knee-cola/evidencija-rezija
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

implemented scanning PDF docs for multiple 2D codes

Browse Source
This commit is contained in:
Nikola Derežić
2025-07-23 13:12:13 +02:00
parent 691715c8b4
commit 36f2f74e64
4 changed files with 201 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

158
app/lib/pdf/barcodeDecoder.ts
View File

@@ -1,5 +1,5 @@
import { PDFPageProxy } from 'pdfjs-dist';
import { BrowserPDF417Reader, BrowserMultiFormatReader } from '@zxing/browser';
import { BrowserPDF417Reader } from '@zxing/browser';
import * as pdfJSx from 'pdfjs-dist';
import { BarcodeFormat, DecodeHintType, Result } from '@zxing/library';
@@ -21,6 +21,12 @@ export type BillInfo = {
description: string,
};
/** Breaks current microtask execution and gives the UI thread a chance to do a re-paint */
const yieldToBrowser = (label:string) => new Promise<boolean>((resolve) => {
setTimeout(() => {
resolve(true);
}, 0);
});
/**
* Decodes a PDF417 barcode
@@ -46,7 +52,7 @@ export type BillInfo = {
* description:Akontacijska rata za 01.2024.
*
*/
const parseHubText = (text: string) => {
const parseHubText = (text: string):BillInfo => {
const [
header,
currency,
@@ -129,7 +135,7 @@ const image2canvas = async function (imageFile:File): Promise<HTMLCanvasElement>
* @param {File} pdfFile - a file containing a PDF document
* @return {Promise<HTMLCanvasElement>} the canvas with the first page of the PDF
*/
const pdf2canvas = async function (pdfFile:File): Promise<HTMLCanvasElement> {
const pdf2canvas = async function (pdfFile:File): Promise<Array<HTMLCanvasElement>> {
const reader = new FileReader();
const data = await new Promise<Uint8Array>((resolve, reject) => {
@@ -144,38 +150,127 @@ const pdf2canvas = async function (pdfFile:File): Promise<HTMLCanvasElement> {
pdfJS.GlobalWorkerOptions.workerSrc = window.location.origin + '/pdf.worker.min.mjs';
const pdf = await pdfJS.getDocument(data).promise;
const page: PDFPageProxy = await pdf.getPage(1);
const scale = 4;
const viewport = page.getViewport({ scale });
const canvases: Array<HTMLCanvasElement> = [];
const canvas = document.createElement('canvas');
const context = canvas.getContext('2d');
canvas.height = viewport.height;
canvas.width = viewport.width;
for(let i = 0; i< pdf.numPages; i++) {
const page: PDFPageProxy = await pdf.getPage(i+1);
await page.render({ canvasContext: context as CanvasRenderingContext2D, viewport }).promise;
const scale = 4;
const viewport = page.getViewport({ scale });
return(canvas);
const canvas = document.createElement('canvas');
const context = canvas.getContext('2d');
canvas.height = viewport.height;
canvas.width = viewport.width;
await page.render({ canvasContext: context as CanvasRenderingContext2D, viewport }).promise;
canvases.push(canvas);
}
return(canvases);
}
export type DecodeResult = {
billInfo: BillInfo,
barcodeImage: string,
};
/**
* Decodes PDF417 from image contained within the given canvas
* Searches the given canvas for all PDF417 codes and decodes them.
* @param {HTMLCanvasElement} canvas - the canvas to search for PDF417 codes
* @return {Promise<Array<DecodeResult> | null>} - an array of decoded results
* */
const decodeFromCanvas = async (canvas:HTMLCanvasElement) => {
const decodeFromCanvas = async (canvas:HTMLCanvasElement): Promise<Array<DecodeResult> | null> => {
try {
const hints = new Map();
hints.set(DecodeHintType.POSSIBLE_FORMATS, [ BarcodeFormat.PDF_417 ]);
hints.set(DecodeHintType.PURE_BARCODE, false);
const codeReader = new BrowserPDF417Reader(hints);
const result = await codeReader.decodeFromCanvas(canvas);
return({
billInfo: parseHubText(result.getText()),
barcodeImage: copyBarcodeImage(canvas, result)
})
const width = canvas.width;
const height = canvas.height;
// Canvas can contain multiple PDF417 codes, so we need to try to find them all
// The issue is that `BrowserPDF417Reader` can only decode one code at a time
// and it will throw an error if it finds more than one code.
// To solve this, we will try splitting the canvas into different number of subsections
// and decode each subsection separately. The best result will be the one with the most codes found.
const splits = [5,4,3,2,1,0];
let bestResult: Array<DecodeResult>|null = null;
for(let splitIx = 0; splitIx < splits.length; splitIx++) {
const split = splits[splitIx];
const overlap = split === 0 ? 0 : Math.round(height / 50); // 50% overlap ensuring that we don't miss any codes that might be split between sections
const sectionHeight = split === 0 ? height : (Math.floor( Math.floor(height / split) + overlap));
const canvasSections = Array.from({ length: split+1 }, (_, i) => {
const sectionCanvas = document.createElement('canvas');
sectionCanvas.width = width;
sectionCanvas.height = sectionHeight;
const sectionContext = sectionCanvas.getContext('2d');
if (!sectionContext) {
throw new Error('Failed to get canvas context');
}
// Calculate the starting Y position for each section
const startY = i===0 ? 0 : i * (sectionHeight) - overlap;
// Draw the section of the original canvas onto the new section canvas
sectionContext.drawImage(canvas, 0, startY, width, sectionHeight, 0, 0, width, sectionHeight);
return sectionCanvas;
});
const codesFoundInSection: Array<DecodeResult> = [];
// Try to decode each section
for (const sectionCanvas of canvasSections) {
try {
// give browser a chance to re-paint
// this is needed to avoid UI freezing when decoding large images
await yieldToBrowser('decodeFromCanvas');
const result = await codeReader.decodeFromCanvas(sectionCanvas);
if (result) {
codesFoundInSection.push({
billInfo: parseHubText(result.getText()),
barcodeImage: copyBarcodeImage(sectionCanvas, result)
});
}
} catch (error) {
// If no code was found in the current section an error will be thrown
// -> we can ignore it
} finally {
}
}
await yieldToBrowser('after decodeFromCanvas');
// IF in this iteration we found less codes than in the previous best result,
// we can stop searching for more codes
// This is because the number of codes found in each section will only decrease
// as we increase the number of sections (split)
if(bestResult && codesFoundInSection.length <= bestResult.length) {
return(bestResult);
}
bestResult = codesFoundInSection;
};
return(bestResult);
} catch(ex:any) {
console.log(ex);
return(null);
@@ -186,7 +281,7 @@ const decodeFromCanvas = async (canvas:HTMLCanvasElement) => {
* Copies bar code from the given canvas
*
*/
const copyBarcodeImage = (canvas:HTMLCanvasElement, decoderResult:Result) => {
const copyBarcodeImage = (canvas:HTMLCanvasElement, decoderResult:Result):string => {
// get coordinates of bar code
const points = decoderResult.getResultPoints();
@@ -241,7 +336,20 @@ const decodeFromFile = async (file:File) => {
case 'image/jpeg':
return(await decodeFromCanvas( await image2canvas(file) ));
case 'application/pdf':
return(await decodeFromCanvas( await pdf2canvas(file) ));
const pageCanvas = await pdf2canvas(file);
// go through each page of the PDF and decode the PDF417 codes
// if there are multiple pages, we will decode each page separately
// and return the results from all pages
const results = (await Promise.all(pageCanvas.map(async (canvas) => {
await yieldToBrowser('decodeFromCanvas');
return await decodeFromCanvas(canvas);
})))
// remove null results (pages with no PDF417 codes)
.filter((result) => result !== null)
// flatten the array of arrays into a single array
.flat();
return(results);
default:
console.error(file.name, 'is not a .pdf file.');
return null;
@@ -253,13 +361,13 @@ const decodeFromFile = async (file:File) => {
* @param {Event} event - The change event from an HTMLInputElement.
* @return {Promise<HTMLCanvasElement | null>} The canvas with the first page of the PDF, or null if the document is not a PDF.
*/
export async function findDecodePdf417(event: React.ChangeEvent<HTMLInputElement>): Promise<{ billInfo: BillInfo, barcodeImage:string } | null> {
export async function findDecodePdf417(event: React.ChangeEvent<HTMLInputElement>): Promise<Array<DecodeResult>|null> {
const file = (event.target as HTMLInputElement).files?.[0];
if(!file) {
console.error('No file was selected.');
return null;
}
return(await decodeFromFile(file));
}

79
app/ui/BillEditForm.tsx
View File

@@ -7,7 +7,7 @@ import { useFormState } from "react-dom";
import { updateOrAddBill } from "../lib/actions/billActions";
import Link from "next/link";
import { formatYearMonth } from "../lib/format";
import { findDecodePdf417 } from "../lib/pdf/barcodeDecoder";
import { DecodeResult, findDecodePdf417 } from "../lib/pdf/barcodeDecoder";
import { useLocale, useTranslations } from "next-intl";
// Next.js does not encode an utf-8 file name correctly when sending a form with a file attachment
@@ -37,10 +37,12 @@ export const BillEditForm:FC<BillEditFormProps> = ({ location, bill }) => {
const handleAction = updateOrAddBillMiddleware.bind(null, locationID, billID, billYear, billMonth);
const [ isScanningPDF, setIsScanningPDF ] = React.useState<boolean>(false);
const [ state, dispatch ] = useFormState(handleAction, initialState);
const [ isPaid, setIsPaid ] = React.useState<boolean>(paid);
const [ payedAmount, setPayedAmount ] = React.useState<string>(initialPayedAmount ? `${initialPayedAmount/100}` : "" );
const [ barcodeImage, setBarcodeImage ] = React.useState<string | undefined>(initialBarcodeImage);
const [ barcodeResults, setBarcodeResults ] = React.useState<Array<DecodeResult> | null>(null);
const billPaid_handleChange = (event: React.ChangeEvent<HTMLInputElement>) => {
setIsPaid(event.target.checked);
@@ -50,19 +52,39 @@ export const BillEditForm:FC<BillEditFormProps> = ({ location, bill }) => {
setPayedAmount(event.target.value);
}
const billAttachment_handleChange = (event: React.ChangeEvent<HTMLInputElement>) => {
findDecodePdf417(event)
.then(result => {
if(result) {
const {
barcodeImage,
billInfo
} = result;
const billAttachment_handleChange = async (event: React.ChangeEvent<HTMLInputElement>) => {
setIsScanningPDF(true);
setPayedAmount(`${billInfo.amount/100}`);
setBarcodeImage(barcodeImage);
}
});
setPayedAmount("");
setBarcodeImage(undefined);
setBarcodeResults(null);
const results = await findDecodePdf417(event);
if(results && results.length > 0) {
if(results.length === 1) {
const {
barcodeImage,
billInfo
} = results[0];
setPayedAmount(`${billInfo.amount/100}`);
setBarcodeImage(barcodeImage);
} else {
setPayedAmount("");
setBarcodeImage(undefined);
setBarcodeResults(results);
}
}
setIsScanningPDF(false);
}
const handleBarcodeSelectClick = (result: DecodeResult) => {
setPayedAmount(`${result.billInfo.amount/100}`);
setBarcodeImage(result.barcodeImage);
setBarcodeResults(null);
}
return(
@@ -98,8 +120,36 @@ export const BillEditForm:FC<BillEditFormProps> = ({ location, bill }) => {
: null
}
<div className="flex">
<input id="billAttachment" name="billAttachment" type="file" className="file-input file-input-bordered grow file-input-xs my-2 block" onChange={billAttachment_handleChange} />
<input id="billAttachment" name="billAttachment" type="file" className="file-input file-input-bordered grow file-input-s my-2 block max-w-[17em] md:max-w-[80em] break-words" onChange={billAttachment_handleChange} />
</div>
{
isScanningPDF &&
<div className="flex flex-row items-center w-full mt-2">
<div className="loading loading-spinner loading-m mr-2"></div>
<span>{t("scanning-pdf")}</span>
</div>
}
{
// if multiple results are found, show them as a list
// and notify the user to select the correct one
barcodeResults && barcodeResults.length > 0 &&
<>
<div className="flex mt-2 ml-2">
<label className="label-text max-w-xs break-words">{t("multiple-barcode-results-notification")}</label>
</div>
<div className="flex">
<label className="label grow pt-0 ml-3">
<ul className="list-none">
{barcodeResults.map((result, index) => (
<li key={index} className="cursor-pointer mt-3" onClick={() => handleBarcodeSelectClick(result)}>
👉 { result.billInfo.description }
</li>
))}
</ul>
</label>
</div>
</>
}
<div id="status-error" aria-live="polite" aria-atomic="true">
{state.errors?.billAttachment &&
state.errors.billAttachment.map((error: string) => (
@@ -141,7 +191,6 @@ export const BillEditForm:FC<BillEditFormProps> = ({ location, bill }) => {
<p className="text-xs my-1">{t.rich('barcode-disclaimer', { br: () => <br /> })}</p>
</div> : null
}
<textarea id="billNotes" name="billNotes" className="textarea textarea-bordered my-2 max-w-lg w-full block" placeholder={t("notes-placeholder")} defaultValue={notes ?? ''}></textarea>
<div id="status-error" aria-live="polite" aria-atomic="true">
{state.errors?.billNotes &&

2
messages/en.json
View File

@@ -69,6 +69,8 @@
"bill-edit-form": {
"bill-name-placeholder": "Bill name",
"paid-checkbox": "Paid",
"scanning-pdf": "🕵️‍♂️ Scanning document for 2D codes ...",
"multiple-barcode-results-notification": "✅ Multiple 2D barcodes found. Pick the one which you want to use:",
"payed-amount": "Amount",
"barcode-disclaimer": "After scanning the code make sure the information is correct.<br></br>We are not liable in case of an incorrect payment.",
"notes-placeholder": "Notes",

2
messages/hr.json
View File

@@ -69,6 +69,8 @@
"bill-edit-form": {
"bill-name-placeholder": "Ime računa",
"paid-checkbox": "Plaćeno",
"scanning-pdf": "🕵️‍♂️ Tražim 2D barkodove unutar dokumenta...",
"multiple-barcode-results-notification": "✅ Pronađeno je više 2D barkodova. Molimo odaberi onaj koji se odnosi na zadani mjesec:",
"payed-amount": "Iznos",
"barcode-disclaimer": "Nakon skeniranja bar koda obavezni provjeri jesu li svi podaci ispravni.<br></br>Ne snosimo odgovornost za slučaj pogrešno provedene uplate.",
"notes-placeholder": "Bilješke",