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
18
from bzrlib.errors import InvalidRevisionId
19
from bzrlib.inventory import Inventory
20
from bzrlib.repository import InterRepository
21
from bzrlib.trace import info
23
from bzrlib.plugins.git import git
24
from bzrlib.plugins.git.repository import LocalGitRepository, GitRepository, GitFormat
25
from bzrlib.plugins.git.remote import RemoteGitRepository
27
from dulwich.objects import Commit
29
from cStringIO import StringIO
32
class BzrFetchGraphWalker(object):
34
def __init__(self, repository, mapping):
35
self.repository = repository
36
self.mapping = mapping
38
self.heads = set(repository.all_revision_ids())
42
revid = self.mapping.revision_id_foreign_to_bzr(sha)
45
def remove(self, revid):
48
self.heads.remove(revid)
49
if revid in self.parents:
50
for p in self.parents[revid]:
55
ret = self.heads.pop()
56
ps = self.repository.get_parent_map([ret])[ret]
57
self.parents[ret] = ps
58
self.heads.update([p for p in ps if not p in self.done])
61
return self.mapping.revision_id_bzr_to_foreign(ret)
62
except InvalidRevisionId:
67
def import_git_blob(repo, mapping, path, blob):
17
from cStringIO import (
21
from dulwich.objects import (
26
from dulwich.object_store import (
39
from bzrlib.errors import (
44
from bzrlib.inventory import (
50
from bzrlib.lru_cache import (
53
from bzrlib.repository import (
56
from bzrlib.revision import (
59
from bzrlib.tsort import (
62
from bzrlib.versionedfile import (
63
FulltextContentFactory,
66
from bzrlib.plugins.git.mapping import (
68
inventory_to_tree_and_blobs,
74
from bzrlib.plugins.git.object_store import (
77
from bzrlib.plugins.git.remote import (
80
from bzrlib.plugins.git.repository import (
87
def import_git_blob(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
88
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
68
89
"""Import a git blob object into a bzr repository.
70
:param repo: bzr repository
91
:param texts: VersionedFiles to add to
71
92
:param path: Path in the tree
72
93
:param blob: A git blob
94
:return: Inventory delta for this file
74
96
file_id = mapping.generate_file_id(path)
75
repo.texts.add_lines((file_id, blob.id),
77
osutils.split_lines(blob.data))
78
inv.add_path(path, "file", file_id)
81
def import_git_tree(repo, mapping, path, tree, inv, lookup_object):
101
# We just have to hope this is indeed utf-8:
102
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
103
ie.executable = executable
104
# See if this has changed at all
109
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
113
if (base_sha == hexsha and base_ie.executable == ie.executable
114
and base_ie.kind == ie.kind):
115
# If nothing has changed since the base revision, we're done
117
if base_sha == hexsha and base_ie.kind == ie.kind:
118
ie.text_size = base_ie.text_size
119
ie.text_sha1 = base_ie.text_sha1
120
ie.symlink_target = base_ie.symlink_target
121
if ie.executable == base_ie.executable:
122
ie.revision = base_ie.revision
124
blob = lookup_object(hexsha)
126
blob = lookup_object(hexsha)
127
if ie.kind == "symlink":
129
ie.symlink_target = blob.data
133
ie.text_size = len(blob.data)
134
ie.text_sha1 = osutils.sha_string(blob.data)
135
# Check what revision we should store
137
for pinv in parent_invs:
138
if pinv.revision_id == base_inv.revision_id:
147
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
148
# found a revision in one of the parents to use
149
ie.revision = pie.revision
151
parent_keys.append((file_id, pie.revision))
152
if ie.revision is None:
153
# Need to store a new revision
154
ie.revision = revision_id
155
assert file_id is not None
156
assert ie.revision is not None
157
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
158
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
161
if file_id in base_inv:
162
old_path = base_inv.id2path(file_id)
165
invdelta = [(old_path, path, file_id, ie)]
166
if base_ie is not None and base_ie.kind == "directory":
167
invdelta.extend(remove_disappeared_children(base_inv, base_ie.children, []))
168
return (invdelta, shamap)
171
def import_git_submodule(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
172
revision_id, parent_invs, shagitmap, lookup_object):
173
raise NotImplementedError(import_git_submodule)
176
def remove_disappeared_children(base_inv, base_children, existing_children):
178
deletable = [v for k,v in base_children.iteritems() if k not in existing_children]
181
ret.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
182
if ie.kind == "directory":
183
deletable.extend(ie.children.values())
187
def import_git_tree(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
188
revision_id, parent_invs, shagitmap, lookup_object):
82
189
"""Import a git tree object into a bzr repository.
84
:param repo: A Bzr repository object
191
:param texts: VersionedFiles object to add to
85
192
:param path: Path in the tree
86
193
:param tree: A git tree object
87
:param inv: Inventory object
194
:param base_inv: Base inventory against which to return inventory delta
195
:return: Inventory delta for this subtree
89
198
file_id = mapping.generate_file_id(path)
90
repo.texts.add_lines((file_id, tree.id),
93
inv.add_path(path, "directory", file_id)
94
for mode, name, hexsha in tree.entries():
95
entry_kind = (mode & 0700000) / 0100000
199
# We just have to hope this is indeed utf-8:
200
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
203
# Newly appeared here
204
ie.revision = revision_id
205
texts.add_lines((file_id, ie.revision), (), [])
206
invdelta.append((None, path, file_id, ie))
208
# See if this has changed at all
210
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
214
if base_sha == hexsha:
215
# If nothing has changed since the base revision, we're done
217
if base_ie.kind != "directory":
218
ie.revision = revision_id
219
texts.add_lines((ie.file_id, ie.revision), (), [])
220
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
221
if base_ie is not None and base_ie.kind == "directory":
222
base_children = base_ie.children
225
# Remember for next time
226
existing_children = set()
229
tree = lookup_object(hexsha)
230
for mode, name, child_hexsha in tree.entries():
96
231
basename = name.decode("utf-8")
100
child_path = urlutils.join(path, name)
102
import_git_tree(repo, mapping, child_path, lookup_object, inv)
103
elif entry_kind == 1:
104
import_git_blob(repo, mapping, child_path, lookup_object, inv)
106
raise AssertionError("Unknown blob kind, perms=%r." % (mode,))
109
def import_git_objects(repo, mapping, object_iter):
232
existing_children.add(basename)
233
child_path = osutils.pathjoin(path, name)
234
if stat.S_ISDIR(mode):
235
subinvdelta, grandchildmodes, subshamap = import_git_tree(
236
texts, mapping, child_path, child_hexsha, base_inv,
237
base_children.get(name), file_id, revision_id, parent_invs, shagitmap,
239
invdelta.extend(subinvdelta)
240
child_modes.update(grandchildmodes)
241
shamap.extend(subshamap)
242
elif S_ISGITLINK(mode): # submodule
243
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
244
texts, mapping, child_path, child_hexsha, base_inv, base_ie.get(name),
245
file_id, revision_id, parent_invs, shagitmap, lookup_object)
246
invdelta.extend(subinvdelta)
247
child_modes.update(grandchildmodes)
248
shamap.extend(subshamap)
250
subinvdelta, subshamap = import_git_blob(texts, mapping,
251
child_path, child_hexsha, base_inv, base_children.get(name), file_id,
252
revision_id, parent_invs, shagitmap, lookup_object,
253
mode_is_executable(mode), stat.S_ISLNK(mode))
254
invdelta.extend(subinvdelta)
255
shamap.extend(subshamap)
256
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
257
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
258
child_modes[child_path] = mode
259
# Remove any children that have disappeared
260
if base_ie is not None and base_ie.kind == "directory":
261
invdelta.extend(remove_disappeared_children(base_inv, base_children,
263
shamap.append((hexsha, "tree", (file_id, revision_id)))
264
return invdelta, child_modes, shamap
267
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
110
269
"""Import a set of git objects into a bzr repository.
112
:param repo: Bazaar repository
271
:param repo: Target Bazaar repository
113
272
:param mapping: Mapping to use
114
273
:param object_iter: Iterator over Git objects.
275
def lookup_object(sha):
277
return object_iter[sha]
279
return target_git_object_retriever[sha]
116
280
# TODO: a more (memory-)efficient implementation of this
118
for o in object_iter:
286
parent_invs_cache = LRUCache(50)
121
287
# Find and convert commit objects
122
for o in objects.iterkeys():
290
pb.update("finding revisions to fetch", len(graph), None)
292
assert isinstance(head, str)
294
o = lookup_object(head)
123
297
if isinstance(o, Commit):
124
298
rev = mapping.import_commit(o)
125
root_trees[rev] = objects[o.tree_sha]
299
if repo.has_revision(rev.revision_id):
301
squash_revision(repo, rev)
302
root_trees[rev.revision_id] = o.tree
303
revisions[rev.revision_id] = rev
304
graph.append((rev.revision_id, rev.parent_ids))
305
target_git_object_retriever._idmap.add_entry(o.id, "commit",
306
(rev.revision_id, o.tree))
307
heads.extend([p for p in o.parents if p not in checked])
308
elif isinstance(o, Tag):
309
heads.append(o.object[1])
311
trace.warning("Unable to import head object %r" % o)
313
# Order the revisions
126
314
# Create the inventory objects
127
for rev, root_tree in root_trees.iteritems():
315
for i, revid in enumerate(topo_sort(graph)):
317
pb.update("fetching revisions", i, len(graph))
318
rev = revisions[revid]
128
319
# We have to do this here, since we have to walk the tree and
129
# we need to make sure to import the blobs / trees with the riht
320
# we need to make sure to import the blobs / trees with the right
130
321
# path; this may involve adding them more than once.
132
def lookup_object(sha):
135
return reconstruct_git_object(repo, mapping, sha)
136
import_git_tree(repo, mapping, "", tree, inv, lookup_object)
137
repo.add_revision(rev.revision_id, rev, inv)
140
def reconstruct_git_commit(repo, rev):
141
raise NotImplementedError(self.reconstruct_git_commit)
144
def reconstruct_git_object(repo, mapping, sha):
146
revid = mapping.revision_id_foreign_to_bzr(sha)
148
rev = repo.get_revision(revid)
149
except NoSuchRevision:
152
return reconstruct_git_commit(rev)
156
raise KeyError("No such object %s" % sha)
323
for parent_id in rev.parent_ids:
325
parent_invs.append(parent_invs_cache[parent_id])
327
parent_inv = repo.get_inventory(parent_id)
328
parent_invs.append(parent_inv)
329
parent_invs_cache[parent_id] = parent_inv
330
if parent_invs == []:
331
base_inv = Inventory(root_id=None)
334
base_inv = parent_invs[0]
335
base_ie = base_inv.root
336
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
337
mapping, "", root_trees[revid], base_inv, base_ie, None, revid,
338
parent_invs, target_git_object_retriever._idmap, lookup_object)
339
target_git_object_retriever._idmap.add_entries(shamap)
340
if unusual_modes != {}:
341
for path, mode in unusual_modes.iteritems():
342
warn_unusual_mode(rev.foreign_revid, path, mode)
343
mapping.import_unusual_file_modes(rev, unusual_modes)
345
basis_id = rev.parent_ids[0]
347
basis_id = NULL_REVISION
348
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
349
inv_delta, rev.revision_id, rev.parent_ids)
350
parent_invs_cache[rev.revision_id] = inv
351
repo.add_revision(rev.revision_id, rev)
352
if "verify" in debug.debug_flags:
353
new_unusual_modes = mapping.export_unusual_file_modes(rev)
354
if new_unusual_modes != unusual_modes:
355
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
356
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
357
for sha1, newobj, path in objs:
358
assert path is not None
359
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
361
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
363
target_git_object_retriever._idmap.commit()
159
366
class InterGitRepository(InterRepository):
161
_matching_repo_format = GitFormat()
368
_matching_repo_format = GitRepositoryFormat()
164
371
def _get_repo_format_to_test():
168
375
"""See InterRepository.copy_content."""
169
376
self.fetch(revision_id, pb, find_ghosts=False)
171
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
378
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
380
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
381
mapping=mapping, fetch_spec=fetch_spec)
384
class InterGitNonGitRepository(InterGitRepository):
385
"""Base InterRepository that copies revisions from a Git into a non-Git
388
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
389
mapping=None, fetch_spec=None):
173
390
if mapping is None:
174
391
mapping = self.source.get_mapping()
177
pb.note("git: %s" % text)
179
info("git: %s" % text)
180
def determine_wants(heads):
181
if revision_id is None:
184
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
392
if revision_id is not None:
393
interesting_heads = [revision_id]
394
elif fetch_spec is not None:
395
interesting_heads = fetch_spec.heads
397
interesting_heads = None
399
def determine_wants(refs):
401
if interesting_heads is None:
402
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
404
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
185
405
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
186
graph_walker = BzrFetchGraphWalker(self.target, mapping)
406
self.fetch_objects(determine_wants, mapping, pb)
410
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
411
def report_git_progress(pb, text):
412
text = text.rstrip("\r\n")
413
g = _GIT_PROGRESS_RE.match(text)
415
(text, pct, current, total) = g.groups()
416
pb.update(text, int(current), int(total))
418
pb.update(text, 0, 0)
421
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
422
"""InterRepository that copies revisions from a remote Git into a non-Git
425
def fetch_objects(self, determine_wants, mapping, pb=None):
427
report_git_progress(pb, text)
428
store = BazaarObjectStore(self.target, mapping)
187
429
self.target.lock_write()
189
import_git_objects(self.target, mapping,
190
self.source.fetch_objects(determine_wants, graph_walker,
431
# FIXME: This should be more efficient
432
heads = self.target.get_graph().heads(self.target.all_revision_ids())
433
graph_walker = store.get_graph_walker(
434
[store._lookup_revision_sha1(head) for head in heads])
437
def record_determine_wants(heads):
438
wants = determine_wants(heads)
439
recorded_wants.extend(wants)
444
create_pb = pb = ui.ui_factory.nested_progress_bar()
446
self.target.start_write_group()
448
objects_iter = self.source.fetch_objects(
449
record_determine_wants, graph_walker,
450
store.get_raw, progress)
451
import_git_objects(self.target, mapping, objects_iter,
452
store, recorded_wants, pb)
454
self.target.commit_write_group()
193
459
self.target.unlock()