wikia_video_stats.py

#!/bin/python
import operator

def main():
    fp = open("reportTotalVideosOnWikiaAll.csv", "r")
    vids_by_provider = {}
    vids_by_wiki = {}

    # Lines are in the csv form provider,count,wiki
    for line in fp.readlines():
        vals = line.split(",")

        if vals[0] not in vids_by_provider:
            vids_by_provider[vals[0]] = 0
        if vals[2] not in vids_by_wiki:
            vids_by_wiki[vals[2]] = 0
        
        # Keep a tally of total videos by wiki, and by provider
        vids_by_provider[vals[0]] += int(vals[1])
        vids_by_wiki[vals[2]] += int(vals[1])
        
    fp.close()

    # Sort both dictionaries based on number of videos creating a list of tuples
    # eg [('minecraftinfinity', 1), ('lyinggame', 1), ('dope101', 1)...]
    sorted_vids_wiki = sorted(vids_by_wiki.iteritems(), key=operator.itemgetter(1), reverse=True)
    # eg [('youtube', 748344), ('ign', 100159), ('screenplay', 60724)...]
    sorted_vids_provider = sorted(vids_by_provider.iteritems(), key=operator.itemgetter(1), reverse=True)

    # Total number of videos, grouped by provider
    print "\nVideos By Provider"
    print "================="
    sum_total = 0
    for vid in sorted_vids_provider:
        sum_total += vid[1]
        print "%s: %s" % (vid[0], vid[1])

    # Total number of videos overall
    print "\nTotal vids"
    print "=========="
    print sum_total

    # Top 10 wikis with most videos (skipping the first 2 which are video and community)
    print "\nTop 10 wikis"
    print "============"
    for i in range(2, 12):
        print "%s: %s" % (sorted_vids_wiki[i][0].strip(), sorted_vids_wiki[i][1])

    
    print "\nTotal wikis with videos"
    print "======================="
    print len(sorted_vids_wiki)

if __name__ == "__main__":
    main()