Update app.py

app.py

@@ -3,6 +3,12 @@ import requests
 import urllib.parse
 import re
 import xmltodict
+from collections import Counter
+import unicodedata
+
+# Función para normalizar palabras clave (elimina tildes y convierte a minúsculas)
+def normalize_keyword(keyword):
+    return ''.join(c for c in unicodedata.normalize('NFD', keyword.lower()) if unicodedata.category(c) != 'Mn')
 
 # Función para obtener sugerencias de DuckDuckGo
 def fetch_duckduckgo_suggestions(query, lang_code="es"):
@@ -12,7 +18,7 @@ def fetch_duckduckgo_suggestions(query, lang_code="es"):
     if response.status_code == 200:
         try:
             data = response.json()
-            return [item['phrase'] for item in data]
+            return [normalize_keyword(item['phrase']) for item in data]
         except ValueError:
             print("Error decodificando JSON de DuckDuckGo")
             return []
@@ -26,7 +32,7 @@ def fetch_google_suggestions(query, lang_code="es"):
     response = requests.get(url)
     if response.status_code == 200:
         try:
-            return response.json()[1]
+            return [normalize_keyword(suggestion) for suggestion in response.json()[1]]
         except ValueError:
             print("Error decodificando JSON de Google")
             return []
@@ -38,16 +44,12 @@ def fetch_youtube_suggestions(query, lang_code="es"):
     encoded_query = urllib.parse.quote(query)
     url = f"http://suggestqueries.google.com/complete/search?client=youtube&hl={lang_code}&q={encoded_query}"
     response = requests.get(url)
-    
     if response.status_code == 200:
         try:
-            # Extraer las sugerencias del formato window.google.ac.h(["..."])
             match = re.search(r'window\.google\.ac\.h\(\["[^"]*",\[(.*?)\],', response.text)
             if match:
                 suggestions_data = match.group(1)
-                # Extraemos las sugerencias
-                suggestions = re.findall(r'\["([^"]+)"', suggestions_data)
-                return suggestions
+                return [normalize_keyword(suggestion) for suggestion in re.findall(r'\["([^"]+)"', suggestions_data)]
             else:
                 print("No se encontraron sugerencias en el formato esperado.")
                 return []
@@ -60,27 +62,17 @@ def fetch_youtube_suggestions(query, lang_code="es"):
 # Función para obtener sugerencias de Bing
 def fetch_bing_suggestions(query, market="en-US"):
     url = "https://api.bing.com/qsml.aspx"
-    params = {
-        "Market": market,
-        "query": query
-    }
-    headers = {
-        "User-agent": "Mozilla/5.0"
-    }
+    params = {"Market": market, "query": query}
+    headers = {"User-agent": "Mozilla/5.0"}
     response = requests.get(url, params=params, headers=headers)
-    
     if response.status_code == 200:
         try:
             obj = xmltodict.parse(response.content)
-            suggestList = []
-            if 'SearchSuggestion' in obj and obj['SearchSuggestion']['Section']:
-                suggestions = obj['SearchSuggestion']['Section']['Item']
-                if isinstance(suggestions, list):
-                    for s in suggestions:
-                        suggestList.append(s['Text'])
-                elif isinstance(suggestions, dict):
-                    suggestList.append(suggestions['Text'])
-            return suggestList
+            suggestions = obj['SearchSuggestion']['Section']['Item']
+            if isinstance(suggestions, list):
+                return [normalize_keyword(s['Text']) for s in suggestions]
+            elif isinstance(suggestions, dict):
+                return [normalize_keyword(suggestions['Text'])]
         except Exception as e:
             print(f"Error procesando la respuesta de Bing: {e}")
             return []
@@ -90,67 +82,27 @@ def fetch_bing_suggestions(query, market="en-US"):
 # Función para obtener sugerencias de Amazon
 def fetch_amazon_suggestions(query, market_id="ATVPDKIKX0DER", alias="aps"):
     url = "https://completion.amazon.com/api/2017/suggestions"
-    params = {
-        "mid": market_id,
-        "alias": alias,
-        "prefix": query
-    }
+    params = {"mid": market_id, "alias": alias, "prefix": query}
     response = requests.get(url, params=params)
-    
     if response.status_code == 200:
         try:
             data = response.json()
-            # Extraemos las sugerencias desde el JSON bajo la clave 'suggestions'
-            return [item['value'] for item in data.get('suggestions', [])]
+            return [normalize_keyword(item['value']) for item in data.get('suggestions', [])]
         except ValueError:
             print("Error decodificando JSON de Amazon")
             return []
     else:
         return []
 
-# Lista de stopwords, artículos, términos de comparativas, preposiciones, pronombres y conjunciones comunes en español
-stopwords = [
-    # Artículos
-    "el", "la", "los", "las", "un", "una", "unos", "unas", 
-    
-    # Conjunciones y preposiciones
-    "y", "o", "de", "por", "para", "con", "sin", "a", "en", 
-    "sobre", "desde", "hacia", "hasta", "entre", "contra", 
-    "bajo", "según", "durante", "ante", "tras", "mediante", 
-    
-    # Términos comunes en comparativas y compras
-    "vs", "review", "opiniones", "opinion", "caracteristicas", 
-    "amazon", "comparativa", "precio", "comprar", "mejor", 
-    "especificaciones", "barato", "oferta", "descuento", 
-    "envio", "gratis", "analisis", "producto", "gama", 
-    "guía", "pros", "contras", "calidad", "rendimiento", 
-    "valoracion", "reseña", "nuevo", "usado", "rebaja", 
-    "disponible", "más", "menos", "mejor", "peor", "versus", 
-    "tamaño", "peso", "marca", "modelo", "compatible", "accesorios",
-    
-    # Otros términos relevantes
-    "novedades", 
-    "actualización", "diferencias", "similitudes", "competencia", 
-    "comparar", "top", "ranking", "valor", "rendimiento", 
-    "durabilidad", "material", "acabado", "resistente"
-]
-
-# Función para expandir la palabra clave añadiendo stopwords, artículos y conjunciones
-def expand_keyword(keyword):
-    expanded_keywords = [keyword]
-    for letter in 'abcdefghijklmnopqrstuvwxyz*_':
-        expanded_keywords.append(keyword + " " + letter)
-        expanded_keywords.append(letter + " " + keyword)
-    # Añadir combinaciones con stopwords, artículos y conjunciones
-    for stopword in stopwords:
-        expanded_keywords.append(f"{stopword} {keyword}")
-        expanded_keywords.append(f"{keyword} {stopword}")
-    return expanded_keywords
+# Función para contar y obtener las top N sugerencias más repetidas
+def get_top_suggestions(suggestions, top_n=3):
+    suggestion_counter = Counter(suggestions)
+    return suggestion_counter.most_common(top_n)
 
 # Función principal actualizada
 def main(keyword):
     expanded_keywords = expand_keyword(keyword)
-    all_suggestions = {}
+    all_suggestions = []
     google_suggestions_all = []
     duckduckgo_suggestions_all = []
     youtube_suggestions_all = []
@@ -160,128 +112,58 @@ def main(keyword):
     # Obtener sugerencias de DuckDuckGo
     for exp_keyword in expanded_keywords:
         suggestions = fetch_duckduckgo_suggestions(exp_keyword)
-        duckduckgo_suggestions_all.extend(suggestions)  # Agregar todas las sugerencias
-        for suggestion in suggestions:
-            if suggestion in all_suggestions:
-                all_suggestions[suggestion] += 1
-            else:
-                all_suggestions[suggestion] = 1
+        duckduckgo_suggestions_all.extend(suggestions)
+        all_suggestions.extend(suggestions)
 
     # Obtener sugerencias de Google
     for exp_keyword in expanded_keywords:
         suggestions = fetch_google_suggestions(exp_keyword)
-        google_suggestions_all.extend(suggestions)  # Agregar todas las sugerencias
-        for suggestion in suggestions:
-            if suggestion in all_suggestions:
-                all_suggestions[suggestion] += 1
-            else:
-                all_suggestions[suggestion] = 1
+        google_suggestions_all.extend(suggestions)
+        all_suggestions.extend(suggestions)
 
     # Obtener sugerencias de YouTube
     for exp_keyword in expanded_keywords:
         suggestions = fetch_youtube_suggestions(exp_keyword)
-        youtube_suggestions_all.extend(suggestions)  # Agregar todas las sugerencias
-        for suggestion in suggestions:
-            if suggestion in all_suggestions:
-                all_suggestions[suggestion] += 1
-            else:
-                all_suggestions[suggestion] = 1
+        youtube_suggestions_all.extend(suggestions)
+        all_suggestions.extend(suggestions)
 
     # Obtener sugerencias de Bing
     for exp_keyword in expanded_keywords:
         suggestions = fetch_bing_suggestions(exp_keyword)
         bing_suggestions_all.extend(suggestions)
-        for suggestion in suggestions:
-            if suggestion in all_suggestions:
-                all_suggestions[suggestion] += 1
-            else:
-                all_suggestions[suggestion] = 1
+        all_suggestions.extend(suggestions)
 
     # Obtener sugerencias de Amazon
     for exp_keyword in expanded_keywords:
         suggestions = fetch_amazon_suggestions(exp_keyword)
         amazon_suggestions_all.extend(suggestions)
-        for suggestion in suggestions:
-            if suggestion in all_suggestions:
-                all_suggestions[suggestion] += 1
-            else:
-                all_suggestions[suggestion] = 1
-
-    # Filtrar las top 3 de cada plataforma con su número de repeticiones
-    google_top_3 = list(set(google_suggestions_all))[:3]
-    duckduckgo_top_3 = list(set(duckduckgo_suggestions_all))[:3]
-    youtube_top_3 = list(set(youtube_suggestions_all))[:3]
-    bing_top_3 = list(set(bing_suggestions_all))[:3]
-    amazon_top_3 = list(set(amazon_suggestions_all))[:3]
-
-    # Ordenar y filtrar las sugerencias más frecuentes combinadas
-    sorted_suggestions = sorted(all_suggestions.items(), key=lambda item: item[1], reverse=True)
-    combined_top_10_suggestions = [sug for sug, freq in sorted_suggestions if freq >= 2][:10]
-    suggestions_str = ", ".join(combined_top_10_suggestions)
-
-    # Crear la lista de todas las palabras clave con su número de repeticiones
-    all_suggestions_str = "<ul>"
-    for suggestion, freq in sorted_suggestions:
-        all_suggestions_str += f"<li>{suggestion} - {freq} repeticiones</li>"
-    all_suggestions_str += "</ul>"
-
-    # Crear el HTML de salida con un botón de copia, pero mostrando solo el Top 3
-    html_output = f"""
-    <div>
-        <b>Sugerencias combinadas de Google, DuckDuckGo, YouTube, Bing y Amazon (Top 10 combinadas):</b> <span id='suggestions_text'>{suggestions_str}</span>
-        <button class="lg secondary svelte-cmf5ev" style="font-size: small; padding: 2px; color: #808080ba; border: none; margin-left: 5px;" 
-        onclick='navigator.clipboard.writeText(document.getElementById("suggestions_text").innerText).then(() => alert("Texto copiado al portapapeles"))'>&nbsp;✂&nbsp;</button>
-    </div>
-    <h4>Top 3 Sugerencias de Google:</h4>
-    <ul>
-    """
-    for suggestion in google_top_3:
-        freq = all_suggestions[suggestion]
+        all_suggestions.extend(suggestions)
+
+    # Obtener las top 3 sugerencias de cada plataforma
+    google_top_3 = get_top_suggestions(google_suggestions_all, top_n=3)
+    duckduckgo_top_3 = get_top_suggestions(duckduckgo_suggestions_all, top_n=3)
+    youtube_top_3 = get_top_suggestions(youtube_suggestions_all, top_n=3)
+    bing_top_3 = get_top_suggestions(bing_suggestions_all, top_n=3)
+    amazon_top_3 = get_top_suggestions(amazon_suggestions_all, top_n=3)
+
+    # Mostrar los resultados
+    html_output = "<h4>Top 3 Sugerencias de Google:</h4><ul>"
+    for suggestion, freq in google_top_3:
         html_output += f"<li>{suggestion} ({freq})</li>"
-    html_output += "</ul>"
-
-    html_output += """
-    <h4>Top 3 Sugerencias de DuckDuckGo:</h4>
-    <ul>
-    """
-    for suggestion in duckduckgo_top_3:
-        freq = all_suggestions[suggestion]
+    html_output += "</ul><h4>Top 3 Sugerencias de DuckDuckGo:</h4><ul>"
+    for suggestion, freq in duckduckgo_top_3:
         html_output += f"<li>{suggestion} ({freq})</li>"
-    html_output += "</ul>"
-
-    html_output += """
-    <h4>Top 3 Sugerencias de YouTube:</h4>
-    <ul>
-    """
-    for suggestion in youtube_top_3:
-        freq = all_suggestions[suggestion]
+    html_output += "</ul><h4>Top 3 Sugerencias de YouTube:</h4><ul>"
+    for suggestion, freq in youtube_top_3:
         html_output += f"<li>{suggestion} ({freq})</li>"
-    html_output += "</ul>"
-
-    html_output += """
-    <h4>Top 3 Sugerencias de Bing:</h4>
-    <ul>
-    """
-    for suggestion in bing_top_3:
-        freq = all_suggestions[suggestion]
+    html_output += "</ul><h4>Top 3 Sugerencias de Bing:</h4><ul>"
+    for suggestion, freq in bing_top_3:
         html_output += f"<li>{suggestion} ({freq})</li>"
-    html_output += "</ul>"
-
-    html_output += """
-    <h4>Top 3 Sugerencias de Amazon:</h4>
-    <ul>
-    """
-    for suggestion in amazon_top_3:
-        freq = all_suggestions[suggestion]
+    html_output += "</ul><h4>Top 3 Sugerencias de Amazon:</h4><ul>"
+    for suggestion, freq in amazon_top_3:
         html_output += f"<li>{suggestion} ({freq})</li>"
     html_output += "</ul>"
 
-    # Agregar la lista completa de todas las palabras clave
-    html_output += """
-    <h4>Lista completa de palabras clave con su número de repeticiones:</h4>
-    """
-    html_output += all_suggestions_str
-
     return html_output
 
 # Interfaz de Gradio
@@ -289,9 +171,9 @@ iface = gr.Interface(
     fn=main,
     inputs="text",
     outputs="html",
-    title="<div style='margin:0 auto;text-align:center'><div style='margin:0 auto;text-align:center'><img style='width:100px;display: inline-table;margin-bottom:-10px' src='https://artxeweb.com/media/files/search.jpg'><p>Sugerencias Combinadas de Google, DuckDuckGo, YouTube, Bing y Amazon</p></div>",
-    description="<p style='margin-bottom:10px;text-align:center;background: #ffffff; padding: 8px; border-radius: 8px; border-width: 1px; border: solid 1px #e5e7eb;'>Ingrese una palabra clave para obtener sugerencias de búsqueda relacionadas de Google, DuckDuckGo, YouTube, Bing y Amazon. Se mostrarán las 3 primeras sugerencias combinadas y también las 3 principales de cada plataforma por separado.</p>",
-    article="<div style='margin-top:10px'><p style='text-align: center !important; background: #ffffff; padding: 5px 30px; border-radius: 8px; border-width: 1px; border: solid 1px #e5e7eb; width: fit-content; margin: auto;'>Desarrollada por <a style='text-decoration: none !important; color: #e12a31 !important;' href='https://artxeweb.com'>© Artxe Web</a></p></div>"
+    title="Sugerencias Combinadas de Google, DuckDuckGo, YouTube, Bing y Amazon",
+    description="Ingrese una palabra clave para obtener sugerencias de búsqueda relacionadas.",
+    article="<p>Desarrollada por © Artxe Web</p>"
 )
 
-iface.launch()
+iface.launch()