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 (
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.repository import (
29
from bzrlib.plugins.git.remote import RemoteGitRepository
20
31
import dulwich as git
21
from dulwich.objects import (
25
from dulwich.object_store import (
37
from bzrlib.errors import (
42
from bzrlib.inventory import (
48
from bzrlib.lru_cache import (
51
from bzrlib.repository import (
54
from bzrlib.revision import (
57
from bzrlib.tsort import (
60
from bzrlib.versionedfile import (
61
FulltextContentFactory,
64
from bzrlib.plugins.git.mapping import (
66
inventory_to_tree_and_blobs,
69
from bzrlib.plugins.git.object_store import (
72
from bzrlib.plugins.git.remote import (
75
from bzrlib.plugins.git.repository import (
82
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
83
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
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):
84
78
"""Import a git blob object into a bzr repository.
86
:param texts: VersionedFiles to add to
80
:param repo: bzr repository
87
81
:param path: Path in the tree
88
82
:param blob: A git blob
89
:return: Inventory delta for this file
91
84
file_id = mapping.generate_file_id(path)
96
# We just have to hope this is indeed utf-8:
97
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
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)
98
93
ie.executable = executable
100
# See if this has changed at all
102
base_ie = base_inv[file_id]
107
base_sha = base_ie.text_id
110
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
114
if (base_sha == hexsha and base_ie.executable == ie.executable
115
and base_ie.kind == ie.kind):
116
# If nothing has changed since the base revision, we're done
118
if base_sha == hexsha:
119
ie.text_size = base_ie.text_size
120
ie.text_sha1 = base_ie.text_sha1
121
ie.symlink_target = base_ie.symlink_target
122
ie.revision = base_ie.revision
124
blob = lookup_object(hexsha)
125
if ie.kind == "symlink":
126
ie.symlink_target = blob.data
130
ie.text_size = len(blob.data)
131
ie.text_sha1 = osutils.sha_string(blob.data)
132
# Check what revision we should store
134
for pinv in parent_invs:
135
if pinv.revision_id == base_inv.revision_id:
144
if pie.text_sha1 == ie.text_sha1:
145
# found a revision in one of the parents to use
146
ie.revision = pie.revision
148
parent_keys.append((file_id, pie.revision))
149
if ie.revision is None:
150
# Need to store a new revision
151
ie.revision = revision_id
152
assert file_id is not None
153
assert ie.revision is not None
154
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
155
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
158
if file_id in base_inv:
159
old_path = base_inv.id2path(file_id)
162
return ([(old_path, path, file_id, ie)], shamap)
165
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
166
revision_id, parent_invs, shagitmap, lookup_object):
96
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
167
97
"""Import a git tree object into a bzr repository.
169
:param texts: VersionedFiles object to add to
99
:param repo: A Bzr repository object
170
100
:param path: Path in the tree
171
101
:param tree: A git tree object
172
:param base_inv: Base inventory against which to return inventory delta
173
:return: Inventory delta for this subtree
102
:param inv: Inventory object
176
104
file_id = mapping.generate_file_id(path)
177
# We just have to hope this is indeed utf-8:
178
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
182
base_ie = base_inv[file_id]
184
# Newly appeared here
186
ie.revision = revision_id
187
texts.add_lines((file_id, ie.revision), (), [])
188
invdelta.append((None, path, file_id, ie))
190
base_sha = base_ie.text_id
191
# See if this has changed at all
194
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
198
if base_sha == hexsha:
199
# If nothing has changed since the base revision, we're done
201
# Remember for next time
202
existing_children = set()
205
tree = lookup_object(hexsha)
206
for mode, name, child_hexsha in tree.entries():
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
207
113
basename = name.decode("utf-8")
208
existing_children.add(basename)
209
child_path = osutils.pathjoin(path, name)
210
if stat.S_ISDIR(mode):
211
subinvdelta, grandchildmodes, subshamap = import_git_tree(texts,
212
mapping, child_path, child_hexsha, base_inv, file_id,
213
revision_id, parent_invs, shagitmap, lookup_object)
214
invdelta.extend(subinvdelta)
215
child_modes.update(grandchildmodes)
216
shamap.extend(subshamap)
218
fs_mode = stat.S_IMODE(mode)
219
symlink = stat.S_ISLNK(mode)
220
subinvdelta, subshamap = import_git_blob(texts, mapping,
221
child_path, child_hexsha, base_inv, file_id, revision_id,
222
parent_invs, shagitmap, lookup_object,
223
bool(fs_mode & 0111), symlink)
224
invdelta.extend(subinvdelta)
225
shamap.extend(subshamap)
226
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
227
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
228
child_modes[child_path] = mode
229
# Remove any children that have disappeared
230
if base_ie is not None:
231
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
234
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
235
if ie.kind == "directory":
236
deletable.extend(ie.children.values())
237
shamap.append((hexsha, "tree", (file_id, revision_id)))
238
return invdelta, child_modes, shamap
241
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
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):
243
130
"""Import a set of git objects into a bzr repository.
245
132
:param repo: Bazaar repository
246
133
:param mapping: Mapping to use
134
:param num_objects: Number of objects.
247
135
:param object_iter: Iterator over Git objects.
249
def lookup_object(sha):
251
return object_iter[sha]
253
return target_git_object_retriever[sha]
254
137
# TODO: a more (memory-)efficient implementation of this
139
for i, (o, _) in enumerate(object_iter):
141
pb.update("fetching objects", i, num_objects)
260
parent_invs_cache = LRUCache(50)
261
146
# Find and convert commit objects
264
pb.update("finding revisions to fetch", len(graph), None)
266
assert isinstance(head, str)
268
o = lookup_object(head)
147
for o in objects.itervalues():
271
148
if isinstance(o, Commit):
272
149
rev = mapping.import_commit(o)
273
if repo.has_revision(rev.revision_id):
275
root_trees[rev.revision_id] = o.tree
150
root_trees[rev.revision_id] = objects[o.tree]
276
151
revisions[rev.revision_id] = rev
277
152
graph.append((rev.revision_id, rev.parent_ids))
278
target_git_object_retriever._idmap.add_entry(o.id, "commit",
279
(rev.revision_id, o.tree))
280
heads.extend([p for p in o.parents if p not in checked])
281
elif isinstance(o, Tag):
282
heads.append(o.object[1])
284
trace.warning("Unable to import head object %r" % o)
286
153
# Order the revisions
287
154
# Create the inventory objects
288
155
for i, revid in enumerate(topo_sort(graph)):
289
156
if pb is not None:
290
157
pb.update("fetching revisions", i, len(graph))
158
root_tree = root_trees[revid]
291
159
rev = revisions[revid]
292
160
# We have to do this here, since we have to walk the tree and
293
# we need to make sure to import the blobs / trees with the right
161
# we need to make sure to import the blobs / trees with the riht
294
162
# path; this may involve adding them more than once.
296
for parent_id in rev.parent_ids:
298
parent_invs.append(parent_invs_cache[parent_id])
300
parent_inv = repo.get_inventory(parent_id)
301
parent_invs.append(parent_inv)
302
parent_invs_cache[parent_id] = parent_inv
303
if parent_invs == []:
304
base_inv = Inventory(root_id=None)
306
base_inv = parent_invs[0]
307
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
308
mapping, "", root_trees[revid], base_inv, None, revid,
309
parent_invs, target_git_object_retriever._idmap, lookup_object)
310
target_git_object_retriever._idmap.add_entries(shamap)
311
if unusual_modes != {}:
312
ret = "unusual modes: \n"
313
for item in unusual_modes.iteritems():
314
ret += "\t%s: %o\n" % item
315
raise AssertionError(ret)
317
basis_id = rev.parent_ids[0]
319
basis_id = NULL_REVISION
320
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
321
inv_delta, rev.revision_id, rev.parent_ids)
322
parent_invs_cache[rev.revision_id] = inv
323
repo.add_revision(rev.revision_id, rev)
324
if "verify" in debug.debug_flags:
325
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
326
for sha1, newobj, path in objs:
327
assert path is not None
328
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
329
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
331
target_git_object_retriever._idmap.commit()
334
class InterGitRepository(InterRepository):
336
_matching_repo_format = GitRepositoryFormat()
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()
339
199
def _get_repo_format_to_test():
343
203
"""See InterRepository.copy_content."""
344
204
self.fetch(revision_id, pb, find_ghosts=False)
346
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
348
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
349
mapping=mapping, fetch_spec=fetch_spec)
352
class InterGitNonGitRepository(InterGitRepository):
353
"""Base InterRepository that copies revisions from a Git into a non-Git
356
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
357
mapping=None, fetch_spec=None):
206
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
358
208
if mapping is None:
359
209
mapping = self.source.get_mapping()
360
if revision_id is not None:
361
interesting_heads = [revision_id]
362
elif fetch_spec is not None:
363
interesting_heads = fetch_spec.heads
365
interesting_heads = None
367
def determine_wants(refs):
369
if interesting_heads is None:
370
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
211
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
212
def determine_wants(heads):
213
if revision_id is None:
372
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
216
ret = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
373
217
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
374
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
store = BazaarObjectStore(self.target, mapping)
386
self.target.lock_read()
388
heads = self.target.get_graph().heads(self.target.all_revision_ids())
391
graph_walker = store.get_graph_walker(
392
[store._lookup_revision_sha1(head) for head in heads])
395
create_pb = pb = ui.ui_factory.nested_progress_bar()
398
def record_determine_wants(heads):
399
wants = determine_wants(heads)
400
recorded_wants.extend(wants)
404
self.target.lock_write()
406
self.target.start_write_group()
408
objects_iter = self.source.fetch_objects(
409
record_determine_wants, graph_walker,
410
store.get_raw, progress)
411
import_git_objects(self.target, mapping, objects_iter,
412
store, recorded_wants, pb)
414
self.target.commit_write_group()
422
def is_compatible(source, target):
423
"""Be compatible with GitRepository."""
424
# FIXME: Also check target uses VersionedFile
425
return (isinstance(source, RemoteGitRepository) and
426
target.supports_rich_root() and
427
not isinstance(target, GitRepository))
430
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
431
"""InterRepository that copies revisions from a remote Git into a non-Git
434
def fetch_objects(self, determine_wants, mapping, pb=None):
435
wants = determine_wants(self.source._git.get_refs())
438
create_pb = pb = ui.ui_factory.nested_progress_bar()
439
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
441
self.target.lock_write()
443
self.target.start_write_group()
445
import_git_objects(self.target, mapping,
446
self.source._git.object_store,
447
target_git_object_retriever, wants, pb)
449
self.target.commit_write_group()
457
def is_compatible(source, target):
458
"""Be compatible with GitRepository."""
459
# FIXME: Also check target uses VersionedFile
460
return (isinstance(source, LocalGitRepository) and
461
target.supports_rich_root() and
462
not isinstance(target, GitRepository))
465
class InterGitGitRepository(InterGitRepository):
466
"""InterRepository that copies between Git repositories."""
468
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
469
mapping=None, fetch_spec=None, branches=None):
218
graph_walker = BzrFetchGraphWalker(self.target, mapping)
221
create_pb = pb = ui.ui_factory.nested_progress_bar()
223
self.target.lock_write()
225
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,
233
self.target.commit_write_group()
241
def is_compatible(source, target):
242
"""Be compatible with GitRepository."""
243
# FIXME: Also check target uses VersionedFile
244
return (isinstance(source, GitRepository) and
245
target.supports_rich_root() and
246
not isinstance(target, GitRepository))
249
class InterGitRepository(InterRepository):
251
_matching_repo_format = GitFormat()
254
def _get_repo_format_to_test():
257
def copy_content(self, revision_id=None, pb=None):
258
"""See InterRepository.copy_content."""
259
self.fetch(revision_id, pb, find_ghosts=False)
261
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
470
263
if mapping is None:
471
264
mapping = self.source.get_mapping()
472
265
def progress(text):
473
trace.info("git: %s", text)
266
info("git: %s", text)
474
267
r = self.target._git
475
if revision_id is not None:
268
if revision_id is None:
269
determine_wants = lambda x: [y for y in x.values() if not y in r.object_store]
476
271
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
477
elif fetch_spec is not None:
478
args = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in fetch_spec.heads]
479
if branches is not None:
480
determine_wants = lambda x: [x[y] for y in branches if not x[y] in r.object_store]
481
elif fetch_spec is None and revision_id is None:
482
determine_wants = r.object_store.determine_wants_all
484
272
determine_wants = lambda x: [y for y in args if not y in r.object_store]
486
graphwalker = r.get_graph_walker()
487
f, commit = r.object_store.add_thin_pack()
274
graphwalker = SimpleFetchGraphWalker(r.heads().values(), r.get_parents)
275
f, commit = r.object_store.add_pack()
489
refs = self.source.fetch_pack(determine_wants, graphwalker,
277
self.source._git.fetch_pack(path, determine_wants, graphwalker, f.write, progress)