python – 如何使用s3对象名作为MRJob映射器的输入,而不是s3对象本身?
作者:互联网
我遗漏了一些关于Yelp的mrjob job库的明显信息.设置MRJob类几乎非常简单.在file或stdin上运行它也是如此.但是,如何在本地或在s3中将文件输入更改为作业,比如在s3存储桶中更改?
像这样的东西.假设我想计算我的S3存储桶中以字符串’foo’开头的所有对象:
import re
class MRCountS3Objects(MRJob):
define mapper(self, _, botoS3Key):
if re.match('^foo', botoS3Key.name):
yield 'foo', 1
define reduce(self, name, occurrences):
yield name, sum(occurrences)
这是一个非常人为的例子,但你可能会得到我的漂移.如何告诉MRJob操作s3对象流,忽略对象的内容?我看到了S3Filesystem.get_s3_keys()method,它让我得到了我需要的流,但我不确定从那里开始.
解决方法:
想到至少一种方法来实现这一目标.您的MRJob具有可以分配给任何迭代器的stdin属性,然后您可以以编程方式运行该作业.例如,此代码应该适用于我的桶的密钥名称:
from mrjob.job import MRJob
from mrjob.emr import EMRJobRunner
class MRS3KeyProcessor(MRJob):
# Do some MRJob stuff.
...
def s3_name_generator(bucket):
"""Generator that returns boto.s3.Key names.
"""
# Could also use raw boto here.
emr = EMRJobRunner()
key_stream = emr.fs.get_s3_keys(bucket)
for key in key_stream:
yield key.name
def main():
# The '-' argument signifies that we use stdin.
mr_job = MRCountS3Objects(['--runner', 'inline', '-'])
stdin = s3_name_generator('my-bucket')
mr_job.stdin = stdin
results = []
with mr_job.make_runner() as runner:
runner.run()
for line in runner.stream_output():
key, value = mr_job.parse_output_line(line)
results.append((key, value))
print(results)
if __name__ == '__main__':
main()
标签:python,mapreduce,boto,elastic-map-reduce,mrjob 来源: https://codeday.me/bug/20190625/1284310.html