mefi_analysis

import csv
import operator
import matplotlib.pyplot as plt

def rape_analysis():
    rape_related_tags = { #python set is nonperformant,
        #although I don't know how much that matters
        "rape": 1,
        "anti-rape": 1,
        "rapeculture": 1,
        "sexualassault": 1,
        "consent": 1
        }
    related_posts = {}
    with open("/home/curuinor/data/mefi_dump/tagdata_mefi.txt", "r") as mefi_tags_file:
        reader = csv.reader(mefi_tags_file, delimiter="\t")
        for row in reader:
            if row[3] in rape_related_tags:
                related_posts[int(row[1])] = set()
    with open("/home/curuinor/data/mefi_dump/commentdata_mefi.txt", "r") as mefi_comment_file:
        reader = csv.reader(mefi_comment_file, delimiter="\t")
        for row in reader:
            if row[1] == "postid":
                continue
            if int(row[1]) in related_posts: #row[1] is post id
                related_posts[int(row[1])].add(row[2])
    return related_posts.items()

if __name__ == "__main__":
    num_users_per_rape_post = sorted(rape_analysis(), key=operator.itemgetter(0))
    xs = map(lambda x: x[0], num_users_per_rape_post)
    print xs
    ys = map(lambda x: len(x[1]), num_users_per_rape_post)
    print ys
    plt.plot(xs, ys)
    plt.title("postid of rape-related discussions vs users involved")
    plt.xlabel("postid")
    plt.ylabel("number of users involved")
    #plt.show()
    plt.savefig("postid_versus_users_involved")