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, NoSuchRevision
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.shamap import GitObjectConverter
30
from bzrlib.plugins.git.remote import RemoteGitRepository
20
32
import dulwich as git
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 (
79
from bzrlib.plugins.git.repository import (
86
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
87
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
33
from dulwich.client import SimpleFetchGraphWalker
34
from dulwich.objects import Commit
36
from cStringIO import StringIO
39
class BzrFetchGraphWalker(object):
40
"""GraphWalker implementation that uses a Bazaar repository."""
42
def __init__(self, repository, mapping):
43
self.repository = repository
44
self.mapping = mapping
46
self.heads = set(repository.all_revision_ids())
50
return iter(self.next, None)
53
revid = self.mapping.revision_id_foreign_to_bzr(sha)
56
def remove(self, revid):
58
if revid in self.heads:
59
self.heads.remove(revid)
60
if revid in self.parents:
61
for p in self.parents[revid]:
66
ret = self.heads.pop()
67
ps = self.repository.get_parent_map([ret])[ret]
68
self.parents[ret] = ps
69
self.heads.update([p for p in ps if not p in self.done])
72
return self.mapping.revision_id_bzr_to_foreign(ret)[0]
73
except InvalidRevisionId:
78
def import_git_blob(repo, mapping, path, blob, inv, parent_invs, executable):
88
79
"""Import a git blob object into a bzr repository.
90
:param texts: VersionedFiles to add to
81
:param repo: bzr repository
91
82
:param path: Path in the tree
92
83
:param blob: A git blob
93
:return: Inventory delta for this file
95
85
file_id = mapping.generate_file_id(path)
100
# We just have to hope this is indeed utf-8:
101
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
86
text_revision = inv.revision_id
87
repo.texts.add_lines((file_id, text_revision),
88
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
89
osutils.split_lines(blob.data))
90
ie = inv.add_path(path, "file", file_id)
91
ie.revision = text_revision
92
ie.text_size = len(blob.data)
93
ie.text_sha1 = osutils.sha_string(blob.data)
102
94
ie.executable = executable
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
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
192
98
"""Import a git tree object into a bzr repository.
194
:param texts: VersionedFiles object to add to
100
:param repo: A Bzr repository object
195
101
:param path: Path in the tree
196
102
:param tree: A git tree object
197
:param base_inv: Base inventory against which to return inventory delta
198
:return: Inventory delta for this subtree
103
:param inv: Inventory object
201
105
file_id = mapping.generate_file_id(path)
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():
106
text_revision = inv.revision_id
107
repo.texts.add_lines((file_id, text_revision),
108
[(file_id, p[file_id].revision) for p in parent_invs if file_id in p],
110
ie = inv.add_path(path, "directory", file_id)
111
ie.revision = text_revision
112
for mode, name, hexsha in tree.entries():
113
entry_kind = (mode & 0700000) / 0100000
233
114
basename = name.decode("utf-8")
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
118
child_path = urlutils.join(path, name)
120
tree = lookup_object(hexsha)
121
import_git_tree(repo, mapping, child_path, tree, inv, parent_invs, lookup_object)
122
elif entry_kind == 1:
123
blob = lookup_object(hexsha)
124
fs_mode = mode & 0777
125
import_git_blob(repo, mapping, child_path, blob, inv, parent_invs, bool(fs_mode & 0111))
127
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
266
130
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
268
132
"""Import a set of git objects into a bzr repository.
270
:param repo: Target Bazaar repository
134
:param repo: Bazaar repository
271
135
:param mapping: Mapping to use
272
136
:param object_iter: Iterator over Git objects.
274
def lookup_object(sha):
276
return object_iter[sha]
278
return target_git_object_retriever[sha]
279
138
# TODO: a more (memory-)efficient implementation of this
285
parent_invs_cache = LRUCache(50)
286
142
# Find and convert commit objects
289
pb.update("finding revisions to fetch", len(graph), None)
291
assert isinstance(head, str)
293
o = lookup_object(head)
143
for o in object_iter.iterobjects():
296
144
if isinstance(o, Commit):
297
145
rev = mapping.import_commit(o)
298
if repo.has_revision(rev.revision_id):
300
squash_revision(repo, rev)
301
root_trees[rev.revision_id] = o.tree
146
root_trees[rev.revision_id] = object_iter[o.tree]
302
147
revisions[rev.revision_id] = rev
303
148
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)
312
149
# Order the revisions
313
150
# Create the inventory objects
314
151
for i, revid in enumerate(topo_sort(graph)):
315
152
if pb is not None:
316
153
pb.update("fetching revisions", i, len(graph))
154
root_tree = root_trees[revid]
317
155
rev = revisions[revid]
318
156
# We have to do this here, since we have to walk the tree and
319
# we need to make sure to import the blobs / trees with the right
157
# we need to make sure to import the blobs / trees with the riht
320
158
# path; this may involve adding them more than once.
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()
160
inv.revision_id = rev.revision_id
161
def lookup_object(sha):
162
if sha in object_iter:
163
return object_iter[sha]
164
return target_git_object_retriever(sha)
165
parent_invs = [repo.get_inventory(r) for r in rev.parent_ids]
166
import_git_tree(repo, mapping, "", root_tree, inv, parent_invs,
168
repo.add_revision(rev.revision_id, rev, inv)
171
def reconstruct_git_object(repo, mapping, sha):
172
import pdb; pdb.set_trace()
176
raise KeyError("No such object %s" % sha)
179
class InterGitNonGitRepository(InterRepository):
181
_matching_repo_format = GitFormat()
368
184
def _get_repo_format_to_test():
372
188
"""See InterRepository.copy_content."""
373
189
self.fetch(revision_id, pb, 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):
388
mapping = self.source.get_mapping()
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("^{}")]
401
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
402
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
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
191
def fetch_objects(self, determine_wants, mapping, pb=None):
412
192
def progress(text):
413
193
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())
194
graph_walker = BzrFetchGraphWalker(self.target, mapping)
463
197
create_pb = pb = ui.ui_factory.nested_progress_bar()
464
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
198
target_git_object_retriever = GitObjectConverter(self.target, mapping)
466
201
self.target.lock_write()
468
203
self.target.start_write_group()
470
import_git_objects(self.target, mapping,
471
self.source._git.object_store,
472
target_git_object_retriever, wants, pb)
205
objects_iter = self.source.fetch_objects(determine_wants,
207
target_git_object_retriever.__getitem__,
209
import_git_objects(self.target, mapping, objects_iter,
210
target_git_object_retriever, pb)
474
212
self.target.commit_write_group()