Identifying Subtitles.swift

// Playground - noun: a place where people can play

import UIKit

enum SubtitleType {
    case SubRip, SubStationAlpha, LRC, SAMI, YouTube, SubViewer, MicroDVD, WebVTT, TTML, Unknown
    
    func description() -> String {
        switch self {
        case .SubRip:
            return "SubRip subtitle (.srt)"
        case .SubStationAlpha:
            return "Advanced Sub Station subtitle (.ass/.ssa)"
        case .LRC:
            return "LRC lyrics (.lrc)"
        case .SAMI:
            return "SAMI subtitle (.smi)"
        case .YouTube:
            return "YouTube Captions/SubViewer subtitle (.sbv/.sub)"
        case .SubViewer:
            return "SubViewer subtitle (.sub)"
        case .MicroDVD:
            return "MicroDVD subtitle (.sub)"
        case .WebVTT:
            return "WebVTT subtitle (.vtt)"
        case .TTML:
            return "TTML subtitle (.xml)"
        case .Unknown:
            return "text / unknown"
        }
    }
}

func identifyText(text: String) -> SubtitleType? {
    let scanner = NSScanner(string: text)

    // check if the subtitle starts with the specified text
    if scanner.scanString("<sami>", intoString: nil) {
        return .SAMI
    }
    
    // check if the subtitle starts with the specified text
    if scanner.scanString("WEBVTT", intoString: nil) {
        return .WebVTT
    }
    
    // check if the subtitle contains at some point the specified text
    if scanner.scanUpToString("[Events]", intoString: nil) {
        // if the scanner is not at the end then we found the text
        if !scanner.atEnd {
            return .SubStationAlpha
        }
        
        // reset the scanner
        scanner.scanLocation = 0
    }

    // check if the subtitle contains at some point the specified text
    if scanner.scanUpToString("[SUBTITLE]", intoString: nil) {
        // if the scanner is not at the end then we found the text
        if !scanner.atEnd {
            return .SubViewer
        }
        
        // reset the scanner
        scanner.scanLocation = 0
    }

    if scanner.scanInt(nil) {
        var timeLine:NSString?

        if scanner.scanCharactersFromSet(NSCharacterSet(charactersInString: "0123456789:.,-> "), intoString: &timeLine) {
            
            if let timeLine = timeLine {
                if timeLine.componentsSeparatedByString(" --> ").count == 2 {
                    return .SubRip
                } else if timeLine.componentsSeparatedByString(",").count == 2 {
                    return .YouTube
                }
            }
        }
        
        // reset the scanner
        scanner.scanLocation = 0
    }
    
    // check if the subtitle starts with square brackets
    if scanner.scanString("[", intoString: nil) {
        // check if the subtitle contains text inside square brackets
        if scanner.scanUpToString("]", intoString: nil) {
            // if the scanner is not at the end then we found the text
            if !scanner.atEnd {
                return .LRC
            }
        }

        // reset the scanner
        scanner.scanLocation = 0
    }
    
    // check if the subtitle contains at some point the specified text
    if scanner.scanUpToString("[0", intoString: nil) {
        // check if the subtitle contains text inside square brackets
        if scanner.scanUpToString("]", intoString: nil) {
            // if the scanner is not at the end then we found the text
            if !scanner.atEnd {
                return .LRC
            }
        }
        
        // reset the scanner
        scanner.scanLocation = 0
    }

    // NOTE: this seems to work, but surely it could be tricked by a badly formed .SubRip or .YouTube ?
    // if scanner.scanCharactersFromSet(NSCharacterSet(charactersInString: "0123456789{}"), intoString: nil) {
    //     return .MicroDVD
    // }
    
    // check if the subtitle starts with curly braces
    if scanner.scanString("{", intoString: nil) {
        // check if the subtitle contains text inside curly braces
        if scanner.scanUpToString("}", intoString: nil) {
            // if the scanner is not at the end then we found the text
            if !scanner.atEnd {
                return .MicroDVD
            }
            
            // reset the scanner
            scanner.scanLocation = 0
        }
    }

    // check if the subtitle contains at some point the specified text
    if scanner.scanUpToString("<tt xmlns=\"http://www.w3.org/ns/ttml\">", intoString: nil) {
        // if the scanner is not at the end then we found the text
        if !scanner.atEnd {
            return .TTML
        }
    }

    return .Unknown
}


if let type = identifyText("[al:title]\n[00:12]some words") {
    type.description()
}

if let type = identifyText("Hello there!\n[00:00]some words") {
    type.description()
}


if let type = identifyText("<SAMI><HEAD>") {
    type.description()
}

if let type = identifyText("[Script Info]\n; This is a Sub Station Alpha v4 script.\n\n[Events]") {
    type.description()
}

if let type = identifyText("1\n00:00:12,000 --> 00:00:15,123\nSome text") {
    type.description()
}

if let type = identifyText("0:00:00.599,0:00:04.160\n>> ALICE: Hi, my name is Alice Miller and this is John Brown\n\n0:00:04.160,0:00:06.770\n>> JOHN: and we're the owners of Miller Bakery.") {
    type.description()
}

if let type = identifyText("{0}{15}Hello!") {
    type.description()
}

if let type = identifyText("[INFORMATION]\n[TITLE]\n[AUTHOR]andycat\n[SOURCE]\n[FILEPATH]\n[DELAY]0\n[COMMENT]Edited with Jubler subtitle editor\n[END INFORMATION]\n[SUBTITLE]\n[COLF]&HFFFFFF,[STYLE]bd,[SIZE]18,[FONT]Arial\n00:00:00.00,00:00:10.00\nThis is my first subtitle!") {
    type.description()
}

if let type = identifyText("WEBVTT\n\n1\n00:00:01.000 --> 00:00:10.000\nThis is the first line of text, displaying from 1-10 seconds\n\n2\n00:00:15.000 --> 00:00:20.000\nAnd the second line of text\nseparated over two lines") {
    type.description()
}

if let type = identifyText("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<tt xmlns=\"http://www.w3.org/ns/ttml\">\n<head>\n<metadata xmlns:ttm=\"http://www.w3.org/ns/ttml#metadata\">\n<ttm:title>Timed Text TTML Example</ttm:title>\n<ttm:copyright>The Authors (c) 2006</ttm:copyright>\n</metadata>") {
    type.description()
}

if let type = identifyText("asdfa sddfs f jsdkf asljdfl") {
    type.description()
}