rag_project/database/chormadb.py

import chromadb
from embedding.embedding import get_embedding
from langchain.text_splitter import RecursiveCharacterTextSplitter
from create_db import load_txt_from_dir

DEFAULT_DB_PATH = r"../knowledge_db"
DEFAULT_PERSIST_PATH = '../vector_db/chroma'
# 创建一个客户端
chroma_client = chromadb.Client()
# 创建一个集合/传统数据库中的一张表
embeddings = get_embedding(embedding='m3e')
collection = chroma_client.get_or_create_collection(name="my_collection",embedding_function=embeddings)

all_docs = load_txt_from_dir(DEFAULT_DB_PATH)

# 切分文档
text_splitter = RecursiveCharacterTextSplitter(
    chunk_size=500, chunk_overlap=150)
split_docs = text_splitter.split_documents(all_docs)