statement-parser-fab/dist/pdf/read-pdf.js

Parse bank and credit card statements. Updated fork with FAB (First Abu Dhabi Bank) support and maintained dependencies.

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.checkThatPdfExists = exports.getPdfDocument = exports.readPdf = void 0;
const fs_extra_1 = require("fs-extra");
const pdf_1 = require("pdfjs-dist/legacy/build/pdf");
async function readPdf(path) {
    checkThatPdfExists(path);
    // Use custom positioning-based extraction instead of pdf-text-reader
    const document = await getPdfDocument(path);
    const pages = [];
    for (let pageNum = 1; pageNum <= document.numPages; pageNum++) {
        const page = await document.getPage(pageNum);
        const textContent = await page.getTextContent();
        // Group text items by Y position (preserve table structure)
        const itemsByY = {};
        textContent.items.forEach((item) => {
            if ('str' in item && item.transform && item.transform[5] !== undefined) {
                const y = Math.round(item.transform[5]);
                if (!itemsByY[y])
                    itemsByY[y] = [];
                const yArray = itemsByY[y];
                if (yArray) {
                    yArray.push(item);
                }
            }
        });
        // Create lines sorted by Y position (top to bottom)
        const lines = [];
        Object.keys(itemsByY)
            .map((y) => parseInt(y))
            .sort((a, b) => b - a) // Sort by Y descending (top to bottom)
            .forEach((y) => {
            const lineData = itemsByY[y];
            if (lineData) {
                const lineItems = lineData.sort((a, b) => a.transform[4] - b.transform[4]); // Sort by X
                const lineText = lineItems
                    .map((item) => item.str)
                    .join(' ')
                    .trim();
                if (lineText) {
                    lines.push(lineText);
                }
            }
        });
        pages.push(lines);
    }
    return pages;
}
exports.readPdf = readPdf;
async function getPdfDocument(path) {
    checkThatPdfExists(path);
    return await (0, pdf_1.getDocument)(createSource(path)).promise;
}
exports.getPdfDocument = getPdfDocument;
function createSource(path) {
    return { url: path, verbosity: pdf_1.VerbosityLevel.ERRORS };
}
function checkThatPdfExists(filePath) {
    if (!(0, fs_extra_1.existsSync)(filePath)) {
        throw new Error(`PDF file "${filePath}" does not exist`);
    }
}
exports.checkThatPdfExists = checkThatPdfExists;