feat(llm): 添加通用图片分析功能并优化边缘系统安全状态恢复

- 新增 IMAGE_ANALYZE 动作类型支持通用图片分析
- 实现图片分析操作服务支持单张及多图片分析
- 添加图片分析的提示词、参考图片和调优参数配置
- 集成 MinIO 服务上传功能支持唯一文件名生成
- 添加 FlowMediaParamResolver 支持 URL 数组解析
- 实现边缘系统恢复运行状态的新 API 接口
- 标记旧的安全状态恢复接口为已弃用
- 更新媒体分析操作服务支持详细的输出选项
- 添加图片分析相关的单元测试验证功能

cmvr-iot-admin/src/main/java/com/cmvr/web/controller/api/EdgeSystemController.java

@@ -63,7 +63,18 @@ public class EdgeSystemController extends BaseController {
         return success(JSON.toJSONString(edgeSystemService.getSafetyState(robotId)));
     }
 
-    @ApiOperation("Clear recoverable edge safety errors")
+    @ApiOperation("Restore edge operational state")
+    @PostMapping("/restoreOperationalState")
+    public AjaxResult restoreOperationalState(@RequestBody @Validated EdgeSafetyRecoveryVO request) {
+        return success(JSON.toJSONString(edgeSystemService.restoreOperationalState(
+                request.getRobotId(), request.getReason(), request.getTimeoutMs())));
+    }
+
+    /**
+     * @deprecated Use /restoreOperationalState.
+     */
+    @Deprecated
+    @ApiOperation("[Deprecated] Clear recoverable edge safety errors")
     @PostMapping("/recoverSafetyState")
     public AjaxResult recoverSafetyState(@RequestBody @Validated EdgeSafetyRecoveryVO request) {
         return success(JSON.toJSONString(edgeSystemService.recoverSafetyState(

cmvr-iot-aima/src/main/java/com/cmvr/aima/support/AimaAnalysisVerdict.java

@@ -31,6 +31,7 @@ public final class AimaAnalysisVerdict {
     }
 
     private static boolean isAnalysisAction(ActionEnum action) {
-        return action == ActionEnum.AUDIO_EVENT_CLASSIFY || action == ActionEnum.VIDEO_ANALYZE;
+        return action == ActionEnum.AUDIO_EVENT_CLASSIFY
+                || action == ActionEnum.VIDEO_ANALYZE;
     }
 }

cmvr-iot-aima/src/test/java/com/cmvr/aima/support/AimaAnalysisVerdictTest.java

@@ -20,6 +20,8 @@ public class AimaAnalysisVerdictTest {
         assertEquals(AimaAnalysisVerdict.NOT_PASSED,
                 AimaAnalysisVerdict.resolve(ActionEnum.VIDEO_ANALYZE,
                         new JSONObject().fluentPut("analysisStatus", "FAILED")));
+        assertNull(AimaAnalysisVerdict.resolve(ActionEnum.IMAGE_ANALYZE,
+                new JSONObject().fluentPut("result", "12.6")));
         assertNull(AimaAnalysisVerdict.resolve(ActionEnum.SLEEP,
                 new JSONObject().fluentPut("passed", true)));
     }

cmvr-iot-api/cmvr-iot-edge/cmvr-iot-grpc-client/src/main/java/com/cmvr/edge/client/service/EdgeSystemService.java

@@ -30,6 +30,13 @@ public interface EdgeSystemService {
 
     SafetyCommand.GetSafetyStateCommand.Feedback getSafetyState(String robotId);
 
+    SafetyCommand.RestoreOperationalStateCommand.Feedback restoreOperationalState(
+            String robotId, String reason, Integer timeoutMs);
+
+    /**
+     * @deprecated Use {@link #restoreOperationalState(String, String, Integer)}.
+     */
+    @Deprecated
     SafetyCommand.RecoverSafetyStateCommand.Feedback recoverSafetyState(
             String robotId, String reason, Integer timeoutMs);
 

cmvr-iot-api/cmvr-iot-edge/cmvr-iot-grpc-client/src/main/java/com/cmvr/edge/client/service/impl/EdgeSystemServiceImpl.java

@@ -135,6 +135,39 @@ public class EdgeSystemServiceImpl implements EdgeSystemService {
     }
 
     @Override
+    public SafetyCommand.RestoreOperationalStateCommand.Feedback restoreOperationalState(
+            String robotId, String reason, Integer timeoutMs) {
+        if (robotId == null || robotId.trim().isEmpty()) {
+            throw new GlobalException("机器人ID不能为空");
+        }
+        SafetyCommand.RestoreOperationalStateCommand.Request request =
+                buildRestoreOperationalStateRequest(reason, timeoutMs);
+        SystemServiceGrpc.SystemServiceBlockingStub stub = grpcServiceManager.getGrpcClient(
+                robotId.trim(), SystemServiceGrpc.SystemServiceBlockingStub.class);
+        SafetyCommand.RestoreOperationalStateCommand.Feedback feedback = stub
+                .withDeadlineAfter(request.getTimeoutMs() + 5000L, TimeUnit.MILLISECONDS)
+                .restoreOperationalState(request);
+        requireSuccessfulHeader(feedback == null || !feedback.hasHeader() ? null : feedback.getHeader(),
+                "恢复机器人运行状态");
+        if (!isSuccessfulRecoveryResult(feedback.getResult())) {
+            throw new GlobalException("恢复机器人运行状态未完成: " + feedback.getResult().name());
+        }
+        return feedback;
+    }
+
+    private SafetyCommand.RestoreOperationalStateCommand.Request buildRestoreOperationalStateRequest(
+            String reason, Integer timeoutMs) {
+        int effectiveTimeoutMs = timeoutMs == null ? 15000 : Math.max(1000, Math.min(timeoutMs, 120000));
+        return SafetyCommand.RestoreOperationalStateCommand.Request.newBuilder()
+                .setRecoveryId(UUID.randomUUID().toString())
+                .setReason(reason == null || reason.trim().isEmpty()
+                        ? "Platform manual operational state restore" : reason.trim())
+                .setTimeoutMs(effectiveTimeoutMs)
+                .build();
+    }
+
+    @Override
+    @Deprecated
     public SafetyCommand.RecoverSafetyStateCommand.Feedback recoverSafetyState(
             String robotId, String reason, Integer timeoutMs) {
         SafetyCommand.GetSafetyStateCommand.Feedback current = getSafetyState(robotId);

cmvr-iot-api/cmvr-iot-edge/cmvr-iot-grpc-lib/src/main/java/cmvr/api/SystemServiceGrpc.java

@@ -263,6 +263,37 @@ public final class SystemServiceGrpc {
     return getRecoverSafetyStateMethod;
   }
 
+  private static volatile io.grpc.MethodDescriptor<cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request,
+      cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback> getRestoreOperationalStateMethod;
+
+  @io.grpc.stub.annotations.RpcMethod(
+      fullMethodName = SERVICE_NAME + '/' + "RestoreOperationalState",
+      requestType = cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request.class,
+      responseType = cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback.class,
+      methodType = io.grpc.MethodDescriptor.MethodType.UNARY)
+  public static io.grpc.MethodDescriptor<cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request,
+      cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback> getRestoreOperationalStateMethod() {
+    io.grpc.MethodDescriptor<cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request, cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback> getRestoreOperationalStateMethod;
+    if ((getRestoreOperationalStateMethod = SystemServiceGrpc.getRestoreOperationalStateMethod) == null) {
+      synchronized (SystemServiceGrpc.class) {
+        if ((getRestoreOperationalStateMethod = SystemServiceGrpc.getRestoreOperationalStateMethod) == null) {
+          SystemServiceGrpc.getRestoreOperationalStateMethod = getRestoreOperationalStateMethod =
+              io.grpc.MethodDescriptor.<cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request, cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback>newBuilder()
+              .setType(io.grpc.MethodDescriptor.MethodType.UNARY)
+              .setFullMethodName(generateFullMethodName(SERVICE_NAME, "RestoreOperationalState"))
+              .setSampledToLocalTracing(true)
+              .setRequestMarshaller(io.grpc.protobuf.ProtoUtils.marshaller(
+                  cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request.getDefaultInstance()))
+              .setResponseMarshaller(io.grpc.protobuf.ProtoUtils.marshaller(
+                  cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback.getDefaultInstance()))
+              .setSchemaDescriptor(new SystemServiceMethodDescriptorSupplier("RestoreOperationalState"))
+              .build();
+        }
+      }
+    }
+    return getRestoreOperationalStateMethod;
+  }
+
   /**
    * Creates a new async stub that supports all call types for the service
    */
@@ -367,6 +398,13 @@ public final class SystemServiceGrpc {
       io.grpc.stub.ServerCalls.asyncUnimplementedUnaryCall(getRecoverSafetyStateMethod(), responseObserver);
     }
 
+    /**
+     */
+    public void restoreOperationalState(cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request request,
+        io.grpc.stub.StreamObserver<cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback> responseObserver) {
+      io.grpc.stub.ServerCalls.asyncUnimplementedUnaryCall(getRestoreOperationalStateMethod(), responseObserver);
+    }
+
     @java.lang.Override public final io.grpc.ServerServiceDefinition bindService() {
       return io.grpc.ServerServiceDefinition.builder(getServiceDescriptor())
           .addMethod(
@@ -425,6 +463,13 @@ public final class SystemServiceGrpc {
                 cmvr.api.SafetyCommand.RecoverSafetyStateCommand.Request,
                 cmvr.api.SafetyCommand.RecoverSafetyStateCommand.Feedback>(
                   this, METHODID_RECOVER_SAFETY_STATE)))
+          .addMethod(
+            getRestoreOperationalStateMethod(),
+            io.grpc.stub.ServerCalls.asyncUnaryCall(
+              new MethodHandlers<
+                cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request,
+                cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback>(
+                  this, METHODID_RESTORE_OPERATIONAL_STATE)))
           .build();
     }
   }
@@ -506,6 +551,14 @@ public final class SystemServiceGrpc {
       io.grpc.stub.ClientCalls.asyncUnaryCall(
           getChannel().newCall(getRecoverSafetyStateMethod(), getCallOptions()), request, responseObserver);
     }
+
+    /**
+     */
+    public void restoreOperationalState(cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request request,
+        io.grpc.stub.StreamObserver<cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback> responseObserver) {
+      io.grpc.stub.ClientCalls.asyncUnaryCall(
+          getChannel().newCall(getRestoreOperationalStateMethod(), getCallOptions()), request, responseObserver);
+    }
   }
 
   /**
@@ -577,6 +630,13 @@ public final class SystemServiceGrpc {
       return io.grpc.stub.ClientCalls.blockingUnaryCall(
           getChannel(), getRecoverSafetyStateMethod(), getCallOptions(), request);
     }
+
+    /**
+     */
+    public cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback restoreOperationalState(cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request request) {
+      return io.grpc.stub.ClientCalls.blockingUnaryCall(
+          getChannel(), getRestoreOperationalStateMethod(), getCallOptions(), request);
+    }
   }
 
   /**
@@ -656,6 +716,14 @@ public final class SystemServiceGrpc {
       return io.grpc.stub.ClientCalls.futureUnaryCall(
           getChannel().newCall(getRecoverSafetyStateMethod(), getCallOptions()), request);
     }
+
+    /**
+     */
+    public com.google.common.util.concurrent.ListenableFuture<cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback> restoreOperationalState(
+        cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request request) {
+      return io.grpc.stub.ClientCalls.futureUnaryCall(
+          getChannel().newCall(getRestoreOperationalStateMethod(), getCallOptions()), request);
+    }
   }
 
   private static final int METHODID_GET_SYSTEM_INFO = 0;
@@ -666,6 +734,7 @@ public final class SystemServiceGrpc {
   private static final int METHODID_EXECUTE_ACTION_QUEUE = 5;
   private static final int METHODID_GET_SAFETY_STATE = 6;
   private static final int METHODID_RECOVER_SAFETY_STATE = 7;
+  private static final int METHODID_RESTORE_OPERATIONAL_STATE = 8;
 
   private static final class MethodHandlers<Req, Resp> implements
       io.grpc.stub.ServerCalls.UnaryMethod<Req, Resp>,
@@ -716,6 +785,10 @@ public final class SystemServiceGrpc {
           serviceImpl.recoverSafetyState((cmvr.api.SafetyCommand.RecoverSafetyStateCommand.Request) request,
               (io.grpc.stub.StreamObserver<cmvr.api.SafetyCommand.RecoverSafetyStateCommand.Feedback>) responseObserver);
           break;
+        case METHODID_RESTORE_OPERATIONAL_STATE:
+          serviceImpl.restoreOperationalState((cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Request) request,
+              (io.grpc.stub.StreamObserver<cmvr.api.SafetyCommand.RestoreOperationalStateCommand.Feedback>) responseObserver);
+          break;
         default:
           throw new AssertionError();
       }
@@ -785,6 +858,7 @@ public final class SystemServiceGrpc {
               .addMethod(getExecuteActionQueueMethod())
               .addMethod(getGetSafetyStateMethod())
               .addMethod(getRecoverSafetyStateMethod())
+              .addMethod(getRestoreOperationalStateMethod())
               .build();
         }
       }

cmvr-iot-api/cmvr-iot-edge/cmvr-iot-grpc-lib/src/main/java/cmvr/api/SystemServiceOuterClass.java

@@ -25,7 +25,7 @@ public final class SystemServiceOuterClass {
     java.lang.String[] descriptorData = {
       "\n\035cmvr/api/system_service.proto\022\010cmvr.ap" +
       "i\032\035cmvr/api/system_command.proto\032\035cmvr/a" +
-      "pi/safety_command.proto2\263\006\n\rSystemServic" +
+      "pi/safety_command.proto2\266\007\n\rSystemServic" +
       "e\022b\n\rGetSystemInfo\022&.cmvr.api.GetSystemI" +
       "nfoCommand.Request\032\'.cmvr.api.GetSystemI" +
       "nfoCommand.Feedback\"\000\022h\n\017GetSystemStatus" +
@@ -46,7 +46,10 @@ public final class SystemServiceOuterClass {
       "Feedback\"\000\022q\n\022RecoverSafetyState\022+.cmvr." +
       "api.RecoverSafetyStateCommand.Request\032,." +
       "cmvr.api.RecoverSafetyStateCommand.Feedb" +
-      "ack\"\000b\006proto3"
+      "ack\"\000\022\200\001\n\027RestoreOperationalState\0220.cmvr" +
+      ".api.RestoreOperationalStateCommand.Requ" +
+      "est\0321.cmvr.api.RestoreOperationalStateCo" +
+      "mmand.Feedback\"\000b\006proto3"
     };
     descriptor = com.google.protobuf.Descriptors.FileDescriptor
       .internalBuildGeneratedFileFrom(descriptorData,

cmvr-iot-api/cmvr-iot-edge/cmvr-iot-grpc-lib/src/main/proto/cmvr/api/safety_command.proto

@@ -184,3 +184,21 @@ message RecoverSafetyStateCommand {
         string recovery_id = 7;
     }
 }
+
+message RestoreOperationalStateCommand {
+    message Request {
+        string recovery_id = 1;
+        string reason = 2;
+        uint32 timeout_ms = 3;
+    }
+
+    message Feedback {
+        CommandHeader.Feedback header = 1;
+        SafetyOperationResult result = 2;
+        uint64 previous_safety_epoch = 3;
+        uint64 current_safety_epoch = 4;
+        SystemAdmissionState system_state = 5;
+        repeated SafetyOperationTargetResult targets = 6;
+        string recovery_id = 7;
+    }
+}

cmvr-iot-api/cmvr-iot-edge/cmvr-iot-grpc-lib/src/main/proto/cmvr/api/system_service.proto

@@ -18,5 +18,8 @@ service SystemService {
     rpc ExecuteActionQueue(ActionQueueCommand.Request) returns (ActionQueueCommand.Feedback) {}
 
     rpc GetSafetyState(GetSafetyStateCommand.Request) returns (GetSafetyStateCommand.Feedback) {}
-    rpc RecoverSafetyState(RecoverSafetyStateCommand.Request) returns (RecoverSafetyStateCommand.Feedback) {}
+    rpc RecoverSafetyState(RecoverSafetyStateCommand.Request) returns (RecoverSafetyStateCommand.Feedback) {
+        option deprecated = true;
+    }
+    rpc RestoreOperationalState(RestoreOperationalStateCommand.Request) returns (RestoreOperationalStateCommand.Feedback) {}
 }

cmvr-iot-common/pom.xml

@@ -156,6 +156,13 @@
             <version>${swagger.annotations.version}</version>
         </dependency>
 
+        <dependency>
+            <groupId>junit</groupId>
+            <artifactId>junit</artifactId>
+            <version>4.13.2</version>
+            <scope>test</scope>
+        </dependency>
+
     </dependencies>
 
 </project>

cmvr-iot-common/src/main/java/com/cmvr/common/core/minio/MinioService.java

@@ -25,6 +25,7 @@ import java.util.Date;
 import java.util.HashMap;
 import java.util.Locale;
 import java.util.Map;
+import java.util.UUID;
 
 @Slf4j
 @Component
@@ -96,16 +97,12 @@ public class MinioService {
         if (StringUtils.isBlank(originalFilename)) {
             throw new RuntimeException();
         }
-        String fileName = System.currentTimeMillis() + "";
+        String fileName = buildUniqueFileName(originalFilename);
-        if (originalFilename.lastIndexOf(".") >= 0) {
-            fileName += originalFilename.substring(originalFilename.lastIndexOf("."));
-        }
         String objectName = DateUtil.format(new Date(), DatePattern.PURE_DATE_PATTERN) + "/" + fileName;
         try {
             InputStream stream = new ByteArrayInputStream(file.getBytes());
             PutObjectArgs objectArgs = PutObjectArgs.builder().bucket(bucketName).object(path + "/" + objectName)
                     .stream(stream, file.getSize(), -1).contentType(file.getContentType()).build();
-            //文件名称相同会覆盖
             minioClient.putObject(objectArgs);
         } catch (Exception e) {
             log.error("上传图片失败,", e);
@@ -121,11 +118,7 @@ public class MinioService {
             throw new RuntimeException("文件名为空");
         }
 
-        // 生成文件名，使用当前时间戳
+        String fileName = buildUniqueFileName(originalFilename);
-        String fileName = System.currentTimeMillis() + "";
-        if (originalFilename.lastIndexOf(".") >= 0) {
-            fileName += originalFilename.substring(originalFilename.lastIndexOf("."));
-        }
 
         // 构建文件在 MinIO 中的路径
         String objectName = DateUtil.format(new Date(), DatePattern.PURE_DATE_PATTERN) + "/" + fileName;
@@ -178,13 +171,12 @@ public class MinioService {
      */
     public String upload(String bucketName, String path, String name, byte[] fileByte, String contentType) {
 
-        String fileName = System.currentTimeMillis() + name.substring(name.lastIndexOf("."));
+        String fileName = buildUniqueFileName(name);
         String objectName = DateUtil.format(new Date(), DatePattern.PURE_DATE_PATTERN) + "/" + fileName;
         try {
             InputStream stream = new ByteArrayInputStream(fileByte);
             PutObjectArgs objectArgs = PutObjectArgs.builder().bucket(bucketName).object(path + "/" + objectName)
                     .stream(stream, fileByte.length, -1).contentType(contentType).build();
-            //文件名称相同会覆盖
             minioClient.putObject(objectArgs);
         } catch (Exception e) {
             log.error("上传文件失败,", e);
@@ -193,6 +185,30 @@ public class MinioService {
         return objectName;
     }
 
+    /**
+     * Generates a collision-resistant object file name while retaining the source extension.
+     */
+    static String buildUniqueFileName(String originalFilename) {
+        String extension = "";
+        if (StringUtils.isNotBlank(originalFilename)) {
+            int separatorIndex = Math.max(originalFilename.lastIndexOf('/'), originalFilename.lastIndexOf('\\'));
+            int extensionIndex = originalFilename.lastIndexOf('.');
+            if (extensionIndex > separatorIndex) {
+                extension = originalFilename.substring(extensionIndex);
+            }
+        }
+        String uuid = UUID.randomUUID().toString().replace("-", "");
+        return System.currentTimeMillis() + "-" + uuid + extension;
+    }
+
+    static String buildUniqueObjectName(String objectName) {
+        String normalized = StringUtils.defaultString(objectName).replace('\\', '/');
+        int separatorIndex = normalized.lastIndexOf('/');
+        String directory = separatorIndex >= 0 ? normalized.substring(0, separatorIndex + 1) : "";
+        String originalFilename = separatorIndex >= 0 ? normalized.substring(separatorIndex + 1) : normalized;
+        return directory + buildUniqueFileName(originalFilename);
+    }
+
 
     /**
      * 删除
@@ -265,21 +281,24 @@ public class MinioService {
      * @param inputStream 输入流
      * @param size 文件大小
      * @param contentType 内容类型
+     * @return 实际写入的唯一对象名称
      */
-    public void uploadStream(String bucketName, String objectName, InputStream inputStream, long size, String contentType) {
+    public String uploadStream(String bucketName, String objectName, InputStream inputStream, long size, String contentType) {
+        String uniqueObjectName = buildUniqueObjectName(objectName);
         try {
             Map<String, String> extraHeaders = new HashMap<>();
             extraHeaders.put("x-amz-acl", "public-read");
             PutObjectArgs objectArgs = PutObjectArgs.builder()
                     .bucket(bucketName)
-                    .object(objectName)
+                    .object(uniqueObjectName)
                     .stream(inputStream, size, -1)
                     .extraHeaders(extraHeaders)
                     .contentType(contentType)
                     .build();
             minioClient.putObject(objectArgs);
+            return uniqueObjectName;
         } catch (Exception e) {
-            log.error("上传文件流失败, objectName: {}", objectName, e);
+            log.error("上传文件流失败, objectName: {}", uniqueObjectName, e);
             throw new RuntimeException("上传文件流失败", e);
         }
     }

cmvr-iot-common/src/test/java/com/cmvr/common/core/minio/MinioServiceTest.java

@@ -0,0 +1,45 @@
+package com.cmvr.common.core.minio;
+
+import org.junit.Test;
+
+import java.util.Set;
+import java.util.concurrent.ConcurrentHashMap;
+import java.util.stream.IntStream;
+
+import static org.junit.Assert.assertEquals;
+import static org.junit.Assert.assertFalse;
+import static org.junit.Assert.assertTrue;
+
+public class MinioServiceTest {
+
+    @Test
+    public void generatesUniqueNamesUnderConcurrency() {
+        int count = 10_000;
+        Set<String> names = ConcurrentHashMap.newKeySet();
+
+        IntStream.range(0, count).parallel()
+                .mapToObj(index -> MinioService.buildUniqueFileName("recording.mp4"))
+                .forEach(names::add);
+
+        assertEquals(count, names.size());
+        assertTrue(names.stream().allMatch(name ->
+                name.matches("\\d{13}-[0-9a-f]{32}\\.mp4")));
+    }
+
+    @Test
+    public void retainsExtensionAndSupportsExtensionlessNames() {
+        String imageName = MinioService.buildUniqueFileName("folder/archive/photo.JPG");
+        String extensionlessName = MinioService.buildUniqueFileName("README");
+
+        assertTrue(imageName.endsWith(".JPG"));
+        assertFalse(extensionlessName.contains("."));
+    }
+
+    @Test
+    public void retainsObjectDirectoryWhenGeneratingStreamUploadName() {
+        String objectName = MinioService.buildUniqueObjectName("inspection/alerts/source.jpg");
+
+        assertTrue(objectName.matches(
+                "inspection/alerts/\\d{13}-[0-9a-f]{32}\\.jpg"));
+    }
+}

cmvr-iot-inspection/src/main/java/com/cmvr/inspection/service/impl/InspectionDetectionAlertServiceImpl.java

@@ -293,10 +293,11 @@ public class InspectionDetectionAlertServiceImpl implements IInspectionDetection
 
         String extension = extensionFor(image.getMediaType());
         String datePath = LocalDate.now().format(DateTimeFormatter.BASIC_ISO_DATE);
-        String objectName = IMAGE_OBJECT_PREFIX + "/" + datePath + "/" + alertId + extension;
+        String requestedObjectName = IMAGE_OBJECT_PREFIX + "/" + datePath + "/" + alertId + extension;
         try (ByteArrayInputStream inputStream = new ByteArrayInputStream(imageBytes))
         {
-            minioService.uploadStream(minioProperties.getBucketName(), objectName, inputStream,
+            String objectName = minioService.uploadStream(
+                    minioProperties.getBucketName(), requestedObjectName, inputStream,
                     imageBytes.length, image.getMediaType());
             return new StoredImage(objectName, buildMinioUrl(objectName));
         }

cmvr-iot-test/src/main/java/com/cmvr/test/enums/ActionEnum.java

@@ -90,6 +90,7 @@ public enum ActionEnum {
     GET_CURRENT_PAGE("LLM", "GET_CURRENT_PAGE", "获取当前页面名称"),
     AUDIO_EVENT_CLASSIFY("LLM", "AUDIO_EVENT_CLASSIFY", "声音类型检测"),
     VIDEO_ANALYZE("LLM", "VIDEO_ANALYZE", "视频智能分析"),
+    IMAGE_ANALYZE("LLM", "IMAGE_ANALYZE", "通用图片分析"),
 
     // 触控交互
     TI_PATH_SEARCH("EDGE", "TI_PATH_SEARCH", "路径搜索"),

cmvr-iot-test/src/main/java/com/cmvr/test/flow/builder/FlowModelBuilder.java

@@ -178,6 +178,10 @@ public class FlowModelBuilder {
 
         // 设置动作类型
         String actionStr = props.getString("action");
+        if (isLegacyImageAnalysisNode(nodeJson, props)) {
+            actionStr = ActionEnum.IMAGE_ANALYZE.getAction();
+            props.put("action", actionStr);
+        }
         ActionEnum action = ActionEnum.fromAction(actionStr);
         if (ObjUtil.isEmpty(action)) {
             throw new IllegalArgumentException("无法识别的action：" + actionStr);
@@ -205,6 +209,28 @@ public class FlowModelBuilder {
         return wrapper;
     }
 
+    /**
+     * Early image-analysis nodes could retain TOUCH_COORDINATES metadata while already
+     * carrying the new component type and profile. Normalize those persisted graphs at runtime.
+     */
+    private static boolean isLegacyImageAnalysisNode(JSONObject nodeJson, JSONObject props) {
+        if (!"imageAnalysis".equalsIgnoreCase(nodeJson.getString("type"))) {
+            return false;
+        }
+        JSONArray params = props.getJSONArray("nodeParams");
+        if (params == null) {
+            return false;
+        }
+        for (int i = 0; i < params.size(); i++) {
+            JSONObject param = params.getJSONObject(i);
+            if ("profileCode".equals(param.getString("name"))
+                    && "common.image_analysis.v1".equals(param.getString("input"))) {
+                return true;
+            }
+        }
+        return false;
+    }
+
     /**
      * 解析分支条件逻辑
      */

cmvr-iot-test/src/main/java/com/cmvr/test/flow/runtime/operator/FlowMediaParamResolver.java

@@ -1,5 +1,7 @@
 package com.cmvr.test.flow.runtime.operator;
 
+import com.alibaba.fastjson2.JSON;
+import com.alibaba.fastjson2.JSONArray;
 import com.alibaba.fastjson2.JSONObject;
 import org.apache.commons.lang3.StringUtils;
 
@@ -21,17 +23,29 @@ public final class FlowMediaParamResolver {
     public static List<String> urls(JSONObject params, String name) {
         Object value = params == null ? null : params.get(name);
         if (value instanceof List<?> values) {
-            List<String> urls = new ArrayList<>();
+            return normalizeUrls(values);
-            for (Object item : values) {
-                String url = normalizeUrl(item);
-                if (url != null && !"null".equalsIgnoreCase(url)) {
-                    urls.add(url);
-                }
-            }
-            return urls;
         }
-        String url = normalizeUrl(value);
+        String text = normalizeUrl(value);
-        return url == null ? Collections.emptyList() : List.of(url);
+        if (text != null && text.startsWith("[") && text.endsWith("]")) {
+            try {
+                JSONArray values = JSON.parseArray(text);
+                return normalizeUrls(values);
+            } catch (Exception ignored) {
+                // Keep treating malformed JSON as a scalar so validation can report it upstream.
+            }
+        }
+        return text == null ? Collections.emptyList() : List.of(text);
+    }
+
+    private static List<String> normalizeUrls(List<?> values) {
+        List<String> urls = new ArrayList<>();
+        for (Object item : values) {
+            String url = normalizeUrl(item);
+            if (url != null && !"null".equalsIgnoreCase(url)) {
+                urls.add(url);
+            }
+        }
+        return urls;
     }
 
     private static String normalizeUrl(Object value) {

cmvr-iot-test/src/main/java/com/cmvr/test/flow/runtime/operator/llm/LLMMediaAnalysisOperateService.java

@@ -11,6 +11,7 @@ import lombok.RequiredArgsConstructor;
 import org.apache.commons.lang3.StringUtils;
 import org.springframework.stereotype.Service;
 
+import java.util.List;
 import java.util.Set;
 import java.util.UUID;
 
@@ -19,27 +20,33 @@ import java.util.UUID;
 public class LLMMediaAnalysisOperateService implements LLMOperateService {
 
     private static final Set<String> VIDEO_ANALYSIS_MODES = Set.of("AUTO", "FAST", "ACCURATE");
-
     private final MediaAnalysisClient mediaAnalysisClient;
 
     @Override
     public boolean supports(ActionEnum action) {
-        return action == ActionEnum.AUDIO_EVENT_CLASSIFY || action == ActionEnum.VIDEO_ANALYZE;
+        return action == ActionEnum.AUDIO_EVENT_CLASSIFY
+                || action == ActionEnum.VIDEO_ANALYZE
+                || action == ActionEnum.IMAGE_ANALYZE;
     }
 
     @Override
     public TaskNodeExecuteResult execute(TaskNodeExecuteMessage message) {
         JSONObject input = message.getInputParams();
         boolean audio = message.getAction() == ActionEnum.AUDIO_EVENT_CLASSIFY;
-        String mediaParam = audio ? "audioUrl" : "videoUrl";
+        boolean image = message.getAction() == ActionEnum.IMAGE_ANALYZE;
+        String mediaParam = audio ? "audioUrl" : (image ? "imageUrl" : "videoUrl");
+        List<String> mediaUrls = FlowMediaParamResolver.urls(input, mediaParam);
         String mediaUrl = FlowMediaParamResolver.lastUrl(input, mediaParam);
         String profileCode = StringUtils.trimToNull(input.getString("profileCode"));
         if (StringUtils.isBlank(mediaUrl)) {
-            throw new GlobalException("{}不能为空", audio ? "音频地址" : "视频地址");
+            throw new GlobalException("{}不能为空", audio ? "音频地址" : (image ? "图片地址" : "视频地址"));
         }
         if (profileCode == null) {
             throw new GlobalException("分析场景不能为空");
         }
+        if (image && mediaUrls.size() > 12) {
+            throw new GlobalException("图片分析一次最多支持12张图片");
+        }
 
         JSONObject context = new JSONObject();
         context.put("taskId", message.getTaskId());
@@ -58,7 +65,7 @@ public class LLMMediaAnalysisOperateService implements LLMOperateService {
                 options.put("expectedLabel", expectedLabel);
                 options.put("expectedLabelName", StringUtils.defaultIfBlank(expectedLabelName, expectedLabel));
             }
-        } else {
+        } else if (!image) {
             options.put("instruction", StringUtils.defaultString(input.getString("instruction")));
             String analysisMode = StringUtils.defaultIfBlank(input.getString("analysisMode"), "AUTO")
                     .trim().toUpperCase();
@@ -67,17 +74,42 @@ public class LLMMediaAnalysisOperateService implements LLMOperateService {
             }
             options.put("analysisMode", analysisMode);
             options.put("decisionPolicy", "FAIL_CLOSED");
+            options.put("detailedOutput", Boolean.TRUE.equals(input.getBoolean("detailedOutput")));
             JSONObject tuning = buildVideoTuning(input.getJSONObject("analysisTuning"));
             if (!tuning.isEmpty()) {
                 options.put("tuning", tuning);
             }
+        } else {
+            String analysisMode = StringUtils.defaultIfBlank(input.getString("analysisMode"), "AUTO")
+                    .trim().toUpperCase();
+            if (!VIDEO_ANALYSIS_MODES.contains(analysisMode)) {
+                throw new GlobalException("图片分析模式无效: {}", analysisMode);
+            }
+            String prompt = resolveImagePrompt(input);
+            if (StringUtils.isBlank(prompt)) {
+                throw new GlobalException("图片分析提示词不能为空");
+            }
+            options.put("prompt", prompt);
+            options.put("analysisMode", analysisMode);
+            String referenceImageUrl = FlowMediaParamResolver.lastUrl(input, "referenceImageUrl");
+            if (StringUtils.isNotBlank(referenceImageUrl)) {
+                options.put("referenceImageUrl", referenceImageUrl);
+            }
+            JSONObject tuning = buildImageTuning(input.getJSONObject("analysisTuning"));
+            if (!tuning.isEmpty()) {
+                options.put("tuning", tuning);
+            }
         }
 
         JSONObject request = new JSONObject();
         request.put("requestId", buildRequestId(message));
-        request.put("analysisType", audio ? "AUDIO_CLASSIFICATION" : "VIDEO_ANALYSIS");
+        request.put("analysisType", audio ? "AUDIO_CLASSIFICATION"
+                : (image ? "IMAGE_ANALYSIS" : "VIDEO_ANALYSIS"));
         request.put("profileCode", profileCode);
         request.put("mediaUrl", mediaUrl);
+        if (image) {
+            request.put("mediaUrls", mediaUrls);
+        }
         request.put("options", options);
         request.put("context", context);
 
@@ -94,6 +126,9 @@ public class LLMMediaAnalysisOperateService implements LLMOperateService {
         output.put("analysisModel", response.getJSONObject("model"));
         output.put("analysisTimingMs", response.getLong("timingMs"));
         output.put("mediaUrl", mediaUrl);
+        if (image) {
+            output.put("mediaUrls", mediaUrls);
+        }
         if (audio) {
             applyAudioVerdict(output, expectedLabel, expectedLabelName);
         }
@@ -127,20 +162,72 @@ public class LLMMediaAnalysisOperateService implements LLMOperateService {
         if (source == null || source.isEmpty()) {
             return tuning;
         }
-        Integer maxFrames = source.getInteger("maxFrames");
+        Double sampleFps = source.getDouble("sampleFps");
         Integer maxWidth = source.getInteger("maxWidth");
         Double confidenceThreshold = source.getDouble("confidenceThreshold");
         Boolean fallbackToAccurate = source.getBoolean("fallbackToAccurate");
-        validateRange("抽帧数量", maxFrames, 2, 16);
+        validateRange("采样帧率", sampleFps, 0.25D, 6D);
         validateRange("图片宽度", maxWidth, 640, 1280);
         validateRange("通过置信度", confidenceThreshold, 0.5D, 0.95D);
-        if (maxFrames != null) tuning.put("maxFrames", maxFrames);
+        if (sampleFps != null) tuning.put("sampleFps", sampleFps);
         if (maxWidth != null) tuning.put("maxWidth", maxWidth);
         if (confidenceThreshold != null) tuning.put("confidenceThreshold", confidenceThreshold);
         if (fallbackToAccurate != null) tuning.put("fallbackToAccurate", fallbackToAccurate);
         return tuning;
     }
 
+    private JSONObject buildImageTuning(JSONObject source) {
+        JSONObject tuning = new JSONObject();
+        if (source == null || source.isEmpty()) {
+            return tuning;
+        }
+        Integer maxWidth = source.getInteger("maxWidth");
+        Integer maxImages = source.getInteger("maxImages");
+        Integer maxOutputTokens = source.getInteger("maxOutputTokens");
+        validateRange("图片宽度", maxWidth, 640, 1600);
+        validateRange("图片数量", maxImages, 1, 12);
+        validateRange("最大输出长度", maxOutputTokens, 32, 4096);
+        if (maxWidth != null) tuning.put("maxWidth", maxWidth);
+        if (maxImages != null) tuning.put("maxImages", maxImages);
+        if (maxOutputTokens != null) tuning.put("maxOutputTokens", maxOutputTokens);
+        return tuning;
+    }
+
+    private String resolveImagePrompt(JSONObject input) {
+        String prompt = StringUtils.trimToNull(input.getString("prompt"));
+        if (prompt != null) {
+            return prompt;
+        }
+
+        // Old image-analysis nodes stored a fixed task type, target and pass criterion.
+        // Convert those fields into one neutral prompt so saved workflows remain runnable
+        // without retaining the old pass/fail semantics.
+        String target = StringUtils.trimToNull(input.getString("targetDescription"));
+        String instruction = StringUtils.trimToNull(input.getString("instruction"));
+        String taskType = StringUtils.trimToEmpty(input.getString("taskType")).toUpperCase();
+        StringBuilder legacyPrompt = new StringBuilder();
+        if (target != null) {
+            legacyPrompt.append(target);
+        }
+        if (instruction != null) {
+            if (!legacyPrompt.isEmpty()) {
+                legacyPrompt.append('\n');
+            }
+            legacyPrompt.append(instruction);
+        }
+        if ("ICON_TEMPLATE_LOCATE".equals(taskType)) {
+            if (!legacyPrompt.isEmpty()) {
+                legacyPrompt.append('\n');
+            }
+            legacyPrompt.append("最后一张图片是参考小图，请在其他图片中查找对应目标。");
+        }
+        if (!legacyPrompt.isEmpty()) {
+            legacyPrompt.append('\n');
+        }
+        legacyPrompt.append("只返回分析结果，不输出通过、未通过或置信度。");
+        return StringUtils.trimToEmpty(legacyPrompt.toString());
+    }
+
     private void validateRange(String name, Number value, double minimum, double maximum) {
         if (value != null && (value.doubleValue() < minimum || value.doubleValue() > maximum)) {
             throw new GlobalException("{}必须在{}到{}之间", name, minimum, maximum);

cmvr-iot-test/src/main/java/com/cmvr/test/service/FlowActionExecutorService.java

@@ -327,7 +327,9 @@ public class FlowActionExecutorService {
     private String executeLLMAction(ActionEnum action, FlowActionRequestVO req) {
         log.info("LLM 执行动作: {}", action);
 
-        if (action == ActionEnum.AUDIO_EVENT_CLASSIFY || action == ActionEnum.VIDEO_ANALYZE) {
+        if (action == ActionEnum.AUDIO_EVENT_CLASSIFY
+                || action == ActionEnum.VIDEO_ANALYZE
+                || action == ActionEnum.IMAGE_ANALYZE) {
             for (LLMOperateService service : llmOperateServices) {
                 if (service.supports(action)) {
                     return resultToString(service.execute(buildSingleNodeMessage(action, req.getPayload())));

cmvr-iot-test/src/test/java/com/cmvr/test/flow/builder/FlowModelBuilderTest.java

@@ -0,0 +1,34 @@
+package com.cmvr.test.flow.builder;
+
+import com.cmvr.test.enums.ActionEnum;
+import org.junit.Test;
+
+import static org.junit.Assert.assertEquals;
+
+public class FlowModelBuilderTest {
+
+    @Test
+    public void normalizesEarlyImageAnalysisNodeWithTouchAction() {
+        String flow = """
+                {
+                  "nodes": [{
+                    "id": "image-1",
+                    "type": "imageAnalysis",
+                    "properties": {
+                      "name": "图片智能分析",
+                      "action": "TOUCH_COORDINATES",
+                      "nodeParams": [
+                        {"name":"profileCode","type":"input","input":"common.image_analysis.v1"},
+                        {"name":"imageUrl","type":"input","input":["http://minio/image.jpg"]}
+                      ]
+                    }
+                  }],
+                  "edges": []
+                }
+                """;
+
+        FlowGraph graph = FlowModelBuilder.buildExecutableGraph(flow);
+
+        assertEquals(ActionEnum.IMAGE_ANALYZE, graph.getNodeMap().get("image-1").getAction());
+    }
+}

cmvr-iot-test/src/test/java/com/cmvr/test/flow/runtime/operator/llm/LLMMediaAnalysisOperateServiceTest.java

@@ -12,6 +12,7 @@ import org.junit.Test;
 import java.util.concurrent.atomic.AtomicReference;
 
 import static org.junit.Assert.assertEquals;
+import static org.junit.Assert.assertNull;
 import static org.junit.Assert.assertTrue;
 
 public class LLMMediaAnalysisOperateServiceTest {
@@ -95,8 +96,9 @@ public class LLMMediaAnalysisOperateServiceTest {
                 .fluentPut("profileCode", "aima.power_video.v1")
                 .fluentPut("videoUrl", "https://files.example/video.mp4")
                 .fluentPut("analysisMode", "FAST")
+                .fluentPut("detailedOutput", true)
                 .fluentPut("analysisTuning", new JSONObject()
-                        .fluentPut("maxFrames", 8)
+                        .fluentPut("sampleFps", 2D)
                         .fluentPut("maxWidth", 960)
                         .fluentPut("confidenceThreshold", 0.8D)
                         .fluentPut("fallbackToAccurate", false))
@@ -110,8 +112,10 @@ public class LLMMediaAnalysisOperateServiceTest {
         assertEquals("FAST", captured.get().getJSONObject("options").getString("analysisMode"));
         assertEquals("FAIL_CLOSED",
                 captured.get().getJSONObject("options").getString("decisionPolicy"));
+        assertEquals(true,
+                captured.get().getJSONObject("options").getBooleanValue("detailedOutput"));
         JSONObject tuning = captured.get().getJSONObject("options").getJSONObject("tuning");
-        assertEquals(8, tuning.getIntValue("maxFrames"));
+        assertEquals(2D, tuning.getDoubleValue("sampleFps"), 0.0001D);
         assertEquals(960, tuning.getIntValue("maxWidth"));
         assertEquals(0.8D, tuning.getDoubleValue("confidenceThreshold"), 0.0001D);
         assertEquals(false, tuning.getBooleanValue("fallbackToAccurate"));
@@ -124,11 +128,85 @@ public class LLMMediaAnalysisOperateServiceTest {
                 .fluentPut("profileCode", "aima.power_video.v1")
                 .fluentPut("videoUrl", "https://files.example/video.mp4")
                 .fluentPut("analysisMode", "FAST")
-                .fluentPut("analysisTuning", new JSONObject().fluentPut("maxFrames", 30));
+                .fluentPut("analysisTuning", new JSONObject().fluentPut("sampleFps", 6.5D));
 
         service.execute(message(ActionEnum.VIDEO_ANALYZE, input));
     }
 
+    @Test
+    public void forwardsMultipleImagesAndReferenceImage() {
+        AtomicReference<JSONObject> captured = new AtomicReference<>();
+        MediaAnalysisClient client = request -> {
+            captured.set(request);
+            return new JSONObject()
+                    .fluentPut("status", "SUCCEEDED")
+                    .fluentPut("analysisType", "IMAGE_ANALYSIS")
+                    .fluentPut("profileCode", "common.image_analysis.v1")
+                    .fluentPut("result", new JSONObject()
+                            .fluentPut("result", new JSONObject().fluentPut("x", 120).fluentPut("y", 80))
+                            .fluentPut("resultText", "{\"x\":120,\"y\":80}"));
+        };
+        LLMMediaAnalysisOperateService service = new LLMMediaAnalysisOperateService(client);
+        JSONObject input = new JSONObject()
+                .fluentPut("profileCode", "common.image_analysis.v1")
+                .fluentPut("imageUrl", new JSONArray()
+                        .fluentAdd("https://files.example/scene-1.jpg")
+                        .fluentAdd("https://files.example/scene-2.jpg"))
+                .fluentPut("referenceImageUrl", "https://files.example/icon.jpg")
+                .fluentPut("prompt", "最后一张图片是参考图，输出匹配图标的中心坐标 JSON")
+                .fluentPut("analysisMode", "ACCURATE")
+                .fluentPut("analysisTuning", new JSONObject()
+                        .fluentPut("maxWidth", 1280)
+                        .fluentPut("maxImages", 8)
+                        .fluentPut("maxOutputTokens", 256));
+
+        TaskNodeExecuteResult result = service.execute(message(ActionEnum.IMAGE_ANALYZE, input));
+
+        assertTrue(result.isSuccess());
+        assertEquals("IMAGE_ANALYSIS", captured.get().getString("analysisType"));
+        assertEquals(2, captured.get().getJSONArray("mediaUrls").size());
+        JSONObject options = captured.get().getJSONObject("options");
+        assertEquals("最后一张图片是参考图，输出匹配图标的中心坐标 JSON", options.getString("prompt"));
+        assertEquals("https://files.example/icon.jpg", options.getString("referenceImageUrl"));
+        assertEquals(8, options.getJSONObject("tuning").getIntValue("maxImages"));
+        assertEquals(256, options.getJSONObject("tuning").getIntValue("maxOutputTokens"));
+        assertEquals(2, result.getOutputParams().getJSONArray("mediaUrls").size());
+        assertNull(result.getOutputParams().get("passed"));
+    }
+
+    @Test
+    public void restoresStaticImageUrlArraySerializedByLegacyFlowParamModel() {
+        AtomicReference<JSONObject> captured = new AtomicReference<>();
+        MediaAnalysisClient client = request -> {
+            captured.set(request);
+            return new JSONObject()
+                    .fluentPut("status", "SUCCEEDED")
+                    .fluentPut("analysisType", "IMAGE_ANALYSIS")
+                    .fluentPut("profileCode", "common.image_analysis.v1")
+                    .fluentPut("result", new JSONObject()
+                            .fluentPut("result", "图片正常")
+                            .fluentPut("resultText", "图片正常"));
+        };
+        LLMMediaAnalysisOperateService service = new LLMMediaAnalysisOperateService(client);
+        JSONObject input = new JSONObject()
+                .fluentPut("profileCode", "common.image_analysis.v1")
+                .fluentPut("taskType", "GENERAL_INSPECTION")
+                .fluentPut("imageUrl", "[\"http://192.168.28.10:9000/cmvr-iot/path%2Fimage.jpg\"]")
+                .fluentPut("targetDescription", "检查图片")
+                .fluentPut("instruction", "满足要求则通过")
+                .fluentPut("analysisTuning", "{\"maxWidth\":1120,\"confidenceThreshold\":0.7,\"maxImages\":12}");
+
+        service.execute(message(ActionEnum.IMAGE_ANALYZE, input));
+
+        assertEquals("http://192.168.28.10:9000/cmvr-iot/path%2Fimage.jpg",
+                captured.get().getString("mediaUrl"));
+        assertEquals(1, captured.get().getJSONArray("mediaUrls").size());
+        assertEquals(1120, captured.get().getJSONObject("options")
+                .getJSONObject("tuning").getIntValue("maxWidth"));
+        assertTrue(captured.get().getJSONObject("options").getString("prompt").contains("检查图片"));
+        assertTrue(captured.get().getJSONObject("options").getString("prompt").contains("满足要求则通过"));
+    }
+
     private TaskNodeExecuteMessage message(ActionEnum action, JSONObject input) {
         TaskNodeExecuteMessage message = new TaskNodeExecuteMessage();
         message.setAction(action);

cmvr-iot-tts/src/main/java/com/cmvr/tts/service/impl/CorpusInfoServiceImpl.java

@@ -153,8 +153,12 @@ public class CorpusInfoServiceImpl implements ICorpusInfoService {
         }
         
         try {
-            String objectName = "corpus/" + DateUtils.datePath() + "/" + corpusId + "_" + System.currentTimeMillis() + ".wav";
+            String uploadPath = "corpus/" + DateUtils.datePath();
-            minioService.upload("cmvr-iot", "corpus/" + DateUtils.datePath(), file);
+            String uploadedObjectName = minioService.upload("cmvr-iot", uploadPath, file);
+            if (StringUtils.isEmpty(uploadedObjectName)) {
+                throw new RuntimeException("上传音频失败");
+            }
+            String objectName = uploadPath + "/" + uploadedObjectName;
             
             corpus.setAudioFilePath(objectName);
             corpus.setFileSize(file.getSize());

cmvr-iot-tts/src/main/java/com/cmvr/tts/service/impl/TtsSynthesizeTaskServiceImpl.java

@@ -285,11 +285,12 @@ public class TtsSynthesizeTaskServiceImpl implements ITtsSynthesizeTaskService {
             
             if (audioBytes != null && audioBytes.length > 0) {
                 // 上传音频到Minio
-                String objectName = "tts/" + DateUtils.datePath() + "/" + task.getTaskCode() + ".wav";
+                String requestedObjectName = "tts/" + DateUtils.datePath() + "/" + task.getTaskCode() + ".wav";
                 ByteArrayInputStream inputStream = new ByteArrayInputStream(audioBytes);
-                
+
                 // 使用MinioService上传
-                minioService.uploadStream(bucketName, objectName, inputStream, audioBytes.length, "audio/wav");
+                String objectName = minioService.uploadStream(
+                        bucketName, requestedObjectName, inputStream, audioBytes.length, "audio/wav");
                 
                 // 更新任务状态
                 task.setStatus("2");