deslop(#610): check-built-links member check — drop no-op BASE escaping, early-continue over one sitePath, collapse nested broken-set if

Co-authored-by: Zack Jackson <ScriptedAlchemy@users.noreply.github.com>

Author: ScriptedAlchemy

website/scripts/check-built-links.mjs

@@ -29,7 +29,7 @@ const ATTRIBUTE = /([a-zA-Z:-]+)\s*=\s*"([^"]*)"/g;
 const ID = /\s(?:id|name)="([^"]*)"/g;
 const SKIPPED_SCHEMES = /^(?:mailto:|javascript:|tel:|data:|https?:\/\/|\/\/)/i;
 const ANCHOR_ELEMENT = /<a\b([^>]*)>([\s\S]*?)<\/a>/gi;
-const API_MEMBER_LINK = new RegExp(`^${BASE.replace(/[/]/g, '\\/')}(?:zh\\/)?api\\/[^#?]+#(.+)$`);
+const API_MEMBER_LINK = new RegExp(`^${BASE}(?:zh/)?api/[^#?]+#(.+)$`);
 const IDENTIFIER = /^[A-Za-z_$][\w$]*$/;
 /** Visible text of an HTML fragment, as a reader compares it: tags (incl. `<wbr/>`) and entities gone, `()` dropped. */
 const visibleText = html => unescapeHtml(html.replace(/<[^>]*>/g, '')).replace(/\(\)$/, '').trim();
@@ -141,18 +141,20 @@ const main = () => {
     if (file.endsWith('.html') && !/^\/(?:zh\/)?api\//.test(pagePath)) {
       for (const [, rawAttributes, inner] of html.matchAll(ANCHOR_ELEMENT)) {
         const href = unescapeHtml(/\shref="([^"]*)"/.exec(rawAttributes)?.[1] ?? '');
-        const member = API_MEMBER_LINK.exec(href.startsWith(ORIGIN) ? href.slice(ORIGIN.length) : href);
-        const label = member ? visibleText(inner) : '';
-        if (!member || !IDENTIFIER.test(label)) continue;
-        const target = resolveTarget(href.replace(/#.*$/, '').replace(ORIGIN, ''));
+        const sitePath = href.startsWith(ORIGIN) ? href.slice(ORIGIN.length) : href;
+        const member = API_MEMBER_LINK.exec(sitePath);
+        if (!member) continue;
+        const label = visibleText(inner);
+        if (!IDENTIFIER.test(label)) continue;
+        const target = resolveTarget(sitePath.replace(/#.*$/, ''));
         if (target === null) continue;
         members += 1;
         const fragment = decode(member[1]);
         const heading = new RegExp(`<h[1-6]\\b[^>]*\\sid="${fragment.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')}"[^>]*>([\\s\\S]*?)</h[1-6]>`).exec(fs.readFileSync(target, 'utf8'));
         if (heading === null) continue;
         const headingLabel = visibleText(heading[1]).replace(/^#/, '').trim();
-        if (headingLabel.toLowerCase() !== label.toLowerCase()) {
-          if (!broken.has(href)) broken.set(href, `${href} — labelled "${label}" but #${fragment} is the heading "${headingLabel}" (e.g. in ${pagePath})`);
+        if (headingLabel.toLowerCase() !== label.toLowerCase() && !broken.has(href)) {
+          broken.set(href, `${href} — labelled "${label}" but #${fragment} is the heading "${headingLabel}" (e.g. in ${pagePath})`);
         }
       }
     }