Add language catalog selector

2026-09-15 23:20:48 +02:00
parent b43c5416f2
commit c8242a844a
10 changed files with 425 additions and 14 deletions
@@ -535,6 +535,32 @@ public struct LanguageSelection: Codable, Equatable, Identifiable, Sendable {
    }
 }

+public struct LanguageCatalogItem: Codable, Equatable, Identifiable, Sendable {
+    public let id: String
+    public var name: String
+    public var familyCategory: String
+    public var aliases: [String]
+    public var searchTerms: [String]
+
+    public init(
+        id: String,
+        name: String,
+        familyCategory: String,
+        aliases: [String] = [],
+        searchTerms: [String] = []
+    ) {
+        self.id = id
+        self.name = name
+        self.familyCategory = familyCategory
+        self.aliases = aliases
+        self.searchTerms = searchTerms
+    }
+
+    public var language: LanguageSelection {
+        LanguageSelection(id: id, name: name)
+    }
+}
+
 public struct DialectSelection: Codable, Equatable, Identifiable, Sendable {
    public let id: String
    public var name: String
@@ -0,0 +1,172 @@
+import Foundation
+
+public struct LocalLanguageCatalog: Sendable {
+    public let items: [LanguageCatalogItem]
+
+    public init(items: [LanguageCatalogItem] = LocalLanguageCatalog.worldwideItems) {
+        self.items = items
+    }
+
+    public var familyCategories: [String] {
+        Array(Set(items.map(\.familyCategory)))
+            .sorted { $0.localizedCaseInsensitiveCompare($1) == .orderedAscending }
+    }
+
+    public func languages(
+        matching query: String,
+        familyCategory: String? = nil
+    ) -> [LanguageCatalogItem] {
+        let searchTokens = Self.searchTokens(for: query)
+        let selectedFamilyCategory = familyCategory?.trimmingCharacters(in: .whitespacesAndNewlines)
+
+        return items
+            .filter { item in
+                let matchesSearch = searchTokens.isEmpty || searchTokens.allSatisfy { token in
+                    Self.searchableTerms(for: item).contains { $0.contains(token) }
+                }
+                let matchesFamily = selectedFamilyCategory?.isEmpty != false || item.familyCategory == selectedFamilyCategory
+                return matchesSearch && matchesFamily
+            }
+            .sorted { $0.name.localizedCaseInsensitiveCompare($1.name) == .orderedAscending }
+    }
+
+    public static let worldwideItems: [LanguageCatalogItem] =
+        middleEastAndNorthAfrica + europe + southAsia + eastAsia + southeastAsia + africa + americas + oceaniaAndPacific
+
+    private static let middleEastAndNorthAfrica: [LanguageCatalogItem] = [
+        item("arabic", "Arabic", "Middle East & North Africa", aliases: ["العربية"], terms: ["عربي"]),
+        item("hebrew", "Hebrew", "Middle East & North Africa", aliases: ["עברית"]),
+        item("persian", "Persian", "Middle East & North Africa", aliases: ["Farsi", "فارسی"]),
+        item("turkish", "Turkish", "Middle East & North Africa", aliases: ["Türkçe"]),
+        item("kurdish", "Kurdish", "Middle East & North Africa", aliases: ["Kurdî"]),
+        item("armenian", "Armenian", "Middle East & North Africa"),
+        item("azerbaijani", "Azerbaijani", "Middle East & North Africa", aliases: ["Azeri"]),
+        item("tamazight", "Tamazight", "Middle East & North Africa", aliases: ["Berber"])
+    ]
+
+    private static let europe: [LanguageCatalogItem] = [
+        item("english", "English", "Europe", aliases: ["English language"]),
+        item("german", "German", "Europe", aliases: ["Deutsch"]),
+        item("french", "French", "Europe", aliases: ["Français"]),
+        item("spanish", "Spanish", "Europe", aliases: ["Español"]),
+        item("portuguese", "Portuguese", "Europe", aliases: ["Português"]),
+        item("italian", "Italian", "Europe", aliases: ["Italiano"]),
+        item("russian", "Russian", "Europe", aliases: ["Русский"]),
+        item("ukrainian", "Ukrainian", "Europe", aliases: ["Українська"]),
+        item("polish", "Polish", "Europe", aliases: ["Polski"]),
+        item("dutch", "Dutch", "Europe", aliases: ["Nederlands"]),
+        item("swedish", "Swedish", "Europe", aliases: ["Svenska"]),
+        item("norwegian", "Norwegian", "Europe", aliases: ["Norsk"]),
+        item("danish", "Danish", "Europe", aliases: ["Dansk"]),
+        item("finnish", "Finnish", "Europe", aliases: ["Suomi"]),
+        item("greek", "Greek", "Europe", aliases: ["Ελληνικά"]),
+        item("romanian", "Romanian", "Europe", aliases: ["Română"]),
+        item("czech", "Czech", "Europe", aliases: ["Čeština"]),
+        item("hungarian", "Hungarian", "Europe", aliases: ["Magyar"]),
+        item("serbian", "Serbian", "Europe"),
+        item("croatian", "Croatian", "Europe"),
+        item("bulgarian", "Bulgarian", "Europe")
+    ]
+
+    private static let southAsia: [LanguageCatalogItem] = [
+        item("hindi", "Hindi", "South Asia", aliases: ["हिन्दी"]),
+        item("urdu", "Urdu", "South Asia", aliases: ["اردو"]),
+        item("punjabi", "Punjabi", "South Asia", aliases: ["ਪੰਜਾਬੀ"]),
+        item("bengali", "Bengali", "South Asia", aliases: ["বাংলা"]),
+        item("tamil", "Tamil", "South Asia", aliases: ["தமிழ்"]),
+        item("telugu", "Telugu", "South Asia", aliases: ["తెలుగు"]),
+        item("marathi", "Marathi", "South Asia", aliases: ["मराठी"]),
+        item("gujarati", "Gujarati", "South Asia", aliases: ["ગુજરાતી"]),
+        item("malayalam", "Malayalam", "South Asia", aliases: ["മലയാളം"]),
+        item("kannada", "Kannada", "South Asia", aliases: ["ಕನ್ನಡ"]),
+        item("sinhala", "Sinhala", "South Asia"),
+        item("nepali", "Nepali", "South Asia", aliases: ["नेपाली"]),
+        item("pashto", "Pashto", "South Asia", aliases: ["پښتو"])
+    ]
+
+    private static let eastAsia: [LanguageCatalogItem] = [
+        item("mandarin-chinese", "Mandarin Chinese", "East Asia", aliases: ["Mandarin", "中文", "普通话"]),
+        item("cantonese", "Cantonese", "East Asia", aliases: ["廣東話"]),
+        item("japanese", "Japanese", "East Asia", aliases: ["日本語"]),
+        item("korean", "Korean", "East Asia", aliases: ["한국어"]),
+        item("mongolian", "Mongolian", "East Asia")
+    ]
+
+    private static let southeastAsia: [LanguageCatalogItem] = [
+        item("indonesian", "Indonesian", "Southeast Asia", aliases: ["Bahasa Indonesia"]),
+        item("malay", "Malay", "Southeast Asia", aliases: ["Bahasa Melayu"]),
+        item("vietnamese", "Vietnamese", "Southeast Asia", aliases: ["Tiếng Việt"]),
+        item("thai", "Thai", "Southeast Asia", aliases: ["ไทย"]),
+        item("tagalog", "Tagalog", "Southeast Asia", aliases: ["Filipino"]),
+        item("burmese", "Burmese", "Southeast Asia", aliases: ["Myanmar"]),
+        item("khmer", "Khmer", "Southeast Asia"),
+        item("lao", "Lao", "Southeast Asia"),
+        item("javanese", "Javanese", "Southeast Asia")
+    ]
+
+    private static let africa: [LanguageCatalogItem] = [
+        item("swahili", "Swahili", "Africa", aliases: ["Kiswahili"]),
+        item("amharic", "Amharic", "Africa", aliases: ["አማርኛ"]),
+        item("hausa", "Hausa", "Africa"),
+        item("yoruba", "Yoruba", "Africa"),
+        item("igbo", "Igbo", "Africa"),
+        item("zulu", "Zulu", "Africa"),
+        item("xhosa", "Xhosa", "Africa"),
+        item("somali", "Somali", "Africa"),
+        item("oromo", "Oromo", "Africa"),
+        item("afrikaans", "Afrikaans", "Africa"),
+        item("wolof", "Wolof", "Africa"),
+        item("shona", "Shona", "Africa")
+    ]
+
+    private static let americas: [LanguageCatalogItem] = [
+        item("haitian-creole", "Haitian Creole", "Americas", aliases: ["Kreyòl"]),
+        item("quechua", "Quechua", "Americas"),
+        item("guarani", "Guarani", "Americas"),
+        item("navajo", "Navajo", "Americas"),
+        item("inuktitut", "Inuktitut", "Americas")
+    ]
+
+    private static let oceaniaAndPacific: [LanguageCatalogItem] = [
+        item("maori", "Maori", "Oceania & Pacific", aliases: ["Te Reo Māori"]),
+        item("hawaiian", "Hawaiian", "Oceania & Pacific"),
+        item("samoan", "Samoan", "Oceania & Pacific"),
+        item("tongan", "Tongan", "Oceania & Pacific"),
+        item("fijian", "Fijian", "Oceania & Pacific")
+    ]
+
+    private static func item(
+        _ id: String,
+        _ name: String,
+        _ familyCategory: String,
+        aliases: [String] = [],
+        terms: [String] = []
+    ) -> LanguageCatalogItem {
+        LanguageCatalogItem(
+            id: id,
+            name: name,
+            familyCategory: familyCategory,
+            aliases: aliases,
+            searchTerms: terms
+        )
+    }
+
+    private static func searchableTerms(for item: LanguageCatalogItem) -> [String] {
+        searchTokens(for: [item.name, item.familyCategory] + item.aliases + item.searchTerms)
+    }
+
+    private static func searchTokens(for query: String) -> [String] {
+        searchTokens(for: [query])
+    }
+
+    private static func searchTokens(for values: [String]) -> [String] {
+        values
+            .joined(separator: " ")
+            .folding(
+                options: [.caseInsensitive, .diacriticInsensitive, .widthInsensitive],
+                locale: Locale(identifier: "en_US_POSIX")
+            )
+            .split(whereSeparator: { $0.isWhitespace || $0 == "-" })
+            .map(String.init)
+    }
+}