fix: Correctly extract page titles when not specified in frontmatter, c…

…loses #146 Co-authored-by: Joe Kleinschmidt <joe@Joes-MacBook-Pro-2.local> Co-authored-by: Christian Flach <cmfcmf.flach@gmail.com> Fixes #146
cmfcmf · Dec 22, 2022 · 1f33151 · 1f33151
1 parent f169f9a
commit 1f33151
Show file tree

Hide file tree

Showing 6 changed files with 107 additions and 3 deletions.
diff --git a/packages/docusaurus-search-local/src/server/parse.test.js b/packages/docusaurus-search-local/src/server/parse.test.js
@@ -116,6 +116,82 @@ describe("parser", () => {
       });
     });
 
+    it("parses page without title in frontmatter and with h1 in Markdown", async () => {
+      const htmlPath = path.join(
+        __dirname,
+        "..",
+        "..",
+        "..",
+        "example-docs/build/docs/next/d-s-l-test-no-title-h1/index.html"
+      );
+      const html = await readFileAsync(htmlPath, "utf-8");
+      expect(html2text(html, "docs")).toEqual({
+        docSidebarParentCategories: ["Docusaurus"],
+        pageTitle: "Page title from markdown",
+        sections: [
+          {
+            content: "bla bla",
+            hash: "",
+            title: "Page title from markdown",
+            tags: [],
+          },
+        ],
+      });
+    });
+
+    it("parses page without title in frontmatter and with h2 in Markdown", async () => {
+      const htmlPath = path.join(
+        __dirname,
+        "..",
+        "..",
+        "..",
+        "example-docs/build/docs/next/d-s-l-test-no-title-h2/index.html"
+      );
+      const html = await readFileAsync(htmlPath, "utf-8");
+      expect(html2text(html, "docs")).toEqual({
+        docSidebarParentCategories: ["Docusaurus"],
+        pageTitle: "d-s-l-test-no-title-h2",
+        sections: [
+          { content: "", hash: "", title: "d-s-l-test-no-title-h2", tags: [] },
+          {
+            content: "bla bla",
+            hash: "#page-title-from-markdown",
+            title: "Page title from markdown",
+            tags: [],
+          },
+        ],
+      });
+    });
+
+    it("parses page without title in frontmatter and with h1 in Markdown but with pre text", async () => {
+      const htmlPath = path.join(
+        __dirname,
+        "..",
+        "..",
+        "..",
+        "example-docs/build/docs/next/d-s-l-test-no-title-h1-pre-text/index.html"
+      );
+      const html = await readFileAsync(htmlPath, "utf-8");
+      expect(html2text(html, "docs")).toEqual({
+        docSidebarParentCategories: ["Docusaurus"],
+        pageTitle: "d-s-l-test-no-title-h1-pre-text",
+        sections: [
+          {
+            content: "some pre text before the actual page title",
+            hash: "",
+            title: "d-s-l-test-no-title-h1-pre-text",
+            tags: [],
+          },
+          {
+            content: "bla bla",
+            hash: "",
+            title: "Page title from markdown",
+            tags: [],
+          },
+        ],
+      });
+    });
+
     it("parses nested sidebar categories", async () => {
       const htmlPath = path.join(
         __dirname,

diff --git a/packages/docusaurus-search-local/src/server/parse.ts b/packages/docusaurus-search-local/src/server/parse.ts
@@ -97,8 +97,7 @@ export function html2text(
 
   if (type === "docs" || type === "blog") {
     const HEADINGS = "h1, h2, h3";
-    const pageTitle = $("article header h1").first().text();
-
+    const pageTitle = $("article h1").first().text();
     const sections: Array<{
       title: string;
       hash: string;

diff --git a/packages/example-docs/docs/d-s-l-test-no-title-h1-pre-text.md b/packages/example-docs/docs/d-s-l-test-no-title-h1-pre-text.md
@@ -0,0 +1,11 @@
+---
+id: d-s-l-test-no-title-h1-pre-text
+# title: DOC TITLE # Deliberately no title here
+sidebar_label: SIDEBAR LABEL
+---
+
+some pre text before the actual page title
+
+# Page title from markdown
+
+bla bla
diff --git a/packages/example-docs/docs/d-s-l-test-no-title-h1.md b/packages/example-docs/docs/d-s-l-test-no-title-h1.md
@@ -0,0 +1,9 @@
+---
+id: d-s-l-test-no-title-h1
+# title: DOC TITLE # Deliberately no title here
+sidebar_label: SIDEBAR LABEL
+---
+
+# Page title from markdown
+
+bla bla
diff --git a/packages/example-docs/docs/d-s-l-test-no-title-h2.md b/packages/example-docs/docs/d-s-l-test-no-title-h2.md
@@ -0,0 +1,9 @@
+---
+id: d-s-l-test-no-title-h2
+# title: DOC TITLE # Deliberately no title here
+sidebar_label: SIDEBAR LABEL
+---
+
+## Page title from markdown
+
+bla bla
diff --git a/packages/example-docs/sidebars.js b/packages/example-docs/sidebars.js
@@ -11,7 +11,7 @@
 
 module.exports = {
   someSidebar: {
-    Docusaurus: ['d-s-l-test', 'd-s-l-test2', 'd-s-l-test3', 'translated'],
+    Docusaurus: ['d-s-l-test', 'd-s-l-test2', 'd-s-l-test3', 'd-s-l-test-no-title-h1-pre-text', 'd-s-l-test-no-title-h1', 'd-s-l-test-no-title-h2', 'translated'],
     SidebarParent: [
       {
         SidebarChild: ['nested_sidebar_doc']