app.py

import gradio as gr
import re
import torch
from transformers import pipeline

# Initialize the translation pipeline
def initialize_translator():
    try:
        device = 0 if torch.cuda.is_available() else -1  # Use GPU if available, otherwise CPU
        return pipeline("translation", model="Helsinki-NLP/opus-mt-tc-big-en-es", device=device)
    except Exception as e:
        raise RuntimeError(f"Failed to initialize translation pipeline: {e}")

translator = initialize_translator()

# Define the custom glossary
custom_glossary = {
    "Change Order": "Orden de Cambio",
    "RFI (Request for Information)": "RFI (Solicitud de Información)",
    "Submittals": "Envíos",
    "Punch List": "Lista de Reparaciones",
    "GMP (Guaranteed Maximum Price)": "GMP (Precio Máximo Garantizado)",
    "Critical Path Method (CPM)": "Método de Ruta Crítica (CPM)",
    "LEED Certification": "Certificación LEED",
    "BIM (Building Information Modeling)": "BIM (Modelado de Información de Construcción)",
    "Retainage": "Retención",
    "Precast Concrete": "Hormigón Prefabricado",
    "Aggregate": "Áridos",
    "Allowance": "Asignación",
    "Apprenticeship": "Aprendizaje",
    "Backfilling": "Relleno",
    "Beam": "Viga",
    "Blueprint": "Plano",
    "Bollard": "Bolardo",
    "Bracing": "Arriostramiento",
    "Brick Masonry": "Albañilería de Ladrillo",
    "Budget Overrun": "Sobrecosto Presupuestario",
    "Building Code": "Código de Construcción",
    "Cantilever": "Voladizo",
    "Caulking": "Sellado",
    "Ceiling Joist": "Vigueta de Techo",
    "Cement": "Cemento",
    "Certificate of Occupancy": "Certificado de Ocupación",
    "Chain Link Fence": "Cerca de Malla",
    "Chipping": "Picado",
    "Cladding": "Revestimiento",
    "Clear Span": "Luz Libre",
    "Construction Drawings": "Planos de Construcción",
    "Construction Schedule": "Cronograma de Construcción",
    "Construction Staging": "Escenario de Construcción",
    "Contingency": "Contingencia",
    "Contract Bond": "Fianza de Contrato",
    "Curtain Wall": "Muro Cortina",
    "Damp Proofing": "Impermeabilización",
    "Dead Load": "Carga Muerta",
    "Demolition": "Demolición",
    "Design Development": "Desarrollo de Diseño",
    "Detail Drawing": "Dibujo Detallado",
    "Drilled Pier": "Pilote Perforado",
    "Drywall": "Tablaroca",
    "Earthmoving": "Movimiento de Tierras",
    "Elevation": "Elevación",
    "Emergency Egress": "Salida de Emergencia",
    "Energy Modeling": "Modelado Energético",
    "Excavation": "Excavación",
    "Expansion Joint": "Junta de Expansión",
    "Exterior Finish": "Acabado Exterior",
    "Fasteners": "Fijaciones",
    "Field Verification": "Verificación en Campo",
    "Final Walkthrough": "Inspección Final",
    "Finishing": "Acabado",
    "Fire Barrier": "Barrera Contra Incendios",
    "Flange": "Brida",
    "Floor Slab": "Losa de Piso",
    "Footing": "Zapata",
    "Foundation Wall": "Muro de Cimentación",
    "Framing": "Estructura",
    "Frost Line": "Línea de Heladas",
    "Grade Beam": "Viga de Cimentación",
    "Grading": "Nivelación",
    "Ground Penetrating Radar": "Radar de Penetración Terrestre",
    "Hardscape": "Paisajismo Duro",
    "HVAC Ductwork": "Conductos HVAC",
    "I-Beam": "Viga en I",
    "Insulation Rating": "Clasificación de Aislamiento",
    "Interior Finish": "Acabado Interior",
    "Joist": "Vigueta",
    "Keyway": "Entalladura",
    "Lath": "Listón",
    "Load Testing": "Prueba de Carga",
    "Louver": "Persiana",
    "Masonry": "Albañilería",
    "Moisture Barrier": "Barrera de Humedad",
    "Mortar": "Mortero",
    "Overhang": "Voladizo",
    "Parapet": "Pretil",
    "Pile": "Pilote",
    "Plaster": "Yeso",
    "Plumbing Fixtures": "Accesorios de Plomería",
    "Post-Tension Slab": "Losa Postensada",
    "Prefabricated Components": "Componentes Prefabricados",
    "Safety Harness": "Arnés de Seguridad",
    "Scaffolding": "Andamios",
    "Structural Integrity": "Integridad Estructural",
    "Structural Steel": "Acero Estructural",
    "Temporary Fence": "Cerca Temporal",
    "Zoning Ordinance": "Ordenanza de Zonificación",
    "Thermal Insulation": "Aislamiento Térmico",
    "Sealant": "Sellador",
    "Retaining Wall": "Muro de Contención",
    "Screw Anchor": "Anclaje de Tornillo",
    "Rain Screen": "Pantalla de Lluvia",
    "Facade Treatment": "Tratamiento de Fachada",
    "Fireproofing": "Protección contra Incendios",
}

# Preprocess text to replace glossary terms with placeholders
def preprocess_text(text, glossary):
    sorted_terms = sorted(glossary.keys(), key=len, reverse=True)
    for term in sorted_terms:
        placeholder = f"[[{term}]]"
        text = re.sub(rf'\b{re.escape(term)}\b', placeholder, text, flags=re.IGNORECASE)
    return text

# Postprocess text to replace placeholders with glossary terms
def postprocess_text(text, glossary):
    for term, translation in glossary.items():
        placeholder = f"[[{term}]]"
        text = text.replace(placeholder, translation)
    return text

# Translate text
def translate_text(text):
    try:
        # Preprocess the text
        preprocessed_text = preprocess_text(text, custom_glossary)
        # Translate the preprocessed text
        translated = translator(preprocessed_text)[0]['translation_text']
        # Postprocess the text
        final_translation = postprocess_text(translated, custom_glossary)
        return final_translation
    except Exception as e:
        return f"An error occurred during translation: {e}"

# Create the Gradio interface
iface = gr.Interface(
    fn=translate_text,
    inputs=gr.Textbox(lines=5, label="Input Text (English)"),
    outputs=gr.Textbox(label="Translated Text (Spanish)"),
    title="English to Spanish Translator",
    description="Enter English text to translate it into Spanish, with accurate translation of construction and project management terms.",
)

# Launch the interface
if __name__ == "__main__":
    iface.launch()