Getting 500x performance drop compared to pure NSRegularExpression

[iby]

Love the syntax. Love the features. But performance goes bad on large strings. Here's the test for digesting a crashlog.

internal func testRegex() {
    let content: String = try! String(contentsOf: PathUtility.getCrashlogUrl(name: "gifox"))
    let nsContent: NSString = content as NSString

    let pattern: String = "^\\s*0x([A-F0-9]+) - \\s*0x([A-F0-9]+) \\s*\\+?(.+?) .*? \\<([A-F0-9]{8}-[A-F0-9]{4}-[A-F0-9]{4}-[A-F0-9]{4}-[A-F0-9]{12})\\>"

    let regex: Regex = try! Regex(pattern: pattern, options: [RegexOptions.caseInsensitive, RegexOptions.anchorsMatchLines])
    let nsRegex: NSRegularExpression = try! NSRegularExpression(pattern: pattern, options: [NSRegularExpression.Options.caseInsensitive, NSRegularExpression.Options.anchorsMatchLines])

    var date: Date
    var intervalAll: Double
    var match: [String]

    date = Date()
    match = regex.findAll(in: content).map({ $0.group(at: 1)! })
    intervalAll = Date().timeIntervalSince(date)
    Swift.print("regex: ", intervalAll)
    Swift.print(match.count, match)

    date = Date()
    match = nsRegex.matches(in: content, range: NSRange(0 ..< content.characters.count)).map({ nsContent.substring(with: $0.rangeAt(1)) as String })
    intervalAll = Date().timeIntervalSince(date)
    Swift.print("nsRegex: ", intervalAll)
    Swift.print(match.count, match)

    // regex:  7.71533399820328
    // 327 ["101e03000", "102186000", "102348000", "1023bf000", "102451000", "1024de000", "1024fc000", "102573000", "1025c2000", "1027dc000", "102812000", "102847000", "1028bf000", "102914000", "102992000", "1029ee000", "102a3b000", "102a49000", "102a64000", "102f78000", "102f92000", "102fbd000", "102fc8000", "102fda000", "103034000", "103215000", "103220000", "103231000", "103243000", "103250000", "1032ee000", "10335e000", "10336c000", "10337d000", "10343d000", "1034e5000", "10352b000", "103560000", "10359b000", "103613000", "1036c3000", "105225000", "105412000", "1058f3000", "1058f9000", "107afe000", "107ce3000", "108a6e000", "108b4e000", "10913f000", "10982f000", "7fffa6cf2000", "7fffa7038000", "7fffa886e000", "7fffa8a30000", "7fffa8b9d000", "7fffa8bb6000", "7fffa90d0000", "7fffa9242000", "7fffa9257000", "7fffa964e000", "7fffa9665000", "7fffa966c000", "7fffa9681000", "7fffa9809000", "7fffa98bc000", "7fffa98bd000", "7fffa98fb000", "7fffa9b7b000", "7fffaa966000", "7fffaa967000", "7fffaaa6f000", "7fffaab9f000", "7fffaac46000", "7fffaace5000", "7fffaad90000", "7fffaade1000", "7fffaadf1000", "7fffaae3f000", "7fffaae7b000", "7fffaae87000", "7fffab094000", "7fffab1fd000", "7fffab5f1000", "7fffab5f2000", "7fffab5f6000", "7fffab900000", "7fffab904000", "7fffab90a000", "7fffab9a2000", "7fffab9bd000", "7fffab9bf000", "7fffab9c2000", "7fffabaa9000", "7fffabab7000", "7fffabbf3000", "7fffabc81000", "7fffabc95000", "7fffabf91000", "7fffac03e000", "7fffac4d8000", "7fffacb5b000", "7fffacd9f000", "7ffface05000", "7fffaceb7000", "7fffacf03000", "7fffacf04000", "7fffacf56000", "7fffad232000", "7fffad266000", "7fffad26f000", "7fffad3dc000", "7fffad48d000", "7fffad4ed000", "7fffad55e000", "7fffad62c000", "7fffad779000", "7fffad7af000", "7fffad853000", "7fffad91e000", "7fffadab5000", "7fffade87000", "7fffadf78000", "7fffae01c000", "7fffae0b2000", "7fffae0b9000", "7fffae10b000", "7fffae26b000", "7fffae270000", "7fffae361000", "7fffae385000", "7fffae3ad000", "7fffae3b0000", "7fffae55c000", "7fffaf165000", "7fffaf17f000", "7fffaf213000", "7fffaf407000", "7fffaf423000", "7fffaf95d000", "7fffb02a1000", "7fffb0480000", "7fffb0489000", "7fffb04d8000", "7fffb04f2000", "7fffb04fe000", "7fffb0501000", "7fffb0505000", "7fffb050a000", "7fffb0514000", "7fffb0520000", "7fffb055d000", "7fffb06d4000", "7fffb107d000", "7fffb1128000", "7fffb1270000", "7fffb14db000", "7fffb15a3000", "7fffb1aca000", "7fffb1aee000", "7fffb1bdc000", "7fffb1bdd000", "7fffb1dde000", "7fffb1e34000", "7fffb2344000", "7fffb2647000", "7fffb26bd000", "7fffb26e7000", "7fffb2a71000", "7fffb2ae1000", "7fffb2e90000", "7fffb4217000", "7fffb4404000", "7fffb5333000", "7fffb534f000", "7fffb535d000", "7fffb5380000", "7fffb547a000", "7fffb5506000", "7fffb5519000", "7fffb5535000", "7fffb553e000", "7fffb5542000", "7fffb5938000", "7fffb6052000", "7fffb6641000", "7fffb6f8d000", "7fffb705d000", "7fffb70ae000", "7fffb711a000", "7fffb74fa000", "7fffb75c5000", "7fffb76de000", "7fffb77b3000", "7fffb77d0000", "7fffb7a29000", "7fffb7cea000", "7fffb7d7a000", "7fffb7eba000", "7fffb7fda000", "7fffb8040000", "7fffb80c0000", "7fffb80cb000", "7fffb8105000", "7fffb81b6000", "7fffb81f6000", "7fffb8449000", "7fffb85ed000", "7fffb9d75000", "7fffb9d82000", "7fffba046000", "7fffba07b000", "7fffba773000", "7fffbadf9000", "7fffbae04000", "7fffbae0c000", "7fffbae0d000", "7fffbae22000", "7fffbaee0000", "7fffbaf2b000", "7fffbb752000", "7fffbba03000", "7fffbbbd8000", "7fffbbbf6000", "7fffbbc1f000", "7fffbbcd0000", "7fffbc128000", "7fffbc5b3000", "7fffbc637000", "7fffbccdc000", "7fffbccf7000", "7fffbcd0f000", "7fffbd9f8000", "7fffbda8a000", "7fffbdb38000", "7fffbdf7c000", "7fffbe672000", "7fffbeb24000", "7fffbebb6000", "7fffbecc5000", "7fffbecdb000", "7fffbecf0000", "7fffbecf2000", "7fffbeefc000", "7fffbf951000", "7fffbfe5c000", "7fffbff33000", "7fffbff61000", "7fffbff8a000", "7fffbffa6000", "7fffc0043000", "7fffc0045000", "7fffc0274000", "7fffc027c000", "7fffc027d000", "7fffc0282000", "7fffc02ef000", "7fffc031b000", "7fffc039b000", "7fffc039c000", "7fffc03ad000", "7fffc03bc000", "7fffc0413000", "7fffc043e000", "7fffc044f000", "7fffc0466000", "7fffc0467000", "7fffc0522000", "7fffc07a5000", "7fffc084a000", "7fffc0873000", "7fffc0883000", "7fffc0889000", "7fffc097c000", "7fffc0ba7000", "7fffc0ba9000", "7fffc0bc3000", "7fffc0bda000", "7fffc0eb5000", "7fffc0f30000", "7fffc1305000", "7fffc130a000", "7fffc1358000", "7fffc1375000", "7fffc13af000", "7fffc13c2000", "7fffc13c4000", "7fffc156d000", "7fffc1606000", "7fffc1614000", "7fffc1704000", "7fffc172e000", "7fffc174e000", "7fffc1753000", "7fffc175e000", "7fffc1766000", "7fffc176f000", "7fffc17f3000", "7fffc1825000", "7fffc182b000", "7fffc182c000", "7fffc1839000", "7fffc183a000", "7fffc1840000", "7fffc1843000", "7fffc1845000", "7fffc185e000", "7fffc185f000", "7fffc18ed000", "7fffc18f1000", "7fffc18f5000", "7fffc190e000", "7fffc1915000", "7fffc193f000", "7fffc1962000", "7fffc19aa000", "7fffc19c9000", "7fffc1a23000", "7fffc1a2d000", "7fffc1a37000", "7fffc1a40000", "7fffc1a4b000", "7fffc1a4f000", "7fffc1a51000", "7fffc1a59000", "7fffc1a6d000", "7fffc1a73000"]
    // nsRegex:  0.0154680013656616
    // 327 ["101e03000", "102186000", "102348000", "1023bf000", "102451000", "1024de000", "1024fc000", "102573000", "1025c2000", "1027dc000", "102812000", "102847000", "1028bf000", "102914000", "102992000", "1029ee000", "102a3b000", "102a49000", "102a64000", "102f78000", "102f92000", "102fbd000", "102fc8000", "102fda000", "103034000", "103215000", "103220000", "103231000", "103243000", "103250000", "1032ee000", "10335e000", "10336c000", "10337d000", "10343d000", "1034e5000", "10352b000", "103560000", "10359b000", "103613000", "1036c3000", "105225000", "105412000", "1058f3000", "1058f9000", "107afe000", "107ce3000", "108a6e000", "108b4e000", "10913f000", "10982f000", "7fffa6cf2000", "7fffa7038000", "7fffa886e000", "7fffa8a30000", "7fffa8b9d000", "7fffa8bb6000", "7fffa90d0000", "7fffa9242000", "7fffa9257000", "7fffa964e000", "7fffa9665000", "7fffa966c000", "7fffa9681000", "7fffa9809000", "7fffa98bc000", "7fffa98bd000", "7fffa98fb000", "7fffa9b7b000", "7fffaa966000", "7fffaa967000", "7fffaaa6f000", "7fffaab9f000", "7fffaac46000", "7fffaace5000", "7fffaad90000", "7fffaade1000", "7fffaadf1000", "7fffaae3f000", "7fffaae7b000", "7fffaae87000", "7fffab094000", "7fffab1fd000", "7fffab5f1000", "7fffab5f2000", "7fffab5f6000", "7fffab900000", "7fffab904000", "7fffab90a000", "7fffab9a2000", "7fffab9bd000", "7fffab9bf000", "7fffab9c2000", "7fffabaa9000", "7fffabab7000", "7fffabbf3000", "7fffabc81000", "7fffabc95000", "7fffabf91000", "7fffac03e000", "7fffac4d8000", "7fffacb5b000", "7fffacd9f000", "7ffface05000", "7fffaceb7000", "7fffacf03000", "7fffacf04000", "7fffacf56000", "7fffad232000", "7fffad266000", "7fffad26f000", "7fffad3dc000", "7fffad48d000", "7fffad4ed000", "7fffad55e000", "7fffad62c000", "7fffad779000", "7fffad7af000", "7fffad853000", "7fffad91e000", "7fffadab5000", "7fffade87000", "7fffadf78000", "7fffae01c000", "7fffae0b2000", "7fffae0b9000", "7fffae10b000", "7fffae26b000", "7fffae270000", "7fffae361000", "7fffae385000", "7fffae3ad000", "7fffae3b0000", "7fffae55c000", "7fffaf165000", "7fffaf17f000", "7fffaf213000", "7fffaf407000", "7fffaf423000", "7fffaf95d000", "7fffb02a1000", "7fffb0480000", "7fffb0489000", "7fffb04d8000", "7fffb04f2000", "7fffb04fe000", "7fffb0501000", "7fffb0505000", "7fffb050a000", "7fffb0514000", "7fffb0520000", "7fffb055d000", "7fffb06d4000", "7fffb107d000", "7fffb1128000", "7fffb1270000", "7fffb14db000", "7fffb15a3000", "7fffb1aca000", "7fffb1aee000", "7fffb1bdc000", "7fffb1bdd000", "7fffb1dde000", "7fffb1e34000", "7fffb2344000", "7fffb2647000", "7fffb26bd000", "7fffb26e7000", "7fffb2a71000", "7fffb2ae1000", "7fffb2e90000", "7fffb4217000", "7fffb4404000", "7fffb5333000", "7fffb534f000", "7fffb535d000", "7fffb5380000", "7fffb547a000", "7fffb5506000", "7fffb5519000", "7fffb5535000", "7fffb553e000", "7fffb5542000", "7fffb5938000", "7fffb6052000", "7fffb6641000", "7fffb6f8d000", "7fffb705d000", "7fffb70ae000", "7fffb711a000", "7fffb74fa000", "7fffb75c5000", "7fffb76de000", "7fffb77b3000", "7fffb77d0000", "7fffb7a29000", "7fffb7cea000", "7fffb7d7a000", "7fffb7eba000", "7fffb7fda000", "7fffb8040000", "7fffb80c0000", "7fffb80cb000", "7fffb8105000", "7fffb81b6000", "7fffb81f6000", "7fffb8449000", "7fffb85ed000", "7fffb9d75000", "7fffb9d82000", "7fffba046000", "7fffba07b000", "7fffba773000", "7fffbadf9000", "7fffbae04000", "7fffbae0c000", "7fffbae0d000", "7fffbae22000", "7fffbaee0000", "7fffbaf2b000", "7fffbb752000", "7fffbba03000", "7fffbbbd8000", "7fffbbbf6000", "7fffbbc1f000", "7fffbbcd0000", "7fffbc128000", "7fffbc5b3000", "7fffbc637000", "7fffbccdc000", "7fffbccf7000", "7fffbcd0f000", "7fffbd9f8000", "7fffbda8a000", "7fffbdb38000", "7fffbdf7c000", "7fffbe672000", "7fffbeb24000", "7fffbebb6000", "7fffbecc5000", "7fffbecdb000", "7fffbecf0000", "7fffbecf2000", "7fffbeefc000", "7fffbf951000", "7fffbfe5c000", "7fffbff33000", "7fffbff61000", "7fffbff8a000", "7fffbffa6000", "7fffc0043000", "7fffc0045000", "7fffc0274000", "7fffc027c000", "7fffc027d000", "7fffc0282000", "7fffc02ef000", "7fffc031b000", "7fffc039b000", "7fffc039c000", "7fffc03ad000", "7fffc03bc000", "7fffc0413000", "7fffc043e000", "7fffc044f000", "7fffc0466000", "7fffc0467000", "7fffc0522000", "7fffc07a5000", "7fffc084a000", "7fffc0873000", "7fffc0883000", "7fffc0889000", "7fffc097c000", "7fffc0ba7000", "7fffc0ba9000", "7fffc0bc3000", "7fffc0bda000", "7fffc0eb5000", "7fffc0f30000", "7fffc1305000", "7fffc130a000", "7fffc1358000", "7fffc1375000", "7fffc13af000", "7fffc13c2000", "7fffc13c4000", "7fffc156d000", "7fffc1606000", "7fffc1614000", "7fffc1704000", "7fffc172e000", "7fffc174e000", "7fffc1753000", "7fffc175e000", "7fffc1766000", "7fffc176f000", "7fffc17f3000", "7fffc1825000", "7fffc182b000", "7fffc182c000", "7fffc1839000", "7fffc183a000", "7fffc1840000", "7fffc1843000", "7fffc1845000", "7fffc185e000", "7fffc185f000", "7fffc18ed000", "7fffc18f1000", "7fffc18f5000", "7fffc190e000", "7fffc1915000", "7fffc193f000", "7fffc1962000", "7fffc19aa000", "7fffc19c9000", "7fffc1a23000", "7fffc1a2d000", "7fffc1a37000", "7fffc1a40000", "7fffc1a4b000", "7fffc1a4f000", "7fffc1a51000", "7fffc1a59000", "7fffc1a6d000", "7fffc1a73000"]
}

[dileping]

Hey @ianbytchek , thanks for reporting. Could you, please, post the log?

[kengruven]

My first thought is that this looks like a symptom of #38 -- detecting extended grapheme clusters can be really slow -- but the NSRegularExpression example also (incorrectly) uses .characters.count here, so there's something else going on in the library, too.

[chriszielinski]

I've managed to track it down to the MatchSequence.Iterator. I believe the performance drop is attributed to the Match initializations that occur within the iterator. You can see the temporary heap allocations are almost doubled.

You can play around with the test cases in the performance branch on my fork.