def fsck_missing_blobs(vol, cwd):
'''Look for blobs in tree or snaps which are not in blobstore.'''
trees = vol.trees()
tree_items = concatMap(lambda t: zipFrom(t,iter(t)))
tree_links = ffilter(uncurry(lambda snap, item: item.is_link()))
broken_tree_links = partial(
ifilter,
uncurry(lambda snap, item: not vol.bs.exists(item.csum())))
checksum_grouper = partial(groupby,
uncurry(lambda snap, item: item.csum()))
def broken_link_printr(csum, snap_items):
print(csum)
for (snap, item) in snap_items:
print( '',
snap.name,
item.to_path(vol.root).relative_to(cwd),
sep='\t')
broken_links_printr = fmap(identify(uncurry(broken_link_printr)))
num_bad_blobs = pipeline(
tree_items,
tree_links,
broken_tree_links,
checksum_grouper,
broken_links_printr,
count)(trees)
return num_bad_blobs
def test_zipFrom():
assert list(zipFrom(1, [2, 3, 4])) == [(1, 2), (1, 3), (1, 4)]
assert list(zipFrom(1, [])) == []
def dbg_ui(argv, cwd):
exitcode = 0
args = docopt(DBG_USAGE, argv)
vol = getvol(cwd)
if args['reverse']:
csum = args['<csum>']
if args['--all']:
trees = vol.trees()
elif args['--snap']:
trees = [vol.snapdb.read(args['--snap'])]
else:
trees = [vol.tree()]
tree_items = concatMap(lambda t: zipFrom(t,iter(t)))
tree_links = ffilter(uncurry(lambda snap, item: item.is_link()))
matching_links = ffilter(uncurry(lambda snap, item: item.csum() == csum))
def link_printr(snap_item):
(snap, item) = snap_item
print(snap.name, item.to_path(vol.root).relative_to(cwd))
links_printr = fmap(identify(link_printr))
pipeline(
tree_items,
tree_links,
matching_links,
links_printr,
consume)(trees)
elif args['key']:
db = vol.keydb
key = args['<key>']
if args['read']:
printNotNone(db.readraw(key))
elif args['delete']:
db.delete(key)
elif args['list']:
for v in db.list(key):
print(v)
elif args['write']:
value = args['<value>']
db.write(key, value)
elif args['walk']:
if args['root']:
printr = json_printr if args.get('--json') else snapshot_printr
printr(encode_snapshot(vol.tree()))
elif args['snap']:
#TODO could add a test for output encoding.
#TODO could add a test for snap format. Leading '/' on paths.
printr = json_printr if args.get('--json') else snapshot_printr
printr(encode_snapshot(vol.snapdb.read(args['<snapshot>'])))
elif args['userdata']:
blobs = vol.bs.blobs()
printr = json_printr if args.get('--json') else strs_printr
printr(blobs)
elif args['keys']:
printr = json_printr if args.get('--json') else strs_printr
printr(vol.keydb.list())
elif args['checksum']:
#TODO <checksum> <full path>
paths = empty_default(map(lambda x: Path(x, cwd), args['<path>']), [vol.root])
for p in paths:
print(p.checksum(), p.relative_to(cwd))
elif args['link']:
f = Path(args['<file>'], cwd)
b = ingest(args['<target>'])
if not vol.bs.exists(b):
print("blob %s doesn't exist" % b)
if args['--remote']:
remote = vol.remotedb.read(args['--remote'])
else:
raise(ValueError("aborting due to missing blob"))
vol.bs.fetch_blob(remote.bs, b)
else:
pass #b exists, can we check its checksum?
vol.bs.link_to_blob(f, b)
elif args['rewrite-links']:
target = Path(args['<target>'], cwd)
for item in vol.tree():
if not item.is_link():
continue
path = item.to_path(vol.root)
new = vol.repair_link(path)
if new is not None:
print("Relinked %s to %s" % (path.relative_to(cwd), new))
elif args['missing']:
tree_csums = pipeline(
ffilter(lambda item: item.is_link()),
fmap(lambda item: item.csum()),
set
)(iter(vol.tree()))
snapNames = args['<snap>']
def missing_printr(csum, pathStrs):
paths = sorted(imap(lambda pathStr: vol.root.join(pathStr), pathStrs))
for path in paths:
print("%s\t%s" % (csum, path.relative_to(cwd)))
missing_csum2pathStr = pipeline(
fmap(vol.snapdb.read),
concatMap(iter),
ffilter(lambda item: item.is_link()),
ffilter(lambda item: not vol.is_ignored(item.to_path(vol.root), None)),
ffilter(lambda item: item.csum() not in tree_csums),
partial(groupby, lambda item: item.csum()),
ffilter(uncurry(lambda csum, items: every(lambda item: not item.to_path(vol.root).exists(), items))),
fmap(uncurry(lambda csum, items: (csum, list(imap(lambda item: item.pathStr(), items))))),
fmap(uncurry(missing_printr)),
count
)(snapNames)
elif args['blobtype']:
for blob in args['<blob>']:
blob = ingest(blob)
#TODO here csum_to_path is really needed.
print(
blob,
maybe("unknown", vol.bs.csum_to_path(blob).filetype()))
elif args['blob']:
for csum in args['<blob>']:
csum = ingest(csum)
#TODO here csum_to_path is needed
print(csum,
vol.bs.csum_to_path(csum).relative_to(cwd))
elif args['s3']:
bucket = args['<bucket>']
prefix = args['<prefix>']
access_id, secret_key = load_s3_creds(None)
s3bs = S3Blobstore(bucket, prefix, access_id, secret_key)
if args['list']:
pipeline(fmap(print), consume)(s3bs.blobs()())
elif args['upload']:
quiet = args.get('--quiet')
print("Fetching remote blobs")
s3_blobs = set(tqdm(s3bs.blobs()(), disable=quiet, desc="Fetching remote blobs", smoothing=1.0, dynamic_ncols=True, maxinterval=1.0))
print("Remote Blobs: %s" % len(s3_blobs))
print("Fetching local blobs") #TODO we are looking at tree, so blobs in snaps won't be sent.
tree_blobs = set(tqdm(pipeline(
ffilter(lambda x: x.is_link()),
fmap(lambda x: x.csum()),
uniq,
)(iter(vol.tree())), disable=quiet, desc="Calculating local blobs", smoothing=1.0, dynamic_ncols=True, maxinterval=1.0))
print("Local Blobs: %s" % len(tree_blobs))
upload_blobs = tree_blobs - s3_blobs
print("Uploading %s blobs to s3" % len(upload_blobs))
with tqdm(desc="Uploading to S3", disable=quiet, total=len(upload_blobs), smoothing=1.0, dynamic_ncols=True, maxinterval=1.0) as pbar:
def update_pbar(blob):
pbar.update(1)
pbar.set_description("Uploaded %s" % blob)
def upload(blob):
s3bs.upload(blob, vol.bs.csum_to_path(blob))()
return blob
all_success = pipeline(
ffilter(lambda x: x not in s3_blobs),
pfmap(upload, workers=2),
fmap(identify(update_pbar)),
partial(every, identity),
)(upload_blobs)
if all_success:
print("Successfully uploaded")
else:
print("Failed to upload")
exitcode = 1
elif args['check']:
num_corrupt_blobs = pipeline(
ffilter(lambda obj: obj['ETag'][1:-1] != obj['blob']),
fmap(identify(lambda obj: print(obj['blob'], obj['ETag'][1:-1]))),
count
)(s3bs.blob_stats()())
if num_corrupt_blobs == 0:
print("All S3 blobs etags match")
else:
exitcode = 2
return exitcode
def farmfs_ui(argv, cwd):
exitcode = 0
args = docopt(UI_USAGE, argv)
if args['mkfs']:
root = userPath2Path(args['<root>'] or ".", cwd)
data = userPath2Path(args['<data>'], cwd) if args.get('<data>') else Path(".farmfs/userdata", root)
mkfs(root, data)
print("FileSystem Created %s using blobstore %s" % (root, data))
else:
vol = getvol(cwd)
paths = empty_default(map(lambda x: userPath2Path(x, cwd), args['<path>']), [vol.root])
def delta_printr(delta):
deltaPath = delta.path(vol.root).relative_to(cwd)
print("diff: %s %s %s" % (delta.mode, deltaPath, delta.csum))
stream_delta_printr = fmap(identify(delta_printr))
def op_printr(op):
(blob_op, tree_op, (desc, path)) = op
print(desc % path.relative_to(cwd))
stream_op_printr = fmap(identify(op_printr))
if args['status']:
get_thawed = fmap(vol.thawed)
pipeline(get_thawed,
concat,
fmap(lambda p: p.relative_to(cwd)),
fmap(print),
consume)(paths)
elif args['freeze']:
def printr(freeze_op):
s = "Imported %s with checksum %s" % \
(freeze_op['path'].relative_to(cwd),
freeze_op['csum'])
if freeze_op['was_dup']:
print(s, "was a duplicate")
else:
print(s)
importer = fmap(vol.freeze)
get_thawed = fmap(vol.thawed)
print_list = fmap(printr)
pipeline(get_thawed, concat, importer, print_list, consume)(paths)
elif args['thaw']:
def printr(path):
print("Exported %s" % path.relative_to(cwd))
exporter = fmap(vol.thaw)
get_frozen = fmap(vol.frozen)
print_list = fmap(printr)
pipeline(get_frozen, concat, exporter, print_list, consume)(paths)
elif args['fsck']:
fsck_actions = {
'--broken': (fsck_missing_blobs, 1),
'--frozen-ignored': (fsck_frozen_ignored, 4),
'--blob-permissions': (fsck_blob_permissions, 8),
'--checksums': (fsck_checksum_mismatches, 2),
}
fsck_tasks = [action for (verb, action) in fsck_actions.items() if args[verb]]
if len(fsck_tasks) == 0:
# No options were specified, run the whole suite.
fsck_tasks = fsck_actions.values()
for foo, fail_code in fsck_tasks:
exitcode = exitcode | (foo(vol, cwd) and fail_code)
elif args['count']:
trees = vol.trees()
tree_items = concatMap(lambda t: zipFrom(t,iter(t)))
tree_links = ffilter(uncurry(lambda snap, item: item.is_link()))
checksum_grouper = partial(groupby,
uncurry(lambda snap, item: item.csum()))
def count_printr(csum, snap_items):
print(csum, count(snap_items))
for (snap, item) in snap_items:
print(snap.name, item.to_path(vol.root).relative_to(cwd))
counts_printr = fmap(identify(uncurry(count_printr)))
pipeline(
tree_items,
tree_links,
checksum_grouper,
counts_printr,
consume
)(trees)
elif args['similarity']:
dir_a = userPath2Path(args['<dir_a>'], cwd)
dir_b = userPath2Path(args['<dir_b>'], cwd)
print("left", "both", "right", "jaccard_similarity", sep="\t")
print(* vol.similarity(dir_a, dir_b), sep="\t")
elif args['gc']:
applyfn = fmap(identity) if args.get('--noop') else fmap(vol.bs.delete_blob)
fns = [fmap(identify(partial(print, "Removing"))),
applyfn,
consume]
pipeline(*fns)(sorted(vol.unused_blobs(vol.items())))
elif args['snap']:
snapdb = vol.snapdb
if args['list']:
#TODO have an optional argument for which remote.
print("\n".join(snapdb.list()))
else:
name = args['<snap>']
if args['delete']:
snapdb.delete(name)
elif args['make']:
snapdb.write(name, vol.tree())
else:
snap = snapdb.read(name)
if args['read']:
for i in snap:
print(i)
elif args['restore']:
tree = vol.tree()
diff = tree_diff(vol.tree(), snap)
pipeline(
stream_delta_printr,
tree_patcher(vol, vol),
stream_op_printr,
stream_op_doer,
consume)(diff)
elif args['diff']:
diff = tree_diff(vol.tree(), snap)
pipeline(stream_delta_printr, consume)(diff)
elif args['remote']:
if args["add"]:
remote_vol = getvol(userPath2Path(args['<root>'], cwd))
vol.remotedb.write(args['<remote>'], remote_vol)
elif args["remove"]:
vol.remotedb.delete(args['<remote>'])
elif args["list"]:
if args["<remote>"]:
remote_vol = vol.remotedb.read(args['<remote>'])
print("\n".join(remote_vol.snapdb.list()))
else:
for remote_name in vol.remotedb.list():
remote_vol = vol.remotedb.read(remote_name)
print(remote_name, remote_vol.root)
elif args['pull'] or args['diff']:
remote_vol = vol.remotedb.read(args['<remote>'])
snap_name = args['<snap>']
remote_snap = remote_vol.snapdb.read(snap_name) if snap_name else remote_vol.tree()
diff = tree_diff(vol.tree(), remote_snap)
if args['pull']:
patcher = tree_patcher(vol, remote_vol)
pipeline(
stream_delta_printr,
patcher,
stream_op_printr,
stream_op_doer,
consume)(diff)
else: # diff
pipeline(stream_delta_printr, consume)(diff)
return exitcode