Update app.py

app.py

@@ -1,23 +1,45 @@
 import os
 import asyncio
 import time
-from typing import Optional, List
-from datetime import datetime
+from typing import Optional
+from datetime import datetime, timezone
 import httpx
 import trafilatura
 import gradio as gr
 from dateutil import parser as dateparser
-from limits import parse
-from limits.aio.storage import MemoryStorage
-from limits.aio.strategies import MovingWindowRateLimiter
-from analytics import record_request, last_n_days_df, last_n_days_avg_time_df
 from fastapi import FastAPI
 from pydantic import BaseModel
-import google.generativeai as genai # <<< NEW IMPORT
+import google.generativeai as genai
+# <<< MISSING IMPORT ADDED BACK >>>
+from analytics import record_request, last_n_days_df, last_n_days_avg_time_df
+
+# --- Prompts ---
+PROMPT_NORMAL = """
+Based on the user's original query, provide a concise summary (3-4 important bullet points) of the following text. Focus only on the most critical information.
+USER'S QUERY: "{query}"
+TEXT TO SUMMARIZE:
+---
+{context_text}
+---
+"""
+PROMPT_DEEP = """
+As a meticulous research analyst, your task is to synthesize the information from the provided web search results into a detailed and comprehensive report.
+**Current Date:** {current_date}.
+**VERY IMPORTANT:** Your top priority is to provide information relevant to this current date and the future. If the user's query is about a recurring event (like an exam), you MUST focus on the upcoming or current event.
+**User's Original Query:** "{query}"
+**Instructions:**
+1.  Combine information from different sources to create a coherent and detailed report.
+2.  Cite source URLs inline, like this: "(Source: http://...)." The URL is provided in the text.
+3.  At the end of your report, create a "## Sources" section and list all the unique URLs you used.
+4.  Use clear markdown with headings and bold text.
+**Provided Search Results:**
+---
+{context_text}
+---
+"""
 
-# --- Core Search Logic (No Changes) ---
+# --- Core Search Logic ---
 async def search_web_logic(query: str, serper_api_key: str, search_type: str, num_results: int) -> str:
-    # ... (Ye function bilkul waisa hi hai jaisa pehle tha, isme koi badlav nahi)
     start_time = time.time()
     if not serper_api_key: return "Error: Serper API Key is required."
     num_results = max(1, min(20, num_results))
@@ -29,7 +51,7 @@ async def search_web_logic(query: str, serper_api_key: str, search_type: str, nu
             resp = await client.post(endpoint, headers=headers, json=payload)
         if resp.status_code != 200: return f"Error: Search API returned status {resp.status_code}."
         results = resp.json().get("news" if search_type == "news" else "organic", [])
-        if not results: return f"No {search_type} results found."
+        if not results: return f"No {search_type} results found for '{query}'."
         urls = [r["link"] for r in results]
         async with httpx.AsyncClient(timeout=20, follow_redirects=True) as client:
             tasks = [client.get(u) for u in urls]; responses = await asyncio.gather(*tasks, return_exceptions=True)
@@ -44,75 +66,102 @@ async def search_web_logic(query: str, serper_api_key: str, search_type: str, nu
             else:
                 domain = meta["link"].split("/")[2].replace("www.", ""); chunk = f"## {meta['title']}\n**Domain:** {domain}\n**URL:** {meta['link']}\n\n{body.strip()}\n"
             chunks.append(chunk)
-        if not chunks: return "Found results, but couldn't extract content."
+        if not chunks: return "Found results for '{query}', but couldn't extract content."
         summary = f"Successfully extracted content from {successful_extractions}/{len(results)} results.\n\n---\n\n"
         await record_request(time.time() - start_time, num_results)
         return summary + "\n---\n".join(chunks)
     except Exception as e:
         return f"An error occurred during web search: {str(e)}"
 
-# <<< NEW: Gemini Summarization Logic >>>
-async def summarize_with_gemini(text_to_summarize: str, query: str, gemini_key: str, model_name: str) -> str:
+# --- Gemini Summarization Logic ---
+async def summarize_with_gemini(text_to_summarize: str, query: str, gemini_key: str, model_name: str, research_mode: str) -> str:
     try:
         genai.configure(api_key=gemini_key)
         model = genai.GenerativeModel(model_name)
+        current_date = datetime.now(timezone.utc).strftime("%Y-%m-%d")
         
-        prompt = f"""
-        Based on the user's original query, provide a concise summary of the following text.
-        Focus on the information that directly answers or relates to the query.
-        Filter out any irrelevant details, advertisements, or boilerplate content.
-        Present the final summary in clean, easy-to-read markdown format.
-
-        USER'S QUERY: "{query}"
-
-        TEXT TO SUMMARIZE:
-        ---
-        {text_to_summarize}
-        ---
-        """
-        
+        if research_mode == 'deep':
+            prompt_template = PROMPT_DEEP
+        else: # Default to normal
+            prompt_template = PROMPT_NORMAL
+            
+        prompt = prompt_template.format(query=query, context_text=text_to_summarize, current_date=current_date)
         response = await model.generate_content_async(prompt)
         return response.text
     except Exception as e:
         return f"\n\n--- ⚠️ Gemini Summarization Failed ---\nError: {str(e)}\nReturning raw text instead."
 
+# --- Main Orchestrator Function ---
+async def search_and_summarize(query, serper_api_key, search_type, num_results, gemini_api_key, gemini_model, research_mode):
+    scraped_text = await search_web_logic(query, serper_api_key, search_type, num_results)
+    
+    if gemini_api_key and "Error:" not in scraped_text:
+        summarized_text = await summarize_with_gemini(scraped_text, query, gemini_api_key, gemini_model, research_mode)
+        if "⚠️ Gemini Summarization Failed" in summarized_text:
+             return scraped_text + summarized_text
+        else:
+             return summarized_text
+    return scraped_text
+
 # --- FastAPI App ---
 app = FastAPI()
+# Add CORS middleware if you plan to call the API from a different domain/frontend
+# from fastapi.middleware.cors import CORSMiddleware
+# app.add_middleware(CORSMiddleware, allow_origins=["*"], allow_credentials=True, allow_methods=["*"], allow_headers=["*"])
 
-# <<< CHANGE: FastAPI request model ko naye fields ke sath update kiya gaya >>>
 class SearchRequest(BaseModel):
     query: str
     serper_api_key: str
     search_type: str = "search"
     num_results: int = 4
-    gemini_api_key: Optional[str] = None # Optional field
-    gemini_model: Optional[str] = "gemini-1.5-flash-latest" # Optional field with default
+    gemini_api_key: Optional[str] = None
+    gemini_model: Optional[str] = "gemini-1.5-flash-latest"
+    research_mode: str = "normal"
 
 @app.post("/api/search")
 async def api_search(request: SearchRequest):
-    # Step 1: Web Search
-    scraped_text = await search_web_logic(
-        request.query, request.serper_api_key, request.search_type, request.num_results
+    result = await search_and_summarize(
+        request.query, request.serper_api_key, request.search_type, request.num_results,
+        request.gemini_api_key, request.gemini_model, request.research_mode
     )
+    return {"result": result}
 
-    # Step 2: (Optional) Summarize with Gemini
-    if request.gemini_api_key and "Error:" not in scraped_text:
-        summarized_text = await summarize_with_gemini(
-            scraped_text, request.query, request.gemini_api_key, request.gemini_model
-        )
-        # Agar summarization fail hota hai, to original text wapas bhej dein
-        if "⚠️ Gemini Summarization Failed" in summarized_text:
-             return {"result": scraped_text + summarized_text}
-        else:
-             return {"result": summarized_text}
-
-    return {"result": scraped_text}
+# --- Gradio App ---
+def create_gradio_app():
+    with gr.Blocks(title="Web Search & Summarize UI") as demo:
+        gr.Markdown("# 🔍 AI Search & Summarize")
+        with gr.Tabs():
+            with gr.Tab("App"):
+                gr.Markdown("### Step 1: Web Search")
+                query_input = gr.Textbox(label="Search Query")
+                serper_api_key_input = gr.Textbox(label="Your Serper API Key", type="password")
+                with gr.Row():
+                    search_type_input = gr.Radio(["search", "news"], value="search", label="Search Type")
+                    num_results_input = gr.Slider(1, 20, value=4, step=1, label="Number of Results")
+                
+                gr.Markdown("### Step 2: AI Summarization")
+                research_mode_input = gr.Radio(["normal", "deep"], value="normal", label="Research Mode", info="Normal for fast summary, Deep for detailed report.")
+                gemini_api_key_input = gr.Textbox(label="Your Gemini API Key", type="password", placeholder="Leave empty to skip summarization")
+                gemini_model_input = gr.Textbox(label="Gemini Model", value="gemini-1.5-flash-latest")
+                search_button = gr.Button("Search & Summarize", variant="primary")
+                output = gr.Textbox(label="Result", lines=25, max_lines=40)
+                
+                search_button.click(
+                    fn=search_and_summarize,
+                    inputs=[query_input, serper_api_key_input, search_type_input, num_results_input, gemini_api_key_input, gemini_model_input, research_mode_input],
+                    outputs=output
+                )
+            with gr.Tab("Analytics"):
+                requests_plot = gr.BarPlot(x="date", y="count", title="Daily Requests")
+                avg_time_plot = gr.BarPlot(x="date", y="avg_time", title="Avg. Response Time (s)")
+                def update_analytics(): return last_n_days_df(14), last_n_days_avg_time_df(14)
+                demo.load(update_analytics, [], [requests_plot, avg_time_plot])
+    return demo
 
-# --- Gradio App (ab iski zaroorat nahi, lekin rakha ja sakta hai) ---
-# ... (Gradio UI ka code abhi ke liye comment out kar sakte hain ya hata sakte hain)
-# ... agar aapko UI bhi chahiye to hum use bhi update kar sakte hain
+# --- Mount and Startup ---
+gradio_ui = create_gradio_app()
+app = gr.mount_gradio_app(app, gradio_ui, path="/")
 
-# --- Server Startup ---
 if __name__ == "__main__":
     import uvicorn
     uvicorn.run(app, host="0.0.0.0", port=7860)