vwa.py

#!/usr/bin/env python3
import json, sys
from hashlib import md5
import numpy as np
import matplotlib.pyplot as plot
from collections import defaultdict
vwa = "/home/tokgeo/Downloads/vwa.json"
vwa_options = "/home/tokgeo/Downloads/vwa_options.json"


def loadDump(path : str) -> dict:
    with open(path, "r") as fobj:
        return json.loads(fobj.read().splitlines()[8])

def getQuestionForHash(hash : str) -> str:
    for i in vwa_options:
        if md5(i["question"].encode("utf-8")).hexdigest() == hash:
            return i["question"]
    raise ValueError("No corresponding question found")

vwa = loadDump(vwa)
vwa_options = loadDump(vwa_options)

for i in range(len(vwa_options)):
    vwa_options[i] = {"question" : vwa_options[i]["question"], "answers" : vwa_options[i]["allowed_answers"].split("|")}

count = defaultdict(lambda: defaultdict(lambda: defaultdict(lambda: 0)))
for i in range(len(vwa)):
    answers = []
    for block in filter(None, vwa[i]["answers"].split("|")):
        question, answer = block.split("::")
        question = getQuestionForHash(question)
        answer = answer.split(";")
        for a in answer:
            count[question][a][vwa[i]["age"]] += 1
        
        answers.append({"question" : question, "answer" : answer})
    vwa[i]["answers"] = answers
colors = ["b", "r", "g"]

for question, answers in count.items():
    fig, ax = plot.subplots()
    bar_width = 0.35
    index = np.arange(len(range(13, 18 + 1)))
    
    for answer, ages in answers.items():
        values = list(ages.values())
        print(values)
        try:
            plot.bar(index, values, bar_width, alpha=1.0, color=colors[list(answers.keys()).index(answer) % len(colors)], label=answer)
        except:
            continue
    
    plot.xlabel("Alter")
    plot.ylabel("Anzahl")
    plot.title(question)
    plot.xticks(index + bar_width / 2, list(map(str, range(13, 18 + 1))))
    plot.legend()
    #plot.tight_layout()
    plot.show()