TranslaSite

Add app.py with custom glossary integration

app.py

@@ -0,0 +1,98 @@
+import gradio as gr
+from transformers import pipeline
+import re
+
+# Initialize the translation pipeline
+translator = pipeline("translation", model="Helsinki-NLP/opus-mt-tc-big-en-es")
+
+# 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",
+    "Greenfield Project": "Proyecto Greenfield",
+    "Brownfield Project": "Proyecto Brownfield",
+    "Constructability Review": "Revisión de Constructibilidad",
+    "Value Engineering": "Ingeniería de Valor",
+    "Design-Build": "Diseño-Construcción",
+    "Cost Estimation": "Estimación de Costos",
+    "Erosion and Sediment Control (ESC)": "Control de Erosión y Sedimentos (ESC)",
+    "OSHA Compliance": "Cumplimiento con OSHA",
+    "Project Closeout": "Cierre del Proyecto",
+    "Integrated Project Delivery (IPD)": "Entrega Integrada de Proyectos (IPD)",
+    "Prefabrication": "Prefabricación",
+    "HVAC (Heating, Ventilation, Air Conditioning)": "HVAC (Calefacción, Ventilación y Aire Acondicionado)",
+    "Sustainability": "Sostenibilidad",
+    "RFP (Request for Proposal)": "RFP (Solicitud de Propuesta)",
+    "Safety Data Sheet (SDS)": "Hoja de Datos de Seguridad (SDS)",
+    "Load-Bearing Wall": "Muro de Carga",
+    "Foundation Types": "Tipos de Cimientos",
+    "Seismic Retrofitting": "Refuerzo Sísmico",
+    "Fireproofing": "Protección contra Incendios",
+    "Commissioning": "Puesta en Servicio",
+    "Facade": "Fachada",
+    "Thermal Insulation": "Aislamiento Térmico",
+    "Waterproofing": "Impermeabilización",
+    "Zoning": "Zonificación",
+    "EPC (Engineering, Procurement, Construction)": "EPC (Ingeniería, Adquisición y Construcción)",
+    "Project Milestone": "Hito del Proyecto",
+    "Subcontractor": "Subcontratista",
+    "Quality Assurance (QA)": "Aseguramiento de Calidad (QA)",
+    "Quality Control (QC)": "Control de Calidad (QC)",
+    "Baseline Schedule": "Cronograma Base",
+    "Material Takeoff": "Lista de Materiales",
+    "Lifecycle Costs": "Costos de Ciclo de Vida",
+    "Building Envelope": "Envolvente del Edificio",
+    "MEP (Mechanical, Electrical, Plumbing)": "MEP (Mecánica, Eléctrica, Fontanería)",
+    "As-Built Drawings": "Planos Tal Como Construidos",
+    "Hazardous Waste Management": "Gestión de Residuos Peligrosos",
+    "Indoor Air Quality (IAQ)": "Calidad del Aire Interior (IAQ)",
+    "Renewable Energy Systems": "Sistemas de Energía Renovable",
+    "Temporary Works": "Obras Temporales",
+    "Construction Waste Diversion": "Desviación de Residuos de Construcción"
+}
+
+def preprocess_text(text, glossary):
+    # Sort terms by length to handle longer terms first
+    sorted_terms = sorted(glossary.keys(), key=len, reverse=True)
+    for term in sorted_terms:
+        # Use a unique placeholder to avoid conflicts
+        placeholder = f"[[{term}]]"
+        # Use regex to replace whole words only
+        text = re.sub(rf'\b{re.escape(term)}\b', placeholder, text)
+    return text
+
+def postprocess_text(text, glossary):
+    for term, translation in glossary.items():
+        placeholder = f"[[{term}]]"
+        text = text.replace(placeholder, translation)
+    return text
+
+def translate_text(text):
+    # Preprocess the text to replace glossary terms with placeholders
+    preprocessed_text = preprocess_text(text, custom_glossary)
+    # Translate the preprocessed text
+    translated = translator(preprocessed_text)[0]['translation_text']
+    # Postprocess the translated text to replace placeholders with actual terms
+    final_translation = postprocess_text(translated, custom_glossary)
+    return final_translation
+
+# 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()