app.py

# app_nlp.py
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel, Field
from typing import Dict

from transformers import pipeline

APP_VERSION = "1.0.0"

app = FastAPI(
    title="Class 8 - NLP Model Serving (Sentiment)",
    version=APP_VERSION,
    description="Serve a small pretrained NLP model via FastAPI."
)

# A small, commonly used sentiment model:
# (It will download on first run.)
SENTIMENT_MODEL_NAME = "distilbert-base-uncased-finetuned-sst-2-english"

# Load once at startup (good pattern)
nlp = pipeline("sentiment-analysis", model=SENTIMENT_MODEL_NAME)

class TextRequest(BaseModel):
    text: str = Field(..., min_length=1, max_length=1000, description="Input text to analyze")

class NLPResponse(BaseModel):
    ok: bool
    model_version: str
    model_name: str
    result: Dict[str, float | str]

@app.get("/health")
def health():
    return {"status": "ok", "model_loaded": True, "model_name": SENTIMENT_MODEL_NAME}

@app.post("/v1/predict", response_model=NLPResponse)
def predict(payload: TextRequest):
    try:
        out = nlp(payload.text)[0]  # {'label': 'POSITIVE', 'score': ...}
        return NLPResponse(
            ok=True,
            model_version=APP_VERSION,
            model_name=SENTIMENT_MODEL_NAME,
            result={"label": out["label"], "score": float(out["score"])}
        )
    except Exception:
        raise HTTPException(status_code=500, detail="NLP inference failed")