Rewrite mta IT using docker cli (#1819)

This change addresses some issues of the testcontainer based testing approach (much repeated code, API not on the right abstraction level). It introduces new methods that make use of the docker cli, and rewrites the mta tests using this method.


Co-authored-by: Stephan Aßmus <stephan.assmus@sap.com>

integration/docker_integration_test_executor.go

@@ -0,0 +1,129 @@
+// +build integration
+
+package main
+
+import (
+	"bytes"
+	"github.com/SAP/jenkins-library/pkg/command"
+	"github.com/SAP/jenkins-library/pkg/log"
+	"io/ioutil"
+	"math/rand"
+	"os"
+	"path"
+	"strconv"
+	"strings"
+	"testing"
+	"time"
+)
+
+// IntegrationTestDockerExecRunner keeps the state of an instance of a docker runner
+type IntegrationTestDockerExecRunner struct {
+	// Runner is the ExecRunner to which all executions are forwarded in the end.
+	Runner        command.Command
+	Image         string
+	User          string
+	TestDir       []string
+	Mounts        map[string]string
+	Environment   map[string]string
+	Setup         []string
+	ContainerName string
+}
+
+// IntegrationTestDockerExecRunnerBundle is used to construct an instance of IntegrationTestDockerExecRunner
+type IntegrationTestDockerExecRunnerBundle struct {
+	Image       string
+	User        string
+	TestDir     []string
+	Mounts      map[string]string
+	Environment map[string]string
+	Setup       []string
+}
+
+func givenThisContainer(t *testing.T, bundle IntegrationTestDockerExecRunnerBundle) IntegrationTestDockerExecRunner {
+	runner := command.Command{}
+
+	// Generate a random container name so we can start a new one for each test method
+	// We don't rely on docker's random name generator for two reasons
+	// First, it is easier to save the name here compared to getting it from stdout
+	// Second, the common prefix allows batch stopping/deleting of containers if so desired
+	// The test code will not automatically delete containers as they might be useful for debugging
+	var seededRand = rand.New(rand.NewSource(time.Now().UnixNano()))
+	containerName := "piper-integration-test-" + strconv.Itoa(seededRand.Int())
+
+	testRunner := IntegrationTestDockerExecRunner{
+		Runner:        runner,
+		Image:         bundle.Image,
+		User:          bundle.User,
+		Mounts:        bundle.Mounts,
+		Setup:         bundle.Setup,
+		ContainerName: containerName,
+	}
+
+	//todo ensure it is a linux binary
+	wd, _ := os.Getwd()
+	localPiper := path.Join(wd, "..", "piper")
+	if localPiper == "" {
+		t.Fatal("Could not locate piper binary to test")
+	}
+
+	projectDir := path.Join(wd, path.Join(bundle.TestDir...))
+
+	// 1. Copy test files to a temp dir in order to avoid non-repeatable test executions because of changed state
+	// 2. Don't remove the temp dir to allow investigation of failed tests. Maybe add an option for cleaning it later?
+	tempDir, err := ioutil.TempDir("", "piper-integration-test")
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	err = copyDir(projectDir, tempDir)
+	if err != nil {
+		t.Fatalf("")
+	}
+
+	//todo mounts
+	//todo env (secrets)
+	err = testRunner.Runner.RunExecutable("docker", "run", "-d", "-u="+testRunner.User,
+		"-v", localPiper+":/piper", "-v", tempDir+":/project",
+		"--name="+testRunner.ContainerName,
+		testRunner.Image,
+		"sleep", "2000")
+	if err != nil {
+		t.Fatalf("Starting test container has failed %s", err)
+	}
+	for _, scriptLine := range testRunner.Setup {
+		err := testRunner.Runner.RunExecutable("docker", "exec", testRunner.ContainerName, "/bin/bash", "-c", scriptLine)
+		if err != nil {
+			t.Fatalf("Running setup script in test container has failed %s", err)
+		}
+	}
+
+	err = testRunner.Runner.RunExecutable("docker", "cp", "piper-command-wrapper.sh", testRunner.ContainerName+":/piper-wrapper")
+	if err != nil {
+		t.Fatalf("Copying command wrapper to container has failed %s", err)
+	}
+	err = testRunner.Runner.RunExecutable("docker", "exec", testRunner.ContainerName, "chmod", "+x", "/piper-wrapper")
+	if err != nil {
+		t.Fatalf("Making command wrapper in container executable has failed %s", err)
+	}
+	return testRunner
+}
+
+func (d *IntegrationTestDockerExecRunner) whenRunningPiperCommand(command string, parameters ...string) error {
+	args := []string{"exec", "--workdir", "/project", d.ContainerName, "/bin/bash", "/piper-wrapper", "/piper", command}
+	args = append(args, parameters...)
+	return d.Runner.RunExecutable("docker", args...)
+}
+
+func (d *IntegrationTestDockerExecRunner) assertHasOutput(t *testing.T, want string) {
+	buffer := new(bytes.Buffer)
+	d.Runner.Stdout(buffer)
+	err := d.Runner.RunExecutable("docker", "exec", d.ContainerName, "cat", "/tmp/test-log.txt")
+	d.Runner.Stdout(log.Writer())
+	if err != nil {
+		t.Fatalf("Failed to get log output of container %s", d.ContainerName)
+	}
+
+	if !strings.Contains(buffer.String(), want) {
+		t.Fatalf("Assertion has failed. Expected output %s in command output.\n%s", want, buffer.String())
+	}
+}