Merge pull request #310 from GSA/seo-expansion

Add experimental seo fields

entities/core-result.entity.ts

@@ -422,6 +422,210 @@ export class CoreResult {
   @Expose({ name: 'cumulative_layout_shift' })
   cumulativeLayoutShift?: number;
 
+  /**
+   * #852
+   * The following fields are experimental and were added in March 2024 for
+   * prototyping purposes. They are not yet used in the application and may be
+   * removed in the future.
+   */
+  @Column({ nullable: true })
+  @Expose({ name: 'meta_keywords_content' })
+  @Exclude()
+  metaKeywordsContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'meta_robots_content' })
+  @Exclude()
+  metaRobotsContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'meta_article_section_content' })
+  @Exclude()
+  metaArticleSectionContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'meta_article_tag_content' })
+  @Exclude()
+  metaArticleTagContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'og_image_final_url' })
+  @Exclude()
+  ogImageFinalUrl?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dcterms_keywords_content' })
+  @Exclude()
+  dctermsKeywordsContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dc_subject_content' })
+  @Exclude()
+  dcSubjectContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dcterms_subject_content' })
+  @Exclude()
+  dctermsSubjectContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dcterms_audience_content' })
+  @Exclude()
+  dctermsAudienceContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dc_type_content' })
+  @Exclude()
+  dcTypeContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dcterms_type_content' })
+  @Exclude()
+  dctermsTypeContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dc_date_content' })
+  @Exclude()
+  dcDateContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dc_date_created_content' })
+  @Exclude()
+  dcDateCreatedContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dcterms_created_content' })
+  @Exclude()
+  dctermsCreatedContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'og_locale_content' })
+  @Exclude()
+  ogLocaleContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'og_site_name_content' })
+  @Exclude()
+  ogSiteNameContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'og_type_content' })
+  @Exclude()
+  ogTypeContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'og_url_content' })
+  @Exclude()
+  ogUrlContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'og_image_alt_content' })
+  @Exclude()
+  ogImageAltContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'revised_content' })
+  @Exclude()
+  revisedContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'last_modified_content' })
+  @Exclude()
+  lastModifiedContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'language_content' })
+  @Exclude()
+  languageContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'date_content' })
+  @Exclude()
+  dateContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'subject_content' })
+  @Exclude()
+  subjectContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'owner_content' })
+  @Exclude()
+  ownerContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'pagename_content' })
+  @Exclude()
+  pagenameContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'dc_title_content' })
+  @Exclude()
+  dcTitleContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'og_site_name' })
+  @Exclude()
+  ogSiteName?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'item_type_content' })
+  @Exclude()
+  itemTypeContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'item_scope_content' })
+  @Exclude()
+  itemScopeContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'item_prop_content' })
+  @Exclude()
+  itemPropContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'vocab_content' })
+  @Exclude()
+  vocabContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'type_of_content' })
+  @Exclude()
+  typeOfContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'propertyContent' })
+  @Exclude()
+  propertyContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'context_content' })
+  @Exclude()
+  contextContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'type_content' })
+  @Exclude()
+  typeContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'html_lang_content' })
+  @Exclude()
+  htmlLangContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'hreflang_content' })
+  @Exclude()
+  hrefLangContent?: string;
+
+  @Column({ nullable: true })
+  @Expose({ name: 'me_content' })
+  @Exclude()
+  meContent?: string;
+  /**
+   * End experimental fields.
+   */
+
   static getColumnNames(): string[] {
     // return class-transformer version of column names
     return Object.keys(classToPlain(new CoreResult()));

entities/scan-data.entity.ts

@@ -30,6 +30,47 @@ export type SeoScan = {
   pageTitle: string;
   metaDescriptionContent: string;
   hreflangCodes: string;
+  // #852 Begin March 2024 experimental fields
+  metaKeywordsContent: string;
+  metaRobotsContent: string;
+  metaArticleSectionContent: string;
+  metaArticleTagContent: string;
+  ogImageFinalUrl: string;
+  dctermsKeywordsContent: string;
+  dcSubjectContent: string;
+  dctermsSubjectContent: string;
+  dctermsAudienceContent: string;
+  dcTypeContent: string;
+  dctermsTypeContent: string;
+  dcDateContent: string;
+  dcDateCreatedContent: string;
+  dctermsCreatedContent: string;
+  ogLocaleContent: string;
+  ogSiteNameContent: string;
+  ogTypeContent: string;
+  ogUrlContent: string;
+  ogImageAltContent: string;
+  revisedContent: string;
+  lastModifiedContent: string;
+  languageContent: string;
+  dateContent: string;
+  subjectContent: string;
+  ownerContent: string;
+  pagenameContent: string;
+  dcTitleContent: string;
+  ogSiteName: string;
+  itemTypeContent: string;
+  itemScopeContent: string;
+  itemPropContent: string;
+  vocabContent: string;
+  typeOfContent: string;
+  propertyContent: string;
+  contextContent: string;
+  typeContent: string;
+  htmlLangContent: string;
+  hrefLangContent: string;
+  meContent: string;
+  // End March 2024 experimental fields
 };
 
 export type ThirdPartyScan = {

libs/core-scanner/src/pages/primary.spec.ts

@@ -50,6 +50,48 @@ describe('primary scanner', () => {
           metaDescriptionContent:
             '18F builds effective, user-centric digital services focused on the interaction between government and the people and businesses it serves.',
           hreflangCodes: '',
+          // #852 Begin March 2024 experimental fields
+          contextContent: null,
+          dateContent: null,
+          dcDateContent: null,
+          dcDateCreatedContent: null,
+          dcSubjectContent: null,
+          dcTitleContent: null,
+          dcTypeContent: null,
+          dctermsAudienceContent: null,
+          dctermsCreatedContent: null,
+          dctermsKeywordsContent: null,
+          dctermsSubjectContent: null,
+          dctermsTypeContent: null,
+          hrefLangContent: '',
+          htmlLangContent: 'en-US',
+          itemPropContent: null,
+          itemScopeContent: null,
+          itemTypeContent: null,
+          languageContent: null,
+          lastModifiedContent: null,
+          meContent: null,
+          metaArticleSectionContent: null,
+          metaArticleTagContent: null,
+          metaKeywordsContent: null,
+          metaRobotsContent: null,
+          ogImageAltContent: null,
+          ogImageFinalUrl:
+            'https://18f.gsa.gov/assets/img/logos/18F-Logo-M.png',
+          ogLocaleContent: null,
+          ogSiteName: null,
+          ogSiteNameContent: null,
+          ogTypeContent: null,
+          ogUrlContent: null,
+          ownerContent: null,
+          pagenameContent: null,
+          propertyContent: 'og:title',
+          revisedContent: null,
+          subjectContent: null,
+          typeContent: 'image/png',
+          typeOfContent: null,
+          vocabContent: null,
+          // End March 2024 experimental fields
         },
         thirdPartyScan: {
           thirdPartyServiceDomains:

libs/core-scanner/src/scans/seo.spec.ts

@@ -26,6 +26,50 @@ describe('seo scan', () => {
         metaDescriptionContent: '',
 
         hreflangCodes: '',
+        // #852 Begin March 2024 experimental fields
+        contextContent: null,
+        dateContent: null,
+        dcDateContent: null,
+        dcDateCreatedContent: null,
+        dcSubjectContent: null,
+        dcTitleContent: null,
+        dcTypeContent: null,
+        dctermsAudienceContent: null,
+        dctermsCreatedContent: null,
+        dctermsKeywordsContent: null,
+        dctermsSubjectContent: null,
+        dctermsTypeContent: null,
+        hrefLangContent: '',
+        htmlLangContent: 'en-US',
+        itemPropContent: null,
+        itemScopeContent: null,
+        itemTypeContent: null,
+        languageContent: null,
+        lastModifiedContent: null,
+        meContent: null,
+        metaArticleSectionContent: null,
+        metaArticleTagContent: null,
+        metaKeywordsContent: null,
+        metaRobotsContent:
+          'index, follow, max-image-preview:large, max-snippet:-1, max-video-preview:-1',
+        ogImageAltContent: null,
+        ogImageFinalUrl:
+          'https://www.arc.gov/wp-content/uploads/2020/08/48982308566_5ce274f0ab_o-washington-scaled.jpg',
+        ogLocaleContent: 'en_US',
+        ogSiteName: null,
+        ogSiteNameContent: 'Appalachian Regional Commission',
+        ogTypeContent: 'website',
+        ogUrlContent: 'https://www.arc.gov/',
+        ownerContent: null,
+        pagenameContent: null,
+        propertyContent: 'og:locale',
+        revisedContent: null,
+        subjectContent: null,
+        typeContent:
+          'text/css,application/rss+xml,application/json,application/rsd+xml,application/wlwmanifest+xml,application/json+oembed,text/xml+oembed',
+        typeOfContent: null,
+        vocabContent: null,
+        // End March 2024 experimental fields
       });
     }, 'arc_gov_dump.mht');
   });