Update app.py

app.py

@@ -3,34 +3,27 @@ import requests
 import time
 from duckduckgo_search import DDGS
 
-# === Model functions ===
-
+# DuckDuckGo search
 def search_articles(name: str) -> str:
-    """Search for 3 newspaper articles containing the name and keywords using DuckDuckGo"""
     keywords = ['founders', 'partners', 'funders', 'owners']
     search_query = f'"{name}" ({" OR ".join(keywords)}) site:news'
     try:
-        print(f"[DEBUG] Search query: {search_query}")
         with DDGS() as ddgs:
             results = list(ddgs.text(search_query, max_results=3))
-            print(f"[DEBUG] Raw results: {results}")
         if not results:
             return f"No articles found for {name}"
-
         articles = []
         for i, result in enumerate(results, 1):
-            article = f"**{i}. {result.get('title', 'No Title')}**\n"
-            article += f"Source: {result.get('href', 'No URL')}\n"
-            article += f"{result.get('body', 'No Body')}\n"
+            article = f"**{i}. {result['title']}**\n"
+            article += f"Source: {result['href']}\n"
+            article += f"{result['body']}\n"
             articles.append(article)
-
         return "\n\n".join(articles)
     except Exception as e:
-        return f"[ERROR] Search failed: {str(e)}"
-
+        return f"Search failed: {str(e)}"
 
+# Call your Modal Mistral endpoint
 def extract_entities(search_results: str) -> str:
-    """Extract entities using Mistral 7B endpoint"""
     modal_endpoint = "https://msoaresdiego--mistral-llm-endpoint-fastapi-app.modal.run/generate"
     prompt = f"""Extract all person names and organization names from the following text.
 Format as:
@@ -40,18 +33,17 @@ Text: {search_results}"""
     try:
         response = requests.post(
             modal_endpoint,
-            json={"prompt": prompt, "max_tokens": 500, "temperature": 0.1}
+            json={"prompt": prompt, "max_tokens": 500, "temperature": 0.1},
+            timeout=30
         )
         if response.status_code == 200:
             return response.json().get("response", "No entities extracted")
         else:
-            return f"[ERROR] API Error: {response.status_code}"
+            return f"API Error: {response.status_code}"
     except Exception as e:
-        return f"[ERROR] Extraction failed: {str(e)}"
-
+        return f"Extraction failed: {str(e)}"
 
 def find_full_names(search_results: str, entities: str) -> str:
-    """Find full names using Mistral 7B endpoint"""
     modal_endpoint = "https://msoaresdiego--mistral-llm-endpoint-fastapi-app.modal.run/generate"
     prompt = f"""Based on the search results, find the full names and titles/roles for these entities:
 Entities: {entities}
@@ -60,100 +52,59 @@ Provide full names with their roles/titles where mentioned."""
     try:
         response = requests.post(
             modal_endpoint,
-            json={"prompt": prompt, "max_tokens": 300, "temperature": 0.1}
+            json={"prompt": prompt, "max_tokens": 300, "temperature": 0.1},
+            timeout=30
         )
         if response.status_code == 200:
             return response.json().get("response", "No full names found")
         else:
-            return f"[ERROR] API Error: {response.status_code}"
+            return f"API Error: {response.status_code}"
     except Exception as e:
-        return f"[ERROR] Full name extraction failed: {str(e)}"
-
-
-# === Gradio interface ===
+        return f"Full name extraction failed: {str(e)}"
 
+# Pipeline with separate status
 def process_name_with_progress(name: str, progress=gr.Progress()):
     if not name.strip():
-        return "", "", ""
-
-    # Initialize outputs
-    search_results = ""
-    entities = ""
-    full_names = ""
+        yield "", "", "", "Please enter a name"
+        return
 
     try:
-        # Step 1: Search for articles
-        progress(0.1, desc="Searching for articles...")
-        yield "", "", ""  # Initial blank yield to reset UI
-
-        search_start = time.time()
+        yield "", "", "", "Searching for articles..."
+        progress(0.1)
         search_results = search_articles(name.strip())
-        search_time = time.time() - search_start
-        print(f"Search took: {search_time:.2f} seconds")
 
-        progress(0.4, desc="Articles found! Extracting entities...")
-        yield search_results, "", ""  # Only show real search results
-
-        # Step 2: Extract entities from search results
-        extract_start = time.time()
+        yield search_results, "", "", "Extracting entities..."
+        progress(0.4)
         entities = extract_entities(search_results)
-        extract_time = time.time() - extract_start
-        print(f"Entity extraction took: {extract_time:.2f} seconds")
-
-        progress(0.7, desc="Entities extracted! Finding full names...")
-        yield search_results, entities, ""  # Show real entities
 
-        # Step 3: Find full names of entities
-        names_start = time.time()
+        yield search_results, entities, "", "Finding full names..."
+        progress(0.7)
         full_names = find_full_names(search_results, entities)
-        names_time = time.time() - names_start
-        print(f"Full name extraction took: {names_time:.2f} seconds")
 
-        progress(1.0, desc="Complete!")
-        yield search_results, entities, full_names  # Final result
+        progress(1.0)
+        yield search_results, entities, full_names, "Done!"
 
     except Exception as e:
-        error_msg = f"Error: {str(e)}"
-        yield search_results or error_msg, entities or error_msg, full_names or error_msg
-
+        error = f"Error: {str(e)}"
+        yield error, error, error, error
 
+# Simple non-streamed debug version
 def process_name_simple(name: str):
-    """Basic version without progress bar - for isolated testing"""
     if not name.strip():
-        return "", "", ""
-
-    print(f"Starting process for: {name}")
-    total_start = time.time()
-
-    print("Step 1: Searching articles...")
-    search_start = time.time()
-    search_results = search_articles(name.strip())
-    search_time = time.time() - search_start
-    print(f"Search completed in: {search_time:.2f}s")
-
-    print("Step 2: Extracting entities...")
-    extract_start = time.time()
-    entities = extract_entities(search_results)
-    extract_time = time.time() - extract_start
-    print(f"Entity extraction in: {extract_time:.2f}s")
-
-    print("Step 3: Finding full names...")
-    names_start = time.time()
-    full_names = find_full_names(search_results, entities)
-    names_time = time.time() - names_start
-    print(f"Full name resolution in: {names_time:.2f}s")
-
-    total_time = time.time() - total_start
-    print(f"Total time: {total_time:.2f}s")
-
-    return search_results, entities, full_names
-
-
-# === Gradio UI ===
+        return "", "", "", "Please enter a name"
+    try:
+        search_results = search_articles(name.strip())
+        entities = extract_entities(search_results)
+        full_names = find_full_names(search_results, entities)
+        return search_results, entities, full_names, "Done!"
+    except Exception as e:
+        error = f"Error: {str(e)}"
+        return error, error, error, error
 
+# UI layout
 with gr.Blocks(title="Name Research Tool") as demo:
-    gr.Markdown("# 🔎 Name Research Tool")
-    gr.Markdown("Enter a business or project name to search for related articles and extract key entities.")
+    gr.Markdown("# 🔍 Name Research Tool")
+    gr.Markdown("Enter a business or project name to search related articles and extract key entities.")
 
     with gr.Row():
         name_input = gr.Textbox(label="Name", placeholder="Enter business or project name")
@@ -161,22 +112,25 @@ with gr.Blocks(title="Name Research Tool") as demo:
             search_btn = gr.Button("Search (Real-time)", variant="primary")
             debug_btn = gr.Button("Search (Debug Mode)", variant="secondary")
 
-    with gr.Column():
-        output1 = gr.Textbox(label="Search Results (with debug)", lines=10, max_lines=30)
-        output2 = gr.Textbox(label="Extracted Entities", lines=5, max_lines=10)
-        output3 = gr.Textbox(label="Full Names", lines=5, max_lines=10)
+    with gr.Row():
+        with gr.Column():
+            output1 = gr.Textbox(label="Search Results", lines=10)
+            output2 = gr.Textbox(label="Extracted Entities", lines=5)
+            output3 = gr.Textbox(label="Full Names & Roles", lines=5)
+        status = gr.Textbox(label="Status", interactive=False)
 
     search_btn.click(
         fn=process_name_with_progress,
         inputs=[name_input],
-        outputs=[output1, output2, output3]
+        outputs=[output1, output2, output3, status]
     )
 
     debug_btn.click(
         fn=process_name_simple,
         inputs=[name_input],
-        outputs=[output1, output2, output3]
+        outputs=[output1, output2, output3, status]
     )
 
 if __name__ == "__main__":
     demo.launch()
+