app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline

# Load your fine-tuned CamemBERT NER model
model_name = "GEODE/camembert-base-edda-span-classification"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForTokenClassification.from_pretrained(model_name)


ner_pipeline = pipeline("token-classification", model=model, tokenizer=tokenizer, aggregation_strategy="simple")

def extract_coordinates(text):
    # Run NER
    entities = ner_pipeline(text)
    
    # Collect coordinate entities
    coords_text = [ent['word'] for ent in entities if ent['entity_group'] == 'Latlong']
    
    if coords_text:
        return "\n".join(coords_text)
    else:
        return "No coordinates found"

with gr.Blocks() as demo:
    gr.Markdown("## Geographic Coordinate Extractor (CamemBERT NER Demo)")
    
    with gr.Row():
        inp = gr.Textbox(
            label="Enter text", 
            placeholder="e.g. * AACH ou ACH, s. f. petite ville d'Allemagne dans le cercle de Souabe, près de la source de l'Aach. Long. 26. 57. lat. 47. 55.",
            lines=3
        )
    
    with gr.Row():
        out_text = gr.Textbox(label="Extracted coordinates")
        #out_map = gr.Map(label="Location on Map")
    
    run_btn = gr.Button("Extract & Show")
    run_btn.click(fn=extract_coordinates, inputs=inp, outputs=out_text)

# Launch
if __name__ == "__main__":
    demo.launch()