fix: migrate metatags audit to AuditBuilder

src/common/audit.js

@@ -21,7 +21,6 @@ export async function defaultMessageSender() {}
 export async function defaultPersister(auditData, context) {
   const { dataAccess } = context;
   const { Audit } = dataAccess;
-
   return Audit.create(auditData);
 }
 

src/metatags/handler.js

@@ -10,16 +10,13 @@
  * governing permissions and limitations under the License.
  */
 
-import {
-  internalServerError, noContent, notFound, ok,
-} from '@adobe/spacecat-shared-http-utils';
-import { composeAuditURL } from '@adobe/spacecat-shared-utils';
-import { retrieveSiteBySiteId } from '../utils/data-access.js';
 import { getObjectFromKey, getObjectKeysUsingPrefix } from '../utils/s3-utils.js';
 import SeoChecks from './seo-checks.js';
-import syncOpportunityAndSuggestions from './opportunityHandler.js';
+import { AuditBuilder } from '../common/audit-builder.js';
+import { noopUrlResolver } from '../common/audit.js';
+import convertToOpportunity from './opportunityHandler.js';
 
-async function fetchAndProcessPageObject(s3Client, bucketName, key, prefix, log) {
+export async function fetchAndProcessPageObject(s3Client, bucketName, key, prefix, log) {
   const object = await getObjectFromKey(s3Client, bucketName, key, log);
   if (!object?.scrapeResult?.tags || typeof object.scrapeResult.tags !== 'object') {
     log.error(`No Scraped tags found in S3 ${key} object`);
@@ -35,88 +32,49 @@ async function fetchAndProcessPageObject(s3Client, bucketName, key, prefix, log)
   };
 }
 
-export default async function auditMetaTags(message, context) {
-  const { type, auditContext = {} } = message;
-  const siteId = message.siteId || message.url;
-  const {
-    dataAccess, log, s3Client,
-  } = context;
-  const { Audit, Configuration } = dataAccess;
-
-  try {
-    const site = await retrieveSiteBySiteId(dataAccess, siteId, log);
-    if (!site) {
-      return notFound('Site not found');
-    }
-    if (!site.getIsLive()) {
-      log.info(`Site ${siteId} is not live`);
-      return ok();
-    }
-    const configuration = await Configuration.findLatest();
-    if (!configuration.isHandlerEnabledForSite(type, site)) {
-      log.info(`Audit type ${type} disabled for site ${siteId}`);
-      return ok();
-    }
-    try {
-      auditContext.finalUrl = await composeAuditURL(site.getBaseURL());
-    } catch (e) {
-      log.error(`Get final URL for siteId ${siteId} failed with error: ${e.message}`, e);
-      return internalServerError(`Internal server error: ${e.message}`);
+export async function auditMetaTagsRunner(baseURL, context, site) {
+  const { log, s3Client } = context;
+  // Fetch site's scraped content from S3
+  const bucketName = context.S3_SCRAPER_BUCKET_NAME;
+  const prefix = `scrapes/${site.getId()}/`;
+  const scrapedObjectKeys = await getObjectKeysUsingPrefix(s3Client, bucketName, prefix, log);
+  const extractedTags = {};
+  const pageMetadataResults = await Promise.all(scrapedObjectKeys.map(
+    (key) => fetchAndProcessPageObject(s3Client, bucketName, key, prefix, log),
+  ));
+  pageMetadataResults.forEach((pageMetadata) => {
+    if (pageMetadata) {
+      Object.assign(extractedTags, pageMetadata);
     }
-    // Fetch site's scraped content from S3
-    const bucketName = context.env.S3_SCRAPER_BUCKET_NAME;
-    const prefix = `scrapes/${siteId}/`;
-    const scrapedObjectKeys = await getObjectKeysUsingPrefix(s3Client, bucketName, prefix, log);
-    const extractedTags = {};
-    const pageMetadataResults = await Promise.all(scrapedObjectKeys.map(
-      (key) => fetchAndProcessPageObject(s3Client, bucketName, key, prefix, log),
-    ));
-    pageMetadataResults.forEach((pageMetadata) => {
-      if (pageMetadata) {
-        Object.assign(extractedTags, pageMetadata);
-      }
-    });
-    const extractedTagsCount = Object.entries(extractedTags).length;
-    if (extractedTagsCount === 0) {
-      log.error(`Failed to extract tags from scraped content for bucket ${bucketName} and prefix ${prefix}`);
-      return notFound('Site tags data not available');
-    }
-    log.info(`Performing SEO checks for ${extractedTagsCount} tags`);
-    // Perform SEO checks
-    const seoChecks = new SeoChecks(log);
-    for (const [pageUrl, pageTags] of Object.entries(extractedTags)) {
-      seoChecks.performChecks(pageUrl || '/', pageTags);
-    }
-    seoChecks.finalChecks();
-    const detectedTags = seoChecks.getDetectedTags();
-    // Prepare Audit result
-    const auditResult = {
-      detectedTags,
-      sourceS3Folder: `${bucketName}/${prefix}`,
-      fullAuditRef: 'na',
-      finalUrl: auditContext.finalUrl,
-    };
-    const auditData = {
-      siteId: site.getId(),
-      isLive: site.getIsLive(),
-      auditedAt: new Date().toISOString(),
-      auditType: type,
-      fullAuditRef: auditResult?.fullAuditRef,
-      auditResult,
-    };
-    // Persist Audit result
-    const audit = await Audit.create(auditData);
-    log.info(`Successfully audited ${siteId} for ${type} type audit`);
-    await syncOpportunityAndSuggestions(
-      siteId,
-      audit.getId(),
-      auditData,
-      dataAccess,
-      log,
-    );
-    return noContent();
-  } catch (e) {
-    log.error(`${type} type audit for ${siteId} failed with error: ${e.message}`, e);
-    return internalServerError(`Internal server error: ${e.message}`);
+  });
+  const extractedTagsCount = Object.entries(extractedTags).length;
+  if (extractedTagsCount === 0) {
+    log.error(`Failed to extract tags from scraped content for bucket ${bucketName} and prefix ${prefix}`);
+  }
+  log.info(`Performing SEO checks for ${extractedTagsCount} tags`);
+  // Perform SEO checks
+  const seoChecks = new SeoChecks(log);
+  for (const [pageUrl, pageTags] of Object.entries(extractedTags)) {
+    seoChecks.performChecks(pageUrl || '/', pageTags);
   }
+  seoChecks.finalChecks();
+  const detectedTags = seoChecks.getDetectedTags();
+
+  const auditResult = {
+    detectedTags,
+    sourceS3Folder: `${bucketName}/${prefix}`,
+    fullAuditRef: 'na',
+    finalUrl: baseURL,
+  };
+
+  return {
+    auditResult,
+    fullAuditRef: baseURL,
+  };
 }
+
+export default new AuditBuilder()
+  .withUrlResolver(noopUrlResolver)
+  .withRunner(auditMetaTagsRunner)
+  .withPostProcessors([convertToOpportunity])
+  .build();

src/metatags/opportunityHandler.js

@@ -12,7 +12,7 @@
 
 import { TITLE, DESCRIPTION, H1 } from './constants.js';
 
-function removeTrailingSlash(url) {
+export function removeTrailingSlash(url) {
   return url.endsWith('/') ? url.slice(0, -1) : url;
 }
 
@@ -140,39 +140,31 @@ function getIssueRanking(tagName, issue) {
 }
 
 /**
- * Updates Meta-tags Opportunity and Suggestions collection with new audit results
- * @param siteId site id of site being audited
- * @param auditId audit id of the latest performed audit
- * @param auditData object containing audit results and some metadata
- * @param dataAccess object containing accessor objects
- * @param log logger
- * @returns {Promise<void>}
+ * @param auditUrl - The URL of the audit
+ * @param auditData - The audit data containing the audit result and additional details.
+ * @param context - The context object containing the data access and logger objects.
  */
-export default async function syncOpportunityAndSuggestions(
-  siteId,
-  auditId,
-  auditData,
-  dataAccess,
-  log,
-) {
+export default async function convertToOpportunity(auditUrl, auditData, context) {
+  const { dataAccess, log } = context;
   const { Opportunity } = dataAccess;
-  log.info(`Syncing opportunity and suggestions for ${siteId}`);
+
+  log.info(`Syncing opportunity and suggestions for ${auditData.siteId}`);
   let metatagsOppty;
+
   try {
-    // Get all opportunities by site-id and new status
-    const opportunities = await Opportunity.allBySiteIdAndStatus(siteId, 'NEW');
-    // Find existing opportunity for meta-tags
+    const opportunities = await Opportunity.allBySiteIdAndStatus(auditData.siteId, 'NEW');
+
     metatagsOppty = opportunities.find((oppty) => oppty.getType() === 'meta-tags');
   } catch (e) {
-    log.error(`Fetching opportunities for siteId ${siteId} failed with error: ${e.message}`);
-    throw new Error(`Failed to fetch opportunities for siteId ${siteId}: ${e.message}`);
+    log.error(`Fetching opportunities for siteId ${auditData.siteId} failed with error: ${e.message}`);
+    throw new Error(`Failed to fetch opportunities for siteId ${auditData.siteId}: ${e.message}`);
   }
 
   try {
     if (!metatagsOppty) {
       const opportunityData = {
-        siteId,
-        auditId,
+        siteId: auditData.siteId,
+        auditId: auditData.id,
         runbook: 'https://adobe.sharepoint.com/:w:/r/sites/aemsites-engineering/_layouts/15/doc2.aspx?sourcedoc=%7B27CF48AA-5492-435D-B17C-01E38332A5CA%7D&file=Experience_Success_Studio_Metatags_Runbook.docx&action=default&mobileredirect=true',
         type: 'meta-tags',
         origin: 'AUTOMATION',
@@ -192,13 +184,14 @@ export default async function syncOpportunityAndSuggestions(
       metatagsOppty = await Opportunity.create(opportunityData);
       log.debug('Meta-tags Opportunity created');
     } else {
-      metatagsOppty.setAuditId(auditId);
+      metatagsOppty.setAuditId(auditData.siteId);
       await metatagsOppty.save();
     }
   } catch (e) {
-    log.error(`Creating meta-tags opportunity for siteId ${siteId} failed with error: ${e.message}`, e);
-    throw new Error(`Failed to create meta-tags opportunity for siteId ${siteId}: ${e.message}`);
+    log.error(`Creating meta-tags opportunity for siteId ${auditData.siteId} failed with error: ${e.message}`, e);
+    throw new Error(`Failed to create meta-tags opportunity for siteId ${auditData.siteId}: ${e.message}`);
   }
+
   const { detectedTags } = auditData.auditResult;
   const suggestions = [];
   // Generate suggestions data to be inserted in meta-tags opportunity suggestions
@@ -216,6 +209,7 @@ export default async function syncOpportunityAndSuggestions(
   });
 
   const buildKey = (data) => `${data.url}|${data.issue}|${data.tagContent}`;
+
   // Sync the suggestions from new audit with old ones
   await syncMetatagsSuggestions({
     opportunity: metatagsOppty,
@@ -225,11 +219,9 @@ export default async function syncOpportunityAndSuggestions(
       opportunityId: metatagsOppty.getId(),
       type: 'METADATA_UPDATE',
       rank: suggestion.rank,
-      data: {
-        ...suggestion,
-      },
+      data: { ...suggestion },
     }),
     log,
   });
-  log.info(`Successfully synced Opportunity And Suggestions for site: ${siteId} and meta-tags audit type.`);
+  log.info(`Successfully synced Opportunity And Suggestions for site: ${auditData.siteId} and meta-tags audit type.`);
 }