fix: improve changeset extraction to use context-aware fallback

The final fallback pattern for extracting changesets now searches within a
~500-character context window around the version string, rather than globally
searching the entire HTML. This prevents the fallback from matching changeset
markers from unrelated versions.

This ensures each version gets paired with the correct changeset, allowing
versions to be discovered even when HTML structure varies.

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>

Author: frostebite

functions/src/logic/ingestUnityVersions/scrapeVersions.ts

@@ -55,15 +55,33 @@ export const scrapeRecentOfficialUnityVersions = async (): Promise<EditorVersion
     processedVersions.add(version);
 
     const escapedVersion = version.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
+    let changeset: string | undefined;
+
     const changesetMatch =
       new RegExp(`unityhub://${escapedVersion}/([a-f0-9]{12})`, 'i').exec(html) ||
-      new RegExp(`${escapedVersion}[^a-f0-9]*([a-f0-9]{12})`, 'i').exec(html) ||
-      /Changeset:\s*([a-f0-9]{12})/i.exec(html);
+      new RegExp(`${escapedVersion}[^a-f0-9]*([a-f0-9]{12})`, 'i').exec(html);
 
     if (changesetMatch?.[1]) {
+      changeset = changesetMatch[1];
+    } else {
+      // As a last resort, search for changeset within a context window near the version
+      // Extract ~500 chars around the version match for local context search
+      const versionIndex = html.indexOf(version);
+      if (versionIndex !== -1) {
+        const contextStart = Math.max(0, versionIndex - 200);
+        const contextEnd = Math.min(html.length, versionIndex + 300);
+        const context = html.substring(contextStart, contextEnd);
+        const contextChangesetMatch = /[Cc]hangeset:\s*([a-f0-9]{12})/i.exec(context);
+        if (contextChangesetMatch?.[1]) {
+          changeset = contextChangesetMatch[1];
+        }
+      }
+    }
+
+    if (changeset) {
       versions.push({
         version,
-        changeset: changesetMatch[1],
+        changeset,
       });
     }
   }