Function parseMarimoFormat

packages/convert/src/marimo-to-deepnote.ts:135–337

(content: string)

 * ```
 */
export function parseMarimoFormat(content: string): MarimoApp {
  const cells: MarimoCell[] = []

  // Extract version
  const versionMatch = /__generated_with\s*=\s*["']([^"']+)["']/.exec(content)
  const generatedWith = versionMatch?.[1]

  // Extract app width - handle both 'marimo.App' and 'mo.App'
  const widthMatch = /(?:marimo|mo)\.App\([^)]*width\s*=\s*["']([^"']+)["']/.exec(content)
  const width = widthMatch?.[1]

  // Extract title - handle both 'marimo.App' and 'mo.App'
  const titleMatch = /(?:marimo|mo)\.App\([^)]*title\s*=\s*["']([^"']+)["']/.exec(content)
  const title = titleMatch?.[1]

  // Parse cells - look for @app.cell and @app.function decorated functions
  // This regex matches @app.cell or @app.function (with optional parameters) followed by def
  // Capture group 1: decorator type ('cell' or 'function')
  // Capture group 2: decorator arguments (if any)
  // Capture group 3: function name
  // Capture group 4: function parameters
  // Capture group 5: function body
  const cellRegex =
    /@app\.(cell|function)(?:\(([^)]*)\))?\s*\n\s*def\s+(\w+)\s*\(([^)]*)\)\s*(?:->.*?)?\s*:\s*\n([\s\S]*?)(?=@app\.cell|@app\.function|if\s+__name__|$)/g

  let match: RegExpExecArray | null = cellRegex.exec(content)

  while (match !== null) {
    const decoratorType = match[1] // 'cell' or 'function'
    const decoratorArgs = match[2] || ''
    const functionName = match[3]
    const params = match[4].trim()
    let body = match[5]

    // @app.function creates a reusable function, convert to code cell
    const isAppFunction = decoratorType === 'function'

    // For @app.function, reconstruct the full function definition
    if (isAppFunction) {
      // Get the return type if present (captured separately)
      const returnTypeMatch = /@app\.function(?:\([^)]*\))?\s*\n\s*def\s+\w+\s*\([^)]*\)\s*(->.*?)?\s*:/.exec(
        content.slice(match.index)
      )
      const returnType = returnTypeMatch?.[1] || ''

      // Reconstruct the function definition
      const funcDef = `def ${functionName}(${params})${returnType}:\n${body}`

      cells.push({
        cellType: 'code',
        content: funcDef.trim(),
        functionName,
        hidden: /hide_code\s*=\s*True/.test(decoratorArgs),
        disabled: /disabled\s*=\s*True/.test(decoratorArgs),
      })

      match = cellRegex.exec(content)
      continue