From b296f21e98d284869d0e12d0e177caa8dbc658ad Mon Sep 17 00:00:00 2001
From: jvoisin <julien.voisin@dustri.org>
Date: Tue, 17 Jun 2025 16:34:13 +0200
Subject: [PATCH] refactor(internal): add an urllib.DomainWithoutWWW function

---
 internal/reader/sanitizer/sanitizer.go | 2 +-
 internal/reader/scraper/scraper.go     | 3 +--
 internal/urllib/url.go                 | 5 +++++
 3 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/internal/reader/sanitizer/sanitizer.go b/internal/reader/sanitizer/sanitizer.go
index 677fc218..41191e2f 100644
--- a/internal/reader/sanitizer/sanitizer.go
+++ b/internal/reader/sanitizer/sanitizer.go
@@ -486,7 +486,7 @@ func isBlockedResource(absoluteURL string) bool {
 }
 
 func isValidIframeSource(iframeSourceURL string) bool {
-	iframeSourceDomain := strings.TrimPrefix(urllib.Domain(iframeSourceURL), "www.")
+	iframeSourceDomain := urllib.DomainWithoutWWW(iframeSourceURL)
 
 	if _, ok := iframeAllowList[iframeSourceDomain]; ok {
 		return true
diff --git a/internal/reader/scraper/scraper.go b/internal/reader/scraper/scraper.go
index cf17e0c5..6e2e9237 100644
--- a/internal/reader/scraper/scraper.go
+++ b/internal/reader/scraper/scraper.go
@@ -93,8 +93,7 @@ func findContentUsingCustomRules(page io.Reader, rules string) (baseURL string,
 }
 
 func getPredefinedScraperRules(websiteURL string) string {
-	urlDomain := urllib.Domain(websiteURL)
-	urlDomain = strings.TrimPrefix(urlDomain, "www.")
+	urlDomain := urllib.DomainWithoutWWW(websiteURL)
 
 	if rules, ok := predefinedRules[urlDomain]; ok {
 		return rules
diff --git a/internal/urllib/url.go b/internal/urllib/url.go
index 56d41d96..259b7a9b 100644
--- a/internal/urllib/url.go
+++ b/internal/urllib/url.go
@@ -83,6 +83,11 @@ func Domain(websiteURL string) string {
 	return parsedURL.Host
 }
 
+// DomainWithoutWWW returns only the domain part of the given URL, with the "www." prefix removed if present.
+func DomainWithoutWWW(websiteURL string) string {
+	return strings.TrimPrefix(Domain(websiteURL), "www.")
+}
+
 // JoinBaseURLAndPath returns a URL string with the provided path elements joined together.
 func JoinBaseURLAndPath(baseURL, path string) (string, error) {
 	if baseURL == "" {