import gradio as gr
from transformers import pipeline

# Load the multilingual NER pipeline
ner = pipeline("ner", model="Davlan/xlm-roberta-base-ner-hrl", grouped_entities=True)

# Inference function
def extract_entities(text):
    results = ner(text)
    return [(ent['word'], ent['entity_group']) for ent in results]

# User instructions and placeholder to show above app
instructions = """
### 🌍 Multilingual Named Entity Recognition (NER) App

🔹 Enter a sentence in **any supported language** (English, French, Hindi, Spanish, etc.)
🔹 The app will detect and classify entities such as:
   - `PER` (Person)
   - `LOC` (Location)
   - `ORG` (Organization)
   - `MISC` (Miscellaneous)

⚠️ Some languages like **Japanese** may show minor inaccuracies.

**Example Input:**
`Barack Obama was born in Hawaii and served as the 44th President of the United States.`

**Expected Output:**
- Barack Obama → PER
- Hawaii → LOC
- United States → LOC
"""

# Gradio interface
with gr.Blocks() as demo:
    gr.Markdown(instructions)
    with gr.Row():
        inp = gr.Textbox(label="Enter Text", placeholder="Type a sentence in any language...", lines=3)
        out = gr.HighlightedText(label="Named Entities")
    btn = gr.Button("Extract Entities")

    btn.click(fn=extract_entities, inputs=inp, outputs=out)

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