Spaces:
Sleeping
Sleeping
himanshu-muk commited on
Commit ·
928cdf1
1
Parent(s): a999b8c
Fix API to directly return flowPages value instead of parsing console logs
Browse files
app.py
CHANGED
|
@@ -1,4 +1,3 @@
|
|
| 1 |
-
import json
|
| 2 |
import asyncio
|
| 3 |
from fastapi import FastAPI, HTTPException
|
| 4 |
from playwright.async_api import async_playwright
|
|
@@ -8,41 +7,41 @@ app = FastAPI()
|
|
| 8 |
|
| 9 |
@app.post("/run")
|
| 10 |
async def run_flow_pages(url: str):
|
| 11 |
-
console_result = None
|
| 12 |
-
|
| 13 |
async with async_playwright() as p:
|
| 14 |
browser = await p.chromium.launch(headless=True)
|
| 15 |
page = await browser.new_page()
|
| 16 |
|
| 17 |
-
# Capture console messages
|
| 18 |
-
async def handle_console(msg):
|
| 19 |
-
nonlocal console_result
|
| 20 |
-
try:
|
| 21 |
-
console_result = json.loads(msg.text())
|
| 22 |
-
except Exception:
|
| 23 |
-
pass # ignore non-JSON logs
|
| 24 |
-
|
| 25 |
-
page.on("console", handle_console)
|
| 26 |
-
|
| 27 |
try:
|
| 28 |
-
await page.goto(url, wait_until="load")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
|
| 30 |
-
|
| 31 |
-
|
|
|
|
|
|
|
|
|
|
| 32 |
|
| 33 |
-
|
| 34 |
-
await asyncio.sleep(1)
|
| 35 |
|
| 36 |
except Exception as e:
|
| 37 |
await browser.close()
|
| 38 |
raise HTTPException(status_code=500, detail=str(e))
|
| 39 |
-
|
| 40 |
-
await browser.close()
|
| 41 |
-
|
| 42 |
-
if console_result is None:
|
| 43 |
-
raise HTTPException(
|
| 44 |
-
status_code=500,
|
| 45 |
-
detail="No JSON response received from flowPages"
|
| 46 |
-
)
|
| 47 |
-
|
| 48 |
-
return console_result
|
|
|
|
|
|
|
| 1 |
import asyncio
|
| 2 |
from fastapi import FastAPI, HTTPException
|
| 3 |
from playwright.async_api import async_playwright
|
|
|
|
| 7 |
|
| 8 |
@app.post("/run")
|
| 9 |
async def run_flow_pages(url: str):
|
|
|
|
|
|
|
| 10 |
async with async_playwright() as p:
|
| 11 |
browser = await p.chromium.launch(headless=True)
|
| 12 |
page = await browser.new_page()
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
try:
|
| 15 |
+
await page.goto(url, wait_until="load", timeout=30000)
|
| 16 |
+
|
| 17 |
+
# Wait a bit for the page to fully load and flowPages to be available
|
| 18 |
+
await asyncio.sleep(2)
|
| 19 |
+
|
| 20 |
+
# Directly evaluate and return flowPages from the page
|
| 21 |
+
flow_pages_data = await page.evaluate("""() => {
|
| 22 |
+
// Check if flowPages exists in various scopes
|
| 23 |
+
if (typeof flowPages !== 'undefined') {
|
| 24 |
+
return flowPages;
|
| 25 |
+
}
|
| 26 |
+
if (typeof window !== 'undefined' && window.flowPages) {
|
| 27 |
+
return window.flowPages;
|
| 28 |
+
}
|
| 29 |
+
if (typeof globalThis !== 'undefined' && globalThis.flowPages) {
|
| 30 |
+
return globalThis.flowPages;
|
| 31 |
+
}
|
| 32 |
+
return null;
|
| 33 |
+
}""")
|
| 34 |
+
|
| 35 |
+
await browser.close()
|
| 36 |
|
| 37 |
+
if flow_pages_data is None:
|
| 38 |
+
raise HTTPException(
|
| 39 |
+
status_code=500,
|
| 40 |
+
detail="flowPages not found on the page"
|
| 41 |
+
)
|
| 42 |
|
| 43 |
+
return flow_pages_data
|
|
|
|
| 44 |
|
| 45 |
except Exception as e:
|
| 46 |
await browser.close()
|
| 47 |
raise HTTPException(status_code=500, detail=str(e))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|