Convert your csv file into locally running qdrant embeddings using this python script.

csv_to_qdrant_embeddings.py

import os
import pandas as pd
from sentence_transformers import SentenceTransformer
from qdrant_client import QdrantClient
from qdrant_client.models import PointStruct, Distance, VectorParams

# Configuration
CSV_FILE = "snippets.csv" # Update with your csv path
QDRANT_HOST = "localhost"  # Change if using a remote Qdrant instance
QDRANT_PORT = 6333
COLLECTION_NAME = "solidity_snippets"

def setup_qdrant():
    """Initialize Qdrant client and create a collection."""
    client = QdrantClient(QDRANT_HOST, port=QDRANT_PORT)

    # Load CSV
    df = pd.read_csv(CSV_FILE, header=None, names=["solidity_code"])

    # Load embedding model
    model = SentenceTransformer("microsoft/codebert-base")

    # Generate embeddings
    df["embedding"] = df["solidity_code"].apply(lambda code: model.encode(code, convert_to_numpy=True))

    # Create a Qdrant collection (if not exists)
    client.recreate_collection(
        collection_name=COLLECTION_NAME,
        vectors_config=VectorParams(size=df["embedding"].iloc[0].shape[0], distance=Distance.COSINE),
    )

    # Insert embeddings into Qdrant
    points = [
        PointStruct(id=i, vector=embedding.tolist(), payload={"code": code})
        for i, (code, embedding) in enumerate(zip(df["solidity_code"], df["embedding"]))
    ]
    client.upsert(collection_name=COLLECTION_NAME, points=points)

    print(f"Inserted {len(df)} embeddings into Qdrant.")

    # Create a snapshot
    client.create_snapshot(collection_name=COLLECTION_NAME)
    print("Snapshot created successfully!")

if __name__ == "__main__":
    setup_qdrant()