Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,11 +1,14 @@
|
|
| 1 |
import os, asyncio
|
| 2 |
from fastapi import FastAPI, HTTPException, Header, Response
|
| 3 |
from pydantic import BaseModel, HttpUrl
|
| 4 |
-
from
|
|
|
|
|
|
|
| 5 |
|
| 6 |
API_KEY = os.getenv("OPENER_API_KEY", "")
|
| 7 |
-
PROFILE = os.getenv("
|
| 8 |
UA = os.getenv("UA", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/129.0.0.0 Safari/537.36")
|
|
|
|
| 9 |
|
| 10 |
app = FastAPI()
|
| 11 |
drv = None
|
|
@@ -15,33 +18,31 @@ worker = None
|
|
| 15 |
class OpenPayload(BaseModel):
|
| 16 |
url: HttpUrl
|
| 17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
async def _worker():
|
| 19 |
while True:
|
| 20 |
url = await queue.get()
|
| 21 |
try:
|
| 22 |
await asyncio.to_thread(drv.get, url)
|
| 23 |
print("[open] done", url)
|
| 24 |
-
except Exception as e:
|
| 25 |
-
print("[open] err", url, e)
|
| 26 |
finally:
|
| 27 |
queue.task_done()
|
| 28 |
|
| 29 |
@app.on_event("startup")
|
| 30 |
async def startup():
|
| 31 |
global drv, queue, worker
|
| 32 |
-
|
| 33 |
-
os.makedirs("downloaded_files", exist_ok=True)
|
| 34 |
-
os.makedirs(PROFILE, exist_ok=True)
|
| 35 |
-
drv = Driver(
|
| 36 |
-
uc=True,
|
| 37 |
-
headless=True,
|
| 38 |
-
incognito=False,
|
| 39 |
-
user_data_dir=PROFILE,
|
| 40 |
-
page_load_strategy="normal",
|
| 41 |
-
agent=UA,
|
| 42 |
-
no_sandbox=True,
|
| 43 |
-
disable_gpu=True,
|
| 44 |
-
)
|
| 45 |
drv.get("data:text/html,ready")
|
| 46 |
queue = asyncio.Queue()
|
| 47 |
worker = asyncio.create_task(_worker())
|
|
|
|
| 1 |
import os, asyncio
|
| 2 |
from fastapi import FastAPI, HTTPException, Header, Response
|
| 3 |
from pydantic import BaseModel, HttpUrl
|
| 4 |
+
from selenium import webdriver
|
| 5 |
+
from selenium.webdriver.chrome.service import Service
|
| 6 |
+
from selenium.webdriver.chrome.options import Options
|
| 7 |
|
| 8 |
API_KEY = os.getenv("OPENER_API_KEY", "")
|
| 9 |
+
PROFILE = os.getenv("CHROME_PROFILE_DIR", "/tmp/chrome-profile")
|
| 10 |
UA = os.getenv("UA", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/129.0.0.0 Safari/537.36")
|
| 11 |
+
CHROMEDRIVER = os.getenv("CHROMEDRIVER", "/usr/bin/chromedriver") # debian usually symlinks here
|
| 12 |
|
| 13 |
app = FastAPI()
|
| 14 |
drv = None
|
|
|
|
| 18 |
class OpenPayload(BaseModel):
|
| 19 |
url: HttpUrl
|
| 20 |
|
| 21 |
+
def _build_driver():
|
| 22 |
+
os.makedirs(PROFILE, exist_ok=True)
|
| 23 |
+
opts = Options()
|
| 24 |
+
opts.add_argument("--headless=new")
|
| 25 |
+
opts.add_argument("--no-sandbox")
|
| 26 |
+
opts.add_argument("--disable-gpu")
|
| 27 |
+
opts.add_argument("--disable-dev-shm-usage")
|
| 28 |
+
opts.add_argument(f"--user-agent={UA}")
|
| 29 |
+
opts.add_argument(f"--user-data-dir={PROFILE}")
|
| 30 |
+
svc = Service(executable_path=CHROMEDRIVER)
|
| 31 |
+
return webdriver.Chrome(service=svc, options=opts)
|
| 32 |
+
|
| 33 |
async def _worker():
|
| 34 |
while True:
|
| 35 |
url = await queue.get()
|
| 36 |
try:
|
| 37 |
await asyncio.to_thread(drv.get, url)
|
| 38 |
print("[open] done", url)
|
|
|
|
|
|
|
| 39 |
finally:
|
| 40 |
queue.task_done()
|
| 41 |
|
| 42 |
@app.on_event("startup")
|
| 43 |
async def startup():
|
| 44 |
global drv, queue, worker
|
| 45 |
+
drv = _build_driver()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
drv.get("data:text/html,ready")
|
| 47 |
queue = asyncio.Queue()
|
| 48 |
worker = asyncio.create_task(_worker())
|