def piratebay_page(root, page): import re from bs4 import BeautifulSoup from urlparse import urljoin from streamajoker.utils import url_get page = int(page) html_data = url_get(urljoin(BASE_URL, "%s/%d/7/100,200,500" % (root, page)), headers=HEADERS) soup = BeautifulSoup(html_data, "html5lib") nodes = soup.findAll("div", "detName") for node in nodes: seeds, peers = map(lambda x: x.text, node.parent.parent.findAll("td")[2:]) magnet_node = node.parent.findAll("a")[1] desc_node = node.parent.findAll("font", "detDesc")[0] size = re.search("Size (.*?),", desc_node.text).group(1) text = "%s (%s S:%s P:%s)" % (node.a.text, size.replace(" ", " "), seeds, peers) yield { "label": text, "path": plugin.url_for("play", uri=magnet_node["href"]), "is_playable": True, } yield { "label": ">> Next page", "path": plugin.url_for("piratebay_page", root=root, page=page + 1), "is_playable": False, }
def btdigg_page(query, sort, page): from bs4 import BeautifulSoup from streamajoker.utils import url_get html_data = url_get("%s/search" % BASE_URL, headers=HEADERS, params={ "order": sort, "q": query, "p": page, }) soup = BeautifulSoup(html_data, "html5lib") name_nodes = soup.findAll("td", "torrent_name") attr_nodes = soup.findAll("table", "torrent_name_tbl")[1::2] for name_node, attr_node in zip(name_nodes, attr_nodes): attrs = attr_node.findAll("span", "attr_val") title = "%s (%s, DLs:%s)" % (name_node.find("a").text, attrs[0].text, attrs[2].text) yield { "label": title, "path": plugin.url_for("play", uri=attr_node.find("a")["href"]), "is_playable": True, } yield { "label": ">> Next page", "path": plugin.url_for("btdigg_page", query=query, sort=sort, page=int(page) + 1), "is_playable": False, }
def search(name, complete=False): from streamajoker.caching import shelf import hashlib search_hash = hashlib.sha1(name).hexdigest() with shelf("com.thetvdb.search.%s" % search_hash) as show: if not show: import re import xml.etree.ElementTree as ET from streamajoker.utils import url_get dom = ET.fromstring( url_get( "%s/api/GetSeries.php" % BASE_URL, params={"seriesname": name}, headers=HEADERS, with_immunicity=False, ) ) if not len(dom): return meta = dom2dict(dom[0]) if not complete: return update_image_urls(meta) show.update(get(meta["id"])) return show
def get_banners(show_id): import xml.etree.ElementTree as ET from streamajoker.utils import url_get r = url_get("%s/banners.xml" % show_base_url(show_id), headers=HEADERS, with_immunicity=False) dom = ET.fromstring(r) if not len(dom): return return update_image_urls(dom2dict(dom))["banner"]
def bitsnoop_page(root, page): from urlparse import urljoin from streamajoker.scrapers import rss from streamajoker.utils import url_get content_type = plugin.request.args_dict.get("content_type") if content_type: plugin.set_content(content_type) page = int(page) page_data = url_get(urljoin(BASE_URL, "%s/%d/" % (root, page)), headers=HEADERS, params={ "fmt": "rss", "sort": "n_s", "dir": "desc", }) return rss.parse(page_data)
def get(show_id): from streamajoker.caching import shelf with shelf("com.thetvdb.show.%s" % show_id) as show: if not show: import xml.etree.ElementTree as ET from streamajoker.utils import url_get dom = ET.fromstring(url_get(show_url(show_id), headers=HEADERS, with_immunicity=False)) if not len(dom): return meta = dom2dict(dom[0]) meta = split_keys(meta, "actors", "genre", "writer") update_image_urls(meta) show.update(meta) return dict(show)
def extratorrent_page(type_="", cid="", search="", page=1): from urlparse import urljoin from streamajoker.scrapers import rss from streamajoker.utils import url_get content_type = plugin.request.args_dict.pop("content_type", None) if content_type: plugin.set_content(content_type) params = { "type": type_, "search": search, "cid": cid, } params.update(plugin.request.args_dict) page_data = url_get(urljoin(BASE_URL, "/rss.xml"), headers=HEADERS, params=params) return rss.parse(page_data, content_type)
def nyaa_page(cats, offset, sort, order, term=""): from streamajoker.scrapers import rss from streamajoker.utils import url_get offset = int(offset) rss_data = url_get(BASE_URL, headers=HEADERS, params={ "cats": cats, "offset": "%d" % offset, "sort": sort, "order": order, "term": term, "page": "rss", }) for item in rss.parse(rss_data): yield item yield { "label": ">> Next page", "path": plugin.url_for(term and "search_result_page" or "default_nyaa_page", cats=cats, sort=sort, order=order, term=term, offset=offset + 1), "is_playable": False, }
def proc_close(): if not proc.poll(): url_get("http://%s/shutdown" % proc.bind_address, with_immunicity=False)
def eztv_shows_by_letter(letter): import re import xbmc import xbmcgui from bs4 import BeautifulSoup from contextlib import nested, closing from itertools import izip, groupby from concurrent import futures from streamajoker.scrapers import ungenerate from streamajoker.utils import terminating, url_get, SafeDialogProgress from streamajoker import tvdb with shelf("it.eztv.shows") as eztv_shows: if not eztv_shows: response = url_get("%s/showlist/" % BASE_URL, headers=HEADERS) soup = BeautifulSoup(response, "html5lib") nodes = soup.findAll("a", "thread_link") for node in nodes: show_id, show_named_id = node["href"].split("/")[2:4] show_name = node.text show_first_letter = show_name[0].lower() if re.match("\d+", show_first_letter): show_first_letter = "0-9" eztv_shows.setdefault(show_first_letter, {}).update({ show_id: { "id": show_id, "named_id": show_named_id, "name": node.text, } }) shows_list = sorted(eztv_shows[letter.lower()].values(), key=lambda x: x["name"].lower()) with closing(SafeDialogProgress(delay_close=0)) as dialog: dialog.create(plugin.name) dialog.update(percent=0, line1="Fetching serie information...", line2="", line3="") state = {"done": 0} def on_serie(future): data = future.result() state["done"] += 1 dialog.update( percent=int(state["done"] * 100.0 / len(shows_list)), line2=data and data["seriesname"] or "", ) with futures.ThreadPoolExecutor(max_workers=5) as pool_tvdb: tvdb_list = [pool_tvdb.submit(tvdb.search, show["name"], True) for show in shows_list] [future.add_done_callback(on_serie) for future in tvdb_list] while not all(job.done() for job in tvdb_list): if dialog.iscanceled(): return xbmc.sleep(100) tvdb_list = [job.result() for job in tvdb_list] for i, (eztv_show, tvdb_show) in enumerate(izip(shows_list, tvdb_list)): if tvdb_show: item = tvdb.get_list_item(tvdb_show) item.update({ "path": plugin.url_for("eztv_get_show_seasons", show_id=eztv_show["id"], tvdb_id=tvdb_show["id"]) }) yield item else: yield { "label": eztv_show["name"], "path": plugin.url_for("eztv_get_show_seasons", show_id=eztv_show["id"]) }
def _eztv_get_show(): return BeautifulSoup(url_get("%s/shows/%s/" % (BASE_URL, show_id), headers=HEADERS), "html5lib")
def _eztv_get_show(): plugin.log.info("Getting show") response = url_get("%s/shows/%s/" % (BASE_URL, show_id), headers=HEADERS) plugin.log.info("Got show") return BeautifulSoup(response, "html5lib")
def _get_all_meta(): r = url_get("%s/all/%s.xml" % (show_base_url(show_id), LANG), headers=HEADERS, with_immunicity=False) dom = ET.fromstring(r) if not len(dom): return return update_image_urls(dom2dict(dom))