'commit'

dsRag/Config/Config.py

@@ -11,7 +11,7 @@ ES_CONFIG = {
     "basic_auth": ("elastic", "jv9h8uwRrRxmDi1dq6u8"),
     "verify_certs": False,
     "ssl_show_warn": False,
-    "default_index": "knowledge_base"
+    "index_name": "knowledge_base"
 }
 
 

dsRag/ElasticSearch/T2_CreatingMapping.py

@@ -31,7 +31,7 @@ mapping = {
 }
 
 # 创建索引
-index_name = Config.ES_CONFIG['default_index']
+index_name = Config.ES_CONFIG['index_name']
 if es.indices.exists(index=index_name):
     es.indices.delete(index=index_name)
     print(f"删除已存在的索引 '{index_name}'")

dsRag/ElasticSearch/T4_InsertData.py

@@ -63,7 +63,7 @@ for filename in os.listdir(txt_dir):
                 'timestamp': timestamp,
                 'embedding': embedding
             }
-            es.index(index=ES_CONFIG['default_index'], document=doc)
+            es.index(index=ES_CONFIG['index_name'], document=doc)
             print(f"文件 {filename} 数据插入成功")
 
 print("所有文件处理完成")

dsRag/ElasticSearch/T5_select_all_data.py

@@ -0,0 +1,49 @@
+from elasticsearch import Elasticsearch
+import warnings
+from Config import Config
+from Config.Config import ES_CONFIG
+
+# 1. 初始化Elasticsearch连接
+warnings.filterwarnings('ignore', category=DeprecationWarning)
+es = Elasticsearch(
+    hosts=Config.ES_CONFIG['hosts'],
+    basic_auth=Config.ES_CONFIG['basic_auth'],
+    verify_certs=False
+)
+
+# 2. 直接在代码中指定要查询的标签
+query_tag = "MATH_DATA_1"  # 可以修改为其他需要的标签
+
+# 3. 构建查询条件
+query = {
+    "query": {
+        "bool": {
+            "must": [
+                {
+                    "term": {
+                        "tags": query_tag
+                    }
+                }
+            ]
+        }
+    },
+    "size": 1000
+}
+
+# 4. 执行查询
+try:
+    results = es.search(index=ES_CONFIG['index_name'], body=query)
+    print(f"查询标签 '{query_tag}' 结果：")
+    if results['hits']['hits']:
+        for hit in results['hits']['hits']:
+            doc = hit['_source']
+            print(f"ID: {hit['_id']}")
+            print(f"标签: {doc['tags']['tags']}")
+            print(f"用户问题: {doc['user_input']}")
+            print(f"时间: {doc['timestamp']}")
+            print(f"向量: {doc['embedding'][:5]}...")
+            print("-" * 40)
+    else:
+        print(f"未找到标签为 '{query_tag}' 的数据。")
+except Exception as e:
+    print(f"查询失败: {e}")