0

我想修改高级教程的 innerjoin-example 以便它可以使用 mapreduce 进行稀疏矩阵乘法（由 Ullman 描述）。因此，我需要第二个 map-reduce 步骤求和结果矩阵中相等位置的值。

不幸的是，我无法将 CsvInnerJoin 类的第一个 reduce 函数的输出获取到 SumJob 的 map 函数中。

import sys
sys.path.append("/home/damian/disco/lib/")
from disco.core import Job, result_iterator
from disco.worker.classic.func import chain_reader
import csv, sys


if __name__ == '__main__':
    input_filename = "input.csv"
    output_filename = "output.csv"
    if len(sys.argv) > 1:
        input_filename = sys.argv[1]
        if len(sys.argv) > 2:
            output_filename = sys.argv[2]

    from CsvInnerJoiner import CsvInnerJoiner
    from SumJob import SumJob

    job = CsvInnerJoiner().run(input=[input_filename])
    job = SumJob().run() (******************)

    with open(output_filename, 'w') as fp:
        writer = csv.writer(fp)
        for url_key, descriptors in result_iterator(job.wait(show=True)):
            writer.writerow([url_key] + descriptors)

CsvInnerJoiner.py 是这个文件：

import sys
sys.path.append("/home/damian/disco/lib/")
from disco.core import Job, result_iterator
from disco.worker.classic.func import chain_reader
import csv, sys
class CsvInnerJoiner(Job):
    partitions = 2
    sort = True

    def map(self, row, params):
        yield row[0], row[1:]

    @staticmethod
    def map_reader(fd, size, url, params):
        reader = csv.reader(fd, delimiter=',')
        for row in reader:
            yield row

    #@staticmethod
def reduce(self, rows_iter, out, params):
    from disco.util import kvgroup
    from itertools import chain
    #for url_key, descriptors in kvgroup(sorted(rows_iter)):
    for url_key, descriptors in kvgroup(rows_iter):
        merged_descriptors = list(chain.from_iterable(descriptors))
        print url_key,"_______",merged_descriptors
        if len(merged_descriptors) > 3:
            Alist = merged_descriptors[:merged_descriptors.index("B")]
            Blist = merged_descriptors[merged_descriptors.index("B"):]
            Alistlength = len(Alist)/3
            Blistlength = len(Blist)/3
            for i in range(Alistlength):
                for j in range(Blistlength):
                    container = int(Alist[3*i+2])*int(Blist[3*j+2])
                    yield [Alist[3*i+1],Blist[3*j+1]],container
                    #out.add(Alist[3*i+1],[Blist[3*j+1],container])

SumJob.py 是这样的：

import sys
sys.path.append("/home/damian/disco/lib/")
from disco.core import Job, result_iterator
from disco.worker.classic.func import chain_reader
import csv, sys


class SumJob(Job):
    map_reader = staticmethod(chain_reader)

    @staticmethod
    def map(self,key_value, params):
        print "KEY::::::",str(key_value[0])
        print "VAL::::::",str(key_value[1])
        yield key_value[0], key_value[1]
    @staticmethod    
    def reduce(self,key_value,out, params):
        Summe = sum(key_value[1])
        out.add(key_value[0],Summe)

问题是我不知道如何更改 (**) 行，以便将第一个 reduce 步骤的第二个输出作为第二个 map 函数的输入。

非常感谢你的帮助！达米安

4

1 回答 1

0

您可以使用 map/reduce 阶段的输出作为另一个阶段的输入（从 的返回job.wait()）。

job1 = SumJob().run(input=[...])
job2 = SumJob().run(input=[...])

output = SomeOtherJob.run(input=[job1.wait(), job2.wait()]).wait(show=True)
for key, value in result_iterator(output):
    print key, value

我不是那段代码对我有用的专家（我实现了pagerank具有许多阶段和多次迭代的算法）。

于 2013-09-23T08:24:37.183 回答