# app/main.py
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
import numpy as np
from services.sms_service import classify_sms, load_trained_model
from schemas.input_schemas import CosineSimilarityInput, CosineSimilarityOutput
from schemas.input_schemas import EmbeddingInput, EmbeddingOutput

# Initialize FastAPI
app = FastAPI()

# Load the models from the 'models' folder
model, vectorizer = load_trained_model()

# Function to compute cosine similarity
def cosine_similarity(vec1, vec2):
    """
    Compute cosine similarity between two vectors.
    """
    norm1 = np.linalg.norm(vec1)
    norm2 = np.linalg.norm(vec2)
    if norm1 == 0 or norm2 == 0:
        return 0.0  # Prevent division by zero
    return np.dot(vec1, vec2) / (norm1 * norm2)

# 🚀 1️⃣ Homepage Endpoint
@app.get("/")
async def home():
    return {"message": "Welcome to SMS Classification API"}

# 📩 2️⃣ SMS Classification Endpoint
class MessageInput(BaseModel):
    message: str

@app.post("/predict_label/")
async def classify_sms_endpoint(input_data: MessageInput):
    """
    Classify an SMS as either 'Transaction' or 'Offer'.
    """
    try:
        return classify_sms(input_data.message, model, vectorizer)
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Unexpected error: {str(e)}")

# 🔢 3️⃣ Cosine Similarity Endpoint
@app.post("/cosine_similarity/", response_model=CosineSimilarityOutput)
async def compute_similarity(input_data: CosineSimilarityInput):
    """
    Compute cosine similarity between two input texts.
    """
    try:
        # Transform the input texts using the TF-IDF vectorizer
        text1_vectorized = vectorizer.transform([input_data.text1])
        text2_vectorized = vectorizer.transform([input_data.text2])
        
        # Compute the cosine similarity between the two text embeddings
        similarity = cosine_similarity(text1_vectorized.toarray(), text2_vectorized.toarray())
        return CosineSimilarityOutput(cosine_similarity=round(float(similarity), 4))
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Error computing similarity: {str(e)}")

# 🧠 4️⃣ Get Embedding of Text Message
@app.post("/get_embedding/", response_model=EmbeddingOutput)
async def get_embedding(input_data: EmbeddingInput):
    """
    Get the embedding (vector representation) of an input text message.
    """
    try:
        # Transform the input text using the TF-IDF vectorizer
        text_embedding = vectorizer.transform([input_data.message]).toarray().tolist()
        return EmbeddingOutput(embedding=text_embedding[0])
    except Exception as e:
        raise HTTPException(status_code=500, detail=f"Error generating embedding: {str(e)}")