docs: fix extra HTML end tags in docs build (adobe#9477)

* add temp script to fail build when errors in build detected

* lint

* try patch

* try another build

* make validation function more generic

* cleanup script

Author: reidbarber

Makefile

@@ -154,16 +154,19 @@ s2-docs:
 	DOCS_ENV=stage PUBLIC_URL=/$(BRANCH_TYPE)/$(HASH) $(MAKE) build-s2-docs
 	cp packages/dev/docs/pages/disallow-robots.txt dist/s2-docs/react-aria/$(BRANCH_TYPE)/$(HASH)/robots.txt
 	cp packages/dev/docs/pages/disallow-robots.txt dist/s2-docs/s2/$(BRANCH_TYPE)/$(HASH)/robots.txt
+	yarn check:s2-docs-build dist/s2-docs
 
 s2-docs-stage:
 	DOCS_ENV=stage PUBLIC_URL=/ $(MAKE) build-s2-docs
 	cp packages/dev/docs/pages/disallow-robots.txt dist/s2-docs/react-aria/robots.txt
 	cp packages/dev/docs/pages/disallow-robots.txt dist/s2-docs/s2/robots.txt
+	yarn check:s2-docs-build dist/s2-docs
 
 s2-docs-production:
 	DOCS_ENV=prod PUBLIC_URL=/ $(MAKE) build-s2-docs
 	cp packages/dev/docs/pages/robots.txt dist/s2-docs/react-aria/robots.txt
 	cp packages/dev/docs/pages/robots.txt dist/s2-docs/s2/robots.txt
+	yarn check:s2-docs-build dist/s2-docs
 	cd starters/docs && yarn install --no-immutable && yarn up react-aria-components
 	cd starters/tailwind && yarn install --no-immutable && yarn up react-aria-components tailwindcss-react-aria-components
 	$(MAKE) build-starters

package.json

@@ -31,6 +31,7 @@
     "build:docs": "DOCS_ENV=staging parcel build 'packages/@react-{spectrum,aria,stately}/*/docs/*.mdx' 'packages/dev/docs/pages/{react-spectrum,releases}/**/*.mdx'",
     "start:s2-docs": "yarn workspace @react-spectrum/s2-docs start",
     "build:s2-docs": "yarn workspace @react-spectrum/s2-docs build",
+    "check:s2-docs-build": "node packages/dev/s2-docs/scripts/validateS2DocsBuild.mjs",
     "build:mcp": "yarn workspace @react-spectrum/mcp build && yarn workspace @react-aria/mcp build",
     "start:mcp": "yarn workspace @react-spectrum/s2-docs generate:md && yarn build:mcp && node packages/dev/mcp/s2/dist/index.js && node packages/dev/mcp/react-aria/dist/index.js",
     "test:mcp": "yarn build:s2-docs && yarn build:mcp && node packages/dev/mcp/scripts/smoke-list-pages.mjs",

packages/dev/s2-docs/scripts/validateS2DocsBuild.mjs

@@ -0,0 +1,108 @@
+#!/usr/bin/env node
+/**
+ * Script to validate s2-docs build output.
+ * - Confirms the build directory exists and contains files.
+ * - Checks for duplicate </script></body></html> occurrences in HTML files.
+ *
+ * Usage: node scripts/validateS2DocsBuild.mjs [directory]
+ * Default directory: ./dist
+ */
+
+import {dirname, join} from 'path';
+import fg from 'fast-glob';
+import {fileURLToPath} from 'url';
+import {readFile, stat} from 'fs/promises';
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = dirname(__filename);
+
+const PATTERN = '</script></body></html>';
+
+async function collectBuildFiles(dir) {
+  const files = await fg(['**/*'], {
+    cwd: dir,
+    onlyFiles: true,
+    dot: true,
+    absolute: true
+  });
+
+  const htmlFiles = files.filter(filePath => filePath.endsWith('.html'));
+  return {htmlFiles, totalFiles: files.length};
+}
+
+async function checkFile(filePath) {
+  const content = await readFile(filePath, 'utf-8');
+  
+  // Count occurrences of the pattern
+  const matches = content.match(new RegExp(PATTERN.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'), 'g'));
+  const count = matches ? matches.length : 0;
+  
+  return {
+    filePath,
+    count,
+    hasDuplicate: count > 1
+  };
+}
+
+async function main() {
+  const targetDir = process.argv[2] || join(__dirname, '..', 'dist');
+  
+  console.log('\nValidating s2-docs build output...');
+  console.log(`Directory: ${targetDir}\n`);
+
+  let dirStats;
+  try {
+    dirStats = await stat(targetDir);
+  } catch {
+    console.error(`Build directory does not exist: ${targetDir}`);
+    process.exit(1);
+  }
+
+  if (!dirStats.isDirectory()) {
+    console.error(`Build path is not a directory: ${targetDir}`);
+    process.exit(1);
+  }
+
+  let htmlFiles;
+  let totalFiles;
+  try {
+    ({htmlFiles, totalFiles} = await collectBuildFiles(targetDir));
+  } catch (error) {
+    console.error(`Error reading build directory: ${error.message}`);
+    process.exit(1);
+  }
+
+  if (totalFiles === 0) {
+    console.error('Build directory is empty. No files found.');
+    process.exit(1);
+  }
+
+  if (htmlFiles.length === 0) {
+    console.error('No HTML files found in the build output.');
+    process.exit(1);
+  }
+
+  console.log(`Found ${htmlFiles.length} HTML files to check.\n`);
+  
+  const results = await Promise.all(htmlFiles.map(checkFile));
+  const duplicates = results.filter(r => r.hasDuplicate);
+  
+  if (duplicates.length === 0) {
+    console.log('✅ All HTML validated for duplicate \'</script></body></html>\' occurrences.');
+  } else {
+    console.log(`❌ Found ${duplicates.length} file(s) with duplicate '</script></body></html>' occurrences:\n`);
+    
+    for (const {filePath, count} of duplicates) {
+      const relativePath = filePath.replace(targetDir, '').replace(/^\//, '');
+      console.log(`   📄 ${relativePath}`);
+      console.log(`      Pattern appears ${count} times (expected: 1)\n`);
+    }
+    
+    process.exit(1);
+  }
+}
+
+main().catch(error => {
+  console.error('Fatal error:', error);
+  process.exit(1);
+});

patches/rsc-html-stream+0.0.6.patch

@@ -0,0 +1,53 @@
+diff --git a/node_modules/rsc-html-stream/server.js b/node_modules/rsc-html-stream/server.js
+index 1234567..abcdefg 100644
+--- a/node_modules/rsc-html-stream/server.js
++++ b/node_modules/rsc-html-stream/server.js
+@@ -14,22 +14,22 @@ export function injectRSCPayload(rscStream, options) {
+   let buffered = [];
+   let timeout = null;
+   function flushBufferedChunks(controller) {
++    // Decode all buffered chunks together so we can reliably detect a trailer that
++    // might be split across chunk boundaries.
++    let combined = '';
+     for (let chunk of buffered) {
+-      let buf = decoder.decode(chunk, {stream: true});
+-      if (buf.endsWith(trailer)) {
+-        buf = buf.slice(0, -trailer.length);
+-      }
+-      controller.enqueue(encoder.encode(buf));
++      combined += decoder.decode(chunk, {stream: true});
+     }
++    combined += decoder.decode();
+ 
+-    let remaining = decoder.decode();
+-    if (remaining.length) {
+-      if (remaining.endsWith(trailer)) {
+-        remaining = remaining.slice(0, -trailer.length);
+-      }
+-      controller.enqueue(encoder.encode(remaining));
++    if (combined.endsWith(trailer)) {
++      combined = combined.slice(0, -trailer.length);
+     }
+ 
++    if (combined.length) {
++      controller.enqueue(encoder.encode(combined));
++    }
++
+     buffered.length = 0;
+     timeout = null;
+   }
+@@ -42,7 +42,13 @@ export function injectRSCPayload(rscStream, options) {
+       }
+ 
+       timeout = setTimeout(async () => {
+-        flushBufferedChunks(controller);
++        try {
++          flushBufferedChunks(controller);
++        } catch (e) {
++          controller.error(e);
++          resolveFlightDataPromise();
++          return;
++        }
+         if (!startedRSC) {
+           startedRSC = true;
+           writeRSCStream(rscStream, controller, nonce)