Update app.py

app.py

@@ -7,7 +7,7 @@ import torch
 import numpy as np
 import time
 import datetime
-import traceback
+import urllib.parse # Para codificar la dirección en mapas
 # Imports para PDF Profesional (Platypus)
 from reportlab.lib.pagesizes import letter
 from reportlab.lib import colors
@@ -25,6 +25,7 @@ import requests
 import sqlite3
 import pandas as pd
 from dotenv import load_dotenv
+import glob
 
 load_dotenv()
 
@@ -65,6 +66,7 @@ def init_db():
             id TEXT PRIMARY KEY,
             date TEXT,
             user TEXT,
+            address TEXT,
             score INTEGER,
             duration TEXT,
             zones_count INTEGER,
@@ -73,19 +75,27 @@ def init_db():
             json_path TEXT
         )
     ''')
+    # Migración segura: verificar si existe la columna address
+    try:
+        c.execute("SELECT address FROM reports LIMIT 1")
+    except sqlite3.OperationalError:
+        try: c.execute("ALTER TABLE reports ADD COLUMN address TEXT")
+        except: pass
+    
     try: c.execute("ALTER TABLE reports ADD COLUMN json_path TEXT")
     except: pass
+    
     conn.commit()
     conn.close()
 
-def save_report_to_db(user, score, duration, zones_count, summary, gemini_status, json_path):
+def save_report_to_db(user, address, score, duration, zones_count, summary, gemini_status, json_path):
     try:
         report_id = str(uuid.uuid4())[:8]
         date_str = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
         conn = sqlite3.connect(DB_NAME)
         c = conn.cursor()
-        c.execute("INSERT INTO reports VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)",
-                  (report_id, date_str, str(user), score, duration, zones_count, summary, gemini_status, json_path))
+        c.execute("INSERT INTO reports VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
+                  (report_id, date_str, str(user), str(address), score, duration, zones_count, summary, gemini_status, json_path))
         conn.commit()
         conn.close()
         return report_id
@@ -96,13 +106,29 @@ def save_report_to_db(user, score, duration, zones_count, summary, gemini_status
 def get_history_df(request: gr.Request):
     try:
         conn = sqlite3.connect(DB_NAME)
-        df = pd.read_sql_query("SELECT date, user, score, duration, zones_count, summary, gemini_status FROM reports ORDER BY date DESC", conn)
+        df = pd.read_sql_query("SELECT date, user, address, score, duration, zones_count, summary, gemini_status FROM reports ORDER BY date DESC", conn)
         conn.close()
         return df
     except: return pd.DataFrame()
 
 init_db()
 
+# ==========================================
+# DIAGNÓSTICO
+# ==========================================
+def run_diagnostics():
+    key = os.getenv("GOOGLE_API_KEY")
+    status = []
+    status.append(f"CV2: {'✅' if OPENCV_AVAILABLE else '❌'}")
+    status.append(f"MoviePy: {'✅' if MOVIEPY_AVAILABLE else '❌'}")
+    if not key: 
+        status.append("Google Key: ❌ (Modo Local)")
+    else: 
+        status.append(f"Google Key: ✅ ({key[:4]}...)")
+    return " | ".join(status)
+
+DIAGNOSTIC_MSG = run_diagnostics()
+
 # ==========================================
 # CONFIGURACIÓN Y MODELOS
 # ==========================================
@@ -154,8 +180,7 @@ class SmartLoader:
         self.defect_pipe = None 
         self.asr_pipe = None
         self.sum_pipe = None 
-        # INICIALIZACIÓN EXPLÍCITA PARA EVITAR ERROR
-        self.sent_pipe = None  
+        self.sent_pipe = None 
         self.status = "Iniciando..."
         self.vision_active = False
 
@@ -165,8 +190,7 @@ class SmartLoader:
             self.zone_pipe = pipeline("image-classification", model=PRIMARY_ZONE)
             self.furn_pipe = pipeline("object-detection", model=PRIMARY_FURN)
             self.vision_active = True
-        except Exception as e:
-            print(f"⚠️ Error Visión Primaria: {e}")
+        except:
             print("⚠️ Usando Respaldo Visión...")
             try:
                 self.zone_pipe = pipeline("image-classification", model=BACKUP_ZONE)
@@ -183,7 +207,6 @@ class SmartLoader:
         try: self.sum_pipe = pipeline("summarization", model=ADVANCED_SUMMARY)
         except: pass
         
-        # Carga segura del modelo de sentimiento
         try: 
             self.sent_pipe = pipeline("text-classification", model=ADVANCED_SENTIMENT)
         except: 
@@ -212,23 +235,28 @@ def extract_json(text):
         return json.loads(match.group(0)) if match else None
     except: return None
 
-def call_gemini_analysis(api_key, inventory_data, transcription):
+def call_gemini_analysis(api_key, inventory_data, transcription, address):
     target_key = api_key or os.getenv("GOOGLE_API_KEY")
     if not target_key: return None
     target_key = target_key.strip()
     
-    summary_str = f"Transcripción: {transcription[:2000]}\n"
+    summary_str = f"Ubicación: {address}\nTranscripción: {transcription[:2000]}\n"
     for z, d in inventory_data.items():
         summary_str += f"- {z}: {len(d['detailed_items'])} items, Defectos: {[x['label'] for x in d['defects']]}\n"
 
+    # PROMPT POTENCIADO PARA RESPUESTA "SENIOR"
     prompt = f"""
-    Actúa como Perito Inmobiliario Senior. Analiza estos datos:
-    {summary_str}
+    Actúa como un **Perito Inmobiliario y Agente Comercial Senior**. Analiza estos datos técnicos del inmueble ubicado en {address}.
+    
+    Datos: {summary_str}
     
-    Genera JSON válido: {{
-        "resumen_ejecutivo": "Texto técnico detallado.",
+    Genera un JSON extendido y profesional con estos campos OBLIGATORIOS:
+    {{
+        "resumen_ejecutivo": "Informe técnico detallado del estado físico (mínimo 100 palabras).",
+        "descripcion_comercial": "Descripción atractiva para venta/alquiler destacando puntos fuertes y ubicación (mínimo 100 palabras).",
+        "conclusion_tecnica": "Veredicto final sobre la viabilidad y estado del inmueble.",
         "presupuesto_estimado": "Tabla markdown de costos.",
-        "habitabilidad": "SI/NO y por qué.",
+        "habitabilidad": "SI/NO y justificación técnica.",
         "valoracion": "Alta/Media/Bajo"
     }}
     """
@@ -243,11 +271,24 @@ def call_gemini_analysis(api_key, inventory_data, transcription):
 def chat_response(message, history, context, api_key):
     target_key = api_key or os.getenv("GOOGLE_API_KEY")
     if not target_key: return "⚠️ Error: Falta configurar GOOGLE_API_KEY."
+    
+    # Intento de recuperación de contexto si está vacío
+    if not context:
+        try:
+            list_of_files = glob.glob(os.path.join(OUTPUT_DIR, '*.json'))
+            if list_of_files:
+                latest_file = max(list_of_files, key=os.path.getctime)
+                with open(latest_file, 'r', encoding='utf-8') as f:
+                    data = json.load(f)
+                    context = f"Resumen: {data.get('summary', '')}\nDatos: {str(data.get('inventory', ''))}"
+        except: pass
+
     if not context: return "⚠️ Primero analiza un video para tener contexto."
+
     payload = {"contents": [{"parts": [{"text": f"Contexto: {context}\nUsuario: {message}"}]}]}
-    response = call_gemini_api(target_key.strip(), payload)
-    if isinstance(response, dict) and "candidates" in response:
-        return response['candidates'][0]['content']['parts'][0]['text']
+    resp = call_gemini_api(target_key.strip(), payload)
+    if isinstance(resp, dict) and "candidates" in resp:
+        return resp['candidates'][0]['content']['parts'][0]['text']
     return "Error Gemini"
 
 # ==========================================
@@ -266,7 +307,6 @@ def transcribe_audio(video_path):
     except: return "Error transcripción."
 
 def analyze_sentiment(text):
-    # ACCESO SEGURO A SENT_PIPE
     pipe = getattr(loader, 'sent_pipe', None)
     if not pipe or len(text) < 5: return "Neutro", 0.0
     try:
@@ -286,14 +326,16 @@ def generate_local_summary_verbose(transcription, inventory_data):
     total_defects = sum([len(d.get('defects', [])) for d in inventory_data.values()])
     zones_list = list(inventory_data.keys())
     
-    summary = f"INFORME TÉCNICO:\n\nSe ha realizado una inspección detallada. "
-    summary += f"Zonas: {', '.join(zones_list)}. "
-    summary += f"Total items: {total_items}. "
+    summary = f"INFORME TÉCNICO DE INSPECCIÓN:\n\n"
+    summary += f"Se ha realizado una inspección visual y auditiva detallada del inmueble. "
+    summary += f"El recorrido abarcó {len(zones_list)} zonas principales: {', '.join(zones_list)}. "
+    summary += f"El sistema de visión artificial inventarió un total de {total_items} elementos.\n\n"
     
     if total_defects > 0:
-        summary += f"ALERTA: {total_defects} anomalías detectadas. Ver plan de mantenimiento.\n"
+        summary += f"HALLAZGOS CRÍTICOS: Se detectaron {total_defects} anomalías que requieren atención. "
+        summary += "Se recomienda consultar la sección de 'Plan de Mantenimiento' para detalles de reparación.\n"
     else:
-        summary += "Estado general: BUENO.\n"
+        summary += "ESTADO GENERAL: El inmueble presenta buenas condiciones de conservación visual, sin patologías graves aparentes en las zonas inspeccionadas.\n"
         
     return summary
 
@@ -315,7 +357,21 @@ def clean_text_for_pdf(text):
     text = text.replace('⚠️', '[!]').replace('✅', '[OK]').replace('📍', '').replace('📦', '').replace('🛠️', '')
     return text.encode('latin-1', 'ignore').decode('latin-1')
 
-def create_pdf_report(inventory_data, transcription, local_summary, gemini_data, score_tuple, sentiment_tuple, duration, report_id):
+# --- GENERACIÓN DE CÓDIGO QR PARA MAPAS ---
+def generate_qr_map(address):
+    try:
+        # Generar enlace de Google Maps
+        maps_url = f"https://www.google.com/maps/search/?api=1&query={urllib.parse.quote(address)}"
+        qr = qrcode.QRCode(box_size=10, border=1)
+        qr.add_data(maps_url)
+        qr.make(fit=True)
+        img = qr.make_image(fill_color="black", back_color="white")
+        temp_qr = tempfile.mktemp(suffix=".png")
+        img.save(temp_qr)
+        return temp_qr
+    except: return None
+
+def create_pdf_report(inventory_data, transcription, local_summary, gemini_data, score_tuple, sentiment_tuple, duration, report_id, address):
     filename = f"ROPINS_Reporte_{report_id}.pdf"
     output_path = os.path.join(OUTPUT_DIR, filename)
 
@@ -339,14 +395,23 @@ def create_pdf_report(inventory_data, transcription, local_summary, gemini_data,
     prop_score = score_tuple[0] if isinstance(score_tuple, tuple) else score_tuple
     sent_label, _ = sentiment_tuple
     
-    # 1. PORTADA
-    story.append(Spacer(1, 2*inch))
+    # --- 1. PORTADA ---
+    story.append(Spacer(1, 1*inch))
     story.append(Paragraph("INFORME DE AUDITORÍA TÉCNICA INMOBILIARIA", styles['TitleProp']))
     story.append(Spacer(1, 0.5*inch))
-    story.append(Paragraph(f"ID Reporte: {report_id}", styles['SubtitleProp']))
-    story.append(Paragraph(f"Fecha de Inspección: {datetime.datetime.now().strftime('%d/%m/%Y')}", styles['SubtitleProp']))
-    story.append(Spacer(1, 2*inch))
+    story.append(Paragraph(f"Ubicación: {clean_text_for_pdf(address)}", styles['SubtitleProp']))
+    story.append(Paragraph(f"ID: {report_id} | Fecha: {datetime.datetime.now().strftime('%d/%m/%Y')}", styles['SubtitleProp']))
+    story.append(Spacer(1, 1*inch))
     
+    # QR de Ubicación
+    qr_path = generate_qr_map(address)
+    if qr_path:
+        im_qr = PlatypusImage(qr_path, width=2*inch, height=2*inch)
+        story.append(Paragraph("Escanea para ver Ubicación en Maps:", styles['SmallProp']))
+        story.append(im_qr)
+        story.append(Spacer(1, 1*inch))
+    
+    # Tabla Métricas
     data_metrics = [
         ["CALIFICACIÓN TÉCNICA", "DURACIÓN VISITA", "TONO DE VOZ"],
         [f"{prop_score}/100", duration, clean_text_for_pdf(sent_label)]
@@ -356,20 +421,35 @@ def create_pdf_report(inventory_data, transcription, local_summary, gemini_data,
         ('BACKGROUND', (0,0), (-1,0), colors.lightgrey),
         ('ALIGN', (0,0), (-1,-1), 'CENTER'),
         ('FONTNAME', (0,0), (-1,-1), 'Helvetica-Bold'),
+        ('FONTSIZE', (0,0), (-1,-1), 12),
         ('BOX', (0,0), (-1,-1), 1, colors.black),
         ('TEXTCOLOR', (0,1), (0,1), colors.green if prop_score > 70 else colors.orange)
     ]))
     story.append(t_metrics)
     story.append(PageBreak())
     
-    # 2. RESUMEN
-    story.append(Paragraph("1. RESUMEN EJECUTIVO", styles['HeadingProp']))
-    final_sum = local_summary
+    # 2. ANÁLISIS DETALLADO
+    story.append(Paragraph("1. ANÁLISIS TÉCNICO Y COMERCIAL", styles['HeadingProp']))
+    
+    resumen_text = local_summary
+    desc_comercial = "Pendiente de análisis comercial."
+    conclusion = "Pendiente de veredicto."
+    
     if gemini_data and isinstance(gemini_data, dict):
-        if "resumen_ejecutivo" in gemini_data:
-            final_sum = gemini_data['resumen_ejecutivo']
+        resumen_text = gemini_data.get('resumen_ejecutivo', resumen_text)
+        desc_comercial = gemini_data.get('descripcion_comercial', desc_comercial)
+        conclusion = gemini_data.get('conclusion_tecnica', conclusion)
             
-    story.append(Paragraph(clean_text_for_pdf(final_sum), styles['BodyProp']))
+    story.append(Paragraph("<b>Resumen Ejecutivo:</b>", styles['BodyProp']))
+    story.append(Paragraph(clean_text_for_pdf(resumen_text), styles['BodyProp']))
+    story.append(Spacer(1, 12))
+    
+    story.append(Paragraph("<b>Descripción Comercial (Venta/Renta):</b>", styles['BodyProp']))
+    story.append(Paragraph(clean_text_for_pdf(desc_comercial), styles['BodyProp']))
+    story.append(Spacer(1, 12))
+    
+    story.append(Paragraph("<b>Conclusión Técnica:</b>", styles['BodyProp']))
+    story.append(Paragraph(clean_text_for_pdf(conclusion), styles['BodyProp']))
     story.append(Spacer(1, 12))
     
     # 3. ZONAS
@@ -388,7 +468,7 @@ def create_pdf_report(inventory_data, transcription, local_summary, gemini_data,
     story.append(Spacer(1, 12))
 
     # 4. MANTENIMIENTO
-    story.append(Paragraph("3. PLAN DE MANTENIMIENTO", styles['HeadingProp']))
+    story.append(Paragraph("3. PLAN DE MANTENIMIENTO SUGERIDO", styles['HeadingProp']))
     budget_txt = "Sin requerimientos críticos."
     if gemini_data and isinstance(gemini_data, dict) and 'presupuesto_estimado' in gemini_data:
         budget_txt = str(gemini_data['presupuesto_estimado'])
@@ -402,7 +482,7 @@ def create_pdf_report(inventory_data, transcription, local_summary, gemini_data,
     story.append(PageBreak())
     
     # 5. DETALLE
-    story.append(Paragraph("4. INSPECCIÓN DETALLADA", styles['TitleProp']))
+    story.append(Paragraph("4. INSPECCIÓN DETALLADA POR ZONA", styles['TitleProp']))
     story.append(Spacer(1, 12))
     
     for zone, data in inventory_data.items():
@@ -423,7 +503,7 @@ def create_pdf_report(inventory_data, transcription, local_summary, gemini_data,
         defs = data.get('defects', [])
         if defs:
             d_str = ', '.join([d['label'] for d in defs])
-            story.append(Paragraph(f"<b>PATOLOGÍAS:</b> {clean_text_for_pdf(d_str)}", styles['RiskProp']))
+            story.append(Paragraph(f"<b>PATOLOGÍAS DETECTADAS:</b> {clean_text_for_pdf(d_str)}", styles['RiskProp']))
         
         items = data.get('detailed_items', [])
         if items:
@@ -487,126 +567,129 @@ def save_json(inventory, transcription, score, summary, report_id):
         json.dump(meta, f, indent=4, ensure_ascii=False)
     return path
 
-def process_full(video_path, api_key, request: gr.Request, progress=gr.Progress()):
-    # 1. Validaciones
-    if not OPENCV_AVAILABLE: return None, "❌ Error: OpenCV faltante.", "", ""
-    if not video_path: return None, "⚠️ Error: Falta video.", "", ""
+def process_full(video_path, address, api_key, request: gr.Request, progress=gr.Progress()):
+    if not OPENCV_AVAILABLE: return None, "❌ Error OpenCV", "", ""
+    if not video_path: return None, "⚠️ Sube un video", "", ""
+    if not address: address = "Sin dirección especificada"
+    
+    user = request.username if request else "anonimo"
+    report_id = str(uuid.uuid4())[:6].upper()
     
-    # 2. Transcripción y Audio
+    # 1. Audio
     progress(0.1, desc="Audio...")
     transcription = transcribe_audio(video_path)
     
-    # 3. Visión
-    try:
-        cap = cv2.VideoCapture(video_path)
-        fps = cap.get(cv2.CAP_PROP_FPS) or 24.0
-        frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
-        duration = frames/fps
-        dur_str = time.strftime('%H:%M:%S', time.gmtime(duration))
-        step = int(fps * 4.0)
-        
-        inventory = {}
-        curr = 0
-        
-        if loader.vision_active:
-            progress(0.3, desc="Visión...")
-            while True:
-                ret, frame = cap.read()
-                if not ret: break
-                if curr % step == 0:
-                    pil = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
-                    if pil.width > 800: pil.thumbnail((800, 800))
-                    
-                    try: 
-                        z_res = loader.zone_pipe(pil)[0]
-                        zone = translate_label(z_res['label'].split(':')[-1].strip())
-                    except: zone = "General"
-                    
-                    if zone not in inventory: 
-                        inventory[zone] = {"detailed_items": [], "defects": [], "materials": [], "image": None, "score": 0}
-                    
-                    if not inventory[zone]["materials"] and loader.mat_pipe:
-                        try:
-                            m = loader.mat_pipe(pil)
-                            inventory[zone]["materials"] = [{"label": translate_label(m[0]['label'])}]
-                        except: pass
-                    
-                    if loader.defect_pipe:
-                        try:
-                            res = loader.defect_pipe(pil, candidate_labels=list(DEFECT_SOLUTIONS_LOCAL.keys()), threshold=0.15)
-                            for d in res:
-                                lbl = translate_label(d['label'])
-                                box = [int(v) for v in d['box'].values()]
-                                if not any(x['label']==lbl for x in inventory[zone]['defects']):
-                                    inventory[zone]['defects'].append({"label": lbl, "score": d['score']})
-                        except: pass
-                    
-                    if loader.furn_pipe:
-                        try:
-                            res = loader.furn_pipe(pil)
-                            for f in res:
-                                if f['score'] > 0.6:
-                                    lbl = translate_label(f['label'])
-                                    box = [int(v) for v in f['box'].values()]
-                                    
-                                    sbox = [max(0, box[0]), max(0, box[1]), min(pil.width, box[2]), min(pil.height, box[3])]
-                                    crop = pil.crop(sbox)
-                                    crop.thumbnail((100, 100))
-                                    inventory[zone]['detailed_items'].append({"label": lbl, "crop": crop, "score": f['score']})
-                        except: pass
-                    
-                    score_frame = len(inventory[zone]['detailed_items']) + len(inventory[zone]['defects'])
-                    if score_frame > len(inventory[zone].get('detailed_items', [])) or not inventory[zone]['image']:
-                        inventory[zone]['image'] = annotate_image(pil.copy(), [], [])
-                curr += 1
-        else:
-            # Fallback
+    # 2. Video
+    cap = cv2.VideoCapture(video_path)
+    fps = cap.get(cv2.CAP_PROP_FPS) or 24.0
+    duration = int(cap.get(cv2.CAP_PROP_FRAME_COUNT)) / fps
+    dur_str = time.strftime('%H:%M:%S', time.gmtime(duration))
+    step = int(fps * 4.0) 
+    
+    inventory = {}
+    curr = 0
+    
+    if loader.vision_active:
+        progress(0.3, desc="Visión...")
+        while True:
             ret, frame = cap.read()
-            if ret:
+            if not ret: break
+            if curr % step == 0:
                 pil = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
-                inventory["Video General"] = {"detailed_items": [], "defects": [], "materials": [], "image": pil}
-        cap.release()
+                if pil.width > 800: pil.thumbnail((800, 800))
+                
+                try: 
+                    z_res = loader.zone_pipe(pil)[0]
+                    zone = translate_label(z_res['label'].split(':')[-1].strip())
+                except: zone = "General"
+                
+                if zone not in inventory: 
+                    inventory[zone] = {"detailed_items": [], "defects": [], "materials": [], "image": None, "score": 0}
+                
+                if not inventory[zone]["materials"] and loader.mat_pipe:
+                    try:
+                        m = loader.mat_pipe(pil)
+                        inventory[zone]["materials"] = [{"label": translate_label(m[0]['label'])}]
+                    except: pass
+                
+                if loader.defect_pipe:
+                    try:
+                        res = loader.defect_pipe(pil, candidate_labels=list(DEFECT_SOLUTIONS_LOCAL.keys()), threshold=0.15)
+                        for d in res:
+                            lbl = translate_label(d['label'])
+                            box = [int(v) for v in d['box'].values()]
+                            if not any(x['label']==lbl for x in inventory[zone]['defects']):
+                                inventory[zone]['defects'].append({"label": lbl, "score": d['score']})
+                    except: pass
+                
+                if loader.furn_pipe:
+                    try:
+                        res = loader.furn_pipe(pil)
+                        for f in res:
+                            if f['score'] > 0.6:
+                                lbl = translate_label(f['label'])
+                                box = [int(v) for v in f['box'].values()]
+                                sbox = [max(0, box[0]), max(0, box[1]), min(pil.width, box[2]), min(pil.height, box[3])]
+                                crop = pil.crop(sbox)
+                                crop.thumbnail((100, 100))
+                                inventory[zone]['detailed_items'].append({"label": lbl, "crop": crop, "score": f['score']})
+                    except: pass
+                
+                score = len(inventory[zone]['detailed_items']) + len(inventory[zone]['defects'])*2
+                if score > inventory[zone]['score'] or not inventory[zone]['image']:
+                    inventory[zone]['image'] = annotate_image(pil.copy(), [], [])
+                    inventory[zone]['score'] = score
+            curr += 1
+    else:
+        ret, frame = cap.read()
+        if ret:
+            pil = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
+            inventory["Video General"] = {"detailed_items": [], "defects": [], "materials": [], "image": pil, "score": 0}
+    cap.release()
     
-    except Exception as e:
-        return None, f"❌ Error Procesamiento: {str(e)}\n{traceback.format_exc()}", "", ""
-
-    # 4. Reporte
-    progress(0.8, desc="Generando...")
-    try:
-        score_val = calculate_score(inventory)
-        sent_val = analyze_sentiment(transcription)
-        
-        local_summary = generate_local_summary_verbose(transcription, inventory)
-        gemini_key = api_key or os.getenv("GOOGLE_API_KEY")
-        gemini_data = call_gemini_analysis(gemini_key, inventory, transcription) if gemini_key else None
-        
-        pdf_path = create_pdf_report(inventory, transcription, local_summary, gemini_data, (score_val, ""), sent_val, dur_str, report_id)
-        
-        final_sum = gemini_data.get('resumen_ejecutivo', local_summary) if gemini_data and "resumen_ejecutivo" in gemini_data else local_summary
-        
-        json_path = save_json(inventory, transcription, score_val, final_sum, report_id)
-        
-        gemini_status = "Online" if gemini_data else "Offline"
-        
-        user_name = getattr(request, 'username', 'anonimo') if request else 'anonimo'
-        save_report_to_db(user_name, score_val, dur_str, len(inventory), final_sum[:100], gemini_status, json_path)
-        
-        html = f"""
-        <div class="result-card">
-            <div class="result-header">
-                <h3 style="margin:0; color:#60a5fa;">✅ Análisis Completado</h3>
-                <span class="score-badge">Score: {score_val}/100</span>
-            </div>
-            <p><b>IA:</b> {gemini_status}</p>
-            <p><i>"{final_sum[:300]}..."</i></p>
+    progress(0.8, desc="Reporte...")
+    score_val = calculate_score(inventory)
+    sent_val = analyze_sentiment(transcription)
+    
+    local_summary = generate_local_summary_verbose(transcription, inventory)
+    gemini_key = api_key or os.getenv("GOOGLE_API_KEY")
+    gemini_data = call_gemini_analysis(gemini_key, inventory, transcription, address) if gemini_key else None
+    
+    pdf_path = create_pdf_report(inventory, transcription, local_summary, gemini_data, (score_val, ""), sent_val, dur_str, report_id, address)
+    final_sum = gemini_data.get('resumen_ejecutivo', local_summary) if gemini_data else local_summary
+    json_path = save_json(inventory, transcription, score_val, final_sum, report_id)
+    
+    gemini_status = "Online" if gemini_data else "Offline"
+    save_report_to_db(user, address, score_val, dur_str, len(inventory), final_sum[:100], gemini_status, json_path)
+    
+    # MAPA HTML
+    encoded_addr = urllib.parse.quote(address)
+    map_html = f'<iframe width="100%" height="250" src="https://maps.google.com/maps?q={encoded_addr}&t=&z=15&ie=UTF8&iwloc=&output=embed" frameborder="0" scrolling="no" marginheight="0" marginwidth="0"></iframe>'
+    
+    html = f"""
+    <div class="result-card">
+        <div class="result-header">
+            <h3 style="margin:0; color:#60a5fa;">✅ Análisis Completado</h3>
+            <span class="score-badge">Score: {score_val}/100</span>
         </div>
-        """
         
-        context = f"Resumen: {final_sum}\nInventario: {str(inventory)}"
-        return [pdf_path, json_path], html, f"Log: {loader.status}", context
+        {map_html}
+        <br>
         
-    except Exception as e:
-        return None, f"❌ Error Generación: {str(e)}\n{traceback.format_exc()}", "", ""
+        <div style="background-color: #111827; padding: 15px; border-radius: 8px; margin-bottom: 20px;">
+            <strong style="color:#60a5fa;">📝 Resumen:</strong>
+            <p style="margin-top:5px; font-style: italic; color:#d1d5db;">"{final_sum[:500]}..."</p>
+        </div>
+        <h4 style="color:#93c5fd;">Resumen de Zonas:</h4>
+        <ul style="color:#e5e7eb;">
+    """
+    for z in inventory.keys():
+        n_items = len(inventory[z]['detailed_items'])
+        html += f"<li><b>{z}:</b> {n_items} elementos identificados.</li>"
+    html += "</ul></div>"
+    
+    context = f"Resumen: {final_sum}\nInventario: {str(inventory)}"
+    return [pdf_path, json_path], html, f"Log: {loader.status}", context
 
 def get_diagnostic_msg():
     key = os.getenv("GOOGLE_API_KEY")
@@ -635,20 +718,21 @@ with gr.Blocks(theme=gr.themes.Soft(primary_hue="blue", neutral_hue="slate"), cs
         with gr.Tab("📹 Análisis"):
             with gr.Row():
                 with gr.Column(scale=4):
+                    address_in = gr.Textbox(label="Dirección del Inmueble (Para Mapa y Reporte)", placeholder="Ej. Av. Calle 26 # 50-20, Bogotá")
                     vid = gr.Video(label="Video", format="mp4")
                     btn = gr.Button("🚀 Generar Informe", variant="primary")
                 with gr.Column(scale=6):
                     files = gr.File(label="Descargas (PDF + JSON)", file_count="multiple", interactive=False)
                     sts = gr.HTML()
             log = gr.Textbox(visible=False)
-            btn.click(process_full, [vid, api_in], [files, sts, log, state])
+            btn.click(process_full, [vid, address_in, api_in], [files, sts, log, state])
 
         with gr.Tab("🤖 Chat"):
             gr.ChatInterface(fn=chat_response, additional_inputs=[state, api_in], title="Asistente ROPINS")
             
         with gr.Tab("📜 Historial"):
             ref = gr.Button("🔄 Actualizar")
-            tbl = gr.Dataframe(headers=["Fecha", "Usuario", "Score", "Duración", "Zonas", "Resumen", "IA"], label="Registros")
+            tbl = gr.Dataframe(headers=["Fecha", "Usuario", "Ubicación", "Score", "Duración", "Zonas", "Resumen", "IA"], label="Registros")
             ref.click(get_history_df, outputs=tbl)
 
 if __name__ == "__main__":