Update app.py

app.py

@@ -35,16 +35,19 @@ def identificar_urls_comunes(serps):
     urls_comunes = {url for url in all_urls if all_urls.count(url) > 1}
     return urls_comunes
 
-def calcular_porcentajes_similaridad(serp_results):
-    urls_por_keyword = [set(entry["URL"] for entry in serp) for serp in serp_results.values()]
+def calcular_porcentajes_similaridad(serp_results, min_length):
     porcentajes = {}
     for combo in combinations(serp_results.keys(), 2):
-        intersection = len(urls_por_keyword[list(serp_results.keys()).index(combo[0])] & urls_por_keyword[list(serp_results.keys()).index(combo[1])])
-        union = len(urls_por_keyword[list(serp_results.keys()).index(combo[0])] | urls_por_keyword[list(serp_results.keys()).index(combo[1])])
+        # Asegurarse de considerar solo hasta min_length resultados para cada keyword
+        urls_serp1 = set(entry["URL"] for entry in serp_results[combo[0]][:min_length])
+        urls_serp2 = set(entry["URL"] for entry in serp_results[combo[1]][:min_length])
+        intersection = len(urls_serp1 & urls_serp2)
+        union = len(urls_serp1 | urls_serp2)
         porcentaje = (intersection / union) * 100 if union != 0 else 0
         porcentajes[combo] = porcentaje
     return porcentajes
 
+
 def generar_html_con_colores(serp_results, urls_comunes):
     url_color_map = {url: color_palette[i % len(color_palette)] for i, url in enumerate(urls_comunes)}
     
@@ -74,13 +77,16 @@ def generar_html_con_colores(serp_results, urls_comunes):
 def analyze_keywords(keywords):
     keywords_list = [keyword.strip() for keyword in keywords.split(',')]
     serp_results = {keyword: buscar_google(keyword) for keyword in keywords_list}
+
+    # Determinar el mínimo número de resultados devueltos entre todas las búsquedas
+    min_length = min(len(serp) for serp in serp_results.values())
     
-    urls_por_keyword = [[entry["URL"] for entry in serp] for serp in serp_results.values()]
+    urls_por_keyword = [[entry["URL"] for entry in serp][:min_length] for serp in serp_results.values()]
     urls_comunes = identificar_urls_comunes(urls_por_keyword)
     
-    html_table = generar_html_con_colores(serp_results, urls_comunes)
+    html_table = generar_html_con_colores(serp_results, urls_comunes, min_length)  # Asegúrate de ajustar esta función para usar min_length
 
-    porcentajes_similaridad = calcular_porcentajes_similaridad(serp_results)
+    porcentajes_similaridad = calcular_porcentajes_similaridad(serp_results, min_length)
     similaridad_html = "<div><strong>Porcentajes de Similaridad:</strong><br>"
     for combo, porcentaje in porcentajes_similaridad.items():
         similaridad_html += f"{combo[0]} & {combo[1]}: {porcentaje:.2f}%<br>"