debug-chapter-detection.js

const path = require('path');
const fs = require('fs').promises;

async function debugChapterDetection() {
  const filePath = path.join(__dirname, 'web/data/samples/Blood and Memory - Bonus Scenes - Alyssa Evans.docx');
  
  console.log('=== DEBUGGING CHAPTER DETECTION ===');
  console.log('File:', filePath);
  console.log('');
  
  try {
    // Step 1: Convert DOCX to HTML
    const mammoth = require('mammoth');
    const arrayBuffer = await fs.readFile(filePath);
    const result = await mammoth.convertToHtml({ buffer: arrayBuffer });
    
    console.log('--- RAW HTML OUTPUT FROM MAMMOTH ---');
    console.log(result.value.substring(0, 2000)); // First 2000 chars
    console.log('...');
    console.log('');
    
    // Step 2: Show what htmlToMarkdownStructure produces
    const htmlToMarkdownStructure = (html) => {
      // STEP 1: Normalize multi-line headings FIRST
      html = html.replace(/<h([1-6])([^>]*)>([\s\S]*?)<\/h\1>/gi, (match, level, attrs, inner) => {
        const cleanInner = inner.replace(/<[^>]*>/g, '').replace(/\s+/g, ' ').trim();
        return `<h${level}${attrs}>${cleanInner}</h${level}>`;
      });
      
      // STEP 2: Split heading content if first paragraph is inlined WITHOUT proper separation
      html = html.replace(/<h([1-6])([^>]*)>(Chapter\s+\d+)([A-Z][a-z][^<]*)<\/h\1>/g, (m, lvl, attrs, cap, rest) => {
        return `<h${lvl}${attrs}>${cap}</h${lvl}>\n<p>${rest}</p>`;
      });
      html = html.replace(/<h([1-6])([^>]*)>((?:Chapter|Ch\.?|Part|Section)\s+[A-Za-z0-9ivxlcdm]+)([A-Z][a-z][^<]*)<\/h\1>/g, (m, lvl, attrs, cap, rest) => {
        return `<h${lvl}${attrs}>${cap}</h${lvl}>\n<p>${rest}</p>`;
      });
      // Only split special chapters when NO subtitle punctuation is present
      html = html.replace(/<h([1-6])([^>]*)>((?:Bonus|Epilogue|Prologue|Final|Extra)\s+Chapter)(?![\s:.\-])([A-Z][a-z][^<]*)<\/h\1>/g, (m, lvl, attrs, cap, rest) => {
        return `<h${lvl}${attrs}>${cap}</h${lvl}>\n<p>${rest}</p>`;
      });
      
      return html
        .replace(/<h1[^>]*>(.*?)<\/h1>/gi, (_m, inner) => `# ${inner.trim()}\n\n`)
        .replace(/<h2[^>]*>(.*?)<\/h2>/gi, (_m, inner) => `## ${inner.trim()}\n\n`)
        .replace(/<h3[^>]*>(.*?)<\/h3>/gi, (_m, inner) => `### ${inner.trim()}\n\n`)
        .replace(/<h4[^>]*>(.*?)<\/h4>/gi, (_m, inner) => `#### ${inner.trim()}\n\n`)
        .replace(/<h5[^>]*>(.*?)<\/h5>/gi, (_m, inner) => `##### ${inner.trim()}\n\n`)
        .replace(/<h6[^>]*>(.*?)<\/h6>/gi, (_m, inner) => `###### ${inner.trim()}\n\n`)
        .replace(/<p[^>]*>(.*?)<\/p>/gi, (match, content) => {
          const text = content.trim();
          const chapterPatterns = [
            /^chapter\s+/i,
            /^ch\.?\s*\d+/i,
            /^\d+\.\s*[^.]/,
            /^part\s+\d+/i,
            /^section\s+\d+/i,
            /^bonus\s+chapter/i,
            /^epilogue\s+chapter/i,
            /^prologue\s+chapter/i,
            /^final\s+chapter/i,
            /^extra\s+chapter/i
          ];
          
          const looksLikeChapter = chapterPatterns.some(pattern => pattern.test(text)) && text.length < 150;
          
          if (looksLikeChapter) {
            return text + '\n\n';
          }
          return text + '\n\n';
        })
        .replace(/<[^>]*>/g, '')
        .replace(/^(#\s+Chapter\s+\d+)([A-Z][a-z]+)/gm, (_, h, rest) => `${h}\n\n${rest}`)
        .replace(/&nbsp;/g, ' ')
        .replace(/&lt;/g, '<')
        .replace(/&gt;/g, '>')
        .replace(/&amp;/g, '&')
        .replace(/[ \t]+/g, ' ')
        .replace(/\n{3,}/g, '\n\n')
        .trim();
    };
    
    const markdown = htmlToMarkdownStructure(result.value);
    
    console.log('--- CONVERTED MARKDOWN STRUCTURE ---');
    console.log(markdown.substring(0, 3000)); // First 3000 chars
    console.log('...');
    console.log('');
    console.log('Full markdown length:', markdown.length, 'characters');
    console.log('');
    
    // Step 3: Show all lines that might be chapter markers
    console.log('--- POTENTIAL CHAPTER MARKER LINES ---');
    const lines = markdown.split(/\r?\n/);
    lines.forEach((line, i) => {
      const trimmed = line.trim();
      if (trimmed && (
        /^#+\s*chapter/i.test(trimmed) ||
        /^chapter/i.test(trimmed) ||
        /^bonus/i.test(trimmed) ||
        /^epilogue/i.test(trimmed) ||
        /^prologue/i.test(trimmed) ||
        /^part\s+\d+/i.test(trimmed) ||
        /^ch\./i.test(trimmed)
      )) {
        console.log(`Line ${i + 1}: "${trimmed}"`);
      }
    });
    
  } catch (error) {
    console.error('Error:', error);
  }
}

debugChapterDetection();