chore: validate html descriptions, warn if invalid

actions/lib/validateHtml.js

@@ -0,0 +1,77 @@
+/**
+ * Validates HTML syntax by checking for balanced opening and closing tags
+ * @param {string} html - The HTML string to validate
+ * @returns {Object} - { valid: boolean, reason: string }
+ */
+function validateHtml(html) {
+  if (typeof html !== 'string') {
+    return { valid: false, reason: 'Input must be a string' };
+  }
+
+  if (html.trim() === '') {
+    return { valid: true, reason: 'Empty string is valid' };
+  }
+
+  const stack = [];
+  const selfClosingTags = new Set([
+    'area', 'base', 'br', 'col', 'embed', 'hr', 'img', 'input',
+    'link', 'meta', 'param', 'source', 'track', 'wbr'
+  ]);
+
+  // Regular expression to match HTML tags
+  const tagRegex = /<\/?([a-zA-Z][a-zA-Z0-9]*)\b[^>]*>/g;
+  let match;
+  let lineNumber = 1;
+  let charPosition = 0;
+
+  while ((match = tagRegex.exec(html)) !== null) {
+    const fullTag = match[0];
+    const tagName = match[1].toLowerCase();
+    const isClosingTag = fullTag.startsWith('</');
+    const isSelfClosing = fullTag.endsWith('/>') || selfClosingTags.has(tagName);
+
+    // Calculate position for error reporting
+    const beforeMatch = html.substring(0, match.index);
+    lineNumber = beforeMatch.split('\n').length;
+    charPosition = match.index - beforeMatch.lastIndexOf('\n') - 1;
+
+    if (isSelfClosing) {
+      // Self-closing tags don't need to be balanced
+      continue;
+    }
+
+    if (isClosingTag) {
+      // Check if we have a matching opening tag
+      if (stack.length === 0) {
+        return {
+          valid: false,
+          reason: `Unexpected closing tag </${tagName}> at line ${lineNumber}, position ${charPosition}`
+        };
+      }
+
+      const lastOpenTag = stack.pop();
+      if (lastOpenTag !== tagName) {
+        return {
+          valid: false,
+          reason: `Mismatched tags: expected </${lastOpenTag}> but found </${tagName}> at line ${lineNumber}, position ${charPosition}`
+        };
+      }
+    } else {
+      // Opening tag - push onto stack
+      stack.push(tagName);
+    }
+  }
+
+  // Check if any opening tags weren't closed
+  if (stack.length > 0) {
+    const unclosedTags = stack.reverse().join(', ');
+    return {
+      valid: false,
+      reason: `Unclosed tags: ${unclosedTags}`
+    };
+  }
+
+  return { valid: true, reason: 'HTML is valid' };
+}
+
+module.exports = { validateHtml }

actions/pdp-renderer/render.js

@@ -5,10 +5,11 @@ const { findDescription, prepareBaseTemplate, getPrimaryImage, generatePriceStri
 const { generateLdJson } = require('./ldJson');
 const { requestSaaS, getProductUrl } = require('../utils');
 const { ProductQuery, ProductByUrlKeyQuery } = require('../queries');
+const { validateHtml } = require('../lib/validateHtml');
 
 const productTemplateCache = {};
 
-function toTemplateProductData(baseProduct) {
+function toTemplateProductData(baseProduct, context) {
   const templateProductData = { ...baseProduct };
   const primaryImage = getPrimaryImage(baseProduct)?.url;
 
@@ -20,6 +21,18 @@ function toTemplateProductData(baseProduct) {
   templateProductData.primaryImage = primaryImage;
   templateProductData.metaTitle = baseProduct.metaTitle || baseProduct.name || 'Product Details';
 
+  const fieldValidations = {
+    metaDescription: validateHtml(templateProductData.metaDescription),
+    shortDescription: validateHtml(templateProductData.shortDescription),
+    description: validateHtml(templateProductData.description)
+  }
+
+  Object.entries(fieldValidations).forEach(([field, validation]) => {
+    if (!validation.valid) {
+      context.logger.warn(`Validation failed for "${field}" field: ${validation.reason}`);
+    }
+  })
+
   return templateProductData;
 }
 
@@ -65,7 +78,7 @@ async function generateProductHtml(sku, urlKey, context) {
   }
 
   // Assign meta tag data for template
-  const templateProductData = toTemplateProductData(baseProduct);
+  const templateProductData = toTemplateProductData(baseProduct, context);
 
   // Generate LD-JSON
   const ldJson = await generateLdJson(baseProduct, context);