ragflow_api_test/chunk_operations.py

from elasticsearch import Elasticsearch

# 初始化 Elasticsearch   用户名elastic，密码infini_rag_flow
es = Elasticsearch(
    [{'host': '127.0.0.1', 'port': 1200, 'scheme': 'http'}],
    basic_auth=('elastic', 'infini_rag_flow')
)

def update_img_id_in_elasticsearch(tenant_id, doc_id, chunk_id, new_img_id):
    """
    在 Elasticsearch 中更新指定文档块的 img_id。
    如果img_id不存在，则增加一个新的 img_id。

    :param tenant_id: 租户 ID
    :param doc_id: 文档 ID
    :param chunk_id: 文档块 ID
    :param new_img_id: 新的 img_id
    :return: 更新结果

    """
    # 构建索引名称
    index_name = f"ragflow_{tenant_id}"  # 这里需要替换为实际的索引名称生成逻辑

    # 构建查询条件
    query = {
        "bool": {
            "must": [
                {"term": {"doc_id": doc_id}},
                {"term": {"_id": chunk_id}}
            ]
        }
    }

    # 搜索目标文档
    result = es.search(index=index_name, body={"query": query})

    # 检查是否找到目标文档
    if result['hits']['total']['value'] == 0:
        return {"code": 102, "message": f"Can't find this chunk {chunk_id}"}
    

    # 获取目标文档的 ID
    hit = result['hits']['hits'][0]
    doc_id_in_es = hit['_id']

    update_body = {
        "doc": {
            "img_id": new_img_id
        }
        }

    # 更新文档
    update_result = es.update(index=index_name, id=doc_id_in_es, body=update_body)
    

    if update_result['result'] == 'updated':
        return {"code": 0, "message": ""}
    else:
        return {"code": 100, "message": "Failed to update img_id"}
    

def list_chunk_information(tenant_id, dataset_id, doc_id=None, chunk_id=None, size=1000):
    """
    列出指定条件下的 chunk 信息

    :param tenant_id: 租户 ID
    :param dataset_id: 数据集 ID
    :param doc_id: 文档 ID（可选）
    :param chunk_id: 文档块 ID（可选）
    :param size: 返回的最大结果数
    :return: chunk 信息列表
    """
    # 构建索引名称
    index_name = f"ragflow_{tenant_id}"  # 需替换为实际索引名生成逻辑
    
    # 构建基础查询
    query = {"bool": {"must": []}}
    
    # 添加可选过滤条件
    if doc_id:
        query["bool"]["must"].append({"term": {"doc_id": doc_id}})

    
    # 执行搜索
    try:
        result = es.search(
            index=index_name,
            body={"query": query, "size": size},
            fields= ["_id"],
            _source=["doc_id", "kb_id", "img_id", "content", "docnm_kwd"]  # 指定返回的字段
        )
        for d in result['hits']['hits']:
            print(d['_id'])
            
        
        # 提取并格式化结果
        chunks = []
        for hit in result['hits']['hits']:
            source = hit['_source']
            chunks.append({
                'doc_id': source.get('doc_id'),
                'kb_id': source.get('kb_id'),
                'img_id': source.get('img_id'),
                'content': source.get('content'),
                'docnm_kwd': source.get('docnm_kwd')
            })
            
        return {"code": 0, "message": "", "data": chunks}
    
    except Exception as e:
        return {"code": 500, "message": str(e), "data": []}

# 示例调用 - 列出特定文档的所有 chunks
if __name__ == "__main__":
    try:
        print(es.info())
    except Exception as e:
        print("连接失败：", e)

    tenant_id = "d669205e57a211f0b9e7324e7f243034"
    tenant_id = "9c73df5a3ebc11f08410c237296aa408"
    dataset_id = "0e6127da574a11f0a59c7e7439a490f8"  #  dataset_id = kb_id
    doc_id = "cbf576385bc911f08f23fedc3996e479"
    doc_id = "323113d8670c11f0b4255ea1d23c381a"  
    chunk_id = "f035247f7de579b0"  #  
    chunk_id = "b2d53baddbfde97c"  # 
    new_img_id =    "10345832587311f0919f3a2728512a4b-f035247f7de579b0" #"new_img_id_12345"
    new_img_id =  "0e6127da574a11f0a59c7e7439a490f8-b2d53baddbfde97c"
    #new_img_id ="c5142bce5ac611f0ae707a8b5ba029cb-thumbnail_fb3cbc165ac611f0b5897a8b5ba029cb.png"



    #chunk_list = list_chunk_information(tenant_id, dataset_id, doc_id=doc_id)
    update_img_id_in_elasticsearch(tenant_id, doc_id, chunk_id,new_img_id)
    # if chunk_list["code"] == 0:
    #     print(f"找到 {len(chunk_list['data'])} 个 chunks")
    #     for chunk in chunk_list['data']:
    #         print(f"Chunk ID: {chunk['kb_id']}, Img ID: {chunk['img_id']}","docnm_kwd:", chunk['docnm_kwd'])
    # else:
    #     print(f"Error: {chunk_list['message']}")