Pythonic in-memory MapReduce.
Project description
Experimental Pythonic MapReduce inspired by Spotify’s luigi framework.
Canonical Word Count Example
Currently there are two MapReduce implementations, one that includes sorting and one that does not. The example below would not benefit from sorting so we can take advantage of the inherent optimization of not sorting. The API is the same but tinymr.memory.MRSerial() sorts after partitioning and again between the reducer() and final_reducer().
import json
import re
import sys
from tinymr.memory import MRSerial
class WordCount(MRSerial):
def __init__(self):
self.pattern = re.compile('[\W_]+')
def mapper(self, item):
for word in item.split():
word = self.pattern.sub('', word)
if word:
yield word.lower(), 1
def reducer(self, key, values):
yield key, sum(values)
def final_reducer(self, pairs):
return {k: tuple(v)[0] for k, v in pairs}
wc = WordCount()
with open('LICENSE.txt') as f:
out = wc(f)
print(json.dumps(out, indent=4, sort_keys=True))
Truncated output:
{
"a": 1,
"above": 2,
"advised": 1,
"all": 1,
"and": 8,
"andor": 1
}
Developing
$ git clone https://github.com/geowurster/tinymr.git
$ cd tinymr
$ pip install -e .\[dev\]
$ py.test tests --cov tinymr --cov-report term-missing
License
See LICENSE.txt
Changelog
See CHANGES.md
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
tinymr-0.1.tar.gz
(16.8 kB
view hashes)
Built Distribution
tinymr-0.1-py2.py3-none-any.whl
(15.1 kB
view hashes)