ctoonparty.py

#!/usr/bin/env python3
"""

Get Rick & Morty episodes ("rick" as in "https://ctoon.party/rick"):
> scrapy runspider ctoonparty.py -o rick.json -a show=rick
(add -a season=2 to get only one season)
Extract links only:
> jq '.[].url' -r < rick.json > rick.txt
> wget -c -i rick.txt

Extract links with subtitles:
> jq -r '.[]|.url,.tracks.en//empty' mlp.json > mlp.txt

"""
import scrapy
import re

def get_best_source(sources):
    ordered = ['1080p', '720p']
    for o in ordered:
        for s in sources:
            label = s.css('::attr(label)').extract_first()
            href = s.css('::attr(src)').extract_first()
            if label == o:
                return (label, href)

def get_tracks(tracks):
    for track in tracks:
        code = track.css('::attr(srclang)').extract_first()
        href = track.css('::attr(src)').extract_first()

        yield (code, href)


class CToonSpider(scrapy.Spider):
    name = 'ctoon'

    def __init__(self, *args, show=None, season=None, **kwargs):
        if not show or '/' in show:
            raise Exception("use -a show=<name>")
        self.show = show
        self.season = int(season) if season else None
        self.base = 'https://ctoon.party/%s' % self.show
        self.start_urls = [self.base]

    def parse(self, response):
        if re.match(self.base + '/[0-9a-z]+/?', response.url):
            # Episode page
            best = get_best_source(response.css('video source'))
            if not best:
                raise Exception()
            tracks = dict(get_tracks(response.css('video track')))
            yield {'url': best[1], 'quality': best[0], 'tracks': tracks}
        else:
            # Index
            if self.season:
                f = '#collapse%02d' % self.season
            else:
                f = ''
            for next_page in response.css(f + ' .ep-entry a'):
                yield response.follow(next_page, self.parse)