Worker.py

# Node Worker menggunakan ZeroMQ untuk memproses query
import zmq
import pymysql
from sklearn.feature_extraction.text import TfidfVectorizer

# Koneksi database lokal untuk worker
def get_db_connection():
    return pymysql.connect(
        host="localhost",
        user="root",
        password="",
        database="db_se",
        cursorclass=pymysql.cursors.DictCursor
    )

def process_query(query):

    connection = get_db_connection()
    try:
        with connection.cursor() as cursor:
            cursor.execute("SELECT document_id, judul, isi_berita FROM documents")
            documents = cursor.fetchall()

        # TF-IDF Processing
        texts = [doc['isi_berita'] for doc in documents]
        vectorizer = TfidfVectorizer()
        tfidf_matrix = vectorizer.fit_transform(texts)
        query_vector = vectorizer.transform([query])

        # Calculate cosine similarity
        cosine_similarities = (tfidf_matrix * query_vector.T).toarray().flatten()
        results = [
            {
                "document_id": doc["document_id"],
                "judul": doc["judul"],
                "content": doc["isi_berita"],
                "device": "bagas",
                "score": float(score),
            }
            for doc, score in zip(documents, cosine_similarities) if score > 0
        ]

        return results
    finally:
        connection.close()

def main():
    context = zmq.Context()
    socket = context.socket(zmq.REP)
    socket.bind("tcp://*:5555")  # Worker mendengarkan di port 5555

    print("Worker berjalan dan siap menerima query...")
    while True:
        message = socket.recv_json()
        query = message.get("query")
        if "query" in message:
            query = message.get("query")
            results = process_query(query)
            socket.send_json({"results": results})
        elif "document_id" in message:
            document_id = message.get("document_id")
            document = get_document_by_id(document_id)
            socket.send_json({"document": document})
        else:
            socket.send_json({"error": "Invalid request"})

def get_document_by_id(document_id):
    connection = get_db_connection()
    try:
        with connection.cursor() as cursor:
            cursor.execute(
                "SELECT document_id, judul, isi_berita FROM documents WHERE document_id = %s", 
                (document_id,)
            )
            document = cursor.fetchone()
        return document
    finally:
        connection.close()

if __name__ == "__main__":
    main()