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 (
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 (
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
75
25
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):
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):
84
73
"""Import a git blob object into a bzr repository.
86
:param texts: VersionedFiles to add to
75
:param repo: bzr repository
87
76
:param path: Path in the tree
88
77
:param blob: A git blob
89
:return: Inventory delta for this file
91
79
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)
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)
98
88
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 and base_ie.kind == ie.kind:
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):
91
def import_git_tree(repo, mapping, path, tree, inv, parent_invs, lookup_object):
167
92
"""Import a git tree object into a bzr repository.
169
:param texts: VersionedFiles object to add to
94
:param repo: A Bzr repository object
170
95
:param path: Path in the tree
171
96
: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
97
:param inv: Inventory object
176
99
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():
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
207
108
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
subinvdelta, subshamap = import_git_blob(texts, mapping,
220
child_path, child_hexsha, base_inv, file_id, revision_id,
221
parent_invs, shagitmap, lookup_object,
222
bool(fs_mode & 0111), stat.S_ISLNK(mode))
223
invdelta.extend(subinvdelta)
224
shamap.extend(subshamap)
225
if mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
226
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
227
child_modes[child_path] = mode
228
# Remove any children that have disappeared
229
if base_ie is not None and base_ie.kind == 'directory':
230
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
233
invdelta.append((base_inv.id2path(ie.file_id), None, ie.file_id, None))
234
if ie.kind == "directory":
235
deletable.extend(ie.children.values())
236
shamap.append((hexsha, "tree", (file_id, revision_id)))
237
return invdelta, child_modes, shamap
240
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
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):
242
125
"""Import a set of git objects into a bzr repository.
244
:param repo: Target Bazaar repository
127
:param repo: Bazaar repository
245
128
:param mapping: Mapping to use
246
129
:param object_iter: Iterator over Git objects.
248
def lookup_object(sha):
250
return object_iter[sha]
252
return target_git_object_retriever[sha]
253
131
# TODO: a more (memory-)efficient implementation of this
133
for i, o in enumerate(object_iter):
135
pb.update("fetching objects", i)
259
parent_invs_cache = LRUCache(50)
260
140
# Find and convert commit objects
263
pb.update("finding revisions to fetch", len(graph), None)
265
assert isinstance(head, str)
267
o = lookup_object(head)
141
for o in objects.itervalues():
270
142
if isinstance(o, Commit):
271
143
rev = mapping.import_commit(o)
272
if repo.has_revision(rev.revision_id):
274
root_trees[rev.revision_id] = o.tree
144
root_trees[rev.revision_id] = objects[o.tree]
275
145
revisions[rev.revision_id] = rev
276
146
graph.append((rev.revision_id, rev.parent_ids))
277
target_git_object_retriever._idmap.add_entry(o.id, "commit",
278
(rev.revision_id, o.tree))
279
heads.extend([p for p in o.parents if p not in checked])
280
elif isinstance(o, Tag):
281
heads.append(o.object[1])
283
trace.warning("Unable to import head object %r" % o)
285
147
# Order the revisions
286
148
# Create the inventory objects
287
149
for i, revid in enumerate(topo_sort(graph)):
288
150
if pb is not None:
289
151
pb.update("fetching revisions", i, len(graph))
152
root_tree = root_trees[revid]
290
153
rev = revisions[revid]
291
154
# We have to do this here, since we have to walk the tree and
292
# we need to make sure to import the blobs / trees with the right
155
# we need to make sure to import the blobs / trees with the riht
293
156
# path; this may involve adding them more than once.
295
for parent_id in rev.parent_ids:
297
parent_invs.append(parent_invs_cache[parent_id])
299
parent_inv = repo.get_inventory(parent_id)
300
parent_invs.append(parent_inv)
301
parent_invs_cache[parent_id] = parent_inv
302
if parent_invs == []:
303
base_inv = Inventory(root_id=None)
305
base_inv = parent_invs[0]
306
inv_delta, unusual_modes, shamap = import_git_tree(repo.texts,
307
mapping, "", root_trees[revid], base_inv, None, revid,
308
parent_invs, target_git_object_retriever._idmap, lookup_object)
309
target_git_object_retriever._idmap.add_entries(shamap)
310
if unusual_modes != {}:
311
ret = "unusual modes: \n"
312
for item in unusual_modes.iteritems():
313
ret += "\t%s: %o\n" % item
314
raise AssertionError(ret)
316
basis_id = rev.parent_ids[0]
318
basis_id = NULL_REVISION
319
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
320
inv_delta, rev.revision_id, rev.parent_ids)
321
parent_invs_cache[rev.revision_id] = inv
322
repo.add_revision(rev.revision_id, rev)
323
if "verify" in debug.debug_flags:
324
objs = inventory_to_tree_and_blobs(inv, repo.texts, mapping)
325
for sha1, newobj, path in objs:
326
assert path is not None
327
oldobj = tree_lookup_path(lookup_object, root_trees[revid], path)
328
assert oldobj == newobj, "%r != %r in %s" % (oldobj, newobj, path)
330
target_git_object_retriever._idmap.commit()
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)
333
187
class InterGitRepository(InterRepository):
335
_matching_repo_format = GitRepositoryFormat()
189
_matching_repo_format = GitFormat()
338
192
def _get_repo_format_to_test():
342
196
"""See InterRepository.copy_content."""
343
197
self.fetch(revision_id, pb, find_ghosts=False)
345
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
347
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
348
mapping=mapping, fetch_spec=fetch_spec)
351
class InterGitNonGitRepository(InterGitRepository):
352
"""Base InterRepository that copies revisions from a Git into a non-Git
355
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
356
mapping=None, fetch_spec=None):
199
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
357
201
if mapping is None:
358
202
mapping = self.source.get_mapping()
359
if revision_id is not None:
360
interesting_heads = [revision_id]
361
elif fetch_spec is not None:
362
interesting_heads = fetch_spec.heads
364
interesting_heads = None
366
def determine_wants(refs):
368
if interesting_heads is None:
369
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
204
pb.note("git: %s", text)
205
def determine_wants(heads):
206
if revision_id is None:
371
ret = [mapping.revision_id_bzr_to_foreign(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
209
ret = [mapping.revision_id_bzr_to_foreign(revision_id)]
372
210
return [rev for rev in ret if not self.target.has_revision(mapping.revision_id_foreign_to_bzr(rev))]
373
self.fetch_objects(determine_wants, mapping, pb)
377
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
378
"""InterRepository that copies revisions from a remote Git into a non-Git
381
def fetch_objects(self, determine_wants, mapping, pb=None):
383
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
384
store = BazaarObjectStore(self.target, mapping)
385
self.target.lock_write()
387
heads = self.target.get_graph().heads(self.target.all_revision_ids())
388
graph_walker = store.get_graph_walker(
389
[store._lookup_revision_sha1(head) for head in heads])
392
def record_determine_wants(heads):
393
wants = determine_wants(heads)
394
recorded_wants.extend(wants)
399
create_pb = pb = ui.ui_factory.nested_progress_bar()
401
self.target.start_write_group()
403
objects_iter = self.source.fetch_objects(
404
record_determine_wants, graph_walker,
405
store.get_raw, progress)
406
import_git_objects(self.target, mapping, objects_iter,
407
store, recorded_wants, pb)
409
self.target.commit_write_group()
417
def is_compatible(source, target):
418
"""Be compatible with GitRepository."""
419
# FIXME: Also check target uses VersionedFile
420
return (isinstance(source, RemoteGitRepository) and
421
target.supports_rich_root() and
422
not isinstance(target, GitRepository))
425
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
426
"""InterRepository that copies revisions from a remote Git into a non-Git
429
def fetch_objects(self, determine_wants, mapping, pb=None):
430
wants = determine_wants(self.source._git.get_refs())
211
graph_walker = BzrFetchGraphWalker(self.target, mapping)
433
214
create_pb = pb = ui.ui_factory.nested_progress_bar()
434
target_git_object_retriever = BazaarObjectStore(self.target, mapping)
436
216
self.target.lock_write()
438
218
self.target.start_write_group()
440
import_git_objects(self.target, mapping,
441
self.source._git.object_store,
442
target_git_object_retriever, wants, pb)
220
import_git_objects(self.target, mapping,
221
iter(self.source.fetch_objects(determine_wants, graph_walker,
444
224
self.target.commit_write_group()
added
removed
fetch.py
