import os
import gradio as gr
from fastapi import FastAPI
from fastapi.responses import RedirectResponse
from pydantic import BaseModel

from app.ml.predictor import EmotionPredictor

app = FastAPI(title="Emotion Classifier", version="1.0.0")

MODEL_ID = os.getenv("MODEL_ID", "AfroLogicInsect/emotionClassifier")
predictor = EmotionPredictor(MODEL_ID)


class PredictRequest(BaseModel):
    text: str


class PredictResponse(BaseModel):
    emotion: str
    confidence: float
    all_emotions: dict[str, float]


# --- existing REST endpoint (unchanged) ---

@app.post("/predict", response_model=PredictResponse)
async def predict(request: PredictRequest):
    result = predictor.predict(request.text)
    return PredictResponse(**result)


@app.get("/health")
async def health():
    return {"status": "ok"}


# --- Gradio UI ---

def classify(text: str):
    if not text.strip():
        return "Please enter some text.", {}
    result = predictor.predict(text)
    label = f"{result['emotion'].upper()}  ({result['confidence']:.1%} confidence)"
    return label, result["all_emotions"]


with gr.Blocks(title="Emotion Classifier") as demo:
    gr.Markdown(
        """
        # Emotion Classifier
        Fine-tuned DistilBERT that detects six emotions: anger, fear, joy, love, sadness, surprise.
        """
    )
    with gr.Row():
        with gr.Column():
            text_input = gr.Textbox(
                label="Input text",
                placeholder="Type something…",
                lines=3,
            )
            submit_btn = gr.Button("Classify", variant="primary")
        with gr.Column():
            label_output = gr.Textbox(label="Prediction")
            scores_output = gr.Label(label="All emotion scores", num_top_classes=6)

    submit_btn.click(
        fn=classify,
        inputs=text_input,
        outputs=[label_output, scores_output],
    )
    gr.Examples(
        examples=[
            ["I am feeling very happy today!"],
            ["This is so frustrating, nothing works."],
            ["I miss my family so much."],
            ["Wow, I did not expect that at all!"],
        ],
        inputs=text_input,
    )
    gr.Markdown(
        """
        ---
        **API endpoint** also available at `/predict` — see `/docs` for the OpenAPI spec.
        """
    )

# Mount Gradio on FastAPI; the UI lives at "/" and the REST API at "/predict"
app = gr.mount_gradio_app(app, demo, path="/")