fix

src/archive/mod.rs

@@ -28,23 +28,36 @@ pub fn read_dir(dir: &PathBuf) -> Vec<String> {
 
 /// Rewrite all URLs in `input` to the format `/s/<domain>/<path..>`
 fn internalize_urls(input: &str, base: &str) -> String {
-    let url_pattern = r#"(?:(https?://([a-zA-Z0-9.-]+))?(/[\w./-]*))"#;
+    let url_pattern = r#"(\s|"|')(?:(<?)(https?:\/\/([a-zA-Z0-9.-]+))?(\/[\w./-]*))"#;
     let re = regex::Regex::new(url_pattern).unwrap();
 
     re.replace_all(input, |caps: &regex::Captures| {
-        if let Some(domain) = caps.get(2) {
+        if caps.get(1).map(|x| x.as_str()).unwrap_or_default() == "<" {
-            let domain = domain.as_str().trim_start_matches("www.");
+            return caps.get(0).unwrap().as_str().to_string();
-            let path = caps.get(3).map_or("", |m| m.as_str());
+        }
+
+        let wrap = caps.get(1).map(|x| x.as_str()).unwrap_or_default();
+
+        if let Some(domain) = caps.get(3) {
+            let domain = domain.as_str();
+            let (protocol, domain) = if domain.starts_with("https://") {
+                ("https", domain.trim_start_matches("https://"))
+            } else {
+                ("http", domain.trim_start_matches("http://"))
+            };
+
+            let domain = domain.trim_start_matches("www.");
+            let path = caps.get(5).map_or("", |m| m.as_str());
 
             // Skip transformation if the domain is in the blacklist
             if check_blacklist(domain) {
-                format!("https://{domain}{path}")
+                format!("{wrap}{protocol}://{domain}{path}")
             } else {
-                format!("/s/{domain}{path}")
+                format!("{wrap}/s/{domain}{path}")
             }
-        } else if let Some(path) = caps.get(3) {
+        } else if let Some(path) = caps.get(5) {
             // Handle relative paths
-            format!("/s/{base}{}", path.as_str())
+            format!("{wrap}/s/{base}{}", path.as_str())
         } else {
             // Default fallback
             caps[0].to_string()