def __init__(self, filename): super(InputReader, self).__init__() self.reader_class_name = "org.apache.hadoop.mapred.TextInputFormat" if "mapred.input.format.class" in os.environ: self.reader_class_name = os.environ["mapred.input.format.class"] self.reader_class = hadoopClassFromName(self.reader_class_name) self.reader = self.reader_class(filename) self.key = self.reader.getKeyClass()() self.value = self.reader.getValueClass()()
def __init__(self, filename): super(InputReader, self).__init__() self.reader_class_name = "org.apache.hadoop.mapred.TextInputFormat" if "mapred.input.format.class" in os.environ: self.reader_class_name = os.environ["mapred.input.format.class"] self.reader_class = hadoopClassFromName(self.reader_class_name) self.reader = self.reader_class(filename) self.key = self.reader.getKeyClass()() self.value = self.reader.getValueClass()()
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
def getValueClass(self): if not self._value_class: self._value_class = hadoopClassFromName("hadoop.io.Text") return self._value_class
def getKeyClass(self): if not self._key_class: self._key_class = hadoopClassFromName("hadoop.io.Text") return self._key_class
def getValueClass(self): if not self._value_class: self._value_class = hadoopClassFromName("hadoop.io.Text") return self._value_class
def getKeyClass(self): if not self._key_class: self._key_class = hadoopClassFromName("hadoop.io.Text") return self._key_class