新增：切片进度查询间隔时间（秒）+清理日志按钮

ragflows/api.py

@@ -114,7 +114,7 @@ def parse_chunks(doc_ids, run=1):
     url = f"{configs.API_URL}/document/run"  # 替换为实际的服务器地址
     data = {"doc_ids":doc_ids,"run":run}
     response = requests.post(url, json=data, headers=configs.get_header())
-    timeutils.print_log(response.text)
+    timeutils.print_log("parse_chunks response:", response.text)
     if response.status_code == 200:
         return response.json()
     else:
@@ -157,10 +157,14 @@ def parse_chunks_with_check(filename):
             fileutils.save(f"{fileutils.get_cache_dir()}/ragflow_fail.txt", f"{timeutils.get_now_str()} {msg}\n")
             return False
 
-        timeutils.print_log(f"[{filename}]解析进度为：{progress}")
+        if configs.ENABLE_PROGRESS_LOG:
+            progress_percent = round(progress * 100, 2)
+            timeutils.print_log(f"[{filename}]解析进度为：{progress_percent}%")
+
         if progress == 1:
             return True
-        time.sleep(1)
+
+        time.sleep(configs.PROGRESS_CHECK_INTERVAL)
 
 
 # 是否请求成功

ragflows/configs.demo.py

@@ -25,6 +25,12 @@
 # 是否仅上传文件。True=仅上传文件， False=上传文件+自动解析
 ONLY_UPLOAD = False
 
+# 是否打印切片进度查询日志。True=打印，False=不打印
+ENABLE_PROGRESS_LOG = True
+
+# 切片进度查询间隔时间（秒）
+PROGRESS_CHECK_INTERVAL = 1
+
 
 def get_header():
     return {'authorization': AUTHORIZATION}

ragflows/main.py

@@ -105,12 +105,13 @@ def main():
         if ragflowdb.exist_name(filename):
             doc_item = ragflowdb.get_doc_item_by_name(filename)
             if configs.ONLY_UPLOAD:
-                timeutils.print_log(f"{file_path} 已存在，跳过")
+                timeutils.print_log(f"{file_path} 已存在，跳过\n")
             elif doc_item.get('progress') == 1:
-                timeutils.print_log(f"{file_path} 已完成切片，跳过")
+                timeutils.print_log(f"{file_path} 已完成切片，跳过\n")
             else:
+                timeutils.print_log(f'{file_path}，开始切片并等待解析完毕')
                 status = api.parse_chunks_with_check(filename)
-                timeutils.print_log(f"{file_path} 切片状态：", status)
+                timeutils.print_log(f"{file_path} 切片状态：", status, "\n")
             continue
 
         # 文件不存在，上传文件=>切片=>解析并等待解析完毕
@@ -121,7 +122,7 @@ def main():
             parser_id=configs.PARSER_ID, 
             run="1"
         )
-        timeutils.print_log(response)
+        timeutils.print_log("upload_file_to_kb response:", response)
         if api.is_succeed(response) is False:
             timeutils.print_log(f'{file_path} 上传失败：{response.get("text")}')
             continue
@@ -133,7 +134,7 @@ def main():
         # 上传成功，开始切片
         timeutils.print_log(f'{file_path}，开始切片并等待解析完毕')
         status = api.parse_chunks_with_check(filename)
-        timeutils.print_log(file_path, "切片状态：", status)
+        timeutils.print_log(file_path, "切片状态：", status, "\n")
 
     timeutils.print_log('all done')
 

ragflows/ragflowdb.py

@@ -40,14 +40,14 @@ def get_doc_list(kb_id):
     doc_ids = db.query_list(sql)
     return doc_ids
 
-@timeutils.monitor
+# @timeutils.monitor
 def get_doc_item(doc_id):
     db = get_db()
     sql = f"select id,name,progress from document where id = '{doc_id}'"
     results = db.query_list(sql)
     return results[0] if results else None
 
-@timeutils.monitor
+# @timeutils.monitor
 def get_doc_item_by_name(name):
     db = get_db()
     kb_id = configs.DIFY_DOC_KB_ID

scripts/launcher.py

@@ -77,10 +77,10 @@ def __init__(self):
         self.log_handlers = []  # 添加日志处理器列表
         self.original_print_log = None  # 保存原始的日志打印函数
         self.title("RagFlow Upload")
-        self.geometry("800x600")
+        self.geometry("800x660")
 
         # 版本和仓库信息
-        self.version = "v1.0.1"  # 版本号
+        self.version = "v1.0.2"  # 版本号
         self.github_repo = "https://github.com/Samge0/ragflow-upload"  # GitHub仓库地址
 
         # 自定义图标
@@ -98,13 +98,16 @@ def __init__(self):
             "PARSER_ID": {"type": str, "label": "解析方式", "default": "naive"},
             "DOC_DIR": {"type": str, "label": "文档目录", "default": "your doc dir"},
             "DOC_SUFFIX": {"type": str, "label": "文档后缀", "default": "md,txt,pdf,docx"},
+            "PROGRESS_CHECK_INTERVAL": {"type": int, "label": "切片进度查询间隔", "default": "1"},
+
             "MYSQL_HOST": {"type": str, "label": "MySQL主机", "default": "localhost"},
             "MYSQL_PORT": {"type": int, "label": "MySQL端口", "default": "5455"},
             "MYSQL_USER": {"type": str, "label": "MySQL用户名", "default": "root"},
             "MYSQL_PASSWORD": {"type": str, "label": "MySQL密码", "default": "infini_rag_flow"},
             "MYSQL_DATABASE": {"type": str, "label": "MySQL数据库", "default": "rag_flow"},
             "DOC_MIN_LINES": {"type": int, "label": "最小行数", "default": "1"},
-            "ONLY_UPLOAD": {"type": bool, "label": "仅上传文件", "default": "False"}
+            "ONLY_UPLOAD": {"type": bool, "label": "仅上传文件", "default": "False"},
+            "ENABLE_PROGRESS_LOG": {"type": bool, "label": "打印切片进度日志", "default": "True"},
         }
 
         self.create_ui()
@@ -171,6 +174,17 @@ def create_ui(self):
         )
         self.run_button.pack(side="left", padx=5, pady=5)
 
+        # 添加清理日志按钮
+        self.clear_log_button = ctk.CTkButton(
+            button_frame,
+            text="清理日志",
+            command=self.clear_log,
+            fg_color=["#757575", "#616161"],  # 灰色
+            hover_color=["#616161", "#424242"],  # 深灰色
+            text_color="white"  # 白色文字
+        )
+        self.clear_log_button.pack(side="left", padx=5, pady=5)
+
         # 日志区域
         log_frame = ctk.CTkFrame(self.main_frame)
         log_frame.pack(fill="both", expand=True, padx=5, pady=5)
@@ -465,6 +479,13 @@ def save_config(self):
         except Exception as e:
             self.log(f"保存配置失败: {str(e)}")
 
+    def clear_log(self):
+        """清理UI界面的日志显示"""
+        self.log_text.configure(state="normal")
+        self.log_text.delete(1.0, "end")
+        self.log_text.configure(state="disabled")
+        self.log("日志已清理")
+
 if __name__ == "__main__":
     ctk.set_appearance_mode("dark")
     ctk.set_default_color_theme("blue")