Update app.py
Browse files
app.py
CHANGED
|
@@ -9,7 +9,6 @@ lang_dict = {
|
|
| 9 |
'Auto': 'auto',
|
| 10 |
'Español': 'es',
|
| 11 |
'English': 'en',
|
| 12 |
-
# Agrega más idiomas según necesites
|
| 13 |
}
|
| 14 |
|
| 15 |
lang_list = [k for k in lang_dict.keys() if k != 'Auto']
|
|
@@ -32,30 +31,30 @@ def split_text(text, limit=4000):
|
|
| 32 |
|
| 33 |
def translate_html_content(text, source_lang, target_lang):
|
| 34 |
soup = BeautifulSoup(text, 'html.parser')
|
| 35 |
-
|
| 36 |
# Traducir nodos de texto
|
| 37 |
for node in soup.find_all(text=True):
|
| 38 |
if isinstance(node, NavigableString) and node.strip():
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
# Dividir en fragmentos si es necesario
|
| 42 |
-
chunks = split_text(text_to_translate)
|
| 43 |
translated_chunks = [translate(chunk, target_lang, source_lang) for chunk in chunks]
|
| 44 |
translated_text = ''.join(translated_chunks)
|
| 45 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
node.replace_with(translated_text)
|
| 47 |
-
|
| 48 |
-
# Asegurar
|
| 49 |
for element in soup.find_all(['p', 'span', 'b', 'strong']):
|
| 50 |
-
# Verificar el nodo anterior
|
| 51 |
if element.previous_sibling and isinstance(element.previous_sibling, NavigableString):
|
| 52 |
-
|
| 53 |
-
element.previous_sibling += ' '
|
| 54 |
-
# Verificar el nodo siguiente
|
| 55 |
if element.next_sibling and isinstance(element.next_sibling, NavigableString):
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
return str(soup)
|
| 60 |
|
| 61 |
def main(Texto, source_lang, target_lang):
|
|
@@ -64,7 +63,7 @@ def main(Texto, source_lang, target_lang):
|
|
| 64 |
|
| 65 |
translated_text = translate_html_content(Texto, source_code, target_code)
|
| 66 |
element_id = f"text-{uuid.uuid4()}"
|
| 67 |
-
|
| 68 |
html_output = f"""
|
| 69 |
<div style="position: relative; min-height: 100px;">
|
| 70 |
<div id="translated-content">{translated_text}</div>
|
|
@@ -91,4 +90,4 @@ iface = gr.Interface(
|
|
| 91 |
article="Desarrollada por © Artxe Web"
|
| 92 |
)
|
| 93 |
|
| 94 |
-
iface.launch()
|
|
|
|
| 9 |
'Auto': 'auto',
|
| 10 |
'Español': 'es',
|
| 11 |
'English': 'en',
|
|
|
|
| 12 |
}
|
| 13 |
|
| 14 |
lang_list = [k for k in lang_dict.keys() if k != 'Auto']
|
|
|
|
| 31 |
|
| 32 |
def translate_html_content(text, source_lang, target_lang):
|
| 33 |
soup = BeautifulSoup(text, 'html.parser')
|
| 34 |
+
|
| 35 |
# Traducir nodos de texto
|
| 36 |
for node in soup.find_all(text=True):
|
| 37 |
if isinstance(node, NavigableString) and node.strip():
|
| 38 |
+
original_text = node.strip()
|
| 39 |
+
chunks = split_text(original_text)
|
|
|
|
|
|
|
| 40 |
translated_chunks = [translate(chunk, target_lang, source_lang) for chunk in chunks]
|
| 41 |
translated_text = ''.join(translated_chunks)
|
| 42 |
+
|
| 43 |
+
# Asegurar que se mantengan los espacios antes y después del nodo
|
| 44 |
+
if node.previous_sibling and not node.previous_sibling.endswith(' '):
|
| 45 |
+
translated_text = ' ' + translated_text
|
| 46 |
+
if node.next_sibling and not node.next_sibling.startswith(' '):
|
| 47 |
+
translated_text = translated_text + ' '
|
| 48 |
+
|
| 49 |
node.replace_with(translated_text)
|
| 50 |
+
|
| 51 |
+
# Asegurar espacios alrededor de etiquetas de bloque
|
| 52 |
for element in soup.find_all(['p', 'span', 'b', 'strong']):
|
|
|
|
| 53 |
if element.previous_sibling and isinstance(element.previous_sibling, NavigableString):
|
| 54 |
+
element.insert_before(NavigableString(' ')) # Añadir espacio antes si no hay
|
|
|
|
|
|
|
| 55 |
if element.next_sibling and isinstance(element.next_sibling, NavigableString):
|
| 56 |
+
element.insert_after(NavigableString(' ')) # Añadir espacio después si no hay
|
| 57 |
+
|
|
|
|
| 58 |
return str(soup)
|
| 59 |
|
| 60 |
def main(Texto, source_lang, target_lang):
|
|
|
|
| 63 |
|
| 64 |
translated_text = translate_html_content(Texto, source_code, target_code)
|
| 65 |
element_id = f"text-{uuid.uuid4()}"
|
| 66 |
+
|
| 67 |
html_output = f"""
|
| 68 |
<div style="position: relative; min-height: 100px;">
|
| 69 |
<div id="translated-content">{translated_text}</div>
|
|
|
|
| 90 |
article="Desarrollada por © Artxe Web"
|
| 91 |
)
|
| 92 |
|
| 93 |
+
iface.launch()
|