Python数据分析时刻PV-MRJob详解

1.1. 前言

这边我们使用Python的M/R框架MRJob来分析.
1.2. M/R步骤

Mapper: 将以行数据解析成 key=hh value=1的形式
Shuffle: 通过Shuffle后的结果会生成以 key 的值排序的 value迭代器
结果如: 09 [1, 1, 1 ... 1, 1]
Reduce: 在这边我们计算出 09 这一小时的访问量
输出如: 09 sum([1, 1, 1 ... 1, 1])
1.3. 代码

cat mr_pv_hour.py
# -*- coding: utf-8 -*-
 
from mrjob.job import MRJob
from ng_line_parser import NgLineParser
 
class MRPVHour(MRJob):
 
    ng_line_parser = NgLineParser()
 
    def mapper(self, _, line):
        self.ng_line_parser.parse(line)
        dy, tm = str(self.ng_line_parser.access_time).split()
        h, m, s = tm.split(':')
        yield h, 1 # 每小时的
        yield 'total', 1 # 所有的
 
    def reducer(self, key, values):
        yield key, sum(values)
 
def main():
    MRPVHour.run()
 
if __name__ == '__main__':
    main()

运行统计和输出结果

python mr_pv_hour.py < www.ttmark.com.access.log
 
No configs found; falling back on auto-configuration
Creating temp directory /tmp/mr_pv_hour.root.20160924.130542.359063
Running step 1 of 1...
reading from STDIN
Streaming final output from /tmp/mr_pv_hour.root.20160924.130542.359063/output...
"00"    31539
"01"    34824
"02"    27895
"03"    29669
"04"    27742
"05"    26797
"06"    29384
"07"    31102
"08"    38257
"09"    43060
"10"    48064
"11"    57923
"12"    56413
"13"    57971
"14"    47260
"15"    46364
"16"    45721
"17"    48884
"18"    49318
"19"    49162
"20"    43641
"21"    42525
"22"    40371
"23"    34953
"total" 988839
Removing temp directory /tmp/mr_pv_hour.root.20160924.130542.359063...