@swiftbrowser/swift 0.0.0-stage → 0.2.0

package/Sources/NaturalLanguage/NaturalLanguage.swift

@@ -0,0 +1,135 @@
+// `import NaturalLanguage` for the browser: `NLLanguage` and
+// `NLLanguageRecognizer`, which apps use to tell what language a text is in
+// (Ice Cubes tags a post with it and asks before posting in another language
+// than the one selected; WishKit compares feedback with the app's language).
+//
+// Apple's recognizer is a trained model; this one is a small, explainable
+// identifier (LanguageIdentifier.swift). It finds the writing system of each
+// letter, which settles most languages that have a script of their own
+// (Greek, Hebrew, Thai, Korean, Georgian…). Within a script shared by several
+// languages it counts their most frequent words and their distinctive
+// letters: Latin-script European languages, Vietnamese and Indonesian, Russian
+// and its Cyrillic neighbours, Arabic, Persian and Urdu, Hindi and Marathi,
+// Chinese (simplified or traditional) and Japanese. Its probabilities are
+// honest about the evidence: a text with a handful of common words is a
+// confident guess, a short or unusual one is not, and a text without letters
+// has no dominant language.
+
+import Foundation
+
+/// A language, by its BCP 47 tag (`"en"`, `"zh-Hans"`), with Apple's constants.
+public struct NLLanguage: RawRepresentable, Hashable, Sendable, Codable {
+    public let rawValue: String
+    public init(rawValue: String) { self.rawValue = rawValue }
+    public init(_ rawValue: String) { self.rawValue = rawValue }
+
+    public static let undetermined = NLLanguage("und")
+    public static let amharic = NLLanguage("am")
+    public static let arabic = NLLanguage("ar")
+    public static let armenian = NLLanguage("hy")
+    public static let bengali = NLLanguage("bn")
+    public static let bulgarian = NLLanguage("bg")
+    public static let burmese = NLLanguage("my")
+    public static let catalan = NLLanguage("ca")
+    public static let cherokee = NLLanguage("chr")
+    public static let croatian = NLLanguage("hr")
+    public static let czech = NLLanguage("cs")
+    public static let danish = NLLanguage("da")
+    public static let dutch = NLLanguage("nl")
+    public static let english = NLLanguage("en")
+    public static let finnish = NLLanguage("fi")
+    public static let french = NLLanguage("fr")
+    public static let georgian = NLLanguage("ka")
+    public static let german = NLLanguage("de")
+    public static let greek = NLLanguage("el")
+    public static let gujarati = NLLanguage("gu")
+    public static let hebrew = NLLanguage("he")
+    public static let hindi = NLLanguage("hi")
+    public static let hungarian = NLLanguage("hu")
+    public static let icelandic = NLLanguage("is")
+    public static let indonesian = NLLanguage("id")
+    public static let italian = NLLanguage("it")
+    public static let japanese = NLLanguage("ja")
+    public static let kannada = NLLanguage("kn")
+    public static let kazakh = NLLanguage("kk")
+    public static let khmer = NLLanguage("km")
+    public static let korean = NLLanguage("ko")
+    public static let lao = NLLanguage("lo")
+    public static let malay = NLLanguage("ms")
+    public static let malayalam = NLLanguage("ml")
+    public static let marathi = NLLanguage("mr")
+    public static let mongolian = NLLanguage("mn")
+    public static let norwegian = NLLanguage("nb")
+    public static let oriya = NLLanguage("or")
+    public static let persian = NLLanguage("fa")
+    public static let polish = NLLanguage("pl")
+    public static let portuguese = NLLanguage("pt")
+    public static let punjabi = NLLanguage("pa")
+    public static let romanian = NLLanguage("ro")
+    public static let russian = NLLanguage("ru")
+    public static let simplifiedChinese = NLLanguage("zh-Hans")
+    public static let sinhalese = NLLanguage("si")
+    public static let slovak = NLLanguage("sk")
+    public static let spanish = NLLanguage("es")
+    public static let swedish = NLLanguage("sv")
+    public static let tamil = NLLanguage("ta")
+    public static let telugu = NLLanguage("te")
+    public static let thai = NLLanguage("th")
+    public static let tibetan = NLLanguage("bo")
+    public static let traditionalChinese = NLLanguage("zh-Hant")
+    public static let turkish = NLLanguage("tr")
+    public static let ukrainian = NLLanguage("uk")
+    public static let urdu = NLLanguage("ur")
+    public static let vietnamese = NLLanguage("vi")
+}
+
+/// Identifies the language of the text it is given (LanguageIdentifier.swift
+/// says how). Strings passed to `processString` add up until `reset()`.
+open class NLLanguageRecognizer {
+    private var text = ""
+    private var cached: [NLLanguage: Double]?
+
+    /// Prior probabilities: each language's likelihood is multiplied by its
+    /// hint, and a language without one by the smallest hint given.
+    open var languageHints: [NLLanguage: Double] = [:] { didSet { cached = nil } }
+    /// When not empty, the only languages the recognizer answers with.
+    open var languageConstraints: [NLLanguage] = [] { didSet { cached = nil } }
+
+    public init() {}
+
+    /// The most likely language of `string`, or nil when it has no letters.
+    open class func dominantLanguage(for string: String) -> NLLanguage? {
+        let recognizer = NLLanguageRecognizer()
+        recognizer.processString(string)
+        return recognizer.dominantLanguage
+    }
+
+    open func processString(_ string: String) {
+        text += text.isEmpty ? string : " " + string
+        cached = nil
+    }
+
+    open func reset() {
+        text = ""
+        cached = nil
+    }
+
+    /// The most likely language of the text so far, or nil when it has no letters.
+    open var dominantLanguage: NLLanguage? {
+        hypotheses.max { $0.value < $1.value || ($0.value == $1.value && $0.key.rawValue > $1.key.rawValue) }?.key
+    }
+
+    /// The `maxHypotheses` most likely languages with their probabilities
+    /// (which add up to at most 1; the rest is doubt).
+    open func languageHypotheses(withMaximum maxHypotheses: Int) -> [NLLanguage: Double] {
+        let ranked = hypotheses.sorted { $0.value > $1.value || ($0.value == $1.value && $0.key.rawValue < $1.key.rawValue) }
+        return Dictionary(uniqueKeysWithValues: ranked.prefix(max(0, maxHypotheses)).map { ($0.key, $0.value) })
+    }
+
+    private var hypotheses: [NLLanguage: Double] {
+        if let cached { return cached }
+        let result = _LanguageIdentifier.hypotheses(for: text, hints: languageHints, constraints: Set(languageConstraints))
+        cached = result
+        return result
+    }
+}

package/Sources/OSLog/OSLog.swift

@@ -0,0 +1,6 @@
+// `import OSLog` for the browser: on Apple platforms the module is where
+// `Logger` lives, over the same unified logging as `os`; here it re-exports
+// the os stand-in (its logs go to standard error, which the page's console
+// shows).
+
+@_exported import os