add tool

api/core/tools/builtin_tool/providers/apo_analysis/tools/trend.py

@@ -0,0 +1,80 @@
+import json
+import numpy as np
+from scipy import stats
+from collections.abc import Generator
+from typing import Any, Optional
+
+from core.tools.builtin_tool.tool import BuiltinTool
+from core.tools.entities.tool_entities import ToolInvokeMessage
+
+
+class TrendTool(BuiltinTool):
+    def _invoke(
+        self,
+        user_id: str,
+        tool_parameters: dict[str, Any],
+        conversation_id: Optional[str] = None,
+        app_id: Optional[str] = None,
+        message_id: Optional[str] = None,
+    ) -> Generator[ToolInvokeMessage, None, None]:
+        metric = tool_parameters.get('metricData')
+        metric_data = json.loads(metric)
+        # threshold = float(tool_parameters.get('threshold'))
+        res = "ok"
+        for serie in metric_data['data']['timeseries']:
+            # for k, v in serie['chart']['chartData'].items():
+            #     v = float(v)
+            #     if v > threshold:
+            #         tmpres['data'][str(k)] = v
+            tmpres = self.analyze_kv_trend(serie['chart']['chartData'], window_size=5)
+            if tmpres.get('trend', "") != "" and tmpres.get('trend', "") != '波动':
+                res = "false"
+        
+        yield self.create_text_message(res)
+    
+    def analyze_kv_trend(self, data_dict, window_size=5, threshold=0.05): 
+    
+        # 提取值序列
+        values = np.array(list(data_dict.values()))
+    
+        # 检查数据有效性
+        if len(values) < window_size:
+            return {"error": "数据长度不足"}
+    
+        # 计算整体趋势（使用线性回归）
+        x = np.arange(len(values))
+        slope, intercept, r_value, p_value, std_err = stats.linregress(x, values)
+    
+        # 计算滑动窗口的局部变化
+        local_trends = []
+        for i in range(len(values) - window_size + 1):
+            window = values[i:i + window_size]
+            window_x = np.arange(window_size)
+            window_slope = stats.linregress(window_x, window)[0]
+            local_trends.append(window_slope)
+    
+        # 分析结果
+        result = {
+            "overall_slope": slope,  # 整体斜率
+            "p_value": p_value,     # 统计显著性
+            "std_dev": np.std(values), # 数据标准差
+            "local_trend_variability": np.std(local_trends), # 局部趋势变化的标准差
+            "data_points": len(values), # 数据点数量
+        }
+    
+        # 判断趋势类型
+        if p_value < threshold and slope > 0:
+            if result["local_trend_variability"] < abs(slope) * 0.5:
+                result["trend"] = "持续升高"
+                result["description"] = "数据呈现显著的持续上升趋势"
+            else:
+                result["trend"] = "波动上升"
+                result["description"] = "数据整体上升但存在一定波动"
+        elif p_value < threshold and slope < 0:
+            result["trend"] = "持续下降"
+            result["description"] = "数据呈现显著的持续下降趋势"
+        else:
+            result["trend"] = "波动"
+            result["description"] = "数据呈现波动状态，无显著趋势"
+    
+        return result

api/core/tools/builtin_tool/providers/apo_analysis/tools/trend.yaml

@@ -0,0 +1,40 @@
+identity:
+  name: trend
+  author: APO
+  label:
+    en_US: trend
+    zh_Hans: 趋势分析
+    pt_BR: trend
+description:
+  human:
+    en_US: threshold judgement
+    zh_Hans: 基于趋势分析，分析数据是波动还是持续变化
+    pt_BR: threshold judgement
+  llm: threshold judgement
+parameters:
+  - name: metricData
+    type: string
+    required: true
+    label:
+      en_US: metricData
+      zh_Hans: metricData
+      pt_BR: metricData
+    human_description:
+      en_US: time series metric data
+      zh_Hans: 时序指标数据
+      pt_BR: time series metric data
+    llm_description: Time series indicator data
+    form: llm
+  - name: threshold
+    type: string
+    required: false
+    label:
+      en_US: threshold
+      zh_Hans: threshold
+      pt_BR: threshold
+    human_description:
+      en_US: threshold
+      zh_Hans: 趋势分析阈值
+      pt_BR: threshold
+    llm_description: threshold
+    form: llm