if repo.has_revision(rev.revision_id):
292
squash_revision(repo, rev)
293
root_trees[rev.revision_id] = o.tree
145
294
revisions[rev.revision_id] = rev
146
295
graph.append((rev.revision_id, rev.parent_ids))
296
target_git_object_retriever._idmap.add_entry(o.id, "commit",
297
(rev.revision_id, o.tree))
298
heads.extend([p for p in o.parents if p not in checked])
299
elif isinstance(o, Tag):
300
heads.append(o.object[1])
302
trace.warning("Unable to import head object %r" % o)
147
304
# Order the revisions
148
305
# Create the inventory objects
149
306
for i, revid in enumerate(topo_sort(graph)):
150
307
if pb is not None:
151
308
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
309
rev = revisions[revid]
154
310
# 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
311
# we need to make sure to import the blobs / trees with the right
156
312
# path; this may involve adding them more than once.
158
inv.revision_id = rev.revision_id
159
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)
314
for parent_id in rev.parent_ids:
316
parent_invs.append(parent_invs_cache[parent_id])
318
parent_inv = repo.get_inventory(parent_id)
319
parent_invs.append(parent_inv)
320
parent_invs_cache[parent_id] = parent_inv
321
if parent_invs == []:
322
base_inv = Inventory(root_id=None)
324
base_inv = parent_invs[0]
325
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
326
mapping, "", root_trees[revid], base_inv, None, revid,
327
parent_invs, target_git_object_retriever._idmap, lookup_object)
328
target_git_object_retriever._idmap.add_entries(shamap)
329
if unusual_modes != {}:
330
for path, mode in unusual_modes.iteritems():
331
warn_unusual_mode(rev.foreign_revid, path, mode)
332
mapping.import_unusual_file_modes(rev, unusual_modes)
334
basis_id = rev.parent_ids[0]
336
basis_id = NULL_REVISION
337
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
338
inv_delta, rev.revision_id, rev.parent_ids)
339
parent_invs_cache[rev.revision_id] = inv
340
repo.add_revision(rev.revision_id, rev)
341
if "verify" in debug.debug_flags:
342
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
343
for sha1, newobj, path in objs:
344
assert path is not None
345
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
347
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
349
target_git_object_retriever._idmap.commit()
187
352
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
354
_matching_repo_format = GitRepositoryFormat()
192
357
def _get_repo_format_to_test():
196
361
"""See InterRepository.copy_content."""
197
362
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
364
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
366
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
367
mapping=mapping, fetch_spec=fetch_spec)
370
class InterGitNonGitRepository(InterGitRepository):
371
"""Base InterRepository that copies revisions from a Git into a non-Git
374
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
375
mapping=None, fetch_spec=None):
201
376
if mapping is None:
202
377
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
378
if revision_id is not None:
379
interesting_heads = [revision_id]
380
elif fetch_spec is not None:
381
interesting_heads = fetch_spec.heads
383
interesting_heads = None
385
def determine_wants(refs):
387
if interesting_heads is None:
388
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
390
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
391
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
211
graph_walker = BzrFetchGraphWalker(self.target, mapping)
392
self.fetch_objects(determine_wants, mapping, pb)
396
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
397
"""InterRepository that copies revisions from a remote Git into a non-Git
400
def fetch_objects(self, determine_wants, mapping, pb=None):
402
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
403
store = BazaarObjectStore(self.target, mapping)
404
self.target.lock_write()
406
heads = self.target.get_graph().heads(self.target.all_revision_ids())
407
graph_walker = store.get_graph_walker(
408
[store._lookup_revision_sha1(head) for head in heads])
411
def record_determine_wants(heads):
412
wants = determine_wants(heads)
413
recorded_wants.extend(wants)
418
create_pb = pb = ui.ui_factory.nested_progress_bar()
420
self.target.start_write_group()
422
objects_iter = self.source.fetch_objects(
423
record_determine_wants, graph_walker,
424
store.get_raw, progress)
425
import_git_objects(self.target, mapping, objects_iter,
426
store, recorded_wants, pb)
428
self.target.commit_write_group()
436
def is_compatible(source, target):
437
"""Be compatible with GitRepository."""
438
# FIXME: Also check target uses VersionedFile
439
return (isinstance(source, RemoteGitRepository) and
440
target.supports_rich_root() and
441
not isinstance(target, GitRepository))
444
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
445
"""InterRepository that copies revisions from a local Git into a non-Git
448
def fetch_objects(self, determine_wants, mapping, pb=None):
449
wants = determine_wants(self.source._git.get_refs())
214
452
create_pb = pb = ui.ui_factory.nested_progress_bar()
453
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
455
self.target.lock_write()
218
457
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
459
import_git_objects(self.target, mapping,
460
self.source._git.object_store,
461
target_git_object_retriever, wants, pb)
224
463
self.target.commit_write_group()