Spaces:
Paused
Paused
Upload centralita_app.py
Browse files- centralita_app.py +470 -0
centralita_app.py
ADDED
|
@@ -0,0 +1,470 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# -*- coding: utf-8 -*-
|
| 2 |
+
"""
|
| 3 |
+
Centralita Torá — HuggingFace Space (Gradio)
|
| 4 |
+
Leer hebreo + español · escuchar (Canto Gregoriano-Hebraico con Voz Divina Especial)
|
| 5 |
+
"""
|
| 6 |
+
import os, glob, json, tempfile, html, asyncio, subprocess, shutil, re
|
| 7 |
+
import urllib.request
|
| 8 |
+
import gradio as gr
|
| 9 |
+
import edge_tts
|
| 10 |
+
try:
|
| 11 |
+
from gtts import gTTS
|
| 12 |
+
except Exception:
|
| 13 |
+
gTTS = None
|
| 14 |
+
|
| 15 |
+
from gradio_theme_fenix import fenix_theme, FENIX_CSS
|
| 16 |
+
|
| 17 |
+
# --- CONFIGURACIÓN DE VOZ Y CANTO ---
|
| 18 |
+
CANTO_GREGORIANO_HEBRAICO = True
|
| 19 |
+
|
| 20 |
+
VOZ_DIVINA = "es-ES-AlvaroNeural"
|
| 21 |
+
VOZ_NARRADOR = "es-ES-AlvaroNeural" # Usamos Álvaro con procesamiento acústico sagrado especial
|
| 22 |
+
|
| 23 |
+
VOZ_HE = "he-IL-AvriNeural"
|
| 24 |
+
|
| 25 |
+
# Perillas de entonación para Narrador
|
| 26 |
+
VOZ_RATE = "-10%"
|
| 27 |
+
VOZ_PITCH = "+5Hz"
|
| 28 |
+
|
| 29 |
+
# Fondo musical
|
| 30 |
+
FONDO = "fondo.mp3"
|
| 31 |
+
FONDO_VOL = 0.18
|
| 32 |
+
|
| 33 |
+
# IA de Estudio (Groq)
|
| 34 |
+
MODELO_ESTUDIA = "openai/gpt-oss-120b"
|
| 35 |
+
CARPETA = "libros"
|
| 36 |
+
GROQ_URL = "https://api.groq.com/openai/v1/chat/completions"
|
| 37 |
+
GROQ_KEY = os.environ.get("GROQ_API_KEY")
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
# --- MOTOR FONÉTICO GREGORIANO-HEBRAICO ---
|
| 41 |
+
def transformar_a_canto_gregoriano_hebraico(texto_sagrado: str) -> str:
|
| 42 |
+
if not texto_sagrado:
|
| 43 |
+
return ""
|
| 44 |
+
|
| 45 |
+
texto = texto_sagrado
|
| 46 |
+
|
| 47 |
+
# Melismas gregorianos en vocales tónicas
|
| 48 |
+
melismas = {
|
| 49 |
+
'á': 'a-a-a-a', 'é': 'e-e-e-e', 'í': 'i-i-i-i',
|
| 50 |
+
'ó': 'o-o-o-o', 'ú': 'u-u-u-u',
|
| 51 |
+
'Á': 'A-a-a-a', 'É': 'E-e-e-e', 'Í': 'I-i-i-i',
|
| 52 |
+
'Ó': 'O-o-o-o', 'Ú': 'U-u-u-u',
|
| 53 |
+
}
|
| 54 |
+
for vocal_con_tilde, melisma in melismas.items():
|
| 55 |
+
texto = texto.replace(vocal_con_tilde, melisma)
|
| 56 |
+
|
| 57 |
+
# Cierres litúrgicos
|
| 58 |
+
def cierre_liturgico(match):
|
| 59 |
+
vocal = match.group(1)
|
| 60 |
+
puntuacion = match.group(2)
|
| 61 |
+
return f"{vocal}-{vocal}-{vocal}{puntuacion} ... "
|
| 62 |
+
|
| 63 |
+
texto = re.sub(r'([aeiouAEIOU])([,.:;])', cierre_liturgico, texto)
|
| 64 |
+
|
| 65 |
+
# Énfasis místico
|
| 66 |
+
texto = re.sub(r'\bDios\b', 'Di-o-o-os', texto, flags=re.IGNORECASE)
|
| 67 |
+
texto = re.sub(r'\bSeñor\b', 'Se-ñoo-or', texto, flags=re.IGNORECASE)
|
| 68 |
+
texto = re.sub(r'\bIsrael\b', 'Is-ra-e-el', texto, flags=re.IGNORECASE)
|
| 69 |
+
|
| 70 |
+
return texto
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
# --- PROCESAMIENTO ACÚSTICO FFMEPG ---
|
| 74 |
+
def procesar_audio_narrador(ruta):
|
| 75 |
+
"""Efecto catedral estándar para el narrador."""
|
| 76 |
+
if not shutil.which("ffmpeg"):
|
| 77 |
+
return ruta
|
| 78 |
+
|
| 79 |
+
filtros = [
|
| 80 |
+
"atempo=0.22",
|
| 81 |
+
"bass=g=8:f=133",
|
| 82 |
+
"treble=g=-22",
|
| 83 |
+
"vibrato=f=3.5:d=0.55",
|
| 84 |
+
"aecho=0.75:0.70:120|240:0.4|0.25"
|
| 85 |
+
]
|
| 86 |
+
salida = ruta[:-4] + "_narrador.mp3"
|
| 87 |
+
cmd = ["ffmpeg", "-y", "-i", ruta, "-af", ", ".join(filtros), "-ac", "2", salida]
|
| 88 |
+
try:
|
| 89 |
+
subprocess.run(cmd, check=True, capture_output=True, timeout=60)
|
| 90 |
+
if os.path.getsize(salida) > 0:
|
| 91 |
+
return salida
|
| 92 |
+
except Exception:
|
| 93 |
+
pass
|
| 94 |
+
return ruta
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
def procesar_audio_divino(ruta):
|
| 98 |
+
"""
|
| 99 |
+
Efecto ACÚSTICO SAGRADO Y PROFUNDO para la Voz de Dios.
|
| 100 |
+
Tono más grave, resonancia sub-bass, eco más amplio y pureza vocal.
|
| 101 |
+
"""
|
| 102 |
+
if not shutil.which("ffmpeg"):
|
| 103 |
+
return ruta
|
| 104 |
+
|
| 105 |
+
filtros = [
|
| 106 |
+
"atempo=0.80", # Más pausado y solemne
|
| 107 |
+
"asetrate=44100*0.45", # Bajar tono (más grave y imponente)
|
| 108 |
+
"aresample=144000", # Reajustar sample rate
|
| 109 |
+
"bass=g=15:f=200", # Graves profundos sub-bass
|
| 110 |
+
"treble=g=-10", # Sonido cálido, sin aristas agudas
|
| 111 |
+
|
| 112 |
+
]
|
| 113 |
+
salida = ruta[:-4] + "_divino.mp3"
|
| 114 |
+
cmd = ["ffmpeg", "-y", "-i", ruta, "-af", ", ".join(filtros), "-ac", "2", salida]
|
| 115 |
+
try:
|
| 116 |
+
subprocess.run(cmd, check=True, capture_output=True, timeout=60)
|
| 117 |
+
if os.path.getsize(salida) > 0:
|
| 118 |
+
return salida
|
| 119 |
+
except Exception:
|
| 120 |
+
pass
|
| 121 |
+
return ruta
|
| 122 |
+
|
| 123 |
+
|
| 124 |
+
def concatenar_audios(lista_rutas):
|
| 125 |
+
"""Concatena varios archivos MP3 en uno solo en orden estricto."""
|
| 126 |
+
if not lista_rutas:
|
| 127 |
+
return None
|
| 128 |
+
if len(lista_rutas) == 1:
|
| 129 |
+
return lista_rutas[0]
|
| 130 |
+
|
| 131 |
+
salida_final = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False).name
|
| 132 |
+
list_file = tempfile.NamedTemporaryFile(suffix=".txt", mode="w", delete=False, encoding="utf-8")
|
| 133 |
+
|
| 134 |
+
for r in lista_rutas:
|
| 135 |
+
list_file.write(f"file '{os.path.abspath(r)}'\n")
|
| 136 |
+
list_file.close()
|
| 137 |
+
|
| 138 |
+
cmd = ["ffmpeg", "-y", "-f", "concat", "-safe", "0", "-i", list_file.name, "-c", "copy", salida_final]
|
| 139 |
+
try:
|
| 140 |
+
subprocess.run(cmd, check=True, capture_output=True, timeout=120)
|
| 141 |
+
os.remove(list_file.name)
|
| 142 |
+
return salida_final
|
| 143 |
+
except Exception:
|
| 144 |
+
if os.path.exists(list_file.name):
|
| 145 |
+
os.remove(list_file.name)
|
| 146 |
+
return lista_rutas[0]
|
| 147 |
+
|
| 148 |
+
|
| 149 |
+
def mezclar_fondo(voz_path):
|
| 150 |
+
if not (os.path.exists(FONDO) and shutil.which("ffmpeg")):
|
| 151 |
+
return voz_path
|
| 152 |
+
salida = voz_path[:-4] + "_mix.mp3"
|
| 153 |
+
filtro = (f"[1:a]volume={FONDO_VOL}[m];"
|
| 154 |
+
f"[0:a][m]amix=inputs=2:duration=first:normalize=0[a]")
|
| 155 |
+
cmd = ["ffmpeg", "-y", "-i", voz_path, "-stream_loop", "-1", "-i", FONDO,
|
| 156 |
+
"-filter_complex", filtro, "-map", "[a]", salida]
|
| 157 |
+
try:
|
| 158 |
+
subprocess.run(cmd, check=True, capture_output=True, timeout=120)
|
| 159 |
+
if os.path.getsize(salida) > 0:
|
| 160 |
+
return salida
|
| 161 |
+
except Exception:
|
| 162 |
+
pass
|
| 163 |
+
return voz_path
|
| 164 |
+
|
| 165 |
+
|
| 166 |
+
# --- SEPARACIÓN DE BLOQUES (NARRADOR vs DIOS) ---
|
| 167 |
+
def segmentar_texto_divino(texto):
|
| 168 |
+
"""
|
| 169 |
+
Detecta patrones como 'Y dijo Dios: ...' o 'Dijo Dios ...'
|
| 170 |
+
y divide el texto en partes ('narrador' o 'dios').
|
| 171 |
+
"""
|
| 172 |
+
patron = r'((?:Y\s+dijo\s+Dios|dijo\s+Dios|Y\s+llamó\s+Dios|Y\s+bendijo\s+Dios)[^:,.–—]*[:,.–—]?\s*)([^.\n]+)'
|
| 173 |
+
bloques = []
|
| 174 |
+
ultimo_idx = 0
|
| 175 |
+
|
| 176 |
+
for match in re.finditer(patron, texto, flags=re.IGNORECASE):
|
| 177 |
+
start, end = match.span()
|
| 178 |
+
if start > ultimo_idx:
|
| 179 |
+
bloques.append(("narrador", texto[ultimo_idx:start]))
|
| 180 |
+
|
| 181 |
+
intro_dios = match.group(1) # Ej: "Y dijo Dios: "
|
| 182 |
+
palabras_dios = match.group(2) # Ej: "Sea la luz"
|
| 183 |
+
|
| 184 |
+
bloques.append(("narrador", intro_dios))
|
| 185 |
+
bloques.append(("dios", palabras_dios))
|
| 186 |
+
ultimo_idx = end
|
| 187 |
+
|
| 188 |
+
if ultimo_idx < len(texto):
|
| 189 |
+
bloques.append(("narrador", texto[ultimo_idx:]))
|
| 190 |
+
|
| 191 |
+
return bloques if bloques else [("narrador", texto)]
|
| 192 |
+
|
| 193 |
+
|
| 194 |
+
# --- GENERACIÓN DE AUDIO DINÁMICA ---
|
| 195 |
+
async def generar_bloque_audio(texto, es_divino):
|
| 196 |
+
texto_tts = transformar_a_canto_gregoriano_hebraico(texto) if CANTO_GREGORIANO_HEBRAICO else texto
|
| 197 |
+
ruta = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False).name
|
| 198 |
+
|
| 199 |
+
voz = VOZ_DIVINA if es_divino else VOZ_NARRADOR
|
| 200 |
+
pitch = "-5Hz" if es_divino else VOZ_PITCH
|
| 201 |
+
rate = "-9%" if es_divino else VOZ_RATE
|
| 202 |
+
|
| 203 |
+
try:
|
| 204 |
+
com = edge_tts.Communicate(texto_tts[:4000], voz, rate=rate, pitch=pitch)
|
| 205 |
+
await com.save(ruta)
|
| 206 |
+
except Exception:
|
| 207 |
+
return None
|
| 208 |
+
|
| 209 |
+
if es_divino:
|
| 210 |
+
return procesar_audio_divino(ruta)
|
| 211 |
+
else:
|
| 212 |
+
return procesar_audio_narrador(ruta)
|
| 213 |
+
|
| 214 |
+
|
| 215 |
+
async def a_voz(texto, idioma):
|
| 216 |
+
if not texto or not texto.strip():
|
| 217 |
+
return None
|
| 218 |
+
|
| 219 |
+
if idioma == "he":
|
| 220 |
+
ruta = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False).name
|
| 221 |
+
try:
|
| 222 |
+
com = edge_tts.Communicate(texto[:4000], VOZ_HE, rate=VOZ_RATE, pitch=VOZ_PITCH)
|
| 223 |
+
await com.save(ruta)
|
| 224 |
+
return mezclar_fondo(ruta)
|
| 225 |
+
except Exception:
|
| 226 |
+
return None
|
| 227 |
+
|
| 228 |
+
# Procesamiento para Español con Voz Divina Diferenciada
|
| 229 |
+
bloques = segmentar_texto_divino(texto)
|
| 230 |
+
audios_segmentos = []
|
| 231 |
+
|
| 232 |
+
for tipo, contenido in bloques:
|
| 233 |
+
if contenido.strip():
|
| 234 |
+
es_divino = (tipo == "dios")
|
| 235 |
+
audio_seg = await generar_bloque_audio(contenido, es_divino)
|
| 236 |
+
if audio_seg:
|
| 237 |
+
audios_segmentos.append(audio_seg)
|
| 238 |
+
|
| 239 |
+
audio_final = concatenar_audios(audios_segmentos)
|
| 240 |
+
if audio_final:
|
| 241 |
+
audio_final = mezclar_fondo(audio_final)
|
| 242 |
+
|
| 243 |
+
return audio_final
|
| 244 |
+
|
| 245 |
+
|
| 246 |
+
async def leer_es(t):
|
| 247 |
+
return await a_voz(t, "es")
|
| 248 |
+
|
| 249 |
+
|
| 250 |
+
async def leer_he(t):
|
| 251 |
+
return await a_voz(t, "he")
|
| 252 |
+
|
| 253 |
+
|
| 254 |
+
# --- CARGA Y MANEJO DE LIBROS ---
|
| 255 |
+
# Nombres de libros en español, para mostrar lindo en el desplegable
|
| 256 |
+
# aunque el archivo se llame en inglés (Genesis.json, Exodus.json, etc.)
|
| 257 |
+
NOMBRES_ES = {
|
| 258 |
+
"Genesis": "Génesis", "Exodus": "Éxodo", "Leviticus": "Levítico",
|
| 259 |
+
"Numbers": "Números", "Deuteronomy": "Deuteronomio", "Joshua": "Josué",
|
| 260 |
+
"Judges": "Jueces", "Ruth": "Rut", "1_Samuel": "1 Samuel", "2_Samuel": "2 Samuel",
|
| 261 |
+
"1_Kings": "1 Reyes", "2_Kings": "2 Reyes", "1_Chronicles": "1 Crónicas",
|
| 262 |
+
"2_Chronicles": "2 Crónicas", "Ezra": "Esdras", "Nehemiah": "Nehemías",
|
| 263 |
+
"Esther": "Ester", "Job": "Job", "Psalms": "Salmos", "Proverbs": "Proverbios",
|
| 264 |
+
"Ecclesiastes": "Eclesiastés", "Song_of_Songs": "Cantar de los Cantares",
|
| 265 |
+
"Isaiah": "Isaías", "Jeremiah": "Jeremías", "Lamentations": "Lamentaciones",
|
| 266 |
+
"Ezekiel": "Ezequiel", "Daniel": "Daniel", "Hosea": "Oseas", "Joel": "Joel",
|
| 267 |
+
"Amos": "Amós", "Obadiah": "Abdías", "Jonah": "Jonás", "Micah": "Miqueas",
|
| 268 |
+
"Nahum": "Nahúm", "Habakkuk": "Habacuc", "Zephaniah": "Sofonías",
|
| 269 |
+
"Haggai": "Hageo", "Zechariah": "Zacarías", "Malachi": "Malaquías",
|
| 270 |
+
}
|
| 271 |
+
|
| 272 |
+
|
| 273 |
+
def _convertir_traduccion_plana(book_id: str, data: dict) -> dict:
|
| 274 |
+
"""
|
| 275 |
+
Convierte el formato plano que genera el traductor
|
| 276 |
+
({"translation": "...", "missing_strong_numbers": [...]})
|
| 277 |
+
a la estructura de capítulos/versículos que usa la Centralita.
|
| 278 |
+
No tenemos separación real de capítulos en este formato, así que
|
| 279 |
+
todo el libro se muestra como Capítulo 1, con una línea = un versículo.
|
| 280 |
+
"""
|
| 281 |
+
texto = data.get("translation", "") or ""
|
| 282 |
+
lineas = [l.strip() for l in texto.split("\n") if l.strip()]
|
| 283 |
+
versiculos = [{"n": i + 1, "he": "", "es": linea} for i, linea in enumerate(lineas)]
|
| 284 |
+
return {
|
| 285 |
+
"id": book_id,
|
| 286 |
+
"es": NOMBRES_ES.get(book_id, book_id),
|
| 287 |
+
"heb": "",
|
| 288 |
+
"capitulos": {"1": versiculos},
|
| 289 |
+
}
|
| 290 |
+
|
| 291 |
+
|
| 292 |
+
def cargar_libros():
|
| 293 |
+
libros = {}
|
| 294 |
+
for ruta in sorted(glob.glob(os.path.join(CARPETA, "*.json"))):
|
| 295 |
+
try:
|
| 296 |
+
with open(ruta, encoding="utf-8") as f:
|
| 297 |
+
d = json.load(f)
|
| 298 |
+
book_id = os.path.splitext(os.path.basename(ruta))[0]
|
| 299 |
+
if "capitulos" in d:
|
| 300 |
+
# Formato completo ya listo (id/es/heb/capitulos)
|
| 301 |
+
lid = d.get("id") or book_id
|
| 302 |
+
libros[lid] = d
|
| 303 |
+
elif "translation" in d:
|
| 304 |
+
# Formato plano del traductor (translation + missing_strong_numbers)
|
| 305 |
+
libros[book_id] = _convertir_traduccion_plana(book_id, d)
|
| 306 |
+
else:
|
| 307 |
+
print(f"Formato no reconocido en {ruta}, se omite.")
|
| 308 |
+
except Exception as e:
|
| 309 |
+
print("No pude leer", ruta, e)
|
| 310 |
+
return libros
|
| 311 |
+
|
| 312 |
+
|
| 313 |
+
LIBROS = cargar_libros()
|
| 314 |
+
ORDEN = ["Genesis", "Exodus", "Leviticus", "Numbers", "Deuteronomy"]
|
| 315 |
+
IDS = [i for i in ORDEN if i in LIBROS] + [i for i in LIBROS if i not in ORDEN]
|
| 316 |
+
NOMBRES = [LIBROS[i].get("es", i) for i in IDS] or ["(sube tus libros)"]
|
| 317 |
+
|
| 318 |
+
|
| 319 |
+
def id_por_nombre(nombre):
|
| 320 |
+
for i in IDS:
|
| 321 |
+
if LIBROS[i].get("es", i) == nombre:
|
| 322 |
+
return i
|
| 323 |
+
return IDS[0] if IDS else None
|
| 324 |
+
|
| 325 |
+
|
| 326 |
+
def versiculos(lid, cap):
|
| 327 |
+
return LIBROS.get(lid, {}).get("capitulos", {}).get(str(cap), [])
|
| 328 |
+
|
| 329 |
+
|
| 330 |
+
def render(nombre, cap):
|
| 331 |
+
lid = id_por_nombre(nombre)
|
| 332 |
+
if not lid:
|
| 333 |
+
return "<div class='pasaje'><p>Aún no has subido libros a la carpeta <b>libros/</b>.</p></div>", "", ""
|
| 334 |
+
try:
|
| 335 |
+
cap = max(1, int(float(cap or 1)))
|
| 336 |
+
except Exception:
|
| 337 |
+
cap = 1
|
| 338 |
+
d = LIBROS[lid]
|
| 339 |
+
vs = versiculos(lid, cap)
|
| 340 |
+
if not vs:
|
| 341 |
+
return f"<div class='pasaje'><p>No hay texto para {html.escape(d.get('es',''))} {cap}.</p></div>", "", ""
|
| 342 |
+
filas = [f"<div class='cab'><span class='h'>{html.escape(d.get('heb',''))}</span><br>"
|
| 343 |
+
f"{html.escape(d.get('es',''))} {cap}</div>"]
|
| 344 |
+
plano_es, plano_he = [], []
|
| 345 |
+
for v in vs:
|
| 346 |
+
he = html.escape(v.get("he", "")); es = html.escape(v.get("es", ""))
|
| 347 |
+
fila = f"<div class='verso'><span class='num'>{v.get('n','')}</span>"
|
| 348 |
+
if he:
|
| 349 |
+
fila += f"<span class='he'>{he}</span>"; plano_he.append(v.get("he", ""))
|
| 350 |
+
if es:
|
| 351 |
+
fila += f"<span class='es'>{es}</span>"; plano_es.append(v.get("es", ""))
|
| 352 |
+
fila += "</div>"
|
| 353 |
+
filas.append(fila)
|
| 354 |
+
return f"<div class='pasaje'>{''.join(filas)}</div>", " ".join(plano_es), " ".join(plano_he)
|
| 355 |
+
|
| 356 |
+
|
| 357 |
+
def estudiar(mensaje, historial, contexto):
|
| 358 |
+
if not GROQ_KEY:
|
| 359 |
+
return "Para el estudio, añade el Secret GROQ_API_KEY en el Space."
|
| 360 |
+
sistema = (
|
| 361 |
+
"Eres un compañero de estudio de la Torá que responde en español, cálido y honesto. "
|
| 362 |
+
"Ofreces el sentido literal (peshat), contexto histórico y lingüístico, capas de la "
|
| 363 |
+
"tradición (midrash, Rashi cuando venga al caso) y reflexión espiritual. Vas al grano.\n\n" + contexto
|
| 364 |
+
)
|
| 365 |
+
mensajes = [{"role": "system", "content": sistema}]
|
| 366 |
+
for m in (historial or []):
|
| 367 |
+
if isinstance(m, dict) and m.get("role") in ("user", "assistant"):
|
| 368 |
+
mensajes.append({"role": m["role"], "content": m["content"]})
|
| 369 |
+
mensajes.append({"role": "user", "content": mensaje})
|
| 370 |
+
cuerpo = {"model": MODELO_ESTUDIA, "temperature": 0.4, "max_tokens": 1200, "messages": mensajes}
|
| 371 |
+
try:
|
| 372 |
+
data = json.dumps(cuerpo).encode("utf-8")
|
| 373 |
+
req = urllib.request.Request(GROQ_URL, data=data, method="POST")
|
| 374 |
+
req.add_header("Content-Type", "application/json")
|
| 375 |
+
req.add_header("Authorization", f"Bearer {GROQ_KEY}")
|
| 376 |
+
with urllib.request.urlopen(req, timeout=90) as r:
|
| 377 |
+
d = json.load(r)
|
| 378 |
+
return (d["choices"][0]["message"]["content"] or "").strip()
|
| 379 |
+
except Exception as e:
|
| 380 |
+
return f"No pude consultar a la IA ahora mismo ({e})."
|
| 381 |
+
|
| 382 |
+
|
| 383 |
+
# ============================ INTERFAZ GRADIO ============================
|
| 384 |
+
with gr.Blocks(theme=fenix_theme(), css=FENIX_CSS, title="Centralita Torá") as demo:
|
| 385 |
+
st_es = gr.State("")
|
| 386 |
+
st_he = gr.State("")
|
| 387 |
+
st_nombre = gr.State(NOMBRES[0])
|
| 388 |
+
st_cap = gr.State(1)
|
| 389 |
+
|
| 390 |
+
gr.HTML("<div id='cabecera'><div class='estrella'>✡</div>"
|
| 391 |
+
"<h1>Centralita Torá</h1><p>hebreo y español · canto Trope con voz celectial diferenciada</p></div>")
|
| 392 |
+
|
| 393 |
+
with gr.Tab("Leer"):
|
| 394 |
+
with gr.Row():
|
| 395 |
+
dd = gr.Dropdown(NOMBRES, value=NOMBRES[0], label="Libro")
|
| 396 |
+
ncap = gr.Number(value=1, precision=0, label="Capítulo", minimum=1)
|
| 397 |
+
ver_btn = gr.Button("Ver capítulo", variant="primary")
|
| 398 |
+
salida = gr.HTML()
|
| 399 |
+
with gr.Row():
|
| 400 |
+
b_es = gr.Button("🔊 Canto Trope + Voz celectial", variant="secondary")
|
| 401 |
+
b_he = gr.Button("🔊 עברית (Hebreo)", variant="secondary")
|
| 402 |
+
audio = gr.Audio(label="Audio Litúrgico", autoplay=True)
|
| 403 |
+
|
| 404 |
+
def ver(nombre, cap):
|
| 405 |
+
md, es, he = render(nombre, cap)
|
| 406 |
+
return md, es, he, nombre, cap
|
| 407 |
+
|
| 408 |
+
ver_btn.click(ver, [dd, ncap], [salida, st_es, st_he, st_nombre, st_cap])
|
| 409 |
+
b_es.click(leer_es, st_es, audio)
|
| 410 |
+
b_he.click(leer_he, st_he, audio)
|
| 411 |
+
|
| 412 |
+
with gr.Tab("Buscar"):
|
| 413 |
+
q = gr.Textbox(label="Palabra o frase (hebreo o español)", placeholder="ej. luz / אור")
|
| 414 |
+
q_btn = gr.Button("Buscar", variant="primary")
|
| 415 |
+
q_out = gr.HTML()
|
| 416 |
+
|
| 417 |
+
def buscar(texto):
|
| 418 |
+
t = (texto or "").strip().lower()
|
| 419 |
+
if not t:
|
| 420 |
+
return "<p>Escribe algo para buscar.</p>"
|
| 421 |
+
filas = []
|
| 422 |
+
for lid in IDS:
|
| 423 |
+
d = LIBROS[lid]
|
| 424 |
+
for c, vs in d.get("capitulos", {}).items():
|
| 425 |
+
for v in vs:
|
| 426 |
+
campos = " ".join(filter(None, [v.get("es"), v.get("he")])).lower()
|
| 427 |
+
if t in campos:
|
| 428 |
+
ref = f"{d.get('es', lid)} {c}:{v.get('n')}"
|
| 429 |
+
txt = html.escape(v.get("es") or v.get("he") or "")
|
| 430 |
+
filas.append(f"<div class='resultado'><span class='ref'>{ref}</span> — {txt}</div>")
|
| 431 |
+
if len(filas) >= 120:
|
| 432 |
+
return "".join(filas)
|
| 433 |
+
return "".join(filas) if filas else "<p>Sin resultados en los libros cargados.</p>"
|
| 434 |
+
|
| 435 |
+
q_btn.click(buscar, q, q_out)
|
| 436 |
+
|
| 437 |
+
with gr.Tab("Estudiar"):
|
| 438 |
+
gr.Markdown("La IA usa el capítulo abierto en **Leer** como contexto.")
|
| 439 |
+
chat = gr.Chatbot(type="messages", height=360)
|
| 440 |
+
pin = gr.Textbox(placeholder="Pregunta sobre el pasaje…", label="")
|
| 441 |
+
with gr.Row():
|
| 442 |
+
enviar = gr.Button("Preguntar", variant="primary")
|
| 443 |
+
comentar = gr.Button("Comentar este capítulo", variant="secondary")
|
| 444 |
+
|
| 445 |
+
def responder(mensaje, historial, es, he, nombre, cap):
|
| 446 |
+
mensaje = (mensaje or "").strip()
|
| 447 |
+
if not mensaje:
|
| 448 |
+
return historial, ""
|
| 449 |
+
contexto = f"Pasaje en pantalla — {nombre} {cap}:\nHebreo: {he}\nEspañol: {es}"
|
| 450 |
+
r = estudiar(mensaje, historial, contexto)
|
| 451 |
+
historial = (historial or []) + [
|
| 452 |
+
{"role": "user", "content": mensaje},
|
| 453 |
+
{"role": "assistant", "content": r},
|
| 454 |
+
]
|
| 455 |
+
return historial, ""
|
| 456 |
+
|
| 457 |
+
enviar.click(responder, [pin, chat, st_es, st_he, st_nombre, st_cap], [chat, pin])
|
| 458 |
+
comentar.click(
|
| 459 |
+
lambda h, es, he, n, c: responder("Comenta y ayúdame a estudiar este capítulo.", h, es, he, n, c),
|
| 460 |
+
[chat, st_es, st_he, st_nombre, st_cap], [chat, pin],
|
| 461 |
+
)
|
| 462 |
+
|
| 463 |
+
with gr.Tab("Cómo subir"):
|
| 464 |
+
gr.Markdown(
|
| 465 |
+
"Sube un JSON por libro a la carpeta `libros/`. Para el estudio, añade el Secret `GROQ_API_KEY`."
|
| 466 |
+
)
|
| 467 |
+
|
| 468 |
+
if __name__ == "__main__":
|
| 469 |
+
demo.launch(ssr_mode=False)
|
| 470 |
+
|