harn-stdlib 0.9.22

/** std/document — portable document conversion helpers. */
type DocumentSourceFormat = "text" | "html" | "markdown"

type PdfRenderer = "auto" | "builtin_text_pdf"

type PdfRenderOptions = {
  renderer?: PdfRenderer,
  source_format?: DocumentSourceFormat,
  title?: string,
  page_width_pt?: int,
  page_height_pt?: int,
  margin_pt?: int,
  font_size_pt?: int,
  line_height_pt?: int,
  max_line_chars?: int,
  metadata?: dict,
  provider_options?: dict,
  renderer_options?: dict,
}

type TextExtractOptions = {source_format?: DocumentSourceFormat}

type PdfWriteResult = {path: string, bytes: int, media_type: string, renderer: string}

/**
 * Return the document/PDF renderers available in this runtime.
 *
 * @effects: []
 * @errors: []
 */
pub fn pdf_capabilities() -> dict {
  return document_pdf_capabilities()
}

/**
 * Extract normalized text from text, HTML, or Markdown input.
 *
 * @effects: []
 * @errors: [document_extract_text]
 */
pub fn extract_text(source: string | bytes | nil, options: TextExtractOptions = {}) -> string {
  return document_extract_text(source, options ?? {})
}

/**
 * Render text-like input to PDF bytes using the selected renderer.
 *
 * The built-in renderer is dependency-free and cross-platform. It preserves
 * readable text layout but is not a browser-grade CSS renderer.
 *
 * @effects: []
 * @errors: [document_render_pdf]
 */
pub fn pdf_bytes(source: string | bytes | nil, options: PdfRenderOptions = {}) -> bytes {
  return document_render_pdf(source, options ?? {})
}

/**
 * Render text-like input to a PDF file through `harness.fs`.
 *
 * @effects: [fs]
 * @errors: [document_render_pdf, fs]
 */
pub fn write_pdf(path: string, source: string | bytes | nil, options: PdfRenderOptions = {}) -> PdfWriteResult {
  let opts = options ?? {}
  let bytes = pdf_bytes(source, opts)
  harness.fs.write_bytes(path, bytes)
  return {
    path: path,
    bytes: bytes_len(bytes),
    media_type: "application/pdf",
    renderer: opts.renderer ?? "builtin_text_pdf",
  }
}