能否通过插件或原生方式从pytest动态生成Jenkins Pipeline阶段？

Great question! While there isn't a fully "native" one-click solution from either pytest or Jenkins to automatically map each test case to a Pipeline stage, there are streamlined ways to achieve this that are more robust than your initial manual approach. Let's walk through them:

1. Optimize Your Existing Approach with Cleaner Pytest Test Collection #

Your current method works, but we can make the test collection more reliable by using pytest's built-in quiet mode for test collection, which outputs a clean, consistent list of test identifiers:

stage("Run Tests in Parallel Stages") {
    steps {
        script {
            // Use -q to get a simplified, line-separated list of test nodes
            def rawTestList = sh(
                script: "pytest tests/ --collect-only -q",
                returnStdout: true
            ).trim().split('\n')
            
            def testStages = [:]
            
            rawTestList.each { testNode ->
                // Split the test node into file path and test name (e.g., tests/test_auth.py::test_login -> [file, name])
                def [testFile, testName] = testNode.split('::')
                
                testStages[testName] = {
                    stage(testName) {
                        sh "pytest ${testFile} -k ${testName}"
                    }
                }
            }
            
            // Run all test stages in parallel for Blue Ocean visibility
            parallel testStages
        }
    }
}

This avoids messy string parsing of unstructured output, since --collect-only -q outputs each test in the standard file.py::test_name format. In Blue Ocean, each parallel stage will show up as a separate, clickable node so you can see exactly which tests passed/failed at a glance.

2. Use Pytest Plugins for Structured Test Data #

For even more reliability, use a pytest plugin to export test metadata in a structured format (like JSON), which is easier to parse in Jenkins:

Step 1: Install the pytest-json-report plugin #

pip install pytest-json-report

Step 2: Parse the JSON Report in Jenkins Pipeline #

stage("Generate & Run Test Stages") {
    steps {
        script {
            // Collect tests and export to JSON
            sh "pytest tests/ --collect-only --json-report --json-report-file=test_catalog.json"
            
            // Read and parse the JSON file
            def testCatalog = readJSON file: 'test_catalog.json'
            
            def testStages = [:]
            
            testCatalog.tests.each { test ->
                def testNode = test.nodeid // Full test identifier (file::name)
                def testName = test.name
                def testFile = testNode.split('::')[0]
                
                testStages[testName] = {
                    stage(testName) {
                        sh "pytest ${testFile} -k ${testName}"
                    }
                }
            }
            
            parallel testStages
        }
    }
}

The JSON report includes additional metadata (like test markers, module paths) if you need to filter or group tests later, making this approach more scalable.

3. Key Notes for Blue Ocean Visibility #

• When using parallel, Blue Ocean will render each test stage as an individual item in the pipeline view, exactly what you're looking for.
• If you need to run tests sequentially instead of in parallel, simply replace parallel testStages with testStages.each { key, value -> value() }—each test will still show up as a separate stage in Blue Ocean.

Are There "Native" Plugins for This? #

As of now, there's no Jenkins plugin or pytest feature that directly generates Pipeline stages per test case out of the box. The above methods are the most practical and maintainable ways to achieve your goal, leveraging pytest's test collection capabilities and Jenkins' scripted Pipeline flexibility.

内容的提问来源于stack exchange，提问作者Aryeh K