Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,47 +1,28 @@
|
|
| 1 |
import streamlit as st
|
| 2 |
from mtranslate import translate
|
|
|
|
| 3 |
import re
|
|
|
|
| 4 |
import html
|
| 5 |
|
| 6 |
# Diccionario de idiomas y sus códigos
|
| 7 |
lang_dict = {
|
|
|
|
| 8 |
'Español': 'es',
|
| 9 |
'English': 'en',
|
| 10 |
-
'Mandarín': 'zh',
|
| 11 |
-
'Hindi': 'hi',
|
| 12 |
-
'Árabe': 'ar',
|
| 13 |
-
'Portugués': 'pt',
|
| 14 |
-
'Bengalí': 'bn',
|
| 15 |
-
'Ruso': 'ru',
|
| 16 |
-
'Japonés': 'ja',
|
| 17 |
-
'Panyabí': 'pa',
|
| 18 |
-
'Alemán': 'de',
|
| 19 |
-
'Javanés': 'jw',
|
| 20 |
-
'Coreano': 'ko',
|
| 21 |
'Francés': 'fr',
|
| 22 |
-
'
|
| 23 |
-
'Turco': 'tr',
|
| 24 |
'Italiano': 'it',
|
| 25 |
-
'
|
| 26 |
-
'
|
| 27 |
-
'
|
| 28 |
-
'
|
| 29 |
-
'Griego': 'el',
|
| 30 |
-
'Neerlandés': 'nl',
|
| 31 |
-
'Sueco': 'sv',
|
| 32 |
-
'Rumano': 'ro',
|
| 33 |
-
'Checo': 'cs',
|
| 34 |
-
'Húngaro': 'hu',
|
| 35 |
-
'Hebreo': 'he',
|
| 36 |
-
'Indonesio': 'id',
|
| 37 |
-
'Nepalí': 'ne',
|
| 38 |
-
'Gallego': 'gl',
|
| 39 |
-
'Catalán': 'ca',
|
| 40 |
-
'Vasco': 'eu'
|
| 41 |
}
|
| 42 |
|
| 43 |
lang_list = list(lang_dict.keys())
|
|
|
|
| 44 |
|
|
|
|
| 45 |
def split_text(text, limit=4000):
|
| 46 |
sentences = re.split(r'([;.])', text)
|
| 47 |
chunks = []
|
|
@@ -56,50 +37,73 @@ def split_text(text, limit=4000):
|
|
| 56 |
chunks.append(chunk)
|
| 57 |
return chunks
|
| 58 |
|
| 59 |
-
def
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
|
| 71 |
def main():
|
| 72 |
-
st.title("
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
st.
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
|
| 97 |
-
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 101 |
|
| 102 |
if "translated_text" not in st.session_state:
|
| 103 |
st.session_state.translated_text = ""
|
| 104 |
|
| 105 |
-
|
|
|
|
|
|
| 1 |
import streamlit as st
|
| 2 |
from mtranslate import translate
|
| 3 |
+
from bs4 import BeautifulSoup, NavigableString
|
| 4 |
import re
|
| 5 |
+
import time
|
| 6 |
import html
|
| 7 |
|
| 8 |
# Diccionario de idiomas y sus códigos
|
| 9 |
lang_dict = {
|
| 10 |
+
'Auto': 'auto',
|
| 11 |
'Español': 'es',
|
| 12 |
'English': 'en',
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
'Francés': 'fr',
|
| 14 |
+
'Alemán': 'de',
|
|
|
|
| 15 |
'Italiano': 'it',
|
| 16 |
+
'Portugués': 'pt',
|
| 17 |
+
'Chino': 'zh',
|
| 18 |
+
'Japonés': 'ja',
|
| 19 |
+
'Ruso': 'ru',
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
}
|
| 21 |
|
| 22 |
lang_list = list(lang_dict.keys())
|
| 23 |
+
source_lang_list = ['Auto'] + lang_list
|
| 24 |
|
| 25 |
+
# --- FUNCIONES ---
|
| 26 |
def split_text(text, limit=4000):
|
| 27 |
sentences = re.split(r'([;.])', text)
|
| 28 |
chunks = []
|
|
|
|
| 37 |
chunks.append(chunk)
|
| 38 |
return chunks
|
| 39 |
|
| 40 |
+
def translate_html_content(text, source_lang, target_lang):
|
| 41 |
+
soup = BeautifulSoup(text, 'html.parser')
|
| 42 |
+
|
| 43 |
+
for node in soup.find_all(text=True):
|
| 44 |
+
if isinstance(node, NavigableString) and node.strip():
|
| 45 |
+
original_text = node.strip()
|
| 46 |
+
chunks = split_text(original_text)
|
| 47 |
+
translated_chunks = [translate(chunk, target_lang, source_lang) for chunk in chunks]
|
| 48 |
+
translated_text = ''.join(translated_chunks)
|
| 49 |
+
|
| 50 |
+
# Asegurar espacios alrededor del texto
|
| 51 |
+
if node.previous_sibling and not node.previous_sibling.endswith(' '):
|
| 52 |
+
translated_text = ' ' + translated_text
|
| 53 |
+
if node.next_sibling and not node.next_sibling.startswith(' '):
|
| 54 |
+
translated_text += ' '
|
| 55 |
+
|
| 56 |
+
node.replace_with(translated_text)
|
| 57 |
+
|
| 58 |
+
return str(soup)
|
| 59 |
|
| 60 |
def main():
|
| 61 |
+
st.title("🌍 Traductor HTML Inteligente")
|
| 62 |
+
|
| 63 |
+
with st.expander("ℹ️ Instrucciones"):
|
| 64 |
+
st.write(
|
| 65 |
+
"- Introduce tu texto o HTML en la caja de texto.\n"
|
| 66 |
+
"- Selecciona el idioma de origen y el idioma de destino.\n"
|
| 67 |
+
"- Haz clic en 'Traducir' para obtener el texto traducido."
|
| 68 |
+
)
|
| 69 |
+
|
| 70 |
+
text = st.text_area("✍️ Introduce el texto o HTML aquí:", height=200)
|
| 71 |
+
col1, col2 = st.columns(2)
|
| 72 |
+
with col1:
|
| 73 |
+
source_lang = st.selectbox("🌐 Idioma de origen:", source_lang_list)
|
| 74 |
+
with col2:
|
| 75 |
+
target_lang = st.selectbox("🎯 Idioma de destino:", lang_list, index=1)
|
| 76 |
+
|
| 77 |
+
if st.button("🚀 Traducir"):
|
| 78 |
+
if text.strip():
|
| 79 |
+
with st.spinner("⏳ Traduciendo..."):
|
| 80 |
+
translated_text = translate_html_content(text, lang_dict[source_lang], lang_dict[target_lang])
|
| 81 |
+
st.session_state.translated_text = translated_text
|
| 82 |
+
time.sleep(2) # Simulación de carga
|
| 83 |
+
|
| 84 |
+
st.subheader("📝 Resultado de la traducción:")
|
| 85 |
+
st.write(translated_text)
|
| 86 |
+
|
| 87 |
+
# Botón de copiar
|
| 88 |
+
copy_button = f"""
|
| 89 |
+
<button onclick="copyToClipboard()">📋 Copiar</button>
|
| 90 |
+
<script>
|
| 91 |
+
function copyToClipboard() {{
|
| 92 |
+
var text = `{html.escape(translated_text)}`;
|
| 93 |
+
navigator.clipboard.writeText(text).then(function() {{
|
| 94 |
+
alert('Texto copiado al portapapeles');
|
| 95 |
+
}}, function(err) {{
|
| 96 |
+
console.error('No se pudo copiar el texto: ', err);
|
| 97 |
+
}});
|
| 98 |
+
}}
|
| 99 |
+
</script>
|
| 100 |
+
"""
|
| 101 |
+
st.markdown(copy_button, unsafe_allow_html=True)
|
| 102 |
+
else:
|
| 103 |
+
st.warning("⚠️ Por favor, introduce un texto antes de traducir.")
|
| 104 |
|
| 105 |
if "translated_text" not in st.session_state:
|
| 106 |
st.session_state.translated_text = ""
|
| 107 |
|
| 108 |
+
# Ejecutar la app
|
| 109 |
+
main()
|