create_db.py

# 首先导入所需第三方库
from langchain.document_loaders import UnstructuredFileLoader
from langchain.document_loaders import UnstructuredMarkdownLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores import Chroma
from langchain.embeddings.huggingface import HuggingFaceEmbeddings
from tqdm import tqdm
import os
from langchain.docstore.document import Document


# 获取文件路径函数
def get_files(dir_path):
    # args：dir_path，目标文件夹路径
    file_list = []
    for filepath, dirnames, filenames in os.walk(dir_path):
        # os.walk 函数将递归遍历指定文件夹
        for filename in filenames:
            # 通过后缀名判断文件类型是否满足要求
            if filename.endswith(".md"):
                # 如果满足要求，将其绝对路径加入到结果列表
                file_list.append(os.path.join(filepath, filename))
            elif filename.endswith(".txt"):
                file_list.append(os.path.join(filepath, filename))
    return file_list


# 加载文件函数
def get_text(dir_path):
    # args：dir_path，目标文件夹路径
    # 首先调用上文定义的函数得到目标文件路径列表
    file_lst = get_files(dir_path)
    # docs 存放加载之后的纯文本对象
    docs = []
    # 遍历所有目标文件
    for one_file in tqdm(file_lst):
        file_type = one_file.split(".")[-1]
        if file_type == "md":
            loader = UnstructuredMarkdownLoader(one_file)
        elif file_type == "txt":
            loader = UnstructuredFileLoader(one_file)
        else:
            # 如果是不符合条件的文件，直接跳过
            continue
        docs.extend(loader.load())
    return docs


# 目标文件夹
tar_dir1 = ["/root/code/law_rag/data/law"]

# 加载目标文件
docs = []
for dir_path in tar_dir1:
    docs.extend(get_text(dir_path))

# 对文本进行分块
split_docs = []
for doc in docs:
    cur_text = doc.page_content.split("\n\n")
    metadata = doc.metadata
    for line in cur_text:
        split_docs.append(Document(page_content=line, metadata=metadata))

# tar_dir2 = ["/root/code/law_rag/data/cases"]
# for dir_path in tar_dir2:
#     split_docs.extend(get_text(dir_path))

# 加载开源词向量模型
embeddings = HuggingFaceEmbeddings(model_name="/root/models/sentence-transformer")

# 构建向量数据库
# 定义持久化路径
persist_directory = "/root/code/law_rag/data_base/vector_db/law"
# 加载数据库
vectordb = Chroma.from_documents(
    documents=split_docs,
    embedding=embeddings,
    persist_directory=persist_directory,  # 允许我们将persist_directory目录保存到磁盘上
)
# 将加载的向量数据库持久化到磁盘上
vectordb.persist()