flutter-rimworld-modman/lib/bbcode_converter.dart

/// Utility class to convert BBCode to Markdown for RimWorld mod descriptions
import 'dart:math' as math;

class BBCodeConverter {
  /// Converts BBCode formatted text to Markdown format
  static String toMarkdown(String bbcode) {
    if (bbcode.isEmpty) return '';
    
    // First, normalize line endings and escape any literal backslashes
    String result = bbcode.replaceAll('\r\n', '\n').replaceAll('\\', '\\\\');
    
    // Ensure paragraphs - double newlines between paragraphs
    // First normalize any consecutive newlines to a single one
    result = result.replaceAll(RegExp(r'\n+'), '\n');
    
    // Then add empty line after each paragraph where needed
    result = result.replaceAll('.\n', '.\n\n');
    result = result.replaceAll('!\n', '!\n\n');
    result = result.replaceAll('?\n', '?\n\n');
    
    // Fix unclosed tags - RimWorld descriptions often have unclosed tags
    final List<String> tagTypes = ['b', 'i', 'color', 'size', 'url', 'code', 'quote'];
    for (final tag in tagTypes) {
      final openCount = '[${tag}'.allMatches(result).length;
      final closeCount = '[/${tag}]'.allMatches(result).length;
      if (openCount > closeCount) {
        result = result + '[/${tag}]' * (openCount - closeCount);
      }
    }
    
    // URLs
    // [url=http://example.com]text[/url] -> [text](http://example.com)
    result = RegExp(r'\[url=([^\]]+)\](.*?)\[/url\]', dotAll: true)
        .allMatches(result)
        .fold(result, (prev, match) {
          final url = match.group(1);
          final text = match.group(2);
          return prev.replaceFirst(
            match.group(0)!, 
            '[$text]($url)'
          );
        });
    
    // Simple URL [url]http://example.com[/url] -> <http://example.com>
    result = result.replaceAllMapped(
      RegExp(r'\[url\](.*?)\[/url\]', dotAll: true), 
      (match) => '<${match.group(1)}>'
    );
    
    // Bold
    result = result.replaceAll('[b]', '**').replaceAll('[/b]', '**');
    
    // Italic
    result = result.replaceAll('[i]', '_').replaceAll('[/i]', '_');
    
    // Headers - ensure they start on their own line
    result = result.replaceAllMapped(
      RegExp(r'\[h1\](.*?)\[/h1\]', dotAll: true), 
      (match) => '\n\n# ${match.group(1)?.trim()}\n\n'
    );
    result = result.replaceAllMapped(
      RegExp(r'\[h2\](.*?)\[/h2\]', dotAll: true), 
      (match) => '\n\n## ${match.group(1)?.trim()}\n\n'
    );
    result = result.replaceAllMapped(
      RegExp(r'\[h3\](.*?)\[/h3\]', dotAll: true), 
      (match) => '\n\n### ${match.group(1)?.trim()}\n\n'
    );
    
    // Lists - handle nested lists too
    result = result.replaceAll('[list]', '\n').replaceAll('[/list]', '\n');
    
    // Handle list items - giving them proper indentation
    result = result.replaceAllMapped(
      RegExp(r'\[\*\](.*?)(?=\[\*\]|\[/list\]|$)', dotAll: true), 
      (match) {
        final content = match.group(1)?.trim() ?? '';
        return '\n* $content\n'; 
      }
    );
    
    // Color - convert to bold since Markdown doesn't support color
    result = result.replaceAllMapped(
      RegExp(r'\[color=([^\]]+)\](.*?)\[/color\]', dotAll: true), 
      (match) {
        final content = match.group(2) ?? '';
        if (content.trim().isEmpty) return '';
        return '**${match.group(2)}**';
      }
    );
    
    // Images
    // [img]url[/img] -> ![Image](url)
    result = result.replaceAllMapped(
      RegExp(r'\[img\](.*?)\[/img\]', dotAll: true), 
      (match) => '![Image](${match.group(1)})'
    );
    
    // Image with size [img width=300]url[/img] -> ![Image](url)
    result = result.replaceAllMapped(
      RegExp(r'\[img[^\]]*width=(\d+)[^\]]*\](.*?)\[/img\]', dotAll: true),
      (match) => '![Image](${match.group(2)})'
    );

    // Tables - convert tables to markdown tables
    if (result.contains('[table]')) {
      // Process tables
      final tableRegex = RegExp(r'\[table\](.*?)\[/table\]', dotAll: true);
      final tables = tableRegex.allMatches(result);
      
      for (final tableMatch in tables) {
        final tableContent = tableMatch.group(1) ?? '';
        final rows = RegExp(r'\[tr\](.*?)\[/tr\]', dotAll: true).allMatches(tableContent);
        
        // Only process tables that have rows
        if (rows.isEmpty) {
          result = result.replaceFirst(tableMatch.group(0)!, '');
          continue;
        }
        
        final markdownTable = StringBuffer('\n');
        var isFirstRow = true;
        
        // First determine the number of columns by examining all rows
        int maxColumns = 0;
        for (final rowMatch in rows) {
          final rowContent = rowMatch.group(1) ?? '';
          final cellCount = RegExp(r'\[td\]').allMatches(rowContent).length;
          maxColumns = math.max(maxColumns, cellCount);
        }
        
        // Ensure we have at least 1 column
        maxColumns = math.max(1, maxColumns);
        
        for (final rowMatch in rows) {
          final rowContent = rowMatch.group(1) ?? '';
          final cells = RegExp(r'\[td\](.*?)\[/td\]', dotAll: true).allMatches(rowContent);
          
          if (cells.isEmpty) continue;
          
          final rowBuffer = StringBuffer('| ');
          
          int cellsAdded = 0;
          for (final cellMatch in cells) {
            final cellContent = cellMatch.group(1)?.trim() ?? '';
            // Clean up any newlines inside cell content
            final cleanCell = cellContent.replaceAll('\n', ' ').trim();
            rowBuffer.write('$cleanCell | ');
            cellsAdded++;
          }
          
          // Add empty cells if needed to maintain table structure
          while (cellsAdded < maxColumns) {
            rowBuffer.write(' | ');
            cellsAdded++;
          }
          
          markdownTable.writeln(rowBuffer.toString());
          
          // Add header separator after first row
          if (isFirstRow) {
            final headerRow = StringBuffer('| ');
            for (int i = 0; i < maxColumns; i++) {
              headerRow.write('--- | ');
            }
            markdownTable.writeln(headerRow.toString());
            isFirstRow = false;
          }
        }
        
        markdownTable.write('\n');
        result = result.replaceFirst(tableMatch.group(0)!, markdownTable.toString());
      }
    }
    
    // Size - remove since Markdown doesn't directly support font size
    result = result.replaceAllMapped(
      RegExp(r'\[size=[^\]]+\](.*?)\[/size\]', dotAll: true), 
      (match) => match.group(1) ?? ''
    );
    
    // Code
    result = result.replaceAll('[code]', '\n```\n').replaceAll('[/code]', '\n```\n');
    
    // Quote
    result = result.replaceAllMapped(
      RegExp(r'\[quote\](.*?)\[/quote\]', dotAll: true), 
      (match) {
        final content = match.group(1)?.trim() ?? '';
        if (content.isEmpty) return '';
        return '\n> ${content.replaceAll('\n', '\n> ')}\n\n';
      }
    );
    
    // Handle any remaining custom BBCode tags - just remove them
    result = result.replaceAllMapped(
      RegExp(r'\[([a-zA-Z0-9_]+)(?:=[^\]]+)?\](.*?)\[/\1\]', dotAll: true),
      (match) => match.group(2) ?? ''
    );
    
    // Handle RimWorld-specific patterns
    // [h1] without closing tag is common
    result = result.replaceAllMapped(
      RegExp(r'\[h1\]([^\[]+)'), 
      (match) => '\n\n# ${match.group(1)?.trim()}\n\n'
    );
    
    // Convert simple newlines to double newlines for proper markdown rendering
    // But avoid doing this for lines that are already marked up as headings, lists, or tables
    result = result.replaceAllMapped(
      RegExp(r'([^\n#*>|])\n([^\n#*>|-])'),
      (match) => '${match.group(1)}\n\n${match.group(2)}'
    );
    
    // Normalize multiple spaces
    result = result.replaceAll(RegExp(r' {2,}'), ' ');
    
    // Remove excessive newlines (more than 2 consecutive)
    result = result.replaceAll(RegExp(r'\n{3,}'), '\n\n');
    
    // Ensure document starts and ends cleanly
    result = result.trim();
    
    return result;
  }
}