weighted_sum.py

"""
``$ articlequality weighted_sum -h``
::
    Extracts probabilities assigned to each class from the output of 
    revscoring score utility and outputs the weighted sum of the article
    quality predicted where each class is represented as a weight sorted
    in a yaml config file.
    
    Usage:
        weighted_sum <weights> [--scores=<path>] [--output=<path>]
    Options:
        -h --help           Show this documentation.
        <weights>           Path to a yaml file containing class weights
        --scores=<path>     Path to a file containting scores generated by 
                            `revscoring score`. [default: <stdin>]
        --output=<path>     Path to a file to write new observations
                            (with "weighted_sum") out to. 
                            [default: <stdout>]
"""
import sys

import mwapi
from docopt import docopt
import yamlconf


def main(argv=None):
    args = docopt(__doc__, argv=argv)

    weights = yamlconf.load(open(args['<weights>']))

    if args['--scores'] == '<stdin>':
        revision_scores = read_revision_scores(sys.stdin)
    else:
        revision_scores = read_revision_scores(open(args['--scores']))

    if args['--output'] == '<stdout>':
        output = sys.stdout
    else:
        output = open(args['--output'])


    run(revision_scores, weights, output)


def read_revision_scores(f):
    for line in f:
        rev_id, json_doc = line.split('\t', 1)
        
        yield int(rev_id), json.loads(json_doc)


def run(revision_scores, weights, output):
    
    for rev_id, score in revision_scores:
        # Apply the weights
        
        # Put the weighted sum in the score doc
        score['weighted_sum'] = weighted_sum

        # Write the output
        output.write("{0}\t{1}".format(rev_id, json.dumps(score)))