perf: preload .bazelignore before configure, do not use globs

walk/config.go

@@ -25,7 +25,6 @@ import (
 	"os"
 	"path"
 	"strings"
-	"sync"
 
 	"github.com/bazelbuild/bazel-gazelle/config"
 	"github.com/bazelbuild/bazel-gazelle/rule"
@@ -42,7 +41,6 @@ type walkConfig struct {
 	excludes []string
 	ignore   bool
 	follow   []string
-	loadOnce *sync.Once
 }
 
 const walkName = "_walk"
@@ -67,7 +65,7 @@ var _ config.Configurer = (*Configurer)(nil)
 type Configurer struct{}
 
 func (*Configurer) RegisterFlags(fs *flag.FlagSet, cmd string, c *config.Config) {
-	wc := &walkConfig{loadOnce: &sync.Once{}}
+	wc := &walkConfig{}
 	c.Exts[walkName] = wc
 	fs.Var(&gzflag.MultiFlag{Values: &wc.excludes}, "exclude", "pattern that should be ignored (may be repeated)")
 }
@@ -84,12 +82,6 @@ func (cr *Configurer) Configure(c *config.Config, rel string, f *rule.File) {
 	*wcCopy = *wc
 	wcCopy.ignore = false
 
-	wc.loadOnce.Do(func() {
-		if err := cr.loadBazelIgnore(c.RepoRoot, wcCopy); err != nil {
-			log.Printf("error loading .bazelignore: %v", err)
-		}
-	})
-
 	if f != nil {
 		for _, d := range f.Directives {
 			switch d.Key {
@@ -114,17 +106,26 @@ func (cr *Configurer) Configure(c *config.Config, rel string, f *rule.File) {
 	c.Exts[walkName] = wcCopy
 }
 
-func (c *Configurer) loadBazelIgnore(repoRoot string, wc *walkConfig) error {
+type isIgnoredFunc = func(string) bool
+
+var nothingIgnored isIgnoredFunc = func(string) bool { return false }
+
+// Repo-wide bazelignore paths.
+// Only applies to files on disk, not generated files.
+// Does not support glob patterns.
+func loadBazelIgnore(repoRoot string) (isIgnoredFunc, error) {
 	ignorePath := path.Join(repoRoot, ".bazelignore")
 	file, err := os.Open(ignorePath)
 	if errors.Is(err, fs.ErrNotExist) {
-		return nil
+		return nothingIgnored, nil
 	}
 	if err != nil {
-		return fmt.Errorf(".bazelignore exists but couldn't be read: %v", err)
+		return nothingIgnored, fmt.Errorf(".bazelignore exists but couldn't be read: %v", err)
 	}
 	defer file.Close()
 
+	excludes := make(map[string]struct{})
+
 	scanner := bufio.NewScanner(file)
 	for scanner.Scan() {
 		ignore := strings.TrimSpace(scanner.Text())
@@ -142,9 +143,15 @@ func (c *Configurer) loadBazelIgnore(repoRoot string, wc *walkConfig) error {
 		// the exclude matching won't work correctly.
 		ignore = path.Clean(ignore)
 
-		wc.excludes = append(wc.excludes, ignore)
+		excludes[ignore] = struct{}{}
+	}
+
+	isIgnored := func(p string) bool {
+		_, ok := excludes[p]
+		return ok
 	}
-	return nil
+
+	return isIgnored, nil
 }
 
 func checkPathMatchPattern(pattern string) error {

walk/walk.go

@@ -117,6 +117,11 @@ func Walk(c *config.Config, cexts []config.Configurer, dirs []string, mode Mode,
 
 	updateRels := NewUpdateFilter(c.RepoRoot, dirs, mode)
 
+	isBazelIgnored, err := loadBazelIgnore(c.RepoRoot)
+	if err != nil {
+		log.Printf("error loading .bazelignore: %v", err)
+	}
+
 	var visit func(*config.Config, string, string, bool)
 	visit = func(c *config.Config, dir, rel string, updateParent bool) {
 		haveError := false
@@ -141,6 +146,10 @@ func Walk(c *config.Config, cexts []config.Configurer, dirs []string, mode Mode,
 			haveError = true
 		}
 
+		if isBazelIgnored(rel) {
+			return
+		}
+
 		c = configure(cexts, knownDirectives, c, rel, f)
 		wc := getWalkConfig(c)
 
@@ -151,6 +160,10 @@ func Walk(c *config.Config, cexts []config.Configurer, dirs []string, mode Mode,
 		var subdirs, regularFiles []string
 		for _, ent := range ents {
 			base := ent.Name()
+			repoRel := path.Join(rel, base)
+			if isBazelIgnored(repoRel) || wc.isExcluded(rel, base) {
+				continue
+			}
 			ent := resolveFileInfo(wc, dir, rel, ent)
 			switch {
 			case ent == nil:
@@ -326,14 +339,14 @@ func findGenFiles(wc *walkConfig, f *rule.File) []string {
 
 func resolveFileInfo(wc *walkConfig, dir, rel string, ent fs.DirEntry) fs.DirEntry {
 	base := ent.Name()
-	if base == "" || wc.isExcluded(rel, base) {
+	if base == "" {
 		return nil
 	}
 	if ent.Type()&os.ModeSymlink == 0 {
 		// Not a symlink, use the original FileInfo.
 		return ent
 	}
-	if !wc.shouldFollow(rel, ent.Name()) {
+	if !wc.shouldFollow(rel, base) {
 		// A symlink, but not one we should follow.
 		return nil
 	}