Regular expression not working in iOS, works fine elsewhere

Question

I have a regular expression to parse pod names from the CocoaPods podfile; I wrote this to be used as part of a vulnerability scan.

This RegEx can be seen to be working in RegexR, as can be seen here. However, when used in NSRegularExpression, no matches are returned from the same input.

I have known NSRegularExpression to be a bit finicky before but I can't work out the problem on this occasion.

The regex is: /(?<=pod ')[A-Za-z/]+(?=')/g

My Swift code is:

guard let podNameRegex = try? NSRegularExpression(pattern: "/(?<=pod ')[A-Za-z/]+(?=')/g", options: [.useUnicodeWordBoundaries]) else {
    fputs("Failed to instantiate pod name regex.", stderr)
    throw NSError()
}

let podfilePodNameMatches = podNameRegex.matches(in: podfileString, options: [], range: NSRangeFromString(podfileString))
let matchStrings = podfilePodNameMatches.map({ (podfileString as NSString).substring(with: $0.range) })

However, no matches are returned.

Does anyone know where the issue lies? TIA

`pattern: "(?<=pod ')[A-Za-z/]+(?=')"`. You should not use regex delimiters and flags like in Perl, JavaScript, or Ruby regex literal notation. BTW, `(?<=pod ')[^']+(?=')` might be more generic for you. Ah, and do not use `.useUnicodeWordBoundaries` — Wiktor Stribiżew, Mar 05 '19 at 15:13

score 1 · Accepted Answer · answered Mar 05 '19 at 15:26

You don't need to use the /{pattern}/g syntax in Swift.

This worked for me in a playground

import UIKit
import PlaygroundSupport

let podfileString = """
platform :ios, '11.0'

inhibit_all_warnings!
use_frameworks!
use_modular_headers!

def shared_pods
    pod 'CocoaLumberjack/Swift'
    pod 'PromiseKit'
    pod 'GZIP'
    pod 'Eureka' => '4.2.0', :configuration => 'Debug'
    pod 'RATreeView'
    pod 'LicensesViewController'
end

target 'DigiMe' do
    pod 'Masonry'
    pod 'Instabug'
    pod 'SAMKeychain'
    pod 'Mixpanel'
    pod 'SDWebImage'
    pod 'FLAnimatedImage'
    pod 'SnapKit', '~> 4.0.0'
    pod 'ObjectMapper'
    pod 'SwiftLint'
    pod 'MarqueeLabel/Swift'
    pod 'Alamofire'
    pod 'PromiseKit/Alamofire'
    pod 'UIColor_Hex_Swift'
    pod 'SwiftyJSON'
    pod 'Hero'
    pod 'GPUImage'
    shared_pods
end

target 'DigiMeTests' do

    ####### SPECTA ######
    pod 'Specta',       '~> 1.0'
    pod 'Expecta',      '~> 1.0'   # expecta matchers
    pod 'OHHTTPStubs/Swift'
    #####################

    shared_pods
end

post_install do |installer|
    installer.pods_project.targets.each do |target|
        target.build_configurations.each do |config|
            preprocessor_definitions = config.build_settings['GCC_PREPROCESSOR_DEFINITIONS']
            preprocessor_definitions = ['$(inherited)'] if preprocessor_definitions == nil
            preprocessor_definitions.push 'MIXPANEL_NO_IFA' if target.to_s.include? 'Mixpanel'
            config.build_settings['GCC_PREPROCESSOR_DEFINITIONS'] = preprocessor_definitions
            config.build_settings['CLANG_ENABLE_CODE_COVERAGE'] = 'NO'
        end
    end
end
"""

let range = NSRange(location: 0, length: podfileString.utf16.count)
let regex = try! NSRegularExpression(pattern: "(?<=pod ')[A-Za-z/]+(?=')")
let matches = regex.matches(in: podfileString, options: [], range: range)
let matchStrings = matches.map({ (podfileString as NSString).substring(with: $0.range) })
print(matchStrings)

Output:

["CocoaLumberjack/Swift", "PromiseKit", "GZIP", "Eureka", "RATreeView", "LicensesViewController", "Masonry", "Instabug", "SAMKeychain", "Mixpanel", "SDWebImage", "FLAnimatedImage", "SnapKit", "ObjectMapper", "SwiftLint", "MarqueeLabel/Swift", "Alamofire", "PromiseKit/Alamofire", "SwiftyJSON", "Hero", "GPUImage", "Specta", "Expecta", "OHHTTPStubs/Swift"]

Thanks for the tip about flags in iOS. Implementing the above code resolved the issue, though, just for the sake of closure, I should not that it was less the RegEx that wasn't working as much as the range I was passing in. Note that I was using `NSRangeFromString` and this answer creates a range with the UTF-16 view of the string. These obviously return different values. I am surprised that UTF-16 is needed though as I thought the standard was UTF-8. Also, +1 for the multi-line String, I hadn't seen that before. — Jacob King, Mar 05 '19 at 15:36
Thanks, Yeah utf16 apparently helps with special characters and emoji's (I'm no expert on that matter, just something picked up along the way), glad to of helped — Scriptable, Mar 05 '19 at 15:38

Regular expression not working in iOS, works fine elsewhere

1 Answers1