def parse(self, path):
buf = open(path, "rb").read()
eof = len(buf)
logging.debug("path: %s", path)
logging.debug("parsing INX of %s(%s) bytes", eof, hex(eof))
self._set_pos(0)
num_entries = read_int(buf, self._pos, 1)
logging.debug("number of entries: %s", num_entries)
self._add_pos(1)
result = INXParseResult(buf)
for i in xrange(num_entries):
logging.debug("entry %s of %s", i+1, num_entries)
content_start = read_int(buf, self._pos, 8)
logging.debug("content_start: %s (%s)", content_start, hex(content_start))
self._add_pos(8)
name_length = read_int(buf, self._pos, 1)
logging.debug("name_length: %s", name_length)
self._add_pos(1)
name = read_str(buf, self._pos, name_length)
logging.debug("name: %s", name)
self._add_pos(name_length)
result.append(INXEntry(result, name, content_start))
return result
def read_vint(f, p):
"""
Decodes variable-length format for position integers.
http://lucene.apache.org/core/old_versioned_docs/versions/3_0_0/fileformats.html#VInt
"""
eof = len(f)
length = 0
total = 0
while True:
if p == eof:
raise EOFError()
i = read_int(f, p, 1)
has_remain = False
if 0b10000000 <= i: #high-order bit is 1
has_remain = True
i = i ^ 0b10000000 #high-order bit to 0
else:
has_remain = False
total = total | i << 7 * length #append low-order 7 bits to the result
length += 1
p += 1
if not has_remain:
break
return VInt(length, total)