Spaces:
Sleeping
Sleeping
Update VisionaryAgent.py
Browse files- VisionaryAgent.py +13 -2
VisionaryAgent.py
CHANGED
|
@@ -13,6 +13,17 @@ import helium
|
|
| 13 |
from typing import List
|
| 14 |
from pydantic import BaseModel, Field
|
| 15 |
from fastapi import UploadFile
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
# Load environment variables (API keys, etc.)
|
| 18 |
from dotenv import load_dotenv
|
|
@@ -56,8 +67,8 @@ def scrape_website(url: str):
|
|
| 56 |
return firecrawl_agent.print_response(f"Extract all relevant business information from {url}, including mission statement, services, case studies, and AI-related content. Provide structured output.")
|
| 57 |
|
| 58 |
# Helium for dynamic websites
|
| 59 |
-
chrome_options = webdriver.ChromeOptions()
|
| 60 |
-
chrome_options.add_argument("--headless")
|
| 61 |
driver = helium.start_chrome(headless=True, options=chrome_options)
|
| 62 |
|
| 63 |
def scrape_dynamic_website(url: str):
|
|
|
|
| 13 |
from typing import List
|
| 14 |
from pydantic import BaseModel, Field
|
| 15 |
from fastapi import UploadFile
|
| 16 |
+
from selenium import webdriver
|
| 17 |
+
import helium
|
| 18 |
+
|
| 19 |
+
# Configure Chrome options for Hugging Face Spaces
|
| 20 |
+
chrome_options = webdriver.ChromeOptions()
|
| 21 |
+
chrome_options.add_argument("--headless") # Runs Chrome in headless mode
|
| 22 |
+
chrome_options.add_argument("--no-sandbox") # Required for running on HF Spaces
|
| 23 |
+
chrome_options.add_argument("--disable-dev-shm-usage") # Prevents memory errors
|
| 24 |
+
chrome_options.add_argument("--disable-gpu") # Disable GPU acceleration (not needed in headless mode)
|
| 25 |
+
chrome_options.add_argument("--remote-debugging-port=9222") # Helps with debugging if needed
|
| 26 |
+
|
| 27 |
|
| 28 |
# Load environment variables (API keys, etc.)
|
| 29 |
from dotenv import load_dotenv
|
|
|
|
| 67 |
return firecrawl_agent.print_response(f"Extract all relevant business information from {url}, including mission statement, services, case studies, and AI-related content. Provide structured output.")
|
| 68 |
|
| 69 |
# Helium for dynamic websites
|
| 70 |
+
# chrome_options = webdriver.ChromeOptions()
|
| 71 |
+
# chrome_options.add_argument("--headless")
|
| 72 |
driver = helium.start_chrome(headless=True, options=chrome_options)
|
| 73 |
|
| 74 |
def scrape_dynamic_website(url: str):
|