Interface: MarkdownOptions

Defined in: src/helpers/types.ts:134

Properties

detectOrientation?

optional detectOrientation?: boolean;

Defined in: src/helpers/types.ts:179

With tableStrategy: "pixels", turn a page without a text layer whose lines run vertically (a sheet scanned sideways, /Rotate not set) before its tables are read. Default true.

---

dpi?

optional dpi?: number;

Defined in: src/helpers/types.ts:149

---

elements?

optional elements?: MarkdownElement[];

Defined in: src/helpers/types.ts:201

Whitelist of markdown elements to emit. When omitted, ALL elements are emitted (backwards-compatible default). When provided, only the listed elements are produced; everything else falls back to plain text.

Combines with the legacy toggles (ignoreCode, hdrInfo: false, tableStrategy: null, writeImages/embedImages): an element is emitted only if BOTH the whitelist allows it AND no legacy option disabled it.

---

embedImages?

optional embedImages?: boolean;

Defined in: src/helpers/types.ts:141

---

explicitTableGrids?

optional explicitTableGrids?: object[];

Defined in: src/helpers/types.ts:183

Explicit grid coordinates for tableStrategy: "explicit".

hLines

hLines: number[];

vLines

vLines: number[];

---

extractWords?

optional extractWords?: boolean;

Defined in: src/helpers/types.ts:187

---

filename?

optional filename?: string | null;

Defined in: src/helpers/types.ts:145

---

fontsizeLimit?

optional fontsizeLimit?: number;

Defined in: src/helpers/types.ts:185

Skip spans whose font size is below this threshold (in pt). Mirrors upstream FONTSIZE_LIMIT.

---

forceText?

optional forceText?: boolean;

Defined in: src/helpers/types.ts:146

---

hdrInfo?

optional hdrInfo?: 
  | false
  | {
  get_header_id: string;
}
  | ((s) => string);

Defined in: src/helpers/types.ts:136

---

ignoreAlpha?

optional ignoreAlpha?: boolean;

Defined in: src/helpers/types.ts:190

Accept invisible (alpha=0) text. Currently a no-op — mupdf.js's walker does not expose alpha.

---

ignoreCode?

optional ignoreCode?: boolean;

Defined in: src/helpers/types.ts:186

---

imageFormat?

optional imageFormat?: "png" | "jpg" | "jpeg";

Defined in: src/helpers/types.ts:143

---

imagePath?

optional imagePath?: string;

Defined in: src/helpers/types.ts:142

---

imageSizeLimit?

optional imageSizeLimit?: number;

Defined in: src/helpers/types.ts:144

---

margins?

optional margins?: 
  | number
  | [number, number]
  | [number, number, number, number];

Defined in: src/helpers/types.ts:148

---

ocr?

optional ocr?: OcrEngine;

Defined in: src/helpers/types.ts:173

OCR engine for table cells. Defaults to RapidOCR (PP-OCRv5 via the optional peer dependencies ppu-paddle-ocr + onnxruntime-node), which is loaded on first use. Any object with recognize(image) works.

---

ocrDpi?

optional ocrDpi?: number;

Defined in: src/helpers/types.ts:181

Resolution the page is rendered at for "pixels" and OCR. Default 300.

---

pages?

optional pages?: number[];

Defined in: src/helpers/types.ts:135

---

pageSeparators?

optional pageSeparators?: boolean;

Defined in: src/helpers/types.ts:147

---

removeRotation?

optional removeRotation?: boolean;

Defined in: src/helpers/types.ts:191

---

showProgress?

optional showProgress?: boolean;

Defined in: src/helpers/types.ts:188

---

tableStrategy?

optional tableStrategy?: "lines_strict" | "lines" | "text" | "explicit" | "pixels" | null;

Defined in: src/helpers/types.ts:154

How tables are found. "pixels" finds ruling lines on the rendered page instead of in the PDF drawings, so it works on scans.

---

textSource?

optional textSource?: TextSource;

Defined in: src/helpers/types.ts:167

Where table cell text comes from, independent of how the table was found:

• "pdf": the PDF text layer.
• "ocr": OCR of every cell; the text layer is ignored.
• "auto" (default): the text layer, and OCR for cells whose text is empty or looks broken (replacement or private-use characters, a letter of another script slipped into a word, mostly symbols). On a page without a text layer that is every cell. Left at the default, a page with a text layer keeps it when no OCR engine can be set up.

The source of each cell is reported in PageChunk.tables[].cells.

---

writeImages?

optional writeImages?: boolean;

Defined in: src/helpers/types.ts:140