extract and test cache checking

Author: dmathieu

lychee-bin/src/commands/check.rs

@@ -10,7 +10,7 @@ use reqwest::Url;
 use tokio::sync::mpsc;
 use tokio_stream::wrappers::ReceiverStream;
 
-use lychee_lib::{Client, Request, Response};
+use lychee_lib::{Client, Request, Response, Uri};
 use lychee_lib::{InputSource, Result};
 use lychee_lib::{ResponseBody, Status};
 
@@ -275,22 +275,25 @@ async fn handle(
     // - Skip caching excluded links; they might not be excluded in the next run.
     // - Skip caching links for which the status code has been explicitly excluded from the cache.
     let status = response.status();
+    if ignore_cache(&uri, status, cache_exclude_status) {
+        return response;
+    }
+
+    cache.insert(uri, status.into());
+    response
+}
+
+fn ignore_cache(uri: &Uri, status: &Status, cache_exclude_status: HashSet<u16>) -> bool {
     let status_code = match status.code() {
         None => 0,
         Some(code) => code.as_u16(),
     };
 
-    if uri.is_file()
+    return uri.is_file()
         || status.is_excluded()
         || status.is_unsupported()
         || status.is_unknown()
-        || cache_exclude_status.contains(&status_code)
-    {
-        return response;
-    }
-
-    cache.insert(uri, status.into());
-    response
+        || cache_exclude_status.contains(&status_code);
 }
 
 fn show_progress(
@@ -338,9 +341,10 @@ fn get_failed_urls(stats: &mut ResponseStats) -> Vec<(InputSource, Url)> {
 
 #[cfg(test)]
 mod tests {
+    use http::StatusCode;
     use log::info;
 
-    use lychee_lib::{CacheStatus, InputSource, ResponseBody, Uri};
+    use lychee_lib::{CacheStatus, ErrorKind, InputSource, ResponseBody, Uri};
 
     use crate::formatters;
 
@@ -389,4 +393,60 @@ mod tests {
         let buf = String::from_utf8_lossy(&buf);
         assert_eq!(buf, "↻ [200] http://127.0.0.1/ | Cached: OK (cached)\n");
     }
+
+    #[test]
+    fn test_ignore_cache() {
+        let mut exclude = HashSet::new();
+
+        // Cache is not ignored
+        assert_eq!(
+            false,
+            ignore_cache(
+                &Uri::try_from("https://[::1]").unwrap(),
+                &Status::Ok(StatusCode::OK),
+                exclude.clone()
+            )
+        );
+
+        // Cache is ignored for file URLs
+        assert_eq!(
+            true,
+            ignore_cache(
+                &Uri::try_from("file:///home").unwrap(),
+                &Status::Ok(StatusCode::OK),
+                exclude.clone()
+            )
+        );
+
+        // Cache is ignored for unsupported status
+        assert_eq!(
+            true,
+            ignore_cache(
+                &Uri::try_from("https://[::1]").unwrap(),
+                &Status::Unsupported(ErrorKind::EmptyUrl),
+                exclude.clone()
+            )
+        );
+
+        // Cache is ignored for unknown status
+        assert_eq!(
+            true,
+            ignore_cache(
+                &Uri::try_from("https://[::1]").unwrap(),
+                &Status::UnknownStatusCode(StatusCode::IM_A_TEAPOT),
+                exclude.clone()
+            )
+        );
+
+        // Cache is ignored for excluded status codes
+        exclude.insert(200);
+        assert_eq!(
+            true,
+            ignore_cache(
+                &Uri::try_from("https://[::1]").unwrap(),
+                &Status::Ok(StatusCode::OK),
+                exclude.clone()
+            )
+        );
+    }
 }