fix(seo): prefix sitemap ignorePatterns with /docs/, validate HowTo steps

baseUrl is /docs/, so Docusaurus emits sitemap routes as /docs/tags/...
and /docs/1.0.0/.../docs/2.0.0/... . The previous bare patterns
(/tags/**, /1.0.0/**, /2.0.0/**) couldn't match those, so tag indexes
and the legacy doc versions were quietly slipping into the generated
sitemap despite the noIndex headers elsewhere. Add the prefixed
patterns; keep the bare ones as defence-in-depth in case baseUrl is
ever flattened.

HowTo.js was emitting steps with empty `text` and auto-generated "Step N"
names whenever authors omitted those fields, producing low-quality
HowTo structured data the rich-results test flags. Filter to entries
that carry both `name` and `text`; drop the schema entirely if none
qualify, and render the visible <ol> from the same filtered list so the
markup and JSON-LD stay in sync.

Signed-off-by: Neha Gupta <gneha21@yahoo.in>

Author: nehagup

docusaurus.config.js

@@ -472,7 +472,17 @@ module.exports = {
           // additionally carry `noIndex: true` via their `versions` config
           // above; excluding from the sitemap signals that they should not
           // be ranked at all.
+          //
+          // Docusaurus matches `ignorePatterns` against the full route path
+          // including `baseUrl` (`/docs/`), so the patterns must carry that
+          // prefix — bare `/tags/**` and `/1.0.0/**` would never match the
+          // emitted `/docs/tags/...` and `/docs/1.0.0/...` routes. Bare
+          // patterns are kept as defence-in-depth in case `baseUrl` is ever
+          // flattened to `/`.
           ignorePatterns: [
+            "/docs/tags/**",
+            "/docs/1.0.0/**",
+            "/docs/2.0.0/**",
             "/tags/**",
             "/1.0.0/**",
             "/2.0.0/**",

src/components/HowTo.js

@@ -42,16 +42,29 @@ export default function HowTo({
     return null;
   }
 
+  // Filter to steps that carry both `name` and `text` per Google's HowTo
+  // requirements. Auto-generating "Step N" placeholders or emitting empty
+  // `text` produces low-quality structured data that the rich-results test
+  // flags. If the author gave us nothing usable, drop the schema entirely
+  // rather than ship a hollow HowTo.
+  const validSteps = steps.filter(
+    (s) => typeof s.name === "string" && s.name.trim() &&
+           typeof s.text === "string" && s.text.trim(),
+  );
+  if (validSteps.length === 0) {
+    return null;
+  }
+
   const schema = {
     "@context": "https://schema.org",
     "@type": "HowTo",
     name,
-    step: steps.map((s, i) => {
+    step: validSteps.map((s, i) => {
       const step = {
         "@type": "HowToStep",
         position: i + 1,
-        name: s.name || `Step ${i + 1}`,
-        text: s.text || "",
+        name: s.name,
+        text: s.text,
       };
       if (s.url) step.url = s.url;
       if (s.image) step.image = s.image;
@@ -105,10 +118,10 @@ export default function HowTo({
                 would produce duplicate ids in the DOM whenever `visible`
                 is enabled. The list is the readable view; `step.url` in
                 the JSON-LD already covers the schema linkage. */}
-            {steps.map((s, i) => (
+            {validSteps.map((s, i) => (
               <li key={i}>
-                <strong>{s.name || `Step ${i + 1}`}</strong>
-                {s.text && <div>{s.text}</div>}
+                <strong>{s.name}</strong>
+                <div>{s.text}</div>
               </li>
             ))}
           </ol>