Update app.py

app.py

@@ -1,161 +1,257 @@
 import gradio as gr
-import aiohttp
-import asyncio
+import requests
 import urllib.parse
 import re
 import xmltodict
 
-# Función asíncrona para obtener sugerencias de DuckDuckGo
-async def fetch_duckduckgo_suggestions(session, query, lang_code="es"):
+# Función para obtener sugerencias de DuckDuckGo
+def fetch_duckduckgo_suggestions(query, lang_code="es"):
     encoded_query = urllib.parse.quote(query)
     url = f"https://duckduckgo.com/ac/?q={encoded_query}&kl={lang_code}"
-    async with session.get(url) as response:
-        if response.status == 200:
-            try:
-                data = await response.json()
-                return [item['phrase'] for item in data]
-            except ValueError:
-                return []
+    response = requests.get(url)
+    if response.status_code == 200:
+        try:
+            data = response.json()
+            return [item['phrase'] for item in data]
+        except ValueError:
+            print("Error decodificando JSON de DuckDuckGo")
+            return []
+    else:
         return []
 
-# Función asíncrona para obtener sugerencias de Google
-async def fetch_google_suggestions(session, query, lang_code="es"):
+# Función para obtener sugerencias de Google
+def fetch_google_suggestions(query, lang_code="es"):
     encoded_query = urllib.parse.quote(query)
     url = f"http://suggestqueries.google.com/complete/search?client=firefox&hl={lang_code}&q={encoded_query}"
-    async with session.get(url) as response:
-        if response.status == 200:
-            try:
-                return (await response.json())[1]
-            except ValueError:
-                return []
+    response = requests.get(url)
+    if response.status_code == 200:
+        try:
+            return response.json()[1]
+        except ValueError:
+            print("Error decodificando JSON de Google")
+            return []
+    else:
         return []
 
-# Función asíncrona para obtener sugerencias de YouTube
-async def fetch_youtube_suggestions(session, query, lang_code="es"):
+# Función para obtener sugerencias de YouTube
+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}"
-    async with session.get(url) as response:
-        if response.status == 200:
-            try:
-                text = await response.text()
-                match = re.search(r'window\.google\.ac\.h\(\["[^"]*",\[(.*?)\],', text)
-                if match:
-                    suggestions_data = match.group(1)
-                    suggestions = re.findall(r'\["([^"]+)"', suggestions_data)
-                    return suggestions
-                return []
-            except Exception:
+    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
+            else:
+                print("No se encontraron sugerencias en el formato esperado.")
                 return []
+        except Exception as e:
+            print(f"Error procesando la respuesta de YouTube: {e}")
+            return []
+    else:
         return []
 
-# Función asíncrona para obtener sugerencias de Bing
-async def fetch_bing_suggestions(session, query, market="en-US"):
+# 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
     }
-    async with session.get(url, params=params) as response:
-        if response.status == 200:
-            try:
-                obj = xmltodict.parse(await response.text())
+    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):
-                    return [s['Text'] for s in suggestions]
+                    for s in suggestions:
+                        suggestList.append(s['Text'])
                 elif isinstance(suggestions, dict):
-                    return [suggestions['Text']]
-                return []
-            except Exception:
-                return []
+                    suggestList.append(suggestions['Text'])
+            return suggestList
+        except Exception as e:
+            print(f"Error procesando la respuesta de Bing: {e}")
+            return []
+    else:
         return []
 
-# Función asíncrona para obtener sugerencias de Amazon
-async def fetch_amazon_suggestions(session, query, market_id="ATVPDKIKX0DER", alias="aps"):
+# 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
     }
-    async with session.get(url, params=params) as response:
-        if response.status == 200:
-            try:
-                data = await response.json()
-                return [item['value'] for item in data.get('suggestions', [])]
-            except ValueError:
-                return []
+    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', [])]
+        except ValueError:
+            print("Error decodificando JSON de Amazon")
+            return []
+    else:
         return []
 
-# Expansión de palabra clave: stopwords, prefijos/sufijos y variantes semánticas
+# Función para expandir la palabra clave
 def expand_keyword(keyword):
-    expanded_keywords = {keyword}
+    expanded_keywords = [keyword]
+    for letter in 'abcdefghijklmnopqrstuvwxyz*_':
+        expanded_keywords.append(keyword + " " + letter)
+        expanded_keywords.append(letter + " " + keyword)
+    return expanded_keywords
 
-    stopwords = ['de', 'para', 'con', 'en', 'y', 'o', 'sin', 'por', 'al']
-    prefixes = ['comprar', 'mejor', 'oferta', 'precio', 'barato']
-    suffixes = ['online', 'gratis', '2023', 'calidad', 'fácil', 'rápido']
-
-    # Combinaciones con letras, stopwords, prefijos y sufijos
-    for word in stopwords:
-        expanded_keywords.add(f"{keyword} {word}")
-        expanded_keywords.add(f"{word} {keyword}")
+# Función principal
+def main(keyword):
+    expanded_keywords = expand_keyword(keyword)
+    all_suggestions = {}
+    google_suggestions_all = []
+    duckduckgo_suggestions_all = []
+    youtube_suggestions_all = []
+    bing_suggestions_all = []
+    amazon_suggestions_all = []
 
-    for prefix in prefixes:
-        expanded_keywords.add(f"{prefix} {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
 
-    for suffix in suffixes:
-        expanded_keywords.add(f"{keyword} {suffix}")
+    # 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
 
-    # Generar variantes de plurales y singulares
-    if not keyword.endswith('s'):
-        expanded_keywords.add(f"{keyword}s")
+    # 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
 
-    # Errores tipográficos comunes
-    if len(keyword) > 2:
-        expanded_keywords.add(keyword[:-1])  # Eliminamos una letra al final
-        expanded_keywords.add(keyword + keyword[-1])  # Duplicamos la última letra
+    # 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
 
-    return list(expanded_keywords)
+    # 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
 
-# Función principal asíncrona para obtener todas las sugerencias
-async def main_async(keyword):
-    expanded_keywords = expand_keyword(keyword)
-    all_suggestions = {}
+    # Filtrar las top 10 de cada plataforma con su número de repeticiones
+    google_top_10 = list(set(google_suggestions_all))[:10]
+    duckduckgo_top_10 = list(set(duckduckgo_suggestions_all))[:10]
+    youtube_top_10 = list(set(youtube_suggestions_all))[:10]
+    bing_top_10 = list(set(bing_suggestions_all))[:10]
+    amazon_top_10 = list(set(amazon_suggestions_all))[:10]
 
-    async with aiohttp.ClientSession() as session:
-        tasks = []
-        for exp_keyword in expanded_keywords:
-            tasks.append(fetch_google_suggestions(session, exp_keyword))
-            tasks.append(fetch_duckduckgo_suggestions(session, exp_keyword))
-            tasks.append(fetch_youtube_suggestions(session, exp_keyword))
-            tasks.append(fetch_bing_suggestions(session, exp_keyword))
-            tasks.append(fetch_amazon_suggestions(session, exp_keyword))
-        
-        results = await asyncio.gather(*tasks)
-
-    # Procesar los resultados
-    for result in results:
-        if result:
-            for suggestion in result:
-                if suggestion in all_suggestions:
-                    all_suggestions[suggestion] += 1
-                else:
-                    all_suggestions[suggestion] = 1
-
-    # Ordenar las sugerencias por su número de repeticiones
+    # Ordenar y filtrar las sugerencias más frecuentes combinadas
     sorted_suggestions = sorted(all_suggestions.items(), key=lambda item: item[1], reverse=True)
-    
-    return sorted_suggestions
+    combined_top_10_suggestions = [sug for sug, freq in sorted_suggestions if freq >= 2][:10]
+    suggestions_str = ", ".join(combined_top_10_suggestions)
 
-# Función de Gradio para mostrar los resultados
-def main(keyword):
-    suggestions = asyncio.run(main_async(keyword))
-    
-    # Crear el HTML de salida con el número de repeticiones
-    html_output = "<h4>Lista completa de palabras clave con su número de repeticiones:</h4><ul>"
-    for suggestion, freq in suggestions:
-        html_output += f"<li>{suggestion} - {freq} repeticiones</li>"
+    # 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
+    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 10 Sugerencias de Google:</h4>
+    <ul>
+    """
+    for suggestion in google_top_10:
+        freq = all_suggestions[suggestion]
+        html_output += f"<li>{suggestion} ({freq})</li>"
+    html_output += "</ul>"
+
+    html_output += """
+    <h4>Top 10 Sugerencias de DuckDuckGo:</h4>
+    <ul>
+    """
+    for suggestion in duckduckgo_top_10:
+        freq = all_suggestions[suggestion]
+        html_output += f"<li>{suggestion} ({freq})</li>"
     html_output += "</ul>"
 
+    html_output += """
+    <h4>Top 10 Sugerencias de YouTube:</h4>
+    <ul>
+    """
+    for suggestion in youtube_top_10:
+        freq = all_suggestions[suggestion]
+        html_output += f"<li>{suggestion} ({freq})</li>"
+    html_output += "</ul>"
+
+    html_output += """
+    <h4>Top 10 Sugerencias de Bing:</h4>
+    <ul>
+    """
+    for suggestion in bing_top_10:
+        freq = all_suggestions[suggestion]
+        html_output += f"<li>{suggestion} ({freq})</li>"
+    html_output += "</ul>"
+
+    html_output += """
+    <h4>Top 10 Sugerencias de Amazon:</h4>
+    <ul>
+    """
+    for suggestion in amazon_top_10:
+        freq = all_suggestions[suggestion]
+        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
@@ -163,9 +259,9 @@ iface = gr.Interface(
     fn=main,
     inputs="text",
     outputs="html",
-    title="Sugerencias Combinadas de Google, DuckDuckGo, YouTube, Bing y Amazon",
-    description="Ingrese una palabra clave para obtener sugerencias de búsqueda relacionadas de varias plataformas. Las sugerencias se ordenan por su número de repeticiones.",
-    article="Desarrollada por © Artxe Web"
+    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 10 primeras sugerencias combinadas y también las 10 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>"
 )
 
 iface.launch()