Rewrite WordPress cross-page links to site-relative paths

Content linked between pages using absolute WordPress URLs -- 100
occurrences across 39 files, many against wp.vicidial.com, the old
WordPress backend host that won't outlive this migration, the rest
against www.vicidial.com in ?page_id=N form.

The export now resolves those IDs against the published page/post set and
emits /slug (or /blog/slug) instead. All 100 resolved; links to
vicihost.com are a separate site and deliberately untouched.

Verified every internal href in the built output corresponds to a page
that actually exists -- no broken links.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

scripts/export-wp.mjs

@@ -82,6 +82,32 @@ function rewriteImageUrls(html, fromDir) {
   );
 }
 
+// Content links between pages are stored as absolute WordPress URLs, many of them against
+// wp.vicidial.com -- the old WordPress backend host, which won't outlive this migration.
+// Turn them into site-relative paths. IDs with no published page/post behind them (deleted
+// pages, a stray revision) are left alone: they are already broken on the live site.
+function rewriteInternalLinks(html, linkMap, unresolved) {
+  return html.replace(
+    /https?:\/\/(?:www\.|wp\.)?vicidial\.(?:com|net)\/\?(?:page_id|p)=(\d+)/gi,
+    (match, id) => {
+      const target = linkMap.get(Number(id));
+      if (!target) {
+        unresolved.add(id);
+        return match;
+      }
+      return target;
+    }
+  );
+}
+
+async function buildLinkMap(conn) {
+  const [rows] = await conn.execute(
+    `SELECT ID, post_type, post_name FROM ${TABLE_PREFIX}posts
+     WHERE post_status = 'publish' AND post_type IN ('page', 'post')`
+  );
+  return new Map(rows.map((r) => [r.ID, r.post_type === 'post' ? `/blog/${r.post_name}` : `/${r.post_name}`]));
+}
+
 async function copyAttachments() {
   // Copy the whole uploads tree rather than just the 82 attachment rows: WordPress generates
   // multiple resized variants per image (e.g. "-1008x1024.png") that content <img> tags/srcsets
@@ -91,7 +117,7 @@ async function copyAttachments() {
   console.log(`Copied uploads tree to ${UPLOADS_DEST}`);
 }
 
-async function exportPostType(conn, postType, destDir, { attachmentPaths, thumbnails }) {
+async function exportPostType(conn, postType, destDir, { attachmentPaths, thumbnails, linkMap, unresolved }) {
   const [rows] = await conn.execute(
     `SELECT ID, post_title, post_name, post_date, post_content, post_excerpt
      FROM ${TABLE_PREFIX}posts WHERE post_type = ? AND post_status = 'publish' ORDER BY post_date DESC`,
@@ -107,6 +133,7 @@ async function exportPostType(conn, postType, destDir, { attachmentPaths, thumbn
       html = wpautop(html);
     }
     html = rewriteImageUrls(html, destDir);
+    html = rewriteInternalLinks(html, linkMap, unresolved);
     const markdown = turndown.turndown(html);
 
     let excerpt = fixLiteralNewlines(row.post_excerpt || '');
@@ -174,8 +201,15 @@ async function main() {
   const thumbnails = new Map(thumbRows.map((r) => [r.post_id, r.meta_value]));
 
   await copyAttachments();
-  await exportPostType(conn, 'page', PAGES_DEST, { attachmentPaths, thumbnails });
-  await exportPostType(conn, 'post', POSTS_DEST, { attachmentPaths, thumbnails });
+  const linkMap = await buildLinkMap(conn);
+  const unresolved = new Set();
+
+  await exportPostType(conn, 'page', PAGES_DEST, { attachmentPaths, thumbnails, linkMap, unresolved });
+  await exportPostType(conn, 'post', POSTS_DEST, { attachmentPaths, thumbnails, linkMap, unresolved });
+
+  if (unresolved.size) {
+    console.warn(`Left ${unresolved.size} link(s) pointing at WordPress IDs with no published page: ${[...unresolved].join(', ')}`);
+  }
   await exportNav(conn);
 
   await conn.end();