def _initialize(self, path, start, length):
self._stream = DataInputStream(FileInputStream(path))
if length == 0:
self._end = self._stream.getPos() + self._stream.length()
else:
self._end = self._stream.getPos() + length
# Parse Header
version_block = self._stream.read(len(VERSION))
self._version = version_block[3]
if self._version > VERSION[3]:
raise VersionMismatchException(VERSION[3], self._version)
if self._version < BLOCK_COMPRESS_VERSION:
# Same as below, but with UTF8 Deprecated Class
raise NotImplementedError
else:
key_class_name = Text.readString(self._stream)
value_class_name = Text.readString(self._stream)
self._key_class = hadoopClassFromName(key_class_name)
self._value_class = hadoopClassFromName(value_class_name)
if ord(self._version) > 2:
self._decompress = self._stream.readBoolean()
else:
self._decompress = False
if self._version >= BLOCK_COMPRESS_VERSION:
self._block_compressed = self._stream.readBoolean()
else:
self._block_compressed = False
# setup compression codec
if self._decompress:
if self._version >= CUSTOM_COMPRESS_VERSION:
codec_class = Text.readString(self._stream)
self._codec = CodecPool().getDecompressor(codec_class)
else:
self._codec = CodecPool().getDecompressor()
self._metadata = Metadata()
if self._version >= VERSION_WITH_METADATA:
self._metadata.readFields(self._stream)
if self._version > 1:
self._sync = self._stream.read(SYNC_HASH_SIZE)
self._header_end = self._stream.getPos()
def getValueClass(self):
if not self._value_class:
self._value_class = hadoopClassFromName(self._value_class_name)
return self._value_class
def getKeyClass(self):
if not self._key_class:
self._key_class = hadoopClassFromName(self._key_class_name)
return self._key_class