Lista de animes de www.anbient.net organizados por score

anbient_score.py

#!/usr/bin/env bash
#
#   Bash Script
#
#   Copyright © Manoel Vilela
#
#

from requests import get
from bs4 import BeautifulSoup


def getNote(url):
    raw = get(url).text
    soup = BeautifulSoup(raw, "html.parser")

    note = 0
    content = []
    for group in soup.find_all(['span', "class=bignote"]):
        content.extend([x for x in group.strings])

    for index, x in enumerate(content):
        if 'Minha nota' in x:
            real_note = content[index + 3]
            note = float(real_note.strip('\t\n'))
            break

    return note


def animeUrls():
    main_url = 'http://www.anbient.net'
    list_url = main_url + '/lista'
    backbone = get(list_url)
    soup = BeautifulSoup(backbone.text, "html.parser")
    anime_urls = {}
    passing = True
    for link in soup.find_all(['tr', 'href', 'a'], href=True):
        name = [x.strip() for x in link.strings if x.strip() not in ('', '\n')]
        url = link['href']
        if passing or "Riccardo Benetti" in name:
            if '.hack//DUSK' in name:
                passing = False
            else:
                continue
        anime_urls[name.pop()] = main_url + url

    return anime_urls


def main():
    print(":: Fetching main list, getting url animes")
    anime_urls = animeUrls()

    # max anime name
    n = len(max(anime_urls, key=lambda x: len(x)))
    anime_format = '{{: <{width}}}'.format(width=n)

    print(":: Get score of each anime")
    anime_score_url = []
    print("{} Score\tUrl".format(anime_format.format('Anime')))
    for anime, url in sorted(anime_urls.items()):
        score = getNote(url)
        anime = anime_format.format(anime)
        print("{} {: >2}\t{}".format(anime, score, url))
        anime_score_url.append((anime, score, url))

    print(":: Sorting...")
    sorted_anime = sorted(anime_score_url, key=lambda x: x[1])
    print("{} Score\tUrl".format(anime_format.format('Anime')))
    for anime, score, url in sorted_anime:
        anime = anime_format.format(anime)
        print("{} {: <2}\t{}".format(anime, score, url))

    print(":: Saving list")
    with open('anbient_list.txt', 'w') as f:
        content = ["{} Score\tUrl".format(anime_format.format('Anime'))]
        for anime, score, url in sorted_anime:
            anime = anime_format.format(anime)
            content.append("{} {}\t{}".format(anime, score, url))
        f.write('\n'.join(content))

if __name__ == '__main__':
    main()