perf(hints): compile link patterns on first use

commit: d5dd8bfbffb6a8f4fc7b92a01d08103f72487d30
author: brice · 3d ago · 2026-10-19 14:57 UTC
parent: 9b0b5e5
signature: GPG VerifiedCD88F9E564E0102646FB6077A1A07C7C7C3E349D · arch-gpg Verified for the committer email

1 file · +15 -7

M internal/adapters/hints/ocr.go +15 -7


@@ -168,14 +168,22 @@ func readWords(ctx context.Context, img *image.Gray) ([]OCRWord, int, error) {
168	168		return words, n, nil
169	169		}
170	170
171		-	var domainPattern = regexp.MustCompile(`(?i)^(?:[a-z0-9](?:[a-z0-9-][a-z0-9])?\.)+([a-z]{2,24})(?::[0-9]+)?(?:[/?#]\S)?$`)
172		-	var schemePattern = regexp.MustCompile(`(?i)^(?:https?\|htps\|file\|ftp):/{1,2}\S+$`)
173		-	var emailPattern = regexp.MustCompile(`(?i)^[a-z0-9.!#$%&'*+/=?^_` + "`" + `{\|}~-]+@(?:[a-z0-9-]+\.)+[a-z]{2,24}$`)
174		-	var fileExtensions = strings.Fields("go md txt toml json yaml yml rs py js ts sh c h png jpg log conf lock mod sum html css")
	171	+	// The link patterns compile on the first OCR link check, not at process
	172	+	// start: with hint mode off nothing here costs memory or CPU.
	173	+	var domainPattern = sync.OnceValue(func() *regexp.Regexp {
	174	+	return regexp.MustCompile(`(?i)^(?:[a-z0-9](?:[a-z0-9-][a-z0-9])?\.)+([a-z]{2,24})(?::[0-9]+)?(?:[/?#]\S)?$`)
	175	+	})
	176	+	var schemePattern = sync.OnceValue(func() *regexp.Regexp { return regexp.MustCompile(`(?i)^(?:https?\|htps\|file\|ftp):/{1,2}\S+$`) })
	177	+	var emailPattern = sync.OnceValue(func() *regexp.Regexp {
	178	+	return regexp.MustCompile(`(?i)^[a-z0-9.!#$%&'*+/=?^_` + "`" + `{\|}~-]+@(?:[a-z0-9-]+\.)+[a-z]{2,24}$`)
	179	+	})
	180	+	var fileExtensions = sync.OnceValue(func() []string {
	181	+	return strings.Fields("go md txt toml json yaml yml rs py js ts sh c h png jpg log conf lock mod sum html css")
	182	+	})
175	183
176	184		func isTextLink(text string) bool {
177	185		text = strings.Trim(text, "()[]<>.,;:'\"")
178		-	if schemePattern.MatchString(text) \|\| (strings.HasPrefix(strings.ToLower(text), "www.") && len(text) > 4) \|\| emailPattern.MatchString(text) {
	186	+	if schemePattern().MatchString(text) \|\| (strings.HasPrefix(strings.ToLower(text), "www.") && len(text) > 4) \|\| emailPattern().MatchString(text) {
179	187		return true
180	188		}
181	189		host := text

@@ -185,11 +193,11 @@ func isTextLink(text string) bool {
185	193		if ip := net.ParseIP(host); ip != nil && ip.To4() != nil {
186	194		return true
187	195		}
188		-	m := domainPattern.FindStringSubmatch(text)
	196	+	m := domainPattern().FindStringSubmatch(text)
189	197		if m == nil {
190	198		return false
191	199		}
192		-	for _, ext := range fileExtensions {
	200	+	for _, ext := range fileExtensions() {
193	201		if strings.EqualFold(m[1], ext) {
194	202		return false
195	203		}

 	return words, n, nil
 }
-var domainPattern = regexp.MustCompile(`(?i)^(?:[a-z0-9](?:[a-z0-9-]*[a-z0-9])?\.)+([a-z]{2,24})(?::[0-9]+)?(?:[/?#]\S*)?$`)
+// The link patterns compile on the first OCR link check, not at process
-var schemePattern = regexp.MustCompile(`(?i)^(?:https?|htps|file|ftp):/{1,2}\S+$`)
+// start: with hint mode off nothing here costs memory or CPU.
-var emailPattern = regexp.MustCompile(`(?i)^[a-z0-9.!#$%&'*+/=?^_` + "`" + `{|}~-]+@(?:[a-z0-9-]+\.)+[a-z]{2,24}$`)
+var domainPattern = sync.OnceValue(func() *regexp.Regexp {
-var fileExtensions = strings.Fields("go md txt toml json yaml yml rs py js ts sh c h png jpg log conf lock mod sum html css")
+	return regexp.MustCompile(`(?i)^(?:[a-z0-9](?:[a-z0-9-]*[a-z0-9])?\.)+([a-z]{2,24})(?::[0-9]+)?(?:[/?#]\S*)?$`)
+})
+var schemePattern = sync.OnceValue(func() *regexp.Regexp { return regexp.MustCompile(`(?i)^(?:https?|htps|file|ftp):/{1,2}\S+$`) })
+var emailPattern = sync.OnceValue(func() *regexp.Regexp {
+	return regexp.MustCompile(`(?i)^[a-z0-9.!#$%&'*+/=?^_` + "`" + `{|}~-]+@(?:[a-z0-9-]+\.)+[a-z]{2,24}$`)
+})
+var fileExtensions = sync.OnceValue(func() []string {
+	return strings.Fields("go md txt toml json yaml yml rs py js ts sh c h png jpg log conf lock mod sum html css")
+})
 func isTextLink(text string) bool {
 	text = strings.Trim(text, "()[]<>.,;:'\"")
-	if schemePattern.MatchString(text) || (strings.HasPrefix(strings.ToLower(text), "www.") && len(text) > 4) || emailPattern.MatchString(text) {
+	if schemePattern().MatchString(text) || (strings.HasPrefix(strings.ToLower(text), "www.") && len(text) > 4) || emailPattern().MatchString(text) {
 		return true
 	}
 	host := text
 	if ip := net.ParseIP(host); ip != nil && ip.To4() != nil {
 		return true
 	}
-	m := domainPattern.FindStringSubmatch(text)
+	m := domainPattern().FindStringSubmatch(text)
 	if m == nil {
 		return false
 	}
-	for _, ext := range fileExtensions {
+	for _, ext := range fileExtensions() {
 		if strings.EqualFold(m[1], ext) {
 			return false
 		}