from sentence_transformers import SentenceTransformer
import numpy as np

courses = [
    {
        "title": "Data Science with Python",
        "description": "Learn the basics of data science, including machine learning and Python programming.",
        "curriculum": "Python, Pandas, Numpy, Machine Learning, Data Visualization"
    },
    {
        "title": "Introduction to Machine Learning",
        "description": "A beginner-friendly course to understand machine learning algorithms.",
        "curriculum": "Supervised Learning, Unsupervised Learning, Regression, Classification"
    },
    {
        "title": "Deep Learning Fundamentals",
        "description": "An introductory course on deep learning and neural networks.",
        "curriculum": "Neural Networks, CNN, RNN, Deep Learning"
    },
]

course_texts = [course["title"] + " " + course["description"] + " " + course["curriculum"] for course in courses]

# Load the model with a fallback option
try:
    model = SentenceTransformer('paraphrase-MiniLM-L6-v2')
except:
    model = SentenceTransformer('all-MiniLM-L6-v2')

# Encode the course texts
course_embeddings = model.encode(course_texts)

def search_courses(query):
    query_embedding = model.encode([query])[0]
    similarities = np.dot(course_embeddings, query_embedding)
    sorted_course_indices = np.argsort(similarities)[::-1]

    results = []
    for idx in sorted_course_indices[:3]:  # Top 3 results
        results.append({
            "title": courses[idx]["title"],
            "description": courses[idx]["description"],
            "curriculum": courses[idx]["curriculum"]
        })

    return results