mindee/src/imageOperations/multiReceiptsExtractor/multiReceiptsExtractor.js

Mindee Client Library for Node.js

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.extractReceipts = extractReceipts;
const pdf_lib_1 = require("@cantoo/pdf-lib");
const errors_1 = require("../../errors");
const extractedMultiReceiptImage_1 = require("./extractedMultiReceiptImage");
const common_1 = require("../common");
/**
 * Given a page and a set of coordinates, extracts & assigns individual receipts to an ExtractedMultiReceiptImage
 * object.
 *
 * @param pdfPage PDF Page to extract from.
 * @param boundingBoxes A set of coordinates delimiting the position of each receipt.
 * @param pageId Id of the page the receipt is extracted from. Caution: this starts at 0, unlike the numbering in PDF
 * pages.
 */
async function extractReceiptsFromPage(pdfPage, boundingBoxes, pageId) {
    const extractedReceiptsRaw = await (0, common_1.extractFromPage)(pdfPage, boundingBoxes);
    const extractedReceipts = [];
    for (let i = 0; i < extractedReceiptsRaw.length; i++) {
        extractedReceipts.push(new extractedMultiReceiptImage_1.ExtractedMultiReceiptImage(extractedReceiptsRaw[i], pageId, i));
    }
    return extractedReceipts;
}
async function loadPdfDoc(inputFile) {
    let pdfDoc;
    if (!["image/jpeg", "image/jpg", "image/png", "application/pdf"].includes(inputFile.mimeType)) {
        throw new errors_1.MindeeMimeTypeError('Unsupported file type "' +
            inputFile.mimeType +
            '" Currently supported types are .png, .jpg and .pdf');
    }
    else if (inputFile.isPdf()) {
        pdfDoc = await pdf_lib_1.PDFDocument.load(inputFile.fileObject, {
            ignoreEncryption: true,
            password: ""
        });
    }
    else {
        pdfDoc = await pdf_lib_1.PDFDocument.create();
        let image;
        if (inputFile.mimeType === "image/png") {
            image = await pdfDoc.embedPng(inputFile.fileObject);
        }
        else {
            image = await pdfDoc.embedJpg(inputFile.fileObject);
        }
        const imageDims = image.scale(1);
        const pageImage = pdfDoc.addPage([imageDims.width, imageDims.height]);
        pageImage.drawImage(image);
    }
    return pdfDoc;
}
/**
 * Extracts individual receipts from multi-receipts documents.
 *
 * @param inputFile File to extract sub-receipts from.
 * @param inference Results of the inference.
 * @returns Individual extracted receipts as an array of ExtractedMultiReceiptImage.
 */
async function extractReceipts(inputFile, inference) {
    const images = [];
    if (!inference.prediction.receipts) {
        throw new errors_1.MindeeError("No possible receipts candidates found for MultiReceipts extraction.");
    }
    const pdfDoc = await loadPdfDoc(inputFile);
    for (let pageId = 0; pageId < pdfDoc.getPageCount(); pageId++) {
        const [page] = await pdfDoc.copyPages(pdfDoc, [pageId]);
        const receiptPositions = inference.pages[pageId].prediction.receipts.map((receipt) => receipt.boundingBox);
        const extractedReceipts = await extractReceiptsFromPage(page, receiptPositions, pageId);
        images.push(...extractedReceipts);
    }
    return images;
}