调整rag_snippet表字段格式

app/services/table_snippet.py

@@ -435,6 +435,8 @@ def merge_snippet_records_from_db(
     for alias_id, alias_info in alias_map.items():
         if alias_id in seen_ids:
             continue
+        if alias_action_id is None and snippet_action_id is None:
+            continue
         merged.append(
             {
                 "id": alias_id,
@@ -520,11 +522,14 @@ def _prepare_rag_payloads(
         else:
             updated_at = updated_at_raw if isinstance(updated_at_raw, datetime) else now
 
+        created_at = updated_at
+
         row = {
             "rag_item_id": rag_item_id,
             "workspace_id": workspace_id,
             "table_id": table_id,
             "version_ts": version_ts,
+            "created_at": created_at,
             "action_result_id": action_result_id,
             "snippet_id": snippet_id,
             "rag_text": rag_text,
@@ -557,6 +562,7 @@ def _upsert_rag_snippet_rows(engine: Engine, rows: Sequence[Dict[str, Any]]) ->
             workspace_id,
             table_id,
             version_ts,
+            created_at,
             action_result_id,
             snippet_id,
             rag_text,
@@ -567,6 +573,7 @@ def _upsert_rag_snippet_rows(engine: Engine, rows: Sequence[Dict[str, Any]]) ->
             :workspace_id,
             :table_id,
             :version_ts,
+            :created_at,
             :action_result_id,
             :snippet_id,
             :rag_text,

file/tableschema/rag_snippet.sql

@@ -7,9 +7,18 @@ CREATE TABLE `rag_snippet` (
   `snippet_id` varchar(255) COLLATE utf8mb4_bin NOT NULL COMMENT '原始 snippet id',
   `rag_text` text COLLATE utf8mb4_bin NOT NULL COMMENT '用于向量化的拼接文本',
   `merged_json` json NOT NULL COMMENT '合并后的 snippet 对象',
-  `updated_at` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
-  PRIMARY KEY (`rag_item_id`),
+  `created_at` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT '写入时间，用于分区',
+  `updated_at` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
+  PRIMARY KEY (`rag_item_id`,`created_at`),
   KEY `idx_action_result` (`action_result_id`),
   KEY `idx_workspace` (`workspace_id`),
   KEY `idx_table_version` (`table_id`,`version_ts`)
-) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin COMMENT='RAG snippet 索引缓存';
+) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin
+PARTITION BY RANGE COLUMNS (`created_at`) (
+  PARTITION p202401 VALUES LESS THAN ('2024-02-01'),
+  PARTITION p202402 VALUES LESS THAN ('2024-03-01'),
+  PARTITION p202403 VALUES LESS THAN ('2024-04-01'),
+  PARTITION p202404 VALUES LESS THAN ('2024-05-01'),
+  PARTITION p202405 VALUES LESS THAN ('2024-06-01'),
+  PARTITION p_future VALUES LESS THAN (MAXVALUE)
+) COMMENT='RAG snippet 索引缓存';

test/test_snippet_rag_ingest.py

@@ -38,6 +38,7 @@ def _setup_sqlite_engine():
                     workspace_id INTEGER,
                     table_id INTEGER,
                     version_ts INTEGER,
+                    created_at TEXT,
                     snippet_id TEXT,
                     rag_text TEXT,
                     merged_json TEXT,