feat: target channel context history, custom website extraction needs, and categorized main menu

services/website_analyzer.py

@@ -38,7 +38,7 @@ class WebsiteAnalyzer:
     def __init__(self, llm: LLMClient):
         self.llm = llm
 
-    async def analyze_website(self, url: str) -> Tuple[bool, Dict[str, Any], str]:
+    async def analyze_website(self, url: str, custom_instructions: str = "") -> Tuple[bool, Dict[str, Any], str]:
         """Discover endpoints and prompt AI to build the extractor schema."""
         clean_url = url.strip()
         if not clean_url.startswith("http://") and not clean_url.startswith("https://"):
@@ -109,11 +109,14 @@ class WebsiteAnalyzer:
             except Exception as e:
                 return False, {}, f"خطا در برقراری ارتباط با وبسایت: {e}"
 
+        custom_block = f"\nUser Extraction Needs / Filtering Instructions:\n{custom_instructions.strip()}\n" if custom_instructions and custom_instructions.strip() else ""
+
         prompt = (
             f"Website Target URL: {clean_url}\n"
+            f"{custom_block}"
             f"Discovery Notes:\n" + "\n".join(discovered_info) + "\n\n"
             f"Data / Feed Sample:\n{sample_data}\n\n"
-            f"Determine the best endpoint_url, parser_type, and field mappings to regularly extract the latest news/articles from this site."
+            f"Determine the best endpoint_url, parser_type, and field mappings to regularly extract the latest news/articles from this site matching the user requirements."
         )
 
         try: