Update app.py

app.py

@@ -2,31 +2,32 @@ import os
 import streamlit as st
 import asyncio
 import nest_asyncio
+import base64
+import tempfile
 from gpt_researcher import GPTResearcher
 from dotenv import load_dotenv
 
-# Apply for async event loop (required for Streamlit + asyncio)
 nest_asyncio.apply()
 load_dotenv()
 
-# Set the Tavily API key (used internally by gpt-researcher)
+# Inject Tavily API key
 os.environ["TAVILY_API_KEY"] = "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"
 
-# Streamlit UI config
 st.set_page_config(page_title="🧠 Super Deep Research Agent", layout="wide")
 st.title("📚 GPT-Powered Super Deep Research Assistant")
 
-# Sidebar configuration
+# --- Sidebar UI ---
 with st.sidebar:
-    st.header("🔍 Research Configuration")
+    st.header("🔍 Setup Research Agent")
     query = st.text_input("📌 Research Topic", "Is AI a threat to creative jobs?")
     report_type = st.selectbox("📄 Report Type", ["research_report", "summary", "detailed_report"])
     tone = st.selectbox("🗣️ Tone", ["objective", "persuasive", "informative"])
     source_type = st.selectbox("📡 Source Scope", ["web", "arxiv", "semantic-scholar", "hybrid"])
     output_format = st.selectbox("📝 Output Format", ["markdown", "text"])
+    export_format = st.selectbox("📤 Export Format", ["Markdown", "PDF", "LaTeX"])
     start = st.button("🚀 Start Deep Research")
 
-# Async logic
+# Async runner
 async def run_research_with_logs(query, report_type, source, tone, fmt, log_callback):
     agent = GPTResearcher(
         query=query,
@@ -34,6 +35,7 @@ async def run_research_with_logs(query, report_type, source, tone, fmt, log_call
         report_source=source,
         report_format=fmt,
         tone=tone,
+        log_fn=log_callback
     )
     await agent.conduct_research()
     report = await agent.write_report()
@@ -42,51 +44,74 @@ async def run_research_with_logs(query, report_type, source, tone, fmt, log_call
     images = agent.get_research_images()
     return report, context, sources, images
 
-# Main research execution
+# File Export Utility
+def export_file(content, export_format):
+    filename_base = f"deep_research_{query.replace(' ', '_')}"
+    
+    if export_format == "Markdown":
+        return content, f"{filename_base}.md", "text/markdown"
+    
+    elif export_format == "LaTeX":
+        latex_content = f"\\documentclass{{article}}\n\\begin{{document}}\n{content}\n\\end{{document}}"
+        return latex_content, f"{filename_base}.tex", "text/plain"
+
+    elif export_format == "PDF":
+        try:
+            from fpdf import FPDF
+        except ImportError:
+            st.error("Please install `fpdf` to enable PDF export: `pip install fpdf`")
+            return None, None, None
+        
+        pdf = FPDF()
+        pdf.add_page()
+        pdf.set_font("Arial", size=12)
+        for line in content.split('\n'):
+            pdf.multi_cell(0, 10, line)
+        
+        temp_path = tempfile.mktemp(suffix=".pdf")
+        pdf.output(temp_path)
+        with open(temp_path, "rb") as f:
+            return f.read(), f"{filename_base}.pdf", "application/pdf"
+    
+    return None, None, None
+
+# --- Main Execution ---
 if start and query:
-    st.info("🤖 Running super deep research agent...")
+    st.info("🤖 Running super deep research...")
+
     log_placeholder = st.empty()
+    log_text = ""
 
-    # Enclosing function to use nonlocal log_text
-    def run_loggable_research():
-        log_text = ""
-
-        def stream_log(msg):
-            nonlocal log_text
-            log_text += f"🟢 {msg}\n"
-            log_placeholder.code(log_text, language="text")
-
-        return asyncio.run(
-            run_research_with_logs(
-                query,
-                report_type,
-                source_type,
-                tone,
-                output_format,
-                log_callback=stream_log
-            )
-        )
+    def stream_log(msg):
+        nonlocal log_text
+        log_text += f"🟢 {msg}\n"
+        log_placeholder.code(log_text, language="text")
 
-    # Run it
-    report, context, sources, images = run_loggable_research()
+    report, context, sources, images = asyncio.run(
+        run_research_with_logs(query, report_type, source_type, tone, output_format, log_callback=stream_log)
+    )
 
-    st.success("✅ Research Complete!")
+    st.success("✅ Report generation complete!")
 
-    # Display report
     st.subheader("📄 Final Report")
     st.markdown(report, unsafe_allow_html=True)
 
-    # Display sources
-    if sources:
-        st.subheader("📚 Sources")
-        for s in sources:
-            st.markdown(f"- [{s.get('title', 'Untitled')}]({s.get('url', '#')})")
+    st.subheader("📚 Sources")
+    for s in sources:
+        st.markdown(f"- [{s.get('title', 'Untitled')}]({s.get('url', '#')})")
 
-    # Display images
     if images:
-        st.subheader("🖼️ Related Images")
+        st.subheader("🖼️ Relevant Images")
         for img in images:
             st.image(img, use_column_width=True)
 
-    # Download button
-    st.download_button("💾 Download Report", report, file_name="deep_research.md", mime="text/markdown")
+    # --- Export Section ---
+    st.subheader("📤 Export Report")
+    file_data, filename, mime = export_file(report, export_format)
+    if file_data:
+        st.download_button(
+            label=f"💾 Download {export_format}",
+            data=file_data,
+            file_name=filename,
+            mime=mime
+        )