mindee
Version:
Mindee Client Library for Node.js
43 lines (42 loc) • 1.84 kB
JavaScript
;
Object.defineProperty(exports, "__esModule", { value: true });
exports.extractFromPage = extractFromPage;
const pdf_lib_1 = require("@cantoo/pdf-lib");
const geometry_1 = require("../../geometry");
/**
* Extracts elements from a page based off of a list of bounding boxes.
*
* @param pdfPage PDF Page to extract from.
* @param polygons List of coordinates to pull the elements from.
*/
async function extractFromPage(pdfPage, polygons) {
const { width, height } = pdfPage.getSize();
const extractedElements = [];
// Manual upscale.
// Fixes issues with the OCR.
const qualityScale = 300 / 72;
for (const polygon of polygons) {
const tempPdf = await pdf_lib_1.PDFDocument.create();
const newWidth = width * ((0, geometry_1.getMinMaxX)(polygon).max - (0, geometry_1.getMinMaxX)(polygon).min);
const newHeight = height * ((0, geometry_1.getMinMaxY)(polygon).max - (0, geometry_1.getMinMaxY)(polygon).min);
const cropped = await tempPdf.embedPage(pdfPage, {
left: (0, geometry_1.getMinMaxX)(polygon).min * width,
right: (0, geometry_1.getMinMaxX)(polygon).max * width,
top: height - ((0, geometry_1.getMinMaxY)(polygon).min * height),
bottom: height - ((0, geometry_1.getMinMaxY)(polygon).max * height),
});
const samplePage = tempPdf.addPage([newWidth * qualityScale, newHeight * qualityScale]);
samplePage.drawRectangle({
x: 0,
y: 0,
width: newWidth * qualityScale,
height: newHeight * qualityScale,
});
samplePage.drawPage(cropped, {
width: newWidth * qualityScale,
height: newHeight * qualityScale,
});
extractedElements.push(await tempPdf.save());
}
return extractedElements;
}