Generate a run summary file on turbo run (#4069)

cli/go.mod

@@ -35,6 +35,7 @@ require (
 	github.com/pyr-sh/dag v1.0.0
 	github.com/sabhiram/go-gitignore v0.0.0-20201211210132-54b8a0bf510f
 	github.com/schollz/progressbar/v3 v3.9.0
+	github.com/segmentio/ksuid v1.0.4
 	github.com/spf13/cobra v1.3.0
 	github.com/spf13/viper v1.12.0
 	github.com/stretchr/testify v1.8.0

cli/go.sum

@@ -420,6 +420,8 @@ github.com/sagikazarmark/crypt v0.3.0/go.mod h1:uD/D+6UF4SrIR1uGEv7bBNkNqLGqUr43
 github.com/schollz/progressbar/v3 v3.9.0 h1:k9SRNQ8KZyibz1UZOaKxnkUE3iGtmGSDt1YY9KlCYQk=
 github.com/schollz/progressbar/v3 v3.9.0/go.mod h1:W5IEwbJecncFGBvuEh4A7HT1nZZ6WNIL2i3qbnI0WKY=
 github.com/sean-/seed v0.0.0-20170313163322-e2103e2c3529/go.mod h1:DxrIzT+xaE7yg65j358z/aeFdxmN0P9QXhEzd20vsDc=
+github.com/segmentio/ksuid v1.0.4 h1:sBo2BdShXjmcugAMwjugoGUdUV0pcxY5mW4xKRn3v4c=
+github.com/segmentio/ksuid v1.0.4/go.mod h1:/XUiZBD3kVx5SmUOl55voK5yeAbBNNIed+2O73XgrPE=
 github.com/shopspring/decimal v1.2.0 h1:abSATXmQEYyShuxI4/vyW3tV1MrKAJzCZ/0zLUXYbsQ=
 github.com/shopspring/decimal v1.2.0/go.mod h1:DKyhrW/HYNuLGql+MJL6WCR6knT2jwCFRcu2hWCYk4o=
 github.com/sirupsen/logrus v1.2.0/go.mod h1:LxeOpSwHxABJmUn/MG1IvRgCAasNZTLOkJPxbbu5VWo=

cli/integration_tests/basic_monorepo/run_summary.t

@@ -0,0 +1,17 @@
+Setup
+  $ . ${TESTDIR}/../setup.sh
+  $ . ${TESTDIR}/setup.sh $(pwd)
+
+  $ rm -rf .turbo/runs
+  $ TURBO_RUN_SUMMARY=true ${TURBO} run build > /dev/null
+# no output, just check for 0 status code
+  $ test -d .turbo/runs
+  $ ls .turbo/runs/*.json | wc -l
+  \s*1 (re)
+
+# Without env var, no summary file is generated
+  $ rm -rf .turbo/runs
+  $ ${TURBO} run build > /dev/null
+# validate with exit code so the test works on macOS and linux
+  $ test -d .turbo/runs
+  [1]

cli/internal/graph/graph.go

@@ -10,6 +10,7 @@ import (
 	"github.com/pyr-sh/dag"
 	"github.com/vercel/turbo/cli/internal/fs"
 	"github.com/vercel/turbo/cli/internal/nodes"
+	"github.com/vercel/turbo/cli/internal/runsummary"
 	"github.com/vercel/turbo/cli/internal/taskhash"
 	"github.com/vercel/turbo/cli/internal/turbopath"
 	"github.com/vercel/turbo/cli/internal/util"
@@ -48,7 +49,7 @@ func (g *CompleteGraph) GetPackageTaskVisitor(
 	taskGraph *dag.AcyclicGraph,
 	getArgs func(taskID string) []string,
 	logger hclog.Logger,
-	visitor func(ctx gocontext.Context, packageTask *nodes.PackageTask) error,
+	visitor func(ctx gocontext.Context, packageTask *nodes.PackageTask, taskSummary *runsummary.TaskSummary) error,
 ) func(taskID string) error {
 	return func(taskID string) error {
 		packageName, taskName := util.GetPackageTaskFromId(taskID)
@@ -62,6 +63,7 @@ func (g *CompleteGraph) GetPackageTaskVisitor(
 			return fmt.Errorf("Could not find definition for task")
 		}
 
+		// TODO: maybe we can remove this PackageTask struct at some point
 		packageTask := &nodes.PackageTask{
 			TaskID:          taskID,
 			Task:            taskName,
@@ -80,22 +82,47 @@ func (g *CompleteGraph) GetPackageTaskVisitor(
 			getArgs(taskName),
 		)
 
+		// Not being able to construct the task hash is a hard error
 		if err != nil {
 			return fmt.Errorf("Hashing error: %v", err)
 		}
 
+		pkgDir := pkg.Dir
 		packageTask.Hash = hash
-		packageTask.HashedEnvVars = g.TaskHashTracker.GetEnvVars(packageTask.TaskID)
-		packageTask.ExpandedInputs = g.TaskHashTracker.GetExpandedInputs(packageTask)
-		packageTask.Framework = g.TaskHashTracker.GetFramework(packageTask.TaskID)
+		envVars := g.TaskHashTracker.GetEnvVars(taskID)
+		expandedInputs := g.TaskHashTracker.GetExpandedInputs(packageTask)
+		framework := g.TaskHashTracker.GetFramework(taskID)
 
+		// Assign remaining fields to packageTask
+		var command string
 		if cmd, ok := pkg.Scripts[taskName]; ok {
-			packageTask.Command = cmd
+			command = cmd
 		}
 
-		packageTask.LogFile = repoRelativeLogFile(packageTask)
+		logFile := repoRelativeLogFile(pkgDir, taskName)
+		packageTask.LogFile = logFile
+		packageTask.Command = command
+
+		summary := &runsummary.TaskSummary{
+			TaskID:                 taskID,
+			Task:                   taskName,
+			Hash:                   hash,
+			Package:                packageName,
+			Dir:                    pkgDir.ToString(),
+			Outputs:                taskDefinition.Outputs.Inclusions,
+			ExcludedOutputs:        taskDefinition.Outputs.Exclusions,
+			LogFile:                logFile,
+			ResolvedTaskDefinition: taskDefinition,
+			ExpandedInputs:         expandedInputs,
+			Command:                command,
+			Framework:              framework,
+			EnvVars: runsummary.TaskEnvVarSummary{
+				Configured: envVars.BySource.Explicit.ToSecretHashable(),
+				Inferred:   envVars.BySource.Prefixed.ToSecretHashable(),
+			},
+		}
 
-		return visitor(ctx, packageTask)
+		return visitor(ctx, packageTask, summary)
 	}
 }
 
@@ -152,6 +179,6 @@ func (g *CompleteGraph) GetPackageJSONFromWorkspace(workspaceName string) (*fs.P
 
 // repoRelativeLogFile returns the path to the log file for this task execution as a
 // relative path from the root of the monorepo.
-func repoRelativeLogFile(pt *nodes.PackageTask) string {
-	return filepath.Join(pt.Pkg.Dir.ToStringDuringMigration(), ".turbo", fmt.Sprintf("turbo-%v.log", pt.Task))
+func repoRelativeLogFile(dir turbopath.AnchoredSystemPath, taskName string) string {
+	return filepath.Join(dir.ToStringDuringMigration(), ".turbo", fmt.Sprintf("turbo-%v.log", taskName))
 }

cli/internal/nodes/packagetask.go

@@ -4,9 +4,7 @@ package nodes
 import (
 	"fmt"
 
-	"github.com/vercel/turbo/cli/internal/env"
 	"github.com/vercel/turbo/cli/internal/fs"
-	"github.com/vercel/turbo/cli/internal/turbopath"
 )
 
 // PackageTask represents running a particular task in a particular package
@@ -21,10 +19,7 @@ type PackageTask struct {
 	Outputs         []string
 	ExcludedOutputs []string
 	LogFile         string
-	ExpandedInputs  map[turbopath.AnchoredUnixPath]string
 	Hash            string
-	HashedEnvVars   env.DetailedMap
-	Framework       string
 }
 
 // OutputPrefix returns the prefix to be used for logging and ui for this task

cli/internal/run/dry_run.go

@@ -19,11 +19,6 @@ import (
 	"github.com/vercel/turbo/cli/internal/util"
 )
 
-// missingTaskLabel is printed when a package is missing a definition for a task that is supposed to run
-// E.g. if `turbo run build --dry` is run, and package-a doesn't define a `build` script in package.json,
-// the RunSummary will print this, instead of the script (e.g. `next build`).
-const missingTaskLabel = "<NONEXISTENT>"
-
 // DryRun gets all the info needed from tasks and prints out a summary, but doesn't actually
 // execute the task.
 func DryRun(
@@ -64,30 +59,20 @@ func DryRun(
 		if err != nil {
 			return err
 		}
-		base.UI.Output(rendered)
+		base.UI.Output(string(rendered))
 		return nil
 	}
 
 	return summary.FormatAndPrintText(base.UI, g.WorkspaceInfos, singlePackage)
 }
 
-func executeDryRun(ctx gocontext.Context, engine *core.Engine, g *graph.CompleteGraph, taskHashTracker *taskhash.Tracker, rs *runSpec, base *cmdutil.CmdBase, turboCache cache.Cache) ([]runsummary.TaskSummary, error) {
-	taskIDs := []runsummary.TaskSummary{}
-
-	dryRunExecFunc := func(ctx gocontext.Context, packageTask *nodes.PackageTask) error {
-		command := missingTaskLabel
-		if packageTask.Command != "" {
-			command = packageTask.Command
-		}
-
-		framework := runsummary.MissingFrameworkLabel
-		if packageTask.Framework != "" {
-			framework = packageTask.Framework
-		}
+func executeDryRun(ctx gocontext.Context, engine *core.Engine, g *graph.CompleteGraph, taskHashTracker *taskhash.Tracker, rs *runSpec, base *cmdutil.CmdBase, turboCache cache.Cache) ([]*runsummary.TaskSummary, error) {
+	taskIDs := []*runsummary.TaskSummary{}
 
+	dryRunExecFunc := func(ctx gocontext.Context, packageTask *nodes.PackageTask, taskSummary *runsummary.TaskSummary) error {
 		isRootTask := packageTask.PackageName == util.RootPkgName
-		if isRootTask && commandLooksLikeTurbo(command) {
-			return fmt.Errorf("root task %v (%v) looks like it invokes turbo and might cause a loop", packageTask.Task, command)
+		if isRootTask && commandLooksLikeTurbo(taskSummary.Command) {
+			return fmt.Errorf("root task %v (%v) looks like it invokes turbo and might cause a loop", packageTask.Task, taskSummary.Command)
 		}
 
 		ancestors, err := engine.GetTaskGraphAncestors(packageTask.TaskID)
@@ -100,34 +85,25 @@ func executeDryRun(ctx gocontext.Context, engine *core.Engine, g *graph.Complete
 			return err
 		}
 
-		hash := packageTask.Hash
-		itemStatus, err := turboCache.Exists(hash)
+		itemStatus, err := turboCache.Exists(packageTask.Hash)
 		if err != nil {
 			return err
 		}
 
-		taskIDs = append(taskIDs, runsummary.TaskSummary{
-			TaskID:                 packageTask.TaskID,
-			Task:                   packageTask.Task,
-			Package:                packageTask.PackageName,
-			Dir:                    packageTask.Dir,
-			Outputs:                packageTask.Outputs,
-			ExcludedOutputs:        packageTask.ExcludedOutputs,
-			LogFile:                packageTask.LogFile,
-			ResolvedTaskDefinition: packageTask.TaskDefinition,
-			ExpandedInputs:         packageTask.ExpandedInputs,
-			Command:                command,
-			Framework:              framework,
-			EnvVars: runsummary.TaskEnvVarSummary{
-				Configured: packageTask.HashedEnvVars.BySource.Explicit.ToSecretHashable(),
-				Inferred:   packageTask.HashedEnvVars.BySource.Prefixed.ToSecretHashable(),
-			},
-
-			Hash:         hash,
-			CacheState:   itemStatus,  // TODO(mehulkar): Move this to PackageTask
-			Dependencies: ancestors,   // TODO(mehulkar): Move this to PackageTask
-			Dependents:   descendents, // TODO(mehulkar): Move this to PackageTask
-		})
+		// Assign some fallbacks if they were missing
+		if taskSummary.Command == "" {
+			taskSummary.Command = runsummary.MissingTaskLabel
+		}
+
+		if taskSummary.Framework == "" {
+			taskSummary.Framework = runsummary.MissingFrameworkLabel
+		}
+
+		taskSummary.CacheState = itemStatus  // TODO(mehulkar): Move this to PackageTask
+		taskSummary.Dependencies = ancestors // TODO(mehulkar): Move this to PackageTask
+		taskSummary.Dependents = descendents // TODO(mehulkar): Move this to PackageTask
+
+		taskIDs = append(taskIDs, taskSummary)
 
 		return nil
 	}

cli/internal/run/real_run.go

@@ -24,6 +24,7 @@ import (
 	"github.com/vercel/turbo/cli/internal/packagemanager"
 	"github.com/vercel/turbo/cli/internal/process"
 	"github.com/vercel/turbo/cli/internal/runcache"
+	"github.com/vercel/turbo/cli/internal/runsummary"
 	"github.com/vercel/turbo/cli/internal/spinner"
 	"github.com/vercel/turbo/cli/internal/taskhash"
 	"github.com/vercel/turbo/cli/internal/turbopath"
@@ -40,6 +41,7 @@ func RealRun(
 	turboCache cache.Cache,
 	packagesInScope []string,
 	base *cmdutil.CmdBase,
+	runSummary *runsummary.RunSummary,
 	packageManager *packagemanager.PackageManager,
 	processes *process.Manager,
 	runState *RunState,
@@ -88,8 +90,10 @@ func RealRun(
 		Concurrency: rs.Opts.runOpts.concurrency,
 	}
 
-	execFunc := func(ctx gocontext.Context, packageTask *nodes.PackageTask) error {
+	taskSummaries := []*runsummary.TaskSummary{}
+	execFunc := func(ctx gocontext.Context, packageTask *nodes.PackageTask, taskSummary *runsummary.TaskSummary) error {
 		deps := engine.TaskGraph.DownEdges(packageTask.TaskID)
+		taskSummaries = append(taskSummaries, taskSummary)
 		// deps here are passed in to calculate the task hash
 		return ec.exec(ctx, packageTask, deps)
 	}
@@ -105,6 +109,9 @@ func RealRun(
 	exitCode := 0
 	exitCodeErr := &process.ChildExit{}
 
+	// Assign tasks after execution
+	runSummary.Tasks = taskSummaries
+
 	for _, err := range errs {
 		if errors.As(err, &exitCodeErr) {
 			if exitCodeErr.ExitCode > exitCode {
@@ -120,6 +127,14 @@ func RealRun(
 	if err := runState.Close(base.UI); err != nil {
 		return errors.Wrap(err, "error with profiler")
 	}
+
+	// Write Run Summary if we wanted to
+	if rs.Opts.runOpts.summarize {
+		if err := runSummary.Save(base.RepoRoot, singlePackage); err != nil {
+			base.UI.Warn(fmt.Sprintf("Failed to write run summary: %s", err))
+		}
+	}
+
 	if exitCode != 0 {
 		return &process.ChildExit{
 			ExitCode: exitCode,

cli/internal/run/run.go

@@ -142,6 +142,10 @@ func configureRun(base *cmdutil.CmdBase, opts *Opts, signalWatcher *signals.Watc
 		opts.cacheOpts.SkipFilesystem = true
 	}
 
+	if os.Getenv("TURBO_RUN_SUMMARY") == "true" {
+		opts.runOpts.summarize = true
+	}
+
 	processes := process.NewManager(base.Logger.Named("processes"))
 	signalWatcher.AddOnClose(processes.Close)
 	return &run{
@@ -345,25 +349,22 @@ func (r *run) run(ctx gocontext.Context, targets []string) error {
 		}
 	}
 
+	// RunSummary contains information that is statically analyzable about
+	// the tasks that we expect to run based on the user command.
+	summary := runsummary.NewRunSummary(
+		r.base.TurboVersion,
+		packagesInScope,
+		runsummary.NewGlobalHashSummary(
+			globalHashable.globalFileHashMap,
+			globalHashable.rootExternalDepsHash,
+			globalHashable.hashedSortedEnvPairs,
+			globalHashable.globalCacheKey,
+			globalHashable.pipeline,
+		),
+	)
+
 	// Dry Run
 	if rs.Opts.runOpts.dryRun {
-		// dryRunSummary contains information that is statically analyzable about
-		// the tasks that we expect to run based on the user command.
-		// Currently, we only emit this on dry runs, but it may be useful for real runs later also.
-		summary := &runsummary.RunSummary{
-			TurboVersion: r.base.TurboVersion,
-			Packages:     packagesInScope,
-			// TODO(mehulkar): passing the globalHashable struct directly caused a type mismatch compilation error
-			GlobalHashSummary: runsummary.NewGlobalHashSummary(
-				globalHashable.globalFileHashMap,
-				globalHashable.rootExternalDepsHash,
-				globalHashable.hashedSortedEnvPairs,
-				globalHashable.globalCacheKey,
-				globalHashable.pipeline,
-			),
-			Tasks: []runsummary.TaskSummary{},
-		}
-
 		return DryRun(
 			ctx,
 			g,
@@ -388,6 +389,7 @@ func (r *run) run(ctx gocontext.Context, targets []string) error {
 		turboCache,
 		packagesInScope,
 		r.base,
+		summary,
 		// Extra arg only for regular runs, dry-run doesn't get this
 		packageManager,
 		r.processes,

cli/internal/run/run_spec.go

@@ -82,4 +82,7 @@ type runOpts struct {
 
 	// logPrefix controls whether we should print a prefix in task logs
 	logPrefix string
+
+	// Whether turbo should create a run summary
+	summarize bool
 }

cli/internal/runsummary/format_json.go

@@ -7,19 +7,19 @@ import (
 )
 
 // FormatJSON returns a json string representing a RunSummary
-func (summary *RunSummary) FormatJSON(singlePackage bool) (string, error) {
+func (summary *RunSummary) FormatJSON(singlePackage bool) ([]byte, error) {
 	if singlePackage {
 		return summary.formatJSONSinglePackage()
 	}
 
 	bytes, err := json.MarshalIndent(summary, "", "  ")
 	if err != nil {
-		return "", errors.Wrap(err, "failed to render JSON")
+		return nil, errors.Wrap(err, "failed to render JSON")
 	}
-	return string(bytes), nil
+	return bytes, nil
 }
 
-func (summary *RunSummary) formatJSONSinglePackage() (string, error) {
+func (summary *RunSummary) formatJSONSinglePackage() ([]byte, error) {
 	singlePackageTasks := make([]singlePackageTaskSummary, len(summary.Tasks))
 
 	for i, task := range summary.Tasks {
@@ -30,8 +30,8 @@ func (summary *RunSummary) formatJSONSinglePackage() (string, error) {
 
 	bytes, err := json.MarshalIndent(spSummary, "", "  ")
 	if err != nil {
-		return "", errors.Wrap(err, "failed to render JSON")
+		return nil, errors.Wrap(err, "failed to render JSON")
 	}
 
-	return string(bytes), nil
+	return bytes, nil
 }