def main(args):
db = getattr(datastore, '{}_db'.format(args.db))()
if args.prefix:
prefix_from = args.prefix
prefix_to = next_key(args.prefix)
it = db.range(prefix_from, prefix_to, include_value=False)
else:
it = db.iterkeys()
# assumes normalized keys with : as separator
def get_next():
k = it.next()
if k.endswith(':'):
logger.warning("Key with `:` suffix: {!r}".format(k))
k = k.rstrip(':')
if k.startswith(':'):
logger.warning("Key with `:` prefix: {!r}".format(k))
k = k.lstrip(':')
bits = k.rpartition(':')
#logger.debug(bits)
return bits[0], bits[2]
try:
head, tail = get_next()
except StopIteration:
print "Empty database"
return
if not head:
# root key
head = tail
path = [head]
count = [1]
while True:
try:
head, tail = get_next()
except StopIteration:
while path:
print '{:10}\t{}'.format(count.pop(), path.pop())
break
if head == path[-1]:
count[-1] += 1
elif path[-1] in head:
path.append(head)
count.append(1)
else:
while path:
prev = path.pop()
prev_count = count.pop()
print '{:10}\t{}'.format(prev_count, prev)
if not path:
break
if path[-1] in head:
path.append(head)
count.append(1)
if not path:
path.append(head)
count.append(1)
def main(args):
db = getattr(datastore, '{}_db'.format(args.db))()
if args.command == 'get':
for key in args.keys:
print json.dumps(db.get(key))
elif args.command == 'list':
for key in args.keys:
for key in db.range(key, next_key(key), include_value=False):
print key
elif args.command == 'list_values':
for key in args.keys:
for key, value in db.range(key, next_key(key)):
print json.dumps(value)
elif args.command == 'shell':
from IPython import embed; embed()
def main(args):
db = getattr(datastore, '{}_db'.format(args.db))()
if args.command == 'get':
for key in args.keys:
print json.dumps(db.get(key))
elif args.command == 'list':
for key in args.keys:
for key in db.range(key, next_key(key), include_value=False):
print key
elif args.command == 'list_values':
for key in args.keys:
for key, value in db.range(key, next_key(key)):
print json.dumps(value)
elif args.command == 'shell':
from IPython import embed
embed()
def main(args):
# index corpus
db = datastore.corpus_db()
if args.prefix:
items = db.range(args.prefix, next_key(args.prefix))
elif args.from_classified:
items = get_classified_items(args.from_classified, db)
else:
items = db.range()
count = search.index(items, 'corpus', create=args.create)
print "Indexed {} documents".format(count)
def load_keys(fromkey, offset, limit):
tokey = next_key(fromkey)
db = datastore.corpus_db()
it = db.range(fromkey, tokey)
return [v for k, v in itertools.islice(it, offset, offset + limit)]
def load_keys(fromkey, offset, limit):
tokey = next_key(fromkey)
db = datastore.corpus_db()
it = db.range(fromkey, tokey)
return [v for k,v in itertools.islice(it, offset, offset + limit)]