Merge branch 'well-known-bug' into 'master'

bcmr: Look for well-known on bare domains

Closes #24

See merge request riftenlabs/riftenlabs-indexer!83

src/bcmr/bcmrdownloader.rs

@@ -36,6 +36,45 @@ const MAX_BCMR_SIZE: usize = 1024 * 1024 * 100; // 100 MB
 
 const MAX_PARALLEL_DOWNLOADS: usize = 5;
 
+const WELL_KNOWN_PATH: &str = "/.well-known/bitcoin-cash-metadata-registry.json";
+
+/// Resolve an on-chain BCMR URI into one or more candidate URLs to try in order.
+///
+/// Handles three cases:
+/// - `ipfs://<cid>` → a single random IPFS gateway URL.
+/// - Bare domain (authority only, no path) → BCMR v2 well-known URI first, then
+///   the bare root URL as a fallback for bug-compatibility with tokens that
+///   served the registry directly at `/`.
+/// - Anything with a path → used as-is (with `https://` prepended if missing).
+fn resolve_uri_candidates(uri: &str) -> Vec<String> {
+    if let Some(stripped) = uri.strip_prefix("ipfs://") {
+        return vec![format!(
+            "{}{}",
+            IPFS_GATEWAYS.choose(&mut thread_rng()).unwrap(),
+            stripped
+        )];
+    }
+
+    let (scheme, rest) = match uri.split_once("://") {
+        Some((s, r)) => (s, r),
+        None => ("https", uri),
+    };
+
+    let (authority, path) = match rest.split_once('/') {
+        Some((a, p)) => (a, format!("/{p}")),
+        None => (rest, String::new()),
+    };
+
+    if path.is_empty() || path == "/" {
+        vec![
+            format!("{scheme}://{authority}{WELL_KNOWN_PATH}"),
+            format!("{scheme}://{authority}/"),
+        ]
+    } else {
+        vec![format!("{scheme}://{authority}{path}")]
+    }
+}
+
 pub struct BCMRDownloader {
     db: SqlitePool,
     keep_running: Arc<AtomicBool>,
@@ -67,31 +106,21 @@ async fn fetch_bcmr(
     let mut errors: Vec<String> = Vec::new();
 
     for url in urls {
-        let url = if let Some(stripped) = url.strip_prefix("ipfs://") {
-            format!(
-                "{}{}",
-                IPFS_GATEWAYS.choose(&mut thread_rng()).unwrap(),
-                stripped
-            )
-        } else if !url.starts_with("https://") {
-            format!("https://{url}")
-        } else {
-            url.to_string()
-        };
+        for resolved in resolve_uri_candidates(url) {
+            let (contents, actual_hash) =
+                match get_url(client, &resolved, DOWNLOAD_TIMEOUT, MAX_BCMR_SIZE).await {
+                    Ok(c) => c,
+                    Err(e) => {
+                        errors.push(e.to_string());
+                        continue;
+                    }
+                };
 
-        let (contents, actual_hash) =
-            match get_url(client, &url, DOWNLOAD_TIMEOUT, MAX_BCMR_SIZE).await {
-                Ok(c) => c,
-                Err(e) => {
-                    errors.push(e.to_string());
-                    continue;
-                }
-            };
-
-        if actual_hash == expected_hash {
-            return (Some((contents, actual_hash)), String::default(), false);
-        } else {
-            candidate = Some((contents, actual_hash))
+            if actual_hash == expected_hash {
+                return (Some((contents, actual_hash)), String::default(), false);
+            } else {
+                candidate = Some((contents, actual_hash))
+            }
         }
     }
 
@@ -272,3 +301,72 @@ impl Drop for BCMRDownloader {
         }
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn bare_domain_resolves_to_well_known_then_root() {
+        let candidates = resolve_uri_candidates("chipnet-16.paryonusd.com");
+        assert_eq!(
+            candidates,
+            vec![
+                "https://chipnet-16.paryonusd.com/.well-known/bitcoin-cash-metadata-registry.json"
+                    .to_string(),
+                "https://chipnet-16.paryonusd.com/".to_string(),
+            ]
+        );
+    }
+
+    #[test]
+    fn bare_domain_with_scheme_and_trailing_slash() {
+        let candidates = resolve_uri_candidates("https://example.com/");
+        assert_eq!(
+            candidates,
+            vec![
+                "https://example.com/.well-known/bitcoin-cash-metadata-registry.json".to_string(),
+                "https://example.com/".to_string(),
+            ]
+        );
+    }
+
+    #[test]
+    fn url_with_path_is_used_as_is() {
+        let candidates = resolve_uri_candidates("https://example.com/path/registry.json");
+        assert_eq!(
+            candidates,
+            vec!["https://example.com/path/registry.json".to_string()]
+        );
+    }
+
+    #[test]
+    fn schemeless_url_with_path_gets_https() {
+        let candidates = resolve_uri_candidates("example.com/registry.json");
+        assert_eq!(
+            candidates,
+            vec!["https://example.com/registry.json".to_string()]
+        );
+    }
+
+    #[test]
+    fn ipfs_uri_uses_a_gateway() {
+        let candidates = resolve_uri_candidates("ipfs://QmAbc123");
+        assert_eq!(candidates.len(), 1);
+        assert!(candidates[0].ends_with("/QmAbc123"));
+        assert!(IPFS_GATEWAYS.iter().any(|gw| candidates[0].starts_with(gw)));
+    }
+
+    #[test]
+    fn domain_with_port_is_treated_as_bare() {
+        let candidates = resolve_uri_candidates("example.com:8443");
+        assert_eq!(
+            candidates,
+            vec![
+                "https://example.com:8443/.well-known/bitcoin-cash-metadata-registry.json"
+                    .to_string(),
+                "https://example.com:8443/".to_string(),
+            ]
+        );
+    }
+}