@dev-workflow/skill 0.1.0

package/tools/lark/scripts/markdown_to_lark_blocks.mjs

@@ -0,0 +1,397 @@
+import { readFileSync } from 'node:fs'
+
+import { parseArgs, printJson } from './lark_api.mjs'
+
+const MAX_TEXT_LENGTH = 1800
+const MAX_TABLE_CELLS = 45
+// 飞书 docx 单次创建表格 row_size 上限为 9，超过返回 1770001 invalid param。
+const MAX_TABLE_ROWS = 9
+const CODE_LANGUAGE_MAP = {
+  javascript: 3,
+  js: 3,
+  typescript: 3,
+  ts: 3,
+  json: 19,
+  bash: 14,
+  sh: 14,
+  shell: 14
+}
+
+function splitLongText(text) {
+  if (text.length <= MAX_TEXT_LENGTH) return [text]
+
+  const chunks = []
+  let rest = text
+
+  while (rest.length > MAX_TEXT_LENGTH) {
+    let cut = rest.lastIndexOf('\n', MAX_TEXT_LENGTH)
+    if (cut < MAX_TEXT_LENGTH * 0.5) cut = rest.lastIndexOf('。', MAX_TEXT_LENGTH)
+    if (cut < MAX_TEXT_LENGTH * 0.5) cut = MAX_TEXT_LENGTH
+
+    chunks.push(rest.slice(0, cut).trim())
+    rest = rest.slice(cut).trim()
+  }
+
+  if (rest) chunks.push(rest)
+  return chunks
+}
+
+function createTextBlock(text) {
+  return {
+    block_type: 2,
+    text: {
+      elements: createTextElements(text)
+    }
+  }
+}
+
+function createBulletBlock(text) {
+  return {
+    block_type: 12,
+    bullet: {
+      elements: createTextElements(text)
+    }
+  }
+}
+
+function createOrderedBlock(text, sequence) {
+  return {
+    block_type: 13,
+    ordered: {
+      elements: createTextElements(text),
+      style: {
+        sequence: String(sequence)
+      }
+    }
+  }
+}
+
+function createCalloutBlock(text, emoji) {
+  const block = {
+    block_type: 19,
+    callout: { background_color: 1, border_color: 1, emoji_id: emoji || 'memo' }
+  }
+  // 文本作为子块由 writeBlocks 经 descendant API 写入，避免污染 JSON 序列化。
+  Object.defineProperty(block, '__calloutText', { value: text })
+  return block
+}
+
+function createQuoteBlock(text) {
+  const block = { block_type: 34, quote_container: {} }
+  Object.defineProperty(block, '__quoteText', { value: text })
+  return block
+}
+
+function createHeadingBlock(level, text) {
+  const blockType = level === 1 ? 3 : level === 2 ? 4 : 5
+  const key = level === 1 ? 'heading1' : level === 2 ? 'heading2' : 'heading3'
+
+  return {
+    block_type: blockType,
+    [key]: {
+      elements: createTextElements(text)
+    }
+  }
+}
+
+function createTextElements(text) {
+  const elements = []
+  // 同时识别行内代码 `code` 与粗体 **bold**。
+  const pattern = /(`[^`\n]+`)|(\*\*[^*\n]+\*\*)/g
+  let lastIndex = 0
+  let match
+
+  while ((match = pattern.exec(text))) {
+    if (match.index > lastIndex) {
+      elements.push({ text_run: { content: text.slice(lastIndex, match.index) } })
+    }
+
+    const token = match[0]
+    if (token.startsWith('`')) {
+      elements.push({
+        text_run: { content: token.slice(1, -1), text_element_style: { inline_code: true } }
+      })
+    } else {
+      elements.push({
+        text_run: { content: token.slice(2, -2), text_element_style: { bold: true } }
+      })
+    }
+
+    lastIndex = pattern.lastIndex
+  }
+
+  if (lastIndex < text.length) {
+    elements.push({ text_run: { content: text.slice(lastIndex) } })
+  }
+
+  return elements.length ? elements : [{ text_run: { content: '' } }]
+}
+
+function normalizeLanguage(language) {
+  const key = language.trim().toLowerCase()
+  return CODE_LANGUAGE_MAP[key] || CODE_LANGUAGE_MAP.text
+}
+
+function createCodeBlock(language, code) {
+  const block = {
+    block_type: 14,
+    code: {
+      elements: [{ text_run: { content: code } }]
+    }
+  }
+  const languageCode = normalizeLanguage(language)
+
+  if (languageCode) {
+    block.code.style = { language: languageCode }
+  }
+
+  return block
+}
+
+function splitTableRow(line) {
+  const trimmed = line.trim().replace(/^\|/, '').replace(/\|$/, '')
+  // 仅按未转义的 | 切分单元格，再把 \| 还原为字面 |（支持单元格内的联合类型如 'A' \| 'B'）。
+  return trimmed.split(/(?<!\\)\|/).map((cell) => cell.trim().replace(/\\\|/g, '|'))
+}
+
+function isTableSeparator(line) {
+  const cells = splitTableRow(line)
+  return cells.length > 1 && cells.every((cell) => /^:?-{3,}:?$/.test(cell))
+}
+
+function isTableStart(lines, index) {
+  return (
+    index + 1 < lines.length && lines[index].includes('|') && isTableSeparator(lines[index + 1])
+  )
+}
+
+function collectTable(lines, startIndex) {
+  const rows = [splitTableRow(lines[startIndex])]
+  let index = startIndex + 2
+
+  while (index < lines.length && lines[index].trim() && lines[index].includes('|')) {
+    rows.push(splitTableRow(lines[index]))
+    index += 1
+  }
+
+  return { rows, nextIndex: index }
+}
+
+function createTableBlocks(rows) {
+  const columnSize = Math.max(...rows.map((row) => row.length))
+  const normalizedRows = rows.map((row) => [
+    ...row,
+    ...Array.from({ length: columnSize - row.length }, () => '')
+  ])
+  const maxRows = Math.max(
+    2,
+    Math.min(MAX_TABLE_ROWS, Math.floor(MAX_TABLE_CELLS / columnSize))
+  )
+  const tables = []
+  const header = normalizedRows[0]
+  const bodyRows = normalizedRows.slice(1)
+  const chunks = []
+
+  if (normalizedRows.length <= maxRows) {
+    chunks.push(normalizedRows)
+  } else {
+    const bodyChunkSize = Math.max(1, maxRows - 1)
+
+    for (let index = 0; index < bodyRows.length; index += bodyChunkSize) {
+      chunks.push([header, ...bodyRows.slice(index, index + bodyChunkSize)])
+    }
+  }
+
+  for (const chunk of chunks) {
+    tables.push(createTableBlock(chunk, columnSize))
+  }
+
+  return tables
+}
+
+const TABLE_TOTAL_WIDTH = 760
+const TABLE_MIN_COL_WIDTH = 56
+
+function displayWidth(text) {
+  let width = 0
+  for (const ch of String(text)) {
+    width += /[一-鿿　-〿＀-￯]/.test(ch) ? 2 : 1
+  }
+  return width
+}
+
+function computeColumnWidths(rows, columnSize) {
+  // 满宽自适应：按每列最长内容（中文计 2 宽）分配，归一到 ~760，长枚举名少换行、行更矮。
+  const weights = Array.from({ length: columnSize }, (_, col) =>
+    Math.max(2, ...rows.map((row) => displayWidth(row[col] ?? '')))
+  )
+  const sum = weights.reduce((acc, value) => acc + value, 0) || columnSize
+  const widths = weights.map((weight) =>
+    Math.max(TABLE_MIN_COL_WIDTH, Math.round((TABLE_TOTAL_WIDTH * weight) / sum))
+  )
+  const diff = TABLE_TOTAL_WIDTH - widths.reduce((acc, value) => acc + value, 0)
+  widths[columnSize - 1] = Math.max(TABLE_MIN_COL_WIDTH, widths[columnSize - 1] + diff)
+  return widths
+}
+
+function createTableBlock(rows, columnSize) {
+  const block = {
+    block_type: 31,
+    table: {
+      property: {
+        row_size: rows.length,
+        column_size: columnSize
+      }
+    }
+  }
+
+  Object.defineProperty(block, '__tableRows', {
+    value: rows
+  })
+  // 表头底色与列宽建表时不被接受，由 writeBlocks 在建表后 PATCH 设置。
+  Object.defineProperty(block, '__columnWidth', {
+    value: computeColumnWidths(rows, columnSize)
+  })
+
+  return block
+}
+
+export function markdownToLarkBlocks(markdown) {
+  const lines = markdown.replace(/\r\n/g, '\n').split('\n')
+  const blocks = []
+  let paragraph = []
+  let codeFence = null
+  let codeLines = []
+  let orderedSequence = 1
+
+  const flushParagraph = () => {
+    const text = paragraph.join('\n').trim()
+    paragraph = []
+
+    if (!text) return
+
+    for (const chunk of splitLongText(text)) {
+      blocks.push(createTextBlock(chunk))
+    }
+  }
+
+  const flushCode = () => {
+    const language = codeFence ? codeFence.replace(/^```/, '').trim() : ''
+    const body = codeLines.join('\n')
+    codeFence = null
+    codeLines = []
+
+    for (const chunk of splitLongText(body)) {
+      blocks.push(createCodeBlock(language, chunk))
+    }
+  }
+
+  for (let index = 0; index < lines.length; index += 1) {
+    const line = lines[index]
+
+    if (line.startsWith('```')) {
+      if (codeFence) {
+        flushCode()
+      } else {
+        flushParagraph()
+        codeFence = line
+        codeLines = []
+      }
+      continue
+    }
+
+    if (isTableStart(lines, index)) {
+      flushParagraph()
+      orderedSequence = 1
+      const table = collectTable(lines, index)
+      blocks.push(...createTableBlocks(table.rows))
+      index = table.nextIndex - 1
+      continue
+    }
+
+    if (codeFence) {
+      codeLines.push(line)
+      continue
+    }
+
+    const heading = /^(#{1,3})\s+(.+)$/.exec(line)
+    if (heading) {
+      flushParagraph()
+      blocks.push(createHeadingBlock(heading[1].length, heading[2].trim()))
+      continue
+    }
+
+    // 高亮块 callout：`> [!TIP] 文本` / `> [!NOTE] ...` / `> [!WARNING] ...`。
+    const calloutMatch = /^>\s*\[!(\w+)\]\s*(.*)$/.exec(line)
+    if (calloutMatch) {
+      flushParagraph()
+      orderedSequence = 1
+      const type = calloutMatch[1].toUpperCase()
+      const emoji = type === 'WARNING' || type === 'WARN' ? 'warning' : type === 'TIP' ? 'bulb' : 'memo'
+      blocks.push(createCalloutBlock(calloutMatch[2].trim(), emoji))
+      continue
+    }
+
+    // 引用块：`> 文本` → 飞书 quote_container（引用样式），由 writeBlocks 经 descendant API 写入。
+    const quote = /^>\s?(.*)$/.exec(line)
+    if (quote) {
+      flushParagraph()
+      orderedSequence = 1
+      const text = quote[1].trim()
+      if (text) blocks.push(createQuoteBlock(text))
+      continue
+    }
+
+    if (line.trim() === '') {
+      flushParagraph()
+      orderedSequence = 1
+      continue
+    }
+
+    // 分割线：独立的 --- / *** / ___（非表格分隔行）→ 飞书 divider 块。
+    if (/^(-{3,}|\*{3,}|_{3,})$/.test(line.trim())) {
+      flushParagraph()
+      orderedSequence = 1
+      blocks.push({ block_type: 22, divider: {} })
+      continue
+    }
+
+    const bullet = /^\s*[-*]\s+(.+)$/.exec(line)
+    if (bullet) {
+      flushParagraph()
+      orderedSequence = 1
+      blocks.push(createBulletBlock(bullet[1].trim()))
+      continue
+    }
+
+    const ordered = /^\s*\d+[.)]\s+(.+)$/.exec(line)
+    if (ordered) {
+      flushParagraph()
+      blocks.push(createOrderedBlock(ordered[1].trim(), orderedSequence))
+      orderedSequence += 1
+      continue
+    }
+
+    orderedSequence = 1
+    paragraph.push(line)
+  }
+
+  if (codeFence) flushCode()
+  flushParagraph()
+
+  return blocks
+}
+
+async function main() {
+  const args = parseArgs()
+  const file = args.file
+  const markdown = file ? readFileSync(file, 'utf8') : readFileSync(0, 'utf8')
+  printJson({ blocks: markdownToLarkBlocks(markdown) })
+}
+
+if (import.meta.url === `file://${process.argv[1]}`) {
+  main().catch((error) => {
+    console.error(error.message)
+    process.exit(1)
+  })
+}