## Kickass Torrent (Videos, Music, Files) # # @website https://kickass.so # @provide-api no (nothing found) # # @using-api no # @results HTML (using search portal) # @stable yes (HTML can change) # @parse url, title, content, seed, leech, magnetlink from urlparse import urljoin from cgi import escape from urllib import quote from lxml import html from operator import itemgetter # engine dependent config categories = ['videos', 'music', 'files'] paging = True # search-url url = 'https://kickass.so/' search_url = url + 'search/{search_term}/{pageno}/' # specific xpath variables magnet_xpath = './/a[@title="Torrent magnet link"]' torrent_xpath = './/a[@title="Download torrent file"]' content_xpath = './/span[@class="font11px lightgrey block"]' # do search-request def request(query, params): params['url'] = search_url.format(search_term=quote(query), pageno=params['pageno']) # FIX: SSLError: hostname 'kickass.so' # doesn't match either of '*.kickass.to', 'kickass.to' params['verify'] = False return params # get response from search-request def response(resp): results = [] dom = html.fromstring(resp.text) search_res = dom.xpath('//table[@class="data"]//tr') # return empty array if nothing is found if not search_res: return [] # parse results for result in search_res[1:]: link = result.xpath('.//a[@class="cellMainLink"]')[0] href = urljoin(url, link.attrib['href']) title = ' '.join(link.xpath('.//text()')) content = escape(html.tostring(result.xpath(content_xpath)[0], method="text")) seed = result.xpath('.//td[contains(@class, "green")]/text()')[0] leech = result.xpath('.//td[contains(@class, "red")]/text()')[0] filesize = result.xpath('.//td[contains(@class, "nobr")]/text()')[0] filesize_multiplier = result.xpath('.//td[contains(@class, "nobr")]//span/text()')[0] files = result.xpath('.//td[contains(@class, "center")][2]/text()')[0] # convert seed to int if possible if seed.isdigit(): seed = int(seed) else: seed = 0 # convert leech to int if possible if leech.isdigit(): leech = int(leech) else: leech = 0 # convert filesize to byte if possible try: filesize = float(filesize) # convert filesize to byte if filesize_multiplier == 'TB': filesize = int(filesize * 1024 * 1024 * 1024 * 1024) elif filesize_multiplier == 'GB': filesize = int(filesize * 1024 * 1024 * 1024) elif filesize_multiplier == 'MB': filesize = int(filesize * 1024 * 1024) elif filesize_multiplier == 'kb': filesize = int(filesize * 1024) except: filesize = None # convert files to int if possible if files.isdigit(): files = int(files) else: files = None magnetlink = result.xpath(magnet_xpath)[0].attrib['href'] torrentfile = result.xpath(torrent_xpath)[0].attrib['href'] # append result results.append({'url': href, 'title': title, 'content': content, 'seed': seed, 'leech': leech, 'filesize': filesize, 'files': files, 'magnetlink': magnetlink, 'torrentfile': torrentfile, 'template': 'torrent.html'}) # return results sorted by seeder return sorted(results, key=itemgetter('seed'), reverse=True)