PdfToImage

Import

import { PdfToImage } from '@stabrise/scaledp/pdf'

Group

Read

Reads

bytes

Writes

image

Needs

pdfjs-dist

Note

Expands one row into several.

Rasterises every page and expands one input row into one row per page, writing the page index to pageCol.

import { Pipeline } from '@stabrise/scaledp'
import { PdfToImage } from '@stabrise/scaledp/pdf'
import { PaddleTextRecognizer } from '@stabrise/scaledp/ocr'

const rows = await new Pipeline([
    new PdfToImage({ resolution: 300 }),
    new PaddleTextRecognizer(),
]).transform(pdfFile)

rows.length          // one per page
rows[0].page         // 0

Choosing a resolution

The scale factor is resolution / 72. 300 DPI suits OCR; 150 halves the render time and the pixel count, and is usually enough for clean digital PDFs. Detection quality falls off below about 150 for body text.

Whatever you choose, use the same number on PdfToDocument if both run — its boxes are emitted in the pixel space of that DPI, and mismatched values put the two sets of boxes in different coordinate systems.

Requirements

Needs pdfjs-dist installed and configure({ pdf: { workerSrc, cMapUrl, standardFontDataUrl } }) pointing at assets you serve. Failures here are wrapped with an actionable message naming the missing setting rather than pdf.js's own — a missing workerSrc is the single most common first-run problem.

renderPage(document, pageNumber, options) is exported separately for rendering one page outside a pipeline; page numbers there are 1-based, as in pdf.js.

Parameters