fix: sites-37260 (#1590)

fix https://jira.corp.adobe.com/browse/SITES-37260

Author: miakobchuk

src/broken-links-guidance/guidance-handler.js

@@ -48,20 +48,58 @@ export default async function handler(message, context) {
     return badRequest('Site ID mismatch');
   }
 
+  // Validate brokenLinks array
+  if (!brokenLinks || !Array.isArray(brokenLinks)) {
+    log.error(`[${opportunity.getType()} Guidance] Invalid brokenLinks format. Expected array, got: ${typeof brokenLinks}. Message: ${JSON.stringify(message)}`);
+    return badRequest('Invalid brokenLinks format');
+  }
+
+  if (brokenLinks.length === 0) {
+    log.info(`[${opportunity.getType()} Guidance] No broken links provided in Mystique response`);
+    return ok();
+  }
+
   await Promise.all(brokenLinks.map(async (brokenLink) => {
     const suggestion = await Suggestion.findById(brokenLink.suggestionId);
     if (!suggestion) {
       log.error(`[${opportunity.getType()}] Suggestion not found for ID: ${brokenLink.suggestionId}`);
       return {};
     }
+
+    const suggestedUrls = brokenLink.suggestedUrls || [];
+
+    // Validate that suggestedUrls is an array
+    if (!Array.isArray(suggestedUrls)) {
+      log.info(
+        `[${opportunity.getType()}] Invalid suggestedUrls format for suggestion ${brokenLink.suggestionId}. `
+        + `Expected array, got: ${typeof suggestedUrls}. Available fields: ${Object.keys(brokenLink).join(', ')}`,
+      );
+    }
+
+    // Filter and validate suggested URLs
+    const validSuggestedUrls = Array.isArray(suggestedUrls) ? suggestedUrls : [];
     const filteredSuggestedUrls = await filterBrokenSuggestedUrls(
-      brokenLink.suggestedUrls,
+      validSuggestedUrls,
       site.getBaseURL(),
     );
+
+    // Handle AI rationale - clear it if all URLs were filtered out
+    // This prevents showing rationale for URLs that don't exist
+    let aiRationale = brokenLink.aiRationale || '';
+    if (filteredSuggestedUrls.length === 0 && validSuggestedUrls.length > 0) {
+      // All URLs were filtered out (likely invalid/broken), clear rationale
+      log.info('All the suggested URLs were filtered out');
+      aiRationale = '';
+    } else if (filteredSuggestedUrls.length === 0 && validSuggestedUrls.length === 0) {
+      // No URLs were provided by Mystique, clear rationale
+      log.info('No suggested URLs provided by Mystique');
+      aiRationale = '';
+    }
+
     suggestion.setData({
       ...suggestion.getData(),
       urlsSuggested: filteredSuggestedUrls,
-      aiRationale: brokenLink.aiRationale,
+      aiRationale,
     });
 
     return suggestion.save();

src/internal-links/handler.js

@@ -226,48 +226,94 @@ export const opportunityAndSuggestionsStep = async (context) => {
   const configuration = await Configuration.findLatest();
   const topPages = await SiteTopPage.allBySiteIdAndSourceAndGeo(site.getId(), 'ahrefs', 'global');
 
+  log.info(
+    `[${AUDIT_TYPE}] [Site: ${site.getId()}] Found ${topPages.length} top pages from Ahrefs`,
+  );
+
   // Filter top pages by audit scope (subpath/locale) if baseURL has a subpath
+  // This determines what alternatives Mystique will see:
+  // - If baseURL is "site.com/en-ca" → only /en-ca alternatives
+  // - If baseURL is "site.com" → ALL locales alternatives
+  // Mystique will then filter by domain (not locale), so cross-locale suggestions
+  // are possible if audit scope includes multiple locales
   const baseURL = site.getBaseURL();
   const filteredTopPages = filterByAuditScope(topPages, baseURL, { urlProperty: 'getUrl' }, log);
 
+  log.info(
+    `[${AUDIT_TYPE}] [Site: ${site.getId()}] After audit scope filtering: ${filteredTopPages.length} top pages available`,
+  );
+
   if (configuration.isHandlerEnabledForSite('broken-internal-links-auto-suggest', site)) {
     const suggestions = await Suggestion.allByOpportunityIdAndStatus(
       opportunity.getId(),
       SuggestionDataAccess.STATUSES.NEW,
     );
 
-    // Filter alternatives per broken link by its locale/subpath
-    const brokenLinksWithFilteredAlternatives = suggestions.map((suggestion) => {
-      const urlFrom = suggestion?.getData()?.urlFrom;
-      const urlTo = suggestion?.getData()?.urlTo;
-
-      // Extract path prefix from broken link to filter alternatives
-      const brokenLinkPathPrefix = extractPathPrefix(urlTo) || extractPathPrefix(urlFrom);
-
-      // Filter alternatives to same locale/subpath as broken link
-      let filteredAlternatives = filteredTopPages.map((page) => page.getUrl());
-      if (brokenLinkPathPrefix) {
-        filteredAlternatives = filteredAlternatives.filter((url) => {
-          const urlPathPrefix = extractPathPrefix(url);
-          return urlPathPrefix === brokenLinkPathPrefix;
-        });
-
-        // Log warning if no alternatives found for this locale
-        if (filteredAlternatives.length === 0) {
-          log.warn(
-            `[${AUDIT_TYPE}] [Site: ${site.getId()}] No alternatives found for broken link `
-            + `with prefix ${brokenLinkPathPrefix}. urlTo: ${urlTo}, urlFrom: ${urlFrom}`,
-          );
-        }
+    // Build broken links array without per-link alternatives
+    // Mystique expects: brokenLinks with only urlFrom, urlTo, suggestionId
+    const brokenLinks = suggestions
+      .map((suggestion) => ({
+        urlFrom: suggestion?.getData()?.urlFrom,
+        urlTo: suggestion?.getData()?.urlTo,
+        suggestionId: suggestion?.getId(),
+      }))
+      .filter((link) => link.urlFrom && link.urlTo && link.suggestionId); // Filter invalid entries
+
+    // Filter alternatives by locales/subpaths present in broken links
+    // This limits suggestions to relevant locales only
+    const allTopPageUrls = filteredTopPages.map((page) => page.getUrl());
+
+    // Extract unique locales/subpaths from broken links
+    const brokenLinkLocales = new Set();
+    brokenLinks.forEach((link) => {
+      const locale = extractPathPrefix(link.urlTo) || extractPathPrefix(link.urlFrom);
+      if (locale) {
+        brokenLinkLocales.add(locale);
       }
+    });
 
+    // Filter alternatives to only include URLs matching broken links' locales
+    // If no locales found (no subpath), include all alternatives
+    // Always ensure alternativeUrls is an array (even if empty)
+    let alternativeUrls = [];
+    if (brokenLinkLocales.size > 0) {
+      alternativeUrls = allTopPageUrls.filter((url) => {
+        const urlLocale = extractPathPrefix(url);
+        // Include if URL matches one of the broken links' locales, or has no locale
+        return !urlLocale || brokenLinkLocales.has(urlLocale);
+      });
+    } else {
+      // No locale prefixes found, include all alternatives
+      alternativeUrls = allTopPageUrls;
+    }
+
+    // Validate before sending to Mystique
+    if (brokenLinks.length === 0) {
+      log.warn(
+        `[${AUDIT_TYPE}] [Site: ${site.getId()}] No valid broken links to send to Mystique. Skipping message.`,
+      );
       return {
-        urlFrom,
-        urlTo,
-        suggestionId: suggestion?.getId(),
-        alternativeUrls: filteredAlternatives,
+        status: 'complete',
       };
-    });
+    }
+
+    if (!opportunity?.getId()) {
+      log.error(
+        `[${AUDIT_TYPE}] [Site: ${site.getId()}] Opportunity ID is missing. Cannot send to Mystique.`,
+      );
+      return {
+        status: 'complete',
+      };
+    }
+
+    if (alternativeUrls.length === 0) {
+      log.warn(
+        `[${AUDIT_TYPE}] [Site: ${site.getId()}] No alternative URLs available. Cannot generate suggestions. Skipping message to Mystique.`,
+      );
+      return {
+        status: 'complete',
+      };
+    }
 
     const message = {
       type: 'guidance:broken-links',
@@ -276,9 +322,9 @@ export const opportunityAndSuggestionsStep = async (context) => {
       deliveryType: site.getDeliveryType(),
       time: new Date().toISOString(),
       data: {
-        alternativeUrls: filteredTopPages.map((page) => page.getUrl()),
-        opportunityId: opportunity?.getId(),
-        brokenLinks: brokenLinksWithFilteredAlternatives,
+        alternativeUrls,
+        opportunityId: opportunity.getId(),
+        brokenLinks,
       },
     };
     await sqs.sendMessage(env.QUEUE_SPACECAT_TO_MYSTIQUE, message);