src/lib/markdownUtils.ts

/**
 * Shared markdown pre-processing utilities used across content pages.
 */

/**
 * Repairs unclosed or misplaced `**` bold markers.
 *
 * Common source examples:
 *   **Parameters CREATE TABLE: value here
 *   **Example CREATE TABLE movies (
 *   **Explanation In the above example ...
 */
export function fixMarkdownBold(text: string): string {
  return text
    .split('\n')
    .map(line => {
      const match = line.match(/^(\s*)\*\*(.*)$/);
      if (!match) return line;

      const [, indent, inner] = match;
      if (inner.includes('**')) return line;

      const colonIdx = inner.indexOf(':');
      if (colonIdx !== -1) {
        return `${indent}**${inner.slice(0, colonIdx)}**:${inner.slice(colonIdx + 1)}`;
      }

      const wordLabelMatch = inner.match(/^(Example|Explanation|Definition|Syntax|Note|Tip|Warning)\b(.*)$/i);
      if (wordLabelMatch) {
        const [, label, rest] = wordLabelMatch;
        return `${indent}**${label}**${rest}`;
      }

      const parenIdx = inner.indexOf('(');
      if (parenIdx !== -1) {
        return `${indent}**${inner.slice(0, parenIdx).trimEnd()}** ${inner.slice(parenIdx).trimStart()}`;
      }

      const spaceIdx = inner.indexOf(' ');
      if (spaceIdx !== -1) {
        return `${indent}**${inner.slice(0, spaceIdx)}** ${inner.slice(spaceIdx + 1)}`;
      }

      return `${indent}**${inner}**`;
    })
    .join('\n');
}

/**
 * Removes common stray artifact patterns from converted markdown.
 */
export function cleanMarkdown(text: string): string {
  return text
    .replace(/\u0026#x09;/g, '')
    .replace(/\\&/g, '&')
    .replace(/--\*\*/g, '')
    .replace(/^\s*\*\*\s*$/gm, '')
    .replace(/^[*\-]+\s*$/gm, '')
    .replace(/\n{3,}/g, '\n\n')
    .trim();
}

/**
 * Robust ASCII table → GFM Markdown converter.
 * Handles all known divider styles: +---+ , ~---~ , +===+, ----+----, and pipe rows.
 * Strips wrapping code fences that sometimes surround tables in the raw data.
 * Replaces empty / NULL cells with "N/A".
 */
export function processAsciiTableToMarkdown(text: string): string {
  const lines = text.split('\n');
  const result: string[] = [];
  let inTable = false;
  let inTableBlock = false;
  let tableRows: string[] = [];

  // Detect any ASCII table border line
  const isDividerLine = (s: string) => /^[+~|][-=~+]+[+~|]$/.test(s) || /^[-=]{3,}$/.test(s);

  const cleanCell = (cell: string) => {
    const v = cell.trim();
    return (v === '' || /^null$/i.test(v)) ? 'N/A' : v;
  };

  const flushTable = () => {
    if (tableRows.length === 0) return;
    // First row = header
    result.push('');
    result.push(tableRows[0]);
    const colCount = tableRows[0].split('|').length - 2;
    result.push('|' + Array(Math.max(1, colCount)).fill(' --- ').join('|') + '|');
    for (let i = 1; i < tableRows.length; i++) result.push(tableRows[i]);
    result.push('');
    tableRows = [];
  };

  for (let i = 0; i < lines.length; i++) {
    const rawLine = lines[i].trim();
    // Strip surrounding backticks for analysis
    const line = rawLine.replace(/^[\`]+|[\`]+$/g, '').trim();

    // Check if this is an opening code block that contains a table
    if (rawLine.startsWith('```') && !inTableBlock) {
      let blockContainsTable = false;
      for (let j = i + 1; j < lines.length; j++) {
        const peekRaw = lines[j].trim();
        if (peekRaw.startsWith('```')) break; // End of block
        const peek = peekRaw.replace(/^[\`]+|[\`]+$/g, '').trim();
        if (isDividerLine(peek)) {
          blockContainsTable = true;
          break;
        }
      }
      if (blockContainsTable) {
        inTableBlock = true;
        continue; // Skip the opening ```
      }
    }

    // Check if this is the closing code block of a table block
    if (rawLine.startsWith('```') && inTableBlock) {
      if (inTable) {
        flushTable();
        inTable = false;
      }
      inTableBlock = false;
      continue; // Skip the closing ```
    }

    if (isDividerLine(line)) {
      inTable = true;
      continue; // skip divider
    }

    // A row is a data row if it contains a pipe.
    const isDataRow = line.includes('|') && line.startsWith('|') && line.endsWith('|');

    if (isDataRow) {
      inTable = true;
      const parts = line.split('|').slice(1, -1);
      tableRows.push(`| ${parts.map(p => cleanCell(p)).join(' | ')} |`);
    } else if (inTable && line === '') {
      continue; // blank inside table
    } else {
      if (inTable) {
        flushTable();
        inTable = false;
      }
      result.push(rawLine);
    }
  }
  if (inTable) flushTable();
  return result.join('\n');
}