simple preference for IA URLs (if available)

1 files changed, 15 insertions, 3 deletions

diff --git a/fatcat-cli/src/download.rs b/fatcat-cli/src/download.rs
index a0f0eeb..53c5a41 100644
--- a/fatcat-cli/src/download.rs
+++ b/fatcat-cli/src/download.rs
@@ -148,13 +148,25 @@ pub fn download_file(
 
     let download_path = final_path.with_extension("partial_download");
 
-    // TODO: only archive.org URLs (?)
     let raw_url = match fe.urls.as_ref() {
         None => return Ok(DownloadStatus::NoPublicFile),
         Some(url_list) if url_list.is_empty() => return Ok(DownloadStatus::NoPublicFile),
         // TODO: remove clone (?)
-        // TODO: better heuristic than "just try first URL"
-        Some(url_list) => url_list[0].url.clone(),
+        Some(url_list) => {
+            // prefer an IA_hosted URL, but fallback to any URL
+            let public_url_list: Vec<&fatcat_openapi::models::FileUrl> = url_list.iter().filter(|v| {
+                if let Ok(url) = Url::parse(&v.url) {
+                    url.host_str() == Some("web.archive.org") || url.host_str() == Some("archive.org")
+                } else {
+                    false
+                }
+            }).collect();
+            if !public_url_list.is_empty() {
+                public_url_list[0].url.clone()
+            } else {
+                url_list[0].url.clone()
+            }
+        },
     };
 
     let mut url = Url::parse(&raw_url)?;