LineFromCorpusFromLine

LineFromCorpusFromLine.swift

protocol LineFromCorpusConvertible: CustomStringConvertible, Codable {
    /// The word, on lowercased form
    var wordForm: WordForm { get }

    /// Part of speech tag, only first major tag, .e.g from line `han    PN.UTR.SIN.DEF.SUB`, we only save `PN`
    var partOfSpeechTag: PartOfSpeech { get }

    /// Base form(s) of word
    var lemgrams: Lemgrams { get }

    /// Whether this is a compound word or not, an example of a compound word is 🇸🇪_"stämband"_,
    /// consisting of the word _"stäm"_ and the word _"band"_.
    var isCompoundWord: Bool { get }

    /// The total frequency, the number of occurences of the word form in the corpus.
    var numberOfOccurencesInCorpus: Int { get }

    /// The relative frequency of the word form in the corpus per 1 million words.
    var relativeFrequencyPerOneMillion: Int { get }

    // MARK: Meta properties
    /// The index of this parsed line in the corpus
    var indexOfLineInCorpus: Int { get }
}

protocol LineFromCorpusConvertibleByProxy: LineFromCorpusConvertible {
    associatedtype FromLine: LineFromCorpusConvertible
    var line: FromLine { get }
}

extension LineFromCorpusConvertibleByProxy {
    var wordForm: WordForm { line.wordForm }
    var partOfSpeechTag: PartOfSpeech { line.partOfSpeechTag }
    var lemgrams: Lemgrams { line.lemgrams }
    var isCompoundWord: Bool { line.isCompoundWord }
    var numberOfOccurencesInCorpus: Int { line.numberOfOccurencesInCorpus }
    var relativeFrequencyPerOneMillion: Int { line.relativeFrequencyPerOneMillion }
    var indexOfLineInCorpus: Int { line.indexOfLineInCorpus }
}

protocol LineFromCorpusFromLineAndOther: LineFromCorpusConvertibleByProxy {
    associatedtype Other
    init(line: FromLine, other: Other)
}

protocol LineFromCorpusFromLine: LineFromCorpusFromLineAndOther where Other == Void {
    init(line: FromLine)
}

extension LineFromCorpusFromLine {
    init(line: FromLine, other: Void) {
        self.init(line: line)
    }
}