fix urls

src/archive/mod.rs

@@ -3,7 +3,7 @@ use std::{collections::HashSet, path::PathBuf};
 use crate::{
     blacklist::{check_blacklist, check_blacklist_path},
     conf::get_config,
-    favicon::download_fav_for
+    favicon::download_fav_for,
 };
 
 mod document;
@@ -27,20 +27,28 @@ pub fn read_dir(dir: &PathBuf) -> Vec<String> {
 }
 
 /// Rewrite all URLs in `input` to the format `/s/<domain>/<path..>`
-fn internalize_urls(input: &str) -> String {
-    let url_pattern = r"https?://([a-zA-Z0-9.-]+)(/[\w./-]*)";
+fn internalize_urls(input: &str, base: &str) -> String {
+    let url_pattern = r#"(?:(https?://([a-zA-Z0-9.-]+))?(/[\w./-]*))"#;
     let re = regex::Regex::new(url_pattern).unwrap();
 
     re.replace_all(input, |caps: &regex::Captures| {
-        let domain = caps[1].trim_start_matches("www.");
-        let path = &caps[2];
+        if let Some(domain) = caps.get(2) {
+            let domain = domain.as_str().trim_start_matches("www.");
+            let path = caps.get(3).map_or("", |m| m.as_str());
 
-        // Dont transform if in blacklist
-        if check_blacklist(domain) {
-            return format!("https://{domain}/{path}");
+            // Skip transformation if the domain is in the blacklist
+            if check_blacklist(domain) {
+                format!("https://{domain}{path}")
+            } else {
+                format!("/s/{domain}{path}")
+            }
+        } else if let Some(path) = caps.get(3) {
+            // Handle relative paths
+            format!("/s/{base}{}", path.as_str())
+        } else {
+            // Default fallback
+            caps[0].to_string()
         }
-
-        format!("/s/{domain}/{path}")
     })
     .to_string()
 }