android fragment回退栈 activity 回退

1.  回退（驳回）

回退的思路就是动态更改节点的流向。先遇水搭桥，最后再过河拆桥。

具体操作如下：

1. 取得当前节点的信息
2. 取得当前节点的上一个节点的信息
3. 保存当前节点的流向
4. 新建流向，由当前节点指向上一个节点
5. 将当前节点的流向设置为上面新建的流向
6. 当前节点完成任务
7. 将当前节点的流向还原
8. 取得之前上个节点的执行人
9. 设置上个节点的assignee为之前的执行人

代码实现起来可能是这样的： 

@Test
public void huitui() throws Exception {
    ProcessEngine processEngine = ProcessEngines.getDefaultProcessEngine();
    TaskService taskService = processEngine.getTaskService();
    Task task = taskService.createTaskQuery().processInstanceId("55001").singleResult();
    backProcess(task);
}

/**
 * 驳回 / 回退
 * 按照这种方法，可以回退至任意节点
 * @param task
 * @throws Exception
 */
public void backProcess(Task task) throws Exception {
    ProcessEngine processEngine = ProcessEngines.getDefaultProcessEngine();
    HistoryService historyService = processEngine.getHistoryService();
    RepositoryService repositoryService = processEngine.getRepositoryService();
    TaskService taskService = processEngine.getTaskService();

    String processInstanceId = task.getProcessInstanceId();

    //  获取所有历史任务（按创建时间降序）
    List<HistoricTaskInstance> hisTaskList = historyService.createHistoricTaskInstanceQuery()
            .processInstanceId(processInstanceId)
            .orderByTaskCreateTime()
            .desc()
            .list();

    List<HistoricActivityInstance> hisActivityList = historyService.createHistoricActivityInstanceQuery()
            .processInstanceId(processInstanceId).list();

    if (CollectionUtils.isEmpty(hisTaskList) || hisTaskList.size() < 2) {
        return;
    }

    //  当前任务
    HistoricTaskInstance currentTask = hisTaskList.get(0);
    //  前一个任务
    HistoricTaskInstance lastTask = hisTaskList.get(1);
    //  当前活动
    HistoricActivityInstance currentActivity = hisActivityList.stream().filter(e -> currentTask.getId().equals(e.getTaskId())).collect(Collectors.toList()).get(0);
    //  前一个活动
    HistoricActivityInstance lastActivity = hisActivityList.stream().filter(e -> lastTask.getId().equals(e.getTaskId())).collect(Collectors.toList()).get(0);

    BpmnModel bpmnModel = repositoryService.getBpmnModel(task.getProcessDefinitionId());

    //  获取前一个活动节点
    FlowNode lastFlowNode = (FlowNode) bpmnModel.getMainProcess().getFlowElement(lastActivity.getActivityId());
    //  获取当前活动节点
    FlowNode currentFlowNode = (FlowNode) bpmnModel.getMainProcess().getFlowElement(currentActivity.getActivityId());

    //  临时保存当前活动的原始方向
    List<SequenceFlow> originalSequenceFlowList = new ArrayList<>();
    originalSequenceFlowList.addAll(currentFlowNode.getOutgoingFlows());
    //  清理活动方向
    currentFlowNode.getOutgoingFlows().clear();

    //  建立新方向
    SequenceFlow newSequenceFlow = new SequenceFlow();
    newSequenceFlow.setId("newSequenceFlowId");
    newSequenceFlow.setSourceFlowElement(currentFlowNode);
    newSequenceFlow.setTargetFlowElement(lastFlowNode);
    List<SequenceFlow> newSequenceFlowList = new ArrayList<>();
    newSequenceFlowList.add(newSequenceFlow);
    //  当前节点指向新的方向
    currentFlowNode.setOutgoingFlows(newSequenceFlowList);

    //  完成当前任务
    taskService.complete(task.getId());

    //  重新查询当前任务
    Task nextTask = taskService.createTaskQuery().processInstanceId(processInstanceId).singleResult();
    if (null != nextTask) {
        taskService.setAssignee(nextTask.getId(), lastTask.getAssignee());
    }

    //  恢复原始方向
    currentFlowNode.setOutgoingFlows(originalSequenceFlowList);
}

以请假为例

<process id="holiday" name="holiday" isExecutable="true">
    <startEvent id="startevent1" name="Start"></startEvent>
    <userTask id="usertask1" name="填写请假单" activiti:assignee="${assignee1}"></userTask>
    <sequenceFlow id="flow1" sourceRef="startevent1" targetRef="usertask1"></sequenceFlow>
    <userTask id="usertask2" name="部门经理审批" activiti:assignee="${assignee2}"></userTask>
    <sequenceFlow id="flow2" sourceRef="usertask1" targetRef="usertask2"></sequenceFlow>
    <userTask id="usertask3" name="人事审批" activiti:candidateUsers="tom,jerry"></userTask>
    <sequenceFlow id="flow3" sourceRef="usertask2" targetRef="usertask3"></sequenceFlow>
    <endEvent id="endevent1" name="End"></endEvent>
    <sequenceFlow id="flow4" sourceRef="usertask3" targetRef="endevent1"></sequenceFlow>
</process>

假设现在已经到“人事审批”这个节点了，当前活动是usertask3

接下来，我们运行上面的代码，回退到上一个节点“部门经理审批”，于是

流程重新从“部门经理审批”节点开始往下走，当流程走完以后

证明，思路正确，写法没啥问题。但是，上面的代码可以简化一下，如下：

/**
 * 跳到最开始的任务节点（直接打回）
 * @param task 当前任务
 */
public void jumpToStart(Task task) {
    ProcessEngine processEngine = ProcessEngines.getDefaultProcessEngine();
    HistoryService historyService = processEngine.getHistoryService();
    RepositoryService repositoryService = processEngine.getRepositoryService();
    TaskService taskService = processEngine.getTaskService();

    String processInstanceId = task.getProcessInstanceId();

    //  获取所有历史任务（按创建时间升序）
    List<HistoricTaskInstance> hisTaskList = historyService.createHistoricTaskInstanceQuery()
            .processInstanceId(processInstanceId)
            .orderByTaskCreateTime()
            .asc()
            .list();

    if (CollectionUtils.isEmpty(hisTaskList) || hisTaskList.size() < 2) {
        return;
    }

    //  第一个任务
    HistoricTaskInstance startTask = hisTaskList.get(0);
    //  当前任务
    HistoricTaskInstance currentTask = hisTaskList.get(hisTaskList.size() - 1);

    BpmnModel bpmnModel = repositoryService.getBpmnModel(task.getProcessDefinitionId());

    //  获取第一个活动节点
    FlowNode startFlowNode = (FlowNode) bpmnModel.getMainProcess().getFlowElement(startTask.getTaskDefinitionKey());
    //  获取当前活动节点
    FlowNode currentFlowNode = (FlowNode) bpmnModel.getMainProcess().getFlowElement(currentTask.getTaskDefinitionKey());

    //  临时保存当前活动的原始方向
    List<SequenceFlow> originalSequenceFlowList = new ArrayList<>();
    originalSequenceFlowList.addAll(currentFlowNode.getOutgoingFlows());
    //  清理活动方向
    currentFlowNode.getOutgoingFlows().clear();

    //  建立新方向
    SequenceFlow newSequenceFlow = new SequenceFlow();
    newSequenceFlow.setId("newSequenceFlowId");
    newSequenceFlow.setSourceFlowElement(currentFlowNode);
    newSequenceFlow.setTargetFlowElement(startFlowNode);
    List<SequenceFlow> newSequenceFlowList = new ArrayList<>();
    newSequenceFlowList.add(newSequenceFlow);
    //  当前节点指向新的方向
    currentFlowNode.setOutgoingFlows(newSequenceFlowList);

    //  完成当前任务
    taskService.complete(task.getId());

    //  重新查询当前任务
    Task nextTask = taskService.createTaskQuery().processInstanceId(processInstanceId).singleResult();
    if (null != nextTask) {
        taskService.setAssignee(nextTask.getId(), startTask.getAssignee());
    }

    //  恢复原始方向
    currentFlowNode.setOutgoingFlows(originalSequenceFlowList);
}

2.  会签

多个人同时处理一个任务，这种任务我们称之为会签任务 。Activiti实现会签是基于多实例任务，将节点设置成多实例，主要通过在UserTask节点的属性上配置。

会签的种类：

• 按数量通过： 达到一定数量的通过表决后，会签通过。
• 按比例通过： 达到一定比例的通过表决后，会签通过。
• 一票否决： 只要有一个表决时否定的，会签通过。
• 一票通过： 只要有一个表决通过的，会签通过。

每个实例有以下变量：

• nrOfInstances: 实例总数
• nrOfActiveInstances: 当前激活的（未完成的）实例总数。 如果串行执行，则改值永远是1
• nrOfCompletedInstances: 已完成的实例总数

条件${nrOfInstances == nrOfCompletedInstances}表示所有人员审批完成后会签结束。

条件${ nrOfCompletedInstances == 1}表示一个人完成审批，该会签就结束。

其他条件依次类推，同时这里也可以写自己添加的流程变量。

相关文档如下：

下面举个例子：

<process id="countersign" name="countersign" isExecutable="true">
    <startEvent id="startevent1" name="Start"></startEvent>
    <userTask id="usertask1" name="申请" activiti:assignee="zhangsan"></userTask>
    <sequenceFlow id="flow1" sourceRef="startevent1" targetRef="usertask1"></sequenceFlow>
    <userTask id="usertask2" name="会签审批" activiti:assignee="${approver}">
        <multiInstanceLoopCharacteristics isSequential="false" 
            activiti:collection="${approverList}" activiti:elementVariable="approver">
            <completionCondition>${nrOfCompletedInstances == nrOfInstances}</completionCondition>
        </multiInstanceLoopCharacteristics>
    </userTask>
    <sequenceFlow id="flow2" sourceRef="usertask1" targetRef="usertask2"></sequenceFlow>
    <userTask id="usertask3" name="备案" activiti:assignee="tianqi"></userTask>
    <sequenceFlow id="flow3" sourceRef="usertask2" targetRef="usertask3"></sequenceFlow>
    <endEvent id="endevent1" name="End"></endEvent>
    <sequenceFlow id="flow4" sourceRef="usertask3" targetRef="endevent1"></sequenceFlow>
</process>

编写代码：

//  部署流程定义
RepositoryService repositoryService = processEngine.getRepositoryService();
Deployment deployment = repositoryService.createDeployment()
                .addClasspathResource("diagram/countersign.bpmn")
                .name("会签示例")
                .key("countersign")
                .deploy();

//  启动流程实例
RuntimeService runtimeService = processEngine.getRuntimeService();
Map<String, Object> variables = new HashMap<>();
variables.put("approverList", Arrays.asList("lisi","wangwu","zhaoliu"));
ProcessInstance processInstance = runtimeService.startProcessInstanceByKey("countersign", variables);

//  完成任务
TaskService taskService = processEngine.getTaskService();
Task task = taskService.createTaskQuery().processInstanceId("107501").taskAssignee("zhaoliu").singleResult();
if (null != task) {
    taskService.complete(task.getId());
}

流程启动后，首先是zhangsan审批

当zhangsan完成自己的任务后，进入会签环节，于是我们看到当前有3个激活的任务

当lisi完成任务以后，当前任务剩下2个

当wangwu和zhaoliu都完成任务了以后，会签任务完成，进入下一个环节

刚才的例子中没有考虑到审批不通过的情况，接下来我们完善一下，考虑下面的流程

<process id="countersign" name="countersign" isExecutable="true">
    <startEvent id="startevent1" name="Start"></startEvent>
    <userTask id="usertask1" name="申请" activiti:assignee="zhangsan"></userTask>
    <sequenceFlow id="flow1" sourceRef="startevent1" targetRef="usertask1"></sequenceFlow>
    <userTask id="usertask2" name="会签审批" activiti:assignee="${approver}">
        <multiInstanceLoopCharacteristics isSequential="false" activiti:collection="${approverList}" activiti:elementVariable="approver">
            <completionCondition>${nrOfCompletedInstances / nrOfInstances == 1 || pass == false}</completionCondition>
        </multiInstanceLoopCharacteristics>
    </userTask>
    <sequenceFlow id="flow2" sourceRef="usertask1" targetRef="usertask2"></sequenceFlow>
    <userTask id="usertask3" name="备案" activiti:assignee="tianqi"></userTask>
    <exclusiveGateway id="exclusivegateway1" name="Exclusive Gateway"></exclusiveGateway>
    <sequenceFlow id="flow5" sourceRef="usertask2" targetRef="exclusivegateway1"></sequenceFlow>
    <sequenceFlow id="flow6" name="通过" sourceRef="exclusivegateway1" targetRef="usertask3">
        <conditionExpression xsi:type="tFormalExpression"><![CDATA[${pass == true}]]></conditionExpression>
    </sequenceFlow>
    <sequenceFlow id="flow7" name="拒绝" sourceRef="exclusivegateway1" targetRef="usertask1">
        <conditionExpression xsi:type="tFormalExpression"><![CDATA[${pass == false}]]></conditionExpression>
    </sequenceFlow>
    <endEvent id="endevent1" name="End"></endEvent>
    <sequenceFlow id="flow8" sourceRef="usertask3" targetRef="endevent1"></sequenceFlow>
</process>

在会签审批完成任务时就要加上流程变量pass了

RuntimeService runtimeService = processEngine.getRuntimeService();
TaskService taskService = processEngine.getTaskService();

Task task = taskService.createTaskQuery().processInstanceId("152501").taskAssignee("lisi").singleResult();
if (null != task) {
    Map<String, Object> variables = new HashMap<>();
    variables.put("pass", true);
//            variables.put("pass", false);
    taskService.complete(task.getId(), variables);

    runtimeService.getVariable(task.getExecutionId(), "nrOfCompletedInstances");
}

zhaoliu审批的时候pass传的false，于是流程又走到zhangsan那里，流程重新又走了一遍才全部完成

回退和会签就先讲到这里