Alanko5/credit-card-scanner-and-validator

Uses the iOS Vision text recognizer framework to visually read credit card numbers and expiration dates. Then uses the Luhn algorithm to make sure the CC number is valid. Also checks agains a list of prefix numbers to determine card type (Mastercard, Visa, Discover, Amex, etc).

Credit and Debit Card Scanner and Validator

Requires iOS 13 or above.

Uses the iOS Vision text recognizer framework to visually read credit and debit card numbers and expiration dates. During the recognition process, it uses the Luhn algorithm to make sure the CC number is valid. It also checks agains a list of prefix numbers to determine card type (Mastercard, Visa, Discover, Amex, etc).

How to Use

It's very simple to use. Add the CCScannerDelegate to your UIViewController, initialize the CCScanner class, set the delegate, and start the scanner. The delegate method will be called returning you the card information.

Here's an extremely easy example to follow:

import UIKit

class ViewController: UIViewController, CCScannerDelegate {
    
    let ccScanner = CCScanner()
    
    override func viewDidLoad() {
        super.viewDidLoad()
        self.ccScanner.delegate = self
    }
    
    @IBAction func startButton() {
        self.ccScanner.startScanner(viewController: self)
    }
    
    func ccScannerCompleted(cardNumber: String, expDate: String, cardType: String) {
        
        // do something with the data
        print(cardNumber, expDate)
    }
}

Why did I make this?

I have noticed lately that many apps on the App Store, including Apple's own Wallet app, now use card scanning to automatically pull credit/debit numbers from a live camera view and populate them into fields. This makes entering payment information a lot easier and faster for customers in apps that don't or can't use Apple Pay. Afterall, the faster the checkout, the more likely you will win that sale, right?

There are also a few other projects here on GitHub that are nice, but they dont have the accuracy for various card numbering formats, nor the validation for worldwide use, to truely be something an international company could depend on. Their code is cleaner than mine though, if that matters. ;)

I don't have an app to use this code in personally, but I was intrigued to see if I could build something better and faster than what a lot of these other apps use. Even Apple's Wallet CC reader is frustratingly slow (and sometimes doesn't work at all) with credit cards that have the numbers in vertical blocks and not in a horizontal line (like the Capitol One Venture card I have), or not being able to get the Exp Date 50% of the time. So I built this, and I think it is better performing, and I'd like to share the code with the world and you to use as you wish. I'm hoping that this side project over time will become useful to people and businesses, especially in places like pop-up shops, third-world countries, consumers on the go, and other places that don't have immediate access to swipe, EMV, or NFC tech.

Optional Settings

• self.ccScanner.recognitionLevel = .normal. You can increase speed or accurancy with the following settings:

.fastest
.fast
.normal (default)
.accurate
.veryaccurate

• self.ccScanner.cards = [.all]. You can allow only certain cards for your business if you'd like. For example, if you are in the USA and only accept Visa and Mastercard, the code would look like self.ccScanner.cards = [.visa, .mastercard]. Here are all of the avaiable card options:

.all (default)
.noneExceptCustom
.visa                     // Visa
.mastercard               // Mastercard
.americanExpress          // American Express
.discover                 // Discover
.chinaTUnion              // China T-Union
.chinaUnionPay            // China Union Pay
.dinersClubInternational  // Diners Club International
.interPayment             // Interpayment
.jcb                      // JCB
.maestroUK                // Maestro UK
.maestro                  // Maestro
.dankort                  // Dankort
.mir                      // MIR
.npsPridnestrovie         // NPS Pridnestrovie
.troy                     // Troy
.utap                     // UTAP
.custom                   // Custom Card (see below)

• You can also add custom cards. For example, if you're Target and you want to add Red Card Debit scanning, you'd simply add the following code. The binRange variable is the first six digits of your card (called an IIN or BIN, see Wikipedia). You can use anywhere between one and six sigits. The lengthRange variable is for card number length. Note: All custom card numbering schemes must follow the Luhn algorithm.

let targetCard = self.ccScanner.createCardType.new(binRange: "639463", lengthRange: "16")
self.ccScanner.addCustomCards(cards: [targetCard])

Both of these variables can be set in ranges, like this:

let targetCardRangeOne = self.ccScanner.createCardType.new(binRange: "639463-639465", lengthRange: "16-19")

You can add as many custom cards, or as many various ranges, as you wish:

self.ccScanner.addCustomCards(cards: [targetCard, oldNavyStoreCard])

Current Known Limitations, Issues, and Extra Details

• During my testing I was able to get about 99.9% accuracy, but every now and then there could be a false positive based on the number arrangement, the numbers in the CCV, and other small text on the back of the card, which by just pure luck of the number ordering could pass the validation check. This is because the code scans the entire image and doesn't break the text down into text blocks using Vision. There is room for improvement here, but overall the accuracy is quite good as it is.

• With expiration dates, some cards list multiple dates not related to the expiration date throughout the text on a card for various internal banking reasons. The code does its best to scan above a certain fractional height of the total image to pull out the correct date using .minimumTextHeight in the CCScanner class, but sometimes that doesn't work perfectly. I added a few extra checks against past dates (meaning you can't get the expiration date of an old card) and future dates (it won't get a date more than 20 years in the current future), and then sort remaining dates by furthest into the future first, and then use that date at the top of the list. It's a bit of guess work to narrow down the best possible choice, but it seems to be working well in my testing.

• This code will not work with cards that have a full 4-digit year as an Exp Date (i.e. it won't find 04/2027). This is something that can be added fairly easily. I am sure there are some cards out there that have the full year (some Amex cards if I remember correctly).

• This code will not pull the name off of the card yet.

• This code will not pull the CCV off of the card yet.

• There are some no-no's I do in the code that I hacked in just to get things up and running as quick as possible, the major one being I force cast a lot. This is just clean-up that needs to be done. The code should run fine without crashing though... I hope.