fix: Added custom error message only for the 404 case and a method to automatically fix the casing of the crawl reference

Author: pjox

src/cli.rs

@@ -95,9 +95,11 @@ fn crawl_name_format(crawl: &str) -> Result<String, String> {
     let main_re = Regex::new(r"^(CC\-MAIN)\-([0-9]{4})\-([0-9]{2})$").unwrap();
     let news_re = Regex::new(r"^(CC\-NEWS)\-([0-9]{4})\-([0-9]{2})$").unwrap();
 
-    if !(main_re.is_match(crawl) || news_re.is_match(crawl)) {
-        return Err("Please use the CC-MAIN-YYYY-WW or the CC-NEWS-YYYY-MM format, make sure your input is propely capitalized".to_string());
+    let crawl_ref = crawl.to_uppercase();
+
+    if !(main_re.is_match(&crawl_ref) || news_re.is_match(&crawl_ref)) {
+        return Err("Please use the CC-MAIN-YYYY-WW or the CC-NEWS-YYYY-MM format.".to_string());
     } else {
-        return Ok(crawl.to_owned());
+        return Ok(crawl_ref);
     }
 }

src/download.rs

@@ -104,17 +104,21 @@ pub async fn download_paths(mut options: DownloadOptions<'_>) -> Result<(), Down
     let resp = client.head(url.as_str()).send().await?;
     match resp.status() {
         status if status.is_success() => (),
-        status if status.is_client_error() => {
+        status if status.as_u16() == 404 => {
             return Err(format!(
-                "\n\nThe reference combination you requested:\n\tCRAWL: {}\n\tSUBSET: {}\n\tURL: {}\n\nDoesn't seem to exist or it is currently not accessible.\n\tError Code: {} {}",
+                "\n\nThe reference combination you requested:\n\tCRAWL: {}\n\tSUBSET: {}\n\tURL: {}\n\nDoesn't seem to exist or it is currently not accessible.\n\tError code: {} {}",
                 snapshot_original_ref, options.data_type, url, status.as_str(), status.canonical_reason().unwrap_or("")
             )
             .into());
         }
-        _ => {
-            return Err(
-                format!("Couldn't download URL: {}. Error: {:?}", url, resp.status()).into(),
-            );
+        status => {
+            return Err(format!(
+                "Couldn't download URL: {}. Error code: {} {}",
+                url,
+                status.as_str(),
+                status.canonical_reason().unwrap_or("")
+            )
+            .into());
         }
     }