update M3U8
This commit is contained in:
GitHub Actions Bot
parent
a162ee1cf0
commit
6cc71c4bd1
5 changed files with 386 additions and 1672 deletions
|
|
@ -1,161 +0,0 @@
|
|||
from functools import partial
|
||||
|
||||
import httpx
|
||||
from playwright.async_api import async_playwright
|
||||
|
||||
from .utils import Cache, Time, get_logger, leagues, network
|
||||
|
||||
log = get_logger(__name__)
|
||||
|
||||
urls: dict[str, dict[str, str | float]] = {}
|
||||
|
||||
CACHE_FILE = Cache("lotus.json", exp=3_600)
|
||||
|
||||
API_CACHE = Cache("lotus-api.json", exp=28_800)
|
||||
|
||||
BASE_URL = "https://lotusgamehd.xyz/api-event.php"
|
||||
|
||||
|
||||
def fix_league(s: str) -> str:
|
||||
return " ".join(x.capitalize() for x in s.split()) if len(s) > 5 else s.upper()
|
||||
|
||||
|
||||
async def refresh_api_cache(
|
||||
client: httpx.AsyncClient,
|
||||
url: str,
|
||||
ts: float,
|
||||
) -> dict[str, dict[str, str]]:
|
||||
log.info("Refreshing API cache")
|
||||
|
||||
try:
|
||||
r = await client.get(url)
|
||||
r.raise_for_status()
|
||||
except Exception as e:
|
||||
log.error(f'Failed to fetch "{url}": {e}')
|
||||
|
||||
return {}
|
||||
|
||||
data = r.json()
|
||||
|
||||
data["timestamp"] = ts
|
||||
|
||||
return data
|
||||
|
||||
|
||||
async def get_events(
|
||||
client: httpx.AsyncClient,
|
||||
event_link: str,
|
||||
cached_keys: set[str],
|
||||
) -> list[dict[str, str]]:
|
||||
now = Time.clean(Time.now())
|
||||
|
||||
if not (api_data := API_CACHE.load(per_entry=False)):
|
||||
api_data = await refresh_api_cache(
|
||||
client,
|
||||
event_link,
|
||||
now.timestamp(),
|
||||
)
|
||||
|
||||
API_CACHE.write(api_data)
|
||||
|
||||
events = []
|
||||
|
||||
for info in api_data.get("days", []):
|
||||
day = Time.from_str(info["day_et"])
|
||||
|
||||
if now.date() != day.date():
|
||||
continue
|
||||
|
||||
for event in info["items"]:
|
||||
event_league = event["league"]
|
||||
|
||||
if event_league == "channel tv":
|
||||
continue
|
||||
|
||||
event_streams: list[dict] = event["streams"]
|
||||
|
||||
if not (event_link := event_streams[0].get("link")):
|
||||
continue
|
||||
|
||||
sport = fix_league(event_league)
|
||||
event_name = event["title"]
|
||||
|
||||
key = f"[{sport}] {event_name} (LOTUS)"
|
||||
|
||||
if cached_keys & {key}:
|
||||
continue
|
||||
|
||||
events.append(
|
||||
{
|
||||
"sport": sport,
|
||||
"event": event_name,
|
||||
"link": event_link,
|
||||
}
|
||||
)
|
||||
|
||||
return events
|
||||
|
||||
|
||||
async def scrape(client: httpx.AsyncClient) -> None:
|
||||
cached_urls = CACHE_FILE.load()
|
||||
cached_count = len(cached_urls)
|
||||
urls.update(cached_urls)
|
||||
|
||||
log.info(f"Loaded {cached_count} event(s) from cache")
|
||||
|
||||
log.info(f'Scraping from "{BASE_URL}"')
|
||||
|
||||
events = await get_events(
|
||||
client,
|
||||
BASE_URL,
|
||||
set(cached_urls.keys()),
|
||||
)
|
||||
|
||||
log.info(f"Processing {len(events)} new URL(s)")
|
||||
|
||||
if events:
|
||||
now = Time.now().timestamp()
|
||||
|
||||
async with async_playwright() as p:
|
||||
browser, context = await network.browser(p, browser="brave")
|
||||
|
||||
for i, ev in enumerate(events, start=1):
|
||||
handler = partial(
|
||||
network.process_event,
|
||||
url=ev["link"],
|
||||
url_num=i,
|
||||
context=context,
|
||||
log=log,
|
||||
)
|
||||
|
||||
url = await network.safe_process(
|
||||
handler,
|
||||
url_num=i,
|
||||
log=log,
|
||||
)
|
||||
|
||||
if url:
|
||||
sport, event = ev["sport"], ev["event"]
|
||||
|
||||
tvg_id, logo = leagues.get_tvg_info(sport, event)
|
||||
|
||||
key = f"[{sport}] {event} (LOTUS)"
|
||||
|
||||
entry = {
|
||||
"url": url,
|
||||
"logo": logo,
|
||||
"base": "https://vividmosaica.com/",
|
||||
"timestamp": now,
|
||||
"id": tvg_id or "Live.Event.us",
|
||||
}
|
||||
|
||||
urls[key] = cached_urls[key] = entry
|
||||
|
||||
await browser.close()
|
||||
|
||||
if new_count := len(cached_urls) - cached_count:
|
||||
log.info(f"Collected and cached {new_count} new event(s)")
|
||||
else:
|
||||
log.info("No new events found")
|
||||
|
||||
CACHE_FILE.write(cached_urls)
|
||||
|
|
@ -1,270 +0,0 @@
|
|||
import asyncio
|
||||
from functools import partial
|
||||
from urllib.parse import urljoin
|
||||
|
||||
import httpx
|
||||
from playwright.async_api import BrowserContext, async_playwright
|
||||
from selectolax.parser import HTMLParser
|
||||
|
||||
from .utils import Cache, Time, get_logger, leagues, network
|
||||
|
||||
log = get_logger(__name__)
|
||||
|
||||
urls: dict[str, dict[str, str | float]] = {}
|
||||
|
||||
CACHE_FILE = Cache("volo.json", exp=10_800)
|
||||
|
||||
HTML_CACHE = Cache("volo-html.json", exp=86_400)
|
||||
|
||||
BASE_URL = "http://volokit2.fun"
|
||||
|
||||
valid_sports = {
|
||||
# "boxing": "Boxing",
|
||||
# "college-football": "CFB",
|
||||
# "mens-college-basketball": "CBB",
|
||||
"mlb": "MLB",
|
||||
"mls": "Soccer",
|
||||
"nba": "NBA",
|
||||
"nfl": "NFL",
|
||||
"nhl": "NHL",
|
||||
# "race": "Racing",
|
||||
# "ufc": "UFC",
|
||||
"wnba": "WNBA",
|
||||
}
|
||||
|
||||
|
||||
def fix_event(s: str) -> str:
|
||||
return " ".join(x.capitalize() for x in s.split())
|
||||
|
||||
|
||||
async def process_event(
|
||||
url: str,
|
||||
url_num: int,
|
||||
context: BrowserContext,
|
||||
timeout: int | float = 10,
|
||||
) -> str | None:
|
||||
|
||||
page = await context.new_page()
|
||||
|
||||
captured: list[str] = []
|
||||
|
||||
got_one = asyncio.Event()
|
||||
|
||||
handler = partial(
|
||||
network.capture_req,
|
||||
captured=captured,
|
||||
got_one=got_one,
|
||||
)
|
||||
|
||||
page.on("request", handler)
|
||||
|
||||
try:
|
||||
await page.goto(
|
||||
url,
|
||||
wait_until="domcontentloaded",
|
||||
timeout=10_000,
|
||||
)
|
||||
|
||||
wait_task = asyncio.create_task(got_one.wait())
|
||||
|
||||
try:
|
||||
|
||||
iframe = page.locator("iframe").first
|
||||
|
||||
src = await iframe.get_attribute("src")
|
||||
|
||||
await page.goto(
|
||||
src,
|
||||
wait_until="domcontentloaded",
|
||||
timeout=10_000,
|
||||
)
|
||||
|
||||
await page.click("#volokit_player")
|
||||
|
||||
await asyncio.wait_for(wait_task, timeout=timeout)
|
||||
except asyncio.TimeoutError:
|
||||
log.warning(f"URL {url_num}) Timed out waiting for M3U8.")
|
||||
return
|
||||
|
||||
finally:
|
||||
if not wait_task.done():
|
||||
wait_task.cancel()
|
||||
|
||||
try:
|
||||
await wait_task
|
||||
except asyncio.CancelledError:
|
||||
pass
|
||||
|
||||
if captured:
|
||||
log.info(f"URL {url_num}) Captured M3U8")
|
||||
return captured[0]
|
||||
|
||||
log.warning(f"URL {url_num}) No M3U8 captured after waiting.")
|
||||
return
|
||||
|
||||
except Exception as e:
|
||||
log.warning(f"URL {url_num}) Exception while processing: {e}")
|
||||
return
|
||||
|
||||
finally:
|
||||
page.remove_listener("request", handler)
|
||||
await page.close()
|
||||
|
||||
|
||||
async def refresh_html_cache(
|
||||
client: httpx.AsyncClient,
|
||||
url: str,
|
||||
sport: str,
|
||||
) -> dict[str, dict[str, str | float]]:
|
||||
|
||||
try:
|
||||
r = await client.get(url)
|
||||
r.raise_for_status()
|
||||
except Exception as e:
|
||||
log.error(f'Failed to fetch "{url}": {e}')
|
||||
|
||||
return {}
|
||||
|
||||
soup = HTMLParser(r.text)
|
||||
now = Time.clean(Time.now())
|
||||
|
||||
events = {}
|
||||
|
||||
for card in soup.css("#events .table .vevent.theevent"):
|
||||
if not (href := card.css_first("a").attributes.get("href")):
|
||||
continue
|
||||
|
||||
name_node = card.css_first(".teamtd.event")
|
||||
time_node = card.css_first(".time")
|
||||
|
||||
if not (name_node and time_node):
|
||||
continue
|
||||
|
||||
name = name_node.text(strip=True)
|
||||
time = time_node.text(strip=True)
|
||||
|
||||
event_sport = valid_sports[sport]
|
||||
event_name = fix_event(name)
|
||||
|
||||
event_dt = Time.from_only_time(
|
||||
time,
|
||||
now.date(),
|
||||
timezone="UTC",
|
||||
)
|
||||
|
||||
key = f"[{event_sport}] {event_name} (VOLO)"
|
||||
|
||||
events[key] = {
|
||||
"sport": event_sport,
|
||||
"event": event_name,
|
||||
"link": href,
|
||||
"event_ts": event_dt.timestamp(),
|
||||
"timestamp": now.timestamp(),
|
||||
}
|
||||
|
||||
return events
|
||||
|
||||
|
||||
async def get_events(
|
||||
client: httpx.AsyncClient,
|
||||
sport_urls: dict[str, str],
|
||||
cached_keys: set[str],
|
||||
) -> list[dict[str, str]]:
|
||||
|
||||
now = Time.clean(Time.now())
|
||||
|
||||
if not (events := HTML_CACHE.load()):
|
||||
tasks = [
|
||||
refresh_html_cache(
|
||||
client,
|
||||
url,
|
||||
sport,
|
||||
)
|
||||
for sport, url in sport_urls.items()
|
||||
]
|
||||
|
||||
results = await asyncio.gather(*tasks)
|
||||
|
||||
events = {k: v for data in results for k, v in data.items()}
|
||||
|
||||
HTML_CACHE.write(events)
|
||||
|
||||
live = []
|
||||
|
||||
start_ts = now.delta(minutes=-30).timestamp()
|
||||
end_ts = now.delta(minutes=30).timestamp()
|
||||
|
||||
for k, v in events.items():
|
||||
if cached_keys & {k}:
|
||||
continue
|
||||
|
||||
if not start_ts <= v["event_ts"] <= end_ts:
|
||||
continue
|
||||
|
||||
live.append({**v})
|
||||
|
||||
return live
|
||||
|
||||
|
||||
async def scrape(client: httpx.AsyncClient) -> None:
|
||||
cached_urls = CACHE_FILE.load()
|
||||
cached_count = len(cached_urls)
|
||||
urls.update(cached_urls)
|
||||
|
||||
log.info(f"Loaded {cached_count} event(s) from cache")
|
||||
|
||||
log.info(f'Scraping from "{BASE_URL}"')
|
||||
|
||||
sport_urls = {sport: urljoin(BASE_URL, f"sport/{sport}") for sport in valid_sports}
|
||||
|
||||
events = await get_events(
|
||||
client,
|
||||
sport_urls,
|
||||
set(cached_urls.keys()),
|
||||
)
|
||||
|
||||
log.info(f"Processing {len(events)} new URL(s)")
|
||||
|
||||
if events:
|
||||
async with async_playwright() as p:
|
||||
browser, context = await network.browser(p, browser="brave")
|
||||
|
||||
for i, ev in enumerate(events, start=1):
|
||||
handler = partial(
|
||||
process_event,
|
||||
url=ev["link"],
|
||||
url_num=i,
|
||||
context=context,
|
||||
)
|
||||
|
||||
url = await network.safe_process(
|
||||
handler,
|
||||
url_num=i,
|
||||
log=log,
|
||||
)
|
||||
|
||||
if url:
|
||||
sport, event, ts = ev["sport"], ev["event"], ev["event_ts"]
|
||||
|
||||
tvg_id, logo = leagues.get_tvg_info(sport, event)
|
||||
|
||||
key = f"[{sport}] {event} (VOLO)"
|
||||
|
||||
entry = {
|
||||
"url": url,
|
||||
"logo": logo,
|
||||
"base": BASE_URL,
|
||||
"timestamp": ts,
|
||||
"id": tvg_id or "Live.Event.us",
|
||||
}
|
||||
|
||||
urls[key] = cached_urls[key] = entry
|
||||
|
||||
await browser.close()
|
||||
|
||||
if new_count := len(cached_urls) - cached_count:
|
||||
log.info(f"Collected and cached {new_count} new event(s)")
|
||||
else:
|
||||
log.info("No new events found")
|
||||
|
||||
CACHE_FILE.write(cached_urls)
|
||||
Loading…
Add table
Add a link
Reference in a new issue