/**
 * Markdown paste detection for the docs editor.
 *
 * Conservatively detects whether pasted plain text is markdown.
 * Requires 2+ signals or 1 strong signal (heading, code block, table)
 * to avoid false positives on normal text.
 *
 * Also detects when clipboard HTML contains unrendered markdown link
 * syntax, indicating the source didn't convert markdown to rich text.
 */

const PATTERNS: readonly RegExp[] = [
  /^#{1,6}\s+\S/m,           // ATX headings (strong)
  /^\s*[-*+]\s+\S/m,         // Unordered lists
  /^\s*\d+\.\s+\S/m,         // Ordered lists
  /^\s*```/m,                 // Fenced code blocks (strong)
  /^\s*>\s+\S/m,             // Blockquotes
  /\[.+?\]\(.+?\)/,          // Links
  /!\[.*?\]\(.+?\)/,         // Images
  /^\s*[-*_]{3,}\s*$/m,      // Horizontal rules
  /^\s*\|.+\|.+\|/m,        // Tables
  /^\s*- \[([ xX])\]/m,     // Task lists
];

// Strong solo signals — a single match is enough
const HEADING_RE = /^#{1,6}\s+\S/m;
const CODE_FENCE_RE = /^\s*```/m;
const TABLE_WITH_SEPARATOR_RE = /^\s*\|.+\|.+\|\s*\n\s*\|[\s:|-]+\|/m;
const MARKDOWN_LINK_RE = /\[.+?\]\(.+?\)/;
const TASK_LIST_RE = /^\s*- \[([ xX])\]/m;

export function looksLikeMarkdown(text: string): boolean {
  if (!text || !text.trim()) return false;

  let signals = 0;
  for (const p of PATTERNS) {
    if (p.test(text)) signals++;
  }

  if (signals >= 2) return true;

  // Strong solo signals
  if (HEADING_RE.test(text)) return true;
  if (CODE_FENCE_RE.test(text)) return true;
  if (TABLE_WITH_SEPARATOR_RE.test(text)) return true;
  if (MARKDOWN_LINK_RE.test(text)) return true;
  if (TASK_LIST_RE.test(text)) return true;

  return false;
}

/**
 * Check whether clipboard HTML contains unrendered markdown link syntax.
 * When apps copy content, they may include both text/html and text/plain.
 * If the HTML contains literal `[text](url)` strings (not rendered as <a> tags),
 * the plain text markdown path should be preferred.
 */
// Looser task list pattern for HTML context (no start-of-line anchor)
const RAW_TASK_LIST_IN_HTML = /- \[([ xX])\]\s/;

export function htmlContainsRawMarkdown(html: string): boolean {
  if (!html) return false;
  // Strip actual <a> tags and their content to avoid false positives,
  // then check if raw markdown link syntax remains in the text content.
  const stripped = html.replace(/<a\b[^>]*>.*?<\/a>/gi, '');
  return MARKDOWN_LINK_RE.test(stripped) || RAW_TASK_LIST_IN_HTML.test(stripped);
}