From a3d1b7cc09ca23ec78d2bbb857dd41e1d4b51a3e Mon Sep 17 00:00:00 2001
From: Jason Bedard <jason+github@jbedard.ca>
Date: Wed, 21 Aug 2024 15:31:47 -0700
Subject: [PATCH] perf: reduce path.join invocations

---
 walk/config.go | 32 +++++++++++++-------------------
 walk/walk.go   | 36 ++++++++++++++++++++++++++----------
 2 files changed, 39 insertions(+), 29 deletions(-)

diff --git a/walk/config.go b/walk/config.go
index c7fc675c3..80808fa27 100644
--- a/walk/config.go
+++ b/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"
@@ -51,15 +49,15 @@ func getWalkConfig(c *config.Config) *walkConfig {
 	return c.Exts[walkName].(*walkConfig)
 }
 
-func (wc *walkConfig) isExcluded(rel, base string) bool {
-	if base == ".git" {
+func (wc *walkConfig) isExcluded(p string) bool {
+	if strings.HasPrefix(p, ".git") && (len(p) < 5 || p[4] == '/') {
 		return true
 	}
-	return matchAnyGlob(wc.excludes, path.Join(rel, base))
+	return matchAnyGlob(wc.excludes, p)
 }
 
-func (wc *walkConfig) shouldFollow(rel, base string) bool {
-	return matchAnyGlob(wc.follow, path.Join(rel, base))
+func (wc *walkConfig) shouldFollow(p string) bool {
+	return matchAnyGlob(wc.follow, p)
 }
 
 var _ config.Configurer = (*Configurer)(nil)
@@ -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 +106,19 @@ 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 {
+func loadBazelIgnore(repoRoot string) (map[string]struct{}, error) {
 	ignorePath := path.Join(repoRoot, ".bazelignore")
 	file, err := os.Open(ignorePath)
 	if errors.Is(err, fs.ErrNotExist) {
-		return nil
+		return nil, nil
 	}
 	if err != nil {
-		return fmt.Errorf(".bazelignore exists but couldn't be read: %v", err)
+		return nil, 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 +136,9 @@ 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{}{}
 	}
-	return nil
+	return excludes, nil
 }
 
 func checkPathMatchPattern(pattern string) error {
diff --git a/walk/walk.go b/walk/walk.go
index 81550a2fa..791aea8ec 100644
--- a/walk/walk.go
+++ b/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)
 
+	bazelExcludes, bazelExcludesErr := loadBazelIgnore(c.RepoRoot)
+	if bazelExcludesErr != nil {
+		log.Printf("error loading .bazelignore: %v", bazelExcludesErr)
+	}
+
 	var visit func(*config.Config, string, string, bool)
 	visit = func(c *config.Config, dir, rel string, updateParent bool) {
 		haveError := false
@@ -142,16 +147,31 @@ func Walk(c *config.Config, cexts []config.Configurer, dirs []string, mode Mode,
 		}
 
 		c = configure(cexts, knownDirectives, c, rel, f)
+
+		if _, bazelExcluded := bazelExcludes[rel]; bazelExcluded {
+			return
+		}
+
 		wc := getWalkConfig(c)
 
-		if wc.isExcluded(rel, ".") {
+		if wc.isExcluded(rel) {
 			return
 		}
 
 		var subdirs, regularFiles []string
 		for _, ent := range ents {
 			base := ent.Name()
-			ent := resolveFileInfo(wc, dir, rel, ent)
+			relBase := path.Join(rel, base)
+
+			if _, bazelExcluded := bazelExcludes[relBase]; bazelExcluded {
+				continue
+			}
+
+			if wc.isExcluded(relBase) {
+				continue
+			}
+
+			ent := resolveFileInfo(wc, c.RepoRoot, relBase, ent)
 			switch {
 			case ent == nil:
 				continue
@@ -317,27 +337,23 @@ func findGenFiles(wc *walkConfig, f *rule.File) []string {
 
 	var genFiles []string
 	for _, s := range strs {
-		if !wc.isExcluded(f.Pkg, s) {
+		if !wc.isExcluded(path.Join(f.Pkg, s)) {
 			genFiles = append(genFiles, s)
 		}
 	}
 	return genFiles
 }
 
-func resolveFileInfo(wc *walkConfig, dir, rel string, ent fs.DirEntry) fs.DirEntry {
-	base := ent.Name()
-	if base == "" || wc.isExcluded(rel, base) {
-		return nil
-	}
+func resolveFileInfo(wc *walkConfig, dir, repoRel string, ent fs.DirEntry) fs.DirEntry {
 	if ent.Type()&os.ModeSymlink == 0 {
 		// Not a symlink, use the original FileInfo.
 		return ent
 	}
-	if !wc.shouldFollow(rel, ent.Name()) {
+	if !wc.shouldFollow(repoRel) {
 		// A symlink, but not one we should follow.
 		return nil
 	}
-	fi, err := os.Stat(path.Join(dir, base))
+	fi, err := os.Stat(path.Join(dir, repoRel))
 	if err != nil {
 		// A symlink, but not one we could resolve.
 		return nil