Merge branch 'main' into copilot/add-centralized-error-boundary

Author: pethers

scripts/article-template.ts

@@ -17,4 +17,4 @@
  * @author Hack23 AB
  * @license Apache-2.0
  */
-export { generateArticleHTML, generateEventCalendar, generateWatchSection, generateArticleLanguageSwitcher, generateSiteFooter, default } from './article-template/index.js';
+export { generateArticleHTML, generateEventCalendar, generateWatchSection, generateArticleLanguageSwitcher, generateSiteFooter, fixHtmlNesting, default } from './article-template/index.js';

scripts/article-template/helpers.ts

@@ -46,9 +46,22 @@ export function getNewsIndexFilename(lang: Language | string): string {
   return `index_${lang}.html`;
 }
 
+/**
+ * Remove orphaned `</p>` tags that appear immediately after `</ul>` or `</ol>`.
+ * Browsers auto-close `<p>` before block-level list elements, so AI-generated
+ * markup of the form `<p>intro</p><ul>…</ul></p>` leaves a dangling `</p>`.
+ * This function removes only that specific trailing `</p>` and does not attempt
+ * any other HTML repair.
+ */
+export function fixHtmlNesting(htmlContent: string): string {
+  return htmlContent.replace(/<\/(ul|ol)>\s*<\/p>/g, '</$1>');
+}
+
 /**
  * Sanitize article body content for JSON-LD structured data.
  * Removes newlines and normalizes whitespace to prevent invalid JSON.
+ * Callers should apply {@link fixHtmlNesting} to the raw HTML *before*
+ * escaping and passing it here, so the regex has a chance to match.
  */
 export function sanitizeArticleBody(htmlContent: string): string {
   return htmlContent

scripts/article-template/index.ts

@@ -15,7 +15,7 @@
  */
 
 export { generateArticleHTML } from './template.js';
-export { generateEventCalendar, generateWatchSection, generateArticleLanguageSwitcher, generateSiteFooter } from './helpers.js';
+export { generateEventCalendar, generateWatchSection, generateArticleLanguageSwitcher, generateSiteFooter, fixHtmlNesting } from './helpers.js';
 
 import { generateArticleHTML } from './template.js';
 import { generateEventCalendar, generateWatchSection, generateArticleLanguageSwitcher, generateSiteFooter } from './helpers.js';

scripts/article-template/template.ts

@@ -18,6 +18,7 @@ import {
   getFooterLabel,
   getNewsIndexFilename,
   sanitizeArticleBody,
+  fixHtmlNesting,
   formatDate,
   generateEventCalendar,
   generateWatchSection,
@@ -57,6 +58,9 @@ export function generateArticleHTML(data: ArticleData): string {
   const formattedDate: string = formatDate(dateObj, lang);
   const isoDate: string = dateObj.toISOString().split('T')[0] ?? '';
 
+  // Fix invalid HTML nesting once so both the rendered body and JSON-LD are consistent
+  const fixedContent: string = fixHtmlNesting(content);
+
   // Fall back to English labels if language not supported
   const typeLabel: string = TYPE_LABELS[lang]?.[type] || TYPE_LABELS.en[type] || 'News';
 
@@ -158,8 +162,8 @@ ${ALL_LANG_CODES.map(l => `  <link rel="alternate" hreflang="${l}" href="https:/
       "height": 630
     },
     "articleSection": "${typeLabel}",
-    "articleBody": "${sanitizeArticleBody(escapeHtml(content))}...",
-    "wordCount": ${Math.ceil(content.length / 5)},
+    "articleBody": "${sanitizeArticleBody(escapeHtml(fixedContent))}...",
+    "wordCount": ${Math.ceil(fixedContent.length / 5)},
     "inLanguage": "${lang}",
     "keywords": "${keywords.join(', ')}",
     "about": {
@@ -270,7 +274,7 @@ ${events.length > 0 ? generateEventCalendar(events as ReadonlyArray<EventGridIte
       ${subtitle}
     </p>
 
-${content}
+${fixedContent}
 
 ${watchPoints.length > 0 ? generateWatchSection(watchPoints as ReadonlyArray<WatchPoint>, lang) : ''}
 

tests/article-template.test.ts

@@ -4,7 +4,7 @@
  */
 
 import { describe, it, expect, afterEach, vi } from 'vitest';
-import { generateArticleHTML, generateArticleLanguageSwitcher, generateSiteFooter } from '../scripts/article-template.js';
+import { generateArticleHTML, generateArticleLanguageSwitcher, generateSiteFooter, fixHtmlNesting } from '../scripts/article-template.js';
 import articleTemplateDefault from '../scripts/article-template.js';
 import type { Language } from '../scripts/types/language.js';
 import type { ArticleData, ArticleCategory, EventGridItem, WatchPoint } from '../scripts/types/article.js';
@@ -890,4 +890,48 @@ describe('Article Template', () => {
       expect(typeof articleTemplateDefault.generateSiteFooter).toBe('function');
     });
   });
+
+  describe('fixHtmlNesting', () => {
+    it('should remove orphaned </p> after </ul>', () => {
+      const input = '<p>intro</p><ul><li>item</li></ul></p>';
+      expect(fixHtmlNesting(input)).toBe('<p>intro</p><ul><li>item</li></ul>');
+    });
+
+    it('should remove orphaned </p> after </ol>', () => {
+      const input = '<p>intro</p><ol><li>step</li></ol></p>';
+      expect(fixHtmlNesting(input)).toBe('<p>intro</p><ol><li>step</li></ol>');
+    });
+
+    it('should handle whitespace between </ul> and </p>', () => {
+      const input = '<ul><li>x</li></ul>  </p>';
+      expect(fixHtmlNesting(input)).toBe('<ul><li>x</li></ul>');
+    });
+
+    it('should leave valid </p> tags unchanged', () => {
+      const input = '<p>valid paragraph</p><p>another</p>';
+      expect(fixHtmlNesting(input)).toBe('<p>valid paragraph</p><p>another</p>');
+    });
+
+    it('should fix the pattern in rendered article HTML body', () => {
+      const data: MockArticleData = {
+        ...mockArticleData,
+        content: '<p>Key proposals include:</p><ul><li><strong>Item A</strong></li></ul></p>'
+      };
+      const html = generateArticleHTML(data as unknown as ArticleData) as string;
+      expect(html).not.toContain('</ul></p>');
+      expect(html).not.toContain('</ol></p>');
+    });
+
+    it('should fix the pattern in JSON-LD articleBody', () => {
+      const data: MockArticleData = {
+        ...mockArticleData,
+        content: '<p>Key proposals include:</p><ul><li><strong>Item A</strong></li></ul></p>'
+      };
+      const html = generateArticleHTML(data as unknown as ArticleData) as string;
+      const jsonLdMatch = html.match(/<script type="application\/ld\+json">([\s\S]*?)<\/script>/);
+      expect(jsonLdMatch).not.toBeNull();
+      // The escaped form of </ul></p> should not appear in JSON-LD articleBody
+      expect(jsonLdMatch![1]).not.toContain('&lt;/ul&gt;&lt;/p&gt;');
+    });
+  });
 });