Fix index.html clean-URL rewrite mangling alternate-format links on index pages

news/changelog-1.10.md

@@ -60,6 +60,7 @@ All changes included in 1.10:
 - ([#13565](https://github.com/quarto-dev/quarto-cli/issues/13565), [#14353](https://github.com/quarto-dev/quarto-cli/issues/14353)): Fix sidebar logo not appearing on secondary sidebars in multi-sidebar website layouts.
 - ([#14562](https://github.com/quarto-dev/quarto-cli/issues/14562)): Fix a heading inside `content-hidden when-format="llms-txt"` (visible in HTML) losing its `<section>` wrapper and `id` in a website with `llms-txt` enabled, which broke its table-of-contents link, anchors, and cross-references.
 - ([#14563](https://github.com/quarto-dev/quarto-cli/issues/14563)): Fix a fatal error when a shortcode is used inside conditional content (e.g. `content-visible when-format="llms-txt"`) in a website with `llms-txt` enabled.
+- ([#14667](https://github.com/quarto-dev/quarto-cli/issues/14667)): Fix the clean-URL rewrite in `quarto-nav.js` breaking links to files whose names merely start with `index.html`, such as the `index.html.md` "Other Formats" links produced when `output-file: index.html` is paired with a markdown format (e.g. by nbdev for llms.txt workflows). The rewrite now only strips `index.html` at the end of the path.
 
 ## Commands
 

src/project/types/website/website-shared.ts

@@ -474,5 +474,5 @@ export function containsHref(href: string, items: SidebarItem[]) {
 
 export function itemHasNavTarget(item: SidebarItem, href: string) {
   return item.href === href ||
-    item.href === href.replace(/\/index\.html/, "/");
+    item.href === href.replace(/\/index\.html(?=[?#]|$)/, "/");
 }

src/resources/projects/website/navigation/quarto-nav.js

@@ -274,7 +274,10 @@ window.document.addEventListener("DOMContentLoaded", function () {
     for (let i = 0; i < links.length; i++) {
       if (links[i].href) {
         links[i].dataset.originalHref = links[i].href;
-        links[i].href = links[i].href.replace(/\/index\.html/, "/");
+        // only strip index.html at the end of the path (optionally followed
+        // by a query or fragment) so that files like index.html.md
+        // (alternate-format twins) are left alone (#14667)
+        links[i].href = links[i].href.replace(/\/index\.html(?=[?#]|$)/, "/");
       }
     }
 

src/webui/quarto-preview/src/server/navigation.ts

@@ -8,7 +8,9 @@ export function navigationHandler() {
 
   return () => {
     const normalizeTarget = (target: string) => {
-      return target.replace(/\/index\.html/, "/")
+      // only strip index.html at the end of the path so that files like
+      // index.html.md (alternate-format twins) are left alone (#14667)
+      return target.replace(/\/index\.html(?=[?#]|$)/, "/")
     };
 
     return (ev: MessageEvent<string>): boolean => {

tests/docs/playwright/website/issue-14667/.gitignore

@@ -0,0 +1,2 @@
+/.quarto/
+/_site/

tests/docs/playwright/website/issue-14667/_quarto.yml

@@ -0,0 +1,7 @@
+project:
+  type: website
+
+website:
+  title: "issue-14667"
+
+format: html

tests/docs/playwright/website/issue-14667/index.qmd

@@ -0,0 +1,18 @@
+---
+title: Home
+---
+
+Links exercising the clean-URL rewrite in `quarto-nav.js` (issue #14667).
+The rewrite must strip `index.html` only at the end of the path, leaving
+names that merely start with `index.html` (like the `index.html.md`
+markdown twins generated for llms.txt workflows) untouched.
+
+- [markdown twin](index.html.md){#lnk-md} — must NOT be rewritten
+- [clean url](sub/index.html){#lnk-clean} — must become `sub/`
+- [fragment](index.html#section){#lnk-frag} — must become `#section` on `./`
+- [query](index.html?a=1){#lnk-query} — must become `./?a=1`
+- [path segment](index.html/foo){#lnk-segment} — must NOT be rewritten
+
+## Section {#section}
+
+Target for the fragment link.

tests/docs/playwright/website/issue-14667/sub/index.qmd

@@ -0,0 +1,5 @@
+---
+title: Sub page
+---
+
+Target page for the clean-URL link on the home page.

tests/integration/playwright/tests/website-index-html-links.spec.ts

@@ -0,0 +1,27 @@
+import { expect, test } from "@playwright/test";
+
+// quarto-nav.js rewrites every link on page load to clean up index.html
+// URLs. The rewrite must only strip index.html at the end of the path
+// (optionally followed by a query or fragment); hrefs that merely start
+// with index.html — like the index.html.md markdown twins generated for
+// llms.txt workflows — must be left alone (#14667).
+test("index.html clean-URL rewrite only strips whole path segments", async ({ page }) => {
+  await page.goto("./website/issue-14667/_site/index.html");
+
+  // data-original-href is stamped by the rewrite loop, so waiting on it
+  // ensures quarto-nav.js has processed the links
+  const mdLink = page.locator("#lnk-md");
+  await expect(mdLink).toHaveAttribute("data-original-href", /\/index\.html\.md$/);
+
+  // a link to an index.html-prefixed file must not be rewritten
+  await expect(mdLink).toHaveAttribute("href", /\/index\.html\.md$/);
+
+  // ...while genuine index.html links still get cleaned up
+  await expect(page.locator("#lnk-clean")).toHaveAttribute("href", /\/sub\/$/);
+  await expect(page.locator("#lnk-frag")).toHaveAttribute("href", /\/_site\/#section$/);
+  await expect(page.locator("#lnk-query")).toHaveAttribute("href", /\/_site\/\?a=1$/);
+
+  // index.html as an intermediate path segment is not the index page of
+  // the link's directory, so it must be left alone too
+  await expect(page.locator("#lnk-segment")).toHaveAttribute("href", /\/index\.html\/foo$/);
+});