e
This commit is contained in:
doms9
parent
00000d9766
commit
00000d950a
1 changed files with 279 additions and 279 deletions
|
|
@ -1,279 +1,279 @@
|
||||||
import asyncio
|
import asyncio
|
||||||
import re
|
import re
|
||||||
from functools import partial
|
from functools import partial
|
||||||
from pathlib import Path
|
from pathlib import Path
|
||||||
from typing import Any
|
from typing import Any
|
||||||
from urllib.parse import urljoin
|
from urllib.parse import urljoin
|
||||||
|
|
||||||
import httpx
|
import httpx
|
||||||
from playwright.async_api import BrowserContext, async_playwright
|
from playwright.async_api import BrowserContext, async_playwright
|
||||||
|
|
||||||
from .utils import Cache, Time, get_logger, leagues, network
|
from .utils import Cache, Time, get_logger, leagues, network
|
||||||
|
|
||||||
log = get_logger(__name__)
|
log = get_logger(__name__)
|
||||||
|
|
||||||
urls: dict[str, dict[str, str | float]] = {}
|
urls: dict[str, dict[str, str | float]] = {}
|
||||||
|
|
||||||
API_FILE = Cache(Path(__file__).parent / "caches" / "strmd_api.json", exp=28_800)
|
API_FILE = Cache(Path(__file__).parent / "caches" / "strmd_api.json", exp=28_800)
|
||||||
|
|
||||||
CACHE_FILE = Cache(Path(__file__).parent / "caches" / "strmd.json", exp=10_800)
|
CACHE_FILE = Cache(Path(__file__).parent / "caches" / "strmd.json", exp=10_800)
|
||||||
|
|
||||||
MIRRORS = ["https://streamed.pk", "https://streami.su", "https://streamed.st"]
|
MIRRORS = ["https://streamed.pk", "https://streami.su", "https://streamed.st"]
|
||||||
|
|
||||||
|
|
||||||
def validate_category(s: str) -> str:
|
def validate_category(s: str) -> str:
|
||||||
if "-" in s:
|
if "-" in s:
|
||||||
return " ".join([i.capitalize() for i in s.split("-")])
|
return " ".join(i.capitalize() for i in s.split("-"))
|
||||||
|
|
||||||
elif s == "fight":
|
elif s == "fight":
|
||||||
return "Fight (UFC/Boxing)"
|
return "Fight (UFC/Boxing)"
|
||||||
|
|
||||||
return s.capitalize()
|
return s.capitalize()
|
||||||
|
|
||||||
|
|
||||||
def get_tvg_id(sport: str, event: str) -> tuple[str | None, str]:
|
def get_tvg_info(sport: str, event: str) -> tuple[str | None, str]:
|
||||||
match sport:
|
match sport:
|
||||||
case "American Football":
|
case "American Football":
|
||||||
if leagues.is_valid(event, "NFL"):
|
if leagues.is_valid(event, "NFL"):
|
||||||
return leagues.info("NFL")
|
return leagues.info("NFL")
|
||||||
|
|
||||||
else:
|
else:
|
||||||
return leagues.info("NCAA")
|
return leagues.info("NCAA")
|
||||||
|
|
||||||
case "Basketball":
|
case "Basketball":
|
||||||
if leagues.is_valid(event, "NBA"):
|
if leagues.is_valid(event, "NBA"):
|
||||||
return leagues.info("NBA")
|
return leagues.info("NBA")
|
||||||
|
|
||||||
elif leagues.is_valid(event, "WNBA"):
|
elif leagues.is_valid(event, "WNBA"):
|
||||||
return leagues.info("WNBA")
|
return leagues.info("WNBA")
|
||||||
|
|
||||||
# NCAA
|
# NCAA
|
||||||
|
|
||||||
else:
|
else:
|
||||||
return leagues.info("Basketball")
|
return leagues.info("Basketball")
|
||||||
|
|
||||||
case "Hockey":
|
case "Hockey":
|
||||||
return leagues.info("NHL")
|
return leagues.info("NHL")
|
||||||
|
|
||||||
case _:
|
case _:
|
||||||
return leagues.info(sport)
|
return leagues.info(sport)
|
||||||
|
|
||||||
|
|
||||||
async def refresh_api_cache(
|
async def refresh_api_cache(
|
||||||
client: httpx.AsyncClient, url: str
|
client: httpx.AsyncClient, url: str
|
||||||
) -> list[dict[str, Any]]:
|
) -> list[dict[str, Any]]:
|
||||||
log.info("Refreshing API cache")
|
log.info("Refreshing API cache")
|
||||||
|
|
||||||
try:
|
try:
|
||||||
r = await client.get(url)
|
r = await client.get(url)
|
||||||
r.raise_for_status()
|
r.raise_for_status()
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
log.error(f'Failed to fetch "{url}"\n{e}')
|
log.error(f'Failed to fetch "{url}"\n{e}')
|
||||||
return {}
|
return {}
|
||||||
|
|
||||||
data = r.json()
|
data = r.json()
|
||||||
|
|
||||||
data[0]["timestamp"] = Time.now().timestamp()
|
data[0]["timestamp"] = Time.now().timestamp()
|
||||||
|
|
||||||
return data
|
return data
|
||||||
|
|
||||||
|
|
||||||
async def process_event(
|
async def process_event(
|
||||||
url: str,
|
url: str,
|
||||||
url_num: int,
|
url_num: int,
|
||||||
context: BrowserContext,
|
context: BrowserContext,
|
||||||
) -> str | None:
|
) -> str | None:
|
||||||
page = await context.new_page()
|
page = await context.new_page()
|
||||||
|
|
||||||
captured: list[str] = []
|
captured: list[str] = []
|
||||||
|
|
||||||
got_one = asyncio.Event()
|
got_one = asyncio.Event()
|
||||||
|
|
||||||
handler = partial(network.capture_req, captured=captured, got_one=got_one)
|
handler = partial(network.capture_req, captured=captured, got_one=got_one)
|
||||||
|
|
||||||
page.on("request", handler)
|
page.on("request", handler)
|
||||||
|
|
||||||
try:
|
try:
|
||||||
await page.goto(
|
await page.goto(
|
||||||
url,
|
url,
|
||||||
wait_until="domcontentloaded",
|
wait_until="domcontentloaded",
|
||||||
timeout=15_000,
|
timeout=15_000,
|
||||||
)
|
)
|
||||||
|
|
||||||
wait_task = asyncio.create_task(got_one.wait())
|
wait_task = asyncio.create_task(got_one.wait())
|
||||||
|
|
||||||
try:
|
try:
|
||||||
await asyncio.wait_for(wait_task, timeout=10)
|
await asyncio.wait_for(wait_task, timeout=10)
|
||||||
except asyncio.TimeoutError:
|
except asyncio.TimeoutError:
|
||||||
log.warning(f"URL {url_num}) Timed out waiting for M3U8.")
|
log.warning(f"URL {url_num}) Timed out waiting for M3U8.")
|
||||||
return
|
return
|
||||||
|
|
||||||
finally:
|
finally:
|
||||||
if not wait_task.done():
|
if not wait_task.done():
|
||||||
wait_task.cancel()
|
wait_task.cancel()
|
||||||
|
|
||||||
try:
|
try:
|
||||||
await wait_task
|
await wait_task
|
||||||
except asyncio.CancelledError:
|
except asyncio.CancelledError:
|
||||||
pass
|
pass
|
||||||
|
|
||||||
if captured:
|
if captured:
|
||||||
log.info(f"URL {url_num}) Captured M3U8")
|
log.info(f"URL {url_num}) Captured M3U8")
|
||||||
|
|
||||||
return captured[-1]
|
return captured[-1]
|
||||||
|
|
||||||
log.warning(f"URL {url_num}) No M3U8 captured after waiting.")
|
log.warning(f"URL {url_num}) No M3U8 captured after waiting.")
|
||||||
return
|
return
|
||||||
|
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
log.warning(f"URL {url_num}) Exception while processing: {e}")
|
log.warning(f"URL {url_num}) Exception while processing: {e}")
|
||||||
return
|
return
|
||||||
|
|
||||||
finally:
|
finally:
|
||||||
page.remove_listener("request", handler)
|
page.remove_listener("request", handler)
|
||||||
await page.close()
|
await page.close()
|
||||||
|
|
||||||
|
|
||||||
async def get_events(
|
async def get_events(
|
||||||
client: httpx.AsyncClient,
|
client: httpx.AsyncClient,
|
||||||
base_url: str,
|
base_url: str,
|
||||||
cached_keys: set[str],
|
cached_keys: set[str],
|
||||||
) -> list[dict[str, str]]:
|
) -> list[dict[str, str]]:
|
||||||
if not (api_data := API_FILE.load(per_entry=False, index=True)):
|
if not (api_data := API_FILE.load(per_entry=False, index=True)):
|
||||||
api_data = await refresh_api_cache(
|
api_data = await refresh_api_cache(
|
||||||
client,
|
client,
|
||||||
urljoin(
|
urljoin(
|
||||||
base_url,
|
base_url,
|
||||||
"api/matches/all-today",
|
"api/matches/all-today",
|
||||||
),
|
),
|
||||||
)
|
)
|
||||||
|
|
||||||
API_FILE.write(api_data)
|
API_FILE.write(api_data)
|
||||||
|
|
||||||
events: list[dict[str, str]] = []
|
events: list[dict[str, str]] = []
|
||||||
|
|
||||||
now = Time.clean(Time.now())
|
now = Time.clean(Time.now())
|
||||||
start_dt = now.delta(minutes=-30)
|
start_dt = now.delta(minutes=-30)
|
||||||
end_dt = now.delta(minutes=30)
|
end_dt = now.delta(minutes=30)
|
||||||
pattern = re.compile(r"[\n\r]+|\s{2,}")
|
pattern = re.compile(r"[\n\r]+|\s{2,}")
|
||||||
|
|
||||||
for event in api_data:
|
for event in api_data:
|
||||||
category = event["category"]
|
category = event["category"]
|
||||||
|
|
||||||
if category == "other":
|
if category == "other":
|
||||||
continue
|
continue
|
||||||
|
|
||||||
sport = validate_category(category)
|
if not (ts := event["date"]):
|
||||||
|
continue
|
||||||
parts = pattern.split(event["title"].strip())
|
|
||||||
name = " | ".join(p.strip() for p in parts if p.strip())
|
start_ts = int(str(ts)[:-3])
|
||||||
|
|
||||||
logo = urljoin(base_url, poster) if (poster := event.get("poster")) else None
|
event_dt = Time.from_ts(start_ts)
|
||||||
|
|
||||||
key = f"[{sport}] {name} (STRMD)"
|
if not start_dt <= event_dt <= end_dt:
|
||||||
|
continue
|
||||||
if cached_keys & {key}:
|
|
||||||
continue
|
sport = validate_category(category)
|
||||||
|
|
||||||
if not (ts := event["date"]):
|
parts = pattern.split(event["title"].strip())
|
||||||
continue
|
name = " | ".join(p.strip() for p in parts if p.strip())
|
||||||
|
|
||||||
start_ts = int(str(ts)[:-3])
|
logo = urljoin(base_url, poster) if (poster := event.get("poster")) else None
|
||||||
|
|
||||||
event_dt = Time.from_ts(start_ts)
|
key = f"[{sport}] {name} (STRMD)"
|
||||||
|
|
||||||
if not start_dt <= event_dt <= end_dt:
|
if cached_keys & {key}:
|
||||||
continue
|
continue
|
||||||
|
|
||||||
sources: list[dict[str, str]] = event["sources"]
|
sources: list[dict[str, str]] = event["sources"]
|
||||||
|
|
||||||
if not sources:
|
if not sources:
|
||||||
continue
|
continue
|
||||||
|
|
||||||
source = sources[0]
|
source = sources[0]
|
||||||
# source = sources[1] if len(sources) > 1 else sources[0]
|
# source = sources[1] if len(sources) > 1 else sources[0]
|
||||||
source_type = source.get("source")
|
source_type = source.get("source")
|
||||||
stream_id = source.get("id")
|
stream_id = source.get("id")
|
||||||
|
|
||||||
if not (source_type and stream_id):
|
if not (source_type and stream_id):
|
||||||
continue
|
continue
|
||||||
|
|
||||||
events.append(
|
events.append(
|
||||||
{
|
{
|
||||||
"sport": sport,
|
"sport": sport,
|
||||||
"event": name,
|
"event": name,
|
||||||
"link": f"https://embedsports.top/embed/{source_type}/{stream_id}/1",
|
"link": f"https://embedsports.top/embed/{source_type}/{stream_id}/1",
|
||||||
"logo": logo,
|
"logo": logo,
|
||||||
"timestamp": event_dt.timestamp(),
|
"timestamp": event_dt.timestamp(),
|
||||||
}
|
}
|
||||||
)
|
)
|
||||||
|
|
||||||
return events
|
return events
|
||||||
|
|
||||||
|
|
||||||
async def scrape(client: httpx.AsyncClient) -> None:
|
async def scrape(client: httpx.AsyncClient) -> None:
|
||||||
cached_urls = CACHE_FILE.load()
|
cached_urls = CACHE_FILE.load()
|
||||||
cached_count = len(cached_urls)
|
cached_count = len(cached_urls)
|
||||||
urls.update(cached_urls)
|
urls.update(cached_urls)
|
||||||
|
|
||||||
log.info(f"Loaded {cached_count} event(s) from cache")
|
log.info(f"Loaded {cached_count} event(s) from cache")
|
||||||
|
|
||||||
if not (base_url := await network.get_base(MIRRORS)):
|
if not (base_url := await network.get_base(MIRRORS)):
|
||||||
log.warning("No working PPV mirrors")
|
log.warning("No working PPV mirrors")
|
||||||
CACHE_FILE.write(cached_urls)
|
CACHE_FILE.write(cached_urls)
|
||||||
return
|
return
|
||||||
|
|
||||||
log.info(f'Scraping from "{base_url}"')
|
log.info(f'Scraping from "{base_url}"')
|
||||||
|
|
||||||
events = await get_events(
|
events = await get_events(
|
||||||
client,
|
client,
|
||||||
base_url,
|
base_url,
|
||||||
set(cached_urls.keys()),
|
set(cached_urls.keys()),
|
||||||
)
|
)
|
||||||
|
|
||||||
log.info(f"Processing {len(events)} new URL(s)")
|
log.info(f"Processing {len(events)} new URL(s)")
|
||||||
|
|
||||||
async with async_playwright() as p:
|
async with async_playwright() as p:
|
||||||
browser, context = await network.browser(p, "brave")
|
browser, context = await network.browser(p, "brave")
|
||||||
|
|
||||||
for i, ev in enumerate(events, start=1):
|
for i, ev in enumerate(events, start=1):
|
||||||
url = await network.safe_process(
|
url = await network.safe_process(
|
||||||
lambda: process_event(
|
lambda: process_event(
|
||||||
ev["link"],
|
ev["link"],
|
||||||
url_num=i,
|
url_num=i,
|
||||||
context=context,
|
context=context,
|
||||||
),
|
),
|
||||||
url_num=i,
|
url_num=i,
|
||||||
log=log,
|
log=log,
|
||||||
)
|
)
|
||||||
|
|
||||||
if url:
|
if url:
|
||||||
sport, event, logo, ts = (
|
sport, event, logo, ts = (
|
||||||
ev["sport"],
|
ev["sport"],
|
||||||
ev["event"],
|
ev["event"],
|
||||||
ev["logo"],
|
ev["logo"],
|
||||||
ev["timestamp"],
|
ev["timestamp"],
|
||||||
)
|
)
|
||||||
|
|
||||||
key = f"[{sport}] {event} (STRMD)"
|
key = f"[{sport}] {event} (STRMD)"
|
||||||
|
|
||||||
tvg_id, pic = get_tvg_id(sport, event)
|
tvg_id, pic = get_tvg_info(sport, event)
|
||||||
|
|
||||||
entry = {
|
entry = {
|
||||||
"url": url,
|
"url": url,
|
||||||
"logo": logo or pic,
|
"logo": logo or pic,
|
||||||
"base": "https://embedsports.top/",
|
"base": "https://embedsports.top/",
|
||||||
"timestamp": ts,
|
"timestamp": ts,
|
||||||
"id": tvg_id or "Live.Event.us",
|
"id": tvg_id or "Live.Event.us",
|
||||||
}
|
}
|
||||||
|
|
||||||
urls[key] = cached_urls[key] = entry
|
urls[key] = cached_urls[key] = entry
|
||||||
|
|
||||||
await browser.close()
|
await browser.close()
|
||||||
|
|
||||||
if new_count := len(cached_urls) - cached_count:
|
if new_count := len(cached_urls) - cached_count:
|
||||||
log.info(f"Collected and cached {new_count} new event(s)")
|
log.info(f"Collected and cached {new_count} new event(s)")
|
||||||
else:
|
else:
|
||||||
log.info("No new events found")
|
log.info("No new events found")
|
||||||
|
|
||||||
CACHE_FILE.write(cached_urls)
|
CACHE_FILE.write(cached_urls)
|
||||||
|
|
|
||||||
Loading…
Add table
Add a link
Reference in a new issue