From 1a3f3a5528b176e4d970d86fb4f22c57b7583d6d Mon Sep 17 00:00:00 2001
From: iacore <noreply+gpg-stub@1a-insec.net>
Date: Mon, 7 Oct 2024 00:45:52 +0000
Subject: [PATCH 1/2] Add check_urls plugin

It checks broken links

Works reasonably well for links from HTML pages to other assets
---
 plugins/check_urls.ts | 108 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 108 insertions(+)
 create mode 100644 plugins/check_urls.ts

diff --git a/plugins/check_urls.ts b/plugins/check_urls.ts
new file mode 100644
index 00000000..8f387680
--- /dev/null
+++ b/plugins/check_urls.ts
@@ -0,0 +1,108 @@
+import { merge } from "lume/core/utils/object.ts";
+import type Site from "lume/core/site.ts";
+import type { Page } from "lume/core/file.ts";
+
+export interface Options {
+  /** The list of extensions this plugin applies to */
+  extensions?: string[];
+}
+
+/** Default options */
+export const defaults: Options = {
+  extensions: [".html"],
+};
+
+/**
+ * This plugin checks broken links in *.html output files.
+ */
+export default function (userOptions?: Options) {
+  const options = merge(defaults, userOptions);
+
+  return (site: Site) => {
+    const url_site = site.options.location;
+    const urls = new Set<string>(); // Set is more performant than arrays
+
+    function scan(
+      url: string | null,
+      page: Page,
+      _element: Element,
+    ): undefined {
+      if (url == null) return;
+
+      const full_url = new URL(url, new URL(page.data.url, url_site));
+      if (full_url.origin != url_site.origin) {
+        return;
+      }
+      full_url.hash = '' // doesn't check hash
+      full_url.search = '' // doesn't check search either
+
+      if (!urls.has(full_url.toString())) {
+        console.warn(`⛓️‍💥 ${page.data.url} -> ${url}`);
+      }
+
+      return;
+    }
+
+    function scanSrcset(
+      attr: string | null,
+      page: Page,
+      element: Element,
+    ): undefined {
+      const srcset = attr != null ? attr.trim().split(",") : [];
+      for (const src of srcset) {
+        const [, url, _rest] = src.trim().match(/^(\S+)(.*)/)!;
+        scan(url, page, element);
+      }
+    }
+
+    site.process("*", (pages) => {
+      urls.clear(); // Clear on rebuild
+      for (const page of pages) {
+        urls.add(new URL(page.data.url, url_site).toString()); // site.url() return the full url if the second argument is true
+      }
+      for (const file of site.files) {
+        urls.add(site.url(file.outputPath, true));
+      }
+    });
+
+    site.process(
+      options.extensions,
+      (pages) =>
+        pages.forEach((page: Page) => {
+          const { document } = page;
+
+          if (!document) {
+            return;
+          }
+
+          for (const element of document.querySelectorAll("[href]")) {
+            scan(element.getAttribute("href"), page, element);
+          }
+
+          for (const element of document.querySelectorAll("[src]")) {
+            scan(element.getAttribute("src"), page, element);
+          }
+
+          for (const element of document.querySelectorAll("video[poster]")) {
+            scan(element.getAttribute("poster"), page, element);
+          }
+
+          for (const element of document.querySelectorAll("[srcset]")) {
+            scanSrcset(
+              element.getAttribute("srcset"),
+              page,
+              element,
+            );
+          }
+
+          for (const element of document.querySelectorAll("[imagesrcset]")) {
+            scanSrcset(
+              element.getAttribute("imagesrcset"),
+              page,
+              element,
+            );
+          }
+        }),
+    );
+  };
+}

From c09f28b24d9cd3e5e4331ae977ed4c94a10ca51e Mon Sep 17 00:00:00 2001
From: iacore <noreply+gpg-stub@1a-insec.net>
Date: Mon, 7 Oct 2024 00:47:50 +0000
Subject: [PATCH 2/2] Run deno fmt

---
 plugins/check_urls.ts | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/plugins/check_urls.ts b/plugins/check_urls.ts
index 8f387680..c671b185 100644
--- a/plugins/check_urls.ts
+++ b/plugins/check_urls.ts
@@ -33,8 +33,8 @@ export default function (userOptions?: Options) {
       if (full_url.origin != url_site.origin) {
         return;
       }
-      full_url.hash = '' // doesn't check hash
-      full_url.search = '' // doesn't check search either
+      full_url.hash = ""; // doesn't check hash
+      full_url.search = ""; // doesn't check search either
 
       if (!urls.has(full_url.toString())) {
         console.warn(`⛓️‍💥 ${page.data.url} -> ${url}`);