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 (
22
from dulwich.object_store import (
35
from bzrlib.errors import (
39
from bzrlib.inventory import (
45
from bzrlib.lru_cache import (
48
from bzrlib.repository import (
51
from bzrlib.revision import (
54
from bzrlib.tsort import (
57
from bzrlib.versionedfile import (
58
FulltextContentFactory,
61
from bzrlib.plugins.git.mapping import (
63
inventory_to_tree_and_blobs,
68
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_inv, base_ie, parent_id,
82
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
73
83
"""Import a git blob object into a bzr repository.
75
:param repo: bzr repository
85
:param texts: VersionedFiles to add to
76
86
:param path: Path in the tree
77
87
:param blob: A git blob
88
:return: Inventory delta for this file
79
90
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)
95
# We just have to hope this is indeed utf-8:
96
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
88
97
ie.executable = executable
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
98
# See if this has changed at all
103
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
107
if (base_sha == hexsha and base_ie.executable == ie.executable
108
and base_ie.kind == ie.kind):
109
# If nothing has changed since the base revision, we're done
111
if base_sha == hexsha and base_ie.kind == ie.kind:
112
ie.text_size = base_ie.text_size
113
ie.text_sha1 = base_ie.text_sha1
114
ie.symlink_target = base_ie.symlink_target
115
if ie.executable == base_ie.executable:
116
ie.revision = base_ie.revision
118
blob = lookup_object(hexsha)
120
blob = lookup_object(hexsha)
121
if ie.kind == "symlink":
123
ie.symlink_target = blob.data
127
ie.text_size = len(blob.data)
128
ie.text_sha1 = osutils.sha_string(blob.data)
129
# Check what revision we should store
131
for pinv in parent_invs:
132
if pinv.revision_id == base_inv.revision_id:
141
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
142
# found a revision in one of the parents to use
143
ie.revision = pie.revision
145
parent_keys.append((file_id, pie.revision))
146
if ie.revision is None:
147
# Need to store a new revision
148
ie.revision = revision_id
149
assert file_id is not None
150
assert ie.revision is not None
151
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), tuple(parent_keys), ie.text_sha1, blob.data)])
152
shamap = [(hexsha, "blob", (ie.file_id, ie.revision))]
156
if base_ie is not None:
157
old_path = base_inv.id2path(file_id)
158
if base_ie.kind == "directory":
159
invdelta.extend(remove_disappeared_children(old_path, base_ie.children, []))
162
invdelta.append((old_path, path, file_id, ie))
163
return (invdelta, shamap)
166
class SubmodulesNotSupported(BzrError):
168
_fmt = """Submodules can not yet be imported (requires nested tree support in Bazaar)."""
172
def import_git_submodule(texts, mapping, path, hexsha, base_inv, base_ie,
173
parent_id, revision_id, parent_invs, shagitmap, lookup_object):
174
raise SubmodulesNotSupported()
177
def remove_disappeared_children(path, base_children, existing_children):
179
deletable = [(osutils.pathjoin(path, k), v) for k,v in base_children.iteritems() if k not in existing_children]
181
(path, ie) = deletable.pop()
182
ret.append((path, None, ie.file_id, None))
183
if ie.kind == "directory":
184
for name, child_ie in ie.children.iteritems():
185
deletable.append((osutils.pathjoin(path, name), child_ie))
189
def import_git_tree(texts, mapping, path, hexsha, base_inv, base_ie, parent_id,
190
revision_id, parent_invs, shagitmap, lookup_object):
92
191
"""Import a git tree object into a bzr repository.
94
:param repo: A Bzr repository object
193
:param texts: VersionedFiles object to add to
95
194
:param path: Path in the tree
96
195
:param tree: A git tree object
97
:param inv: Inventory object
196
:param base_inv: Base inventory against which to return inventory delta
197
:return: Inventory delta for this subtree
99
200
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
201
# We just have to hope this is indeed utf-8:
202
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
205
# Newly appeared here
206
ie.revision = revision_id
207
texts.insert_record_stream([FulltextContentFactory((file_id, ie.revision), (), None, "")])
208
invdelta.append((None, path, file_id, ie))
210
# See if this has changed at all
212
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
216
if base_sha == hexsha:
217
# If nothing has changed since the base revision, we're done
219
if base_ie.kind != "directory":
220
ie.revision = revision_id
221
texts.insert_record_stream([FulltextContentFactory((ie.file_id, ie.revision), (), None, "")])
222
invdelta.append((base_inv.id2path(ie.file_id), path, ie.file_id, ie))
223
if base_ie is not None and base_ie.kind == "directory":
224
base_children = base_ie.children
227
# Remember for next time
228
existing_children = set()
231
tree = lookup_object(hexsha)
232
for mode, name, child_hexsha in tree.entries():
108
233
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):
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
base_children.get(basename), file_id, revision_id, parent_invs, shagitmap,
241
invdelta.extend(subinvdelta)
242
child_modes.update(grandchildmodes)
243
shamap.extend(subshamap)
244
elif S_ISGITLINK(mode): # submodule
245
subinvdelta, grandchildmodes, subshamap = import_git_submodule(
246
texts, mapping, child_path, child_hexsha, base_inv, base_children.get(basename),
247
file_id, revision_id, parent_invs, shagitmap, lookup_object)
248
invdelta.extend(subinvdelta)
249
child_modes.update(grandchildmodes)
250
shamap.extend(subshamap)
252
subinvdelta, subshamap = import_git_blob(texts, mapping,
253
child_path, child_hexsha, base_inv, base_children.get(basename), file_id,
254
revision_id, parent_invs, shagitmap, lookup_object,
255
mode_is_executable(mode), stat.S_ISLNK(mode))
256
invdelta.extend(subinvdelta)
257
shamap.extend(subshamap)
258
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
259
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
260
child_modes[child_path] = mode
261
# Remove any children that have disappeared
262
if base_ie is not None and base_ie.kind == "directory":
263
invdelta.extend(remove_disappeared_children(base_inv.id2path(file_id),
264
base_children, existing_children))
265
shamap.append((hexsha, "tree", (file_id, revision_id)))
266
return invdelta, child_modes, shamap
269
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
125
271
"""Import a set of git objects into a bzr repository.
127
:param repo: Bazaar repository
273
:param repo: Target Bazaar repository
128
274
:param mapping: Mapping to use
129
275
:param object_iter: Iterator over Git objects.
277
def lookup_object(sha):
279
return object_iter[sha]
281
return target_git_object_retriever[sha]
131
282
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
288
parent_invs_cache = LRUCache(50)
140
289
# Find and convert commit objects
141
for o in objects.itervalues():
292
pb.update("finding revisions to fetch", len(graph), None)
294
assert isinstance(head, str)
296
o = lookup_object(head)
298
trace.mutter('missing head %s', head)
142
300
if isinstance(o, Commit):
143
301
rev = mapping.import_commit(o)
144
root_trees[rev.revision_id] = objects[o.tree]
302
if repo.has_revision(rev.revision_id):
304
squash_revision(repo, rev)
305
root_trees[rev.revision_id] = o.tree
145
306
revisions[rev.revision_id] = rev
146
307
graph.append((rev.revision_id, rev.parent_ids))
308
target_git_object_retriever._idmap.add_entry(o.id, "commit",
309
(rev.revision_id, o.tree))
310
heads.extend([p for p in o.parents if p not in checked])
311
elif isinstance(o, Tag):
312
heads.append(o.object[1])
314
trace.warning("Unable to import head object %r" % o)
147
316
# Order the revisions
148
317
# Create the inventory objects
149
318
for i, revid in enumerate(topo_sort(graph)):
150
319
if pb is not None:
151
320
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
153
321
rev = revisions[revid]
154
322
# 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
323
# we need to make sure to import the blobs / trees with the right
156
324
# 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)
326
for parent_id in rev.parent_ids:
328
parent_invs.append(parent_invs_cache[parent_id])
330
parent_inv = repo.get_inventory(parent_id)
331
parent_invs.append(parent_inv)
332
parent_invs_cache[parent_id] = parent_inv
333
if parent_invs == []:
334
base_inv = Inventory(root_id=None)
337
base_inv = parent_invs[0]
338
base_ie = base_inv.root
339
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
340
mapping, "", root_trees[revid], base_inv, base_ie, None, revid,
341
parent_invs, target_git_object_retriever._idmap, lookup_object)
342
target_git_object_retriever._idmap.add_entries(shamap)
343
if unusual_modes != {}:
344
for path, mode in unusual_modes.iteritems():
345
warn_unusual_mode(rev.foreign_revid, path, mode)
346
mapping.import_unusual_file_modes(rev, unusual_modes)
348
basis_id = rev.parent_ids[0]
350
basis_id = NULL_REVISION
352
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
353
inv_delta, rev.revision_id, rev.parent_ids,
355
parent_invs_cache[rev.revision_id] = inv
356
repo.add_revision(rev.revision_id, rev)
357
if "verify" in debug.debug_flags:
358
new_unusual_modes = mapping.export_unusual_file_modes(rev)
359
if new_unusual_modes != unusual_modes:
360
raise AssertionError("unusual modes don't match: %r != %r" % (unusual_modes, new_unusual_modes))
361
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping, unusual_modes)
362
for sha1, newobj, path in objs:
363
assert path is not None
364
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
366
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
368
target_git_object_retriever._idmap.commit()
187
371
class InterGitRepository(InterRepository):
189
_matching_repo_format = GitFormat()
373
_matching_repo_format = GitRepositoryFormat()
192
376
def _get_repo_format_to_test():
196
380
"""See InterRepository.copy_content."""
197
381
self.fetch(revision_id, pb, find_ghosts=False)
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
383
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
385
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
386
mapping=mapping, fetch_spec=fetch_spec)
389
class InterGitNonGitRepository(InterGitRepository):
390
"""Base InterRepository that copies revisions from a Git into a non-Git
393
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
394
mapping=None, fetch_spec=None):
201
395
if mapping is None:
202
396
mapping = self.source.get_mapping()
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
397
if revision_id is not None:
398
interesting_heads = [revision_id]
399
elif fetch_spec is not None:
400
interesting_heads = fetch_spec.heads
402
interesting_heads = None
404
def determine_wants(refs):
406
if interesting_heads is None:
407
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
409
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
210
410
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)
411
pack_hint = self.fetch_objects(determine_wants, mapping, pb)
412
if pack_hint is not None and self.target._format.pack_compresses:
413
self.target.pack(hint=pack_hint)
414
if interesting_heads is not None:
415
present_interesting_heads = self.target.has_revisions(interesting_heads)
416
missing_interesting_heads = set(interesting_heads) - present_interesting_heads
417
if missing_interesting_heads:
418
raise AssertionError("Missing interesting heads: %r" % missing_interesting_heads)
422
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
423
def report_git_progress(pb, text):
424
text = text.rstrip("\r\n")
425
g = _GIT_PROGRESS_RE.match(text)
427
(text, pct, current, total) = g.groups()
428
pb.update(text, int(current), int(total))
430
pb.update(text, 0, 0)
433
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
434
"""InterRepository that copies revisions from a remote Git into a non-Git
437
def get_target_heads(self):
438
# FIXME: This should be more efficient
439
all_revs = self.target.all_revision_ids()
440
parent_map = self.target.get_parent_map(all_revs)
442
map(all_parents.update, parent_map.itervalues())
443
return set(all_revs) - all_parents
445
def fetch_objects(self, determine_wants, mapping, pb=None):
447
report_git_progress(pb, text)
448
store = BazaarObjectStore(self.target, mapping)
449
self.target.lock_write()
451
heads = self.get_target_heads()
452
graph_walker = store.get_graph_walker(
453
[store._lookup_revision_sha1(head) for head in heads])
456
def record_determine_wants(heads):
457
wants = determine_wants(heads)
458
recorded_wants.extend(wants)
463
create_pb = pb = ui.ui_factory.nested_progress_bar()
465
self.target.start_write_group()
467
objects_iter = self.source.fetch_objects(
468
record_determine_wants, graph_walker,
469
store.get_raw, progress)
470
import_git_objects(self.target, mapping, objects_iter,
471
store, recorded_wants, pb)
473
pack_hint = self.target.commit_write_group()
482
def is_compatible(source, target):
483
"""Be compatible with GitRepository."""
484
# FIXME: Also check target uses VersionedFile
485
return (isinstance(source, RemoteGitRepository) and
486
target.supports_rich_root() and
487
not isinstance(target, GitRepository))
490
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
491
"""InterRepository that copies revisions from a local Git into a non-Git
494
def fetch_objects(self, determine_wants, mapping, pb=None):
495
wants = determine_wants(self.source._git.get_refs())
214
498
create_pb = pb = ui.ui_factory.nested_progress_bar()
499
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
216
501
self.target.lock_write()
218
503
self.target.start_write_group()
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
505
import_git_objects(self.target, mapping,
506
self.source._git.object_store,
507
target_git_object_retriever, wants, pb)
224
self.target.commit_write_group()
509
pack_hint = self.target.commit_write_group()
226
512
self.target.unlock()
added
removed
fetch.py
