Update app.py

app.py

@@ -60,57 +60,17 @@ def remove_duplicate_paragraphs(text: str) -> str:
             seen.add(clean_p)
     return "\n\n".join(unique_paragraphs)
 
-def extract_text_from_excel(path: str) -> str:
-    all_text = []
-    xls = pd.ExcelFile(path)
-    for sheet_name in xls.sheet_names:
-        try:
-            df = xls.parse(sheet_name).astype(str).fillna("")
-        except Exception:
-            continue
-        for _, row in df.iterrows():
-            non_empty = [cell.strip() for cell in row if cell.strip()]
-            if len(non_empty) >= 2:
-                text_line = " | ".join(non_empty)
-                if len(text_line) > 15:
-                    all_text.append(f"[{sheet_name}] {text_line}")
-    return "\n".join(all_text)
-
-def extract_text_from_csv(path: str) -> str:
-    all_text = []
-    try:
-        df = pd.read_csv(path).astype(str).fillna("")
-    except Exception:
-        return ""
-    for _, row in df.iterrows():
-        non_empty = [cell.strip() for cell in row if cell.strip()]
-        if len(non_empty) >= 2:
-            text_line = " | ".join(non_empty)
-            if len(text_line) > 15:
-                all_text.append(text_line)
-    return "\n".join(all_text)
-
-def extract_text_from_pdf(path: str) -> str:
-    import logging
-    logging.getLogger("pdfminer").setLevel(logging.ERROR)
-    all_text = []
-    try:
-        with pdfplumber.open(path) as pdf:
-            for page in pdf.pages:
-                text = page.extract_text()
-                if text:
-                    all_text.append(text.strip())
-    except Exception:
-        return ""
-    return "\n".join(all_text)
-
 def extract_text(file_path: str) -> str:
     if file_path.endswith(".xlsx"):
-        return extract_text_from_excel(file_path)
+        return pd.read_excel(file_path).astype(str).fillna("").to_string(index=False)
     elif file_path.endswith(".csv"):
-        return extract_text_from_csv(file_path)
+        return pd.read_csv(file_path).astype(str).fillna("").to_string(index=False)
     elif file_path.endswith(".pdf"):
-        return extract_text_from_pdf(file_path)
+        try:
+            with pdfplumber.open(file_path) as pdf:
+                return "\n".join(page.extract_text() or '' for page in pdf.pages)
+        except Exception:
+            return ""
     else:
         return ""
 
@@ -136,6 +96,9 @@ def batch_chunks(chunks: List[str], batch_size: int = BATCH_SIZE) -> List[List[s
 def build_prompt(chunk: str) -> str:
     return f"""### Unstructured Clinical Records\n\nAnalyze the clinical notes below and summarize with:\n- Diagnostic Patterns\n- Medication Issues\n- Missed Opportunities\n- Inconsistencies\n- Follow-up Recommendations\n\n---\n\n{chunk}\n\n---\nRespond concisely in bullet points with clinical reasoning."""
 
+def remove_non_ascii(text):
+    return ''.join(c for c in text if ord(c) < 256)
+
 def init_agent() -> TxAgent:
     tool_path = os.path.join(tool_cache_dir, "new_tool.json")
     if not os.path.exists(tool_path):
@@ -224,189 +187,35 @@ Avoid repeating the same points multiple times.
     final_response = remove_duplicate_paragraphs(final_response)
     return final_response
 
-def remove_non_ascii(text):
-    return ''.join(c for c in text if ord(c) < 256)
-
-def generate_pdf_report_with_charts(summary: str, report_path: str, detailed_batches: List[str] = None):
-    chart_dir = os.path.join(os.path.dirname(report_path), "charts")
-    os.makedirs(chart_dir, exist_ok=True)
-
-    # Prepare data
-    categories = ['Diagnostics', 'Medications', 'Missed', 'Inconsistencies', 'Follow-up']
-    values = [4, 2, 3, 1, 5]
-
-    # Chart 1: Bar
-    bar_chart_path = os.path.join(chart_dir, "bar_chart.png")
-    plt.figure(figsize=(6, 4))
-    plt.bar(categories, values)
-    plt.title('Clinical Issues Overview')
-    plt.tight_layout()
-    plt.savefig(bar_chart_path)
-    plt.close()
-
-    # Chart 2: Pie
-    pie_chart_path = os.path.join(chart_dir, "pie_chart.png")
-    plt.figure(figsize=(6, 6))
-    plt.pie(values, labels=categories, autopct='%1.1f%%')
-    plt.title('Issue Distribution')
-    plt.tight_layout()
-    plt.savefig(pie_chart_path)
-    plt.close()
-
-    # Chart 3: Line
-    trend_chart_path = os.path.join(chart_dir, "trend_chart.png")
-    plt.figure(figsize=(6, 4))
-    plt.plot(categories, values, marker='o')
-    plt.title('Trend Analysis')
-    plt.tight_layout()
-    plt.savefig(trend_chart_path)
-    plt.close()
-
-    # PDF init
-    pdf_path = report_path.replace('.md', '.pdf')
-    pdf = FPDF()
-    pdf.set_auto_page_break(auto=True, margin=15)
-
-    # === Title Page ===
-    pdf.add_page()
-    pdf.set_font("Arial", 'B', 24)
-    pdf.cell(0, 20, remove_non_ascii("Final Medical Report"), ln=True, align='C')
-    pdf.set_font("Arial", '', 14)
-    pdf.cell(0, 10, datetime.now().strftime("Generated on %B %d, %Y at %H:%M"), ln=True, align='C')
-    pdf.ln(20)
-    pdf.set_font("Arial", 'I', 12)
-    pdf.multi_cell(0, 10, remove_non_ascii(
-        "This report contains a professional summary of clinical observations, potential inconsistencies, and follow-up recommendations based on the uploaded medical document."
-    ), align="C")
-
-    # === Summary Section ===
-    pdf.add_page()
-    pdf.set_font("Arial", 'B', 16)
-    pdf.cell(0, 10, remove_non_ascii("Final Summary"), ln=True)
-    pdf.set_draw_color(200, 200, 200)
-    pdf.line(10, pdf.get_y(), 200, pdf.get_y())
-    pdf.ln(5)
-    pdf.set_font("Arial", '', 12)
-    for line in summary.split("\n"):
-        clean_line = remove_non_ascii(line.strip())
-        if clean_line:
-            pdf.multi_cell(0, 8, txt=clean_line)
-
-    # === Charts Section ===
-    pdf.add_page()
-    pdf.set_font("Arial", 'B', 16)
-    pdf.cell(0, 10, remove_non_ascii("Statistical Overview"), ln=True)
-    pdf.line(10, pdf.get_y(), 200, pdf.get_y())
-    pdf.ln(5)
-
-    pdf.set_font("Arial", 'B', 12)
-    pdf.cell(0, 10, remove_non_ascii("1. Clinical Issues Overview"), ln=True)
-    pdf.image(bar_chart_path, w=180)
-    pdf.ln(5)
-
-    pdf.cell(0, 10, remove_non_ascii("2. Issue Distribution"), ln=True)
-    pdf.image(pie_chart_path, w=150)
-    pdf.ln(5)
-
-    pdf.cell(0, 10, remove_non_ascii("3. Trend Analysis"), ln=True)
-    pdf.image(trend_chart_path, w=180)
-
-    # === Detailed Tool Outputs ===
-    if detailed_batches:
-        pdf.add_page()
-        pdf.set_font("Arial", 'B', 16)
-        pdf.cell(0, 10, remove_non_ascii("Detailed Tool Insights"), ln=True)
-        pdf.line(10, pdf.get_y(), 200, pdf.get_y())
-        pdf.ln(5)
-
-        for idx, detail in enumerate(detailed_batches):
-            pdf.set_font("Arial", 'B', 13)
-            pdf.cell(0, 10, remove_non_ascii(f"Tool Output #{idx + 1}"), ln=True)
-            pdf.set_font("Arial", '', 11)
-            for line in remove_non_ascii(detail).split("\n"):
-                pdf.multi_cell(0, 8, txt=line.strip())
-            pdf.ln(3)
-
-    pdf.output(pdf_path)
-    return pdf_path
-
-def process_report(agent, file, messages: List[Dict[str, str]]) -> Tuple[List[Dict[str, str]], Union[str, None]]:
+def handle_analysis(file):
+    messages = []
     if not file or not hasattr(file, "name"):
-        messages.append({"role": "assistant", "content": "❌ Please upload a valid file."})
-        return messages, None
-
-    start_time = time.time()
-    messages.append({"role": "user", "content": f"📂 Processing file: {os.path.basename(file.name)}"})
-
+        return "❌ Please upload a valid file.", None
     try:
         extracted = extract_text(file.name)
         if not extracted:
-            messages.append({"role": "assistant", "content": "❌ Could not extract text."})
-            return messages, None
+            return "❌ Could not extract text.", None
 
         chunks = split_text(extracted)
         batches = batch_chunks(chunks, batch_size=BATCH_SIZE)
-        messages.append({"role": "assistant", "content": f"🔍 Split into {len(batches)} batches. Analyzing..."})
-
         batch_results = analyze_batches(agent, batches)
-        all_tool_outputs = batch_results.copy()
         valid = [res for res in batch_results if not res.startswith("❌")]
 
         if not valid:
-            messages.append({"role": "assistant", "content": "❌ No valid batch outputs."})
-            return messages, None
+            return "❌ No valid batch outputs.", None
 
         summary = generate_final_summary(agent, "\n\n".join(valid))
-
-        report_path = os.path.join(report_dir, f"report_{datetime.now().strftime('%Y%m%d_%H%M%S')}.md")
+        report_path = os.path.join(report_dir, f"report_{datetime.now().strftime('%Y%m%d_%H%M%S')}.txt")
         with open(report_path, 'w', encoding='utf-8') as f:
-            f.write(f"# Final Medical Report\n\n{summary}")
-
-        pdf_path = generate_pdf_report_with_charts(summary, report_path, detailed_batches=all_tool_outputs)
-
-        end_time = time.time()
-        elapsed_time = end_time - start_time
-
-        messages.append({"role": "assistant", "content": f"📊 **Final Report:**\n\n{summary}"})
-        messages.append({"role": "assistant", "content": f"✅ Report generated in **{elapsed_time:.2f} seconds**.\n\n📥 PDF report ready: {os.path.basename(pdf_path)}"})
-
-        return messages, pdf_path
-
+            f.write(summary)
+        return summary, report_path
     except Exception as e:
-        messages.append({"role": "assistant", "content": f"❌ Error: {str(e)}"})
-        return messages, None
-
-def create_ui(agent):
-    with gr.Blocks(css="""
-        html, body, .gradio-container { background: #0e1621; color: #e0e0e0; padding: 16px; }
-        button.svelte-1ipelgc { background: linear-gradient(to right, #1e88e5, #0d47a1) !important; border: 1px solid #0d47a1 !important; color: white !important; font-weight: bold !important; padding: 10px 20px !important; border-radius: 8px !important; }
-        button.svelte-1ipelgc:hover { background: linear-gradient(to right, #2196f3, #1565c0) !important; border: 1px solid #1565c0 !important; }
-        .gr-column { align-items: center !important; gap: 12px; }
-        .gr-file, .gr-button { width: 100% !important; max-width: 400px; }
-    """) as demo:
-        gr.Markdown("""
-        <h2 style='text-align:center;'>📄 CPS: Clinical Patient Support System</h2>
-        <p style='text-align:center;'>Analyze and summarize unstructured medical files using AI (optimized for A100 GPU).</p>
-        """)
-
-        with gr.Column():
-            chatbot = gr.Chatbot(label="🧠 CPS Assistant", height=480, type="messages")
-            upload = gr.File(label="📂 Upload Medical File", file_types=[".xlsx", ".csv", ".pdf"])
-            analyze = gr.Button("🧠 Analyze")
-            download = gr.File(label="📥 Download Report", visible=False, interactive=False)
-
-        state = gr.State(value=[])
-
-        def handle_analysis(file, chat):
-            messages, report_path = process_report(agent, file, chat)
-            return messages, gr.update(visible=bool(report_path), value=report_path), messages
-
-        analyze.click(fn=handle_analysis, inputs=[upload, state], outputs=[chatbot, download, state])
-
-    return demo
+        return f"❌ Error: {str(e)}", None
 
 if __name__ == "__main__":
     agent = init_agent()
-    ui = create_ui(agent)
-    ui.launch(server_name="0.0.0.0", server_port=7860, allowed_paths=["/data/hf_cache/reports"], share=False)
-        
+    gr.Interface(
+        fn=handle_analysis,
+        inputs=gr.File(file_types=[".pdf", ".csv", ".xlsx"]),
+        outputs=[gr.Textbox(label="Summary"), gr.File(label="Download Report")]
+    ).queue().launch(server_name="0.0.0.0", server_port=7860, allowed_paths=["/data/hf_cache/reports"], share=False)