我尝试使用python的hadoop流来计算输入键的平均值。以下是Map器、合路器和减速机的代码:
# mapper:
import sys
def map(argv):
line = sys.stdin.readline()
try:
while line:
word, num = line.split()
num = int(num)
print word+'\t'+str(num)
line = sys.stdin.readline()
except Exception, ex:
print 'mapper ex:'+str(ex)
return None
if __name__ == "__main__":
map(sys.argv)
# combiner
import sys
def combine(argv):
line = sys.stdin.readline()
cur_word = ''
cur_num = 0
cur_times = 0
try:
while line:
word, num = line.split('\t')
if word != cur_word:
if cur_word != '':
print cur_word+'\t'+str(cur_num)+'\t'+str(cur_times)
cur_word = word
cur_num = 0
cur_times = 0
cur_num += int(num)
cur_times += 1
line = sys.stdin.readline()
print cur_word+'\t'+str(cur_num)+'\t'+str(cur_times)
except Exception, ex:
print 'except:{0}'.format(ex)
return None
if __name__ == "__main__":
combine(sys.argv)
# reducer
import sys
def reduce(argv):
line = sys.stdin.readline()
cur_word = ''
cur_num = 0
cur_times = 0
try:
while line:
word, num, times = line.split('\t')
if word != cur_word:
if cur_word != '':
if cur_times != 0:
avr = cur_num / cur_times
print cur_word+'\t'+str(cur_num)+'\t'+str(cur_times)+'\t'+str(avr)
else:
print cur_word+'\t'+str(cur_num)+'\t'+str(cur_times)+'\t'+'0'
cur_word = word
cur_num = 0
cur_times = 0
cur_num += int(num)
cur_times += int(times)
line = sys.stdin.readline()
if cur_times != 0:
avr = cur_num / cur_times
print cur_word+'\t'+str(cur_num)+'\t'+str(cur_times)+'\t'+str(avr)
else:
print cur_word+'\t'+str(cur_num)+'\t'+str(cur_times)+'\t'+'0'
except Exception, ex:
print 'except:{0}'.format(ex)
return None
if __name__ == "__main__":
reduce(sys.argv)
这似乎是一个简单的过程,不是吗?但是reduce每次都失败了。但是,如果我不使用combiner,而是使用combiner.py作为减速机,它就可以工作了。
有人愿意帮忙吗,非常感谢。
暂无答案!
目前还没有任何答案,快来回答吧!