fix(llmage): 修复backup_accounted_llmusage中sor.execute()返回值误用（MySQL cursor.execute返回None非rowcount）

dat/qwen3.7-max.txt

@@ -0,0 +1,53 @@
+    已生成完整的 qwen3.7-max 配置SQL。以下是配置方案：
+    
+    模型摘要
+    - 模型名称: 千问3.7-Max
+    - API model: qwen3.7-max (严格按API文档填写)
+    - 分类: text2text (文生文)
+    - 供应商: 阿里百炼 (ali-qwen)
+    - 接口: OpenAI兼容 /chat/completions，同步流式
+    
+    复用vs新建
+    - upapp: 复用 ali-qwen (阿里百炼)
+    - upappkey: 复用 gCMl-BnrTrfoqWAEPPoH8 (API Key已配置)
+    - uapi: 复用 t2t (BY2cA4RD_axNxTG4m9Coa)
+    - uapiio: 复用 Is8l4TGkcZcqFSjbbeIK2 (文本会话)
+    - llm: 新建
+    - llm_api_map: 新建
+    - pricing_program: 新建
+    - pricing_program_timing: 新建
+    
+    生成的SQL (4条)
+    
+    sql
+    -- 1. llm表
+    INSERT INTO llm VALUES ('Jt26VKlUFsJxABuVLauBc','千问3.7-Max','qwen3.7-max','通义千问3.7系列旗舰模型，相较3.6版本在推理能力、代码能力、多模态理解等方面全面升级。适用于复杂任务处理、智能体开发、代码生成等高级应用场景。','text2text','qwen','ali-qwen','t2t','6fadgewjraOyvxC_EkHou','0','2026-05-23','9999-12-31','',NULL,10.00);
+    
+    -- 2. llm_api_map表
+    INSERT INTO llm_api_map VALUES ('rFS0Ag-epBUxE5LCF6glW','Jt26VKlUFsJxABuVLauBc','text2text','t2t',NULL,NULL,'0B-JNSUEelsY5TQ_Sawa7','1');
+    
+    -- 3. pricing_program表
+    INSERT INTO pricing_program VALUES ('0B-JNSUEelsY5TQ_Sawa7','qwen3.7-max定价','fields:\r\n  formula:\r\n    label: 计算公式\r\n    type: str\r\n  model:\r\n    label: 模型\r\n    type: str\r\n  prompt_tokens:\r\n    label: 输入词元数\r\n    type: int\r\n    value_mode: between\r\nformula: null\r\npricings:\r\n- formula: 2.5  prompt_tokens / 1000000 + completion_tokens  10 /1000000\r\n  model: qwen3.7-max\r\n  prompt_tokens: -0.1 ~= 32000\r\n- formula: 4  prompt_tokens / 1000000 + completion_tokens  16 /1000000\r\n  model: qwen3.7-max\r\n  prompt_tokens: 32000 ~= 128000\r\n- formula: 7  prompt_tokens / 1000000 + completion_tokens  28 /1000000\r\n  model: qwen3.7-max\r\n  prompt_tokens: 128000 ~= 256000\r\n- formula: 10  prompt_tokens / 1000000 + completion_tokens  40 /1000000\r\n  model: qwen3.7-max\r\n  prompt_tokens: 256000 ~= 1000000','2026-05-23','9999-12-31');
+    
+    -- 4. pricing_program_timing表
+    INSERT INTO pricing_program_timing VALUES ('C-IIEG_9_O9beN8S2vdUI','0B-JNSUEelsY5TQ_Sawa7','qwen3.7-max计费','fields:\r\n  formula:\r\n    label: 计算公式\r\n    type: str\r\n  model:\r\n    label: 模型\r\n    type: str\r\n  prompt_tokens:\r\n    label: 输入词元数\r\n    type: int\r\n    value_mode: between\r\nformula: null\r\npricings:\r\n- formula: 2.5  prompt_tokens / 1000000 + completion_tokens  10 /1000000\r\n  model: qwen3.7-max\r\n  prompt_tokens: -0.1 ~= 32000\r\n- formula: 4  prompt_tokens / 1000000 + completion_tokens  16 /1000000\r\n  model: qwen3.7-max\r\n  prompt_tokens: 32000 ~= 128000\r\n- formula: 7  prompt_tokens / 1000000 + completion_tokens  28 /1000000\r\n  model: qwen3.7-max\r\n  prompt_tokens: 128000 ~= 256000\r\n- formula: 10  prompt_tokens / 1000000 + completion_tokens  40 /1000000\r\n  model: qwen3.7-max\r\n  prompt_tokens: 256000 ~= 1000000','2026-05-23','9999-12-31');
+    
+    
+    验证查询
+    sql
+    SELECT id, name, model, upappid, apiname FROM llm WHERE model = 'qwen3.7-max';
+    SELECT id, llmid, llmcatelogid, apiname, ppid FROM llm_api_map WHERE llmid = 'Jt26VKlUFsJxABuVLauBc';
+    SELECT id, name FROM pricing_program WHERE id = '0B-JNSUEelsY5TQ_Sawa7';
+    
+    
+    ppid三处一致
+    - pricing_program.id = 0B-JNSUEelsY5TQ_Sawa7
+    - pricing_program_timing.ppid = 0B-JNSUEelsY5TQ_Sawa7
+    - llm_api_map.ppid = 0B-JNSUEelsY5TQ_Sawa7
+    
+    注意事项
+    1. 定价参考qwen3.6-plus的阶梯式计费模式，具体价格需根据百炼官方定价调整
+    2. qwen3.7-max使用已有的t2t uapi，OpenAI兼容接口
+    3. 如果qwen3.7-max有特殊的推理模式（如思考/非思考切换），可能需要额外的配置
+    4. 之前提到的 startReasoning is not defined 前端报错问题需要单独排查，可能与harnessed_reasoning的bricks前端代码有关
+

llmage/accounting.py

@@ -240,28 +240,36 @@ async def llm_accoung_failed(luid, reason=None):
 
 
 async def backup_accounted_llmusage(cutoff_date):
-	"""Backup accounted records with use_date < cutoff_date to history table using single SQL statements."""
+	"""Backup accounted records with use_date < cutoff_date to history table."""
 	env = ServerEnv()
 	ts = env.timestampstr()
 	async with get_sor_context(env, 'llmage') as sor:
+		# Step 0: Count records to backup
+		count_sql = """SELECT COUNT(*) as cnt FROM llmusage
+WHERE accounting_status='accounted' AND use_date < ${cutoff_date}$"""
+		count_recs = await sor.sqlExe(count_sql, {'cutoff_date': cutoff_date})
+		total = count_recs[0].cnt if count_recs else 0
+		if total == 0:
+			debug(f'backup_accounted_llmusage: no records to backup for use_date < {cutoff_date}')
+			return 0
+		debug(f'backup_accounted_llmusage: {total} records to backup')
+		
 		# Step 1: INSERT INTO history SELECT from main table
 		insert_sql = """INSERT INTO llmusage_history 
 (id, llmid, use_date, use_time, userid, usages, ioinfo, transno, responsed_seconds, finish_seconds, status, taskid, amount, cost, userorgid, ownerid, accounting_status, backup_time)
 SELECT id, llmid, use_date, use_time, userid, usages, ioinfo, transno, responsed_seconds, finish_seconds, status, taskid, amount, cost, userorgid, ownerid, accounting_status, ${ts}$
 FROM llmusage
 WHERE accounting_status='accounted' AND use_date < ${cutoff_date}$"""
-		result = await sor.execute(insert_sql, {'cutoff_date': cutoff_date, 'ts': ts})
-		inserted = result if isinstance(result, int) else 0
-		debug(f'backup_accounted_llmusage: {inserted} records inserted to history')
+		await sor.execute(insert_sql, {'cutoff_date': cutoff_date, 'ts': ts})
+		debug(f'backup_accounted_llmusage: inserted {total} records to history')
 		
-		if inserted > 0:
 		# Step 2: DELETE from main table
 		delete_sql = """DELETE FROM llmusage
 WHERE accounting_status='accounted' AND use_date < ${cutoff_date}$"""
 		await sor.execute(delete_sql, {'cutoff_date': cutoff_date})
-			debug(f'backup_accounted_llmusage: {inserted} records deleted from main table')
+		debug(f'backup_accounted_llmusage: deleted {total} records from main table')
 	
-	return inserted
+	return total
 
 
 async def get_failed_accounting_records(filters=None, page=1, page_size=50):