Update app.py
Browse files
app.py
CHANGED
|
@@ -3,42 +3,12 @@ from mtranslate import translate
|
|
| 3 |
import re
|
| 4 |
import json
|
| 5 |
|
| 6 |
-
# Diccionario de idiomas
|
| 7 |
lang_dict = {
|
| 8 |
'Auto': 'auto',
|
| 9 |
'Español': 'es',
|
| 10 |
'English': 'en',
|
| 11 |
-
|
| 12 |
-
'Hindi': 'hi',
|
| 13 |
-
'Árabe': 'ar',
|
| 14 |
-
'Portugués': 'pt',
|
| 15 |
-
'Bengalí': 'bn',
|
| 16 |
-
'Ruso': 'ru',
|
| 17 |
-
'Japonés': 'ja',
|
| 18 |
-
'Panyabí': 'pa',
|
| 19 |
-
'Alemán': 'de',
|
| 20 |
-
'Javanés': 'jw',
|
| 21 |
-
'Coreano': 'ko',
|
| 22 |
-
'Francés': 'fr',
|
| 23 |
-
'Vietnamita': 'vi',
|
| 24 |
-
'Turco': 'tr',
|
| 25 |
-
'Italiano': 'it',
|
| 26 |
-
'Ucraniano': 'uk',
|
| 27 |
-
'Tailandés': 'th',
|
| 28 |
-
'Guyaratí': 'gu',
|
| 29 |
-
'Polaco': 'pl',
|
| 30 |
-
'Griego': 'el',
|
| 31 |
-
'Neerlandés': 'nl',
|
| 32 |
-
'Sueco': 'sv',
|
| 33 |
-
'Rumano': 'ro',
|
| 34 |
-
'Checo': 'cs',
|
| 35 |
-
'Húngaro': 'hu',
|
| 36 |
-
'Hebreo': 'he',
|
| 37 |
-
'Indonesio': 'id',
|
| 38 |
-
'Nepalí': 'ne',
|
| 39 |
-
'Gallego': 'gl',
|
| 40 |
-
'Catalán': 'ca',
|
| 41 |
-
'Vasco': 'eu'
|
| 42 |
}
|
| 43 |
|
| 44 |
lang_list = [k for k in lang_dict.keys() if k != 'Auto']
|
|
@@ -59,28 +29,31 @@ def split_text(text, limit=4000):
|
|
| 59 |
return chunks
|
| 60 |
|
| 61 |
def translate_html_content(text, source_lang, target_lang):
|
| 62 |
-
# Patrón para identificar etiquetas HTML
|
| 63 |
html_tag_pattern = re.compile(r'<[^>]+>')
|
| 64 |
|
| 65 |
-
# Separar
|
| 66 |
parts = html_tag_pattern.split(text)
|
| 67 |
tags = html_tag_pattern.findall(text)
|
| 68 |
|
| 69 |
-
# Traducir solo las partes
|
| 70 |
translated_parts = []
|
| 71 |
for part in parts:
|
| 72 |
-
if part.strip(): #
|
| 73 |
chunks = split_text(part)
|
| 74 |
translated_chunks = [translate(chunk, target_lang, source_lang) for chunk in chunks]
|
| 75 |
-
|
|
|
|
|
|
|
|
|
|
| 76 |
else:
|
| 77 |
-
translated_parts.append(part) # Mantener partes vacías o
|
| 78 |
|
| 79 |
-
# Reconstruir el texto
|
| 80 |
translated_text = ''
|
| 81 |
for i in range(len(parts)):
|
| 82 |
translated_text += translated_parts[i]
|
| 83 |
-
if i < len(tags):
|
| 84 |
translated_text += tags[i]
|
| 85 |
|
| 86 |
return translated_text
|
|
@@ -90,6 +63,7 @@ def main(Texto, source_lang, target_lang):
|
|
| 90 |
target_code = lang_dict[target_lang]
|
| 91 |
|
| 92 |
translated_text = translate_html_content(Texto, source_code, target_code)
|
|
|
|
| 93 |
escaped_text = json.dumps(translated_text)[1:-1]
|
| 94 |
|
| 95 |
html_output = f"""
|
|
@@ -103,6 +77,7 @@ def main(Texto, source_lang, target_lang):
|
|
| 103 |
"""
|
| 104 |
return html_output
|
| 105 |
|
|
|
|
| 106 |
iface = gr.Interface(
|
| 107 |
fn=main,
|
| 108 |
inputs=[
|
|
@@ -111,9 +86,9 @@ iface = gr.Interface(
|
|
| 111 |
gr.Dropdown(lang_list, value="Español", label="Idioma destino")
|
| 112 |
],
|
| 113 |
outputs="text",
|
| 114 |
-
title="
|
| 115 |
-
description="
|
| 116 |
-
article="
|
| 117 |
)
|
| 118 |
|
| 119 |
iface.launch()
|
|
|
|
| 3 |
import re
|
| 4 |
import json
|
| 5 |
|
| 6 |
+
# Diccionario de idiomas (sin cambios)
|
| 7 |
lang_dict = {
|
| 8 |
'Auto': 'auto',
|
| 9 |
'Español': 'es',
|
| 10 |
'English': 'en',
|
| 11 |
+
# ... resto del diccionario ...
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
}
|
| 13 |
|
| 14 |
lang_list = [k for k in lang_dict.keys() if k != 'Auto']
|
|
|
|
| 29 |
return chunks
|
| 30 |
|
| 31 |
def translate_html_content(text, source_lang, target_lang):
|
| 32 |
+
# Patrón para identificar etiquetas HTML
|
| 33 |
html_tag_pattern = re.compile(r'<[^>]+>')
|
| 34 |
|
| 35 |
+
# Separar texto y etiquetas
|
| 36 |
parts = html_tag_pattern.split(text)
|
| 37 |
tags = html_tag_pattern.findall(text)
|
| 38 |
|
| 39 |
+
# Traducir solo las partes de texto
|
| 40 |
translated_parts = []
|
| 41 |
for part in parts:
|
| 42 |
+
if part.strip(): # Solo traducir si hay texto no vacío
|
| 43 |
chunks = split_text(part)
|
| 44 |
translated_chunks = [translate(chunk, target_lang, source_lang) for chunk in chunks]
|
| 45 |
+
translated_part = ''.join(translated_chunks)
|
| 46 |
+
# Limpiar caracteres escapados no deseados
|
| 47 |
+
translated_part = translated_part.replace('\\"', '"').replace('\\n', '\n').replace('\\\\', '\\')
|
| 48 |
+
translated_parts.append(translated_part)
|
| 49 |
else:
|
| 50 |
+
translated_parts.append(part) # Mantener partes vacías o espacios
|
| 51 |
|
| 52 |
+
# Reconstruir el texto
|
| 53 |
translated_text = ''
|
| 54 |
for i in range(len(parts)):
|
| 55 |
translated_text += translated_parts[i]
|
| 56 |
+
if i < len(tags):
|
| 57 |
translated_text += tags[i]
|
| 58 |
|
| 59 |
return translated_text
|
|
|
|
| 63 |
target_code = lang_dict[target_lang]
|
| 64 |
|
| 65 |
translated_text = translate_html_content(Texto, source_code, target_code)
|
| 66 |
+
# Escapar solo para el botón de copiar, no para la salida visual
|
| 67 |
escaped_text = json.dumps(translated_text)[1:-1]
|
| 68 |
|
| 69 |
html_output = f"""
|
|
|
|
| 77 |
"""
|
| 78 |
return html_output
|
| 79 |
|
| 80 |
+
# Configuración de la interfaz (sin cambios)
|
| 81 |
iface = gr.Interface(
|
| 82 |
fn=main,
|
| 83 |
inputs=[
|
|
|
|
| 86 |
gr.Dropdown(lang_list, value="Español", label="Idioma destino")
|
| 87 |
],
|
| 88 |
outputs="text",
|
| 89 |
+
title="Traducción sin límites",
|
| 90 |
+
description="Ingresa el texto que deseas traducir y selecciona los idiomas. ¡No hay límites!",
|
| 91 |
+
article="Desarrollada por © Artxe Web"
|
| 92 |
)
|
| 93 |
|
| 94 |
iface.launch()
|