python实现Doris的streamLoad

# ===================== StreamLoad函数【新增】 =====================
def json_serial(obj):
    if isinstance(obj, (datetime, date)):
        return obj.strftime("%Y-%m-%d %H:%M:%S")
    raise TypeError(f"Object of type {type(obj).name} is not JSON serializable")

def doris_stream_load(data_rows: List[list]) -> bool:
    """Doris StreamLoad写入明细表ods_jice_detail_data"""
    print(f"【StreamLoad实际接收行数】{len(data_rows)}", flush=True)
    if not data_rows:
        return True
    url = f"http://{doris_load_host}/api/{target_doris_ods_db}/{ods_table_name}/_stream_load"
    headers = {
        "Expect": "100-continue",
        "format": "json",
        "read_json_by_line": "true",
        "Content-Type": "application/json",
        "strip_outer_array": "false",
        "label": f"jice-{uuid.uuid4()}"
    }
    auth = (doris_user, doris_pwd)
    body = "\n".join([json.dumps(line, ensure_ascii=False, default=json_serial) for line in data_rows])
    try:
        resp = requests.put(url, headers=headers, auth=auth, data=body.encode("utf-8"), timeout=120, allow_redirects=False)
        #print(f"【StreamLoad http状态码】{resp.status_code}", flush=True)
        # 处理FE 307重定向，携带完整header请求BE
        if resp.status_code == 307:
            redirect_url = resp.headers.get("Location")
            #print(f"【检测到307重定向，跳转地址：{redirect_url}】", flush=True)
            headers["label"] = f"jice-{uuid.uuid4()}"
            resp = requests.put(redirect_url, headers=headers, auth=auth, data=body.encode("utf-8"), timeout=120)
            #print(f"【重定向后http状态码】{resp.status_code}", flush=True)
        resp_text = resp.text.strip()
        #print(f"【Doris原始返回内容】{resp_text}", flush=True)
        if not resp_text:
            print(f"【StreamLoad严重失败】Doris返回空响应！批次行数:{len(data_rows)}", flush=True)
            print(f"响应头信息：{resp.headers}", flush=True)
            return False
        try:
            res_json = json.loads(resp_text)
        except json.JSONDecodeError as je:
            print(f"【响应非合法JSON】原始返回内容：{resp_text}", flush=True)
            print(f"JSON解析异常：{str(je)}", flush=True)
            traceback.print_exc()
            return False
        #print(f"【Doris返回完整信息】{res_json}", flush=True)
        status = res_json.get("Status")
        load_rows = res_json.get("NumberLoadedRows", 0)
        print(f"【导入统计】请求行数:{len(data_rows)}, Doris实际加载行数:{load_rows}", flush=True)
        if status == "Success" and load_rows == len(data_rows):
            print(f"【StreamLoad成功】LoadId:{res_json.get('LoadId')}", flush=True)
            return True
        else:
            # 这里包含 Publish Timeout、ETL失败、数据格式错误等隐性失败
            print(f"【StreamLoad业务失败】状态:{status},请求{len(data_rows)}行, 实际{load_rows}行,原因:{res_json.get('Message')}", flush=True)
            return False
    except Exception as e:
        print(f"【StreamLoad网络异常】异常信息:{str(e)}", flush=True)
        traceback.print_exc()
        return False