embed_store.py

from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import Qdrant
from qdrant_client import QdrantClient


def get_embeddings():
    return HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")


def store_embeddings(chunks, embeddings):
    client = QdrantClient(url="http://localhost:6333")

    collection_name = "repo_docs"

    # create collection manually (safe + explicit)
    client.recreate_collection(
        collection_name=collection_name,
        vectors_config={
            "size": 384,  # MiniLM embedding size
            "distance": "Cosine"
        }
    )

    texts = [c["content"] for c in chunks]
    metadatas = [
        {
            "path": c["path"],
            "type": c["type"],
            "file_name": c["file_name"]
        }
        for c in chunks
    ]

    vectorstore = Qdrant(
        client=client,
        collection_name=collection_name,
        embeddings=embeddings,
    )

    vectorstore.add_texts(texts, metadatas=metadatas)

    return vectorstore