astro_avtourist/frontend/scripts/analyze-posts.ts

import PocketBase from 'pocketbase';

const PB_URL = 'http://127.0.0.1:8090';
const pb = new PocketBase(PB_URL);

interface Post {
  id: string;
  title: string;
  content: string;
  slug: string;
  category: string;
  draft: boolean;
}

interface AnalysisResult {
  title: string;
  slug: string;
  category: string;
  hasClaim: boolean;
  hasTakeaway: boolean;
  hasPronouns: boolean;
  hasProof: boolean;
  h2Issues: string[];
  problems: string[];
}

function analyzePost(post: Post): AnalysisResult {
  const result: AnalysisResult = {
    title: post.title,
    slug: post.slug,
    category: post.category || 'unknown',
    hasClaim: false,
    hasTakeaway: false,
    hasPronouns: false,
    hasProof: false,
    h2Issues: [],
    problems: []
  };

  const content = post.content || '';
  
  // Check for Claim patterns directly in content (including HTML)
  // Look for new first paragraph that starts with Claim-like content
  // Check first 1500 chars to catch posts where Claim may have been inserted
  const claimPatterns = [
    { pattern: /За\s+\d+/i, name: 'За год' },
    { pattern: /В\s+\d+\s+(год|месяц)/i, name: 'В год/месяц' },
    { pattern: /\d+%/i, name: 'Процент' },
    { pattern: /Вы можете/i, name: 'Вы можете' },
  ];
  
  const prefix = content.substring(0, 3000);
  result.hasClaim = claimPatterns.some(p => p.pattern.test(prefix));
  
  // Proof: contains numbers
  const numberPatterns = [/\d+%/g, /\d+\s+(лет|месяцев|дней|тысяч|рублей)/gi];
  result.hasProof = numberPatterns.some(p => p.test(content));
  
  // Pronouns check
  const pronouns = /\b(мы|наш|наша|наши|их|он|она|оно|этот|эта|эти)\b/gi;
  result.hasPronouns = pronouns.test(content);
  
  // Takeaway check
  const takeawayPatterns = [
    /что\s+делать/i,
    /обращайтесь/i,
    /позвоните/i,
    /закажите/i,
    /получите/i,
    /напишите/i,
    /свяжитесь/i
  ];
  result.hasTakeaway = takeawayPatterns.some(p => p.test(content));
  
  // H2 analysis - check all H2s in content
  const h2Matches = content.match(/<h2[^>]*>[^<]+<\/h2>/gi) || [];
  const badH2s = ['проблема', 'подход', 'ситуация', 'что было', 'решение', 'как помочь', 'поможет'];
  h2Matches.forEach(h2 => {
    const h2Text = h2.toLowerCase();
    if (badH2s.some(b => h2Text.includes(b))) {
      result.h2Issues.push(h2.replace(/<[^>]+>/g, ''));
    }
  });
  
  // Collect problems
  if (!result.hasClaim) result.problems.push('Нет Claim в первом абзаце');
  if (!result.hasTakeaway) result.problems.push('Нет Takeaway');
  if (result.hasPronouns) result.problems.push('Есть местоимения');
  if (!result.hasProof) result.problems.push('Нет конкретных цифр');
  if (result.h2Issues.length > 0) result.problems.push(`Размытые H2: ${result.h2Issues.join(', ')}`);
  
  return result;
}

async function main() {
  console.log('📊 Анализ постов блога по методологии Answer Unit\n');
  console.log('='.repeat(60));
  
  try {
    const postsResult = await pb.collection('posts').getList(1, 500);
    const posts = postsResult.items as unknown as Post[];
    
    console.log(`Найдено постов: ${posts.length}\n`);
    
    // Filter for non-draft posts
    const activePosts = posts.filter(p => !p.draft);
    console.log(`Опубликованных постов: ${activePosts.length}\n`);
    
    const results = activePosts.map(analyzePost);
    
    // Stats
    const withClaim = results.filter(r => r.hasClaim).length;
    const withTakeaway = results.filter(r => r.hasTakeaway).length;
    const withPronouns = results.filter(r => r.hasPronouns).length;
    const withProof = results.filter(r => r.hasProof).length;
    const withH2Issues = results.filter(r => r.h2Issues.length > 0).length;
    const problemCount = results.filter(r => r.problems.length > 0).length;
    
    const n = activePosts.length;
    console.log('📈 ОБЩАЯ СТАТИСТИКА:');
    console.log('-'.repeat(40));
    console.log(`  ✓ Есть Claim: ${withClaim}/${n} (${n > 0 ? Math.round(withClaim/n*100) : 0}%)`);
    console.log(`  ✓ Есть Takeaway: ${withTakeaway}/${n} (${n > 0 ? Math.round(withTakeaway/n*100) : 0}%)`);
    console.log(`  ✗ Есть местоимения: ${withPronouns}/${n} (${n > 0 ? Math.round(withPronouns/n*100) : 0}%)`);
    console.log(`  ✓ Есть Proof (цифры): ${withProof}/${n} (${n > 0 ? Math.round(withProof/n*100) : 0}%)`);
    console.log(`  ✗ Проблемы с H2: ${withH2Issues}/${n}`);
    console.log(`  ⚠️ Постов с проблемами: ${problemCount}/${n}\n`);
    
    // Problem posts
    const problemPosts = results.filter(r => r.problems.length > 0);
    
    if (problemPosts.length > 0) {
      console.log('❌ ПОСТЫ С ПРОБЛЕМАМИ:');
      console.log('-'.repeat(40));
      problemPosts.forEach((r, i) => {
        console.log(`\n${i+1}. ${r.title}`);
        console.log(`   slug: ${r.slug}`);
        console.log(`   category: ${r.category}`);
        r.problems.forEach(p => console.log(`   - ${p}`));
      });
    }
    
    // Good posts
    const goodPosts = results.filter(r => r.problems.length === 0);
    if (goodPosts.length > 0) {
      console.log('\n\n✅ ПОСТЫ БЕЗ ПРОБЛЕМ:');
      console.log('-'.repeat(40));
      goodPosts.forEach(r => {
        console.log(`  • ${r.title}`);
      });
    }
    
  } catch (error) {
    console.error('Ошибка:', error);
  }
}

main();