feat: Support shard by scenario (#940)

* feat: Support shard by scenario

* unique names

* refine JSON schema

* refine JSON schema

* Update internal/cucumber/config.go

Co-authored-by: Alex Plischke <alex.plischke@saucelabs.com>

* Update internal/cucumber/scenario/scenario.go

Co-authored-by: Alex Plischke <alex.plischke@saucelabs.com>

* read feature separately

---------

Co-authored-by: Alex Plischke <alex.plischke@saucelabs.com>

api/saucectl.schema.json

@@ -2468,11 +2468,12 @@
                   ]
                 },
                 "shard": {
-                  "description": "When sharding is configured, saucectl automatically splits the tests (e.g. by spec or concurrency) so that they can easily run in parallel.",
+                  "description": "When sharding is configured, saucectl automatically splits the tests (e.g. by spec, concurrency or scenario) so that they can easily run in parallel.",
                   "enum": [
                     "",
                     "concurrency",
-                    "spec"
+                    "spec",
+                    "scenario"
                   ]
                 },
                 "shardTagsEnabled": {

api/v1alpha/framework/playwright-cucumberjs.schema.json

@@ -142,11 +142,12 @@
             ]
           },
           "shard": {
-            "description": "When sharding is configured, saucectl automatically splits the tests (e.g. by spec or concurrency) so that they can easily run in parallel.",
+            "description": "When sharding is configured, saucectl automatically splits the tests (e.g. by spec, concurrency or scenario) so that they can easily run in parallel.",
             "enum": [
               "",
               "concurrency",
-              "spec"
+              "spec",
+              "scenario"
             ]
           },
           "shardTagsEnabled": {

internal/cucumber/config.go

@@ -12,6 +12,7 @@ import (
 	"github.com/rs/zerolog/log"
 	"github.com/saucelabs/saucectl/internal/concurrency"
 	"github.com/saucelabs/saucectl/internal/config"
+	"github.com/saucelabs/saucectl/internal/cucumber/scenario"
 	"github.com/saucelabs/saucectl/internal/cucumber/tag"
 	"github.com/saucelabs/saucectl/internal/fpath"
 	"github.com/saucelabs/saucectl/internal/insights"
@@ -231,7 +232,7 @@ func shardSuites(rootDir string, suites []Suite, ccy int) ([]Suite, error) {
 	var shardedSuites []Suite
 
 	for _, s := range suites {
-		if s.Shard != "spec" && s.Shard != "concurrency" {
+		if s.Shard == "" {
 			shardedSuites = append(shardedSuites, s)
 			continue
 		}
@@ -291,6 +292,15 @@ func shardSuites(rootDir string, suites []Suite, ccy int) ([]Suite, error) {
 				shardedSuites = append(shardedSuites, replica)
 			}
 		}
+		if s.Shard == "scenario" {
+			scenarios := scenario.List(os.DirFS(rootDir), testFiles)
+			for _, name := range scenario.GetUniqueNames(scenarios) {
+				replica := s
+				replica.Name = fmt.Sprintf("%s - %s", s.Name, name)
+				replica.Options.Name = fmt.Sprintf("^%s$", name)
+				shardedSuites = append(shardedSuites, replica)
+			}
+		}
 	}
 
 	return shardedSuites, nil

internal/cucumber/scenario/scenario.go

@@ -0,0 +1,55 @@
+package scenario
+
+import (
+	"io/fs"
+
+	gherkin "github.com/cucumber/gherkin/go/v28"
+	messages "github.com/cucumber/messages/go/v24"
+	"github.com/rs/zerolog/log"
+)
+
+// List parses the provided files and returns a list of scenarios.
+func List(sys fs.FS, files []string) []*messages.Pickle {
+	uuid := &messages.UUID{}
+
+	var scenarios []*messages.Pickle
+	for _, filename := range files {
+		scenarios = append(scenarios, ReadFile(sys, filename, uuid)...)
+	}
+	return scenarios
+}
+
+// ReadFile reads a feature file and returns the parsed list of scenarios.
+func ReadFile(sys fs.FS, filename string, uuid *messages.UUID) []*messages.Pickle {
+	f, err := sys.Open(filename)
+	if err != nil {
+		log.Warn().Str("filename", filename).Msgf("Failed to open the file: %v", err)
+		return nil
+	}
+	defer f.Close()
+
+	doc, err := gherkin.ParseGherkinDocument(f, uuid.NewId)
+	if err != nil {
+		log.Warn().
+			Str("filename", filename).
+			Msg("Could not parse file. It will be excluded from sharded execution.")
+		return nil
+	}
+	return gherkin.Pickles(*doc, filename, uuid.NewId)
+}
+
+// GetUniqueNames extracts and returns unique scenario names.
+func GetUniqueNames(scenarios []*messages.Pickle) []string {
+	uniqueMap := make(map[string]bool)
+
+	for _, s := range scenarios {
+		uniqueMap[s.Name] = true
+	}
+
+	var names []string
+	for name := range uniqueMap {
+		names = append(names, name)
+	}
+
+	return names
+}

internal/cucumber/tag/matcher.go

@@ -4,10 +4,9 @@ package tag
 import (
 	"io/fs"
 
-	gherkin "github.com/cucumber/gherkin/go/v28"
 	messages "github.com/cucumber/messages/go/v24"
 	tagexpressions "github.com/cucumber/tag-expressions/go/v6"
-	"github.com/rs/zerolog/log"
+	"github.com/saucelabs/saucectl/internal/cucumber/scenario"
 )
 
 // MatchFiles finds feature files that include scenarios with tags that match the given tag expression.
@@ -23,21 +22,7 @@ func MatchFiles(sys fs.FS, files []string, tagExpression string) (matched []stri
 	uuid := &messages.UUID{}
 
 	for _, filename := range files {
-		f, err := sys.Open(filename)
-		if err != nil {
-			continue
-		}
-		defer f.Close()
-
-		doc, err := gherkin.ParseGherkinDocument(f, uuid.NewId)
-		if err != nil {
-			log.Warn().
-				Str("filename", filename).
-				Msg("Could not parse file. It will be excluded from sharded execution.")
-			continue
-		}
-		scenarios := gherkin.Pickles(*doc, filename, uuid.NewId)
-
+		scenarios := scenario.ReadFile(sys, filename, uuid)
 		hasMatch := false
 		for _, s := range scenarios {
 			if match(s.Tags, tagMatcher) {