fix(javascript): distinguish HTML script loads from data

src/domain/javascript/htmlArtifactReferences.test.ts

@@ -30,8 +30,13 @@ it.each([
     '<svg><foreignObject><script src="actual.js"></script></foreignObject></svg>',
     ["actual.js"],
   ],
-  // The existing extraction contract does not classify script MIME types.
   ['<script type="application/json" src="actual.js"></script>', ["actual.js"]],
+  ['<script type="importmap" src="actual.js"></script>', ["actual.js"]],
+  ['<script type="speculationrules" src="actual.js"></script>', ["actual.js"]],
+  ['<script language="vbscript" src="actual.js"></script>', ["actual.js"]],
+  ['<script type=" Text/JavaScript " src="actual.js"></script>', ["actual.js"]],
+  ['<script type="MODULE" src="actual.js"></script>', ["actual.js"]],
+  ['<script type=" module " src="actual.js"></script>', ["actual.js"]],
 ])("extracts document references from %s", (text, expected) => {
   expect(
     htmlArtifactReferences(text).scripts.map(({ scriptPath }) => scriptPath),
@@ -49,6 +54,7 @@ it("retains exact opening-tag coordinates rather than the script body", () => {
     scripts: [
       {
         scriptPath: "actual.js",
+        scriptType: "classic",
         location: {
           start: { line: 3, column: 0 },
           end: { line: 4, column: ' title=">" src=actual.js>'.length },
@@ -143,3 +149,21 @@ it("counts every inline body but retains bounded locations", () => {
     },
   });
 });
+
+it.each([
+  ['type=" Text/JavaScript "', "classic"],
+  ['type="MODULE"', "module"],
+  ['type="application/json"', null],
+  ['type="importmap"', null],
+  ['type="speculationrules"', null],
+  ['language="vbscript"', null],
+  ['type=" module "', null],
+])(
+  "retains external src with parsed JavaScript type for %s",
+  (attributes, scriptType) => {
+    expect(
+      htmlArtifactReferences(`<script ${attributes} src="source.js"></script>`)
+        .scripts,
+    ).toMatchObject([{ scriptPath: "source.js", scriptType }]);
+  },
+);

src/domain/javascript/htmlArtifactReferences.ts

@@ -11,6 +11,8 @@ import type {
 
 interface HtmlScriptReference {
   readonly scriptPath: string;
+  /** Null for data blocks and other non-JavaScript script types. */
+  readonly scriptType: HtmlInlineScript["scriptType"] | null;
   /** The opening tag's range in parse5 coordinates. */
   readonly location: JavaScriptSourceRange;
 }
@@ -54,7 +56,7 @@ const JAVASCRIPT_MIME_TYPES = new Set([
  * MIME check; `module` must match the type string exactly, ignoring ASCII
  * case. Import maps, speculation rules and data blocks are not JavaScript.
  */
-const inlineScriptType = (
+const javaScriptType = (
   attrs: readonly { readonly name: string; readonly value: string }[],
 ): HtmlInlineScript["scriptType"] | null => {
   const type = attrs.find(({ name }) => name === "type");
@@ -110,11 +112,13 @@ export const htmlArtifactReferences = (
         if (href !== undefined) baseHref = href.value;
       }
       if (node.tagName === "script") {
+        const scriptType = javaScriptType(node.attrs);
         const src = node.attrs.find(({ name }) => name === "src");
         const location = node.sourceCodeLocation?.startTag;
         if (src !== undefined && src.value !== "" && location !== undefined) {
           scripts.push({
             scriptPath: src.value,
+            scriptType,
             location: {
               start: sourcePoint(location.startLine, location.startCol),
               end: sourcePoint(location.endLine, location.endCol),
@@ -122,8 +126,6 @@ export const htmlArtifactReferences = (
           });
         }
         const inline = src === undefined ? inlineScriptBody(node) : undefined;
-        const scriptType =
-          inline === undefined ? null : inlineScriptType(node.attrs);
         if (inline !== undefined && scriptType !== null) {
           inlineScriptCount += 1;
           if (inlineScripts.length < HTML_INLINE_SCRIPT_LOCATION_LIMIT)