feat: add continue-on-error job-level support

WORKFLOW_SUPPORT.md

@@ -27,7 +27,7 @@ syntax](https://docs.github.com/en/actions/writing-workflows/workflow-syntax-for
   - [ ] jobs.<job_id>.timeout-minutes
   - [ ] jobs.<job_id>.strategy
   - [ ] jobs.<job_id>.container
-  - [ ] jobs.<job_id>.continue-on-error
+  - [x] jobs.<job_id>.continue-on-error
   - [ ] jobs.<job_id>.services
   - [ ] jobs.<job_id>.uses
   - [ ] jobs.<job_id>.with

internal/commands/execute_workflow.go

@@ -38,10 +38,6 @@ func ExecuteWorkflow(configuration Configuration, workflowFile string) error {
 	}
 	taskResult := runnerInstance.RunWorkflow(*workflow)
 
-	if !taskResult.HasError() {
-		return nil
-	}
-
 	for _, job := range taskResult.Children {
 		if job.Status == "success" {
 			logger.Info(logger.Green("Job %s: %s"), job.TaskId, job.Status)
@@ -50,5 +46,9 @@ func ExecuteWorkflow(configuration Configuration, workflowFile string) error {
 		}
 	}
 
+	if taskResult.Failed() {
 		return fmt.Errorf("Task %s failed with at least 1 error.", taskResult.TaskId)
 	}
+
+	return nil
+}

internal/runner/mock_container_driver.go

@@ -1,6 +1,7 @@
 package runner
 
 import (
+	"fmt"
 	"strings"
 )
 
@@ -52,10 +53,19 @@ func (d *MockDriver) Exec(containerName string, command string, cwd string) Comm
 	if _, init := d.calls["Exec"]; !init {
 		d.calls["Exec"] = []MockCall{}
 	}
-	d.calls["Exec"] = append(d.calls["Exec"], MockCall{fname: "Exec", args: []string{containerName, command, cwd}})
 
-	if _, mocked := d.mockedCalls["Exec"][strings.Join([]string{containerName, command, cwd}, " ")]; mocked {
-		return d.mockedCalls["Exec"][strings.Join([]string{containerName, command, cwd}, " ")]
+	args := []string{containerName, command, cwd}
+	d.calls["Exec"] = append(d.calls["Exec"], MockCall{fname: "Exec", args: args})
+
+	mockKeys := []string{
+		fmt.Sprintf("nthcall::%d", len(d.calls["Exec"])),
+		fmt.Sprintf("withargs::%s", strings.Join(args, " ")),
+	}
+
+	for _, mockKey := range mockKeys {
+		if _, mocked := d.mockedCalls["Exec"][mockKey]; mocked {
+			return d.mockedCalls["Exec"][mockKey]
+		}
 	}
 
 	return CommandResult{}
@@ -65,7 +75,20 @@ func (d *MockDriver) Exec(containerName string, command string, cwd string) Comm
 //
 // The mocked call is reused as long as it's defined within the mock driver.
 func (d *MockDriver) WithMockedCall(fn string, returnValue CommandResult, args ...string) {
-	mockKey := strings.Join(args, " ")
+	mockKey := fmt.Sprintf("withargs::%s", strings.Join(args, " "))
+
+	if _, initialized := d.mockedCalls[fn]; !initialized {
+		d.mockedCalls[fn] = map[string]CommandResult{}
+	}
+
+	d.mockedCalls[fn][mockKey] = returnValue
+}
+
+// Mocks the nth call to <fn> to return <returnValue>.
+//
+// The mocked call is reused as long as it's defined within the mock driver.
+func (d *MockDriver) WithNthMockedCall(fn string, callIndex int, returnValue CommandResult) {
+	mockKey := fmt.Sprintf("nthcall::%d", callIndex)
 
 	if _, initialized := d.mockedCalls[fn]; !initialized {
 		d.mockedCalls[fn] = map[string]CommandResult{}

internal/runner/runner.go

@@ -64,14 +64,22 @@ func (r *Runner) RunWorkflow(workflow workflow.Workflow) TaskTracker {
 				logger.Info("Using image %s (label: %s)", runnerImage, job.RunsOn)
 
 				if pullError := runner.Driver.Pull(runnerImage); pullError != nil {
-					jobTracker.SetStatus("failed").SetError(pullError)
+					jobTracker.SetError(pullError)
+
+					if !job.ContinueOnError {
+						jobTracker.SetStatus("failed")
 						return
 					}
 
+				}
+
 				if runError := runner.RunJobInContainer(runnerImage, containerName, jobContext); runError != nil {
-					jobTracker.SetStatus("failed").SetError(runError)
+					jobTracker.SetError(runError)
+					if !job.ContinueOnError {
+						jobTracker.SetStatus("failed")
 						return
 					}
+				}
 
 				jobTracker.SetStatus("success")
 				runner.deferred.RunDeferredTasksInScope(fmt.Sprintf("job-%s", containerName))

internal/runner/runner_job_test.go

@@ -7,6 +7,83 @@ import (
 	"testing"
 )
 
+func TestJobErrorsWorkflowErrorsIfJobContinueOnErrorUndefined(t *testing.T) {
+	workflowSample := `
+jobs:
+  jobA:
+    runs-on: test
+    steps:
+      - run: exit 1
+    `
+
+	workflow, _ := workflow.FromYamlBytes([]byte(workflowSample))
+	mockDriver := NewMockDriver()
+
+	runner := NewRunner(&mockDriver, map[string]string{"test": "test"})
+
+	jobContext := context.WithValue(context.Background(), "workflow", *workflow)
+	jobContext = context.WithValue(jobContext, "currentJob", workflow.Jobs["jobA"])
+
+	mockDriver.WithNthMockedCall("Exec", 1, CommandResult{Error: errors.New("exit 1!"), ExitCode: 1})
+	task := runner.RunWorkflow(*workflow)
+
+	if !task.Failed() {
+		t.Error("Expected task to fail, but it succeeded.")
+	}
+}
+func TestJobErrorsWorkflowErrorsIfJobContinueOnErrorFalsy(t *testing.T) {
+	workflowSample := `
+jobs:
+  jobA:
+    runs-on: test
+    continue-on-error: false
+    steps:
+      - run: exit 1
+    `
+
+	workflow, _ := workflow.FromYamlBytes([]byte(workflowSample))
+	mockDriver := NewMockDriver()
+
+	runner := NewRunner(&mockDriver, map[string]string{"test": "test"})
+
+	jobContext := context.WithValue(context.Background(), "workflow", *workflow)
+	jobContext = context.WithValue(jobContext, "currentJob", workflow.Jobs["jobA"])
+
+	mockDriver.WithNthMockedCall("Exec", 1, CommandResult{Error: errors.New("exit 1!"), ExitCode: 1})
+	task := runner.RunWorkflow(*workflow)
+
+	if !task.Failed() {
+		t.Error("Expected task to fail, but it succeeded.")
+	}
+}
+
+func TestJobSkipsErrorWorkflowSucceedsIfContinueOnErrorTruthy(t *testing.T) {
+	workflowSample := `
+jobs:
+  jobA:
+    runs-on: test
+    continue-on-error: true
+    steps:
+      - run: exit 1
+    `
+
+	workflow, _ := workflow.FromYamlBytes([]byte(workflowSample))
+	mockDriver := NewMockDriver()
+
+	mockDriver.WithMockedCall("Exec", CommandResult{Error: errors.New("exit 1!"), ExitCode: 1}, "testContainer", "exit 1")
+	runner := NewRunner(&mockDriver, map[string]string{"test": "test"})
+
+	jobContext := context.WithValue(context.Background(), "workflow", *workflow)
+	jobContext = context.WithValue(jobContext, "currentJob", workflow.Jobs["jobA"])
+
+	task := runner.RunWorkflow(*workflow)
+
+	if task.Failed() {
+		t.Error("Expected task to succeed, but it failed.")
+	}
+
+}
+
 func TestJobStepSkipsErrorIfContinueOnErrorTruthy(t *testing.T) {
 	workflowSample := `
 jobs:

internal/runner/task.go

@@ -52,3 +52,17 @@ func (t TaskTracker) HasError() bool {
 
 	return false
 }
+
+func (t TaskTracker) Failed() bool {
+	if t.Status == "failed" {
+		return true
+	}
+
+	for _, child := range t.Children {
+		if child.Failed() {
+			return true
+		}
+	}
+
+	return false
+}

internal/runner/task_test.go

@@ -27,3 +27,30 @@ func TestTaskHasErrorReturnsTrueIfAnyJobHasErrors(t *testing.T) {
 		t.Errorf("Expected true, got false.")
 	}
 }
+
+func TestTaskFailedReturnsTrueIfAnyChildHasFailed(t *testing.T) {
+	task := NewTaskTracker(nil)
+	NewTaskTracker(task).SetStatus("failed")
+
+	if !task.Failed() {
+		t.Error("Expected failed = true, got false.")
+	}
+}
+
+func TestTaskFailedReturnsTrueIfTaskFailed(t *testing.T) {
+	task := NewTaskTracker(nil).SetStatus("failed")
+	NewTaskTracker(task)
+
+	if !task.Failed() {
+		t.Error("Expected failed = true, got false.")
+	}
+}
+
+func TeskTaskFailedReturnsFalseIfNeitherSelfOrChildrenFailed(t *testing.T) {
+	task := NewTaskTracker(nil).SetStatus("success")
+	NewTaskTracker(task)
+
+	if task.Failed() {
+		t.Error("Expected failed = false, got true.")
+	}
+}

internal/workflow/models.go

@@ -10,6 +10,8 @@ type Job struct {
 	Needs  []string `yaml:"needs"`
 	// Job name; this isn't guaranteed to be unique.
 	Name string `yaml:"name"`
+	// If truthy, job-level failure does not bubble up to the workflow.
+	ContinueOnError bool `yaml:"continue-on-error"`
 	Defaults        struct {
 		Run struct {
 			WorkingDirectory string `yaml:"working-directory"`