Remove .html from internal links

src/lib/archive.ts

@@ -122,13 +122,12 @@ export function speechCategorySlug(category: string): string {
 
 export function categoryHref(entry: ArchiveEntry, category: string): string {
   if (entry.data.type === "speech") {
-    return `/speeches/categories/${speechCategorySlug(category)}.html`;
+    return `/speeches/categories/${speechCategorySlug(category)}`;
   }
 
-  return `/blog/categories/${blogCategorySlug(category)}.html`;
+  return `/blog/categories/${blogCategorySlug(category)}`;
 }
 
 export function tagHref(tag: string): string {
-  return `/blog/tags/${blogTagSlug(tag)}.html`;
+  return `/blog/tags/${blogTagSlug(tag)}`;
 }
-

src/lib/speeches.ts

@@ -1,5 +1,6 @@
 import fs from "node:fs";
 import path from "node:path";
+import { stripHtmlExtensionFromInternalUrl } from "./urls";
 
 const repoRoot = process.cwd();
 const speechesRoot = path.join(repoRoot, "content", "speeches");
@@ -190,7 +191,7 @@ function itemFromFile(filePath: string): SpeechArchiveItem {
   return {
     id: relativePath.replace(/\.md$/, ""),
     title,
-    url: frontmatter.source ?? "#",
+    url: stripHtmlExtensionFromInternalUrl(frontmatter.source) ?? "#",
     year,
     speaker,
     category,

src/lib/urls.ts

@@ -0,0 +1,39 @@
+const internalHosts = new Set(["familyfed.ie", "www.familyfed.ie", "familyfed.bcgen.ie"]);
+const ignoredProtocolPattern = /^(?:mailto|tel|javascript|data|blob):/i;
+
+function splitSuffix(value: string): { pathname: string; suffix: string } {
+  const match = value.match(/^([^?#]*)([?#].*)?$/);
+  return {
+    pathname: match?.[1] ?? value,
+    suffix: match?.[2] ?? "",
+  };
+}
+
+function stripHtmlPath(pathname: string): string {
+  return pathname.replace(/\.html$/i, "");
+}
+
+export function stripHtmlExtensionFromInternalUrl<T extends string | undefined>(value: T): T {
+  if (!value || value.startsWith("#") || value.startsWith("?") || value.startsWith("//") || ignoredProtocolPattern.test(value)) {
+    return value;
+  }
+
+  const absoluteMatch = value.match(/^(https?:\/\/([^/?#]+))([^?#]*)([?#].*)?$/i);
+  if (absoluteMatch) {
+    const [, origin, host, pathname = "", suffix = ""] = absoluteMatch;
+    if (!internalHosts.has(host.toLowerCase())) {
+      return value;
+    }
+
+    return `${origin}${stripHtmlPath(pathname)}${suffix}` as T;
+  }
+
+  const { pathname, suffix } = splitSuffix(value);
+  return `${stripHtmlPath(pathname)}${suffix}` as T;
+}
+
+export function normalizeInternalHtmlLinks(html = ""): string {
+  return html.replace(/\b(href|action)=(["'])(.*?)\2/gis, (_match, attr, quote, value) => {
+    return `${attr}=${quote}${stripHtmlExtensionFromInternalUrl(value)}${quote}`;
+  });
+}