fix(preload): preserve modulepreload scanner order

moli-renderer-v8/src/modulepreload.rs

@@ -64,7 +64,13 @@ pub(crate) fn modulepreload_href(element: &Element) -> Option<&str> {
 }
 
 pub(crate) fn modulepreload_as_state(element: &Element) -> ModulepreloadAsState {
-    match link_as_destination(element.attribute("as")) {
+    modulepreload_as_state_from_attribute(element.attribute("as"))
+}
+
+pub(crate) fn modulepreload_as_state_from_attribute(
+    as_attribute: Option<&str>,
+) -> ModulepreloadAsState {
+    match link_as_destination(as_attribute) {
         LinkAsDestination::None
         | LinkAsDestination::Script
         | LinkAsDestination::AudioWorklet
@@ -126,7 +132,11 @@ fn modulepreload_attributes_for_state(state: ModulepreloadAsState) -> Option<Mod
 }
 
 pub(crate) fn modulepreload_media_matches(element: &Element) -> bool {
-    let Some(media) = element.attribute("media").map(str::trim) else {
+    modulepreload_media_attribute_matches(element.attribute("media"))
+}
+
+pub(crate) fn modulepreload_media_attribute_matches(media: Option<&str>) -> bool {
+    let Some(media) = media.map(str::trim) else {
         return true;
     };
     if media.is_empty() {

moli-renderer-v8/src/runtime/phase_one/tests/extracted/preload_scanning_and_csp.rs

@@ -134,6 +134,7 @@ fn html_preload_scanner_stops_module_preloads_after_import_map() {
         r#"
                 <script type="module" src="/before.mjs"></script>
                 <script type="importmap">{"integrity": {}}</script>
+                <link rel="modulepreload" href="/after-preload.mjs">
                 <script type="module" src="/after.mjs"></script>
                 <script src="/classic.js"></script>
             "#,
@@ -1975,3 +1976,23 @@ fn buffered_preload_scanner_clears_nonnonceable_script_nonces() {
         "speculative requests must use the same nonceability gate as parser execution"
     );
 }
+
+#[test]
+fn html_preload_scanner_preserves_modulepreload_before_module_script() {
+    let final_url = Url::parse("https://example.test/docs/page.html").expect("test url");
+    let requests = collect_preloadable_external_script_requests_from_html(
+        &final_url,
+        r#"
+                <link rel="modulepreload" href="/entry.mjs" integrity="sha384-invalid">
+                <script type="module" src="/entry.mjs"></script>
+            "#,
+    );
+
+    assert_eq!(requests.len(), 2);
+    assert_eq!(requests[0].url, requests[1].url);
+    assert_eq!(
+        requests[0].fetch_metadata.integrity.as_deref(),
+        Some("sha384-invalid")
+    );
+    assert_eq!(requests[1].fetch_metadata.integrity, None);
+}

moli-renderer-v8/src/runtime/phase_one/tests/extracted/streaming_input.rs

@@ -28,27 +28,7 @@ fn buffered_html_preload_scan_collects_future_external_scripts_without_importmap
         ]
     );
 }
-#[test]
-fn buffered_html_preload_scan_leaves_modulepreload_to_native_module_map() {
-    let final_url = Url::parse("https://example.test/docs/page.html").expect("test url");
-    let requests = collect_preloadable_external_script_requests_from_html(
-        &final_url,
-        r#"
-                <link rel="dns-prefetch modulepreload" href="/entry.mjs">
-                <link rel="MODULEPRELOAD" href="/entry.mjs">
-                <link rel="preload" as="script" href="/classic.js">
-                <link rel="modulepreload" as="style" href="/theme.css">
-                <link rel="modulepreload" href="/theme.css?version=1">
-                <link rel="modulepreload" href="data:text/javascript,export%20default%201">
-            "#,
-    );
 
-    assert_eq!(
-        requests,
-        Vec::new(),
-        "modulepreload must not enter the legacy script-text preload cache; the parser publishes exact link candidates to the native module map"
-    );
-}
 #[test]
 fn incremental_html_preload_scanner_handles_split_script_tag_boundaries() {
     let final_url = Url::parse("https://example.test/docs/page.html").expect("test url");
@@ -432,3 +412,35 @@ fn full_body_phase_one_parks_async_subresource_terminal_for_page_owner() {
         full_body_phase_one_parks_async_subresource_terminal_for_page_owner_inner,
     );
 }
+
+#[test]
+fn buffered_html_preload_scan_routes_script_like_modulepreloads_to_native_module_map() {
+    let final_url = Url::parse("https://example.test/docs/page.html").expect("test url");
+    let requests = collect_preloadable_external_script_requests_from_html(
+        &final_url,
+        r#"
+                <link rel="dns-prefetch modulepreload" href="/entry.mjs">
+                <link rel="MODULEPRELOAD" href="/entry.mjs">
+                <link rel="preload" as="script" href="/classic.js">
+                <link rel="modulepreload" as="style" href="/theme.css">
+                <link rel="modulepreload" href="/theme.css?version=1">
+                <link rel="modulepreload" href="data:text/javascript,export%20default%201">
+            "#,
+    );
+
+    assert!(
+        requests
+            .iter()
+            .all(|request| request.kind_hint == crate::types::ScriptKind::Module),
+        "scanner modulepreloads must take the native module-map admission path"
+    );
+    assert_eq!(
+        preload_request_urls(requests),
+        vec![
+            Url::parse("https://example.test/entry.mjs").expect("entry module URL"),
+            Url::parse("https://example.test/theme.css?version=1")
+                .expect("default script-like module URL"),
+        ],
+        "the scanner should dedupe script-like modulepreloads while leaving typed and data candidates to the parser"
+    );
+}

moli-renderer-v8/src/runtime/script_preloads.rs

@@ -15,6 +15,10 @@ use url::Url;
 use moli_script::script_element_nonce_is_nonceable;
 use parking_lot::Mutex;
 
+use crate::modulepreload::{
+    ModulepreloadAsState, modulepreload_as_state_from_attribute,
+    modulepreload_media_attribute_matches, resolve_parser_network_resource_url,
+};
 use crate::network::ResourceRequestClient;
 use crate::page_task_queue::RendererOwnerWakeSender;
 use crate::parser::{PreparedScript, ScriptSource};
@@ -1410,6 +1414,58 @@ impl HtmlPreloadScannerSink {
         }
     }
 
+    fn maybe_collect_modulepreload(&self, tag: &Tag) {
+        if self.seen_import_map.get()
+            || (matches!(self.meta_csp_mode, MetaCspScannerMode::StopAfterMeta)
+                && self.seen_meta_csp_count.get() != 0)
+        {
+            return;
+        }
+        let Some(rel) = html_attr_value(&tag.attrs, "rel") else {
+            return;
+        };
+        if !link_rel_includes_token(&rel, "modulepreload")
+            || modulepreload_as_state_from_attribute(html_attr_value(&tag.attrs, "as").as_deref())
+                != ModulepreloadAsState::ScriptLike
+            || !modulepreload_media_attribute_matches(
+                html_attr_value(&tag.attrs, "media").as_deref(),
+            )
+        {
+            return;
+        }
+        let Some(href) = html_attr_value(&tag.attrs, "href") else {
+            return;
+        };
+        let Some(url) = resolve_parser_network_resource_url(&self.base_url.borrow(), &href) else {
+            return;
+        };
+        let kind_hint = crate::types::ScriptKind::Module;
+        let mode_hint = crate::types::ScriptMode::ModuleInOrder;
+        let fetch_metadata = crate::planning::ScriptFetchMetadata::from_script_attributes(
+            html_attr_value(&tag.attrs, "crossorigin").as_deref(),
+            html_attr_value(&tag.attrs, "referrerpolicy").as_deref(),
+            None,
+            html_attr_value(&tag.attrs, "integrity").as_deref(),
+            html_attr_value(&tag.attrs, "nonce").as_deref(),
+            html_attr_value(&tag.attrs, "fetchpriority").as_deref(),
+        );
+        let Some(key) = BufferedScriptPreloadKey::new(url.clone(), kind_hint, &fetch_metadata)
+        else {
+            return;
+        };
+        let mut requests = self.requests.borrow_mut();
+        if requests.seen.insert(BufferedPreloadKey::Script(key)) {
+            requests.script_requests.push(BufferedScriptPreloadRequest {
+                url,
+                initiator_url: self.final_url.clone(),
+                kind_hint,
+                mode_hint,
+                resource_type_hint: moli_fetch::RequestResourceType::Script,
+                fetch_metadata,
+            });
+        }
+    }
+
     fn maybe_update_base_url(&self, tag: &Tag) {
         if self.has_valid_base.get() {
             return;
@@ -1630,6 +1686,7 @@ impl TokenSink for HtmlPreloadScannerSink {
                 TokenSinkResult::RawData(ScriptData)
             }
             "link" => {
+                self.maybe_collect_modulepreload(&tag);
                 self.maybe_collect_stylesheet_preload(&tag);
                 TokenSinkResult::Continue
             }