Update app.py

app.py

@@ -3,7 +3,7 @@ from fastapi.responses import PlainTextResponse
 from urllib.parse import unquote
 import uvicorn
 from scrapling import AsyncFetcher, StealthyFetcher 
-from markitdown import MarkItDown
+from markitdown._markitdown import HtmlConverter
 import tempfile
 import os
 import asyncio
@@ -13,7 +13,7 @@ app = FastAPI()
 
 async_fetcher = AsyncFetcher(auto_match=True)  # Use AsyncFetcher
 stealthy_fetcher = StealthyFetcher()  # Keep StealthyFetcher (it handles its own async internally)
-md = MarkItDown()
+md = HtmlConverter()
 
 
 async def stealthy_scraper(url):
@@ -37,17 +37,8 @@ async def text_scraper(url):
 
 async def convert_html_to_md(html):
     """Converts HTML to Markdown using MarkItDown (assuming it's synchronous)."""
-    with tempfile.NamedTemporaryFile(suffix=".html", delete=False) as temp_file:
-        # Asynchronously write to the temporary file
-        await asyncio.to_thread(temp_file.write, html.encode('utf-8'))
-        await asyncio.to_thread(temp_file.flush)
-        temp_file_path = temp_file.name
-        # Use asyncio.to_thread for synchronous operations within async functions
-        md_text = await asyncio.to_thread(md.convert, temp_file_path)
-        x = md_text.text_content
-    # Asynchronously remove file
-    await asyncio.to_thread(os.remove, temp_file_path)
-    return x
+    md_text = await asyncio.to_thread(md._convert, html)
+    return md_text.text_content
 
 
 @app.get("/read/{url:path}", response_class=PlainTextResponse)