This commit is contained in:
doms9 2025-09-01 14:05:33 -04:00
parent 00000d92c5
commit 00000d9172
5 changed files with 195 additions and 65 deletions

View file

@ -3,7 +3,7 @@ import asyncio
from pathlib import Path
import httpx
from scrape import logger, tvpass # , fstv
from scrape import ace, logger, tvpass # , fstv
log = logger.get_logger(__name__)
@ -40,11 +40,13 @@ async def vanilla_fetch() -> tuple[list[str], int]:
async def main() -> None:
await tvpass.main(client)
await ace.main(client)
# await fstv.main(client)
base_m3u8, tvg_chno = await vanilla_fetch()
additions = tvpass.urls # | fstv.urls
additions = tvpass.urls | ace.urls # | fstv.urls
lines = [
f'#EXTINF:-1 tvg-chno="{chnl_num}" tvg-id="(N/A)" tvg-name="{event}" tvg-logo="{info["logo"]}" group-title="Live Events",{event}\n{info["url"]}'

128
M3U8/scrape/ace.py Normal file
View file

@ -0,0 +1,128 @@
import asyncio
import re
from urllib.parse import urljoin
import httpx
from selectolax.parser import HTMLParser, Node
from .fstv import get_base
from .logger import get_logger
from .tvpass import logos
log = get_logger(__name__)
urls: dict[str, dict[str, str]] = {}
mirrors = [
"https://aceztrims.pages.dev/",
"https://acestrlms.pages.dev/",
]
def is_valid_href(a: Node) -> bool:
href = a.attributes.get("href", "")
return href.startswith("/") and href != "/news/"
async def get_schedule(client: httpx.AsyncClient, base_url: str) -> list[dict]:
log.info(f'Scraping schedule from "{base_url}"')
try:
r = await client.get(base_url)
r.raise_for_status()
except Exception as e:
log.error(f'Failed to fetch "{base_url}": {e}')
return []
html = re.sub(r"<!--.*?-->", "", r.text, flags=re.DOTALL)
tree = HTMLParser(html)
events = []
for a in filter(is_valid_href, tree.css("a[href]")):
href = a.attributes.get("href", "")
title_text = a.text(strip=True)
after_time = (
title_text.split("//", 1)[1].strip() if "//" in title_text else title_text
)
if " - " in after_time:
sport, event_name = [x.strip() for x in after_time.split(" - ", 1)]
else:
sport, event_name = "", after_time
events.append(
{"sport": sport, "event": event_name, "href": urljoin(base_url, href)}
)
return events
async def get_m3u8_links(client: httpx.AsyncClient, url: str) -> list[str]:
try:
r = await client.get(url)
r.raise_for_status()
except Exception as e:
log.error(f'Failed to fetch "{url}": {e}')
return []
html = re.sub(r"<!--.*?-->", "", r.text, flags=re.DOTALL)
tree = HTMLParser(html)
m3u8_links = []
for btn in tree.css("button[onclick]"):
onclick = btn.attributes.get("onclick", "")
if match := re.search(r"src\s*=\s*['\"](.*?)['\"]", onclick):
link = match[1]
if ".m3u8" in link:
m3u8_links.append(link)
if iframe := tree.css_first("iframe#iframe"):
src = iframe.attributes.get("src", "")
if ".m3u8" in src and src not in m3u8_links:
m3u8_links.insert(
0,
src.split("https://cors.ricohspaces.app/")[-1],
)
return m3u8_links
async def main(client: httpx.AsyncClient) -> None:
if not (base_url := await get_base(client, mirrors)):
log.warning("No working ace mirrors")
return
schedule = await get_schedule(client, base_url)
tasks = [get_m3u8_links(client, item["href"]) for item in schedule]
results = await asyncio.gather(*tasks)
for item, m3u8_urls in zip(schedule, results):
if not m3u8_urls:
continue
for i, link in enumerate(m3u8_urls, start=1):
sport, event = item["sport"], item["event"]
urls[f"[{sport}] {event} (S{i})"] = {
"url": link,
"logo": logos.get(
sport,
"https://i.gyazo.com/ec27417a9644ae517196494afa72d2b9.png",
),
}
log.info(f"Collected {len(urls)} events")
# add caching

View file

@ -29,7 +29,7 @@ async def check_status(client: httpx.AsyncClient, url: str) -> bool:
return r.status_code == 200
async def get_base(client: httpx.AsyncClient) -> str:
async def get_base(client: httpx.AsyncClient, mirrors: list[str]) -> str:
tasks = [check_status(client, link) for link in mirrors]
results = await asyncio.gather(*tasks)
@ -103,7 +103,7 @@ async def fetch_m3u8(client: httpx.AsyncClient, url: str) -> tuple[str, list[str
async def main(client: httpx.AsyncClient) -> None:
if not (base_url := await get_base(client)):
if not (base_url := await get_base(client, mirrors)):
log.warning("No working FSTV mirrors")
return

0
M3U8/scrape/star.py Normal file
View file

View file

@ -1,62 +1,62 @@
{
"[WNBA] Atlanta Dream @ Connecticut Sun": {
"url": "http://origin.thetvapp.to/hls/nbc-sports-boston/mono.m3u8",
"logo": "https://i.gyazo.com/02d665a5704118d195dbcd5fa20d5462.png"
},
"[MLB] Toronto Blue Jays @ Cincinnati Reds": {
"url": "http://origin.thetvapp.to/hls/mlb-04/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] New York Mets @ Detroit Tigers": {
"url": "http://origin.thetvapp.to/hls/mlb-21/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Cleveland Guardians @ Boston Red Sox": {
"url": "http://origin.thetvapp.to/hls/mlb-22/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Los Angeles Angels @ Houston Astros": {
"url": "http://origin.thetvapp.to/hls/mlb-06/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Oakland Athletics @ St. Louis Cardinals": {
"url": "http://origin.thetvapp.to/hls/mlb-08/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Atlanta Braves @ Chicago Cubs": {
"url": "http://origin.thetvapp.to/hls/mlb-23/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] San Francisco Giants @ Colorado Rockies": {
"url": "http://origin.thetvapp.to/hls/mlb-25/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Philadelphia Phillies @ Milwaukee Brewers": {
"url": "http://origin.thetvapp.to/hls/mlb-16/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Baltimore Orioles @ San Diego Padres": {
"url": "http://origin.thetvapp.to/hls/mlb-02/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Seattle Mariners @ Tampa Bay Rays": {
"url": "http://origin.thetvapp.to/hls/mlb-11/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[NCAAF] TCU Horned Frogs @ North Carolina Tar Heels": {
"url": "http://origin.thetvapp.to/hls/ESPN/mono.m3u8",
"logo": "https://i.gyazo.com/ca63b40c86e757436de9d34d369b24f8.png"
},
"[WNBA] Dallas Wings @ Minnesota Lynx": {
"url": "http://origin.thetvapp.to/hls/NBATV/mono.m3u8",
"logo": "https://i.gyazo.com/02d665a5704118d195dbcd5fa20d5462.png"
},
"[MLB] Texas Rangers @ Arizona Diamondbacks": {
"url": "http://origin.thetvapp.to/hls/mlb-09/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[WNBA] Los Angeles Sparks @ Seattle Storm": {
"url": "http://origin.thetvapp.to/hls/NBATV/mono.m3u8",
"logo": "https://i.gyazo.com/02d665a5704118d195dbcd5fa20d5462.png"
}
{
"[WNBA] Atlanta Dream @ Connecticut Sun": {
"url": "http://origin.thetvapp.to/hls/nbc-sports-boston/mono.m3u8",
"logo": "https://i.gyazo.com/02d665a5704118d195dbcd5fa20d5462.png"
},
"[MLB] Toronto Blue Jays @ Cincinnati Reds": {
"url": "http://origin.thetvapp.to/hls/mlb-04/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] New York Mets @ Detroit Tigers": {
"url": "http://origin.thetvapp.to/hls/mlb-21/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Cleveland Guardians @ Boston Red Sox": {
"url": "http://origin.thetvapp.to/hls/mlb-22/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Los Angeles Angels @ Houston Astros": {
"url": "http://origin.thetvapp.to/hls/mlb-06/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Oakland Athletics @ St. Louis Cardinals": {
"url": "http://origin.thetvapp.to/hls/mlb-08/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Atlanta Braves @ Chicago Cubs": {
"url": "http://origin.thetvapp.to/hls/mlb-23/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] San Francisco Giants @ Colorado Rockies": {
"url": "http://origin.thetvapp.to/hls/mlb-25/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Philadelphia Phillies @ Milwaukee Brewers": {
"url": "http://origin.thetvapp.to/hls/mlb-16/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Baltimore Orioles @ San Diego Padres": {
"url": "http://origin.thetvapp.to/hls/mlb-02/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[MLB] Seattle Mariners @ Tampa Bay Rays": {
"url": "http://origin.thetvapp.to/hls/mlb-11/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[NCAAF] TCU Horned Frogs @ North Carolina Tar Heels": {
"url": "http://origin.thetvapp.to/hls/ESPN/mono.m3u8",
"logo": "https://i.gyazo.com/ca63b40c86e757436de9d34d369b24f8.png"
},
"[WNBA] Dallas Wings @ Minnesota Lynx": {
"url": "http://origin.thetvapp.to/hls/NBATV/mono.m3u8",
"logo": "https://i.gyazo.com/02d665a5704118d195dbcd5fa20d5462.png"
},
"[MLB] Texas Rangers @ Arizona Diamondbacks": {
"url": "http://origin.thetvapp.to/hls/mlb-09/mono.m3u8",
"logo": "https://i.gyazo.com/0fe7865ef2f06c9507791b24f04dbca8.png"
},
"[WNBA] Los Angeles Sparks @ Seattle Storm": {
"url": "http://origin.thetvapp.to/hls/NBATV/mono.m3u8",
"logo": "https://i.gyazo.com/02d665a5704118d195dbcd5fa20d5462.png"
}
}