Momento/memento-note/lib/editor/markdown-export.ts

/**
 * markdown-export.ts
 * Utilities for TipTap HTML ↔ Markdown conversion.
 *
 * Uses:
 *   - turndown (+ turndown-plugin-gfm) : HTML → Markdown
 *   - marked                           : Markdown → HTML
 */

import TurndownService from 'turndown'
import { tables, taskListItems, strikethrough } from 'turndown-plugin-gfm'
import { marked } from 'marked'

// ── Markdown heuristic detection ────────────────────────────────────────────

const MARKDOWN_PATTERNS = [
  /^#{1,6}\s/m,           // headings
  /^\s*[-*+]\s/m,         // unordered list
  /^\s*\d+\.\s/m,         // ordered list
  /^\s*>\s/m,             // blockquote
  /^```/m,                // code fence
  /`[^`]+`/,              // inline code
  /\*\*[^*]+\*\*/,        // bold
  /\*[^*]+\*/,            // italic
  /^[|].+[|]/m,           // table
  /\[.+\]\(.+\)/,         // link
  /!\[.+\]\(.+\)/,        // image
  /~~[^~]+~~/,            // strikethrough
]

/**
 * Returns true if the given plain text looks like it contains Markdown syntax.
 * Used by the paste handler to decide whether to convert before inserting.
 */
export function looksLikeMarkdown(text: string): boolean {
  if (!text || text.trim().length < 3) return false
  return MARKDOWN_PATTERNS.some((re) => re.test(text))
}

// ── Turndown service factory ─────────────────────────────────────────────────

function createTurndownService(): TurndownService {
  const td = new TurndownService({
    headingStyle: 'atx',
    hr: '---',
    bulletListMarker: '-',
    codeBlockStyle: 'fenced',
    fence: '```',
    emDelimiter: '_',
    strongDelimiter: '**',
    linkStyle: 'inlined',
    linkReferenceStyle: 'full',
  })

  // GFM plugins: tables + task lists + strikethrough
  td.use([tables, taskListItems, strikethrough])

  // Custom rule: liveBlock → HTML comment
  td.addRule('liveBlock', {
    filter(node) {
      return (
        node.nodeName === 'DIV' &&
        (node as HTMLElement).hasAttribute('data-live-block')
      )
    },
    replacement(_content, node) {
      const el = node as HTMLElement
      const sourceNoteId = el.getAttribute('sourcenoteId') || el.getAttribute('sourcenoteId') || el.getAttribute('sourcenoteid') || ''
      const blockId = el.getAttribute('blockId') || el.getAttribute('blockid') || ''
      return `\n\n<!-- live-block: ${sourceNoteId}#${blockId} -->\n\n`
    },
  })

  // Custom rule: structuredViewBlock → HTML comment
  td.addRule('structuredViewBlock', {
    filter(node) {
      return (
        node.nodeName === 'DIV' &&
        (node as HTMLElement).hasAttribute('data-structured-view-block')
      )
    },
    replacement(_content, node) {
      const el = node as HTMLElement
      const attrs: Record<string, string> = {}
      for (const attr of Array.from(el.attributes)) {
        if (attr.name !== 'data-structured-view-block') {
          attrs[attr.name] = attr.value
        }
      }
      return `\n\n<!-- structured-view: ${JSON.stringify(attrs)} -->\n\n`
    },
  })

  return td
}

// Singleton (lazy-init) — safe for server + client usage
let _tdService: TurndownService | null = null
function getTurndownService(): TurndownService {
  if (!_tdService) _tdService = createTurndownService()
  return _tdService
}

// ── Custom node pre-processor ─────────────────────────────────────────────

// Sentinel prefix — alphanumeric only to avoid Markdown escaping by turndown
const SENTINEL_PREFIX = 'MOMENTOBLOCKSENTINEL'

interface BlockPlaceholder {
  key: string
  comment: string
}

/**
 * Pre-process HTML before passing to turndown:
 * - Replace empty custom node divs (liveBlock, structuredViewBlock) with text
 *   placeholders so they survive turndown processing (turndown drops blank nodes
 *   and strips HTML comments).
 * - Return the modified HTML and a map of placeholder → HTML comment.
 */
function preprocessCustomNodes(html: string): { html: string; placeholders: BlockPlaceholder[] } {
  const placeholders: BlockPlaceholder[] = []

  // liveBlock: <div data-live-block="true" sourceNoteId="..." blockId="..."></div>
  let result = html.replace(
    /<div([^>]*?data-live-block[^>]*?)>\s*<\/div>/gi,
    (_match, attrs) => {
      const snId = (attrs.match(/sourcenoteid="([^"]*)"/i) || attrs.match(/sourcenoteid='([^']*)'/i) || [])[1] || ''
      const bId = (attrs.match(/blockid="([^"]*)"/i) || attrs.match(/blockid='([^']*)'/i) || [])[1] || ''
      const key = `${SENTINEL_PREFIX}LIVEBLOCK${placeholders.length}`
      placeholders.push({ key, comment: `<!-- live-block: ${snId}#${bId} -->` })
      return `<p>${key}</p>`
    }
  )

  // structuredViewBlock: <div data-structured-view-block="true" ...></div>
  result = result.replace(
    /<div([^>]*?data-structured-view-block[^>]*?)>\s*<\/div>/gi,
    (_match, attrs) => {
      const attrMap: Record<string, string> = {}
      const attrRe = /(data-[a-z-]+)="([^"]*)"/gi
      let m: RegExpExecArray | null
      while ((m = attrRe.exec(attrs)) !== null) {
        if (m[1] !== 'data-structured-view-block') attrMap[m[1]] = m[2]
      }
      const key = `${SENTINEL_PREFIX}SVBLOCK${placeholders.length}`
      placeholders.push({ key, comment: `<!-- structured-view: ${JSON.stringify(attrMap)} -->` })
      return `<p>${key}</p>`
    }
  )

  return { html: result, placeholders }
}

/**
 * Post-process the markdown output: replace sentinel placeholders with HTML comments.
 */
function postprocessPlaceholders(md: string, placeholders: BlockPlaceholder[]): string {
  let result = md
  for (const { key, comment } of placeholders) {
    result = result.replace(key, `\n\n${comment}\n\n`)
  }
  return result
}

// ── HTML → Markdown ──────────────────────────────────────────────────────────

/**
 * Convert a TipTap-generated HTML string to GitHub-Flavored Markdown.
 * Custom nodes (liveBlock, structuredViewBlock) are serialised as HTML comments.
 */
export function tiptapHTMLToMarkdown(html: string): string {
  if (!html || html.trim() === '') return ''
  const { html: preprocessed, placeholders } = preprocessCustomNodes(html)
  const td = getTurndownService()
  const md = td.turndown(preprocessed).trim()
  return postprocessPlaceholders(md, placeholders).trim()
}

// ── Markdown → HTML ──────────────────────────────────────────────────────────

/**
 * Convert a Markdown string to HTML suitable for injection into TipTap via
 * `editor.commands.setContent(html)`.
 *
 * Uses marked with GFM enabled (tables, task lists, line breaks).
 */
export function markdownToHTML(markdown: string): string {
  if (!markdown || markdown.trim() === '') return ''

  // marked v18+ uses synchronous parse by default when no async tokens
  const html = marked.parse(markdown, {
    gfm: true,
    breaks: false,
  }) as string

  return html
}

// ── Title extraction from Markdown ──────────────────────────────────────────

/**
 * Extract the first H1 title from a Markdown string.
 * Returns null if no H1 is found.
 */
export function extractMarkdownTitle(markdown: string): string | null {
  const match = markdown.match(/^#\s+(.+)/m)
  return match ? match[1].trim() : null
}