To get this branch, use:
bzr branch
http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar
« back to all changes in this revision
Viewing changes to fetch.py
- browse files at revision 0.257.1
- compare with another revision
- download diff
- download tarball
- view history from revision 0.257.1
- Committer: Jelmer Vernooij
- Date: 2010-06-28 22:30:34 UTC
- mto: (0.200.953 trunk)
- mto: This revision was merged to the branch mainline in revision 6960.
- Revision ID: jelmer@samba.org-20100628223034-vylrgdyakmqoupl6
use transport repo objects even for local access.
- files added:
- hg.py
- files renamed:
- shamap.py => cache.py
- tests/test_shamap.py => tests/test_cache.py
- files modified:
- HACKING
added
removed
fetch.py
1
# Copyright (C) 2008 Jelmer Vernooij <jelmer@samba.org>
1
# Copyright (C) 2008-2010 Jelmer Vernooij <jelmer@samba.org>
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
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
16
16
17
from cStringIO import (
18
StringIO,
19
)
20
import dulwich as git
21
17
from dulwich.objects import (
22
18
Commit,
23
19
Tag,
20
Tree,
24
21
S_ISGITLINK,
25
22
)
26
23
from dulwich.object_store import (
27
24
tree_lookup_path,
28
25
)
26
from itertools import (
27
imap,
28
)
29
import posixpath
30
import re
29
31
import stat
30
32
31
33
from bzrlib import (
33
35
osutils,
34
36
trace,
35
37
ui,
36
urlutils,
37
38
)
38
39
from bzrlib.errors import (
39
InvalidRevisionId,
40
BzrError,
40
41
NoSuchId,
41
NoSuchRevision,
42
42
)
43
43
from bzrlib.inventory import (
44
44
Inventory,
45
45
InventoryDirectory,
46
46
InventoryFile,
47
47
InventoryLink,
48
)
49
from bzrlib.lru_cache import (
50
LRUCache,
48
TreeReference,
51
49
)
52
50
from bzrlib.repository import (
53
51
InterRepository,
55
53
from bzrlib.revision import (
56
54
NULL_REVISION,
57
55
)
56
from bzrlib.revisiontree import (
57
RevisionTree,
58
)
58
59
from bzrlib.tsort import (
59
60
topo_sort,
60
61
)
61
62
from bzrlib.versionedfile import (
62
FulltextContentFactory,
63
ChunkedContentFactory,
63
64
)
64
65
65
66
from bzrlib.plugins.git.mapping import (
66
67
DEFAULT_FILE_MODE,
67
inventory_to_tree_and_blobs,
68
68
mode_is_executable,
69
squash_revision,
70
text_to_blob,
69
mode_kind,
71
70
warn_unusual_mode,
72
71
)
73
72
from bzrlib.plugins.git.object_store import (
74
73
BazaarObjectStore,
74
LRUTreeCache,
75
_tree_to_objects,
75
76
)
76
77
from bzrlib.plugins.git.remote import (
77
78
RemoteGitRepository,
78
79
)
79
80
from bzrlib.plugins.git.repository import (
80
GitRepository,
81
GitRepository,
81
82
GitRepositoryFormat,
82
83
LocalGitRepository,
83
84
)
84
85
85
86
86
def import_git_blob(texts, mapping, path, hexsha, base_inv, parent_id,
87
revision_id, parent_invs, shagitmap, lookup_object, executable, symlink):
87
def import_git_blob(texts, mapping, path, name, (base_hexsha, hexsha),
88
base_inv, parent_id, revision_id,
89
parent_invs, lookup_object, (base_mode, mode), store_updater,
90
lookup_file_id):
88
91
"""Import a git blob object into a bzr repository.
89
92
90
93
:param texts: VersionedFiles to add to
92
95
:param blob: A git blob
93
96
:return: Inventory delta for this file
94
97
"""
95
file_id = mapping.generate_file_id(path)
96
if symlink:
98
if mapping.is_control_file(path):
99
return []
100
if base_hexsha == hexsha and base_mode == mode:
101
# If nothing has changed since the base revision, we're done
102
return []
103
file_id = lookup_file_id(path)
104
if stat.S_ISLNK(mode):
97
105
cls = InventoryLink
98
106
else:
99
107
cls = InventoryFile
100
# We just have to hope this is indeed utf-8:
101
ie = cls(file_id, urlutils.basename(path).decode("utf-8"), parent_id)
102
ie.executable = executable
103
# See if this has changed at all
104
try:
105
base_ie = base_inv[file_id]
106
except NoSuchId:
107
base_ie = None
108
base_sha = None
109
else:
110
try:
111
base_sha = shagitmap.lookup_blob(file_id, base_ie.revision)
112
except KeyError:
113
base_sha = None
114
else:
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
118
return [], []
119
if base_sha == hexsha and base_ie.kind == ie.kind:
108
ie = cls(file_id, name.decode("utf-8"), parent_id)
109
ie.executable = mode_is_executable(mode)
110
if base_hexsha == hexsha and mode_kind(base_mode) == mode_kind(mode):
111
base_ie = base_inv[base_inv.path2id(path)]
120
112
ie.text_size = base_ie.text_size
121
113
ie.text_sha1 = base_ie.text_sha1
122
114
ie.symlink_target = base_ie.symlink_target
132
124
ie.text_size = None
133
125
ie.text_sha1 = None
134
126
else:
135
ie.text_size = len(blob.data)
136
ie.text_sha1 = osutils.sha_string(blob.data)
127
ie.text_size = sum(imap(len, blob.chunked))
128
ie.text_sha1 = osutils.sha_strings(blob.chunked)
137
129
# Check what revision we should store
138
130
parent_keys = []
139
131
for pinv in parent_invs:
140
if pinv.revision_id == base_inv.revision_id:
141
pie = base_ie
142
if pie is None:
143
continue
144
else:
145
try:
146
pie = pinv[file_id]
147
except NoSuchId:
148
continue
149
if pie.text_sha1 == ie.text_sha1 and pie.executable == ie.executable and pie.symlink_target == ie.symlink_target:
132
try:
133
pie = pinv[file_id]
134
except NoSuchId:
135
continue
136
if (pie.text_sha1 == ie.text_sha1 and
137
pie.executable == ie.executable and
138
pie.symlink_target == ie.symlink_target):
150
139
# found a revision in one of the parents to use
151
140
ie.revision = pie.revision
152
141
break
153
parent_keys.append((file_id, pie.revision))
142
parent_key = (file_id, pie.revision)
143
if not parent_key in parent_keys:
144
parent_keys.append(parent_key)
154
145
if ie.revision is None:
155
146
# Need to store a new revision
156
147
ie.revision = revision_id
157
assert file_id is not None
158
148
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))]
161
else:
162
shamap = []
163
if file_id in base_inv:
164
old_path = base_inv.id2path(file_id)
149
if ie.kind == 'symlink':
150
chunks = []
151
else:
152
chunks = blob.chunked
153
texts.insert_record_stream([
154
ChunkedContentFactory((file_id, ie.revision),
155
tuple(parent_keys), ie.text_sha1, chunks)])
156
invdelta = []
157
if base_hexsha is not None:
158
old_path = path # Renames are not supported yet
159
if stat.S_ISDIR(base_mode):
160
invdelta.extend(remove_disappeared_children(base_inv, old_path,
161
lookup_object(base_hexsha), [], lookup_object))
165
162
else:
166
163
old_path = None
167
invdelta = [(old_path, path, file_id, ie)]
168
invdelta.extend(remove_disappeared_children(base_inv, base_ie, []))
169
return (invdelta, shamap)
170
171
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)
175
176
177
def remove_disappeared_children(base_inv, base_ie, existing_children):
178
if base_ie is None or base_ie.kind != 'directory':
179
return []
164
invdelta.append((old_path, path, file_id, ie))
165
if base_hexsha != hexsha:
166
store_updater.add_object(blob, ie, path)
167
return invdelta
168
169
170
class SubmodulesRequireSubtrees(BzrError):
171
_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'."""
172
internal = False
173
174
175
def import_git_submodule(texts, mapping, path, name, (base_hexsha, hexsha),
176
base_inv, parent_id, revision_id, parent_invs, lookup_object,
177
(base_mode, mode), store_updater, lookup_file_id):
178
if base_hexsha == hexsha and base_mode == mode:
179
return [], {}
180
file_id = lookup_file_id(path)
181
ie = TreeReference(file_id, name.decode("utf-8"), parent_id)
182
ie.revision = revision_id
183
if base_hexsha is None:
184
oldpath = None
185
else:
186
oldpath = path
187
ie.reference_revision = mapping.revision_id_foreign_to_bzr(hexsha)
188
texts.insert_record_stream([
189
ChunkedContentFactory((file_id, ie.revision), (), None, [])])
190
invdelta = [(oldpath, path, file_id, ie)]
191
return invdelta, {}
192
193
194
def remove_disappeared_children(base_inv, path, base_tree, existing_children,
195
lookup_object):
196
"""Generate an inventory delta for removed children.
197
198
:param base_inv: Base inventory against which to generate the
199
inventory delta.
200
:param path: Path to process
201
:param base_tree: Git Tree base object
202
:param existing_children: Children that still exist
203
:param lookup_object: Lookup a git object by its SHA1
204
:return: Inventory delta, as list
205
"""
180
206
ret = []
181
deletable = [v for k,v in base_ie.children.iteritems() if k not in existing_children]
182
while deletable:
183
ie = deletable.pop()
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())
207
for name, mode, hexsha in base_tree.iteritems():
208
if name in existing_children:
209
continue
210
c_path = posixpath.join(path, name.decode("utf-8"))
211
file_id = base_inv.path2id(c_path)
212
assert file_id is not None
213
ret.append((c_path, None, file_id, None))
214
if stat.S_ISDIR(mode):
215
ret.extend(remove_disappeared_children(
216
base_inv, c_path, lookup_object(hexsha), [], lookup_object))
187
217
return ret
188
218
189
219
190
def import_git_tree(texts, mapping, path, hexsha, base_inv, parent_id,
191
revision_id, parent_invs, shagitmap, lookup_object):
220
def import_git_tree(texts, mapping, path, name, (base_hexsha, hexsha),
221
base_inv, parent_id, revision_id, parent_invs,
222
lookup_object, (base_mode, mode), store_updater,
223
lookup_file_id, allow_submodules=False):
192
224
"""Import a git tree object into a bzr repository.
193
225
194
226
:param texts: VersionedFiles object to add to
197
229
:param base_inv: Base inventory against which to return inventory delta
198
230
:return: Inventory delta for this subtree
199
231
"""
232
if base_hexsha == hexsha and base_mode == mode:
233
# If nothing has changed since the base revision, we're done
234
return [], {}
200
235
invdelta = []
201
file_id = mapping.generate_file_id(path)
236
file_id = lookup_file_id(path)
202
237
# We just have to hope this is indeed utf-8:
203
ie = InventoryDirectory(file_id, urlutils.basename(path.decode("utf-8")),
204
parent_id)
205
try:
206
base_ie = base_inv[file_id]
207
except NoSuchId:
208
# Newly appeared here
209
base_ie = None
238
ie = InventoryDirectory(file_id, name.decode("utf-8"), parent_id)
239
tree = lookup_object(hexsha)
240
if base_hexsha is None:
241
base_tree = None
242
old_path = None # Newly appeared here
243
else:
244
base_tree = lookup_object(base_hexsha)
245
old_path = path # Renames aren't supported yet
246
if base_tree is None or type(base_tree) is not Tree:
210
247
ie.revision = revision_id
211
texts.add_lines((file_id, ie.revision), (), [])
212
invdelta.append((None, path, file_id, ie))
213
else:
214
# See if this has changed at all
215
try:
216
base_sha = shagitmap.lookup_tree(file_id, base_inv.revision_id)
217
except KeyError:
218
pass
219
else:
220
if base_sha == hexsha:
221
# If nothing has changed since the base revision, we're done
222
return [], {}, []
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))
248
invdelta.append((old_path, path, ie.file_id, ie))
249
texts.insert_record_stream([
250
ChunkedContentFactory((ie.file_id, ie.revision), (), None, [])])
227
251
# Remember for next time
228
252
existing_children = set()
229
253
child_modes = {}
230
shamap = []
231
tree = lookup_object(hexsha)
232
for mode, name, child_hexsha in tree.entries():
233
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)
250
else:
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,
254
for child_mode, name, child_hexsha in tree.entries():
255
existing_children.add(name)
256
child_path = posixpath.join(path, name)
257
if type(base_tree) is Tree:
258
try:
259
child_base_mode, child_base_hexsha = base_tree[name]
260
except KeyError:
261
child_base_hexsha = None
262
child_base_mode = 0
263
else:
264
child_base_hexsha = None
265
child_base_mode = 0
266
if stat.S_ISDIR(child_mode):
267
subinvdelta, grandchildmodes = import_git_tree(texts, mapping,
268
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
269
file_id, revision_id, parent_invs, lookup_object,
270
(child_base_mode, child_mode), store_updater, lookup_file_id,
271
allow_submodules=allow_submodules)
272
elif S_ISGITLINK(child_mode): # submodule
273
if not allow_submodules:
274
raise SubmodulesRequireSubtrees()
275
subinvdelta, grandchildmodes = import_git_submodule(texts, mapping,
276
child_path, name, (child_base_hexsha, child_hexsha), base_inv,
277
file_id, revision_id, parent_invs, lookup_object,
278
(child_base_mode, child_mode), store_updater, lookup_file_id)
279
else:
280
subinvdelta = import_git_blob(texts, mapping, child_path, name,
281
(child_base_hexsha, child_hexsha), base_inv, file_id,
282
revision_id, parent_invs, lookup_object,
283
(child_base_mode, child_mode), store_updater, lookup_file_id)
284
grandchildmodes = {}
285
child_modes.update(grandchildmodes)
286
invdelta.extend(subinvdelta)
287
if child_mode not in (stat.S_IFDIR, DEFAULT_FILE_MODE,
258
288
stat.S_IFLNK, DEFAULT_FILE_MODE|0111):
259
child_modes[child_path] = mode
289
child_modes[child_path] = child_mode
260
290
# 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
264
265
266
def import_git_objects(repo, mapping, object_iter, target_git_object_retriever,
267
heads, pb=None):
291
if base_tree is not None and type(base_tree) is Tree:
292
invdelta.extend(remove_disappeared_children(base_inv, old_path,
293
base_tree, existing_children, lookup_object))
294
store_updater.add_object(tree, ie, path)
295
return invdelta, child_modes
296
297
298
def verify_commit_reconstruction(target_git_object_retriever, lookup_object,
299
o, rev, ret_tree, parent_trees, mapping, unusual_modes):
300
new_unusual_modes = mapping.export_unusual_file_modes(rev)
301
if new_unusual_modes != unusual_modes:
302
raise AssertionError("unusual modes don't match: %r != %r" % (
303
unusual_modes, new_unusual_modes))
304
# Verify that we can reconstruct the commit properly
305
rec_o = target_git_object_retriever._reconstruct_commit(rev, o.tree, True)
306
if rec_o != o:
307
raise AssertionError("Reconstructed commit differs: %r != %r" % (
308
rec_o, o))
309
diff = []
310
new_objs = {}
311
for path, obj, ie in _tree_to_objects(ret_tree, parent_trees,
312
target_git_object_retriever._cache.idmap, unusual_modes, mapping.BZR_DUMMY_FILE):
313
old_obj_id = tree_lookup_path(lookup_object, o.tree, path)[1]
314
new_objs[path] = obj
315
if obj.id != old_obj_id:
316
diff.append((path, lookup_object(old_obj_id), obj))
317
for (path, old_obj, new_obj) in diff:
318
while (old_obj.type_name == "tree" and
319
new_obj.type_name == "tree" and
320
sorted(old_obj) == sorted(new_obj)):
321
for name in old_obj:
322
if old_obj[name][0] != new_obj[name][0]:
323
raise AssertionError("Modes for %s differ: %o != %o" %
324
(path, old_obj[name][0], new_obj[name][0]))
325
if old_obj[name][1] != new_obj[name][1]:
326
# Found a differing child, delve deeper
327
path = posixpath.join(path, name)
328
old_obj = lookup_object(old_obj[name][1])
329
new_obj = new_objs[path]
330
break
331
raise AssertionError("objects differ for %s: %r != %r" % (path,
332
old_obj, new_obj))
333
334
335
def import_git_commit(repo, mapping, head, lookup_object,
336
target_git_object_retriever, trees_cache):
337
o = lookup_object(head)
338
rev = mapping.import_commit(o,
339
lambda x: target_git_object_retriever.lookup_git_sha(x)[1][0])
340
# We have to do this here, since we have to walk the tree and
341
# we need to make sure to import the blobs / trees with the right
342
# path; this may involve adding them more than once.
343
parent_trees = trees_cache.revision_trees(rev.parent_ids)
344
if parent_trees == []:
345
base_inv = Inventory(root_id=None)
346
base_tree = None
347
base_mode = None
348
else:
349
base_inv = parent_trees[0].inventory
350
base_tree = lookup_object(o.parents[0]).tree
351
base_mode = stat.S_IFDIR
352
store_updater = target_git_object_retriever._get_updater(rev)
353
store_updater.add_object(o, None, None)
354
lookup_file_id = mapping.get_fileid_map(lookup_object, o.tree).lookup_file_id
355
inv_delta, unusual_modes = import_git_tree(repo.texts,
356
mapping, "", u"", (base_tree, o.tree), base_inv,
357
None, rev.revision_id, [p.inventory for p in parent_trees],
358
lookup_object, (base_mode, stat.S_IFDIR), store_updater,
359
lookup_file_id,
360
allow_submodules=getattr(repo._format, "supports_tree_reference", False))
361
store_updater.finish()
362
if unusual_modes != {}:
363
for path, mode in unusual_modes.iteritems():
364
warn_unusual_mode(rev.foreign_revid, path, mode)
365
mapping.import_unusual_file_modes(rev, unusual_modes)
366
try:
367
basis_id = rev.parent_ids[0]
368
except IndexError:
369
basis_id = NULL_REVISION
370
base_inv = None
371
rev.inventory_sha1, inv = repo.add_inventory_by_delta(basis_id,
372
inv_delta, rev.revision_id, rev.parent_ids,
373
base_inv)
374
ret_tree = RevisionTree(repo, inv, rev.revision_id)
375
trees_cache.add(ret_tree)
376
repo.add_revision(rev.revision_id, rev)
377
if "verify" in debug.debug_flags:
378
verify_commit_reconstruction(target_git_object_retriever,
379
lookup_object, o, rev, ret_tree, parent_trees, mapping,
380
unusual_modes)
381
382
383
def import_git_objects(repo, mapping, object_iter,
384
target_git_object_retriever, heads, pb=None, limit=None):
268
385
"""Import a set of git objects into a bzr repository.
269
386
270
387
:param repo: Target Bazaar repository
271
388
:param mapping: Mapping to use
272
389
:param object_iter: Iterator over Git objects.
390
:return: Tuple with pack hints and last imported revision id
273
391
"""
274
392
def lookup_object(sha):
275
393
try:
276
394
return object_iter[sha]
277
395
except KeyError:
278
396
return target_git_object_retriever[sha]
279
# TODO: a more (memory-)efficient implementation of this
280
397
graph = []
281
root_trees = {}
282
revisions = {}
283
398
checked = set()
284
heads = list(heads)
285
parent_invs_cache = LRUCache(50)
399
heads = list(set(heads))
400
trees_cache = LRUTreeCache(repo)
286
401
# Find and convert commit objects
287
402
while heads:
288
403
if pb is not None:
294
409
except KeyError:
295
410
continue
296
411
if isinstance(o, Commit):
297
rev = mapping.import_commit(o)
412
rev = mapping.import_commit(o, lambda x: None)
298
413
if repo.has_revision(rev.revision_id):
299
414
continue
300
squash_revision(repo, rev)
301
root_trees[rev.revision_id] = o.tree
302
revisions[rev.revision_id] = rev
303
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))
415
graph.append((o.id, o.parents))
306
416
heads.extend([p for p in o.parents if p not in checked])
307
417
elif isinstance(o, Tag):
308
heads.append(o.object[1])
418
if o.object[1] not in checked:
419
heads.append(o.object[1])
309
420
else:
310
421
trace.warning("Unable to import head object %r" % o)
311
checked.add(head)
422
checked.add(o.id)
423
del checked
312
424
# Order the revisions
313
425
# Create the inventory objects
314
for i, revid in enumerate(topo_sort(graph)):
315
if pb is not None:
316
pb.update("fetching revisions", i, len(graph))
317
rev = revisions[revid]
318
# 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
320
# path; this may involve adding them more than once.
321
parent_invs = []
322
for parent_id in rev.parent_ids:
426
batch_size = 1000
427
revision_ids = topo_sort(graph)
428
pack_hints = []
429
if limit is not None:
430
revision_ids = revision_ids[:limit]
431
last_imported = None
432
for offset in range(0, len(revision_ids), batch_size):
433
target_git_object_retriever.start_write_group()
434
try:
435
repo.start_write_group()
323
436
try:
324
parent_invs.append(parent_invs_cache[parent_id])
325
except KeyError:
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)
437
for i, head in enumerate(
438
revision_ids[offset:offset+batch_size]):
439
if pb is not None:
440
pb.update("fetching revisions", offset+i,
441
len(revision_ids))
442
import_git_commit(repo, mapping, head, lookup_object,
443
target_git_object_retriever, trees_cache)
444
last_imported = head
445
except:
446
repo.abort_write_group()
447
raise
448
else:
449
hint = repo.commit_write_group()
450
if hint is not None:
451
pack_hints.extend(hint)
452
except:
453
target_git_object_retriever.abort_write_group()
454
raise
331
455
else:
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)
341
try:
342
basis_id = rev.parent_ids[0]
343
except IndexError:
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)
357
if oldobj != newobj:
358
raise AssertionError("%r != %r in %s" % (oldobj, newobj, path))
359
360
target_git_object_retriever._idmap.commit()
456
target_git_object_retriever.commit_write_group()
457
return pack_hints, last_imported
361
458
362
459
363
460
class InterGitRepository(InterRepository):
372
469
"""See InterRepository.copy_content."""
373
470
self.fetch(revision_id, pb, find_ghosts=False)
374
471
375
def fetch(self, revision_id=None, pb=None, find_ghosts=False, mapping=None,
376
fetch_spec=None):
377
self.fetch_refs(revision_id=revision_id, pb=pb, find_ghosts=find_ghosts,
378
mapping=mapping, fetch_spec=fetch_spec)
379
380
472
381
473
class InterGitNonGitRepository(InterGitRepository):
382
"""Base InterRepository that copies revisions from a Git into a non-Git
474
"""Base InterRepository that copies revisions from a Git into a non-Git
383
475
repository."""
384
476
385
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
477
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
386
478
mapping=None, fetch_spec=None):
387
479
if mapping is None:
388
480
mapping = self.source.get_mapping()
398
490
if interesting_heads is None:
399
491
ret = [sha for (ref, sha) in refs.iteritems() if not ref.endswith("^{}")]
400
492
else:
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)
493
ret = [self.source.lookup_bzr_revision_id(revid)[0] for revid in interesting_heads if revid not in (None, NULL_REVISION)]
494
return [rev for rev in ret if not self.target.has_revision(self.source.lookup_foreign_revision_id(rev))]
495
(pack_hint, _) = self.fetch_objects(determine_wants, mapping, pb)
496
if pack_hint is not None and self.target._format.pack_compresses:
497
self.target.pack(hint=pack_hint)
404
498
return self._refs
405
499
406
500
501
_GIT_PROGRESS_RE = re.compile(r"(.*?): +(\d+)% \((\d+)/(\d+)\)")
502
def report_git_progress(pb, text):
503
text = text.rstrip("\r\n")
504
g = _GIT_PROGRESS_RE.match(text)
505
if g is not None:
506
(text, pct, current, total) = g.groups()
507
pb.update(text, int(current), int(total))
508
else:
509
pb.update(text, 0, 0)
510
511
407
512
class InterRemoteGitNonGitRepository(InterGitNonGitRepository):
408
"""InterRepository that copies revisions from a remote Git into a non-Git
513
"""InterRepository that copies revisions from a remote Git into a non-Git
409
514
repository."""
410
515
411
def fetch_objects(self, determine_wants, mapping, pb=None):
516
def get_target_heads(self):
517
# FIXME: This should be more efficient
518
all_revs = self.target.all_revision_ids()
519
parent_map = self.target.get_parent_map(all_revs)
520
all_parents = set()
521
map(all_parents.update, parent_map.itervalues())
522
return set(all_revs) - all_parents
523
524
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
412
525
def progress(text):
413
pb.update("git: %s" % text.rstrip("\r\n"), 0, 0)
526
report_git_progress(pb, text)
414
527
store = BazaarObjectStore(self.target, mapping)
415
528
self.target.lock_write()
416
529
try:
417
heads = self.target.get_graph().heads(self.target.all_revision_ids())
530
heads = self.get_target_heads()
418
531
graph_walker = store.get_graph_walker(
419
532
[store._lookup_revision_sha1(head) for head in heads])
420
533
recorded_wants = []
423
536
wants = determine_wants(heads)
424
537
recorded_wants.extend(wants)
425
538
return wants
426
539
427
540
create_pb = None
428
541
if pb is None:
429
542
create_pb = pb = ui.ui_factory.nested_progress_bar()
430
543
try:
431
self.target.start_write_group()
432
try:
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)
438
finally:
439
self.target.commit_write_group()
544
objects_iter = self.source.fetch_objects(record_determine_wants,
545
graph_walker, store.get_raw, progress)
546
return import_git_objects(self.target, mapping,
547
objects_iter, store, recorded_wants, pb, limit)
440
548
finally:
441
549
if create_pb:
442
550
create_pb.finished()
446
554
@staticmethod
447
555
def is_compatible(source, target):
448
556
"""Be compatible with GitRepository."""
449
# FIXME: Also check target uses VersionedFile
450
return (isinstance(source, RemoteGitRepository) and
557
return (isinstance(source, RemoteGitRepository) and
451
558
target.supports_rich_root() and
452
not isinstance(target, GitRepository))
559
not isinstance(target, GitRepository) and
560
target.texts is not None)
453
561
454
562
455
563
class InterLocalGitNonGitRepository(InterGitNonGitRepository):
456
"""InterRepository that copies revisions from a local Git into a non-Git
564
"""InterRepository that copies revisions from a local Git into a non-Git
457
565
repository."""
458
566
459
def fetch_objects(self, determine_wants, mapping, pb=None):
567
def fetch_objects(self, determine_wants, mapping, pb=None, limit=None):
568
"""Fetch objects.
569
"""
460
570
wants = determine_wants(self.source._git.get_refs())
461
571
create_pb = None
462
572
if pb is None:
465
575
try:
466
576
self.target.lock_write()
467
577
try:
468
self.target.start_write_group()
469
try:
470
import_git_objects(self.target, mapping,
471
self.source._git.object_store,
472
target_git_object_retriever, wants, pb)
473
finally:
474
self.target.commit_write_group()
578
return import_git_objects(self.target, mapping,
579
self.source._git.object_store,
580
target_git_object_retriever, wants, pb, limit)
475
581
finally:
476
582
self.target.unlock()
477
583
finally:
481
587
@staticmethod
482
588
def is_compatible(source, target):
483
589
"""Be compatible with GitRepository."""
484
# FIXME: Also check target uses VersionedFile
485
return (isinstance(source, LocalGitRepository) and
590
return (isinstance(source, LocalGitRepository) and
486
591
target.supports_rich_root() and
487
not isinstance(target, GitRepository))
592
not isinstance(target, GitRepository) and
593
target.texts is not None)
488
594
489
595
490
596
class InterGitGitRepository(InterGitRepository):
491
597
"""InterRepository that copies between Git repositories."""
492
598
493
def fetch_refs(self, revision_id=None, pb=None, find_ghosts=False,
599
def fetch_objects(self, determine_wants, mapping, pb=None):
600
def progress(text):
601
trace.note("git: %s", text)
602
graphwalker = self.target._git.get_graph_walker()
603
if (isinstance(self.source, LocalGitRepository) and
604
isinstance(self.target, LocalGitRepository)):
605
return self.source._git.fetch(self.target._git, determine_wants,
606
progress)
607
elif (isinstance(self.source, LocalGitRepository) and
608
isinstance(self.target, RemoteGitRepository)):
609
raise NotImplementedError
610
elif (isinstance(self.source, RemoteGitRepository) and
611
isinstance(self.target, LocalGitRepository)):
612
f, commit = self.target._git.object_store.add_thin_pack()
613
try:
614
refs = self.source._git.fetch_pack(determine_wants,
615
graphwalker, f.write, progress)
616
commit()
617
return refs
618
except:
619
f.close()
620
raise
621
else:
622
raise AssertionError
623
624
def fetch(self, revision_id=None, pb=None, find_ghosts=False,
494
625
mapping=None, fetch_spec=None, branches=None):
495
626
if mapping is None:
496
627
mapping = self.source.get_mapping()
497
def progress(text):
498
trace.info("git: %s", text)
499
628
r = self.target._git
500
629
if revision_id is not None:
501
630
args = [mapping.revision_id_bzr_to_foreign(revision_id)[0]]
507
636
determine_wants = r.object_store.determine_wants_all
508
637
else:
509
638
determine_wants = lambda x: [y for y in args if not y in r.object_store]
639
return self.fetch_objects(determine_wants, mapping)[0]
510
640
511
graphwalker = r.get_graph_walker()
512
f, commit = r.object_store.add_thin_pack()
513
try:
514
refs = self.source.fetch_pack(determine_wants, graphwalker,
515
f.write, progress)
516
commit()
517
return refs
518
except:
519
f.close()
520
raise
521
641
522
642
@staticmethod
523
643
def is_compatible(source, target):
524
644
"""Be compatible with GitRepository."""
525
return (isinstance(source, GitRepository) and
645
return (isinstance(source, GitRepository) and
526
646
isinstance(target, GitRepository))
Loggerhead is a web-based interface for Breezy
Version: 2.0.2.dev0