Select module files with ancestor boundary lookups

internal/validation/scope_performance_test.go

@@ -0,0 +1,38 @@
+package validation
+
+import (
+	"fmt"
+	"testing"
+
+	"gitea.maximumdirect.net/eric/ReleaseKit/internal/git"
+)
+
+func BenchmarkSelectInputEntriesByModuleCount(b *testing.B) {
+	const retainedFiles = 20000
+	for _, boundaries := range []int{0, 1000, 4000} {
+		b.Run(fmt.Sprintf("boundaries=%d", boundaries), func(b *testing.B) {
+			entries := make([]git.TreeEntry, 0, retainedFiles+2*boundaries)
+			entries = append(entries, git.TreeEntry{Path: "go.mod", Type: "blob", Mode: "100644"})
+			for file := 1; file < retainedFiles; file++ {
+				entries = append(entries, git.TreeEntry{Path: fmt.Sprintf("source/file-%06d.txt", file), Type: "blob", Mode: "100644"})
+			}
+			for module := 0; module < boundaries; module++ {
+				directory := fmt.Sprintf("modules/unit-%06d", module)
+				entries = append(entries,
+					git.TreeEntry{Path: directory + "/go.mod", Type: "blob", Mode: "100644"},
+					git.TreeEntry{Path: directory + "/source.go", Type: "blob", Mode: "100644"})
+			}
+			byPath := make(map[string]int, len(entries))
+			for i, entry := range entries {
+				byPath[entry.Path] = i
+			}
+			b.ReportAllocs()
+			for b.Loop() {
+				selected, err := selectInputEntries(entries, byPath, ".", []string{})
+				if err != nil || len(selected) != retainedFiles {
+					b.Fatalf("selected %d files, error=%v; want %d", len(selected), err, retainedFiles)
+				}
+			}
+		})
+	}
+}

internal/validation/tree.go

@@ -95,7 +95,7 @@ func (tree *CandidateTree) InputEntries() []git.TreeEntry {
 }
 
 func selectInputEntries(entries []git.TreeEntry, byPath map[string]int, modulePath string, sharedFiles []string) ([]git.TreeEntry, error) {
-	boundaries := make([]string, 0)
+	boundaries := make(map[string]struct{})
 	for _, entry := range entries {
 		moduleRelative, ok := relativeToModule(entry.Path, modulePath)
 		if !ok || moduleRelative == "go.mod" || path.Base(moduleRelative) != "go.mod" {
@@ -104,7 +104,7 @@ func selectInputEntries(entries []git.TreeEntry, byPath map[string]int, modulePa
 		if insideVendor(moduleRelative) {
 			continue
 		}
-		boundaries = append(boundaries, path.Dir(entry.Path))
+		boundaries[path.Dir(entry.Path)] = struct{}{}
 	}
 	selected := make(map[string]git.TreeEntry)
 	for _, entry := range entries {
@@ -113,8 +113,8 @@ func selectInputEntries(entries []git.TreeEntry, byPath map[string]int, modulePa
 			continue
 		}
 		excluded := false
-		for _, boundary := range boundaries {
-			if entry.Path == boundary || strings.HasPrefix(entry.Path, boundary+"/") {
+		for ancestor := entry.Path; ancestor != "." && ancestor != modulePath; ancestor = path.Dir(ancestor) {
+			if _, isBoundary := boundaries[ancestor]; isBoundary {
 				excluded = true
 				break
 			}