main.py

import joblib
import os
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel

# ------------------------------------------------------------------
# App setup
# ------------------------------------------------------------------
app = FastAPI(
    title="GitHub Spam Detector API",
    description="Predicts whether a GitHub comment is spam (1) or not spam (0). "
                "Trained on 100k+ real GitHub comments using TF-IDF + LinearSVC.",
    version="1.0.0",
)

# ------------------------------------------------------------------
# Model loading (once at startup, never per-request)
# ------------------------------------------------------------------
MODEL_PATH = os.getenv("MODEL_PATH", "spam_detector_model_cv.pkl")

@app.on_event("startup")
def load_model():
    global model
    if not os.path.exists(MODEL_PATH):
        raise RuntimeError(f"Model file not found at: {MODEL_PATH}")
    model = joblib.load(MODEL_PATH)
    print(f"Model loaded successfully from {MODEL_PATH}")


# ------------------------------------------------------------------
# Request / Response schemas
# ------------------------------------------------------------------
class PredictRequest(BaseModel):
    text: str

    class Config:
        json_schema_extra = {
            "example": {
                "text": "Please fix the bug at line 56, it causes a null pointer exception."
            }
        }


class PredictResponse(BaseModel):
    text: str
    prediction: int          # 0 = Not Spam, 1 = Spam
    label: str               # Human-readable label


# ------------------------------------------------------------------
# Endpoints
# ------------------------------------------------------------------
@app.get("/", tags=["Health"])
def root():
    """Health check endpoint."""
    return {"status": "ok", "message": "Spam Detector API is running."}


@app.post("/predict", response_model=PredictResponse, tags=["Prediction"])
def predict(request: PredictRequest):
    """
    Predict whether a piece of text is spam or not.

    - **text**: The comment / text to classify.

    Returns:
    - **prediction**: `0` (Not Spam) or `1` (Spam)
    - **label**: Human-readable string `"spam"` or `"not_spam"`
    """
    if not request.text or not request.text.strip():
        raise HTTPException(status_code=422, detail="Input text cannot be empty.")

    raw_pred = model.predict([request.text])[0]
    prediction = int(raw_pred)
    label = "spam" if prediction == 1 else "not_spam"

    return PredictResponse(
        text=request.text,
        prediction=prediction,
        label=label,
    )