dl/commands/pygg.py

226 lines
6.3 KiB
Python
Raw Normal View History

2022-11-13 16:33:32 +00:00
#!/usr/bin/env python3
2022-11-13 16:14:26 +00:00
import argparse
2023-01-19 21:56:22 +00:00
import io
import json
2022-12-24 10:47:15 +00:00
import os
2022-11-13 16:14:26 +00:00
import re
2022-12-24 10:47:15 +00:00
import urllib
2022-11-13 16:14:26 +00:00
2022-12-24 10:47:15 +00:00
import bs4
import charset_normalizer
2022-12-24 10:47:15 +00:00
import requests
2022-11-13 16:14:26 +00:00
2022-12-24 10:47:15 +00:00
2023-03-20 13:47:56 +00:00
YGG_DOMAIN = "www3.yggtorrent.do"
2023-01-19 21:56:22 +00:00
FLARE_ENDPOINT = os.getenv("FLARE_ENDPOINT")
2022-11-13 16:14:26 +00:00
2023-01-19 21:56:22 +00:00
class FlareRequests(requests.Session):
def request(self, method, url, params=None, data=None, **kwargs):
if not FLARE_ENDPOINT:
return super().request(method, url, params, data, **kwargs)
sessions = requests.post(FLARE_ENDPOINT, json={"cmd": "sessions.list"}).json()
if "sessions" in sessions and len(sessions["sessions"]) > 0:
FLARE_SESSION = sessions["sessions"][0]
else:
response = requests.post(FLARE_ENDPOINT, json={"cmd": "sessions.create"})
session = response.json()
if "session" in session:
FLARE_SESSION = session["session"]
else:
raise requests.RequestException(response)
if params:
url += "&" if len(url.split("?")) > 1 else "?"
url = f"{url}{urllib.parse.urlencode(params)}"
post_data = {
"cmd": f"request.{method.lower()}",
"session": FLARE_SESSION,
"url": url,
}
if data:
post_data["postData"] = urllib.parse.urlencode(data)
try:
response = requests.post(
FLARE_ENDPOINT,
json=post_data,
)
content = response.json()
2023-01-24 16:21:33 +00:00
if "solution" in content and content["solution"]:
2023-01-19 21:56:22 +00:00
solution = content["solution"]
raw = solution["response"].encode()
encoding = charset_normalizer.detect(raw)
2023-01-19 21:56:22 +00:00
resolved = requests.Response()
resolved.status_code = solution["status"]
resolved.headers = solution["headers"]
resolved.raw = io.BytesIO(raw)
2023-01-19 21:56:22 +00:00
resolved.url = url
resolved.encoding = encoding["encoding"]
2023-01-19 21:56:22 +00:00
resolved.reason = content["status"]
resolved.cookies = solution["cookies"]
return resolved
2023-01-24 16:21:33 +00:00
raise requests.RequestException(content["message"], response=response)
2023-01-19 21:56:22 +00:00
except requests.RequestException:
session = requests.post(
FLARE_ENDPOINT,
json={"cmd": "sessions.destroy", "session": FLARE_SESSION},
)
2023-01-24 16:21:33 +00:00
raise requests.RequestException(content["message"], response=response)
2023-01-19 21:56:22 +00:00
session = FlareRequests()
2022-11-13 16:14:26 +00:00
parser = argparse.ArgumentParser()
parser.add_argument("-u", "--uploader", action="append")
2022-12-24 10:47:15 +00:00
parser.add_argument("-b", "--blacklist", action="append", default=["dvd", "iso"])
2022-11-13 16:14:26 +00:00
parser.add_argument("-y", "--year", type=int)
2022-11-13 17:24:28 +00:00
parser.add_argument("-s", "--size", type=int, default=10)
2022-11-17 21:12:00 +00:00
parser.add_argument("-d", "--downloads", type=int, default=20)
2022-11-13 16:14:26 +00:00
parser.add_argument("query")
args = parser.parse_args()
def parse_size(size):
units = {"o": 1, "Ko": 10**3, "Mo": 10**6, "Go": 10**9, "To": 10**12}
match = re.search("([0-9.]+)([^0-9]+)", size)
number = match.group(1).strip()
unit = match.group(2).strip()
return int(float(number) * units[unit])
2022-11-17 00:29:15 +00:00
def check_files(id):
2023-01-20 15:07:32 +00:00
req = session.get(f"https://{YGG_DOMAIN}/engine/get_files", params={"torrent": id})
2023-01-20 15:05:04 +00:00
res = bs4.BeautifulSoup(req.text, "html.parser")
pre = res.select_one("pre")
jhtml = json.loads(pre.get_text())
html = bs4.BeautifulSoup(jhtml["html"], "html.parser")
2022-11-13 16:14:26 +00:00
trs = html.select("tr")
2022-11-17 00:29:15 +00:00
return len(trs) == 1 and "mkv" in trs[0].get_text().lower()
2022-11-13 16:14:26 +00:00
2022-11-14 17:42:52 +00:00
def search_ygg(query, multi, full):
2022-11-13 16:14:26 +00:00
ygg_params = {
2022-11-14 13:51:25 +00:00
"name": query,
2022-11-13 16:14:26 +00:00
"category": "2145",
"sub_category": "2183",
"do": "search",
"order": "asc",
"sort": "publish_date",
}
2022-11-14 17:42:52 +00:00
if full and args.year:
ygg_params["name"] += f" {args.year}"
2022-11-13 16:14:26 +00:00
if multi:
2022-11-13 19:19:44 +00:00
ygg_params["option_langue:multiple[]"] = "4"
2022-11-13 16:14:26 +00:00
2023-01-26 22:27:58 +00:00
try:
req = session.get(f"https://{YGG_DOMAIN}/engine/search", params=ygg_params)
except requests.RequestException as e:
print(
f"Request failed. {e}\n"
f"https://{YGG_DOMAIN}/engine/search?{urllib.parse.urlencode(ygg_params)}"
)
return
2022-12-24 13:49:15 +00:00
html = bs4.BeautifulSoup(
req.text,
"html.parser",
)
2022-11-13 16:14:26 +00:00
trs = html.select("table.table tr")
if len(trs) > 1:
for i, tr in enumerate(trs):
if not i:
continue
tds = tr.find_all("td")
size = tds[5].get_text()
2022-11-17 00:12:13 +00:00
downloads = tds[6].get_text()
2022-11-13 16:14:26 +00:00
name = tds[1].get_text().lower().strip()
2022-11-13 17:24:28 +00:00
if parse_size(size) > parse_size(f"{args.size}Go"):
2022-11-13 16:14:26 +00:00
continue
2022-11-17 21:12:00 +00:00
if int(downloads) < args.downloads:
2022-11-17 00:12:13 +00:00
continue
2023-03-22 23:24:28 +00:00
if any(word and word.lower() in name for word in args.blacklist):
2022-11-13 16:14:26 +00:00
continue
2022-11-14 17:11:22 +00:00
if args.year and str(args.year) not in name:
2022-11-14 13:51:25 +00:00
continue
2022-11-13 16:14:26 +00:00
if args.uploader and not any(
2023-03-22 23:24:28 +00:00
uploader and uploader.lower() in name for uploader in args.uploader
2022-11-13 16:14:26 +00:00
):
continue
link = tds[1].a["href"]
id = link.split("/")[-1].split("-")[0]
2022-11-17 00:29:15 +00:00
if not check_files(id):
2022-11-13 16:14:26 +00:00
continue
2022-11-14 13:51:25 +00:00
print(f"{name} {link}")
2022-11-13 16:14:26 +00:00
exit(0)
query_string = {"query": args.query, "filters": "type:movie"}
if args.year:
2022-11-13 17:24:28 +00:00
query_string["filters"] += f" AND year:{args.year}"
2022-11-13 16:14:26 +00:00
2022-12-24 10:47:15 +00:00
tvdb = requests.post(
2022-11-13 16:14:26 +00:00
"https://tvshowtime-dsn.algolia.net/1/indexes/TVDB/query",
params={
"x-algolia-application-id": "tvshowtime",
"x-algolia-api-key": "c9d5ec1316cec12f093754c69dd879d3",
},
2022-12-24 10:47:15 +00:00
json={"params": urllib.parse.urlencode(query_string)},
2022-11-13 16:14:26 +00:00
)
2022-11-13 17:24:28 +00:00
tvdata = tvdb.json()
2022-11-13 16:14:26 +00:00
if not tvdata["nbHits"] > 0:
print("Can't find query on TheTVDB")
exit(1)
eng = tvdata["hits"][0]["name"]
fra = (
tvdata["hits"][0]["translations"]["fra"]
if "fra" in tvdata["hits"][0]["translations"]
else args.query
)
2022-11-14 17:42:52 +00:00
search_ygg(args.query, True, True)
search_ygg(fra, True, True)
search_ygg(eng, True, True)
search_ygg(args.query, False, True)
search_ygg(fra, False, True)
search_ygg(eng, False, True)
if args.year:
search_ygg(args.query, True, False)
search_ygg(fra, True, False)
search_ygg(eng, True, False)
search_ygg(args.query, False, False)
search_ygg(fra, False, False)
search_ygg(eng, False, False)
2022-11-13 17:24:28 +00:00
print("No results :(")