✨ any mime

src/archive/mod.rs

@@ -32,7 +32,7 @@ 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 {
+pub fn internalize_urls(input: &str, base: &str) -> String {
     // todo : fix regex, domains without path are not captured
     let url_pattern = r#"(\ |"|')(?:(<?)(https?:\/\/([a-zA-Z0-9.-]+))?(\/[\w./-]*))"#;
     let re = regex::Regex::new(url_pattern).unwrap();
@@ -172,7 +172,7 @@ impl WebsiteArchive {
         std::fs::create_dir_all(&folder_name).unwrap();
 
         let timestamp = chrono::Utc::now().format("%Y-%m-%d").to_string();
-        let filename = folder_name.join(format!("index_{timestamp}.html"));
+        let filename = folder_name.join(format!("index_{timestamp}"));
 
         log::info!("Archiving {url} to {}", filename.to_str().unwrap());
 
@@ -238,10 +238,17 @@ fn run_command(cmd: &[&str]) {
     let child = cmd_setup.spawn().unwrap();
 
     let status = child.wait_with_output().unwrap();
-    assert!(status.status.success());
+    if !status.status.success() {
+        log::warn!(
+            "Command {cmd:?} exited with code {}",
+            status.status.code().unwrap_or_default()
+        )
+    }
 }
 
 pub async fn index_archive_db(arc: &WebsiteArchive) {
+    // TODO : more index attrs size,mime
+
     log::info!("Indexing archive");
 
     for dom in arc.domains() {