chore: compile regexs only once

internetarchive · Aug 23, 2024 · 25c8f40 · 25c8f40
1 parent ed0f33b
commit 25c8f40
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 13 deletions.
diff --git a/internal/pkg/crawl/assets.go b/internal/pkg/crawl/assets.go
@@ -12,6 +12,9 @@ import (
 	"github.com/internetarchive/Zeno/internal/pkg/utils"
 )
 
+var backgroundImageRegex = regexp.MustCompile(`(?:\(['"]?)(.*?)(?:['"]?\))`)
+var urlRegex = regexp.MustCompile(`(?m)url\((.*?)\)`)
+
 func (c *Crawl) extractAssets(base *url.URL, item *queue.Item, doc *goquery.Document) (assets []*url.URL, err error) {
 	var rawAssets []string
 	var URL = utils.URLToString(item.URL)
@@ -45,8 +48,7 @@ func (c *Crawl) extractAssets(base *url.URL, item *queue.Item, doc *goquery.Docu
 	doc.Find("*").Each(func(index int, item *goquery.Selection) {
 		style, exists := item.Attr("style")
 		if exists {
-			re := regexp.MustCompile(`(?:\(['"]?)(.*?)(?:['"]?\))`)
-			matches := re.FindAllStringSubmatch(style, -1)
+			matches := backgroundImageRegex.FindAllStringSubmatch(style, -1)
 
 			for match := range matches {
 				if len(matches[match]) > 0 {
@@ -110,9 +112,7 @@ func (c *Crawl) extractAssets(base *url.URL, item *queue.Item, doc *goquery.Docu
 
 	if !utils.StringInSlice("style", c.DisabledHTMLTags) {
 		doc.Find("style").Each(func(index int, item *goquery.Selection) {
-			re := regexp.MustCompile(`(?m)url\((.*?)\)`)
-			matches := re.FindAllStringSubmatch(item.Text(), -1)
-
+			matches := urlRegex.FindAllStringSubmatch(item.Text(), -1)
 			for match := range matches {
 				matchReplacement := matches[match][1]
 				matchReplacement = strings.Replace(matchReplacement, "'", "", -1)

diff --git a/internal/pkg/crawl/sitespecific/truthsocial/truthsocial.go b/internal/pkg/crawl/sitespecific/truthsocial/truthsocial.go
@@ -7,15 +7,10 @@ import (
 	"strings"
 )
 
-func IsTruthSocialURL(URL string) bool {
-	regexPattern := `https?://truthsocial\.com/@[A-Za-z0-9_]+/posts/\d+`
+var truthSocialPostURLRegex = regexp.MustCompile(`https?://truthsocial\.com/@[A-Za-z0-9_]+/posts/\d+`)
 
-	match, err := regexp.MatchString(regexPattern, URL)
-	if err != nil {
-		return false
-	}
-
-	return match
+func IsTruthSocialURL(URL string) bool {
+	return truthSocialPostURLRegex.MatchString(URL)
 }
 
 func extractPostID(URL string) (string, error) {