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 dulwich.objects import (
23
from dulwich.object_store import (
36
from bzrlib.errors import (
40
from bzrlib.inventory import (
47
from bzrlib.repository import (
50
from bzrlib.revision import (
53
from bzrlib.tsort import (
56
from bzrlib.versionedfile import (
57
ChunkedContentFactory,
60
from bzrlib.plugins.git.mapping import (
62
inventory_to_tree_and_blobs,
67
from bzrlib.plugins.git.object_store import (
71
from bzrlib.plugins.git.remote import (
25
74
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):
81
def import_git_blob(texts, mapping, path, hexsha, base_hexsha,
82
base_inv, base_inv_shamap, base_ie, parent_id, revision_id,
83
parent_invs, lookup_object, executable, symlink):
73
84
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
86
:param texts: VersionedFiles to add to
76
87
:param path: Path in the tree
77
88
:param blob: A git blob
89
:return: Inventory delta for this file
79
91
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)
96
# We just have to hope this is indeed utf-8:
97
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
98
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
99
# See if this has changed at all
100
if (base_hexsha == hexsha and base_ie.executable == ie.executable
101
and base_ie.kind == ie.kind):
102
# If nothing has changed since the base revision, we're done
104
if base_hexsha == hexsha and base_ie.kind == ie.kind:
105
ie.text_size = base_ie.text_size
106
ie.text_sha1 = base_ie.text_sha1
107
ie.symlink_target = base_ie.symlink_target
108
if ie.executable == base_ie.executable:
109
ie.revision = base_ie.revision
111
blob = lookup_object(hexsha)
113
blob = lookup_object(hexsha)
114
if ie.kind == "symlink":
116
ie.symlink_target = blob.data
120
ie.text_size = len(blob.data)
121
ie.text_sha1 = osutils.sha_string(blob.data)
122
# Check what revision we should store
124
for pinv in parent_invs:
125
if pinv.revision_id == base_inv.revision_id:
134
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
135
# found a revision in one of the parents to use
136
ie.revision = pie.revision
138
parent_keys.append((file_id, pie.revision))
139
if ie.revision is None:
140
# Need to store a new revision
141
ie.revision = revision_id
142
assert file_id is not None
143
assert ie.revision is not None
144
if ie.kind == 'symlink':
148
chunks = blob.chunked
149
except AttributeError: # older version of dulwich
151
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, chunks)])
152
shamap = { ie.file_id: hexsha }
154
if base_ie is not None:
155
old_path = base_inv.id2path(file_id)
156
if base_ie.kind == "directory":
157
invdelta.extend(remove_disappeared_children(old_path, base_ie.children, []))
160
invdelta.append((old_path, path, file_id, ie))
161
return (invdelta, shamap)
164
class SubmodulesRequireSubtrees(BzrError):
165
_fmt = """The repository you are fetching from contains submodules. To continue, upgrade your Bazaar repository to a format that supports nested trees, such as 'development-subtree'."""
169
def import_git_submodule(texts, mapping, path, hexsha, base_hexsha,
170
base_inv, base_ie, parent_id, revision_id, parent_invs, lookup_object):
171
file_id = mapping.generate_file_id(path)
172
ie = TreeReference(file_id, urlutils.basename(path.decode("utf-8")),
174
ie.revision = revision_id
179
if (base_ie.kind == ie.kind and
180
base_ie.reference_revision == ie.reference_revision):
181
ie.revision = base_ie.revision
182
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
183
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
184
invdelta = [(oldpath, path, file_id, ie)]
185
return invdelta, {}, {}
188
def remove_disappeared_children(path, base_children, existing_children):
190
deletable = [(osutils.pathjoin(path, k), v) for k,v in base_children.iteritems() if k not in existing_children]
192
(path, ie) = deletable.pop()
193
ret.append((path, None, ie.file_id, None))
194
if ie.kind == "directory":
195
for name, child_ie in ie.children.iteritems():
196
deletable.append((osutils.pathjoin(path, name), child_ie))
200
def import_git_tree(texts, mapping, path, hexsha, base_hexsha, base_inv,
201
base_inv_shamap, base_ie, parent_id, revision_id, parent_invs,
202
lookup_object, allow_submodules=False):
92
203
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
205
:param texts: VersionedFiles object to add to
95
206
:param path: Path in the tree
96
207
:param tree: A git tree object
97
:param inv: Inventory object
208
:param base_inv: Base inventory against which to return inventory delta
209
:return: Inventory delta for this subtree
99
212
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
213
# We just have to hope this is indeed utf-8:
214
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
217
# Newly appeared here
218
ie.revision = revision_id
219
texts.insert_record_stream([ChunkedContentFactory((file_id, ie.revision), (), None, [])])
220
invdelta.append((None, path, file_id, ie))
222
if base_hexsha == hexsha:
223
# If nothing has changed since the base revision, we're done
225
if base_ie.kind != "directory":
226
ie.revision = revision_id
227
texts.insert_record_stream([ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
228
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
229
tree = lookup_object(hexsha)
230
if base_hexsha is None:
233
base_tree = lookup_object(base_hexsha)
235
if base_ie is not None and base_ie.kind == "directory":
236
base_children = base_ie.children
239
# Remember for next time
240
existing_children = set()
243
for mode, name, child_hexsha in tree.entries():
108
244
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):
245
existing_children.add(basename)
246
child_path = osutils.pathjoin(path, name)
247
if type(base_tree) is Tree:
249
child_base_mode, child_base_hexsha = base_tree[name]
251
child_base_hexsha = None
254
child_base_hexsha = None
256
if stat.S_ISDIR(mode):
257
subinvdelta, grandchildmodes, subshamap = import_git_tree(
258
texts, mapping, child_path, child_hexsha, child_base_hexsha,
259
base_inv, base_inv_shamap, base_children.get(basename),
260
file_id, revision_id, parent_invs, lookup_object,
261
allow_submodules=allow_submodules)
262
elif S_ISGITLINK(mode): # submodule
263
if not allow_submodules:
264
raise SubmodulesRequireSubtrees()
265
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
266
texts, mapping, child_path, child_hexsha, child_base_hexsha,
267
base_inv, base_children.get(basename),
268
file_id, revision_id, parent_invs, lookup_object)
270
subinvdelta, subshamap = import_git_blob(texts, mapping,
271
child_path, child_hexsha, child_base_hexsha,
272
base_inv, base_inv_shamap,
273
base_children.get(basename), file_id,
274
revision_id, parent_invs, lookup_object,
275
mode_is_executable(mode), stat.S_ISLNK(mode))
277
child_modes.update(grandchildmodes)
278
invdelta.extend(subinvdelta)
279
shamap.update(subshamap)
280
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
281
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
282
child_modes[child_path] = mode
283
# Remove any children that have disappeared
284
if base_ie is not None and base_ie.kind == "directory":
285
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
286
base_children, existing_children))
287
shamap[file_id] = hexsha
288
return invdelta, child_modes, shamap
291
def import_git_commit(repo, mapping, head, lookup_object,
292
target_git_object_retriever, parent_invs_cache):
293
o = lookup_object(head)
294
rev = mapping.import_commit(o)
295
# We have to do this here, since we have to walk the tree and
296
# we need to make sure to import the blobs / trees with the right
297
# path; this may involve adding them more than once.
298
parent_invs = parent_invs_cache.get_inventories(rev.parent_ids)
299
if parent_invs == []:
300
base_inv = Inventory(root_id=None)
302
base_inv_shamap = None # Should never be accessed
305
base_inv = parent_invs[0]
306
base_ie = base_inv.root
307
base_inv_shamap = target_git_object_retriever._idmap.get_inventory_sha_map(base_inv.revision_id)
308
base_tree = lookup_object(o.parents[0]).tree
309
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
310
mapping, "", o.tree, base_tree, base_inv, base_inv_shamap,
311
base_ie, None, rev.revision_id, parent_invs, lookup_object,
312
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
314
for (oldpath, newpath, fileid, new_ie) in inv_delta:
316
entries.append((fileid, None, None, None))
318
if new_ie.kind in ("file", "symlink"):
319
entries.append((fileid, "blob", shamap[fileid], new_ie.revision))
320
elif new_ie.kind == "directory":
321
entries.append((fileid, "tree", shamap[fileid], rev.revision_id))
324
target_git_object_retriever._idmap.add_entries(rev.revision_id,
325
rev.parent_ids, head, o.tree, entries)
326
if unusual_modes != {}:
327
for path, mode in unusual_modes.iteritems():
328
warn_unusual_mode(rev.foreign_revid, path, mode)
329
mapping.import_unusual_file_modes(rev, unusual_modes)
331
basis_id = rev.parent_ids[0]
333
basis_id = NULL_REVISION
335
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
336
inv_delta, rev.revision_id, rev.parent_ids,
338
parent_invs_cache.add(rev.revision_id, inv)
339
repo.add_revision(rev.revision_id, rev)
340
if "verify" in debug.debug_flags:
341
new_unusual_modes = mapping.export_unusual_file_modes(rev)
342
if new_unusual_modes != unusual_modes:
343
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
344
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
345
for newsha1, newobj, path in objs:
346
assert path is not None
350
(oldmode, oldsha1) = tree_lookup_path(lookup_object, o.tree, path)
351
if oldsha1 != newsha1:
352
raise AssertionError("%r != %r in %s" % (oldsha1, newsha1, path))
355
def import_git_objects(repo, mapping, object_iter,
356
target_git_object_retriever, heads, pb=None, limit=None):
125
357
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
359
:param repo: Target Bazaar repository
128
360
:param mapping: Mapping to use
129
361
:param object_iter: Iterator over Git objects.
362
:return: Tuple with pack hints and last imported revision id
364
def lookup_object(sha):
366
return object_iter[sha]
368
return target_git_object_retriever[sha]
131
369
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
372
heads = list(set(heads))
373
parent_invs_cache = LRUInventoryCache(repo)
374
target_git_object_retriever.start_write_group() # FIXME: try/finally
140
375
# Find and convert commit objects
141
for o in objects.itervalues():
378
pb.update("finding revisions to fetch", len(graph), None)
380
assert isinstance(head, str)
382
o = lookup_object(head)
142
385
if isinstance(o, Commit):
143
386
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
145
revisions[rev.revision_id] = rev
146
graph.append((rev.revision_id, rev.parent_ids))
387
if repo.has_revision(rev.revision_id):
389
squash_revision(repo, rev)
390
graph.append((o.id, o.parents))
391
heads.extend([p for p in o.parents if p not in checked])
392
elif isinstance(o, Tag):
393
if o.object[1] not in checked:
394
heads.append(o.object[1])
396
trace.warning("Unable to import head object %r" % o)
147
399
# Order the revisions
148
400
# Create the inventory objects
149
for i, revid in enumerate(topo_sort(graph)):
151
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
rev = revisions[revid]
154
# 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
156
# 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)
402
revision_ids = topo_sort(graph)
404
if limit is not None:
405
revision_ids = revision_ids[:limit]
407
for offset in range(0, len(revision_ids), batch_size):
408
repo.start_write_group()
410
for i, head in enumerate(revision_ids[offset:offset+batch_size]):
412
pb.update("fetching revisions", offset+i, len(revision_ids))
413
import_git_commit(repo, mapping, head, lookup_object,
414
target_git_object_retriever,
418
repo.abort_write_group()
421
hint = repo.commit_write_group()
423
pack_hints.extend(hint)
424
target_git_object_retriever.commit_write_group()
425
return pack_hints, last_imported
187
428
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
430
_matching_repo_format = GitRepositoryFormat()
192
433
def _get_repo_format_to_test():
196
437
"""See InterRepository.copy_content."""
197
438
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
440
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
441
mapping=None, fetch_spec=None):
442
self.fetch_refs(revision_id=revision_id, pb=pb,
443
find_ghosts=find_ghosts, mapping=mapping, fetch_spec=fetch_spec)
446
class InterGitNonGitRepository(InterGitRepository):
447
"""Base InterRepository that copies revisions from a Git into a non-Git
450
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
451
mapping=None, fetch_spec=None):
201
452
if mapping is None:
202
453
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
454
if revision_id is not None:
455
interesting_heads = [revision_id]
456
elif fetch_spec is not None:
457
interesting_heads = fetch_spec.heads
459
interesting_heads = None
461
def determine_wants(refs):
463
if interesting_heads is None:
464
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
466
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
467
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)
468
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
469
if pack_hint is not None and self.target._format.pack_compresses:
470
self.target.pack(hint=pack_hint)
471
if interesting_heads is not None:
472
present_interesting_heads = self.target.has_revisions(interesting_heads)
473
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
474
if missing_interesting_heads:
475
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
479
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
480
def report_git_progress(pb, text):
481
text = text.rstrip("\r\n")
482
g = _GIT_PROGRESS_RE.match(text)
484
(text, pct, current, total) = g.groups()
485
pb.update(text, int(current), int(total))
487
pb.update(text, 0, 0)
490
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
491
"""InterRepository that copies revisions from a remote Git into a non-Git
494
def get_target_heads(self):
495
# FIXME: This should be more efficient
496
all_revs = self.target.all_revision_ids()
497
parent_map = self.target.get_parent_map(all_revs)
499
map(all_parents.update, parent_map.itervalues())
500
return set(all_revs) - all_parents
502
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
504
report_git_progress(pb, text)
505
store = BazaarObjectStore(self.target, mapping)
506
self.target.lock_write()
508
heads = self.get_target_heads()
509
graph_walker = store.get_graph_walker(
510
[store._lookup_revision_sha1(head) for head in heads])
513
def record_determine_wants(heads):
514
wants = determine_wants(heads)
515
recorded_wants.extend(wants)
520
create_pb = pb = ui.ui_factory.nested_progress_bar()
522
objects_iter = self.source.fetch_objects(
523
record_determine_wants, graph_walker,
524
store.get_raw, progress)
525
return import_git_objects(self.target, mapping,
526
objects_iter, store, recorded_wants, pb, limit)
534
def is_compatible(source, target):
535
"""Be compatible with GitRepository."""
536
# FIXME: Also check target uses VersionedFile
537
return (isinstance(source, RemoteGitRepository) and
538
target.supports_rich_root() and
539
not isinstance(target, GitRepository))
542
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
543
"""InterRepository that copies revisions from a local Git into a non-Git
546
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
549
wants = determine_wants(self.source._git.get_refs())
214
552
create_pb = pb = ui.ui_factory.nested_progress_bar()
553
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
555
self.target.lock_write()
218
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
224
self.target.commit_write_group()
557
return import_git_objects(self.target, mapping,
558
self.source._git.object_store,
559
target_git_object_retriever, wants, pb, limit)
226
561
self.target.unlock()
added
removed
fetch.py
