@rankcli/agent-runtime 0.0.14 → 0.0.16

package/src/analyzers/structured-data-analyzer.ts

@@ -1,590 +0,0 @@
-/**
- * Structured Data / Schema.org Analyzer
- * 
- * Validates JSON-LD structured data for SEO and AI search optimization.
- * Checks for common schema types, required properties, and best practices.
- */
-
-import * as cheerio from 'cheerio';
-import type { AuditIssue } from '../audit/types.js';
-
-export interface StructuredDataResult {
-  score: number;
-  schemas: SchemaItem[];
-  hasOrganization: boolean;
-  hasWebSite: boolean;
-  hasWebPage: boolean;
-  hasBreadcrumb: boolean;
-  hasArticle: boolean;
-  hasProduct: boolean;
-  hasFAQ: boolean;
-  hasHowTo: boolean;
-  hasLocalBusiness: boolean;
-  hasEvent: boolean;
-  hasReview: boolean;
-  issues: AuditIssue[];
-  recommendations: string[];
-}
-
-export interface SchemaItem {
-  type: string;
-  isValid: boolean;
-  errors: string[];
-  warnings: string[];
-  properties: string[];
-}
-
-// Required and recommended properties for common schema types
-const SCHEMA_REQUIREMENTS: Record<string, { required: string[]; recommended: string[] }> = {
-  'Organization': {
-    required: ['name', 'url'],
-    recommended: ['logo', 'sameAs', 'contactPoint', 'address'],
-  },
-  'WebSite': {
-    required: ['name', 'url'],
-    recommended: ['potentialAction', 'publisher'],
-  },
-  'WebPage': {
-    required: ['name'],
-    recommended: ['description', 'url', 'breadcrumb', 'mainEntity'],
-  },
-  'Article': {
-    required: ['headline', 'author', 'datePublished'],
-    recommended: ['image', 'dateModified', 'publisher', 'description', 'mainEntityOfPage'],
-  },
-  'BlogPosting': {
-    required: ['headline', 'author', 'datePublished'],
-    recommended: ['image', 'dateModified', 'publisher', 'description'],
-  },
-  'NewsArticle': {
-    required: ['headline', 'author', 'datePublished', 'image'],
-    recommended: ['dateModified', 'publisher', 'description'],
-  },
-  'Product': {
-    required: ['name'],
-    recommended: ['image', 'description', 'offers', 'brand', 'sku', 'aggregateRating', 'review'],
-  },
-  'Offer': {
-    required: ['price', 'priceCurrency'],
-    recommended: ['availability', 'url', 'priceValidUntil'],
-  },
-  'FAQPage': {
-    required: ['mainEntity'],
-    recommended: [],
-  },
-  'Question': {
-    required: ['name', 'acceptedAnswer'],
-    recommended: [],
-  },
-  'HowTo': {
-    required: ['name', 'step'],
-    recommended: ['image', 'totalTime', 'estimatedCost', 'supply', 'tool'],
-  },
-  'BreadcrumbList': {
-    required: ['itemListElement'],
-    recommended: [],
-  },
-  'LocalBusiness': {
-    required: ['name', 'address'],
-    recommended: ['telephone', 'openingHours', 'geo', 'image', 'priceRange'],
-  },
-  'Event': {
-    required: ['name', 'startDate', 'location'],
-    recommended: ['endDate', 'description', 'image', 'offers', 'performer', 'organizer'],
-  },
-  'Review': {
-    required: ['itemReviewed', 'reviewRating', 'author'],
-    recommended: ['reviewBody', 'datePublished'],
-  },
-  'AggregateRating': {
-    required: ['ratingValue', 'reviewCount'],
-    recommended: ['bestRating', 'worstRating'],
-  },
-  'Person': {
-    required: ['name'],
-    recommended: ['url', 'image', 'jobTitle', 'sameAs'],
-  },
-  'VideoObject': {
-    required: ['name', 'thumbnailUrl', 'uploadDate'],
-    recommended: ['description', 'contentUrl', 'duration', 'embedUrl'],
-  },
-  'ImageObject': {
-    required: ['contentUrl'],
-    recommended: ['caption', 'creditText', 'creator'],
-  },
-  'SoftwareApplication': {
-    required: ['name'],
-    recommended: ['operatingSystem', 'applicationCategory', 'offers', 'aggregateRating'],
-  },
-  'Recipe': {
-    required: ['name', 'recipeIngredient', 'recipeInstructions'],
-    recommended: ['image', 'author', 'prepTime', 'cookTime', 'totalTime', 'nutrition'],
-  },
-  'Course': {
-    required: ['name', 'provider'],
-    recommended: ['description', 'offers'],
-  },
-};
-
-/**
- * Parse and validate JSON-LD structured data
- */
-function parseJsonLd(script: string): { data: any; error?: string } {
-  try {
-    const data = JSON.parse(script);
-    return { data };
-  } catch (e) {
-    return { data: null, error: `Invalid JSON: ${(e as Error).message}` };
-  }
-}
-
-/**
- * Get all types from a schema object (handles @graph and nested types)
- */
-function extractTypes(data: any): string[] {
-  const types: string[] = [];
-  
-  if (!data) return types;
-  
-  if (Array.isArray(data)) {
-    for (const item of data) {
-      types.push(...extractTypes(item));
-    }
-  } else if (typeof data === 'object') {
-    if (data['@type']) {
-      const schemaType = Array.isArray(data['@type']) ? data['@type'] : [data['@type']];
-      types.push(...schemaType);
-    }
-    if (data['@graph']) {
-      types.push(...extractTypes(data['@graph']));
-    }
-  }
-  
-  return types;
-}
-
-/**
- * Validate a schema object against requirements
- */
-function validateSchema(data: any, type: string): SchemaItem {
-  const errors: string[] = [];
-  const warnings: string[] = [];
-  const properties: string[] = [];
-  
-  if (!data || typeof data !== 'object') {
-    return { type, isValid: false, errors: ['Invalid schema object'], warnings, properties };
-  }
-  
-  // Extract properties
-  for (const key of Object.keys(data)) {
-    if (!key.startsWith('@')) {
-      properties.push(key);
-    }
-  }
-  
-  // Check requirements
-  const requirements = SCHEMA_REQUIREMENTS[type];
-  if (requirements) {
-    // Check required properties
-    for (const prop of requirements.required) {
-      if (!data[prop]) {
-        errors.push(`Missing required property: ${prop}`);
-      }
-    }
-    
-    // Check recommended properties
-    for (const prop of requirements.recommended) {
-      if (!data[prop]) {
-        warnings.push(`Missing recommended property: ${prop}`);
-      }
-    }
-  }
-  
-  // Type-specific validations
-  if (type === 'Article' || type === 'BlogPosting' || type === 'NewsArticle') {
-    // Check for proper author format
-    if (data.author && typeof data.author === 'string') {
-      warnings.push('author should be a Person or Organization object, not a string');
-    }
-    // Check for proper publisher format
-    if (data.publisher && typeof data.publisher === 'string') {
-      warnings.push('publisher should be an Organization object, not a string');
-    }
-    // Check date format
-    if (data.datePublished && !/^\d{4}-\d{2}-\d{2}/.test(data.datePublished)) {
-      warnings.push('datePublished should be in ISO 8601 format (YYYY-MM-DD)');
-    }
-  }
-  
-  if (type === 'Product') {
-    // Check for offers
-    if (!data.offers) {
-      warnings.push('Product without offers/price - may not show rich results');
-    }
-  }
-  
-  if (type === 'FAQPage') {
-    // Check for valid Q&A structure
-    if (data.mainEntity) {
-      const questions = Array.isArray(data.mainEntity) ? data.mainEntity : [data.mainEntity];
-      for (let i = 0; i < questions.length; i++) {
-        const q = questions[i];
-        if (!q.name && !q['@type']?.includes('Question')) {
-          errors.push(`FAQ item ${i + 1} is not a valid Question`);
-        }
-        if (!q.acceptedAnswer) {
-          errors.push(`FAQ item ${i + 1} missing acceptedAnswer`);
-        }
-      }
-    }
-  }
-  
-  if (type === 'BreadcrumbList') {
-    // Check for proper item list
-    if (data.itemListElement) {
-      const items = Array.isArray(data.itemListElement) ? data.itemListElement : [data.itemListElement];
-      items.forEach((item: Record<string, unknown>, i: number) => {
-        if (!item.position) {
-          errors.push(`Breadcrumb item ${i + 1} missing position`);
-        }
-        if (!item.name && !(item.item as Record<string, unknown>)?.name) {
-          errors.push(`Breadcrumb item ${i + 1} missing name`);
-        }
-      });
-    }
-  }
-  
-  if (type === 'LocalBusiness') {
-    // Check for geo coordinates
-    if (data.geo) {
-      if (!data.geo.latitude || !data.geo.longitude) {
-        warnings.push('geo should include latitude and longitude');
-      }
-    }
-    // Check opening hours format
-    if (data.openingHours && typeof data.openingHours === 'string') {
-      warnings.push('openingHours should be an array of OpeningHoursSpecification objects for rich results');
-    }
-  }
-  
-  return {
-    type,
-    isValid: errors.length === 0,
-    errors,
-    warnings,
-    properties,
-  };
-}
-
-/**
- * Analyze all structured data on a page
- */
-export function analyzeStructuredData(html: string, url: string): StructuredDataResult {
-  const $ = cheerio.load(html);
-  const schemas: SchemaItem[] = [];
-  const issues: AuditIssue[] = [];
-  const recommendations: string[] = [];
-  let score = 100;
-  
-  // Track schema types
-  const foundTypes = new Set<string>();
-  
-  // Parse all JSON-LD scripts
-  $('script[type="application/ld+json"]').each((_, el) => {
-    const content = $(el).html();
-    if (!content) return;
-    
-    const { data, error } = parseJsonLd(content);
-    
-    if (error) {
-      issues.push({
-        code: 'SCHEMA_INVALID_JSON',
-        severity: 'critical',
-        category: 'technical',
-        title: 'Invalid JSON-LD syntax',
-        description: error,
-        impact: 'Schema will not be parsed by search engines',
-        howToFix: 'Fix JSON syntax errors. Use a JSON validator.',
-        affectedUrls: [url],
-      });
-      score -= 20;
-      return;
-    }
-    
-    // Extract and validate all types
-    const types = extractTypes(data);
-    
-    // Process each schema item
-    const processItem = (item: any) => {
-      if (!item || typeof item !== 'object') return;
-      
-      let itemType = item['@type'];
-      if (Array.isArray(itemType)) itemType = itemType[0];
-      
-      if (itemType) {
-        foundTypes.add(itemType);
-        const validation = validateSchema(item, itemType);
-        schemas.push(validation);
-        
-        if (!validation.isValid) {
-          for (const err of validation.errors) {
-            issues.push({
-              code: 'SCHEMA_VALIDATION_ERROR',
-              severity: 'warning',
-              category: 'technical',
-              title: `${itemType} schema error: ${err}`,
-              description: `The ${itemType} schema is missing required properties or has invalid data.`,
-              impact: 'May not qualify for rich results',
-              howToFix: `Add the missing property: ${err.replace('Missing required property: ', '')}`,
-              affectedUrls: [url],
-            });
-            score -= 5;
-          }
-        }
-      }
-      
-      // Process @graph
-      if (item['@graph'] && Array.isArray(item['@graph'])) {
-        item['@graph'].forEach(processItem);
-      }
-    };
-    
-    if (Array.isArray(data)) {
-      data.forEach(processItem);
-    } else {
-      processItem(data);
-    }
-  });
-  
-  // Check for essential schemas
-  const hasOrganization = foundTypes.has('Organization');
-  const hasWebSite = foundTypes.has('WebSite');
-  const hasWebPage = foundTypes.has('WebPage');
-  const hasBreadcrumb = foundTypes.has('BreadcrumbList');
-  const hasArticle = foundTypes.has('Article') || foundTypes.has('BlogPosting') || foundTypes.has('NewsArticle');
-  const hasProduct = foundTypes.has('Product');
-  const hasFAQ = foundTypes.has('FAQPage');
-  const hasHowTo = foundTypes.has('HowTo');
-  const hasLocalBusiness = foundTypes.has('LocalBusiness');
-  const hasEvent = foundTypes.has('Event');
-  const hasReview = foundTypes.has('Review') || foundTypes.has('AggregateRating');
-  
-  // Generate recommendations
-  if (!hasOrganization) {
-    recommendations.push('Add Organization schema for brand knowledge panel');
-    score -= 5;
-  }
-  
-  if (!hasWebSite) {
-    recommendations.push('Add WebSite schema with SearchAction for sitelinks search box');
-    score -= 3;
-  }
-  
-  if (!hasBreadcrumb) {
-    recommendations.push('Add BreadcrumbList schema for breadcrumb rich results');
-    score -= 3;
-  }
-  
-  // Check for page-type specific schemas
-  if (!hasArticle && !hasProduct && !hasFAQ && !hasHowTo && !hasLocalBusiness && !hasEvent) {
-    recommendations.push('Add page-specific schema (Article, Product, FAQ, HowTo, etc.)');
-    score -= 10;
-    
-    issues.push({
-      code: 'SCHEMA_NO_RICH_RESULT_TYPE',
-      severity: 'warning',
-      category: 'technical',
-      title: 'No rich result schema found',
-      description: 'Page lacks schema types that enable rich results (Article, Product, FAQ, etc.)',
-      impact: 'Missing opportunity for enhanced search results',
-      howToFix: 'Add appropriate schema type for your content (e.g., Article for blog posts, Product for products, FAQ for Q&A)',
-      affectedUrls: [url],
-    });
-  }
-  
-  // Check for no schema at all
-  if (schemas.length === 0) {
-    score -= 20;
-    issues.push({
-      code: 'SCHEMA_NONE_FOUND',
-      severity: 'warning',
-      category: 'technical',
-      title: 'No structured data found',
-      description: 'No JSON-LD structured data detected on the page.',
-      impact: 'Missing rich results, reduced AI understanding, lower search visibility',
-      howToFix: 'Add JSON-LD structured data. At minimum, include Organization and WebPage schemas.',
-      affectedUrls: [url],
-    });
-  }
-  
-  // AI-specific recommendations
-  if (schemas.length > 0 && !hasFAQ) {
-    recommendations.push('Consider adding FAQPage schema - excellent for AI search citations');
-  }
-  
-  if (!foundTypes.has('Person') && hasArticle) {
-    recommendations.push('Use Person schema for authors to improve E-E-A-T signals');
-  }
-  
-  return {
-    score: Math.max(0, Math.min(100, score)),
-    schemas,
-    hasOrganization,
-    hasWebSite,
-    hasWebPage,
-    hasBreadcrumb,
-    hasArticle,
-    hasProduct,
-    hasFAQ,
-    hasHowTo,
-    hasLocalBusiness,
-    hasEvent,
-    hasReview,
-    issues,
-    recommendations,
-  };
-}
-
-/**
- * Generate recommended schema for a page type
- */
-export function generateSchemaTemplate(pageType: 'article' | 'product' | 'faq' | 'local-business' | 'website', options: {
-  siteName: string;
-  siteUrl: string;
-  authorName?: string;
-  organizationName?: string;
-}): string {
-  const { siteName, siteUrl, authorName, organizationName } = options;
-  
-  const baseSchemas = [
-    {
-      '@context': 'https://schema.org',
-      '@type': 'Organization',
-      name: organizationName || siteName,
-      url: siteUrl,
-      logo: `${siteUrl}/logo.png`,
-    },
-    {
-      '@context': 'https://schema.org',
-      '@type': 'WebSite',
-      name: siteName,
-      url: siteUrl,
-      potentialAction: {
-        '@type': 'SearchAction',
-        target: {
-          '@type': 'EntryPoint',
-          urlTemplate: `${siteUrl}/search?q={search_term_string}`,
-        },
-        'query-input': 'required name=search_term_string',
-      },
-    },
-  ];
-  
-  const pageSchemas: Record<string, object> = {
-    article: {
-      '@context': 'https://schema.org',
-      '@type': 'Article',
-      headline: '{{title}}',
-      description: '{{description}}',
-      image: '{{image_url}}',
-      datePublished: '{{date_published}}',
-      dateModified: '{{date_modified}}',
-      author: {
-        '@type': 'Person',
-        name: authorName || '{{author_name}}',
-        url: '{{author_url}}',
-      },
-      publisher: {
-        '@type': 'Organization',
-        name: organizationName || siteName,
-        logo: {
-          '@type': 'ImageObject',
-          url: `${siteUrl}/logo.png`,
-        },
-      },
-    },
-    product: {
-      '@context': 'https://schema.org',
-      '@type': 'Product',
-      name: '{{product_name}}',
-      description: '{{product_description}}',
-      image: '{{product_image}}',
-      brand: {
-        '@type': 'Brand',
-        name: '{{brand_name}}',
-      },
-      sku: '{{sku}}',
-      offers: {
-        '@type': 'Offer',
-        price: '{{price}}',
-        priceCurrency: 'USD',
-        availability: 'https://schema.org/InStock',
-        url: '{{product_url}}',
-      },
-      aggregateRating: {
-        '@type': 'AggregateRating',
-        ratingValue: '{{rating}}',
-        reviewCount: '{{review_count}}',
-      },
-    },
-    faq: {
-      '@context': 'https://schema.org',
-      '@type': 'FAQPage',
-      mainEntity: [
-        {
-          '@type': 'Question',
-          name: '{{question_1}}',
-          acceptedAnswer: {
-            '@type': 'Answer',
-            text: '{{answer_1}}',
-          },
-        },
-        {
-          '@type': 'Question',
-          name: '{{question_2}}',
-          acceptedAnswer: {
-            '@type': 'Answer',
-            text: '{{answer_2}}',
-          },
-        },
-      ],
-    },
-    'local-business': {
-      '@context': 'https://schema.org',
-      '@type': 'LocalBusiness',
-      name: '{{business_name}}',
-      description: '{{description}}',
-      url: siteUrl,
-      telephone: '{{phone}}',
-      address: {
-        '@type': 'PostalAddress',
-        streetAddress: '{{street}}',
-        addressLocality: '{{city}}',
-        addressRegion: '{{state}}',
-        postalCode: '{{zip}}',
-        addressCountry: '{{country}}',
-      },
-      geo: {
-        '@type': 'GeoCoordinates',
-        latitude: '{{latitude}}',
-        longitude: '{{longitude}}',
-      },
-      openingHoursSpecification: [
-        {
-          '@type': 'OpeningHoursSpecification',
-          dayOfWeek: ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday'],
-          opens: '09:00',
-          closes: '17:00',
-        },
-      ],
-    },
-    website: baseSchemas[1],
-  };
-  
-  const result: object[] = [...baseSchemas];
-  if (pageType !== 'website' && pageSchemas[pageType]) {
-    result.push(pageSchemas[pageType]);
-  }
-  
-  return JSON.stringify(result, null, 2);
-}