fix: Mejorar segmentación de traducciones para textos largos
- Limitar cada segmento a 400 caracteres - Dividir textos largos por palabras para evitar truncamiento - Limpiar logs de debug
This commit is contained in:
@@ -49,58 +49,68 @@ REVERSE_MAPPING = {}
|
|||||||
FLAG_MAPPING = {}
|
FLAG_MAPPING = {}
|
||||||
NAME_TO_CODE = {}
|
NAME_TO_CODE = {}
|
||||||
|
|
||||||
async def _do_translate_request(session, url, text, target_code):
|
async def _do_translate_request(session, url, text, target_code, max_length=500):
|
||||||
"""Función interna para realizar una única petición de traducción."""
|
"""Función interna para realizar una única petición de traducción."""
|
||||||
if not text.strip() or not re.search(r'[a-zA-Z\u00C0-\u017F]', text):
|
if not text.strip() or not re.search(r'[a-zA-Z\u00C0-\u017F]', text):
|
||||||
return text
|
return text
|
||||||
|
|
||||||
payload = {
|
chunks = []
|
||||||
"q": text,
|
if len(text) > max_length:
|
||||||
"source": "auto",
|
parts = text.split(' ')
|
||||||
"target": target_code,
|
current_chunk = ''
|
||||||
"format": "html"
|
for part in parts:
|
||||||
}
|
if len(current_chunk) + len(part) + 1 <= max_length:
|
||||||
|
current_chunk += (' ' if current_chunk else '') + part
|
||||||
try:
|
|
||||||
async with session.post(url, json=payload, timeout=30) as resp:
|
|
||||||
if resp.status == 200:
|
|
||||||
data = await resp.json()
|
|
||||||
translated = data.get("translatedText", text)
|
|
||||||
print(f"[TRANSLATE] Segmento traducido: '{text[:30]}...' -> '{translated[:30]}...'")
|
|
||||||
return translated
|
|
||||||
else:
|
else:
|
||||||
print(f"[TRANSLATE] Error HTTP {resp.status}")
|
if current_chunk:
|
||||||
return text
|
chunks.append(current_chunk)
|
||||||
except Exception as e:
|
current_chunk = part
|
||||||
print(f"[TRANSLATE] Error en petición: {e}")
|
if current_chunk:
|
||||||
return text
|
chunks.append(current_chunk)
|
||||||
|
else:
|
||||||
|
chunks.append(text)
|
||||||
|
|
||||||
|
results = []
|
||||||
|
for chunk in chunks:
|
||||||
|
payload = {
|
||||||
|
"q": chunk,
|
||||||
|
"source": "auto",
|
||||||
|
"target": target_code,
|
||||||
|
"format": "html"
|
||||||
|
}
|
||||||
|
|
||||||
|
try:
|
||||||
|
async with session.post(url, json=payload, timeout=30) as resp:
|
||||||
|
if resp.status == 200:
|
||||||
|
data = await resp.json()
|
||||||
|
translated = data.get("translatedText", chunk)
|
||||||
|
results.append(translated)
|
||||||
|
else:
|
||||||
|
results.append(chunk)
|
||||||
|
except Exception as e:
|
||||||
|
print(f"[TRANSLATE] Error en petición: {e}")
|
||||||
|
results.append(chunk)
|
||||||
|
|
||||||
|
return ' '.join(results)
|
||||||
|
|
||||||
async def translate_text(text: str, target_lang: str) -> str:
|
async def translate_text(text: str, target_lang: str) -> str:
|
||||||
url = get_libretranslate_url()
|
url = get_libretranslate_url()
|
||||||
if not url:
|
if not url:
|
||||||
print(f"[TRANSLATE] URL no configurada")
|
|
||||||
return text
|
return text
|
||||||
|
|
||||||
print(f"[TRANSLATE] target_lang recibido: '{target_lang}'")
|
|
||||||
|
|
||||||
target_code = NAME_TO_CODE.get(target_lang, target_lang)
|
target_code = NAME_TO_CODE.get(target_lang, target_lang)
|
||||||
print(f"[TRANSLATE] target_code resuelto: '{target_code}'")
|
|
||||||
print(f"[TRANSLATE] texto a traducir: '{text[:100]}...' (largo: {len(text)})")
|
|
||||||
|
|
||||||
segments = re.split(r'([.!?]+\s*|\n+)', text)
|
segments = re.split(r'([.!?]+\s*|\n+)', text)
|
||||||
segments = [s for s in segments if s]
|
segments = [s for s in segments if s]
|
||||||
print(f"[TRANSLATE] Segmentos: {len(segments)}")
|
|
||||||
|
|
||||||
async with aiohttp.ClientSession() as session:
|
async with aiohttp.ClientSession() as session:
|
||||||
tasks = []
|
tasks = []
|
||||||
for segment in segments:
|
for segment in segments:
|
||||||
tasks.append(_do_translate_request(session, url, segment, target_code))
|
tasks.append(_do_translate_request(session, url, segment, target_code, max_length=400))
|
||||||
|
|
||||||
translated_segments = await asyncio.gather(*tasks)
|
translated_segments = await asyncio.gather(*tasks)
|
||||||
|
|
||||||
result = "".join(translated_segments)
|
return "".join(translated_segments)
|
||||||
print(f"[TRANSLATE] Resultado final: '{result[:100]}...'")
|
|
||||||
return result
|
|
||||||
|
|
||||||
def translate_text_sync(text: str, target_lang: str) -> str:
|
def translate_text_sync(text: str, target_lang: str) -> str:
|
||||||
"""Versión síncrona de translate_text utilizando un hilo separado."""
|
"""Versión síncrona de translate_text utilizando un hilo separado."""
|
||||||
|
|||||||
@@ -30,8 +30,6 @@ class TranslationButton(discord.ui.Button):
|
|||||||
return
|
return
|
||||||
|
|
||||||
original_msg_id = interaction.message.reference.message_id
|
original_msg_id = interaction.message.reference.message_id
|
||||||
print(f"[UI] Traduciendo mensaje {original_msg_id} a {self.lang_code}")
|
|
||||||
|
|
||||||
db_msg = get_message(original_msg_id)
|
db_msg = get_message(original_msg_id)
|
||||||
|
|
||||||
if not db_msg:
|
if not db_msg:
|
||||||
@@ -41,17 +39,11 @@ class TranslationButton(discord.ui.Button):
|
|||||||
text = db_msg['content']
|
text = db_msg['content']
|
||||||
mentions_map = db_msg['mentions_map']
|
mentions_map = db_msg['mentions_map']
|
||||||
|
|
||||||
print(f"[UI] Texto original: {text}")
|
|
||||||
|
|
||||||
# Traducción
|
|
||||||
cached = get_cached_translation(original_msg_id, self.lang_code)
|
cached = get_cached_translation(original_msg_id, self.lang_code)
|
||||||
if cached:
|
if cached:
|
||||||
translated = cached
|
translated = cached
|
||||||
print(f"[UI] Usando traducción cacheada: {translated}")
|
|
||||||
else:
|
else:
|
||||||
print(f"[UI] Traduciendo con translate_text...")
|
|
||||||
translated = await translate_text(text, self.lang_code)
|
translated = await translate_text(text, self.lang_code)
|
||||||
print(f"[UI] Traducción resultado: {translated}")
|
|
||||||
save_translation(original_msg_id, self.lang_code, translated)
|
save_translation(original_msg_id, self.lang_code, translated)
|
||||||
|
|
||||||
translated = html.unescape(translated)
|
translated = html.unescape(translated)
|
||||||
@@ -78,9 +70,7 @@ class TranslationButton(discord.ui.Button):
|
|||||||
await interaction.edit_original_response(content=translated, view=new_view)
|
await interaction.edit_original_response(content=translated, view=new_view)
|
||||||
|
|
||||||
except Exception as e:
|
except Exception as e:
|
||||||
import traceback
|
|
||||||
print(f"[ERROR UI] {e}")
|
print(f"[ERROR UI] {e}")
|
||||||
traceback.print_exc()
|
|
||||||
await interaction.followup.send(f"❌ Error: {str(e)}", ephemeral=True)
|
await interaction.followup.send(f"❌ Error: {str(e)}", ephemeral=True)
|
||||||
|
|
||||||
class PersistentTranslationView(discord.ui.View):
|
class PersistentTranslationView(discord.ui.View):
|
||||||
@@ -90,21 +80,6 @@ class PersistentTranslationView(discord.ui.View):
|
|||||||
for lang in db_langs:
|
for lang in db_langs:
|
||||||
self.add_item(TranslationButton(lang['name'], lang['code'], lang.get('flag', '')))
|
self.add_item(TranslationButton(lang['name'], lang['code'], lang.get('flag', '')))
|
||||||
|
|
||||||
_global_views = {}
|
|
||||||
|
|
||||||
def get_message_translation_view(guild_id: int):
|
|
||||||
view_key = f"msg_trans_{guild_id}"
|
|
||||||
if view_key not in _global_views:
|
|
||||||
_global_views[view_key] = discord.ui.View(timeout=None)
|
|
||||||
active_codes = get_active_languages(guild_id)
|
|
||||||
if not active_codes:
|
|
||||||
active_codes = get_bot_languages("discord")
|
|
||||||
db_langs = get_available_languages()
|
|
||||||
for lang in db_langs:
|
|
||||||
if lang['code'] in active_codes:
|
|
||||||
_global_views[view_key].add_item(TranslationButton(lang['name'], lang['code'], lang.get('flag', '')))
|
|
||||||
return _global_views[view_key]
|
|
||||||
|
|
||||||
class ConfigSelect(discord.ui.Select):
|
class ConfigSelect(discord.ui.Select):
|
||||||
def __init__(self, guild_id: int, bot_type: str = "discord"):
|
def __init__(self, guild_id: int, bot_type: str = "discord"):
|
||||||
lang_mapping = get_lang_mapping(bot_type)
|
lang_mapping = get_lang_mapping(bot_type)
|
||||||
|
|||||||
Reference in New Issue
Block a user