logging: 格式日志输出 (#268)

* logging: 统一日志格式输出

---------

Co-authored-by: Bob Chang <bob.chang@amway.com>
Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com>

cli_demo.py

@@ -31,13 +31,13 @@ if __name__ == "__main__":
                                                                      chat_history=history,
                                                                      streaming=STREAMING):
             if STREAMING:
-                print(resp["result"][last_print_len:], end="", flush=True)
+                logger.info(resp["result"][last_print_len:], end="", flush=True)
                 last_print_len = len(resp["result"])
             else:
-                print(resp["result"])
+                logger.info(resp["result"])
         if REPLY_WITH_SOURCE:
             source_text = [f"""出处 [{inum + 1}] {os.path.split(doc.metadata['source'])[-1]}：\n\n{doc.page_content}\n\n"""
                            # f"""相关度：{doc.metadata['score']}\n\n"""
                            for inum, doc in
                            enumerate(resp["source_documents"])]
-            print("\n\n" + "\n\n".join(source_text))
+            logger.info("\n\n" + "\n\n".join(source_text))

configs/model_config.py

@@ -1,6 +1,13 @@
 import torch.cuda
 import torch.backends
 import os
+import logging
+import uuid
+
+LOG_FORMAT = "%(levelname) -5s %(asctime)s" "-1d: %(message)s"
+logger = logging.getLogger()
+logger.setLevel(logging.INFO)
+logging.basicConfig(format=LOG_FORMAT)
 
 embedding_model_dict = {
     "ernie-tiny": "nghuyong/ernie-3.0-nano-zh",
@@ -63,3 +70,13 @@ LLM_HISTORY_LEN = 3
 VECTOR_SEARCH_TOP_K = 5
 
 NLTK_DATA_PATH = os.path.join(os.path.dirname(os.path.dirname(__file__)), "nltk_data")
+
+FLAG_USER_NAME = uuid.uuid4().hex
+
+logger.info(f"""
+loading model config
+llm device: {LLM_DEVICE}
+embedding device: {EMBEDDING_DEVICE}
+dir: {os.path.dirname(os.path.dirname(__file__))}
+flagging username: {FLAG_USER_NAME}
+""")

models/chatglm_llm.py

@@ -126,8 +126,7 @@ class ChatGLM(LLM):
                 model_config.pre_seq_len = prefix_encoder_config['pre_seq_len']
                 model_config.prefix_projection = prefix_encoder_config['prefix_projection']
             except Exception as e:                
-                print(e)
+                logger.error(f"加载PrefixEncoder config.json失败: {e}")
-                print("加载PrefixEncoder config.json失败")
         self.model = AutoModel.from_pretrained(model_name_or_path, config=model_config, trust_remote_code=True,
                                                **kwargs)
         if LLM_LORA_PATH and use_lora:
@@ -165,8 +164,7 @@ class ChatGLM(LLM):
                 self.model.transformer.prefix_encoder.load_state_dict(new_prefix_state_dict)
                 self.model.transformer.prefix_encoder.float()
             except Exception as e:              
-                print(e)
+                logger.error(f"加载PrefixEncoder模型参数失败:{e}")
-                print("加载PrefixEncoder模型参数失败")
 
         self.model = self.model.eval()
 
@@ -177,8 +175,8 @@ if __name__ == "__main__":
                    llm_device=LLM_DEVICE, )
     last_print_len = 0
     for resp, history in llm._call("你好", streaming=True):
-        print(resp[last_print_len:], end="", flush=True)
+        logger.info(resp[last_print_len:], end="", flush=True)
         last_print_len = len(resp)
     for resp, history in llm._call("你好", streaming=False):
-        print(resp)
+        logger.info(resp)
     pass

webui.py

@@ -4,7 +4,6 @@ import shutil
 from chains.local_doc_qa import LocalDocQA
 from configs.model_config import *
 import nltk
-import uuid
  
 nltk.data.path = [NLTK_DATA_PATH] + nltk.data.path
 
@@ -27,9 +26,7 @@ llm_model_dict_list = list(llm_model_dict.keys())
 
 local_doc_qa = LocalDocQA()
 
-logger = gr.CSVLogger()
+flag_csv_logger = gr.CSVLogger()
-username = uuid.uuid4().hex
-
 
 def get_answer(query, vs_path, history, mode,
                streaming: bool = STREAMING):
@@ -54,23 +51,24 @@ def get_answer(query, vs_path, history, mode,
             history[-1][-1] = resp + (
                 "\n\n当前知识库为空，如需基于知识库进行问答，请先加载知识库后，再进行提问。" if mode == "知识库问答" else "")
             yield history, ""
-    logger.flag([query, vs_path, history, mode], username=username)
+    logger.info(f"flagging: username={FLAG_USER_NAME},query={query},vs_path={vs_path},mode={mode},history={history}")
+    flag_csv_logger.flag([query, vs_path, history, mode], username=FLAG_USER_NAME)
 
 def init_model():
     try:
         local_doc_qa.init_cfg()
         local_doc_qa.llm._call("你好")
         reply = """模型已成功加载，可以开始对话，或从右侧选择模式后开始对话"""
-        print(reply)
+        logger.info(reply)
         return reply
     except Exception as e:
-        print(e)
+        logger.error(e)
         reply = """模型未成功加载，请到页面左上角"模型配置"选项卡中重新选择后点击"加载模型"按钮"""
         if str(e) == "Unknown platform: darwin":
-            print("该报错可能因为您使用的是 macOS 操作系统，需先下载模型至本地后执行 Web UI，具体方法请参考项目 README 中本地部署方法及常见问题："
+            logger.info("该报错可能因为您使用的是 macOS 操作系统，需先下载模型至本地后执行 Web UI，具体方法请参考项目 README 中本地部署方法及常见问题："
                   " https://github.com/imClumsyPanda/langchain-ChatGLM")
         else:
-            print(reply)
+            logger.info(reply)
         return reply
 
 
@@ -83,11 +81,11 @@ def reinit_model(llm_model, embedding_model, llm_history_len, use_ptuning_v2, us
                               use_lora=use_lora,
                               top_k=top_k, )
         model_status = """模型已成功重新加载，可以开始对话，或从右侧选择模式后开始对话"""
-        print(model_status)
+        logger.info(model_status)
     except Exception as e:
-        print(e)
+        logger.error(e)
         model_status = """模型未成功重新加载，请到页面左上角"模型配置"选项卡中重新选择后点击"加载模型"按钮"""
-        print(model_status)
+        logger.info(model_status)
     return history + [[None, model_status]]
 
 
@@ -109,7 +107,7 @@ def get_vector_store(vs_id, files, history):
     else:
         file_status = "模型未完成加载，请先在加载模型后再导入文件"
         vs_path = None
-    print(file_status)
+    logger.info(file_status)
     return vs_path, None, history + [[None, file_status]]
 
 
@@ -235,7 +233,7 @@ with gr.Blocks(css=block_css) as demo:
                                              inputs=[select_vs, folder_files, chatbot],
                                              outputs=[vs_path, folder_files, chatbot],
                                              )
-                    logger.setup([query, vs_path, chatbot, mode], "flagged")
+                    flag_csv_logger.setup([query, vs_path, chatbot, mode], "flagged")
                     query.submit(get_answer,
                                  [query, vs_path, chatbot, mode],
                                  [chatbot, query])