Fix C API metadata handling (#2861)

Currently, we're looking for an ID attribute starting with `group__` to
work out if a page is a module or not. This isn't very robust because
the `group__` ID only appears if there's a `doxygengroup` in the page,
which isn't always the case.

Instead, this PR treats all pages as modules except the index page and
the release notes page, which we identify by their `<h1>`.

Author: frankharkins

scripts/js/lib/api/processHtml.test.ts

@@ -16,7 +16,7 @@ import {
   addLanguageClassToCodeBlocks,
   loadImages,
   handleSphinxDesignCards,
-  maybeSetModuleMetadata,
+  maybeSetPythonModuleMetadata,
   renameAllH1s,
   removeHtmlExtensionsInRelativeLinks,
   removeDownloadSourceCode,
@@ -395,12 +395,12 @@ test("handleFootnotes()", () => {
     <span class="label"><span class="fn-bracket">[</span><a role="doc-backlink" href="#id2">1</a><span class="fn-bracket">]</span></span></aside></aside>`);
 });
 
-test.describe("maybeSetModuleMetadata()", () => {
+test.describe("maybeSetPythonModuleMetadata()", () => {
   test("not a module", () => {
     const html = `<h1>Hello</h1>`;
     const meta: Metadata = {};
     const doc = CheerioDoc.load(html);
-    maybeSetModuleMetadata(doc.$, doc.$main, meta, { isCApi: false });
+    maybeSetPythonModuleMetadata(doc.$, doc.$main, meta);
     doc.expectHtml(html);
     expect(meta).toEqual({});
   });
@@ -412,7 +412,7 @@ test.describe("maybeSetModuleMetadata()", () => {
   ): void => {
     const meta: Metadata = {};
     const doc = CheerioDoc.load(html);
-    maybeSetModuleMetadata(doc.$, doc.$main, meta, { isCApi });
+    maybeSetPythonModuleMetadata(doc.$, doc.$main, meta);
     doc.expectHtml(html);
     expect(meta).toEqual({
       apiType: "module",
@@ -441,14 +441,6 @@ test.describe("maybeSetModuleMetadata()", () => {
       "qiskit_ibm_provider.transpiler.passes.basis",
     );
   });
-
-  test("C API group", () => {
-    checkModuleFound(
-      `<section id="group__QkSparseObservable"><h1>Hello</h1></section>`,
-      "QkSparseObservable",
-      true,
-    );
-  });
 });
 
 test.describe("processMembersAndSetMeta()", () => {

scripts/js/lib/api/processHtml.ts

@@ -47,14 +47,14 @@ export async function processHtml(
     determineGithubUrl,
     releaseNotesTitle,
     hasSeparateReleaseNotes,
-    isCApi,
   } = options;
   const $ = load(html);
   const $main = $(`[role='main']`);
 
   const isReleaseNotes =
     fileName.endsWith("release_notes.html") ||
     fileName.endsWith("release-notes.html");
+  const isIndex = fileName.endsWith("index.html");
   const images = loadImages(
     $,
     $main,
@@ -84,7 +84,9 @@ export async function processHtml(
 
   const meta: Metadata = {};
   await processMembersAndSetMeta($, $main, meta, options);
-  maybeSetModuleMetadata($, $main, meta, options);
+  if (options.isCApi) maybeSetCMetadata($main, meta, isReleaseNotes, isIndex);
+  else maybeSetPythonModuleMetadata($, $main, meta);
+
   if (meta.apiType === "module") {
     updateModuleHeadings($, $main);
   }
@@ -435,13 +437,26 @@ export async function processMembersAndSetMeta(
   }
 }
 
-export function maybeSetModuleMetadata(
+function maybeSetCMetadata(
+  $main: Cheerio<any>,
+  meta: Metadata,
+  isReleaseNotes: boolean,
+  isIndex: boolean,
+): void {
+  // Every page in the C API should be displayed as a module except the index
+  // and the release notes.
+  if (isIndex || isReleaseNotes) return;
+  const topHeadingText = $main.find("h1").first().text();
+  meta.apiType = "module";
+  meta.apiName = topHeadingText;
+}
+
+export function maybeSetPythonModuleMetadata(
   $: CheerioAPI,
   $main: Cheerio<any>,
   meta: Metadata,
-  options: { isCApi: boolean },
 ): void {
-  const modulePrefix = options.isCApi ? "group__" : "module-";
+  const modulePrefix = "module-";
   const moduleIdWithPrefix = $main
     .find("span, section, div.section")
     .toArray()