i18n: add i18n engine (#83)

Adds a translation engine

CMakeLists.txt

@@ -120,6 +120,15 @@ add_test(
   COMMAND hyprutils_beziercurve "beziercurve")
 add_dependencies(tests hyprutils_beziercurve)
 
+add_executable(hyprutils_i18n "tests/i18n.cpp")
+target_link_libraries(hyprutils_i18n PRIVATE hyprutils PkgConfig::deps)
+add_test(
+  NAME "I18n"
+  WORKING_DIRECTORY ${CMAKE_SOURCE_DIR}/tests
+  COMMAND hyprutils_i18n "i18n")
+add_dependencies(tests hyprutils_i18n)
+
+
 # Installation
 install(TARGETS hyprutils)
 install(DIRECTORY "include/hyprutils" DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})

include/hyprutils/i18n/I18nEngine.hpp

@@ -0,0 +1,55 @@
+#pragma once
+
+#include "../memory/UniquePtr.hpp"
+
+#include <cstdint>
+#include <string>
+#include <unordered_map>
+#include <functional>
+
+namespace Hyprutils::I18n {
+    struct SI18nEngineImpl;
+
+    typedef std::unordered_map<std::string, std::string>         translationVarMap;
+    typedef std::function<std::string(const translationVarMap&)> translationFn;
+
+    class CI18nLocale {
+      public:
+        ~CI18nLocale() = default;
+
+        std::string locale();
+        std::string stem();
+        std::string full();
+
+      private:
+        CI18nLocale(std::string fullLocale);
+
+        std::string m_locale, m_rawFullLocale;
+
+        friend class CI18nEngine;
+    };
+
+    class CI18nEngine {
+      public:
+        CI18nEngine();
+        ~CI18nEngine();
+        /*
+            Register a translation entry. The internal translation db is kept as a vector,
+            so make sure your keys are linear, don't use e.g. 2 billion as that will call
+            .resize() on the vec to 2 billion.
+
+            If you pass a Fn, you can do logic, e.g. "1 point" vs "2 points".
+        */
+        void        registerEntry(const std::string& locale, uint64_t key, std::string&& translationUTF8);
+        void        registerEntry(const std::string& locale, uint64_t key, translationFn&& translationFn);
+
+        void        setFallbackLocale(const std::string& locale);
+
+        std::string localizeEntry(const std::string& locale, uint64_t key, const translationVarMap& map);
+
+        CI18nLocale getSystemLocale();
+
+      private:
+        Memory::CUniquePointer<SI18nEngineImpl> m_impl;
+    };
+}

src/i18n/I18nEngine.cpp

@@ -0,0 +1,110 @@
+#include "I18nEngine.hpp"
+
+#include <algorithm>
+#include <format>
+#include <locale>
+#include <hyprutils/string/String.hpp>
+
+using namespace Hyprutils::I18n;
+using namespace Hyprutils;
+using namespace Hyprutils::String;
+
+CI18nEngine::CI18nEngine() : m_impl(Memory::makeUnique<SI18nEngineImpl>()) {
+    ;
+}
+CI18nEngine::~CI18nEngine() = default;
+
+void CI18nEngine::registerEntry(const std::string& locale, uint64_t key, std::string&& translationUTF8) {
+    auto& entryVec = m_impl->entries[locale];
+
+    if (entryVec.size() <= key)
+        entryVec.resize(key + 1);
+
+    entryVec[key].entry  = std::move(translationUTF8);
+    entryVec[key].exists = true;
+}
+
+void CI18nEngine::registerEntry(const std::string& locale, uint64_t key, translationFn&& translationFn) {
+    auto& entryVec = m_impl->entries[locale];
+
+    if (entryVec.size() <= key)
+        entryVec.resize(key + 1);
+
+    entryVec[key].fn     = std::move(translationFn);
+    entryVec[key].exists = true;
+}
+
+void CI18nEngine::setFallbackLocale(const std::string& locale) {
+    m_impl->fallbackLocale = locale;
+}
+
+std::string CI18nEngine::localizeEntry(const std::string& locale, uint64_t key, const translationVarMap& map) {
+    SI18nTranslationEntry* entry = nullptr;
+
+    if (m_impl->entries.contains(locale) && m_impl->entries[locale].size() > key)
+        entry = &m_impl->entries[locale][key];
+
+    if (locale.contains('_')) {
+
+        if (!entry || !entry->exists) {
+            // try to fall back to lang_LANG
+            auto stem      = locale.substr(0, locale.find('_'));
+            auto stemUpper = stem;
+            std::ranges::transform(stemUpper, stemUpper.begin(), ::toupper);
+            auto newLocale = std::format("{}_{}", stem, stemUpper);
+            if (m_impl->entries.contains(newLocale) && m_impl->entries[newLocale].size() > key)
+                entry = &m_impl->entries[newLocale][key];
+        }
+
+        if (!entry || !entry->exists) {
+            // try to fall back to any lang prefixed with our prefix
+
+            auto stem = locale.substr(0, locale.find('_') + 1);
+            for (const auto& [k, v] : m_impl->entries) {
+                if (k.starts_with(stem) || k == stem) {
+                    if (m_impl->entries.contains(k) && m_impl->entries[k].size() > key)
+                        entry = &m_impl->entries[k][key];
+
+                    if (entry && entry->exists)
+                        break;
+                }
+            }
+        }
+    } else {
+        // locale doesn't have a _, e.g. pl
+        // find any locale that has the same stem
+        for (const auto& [k, v] : m_impl->entries) {
+            if (k.starts_with(locale + "_") || k == locale) {
+                if (m_impl->entries.contains(k) && m_impl->entries[k].size() > key)
+                    entry = &m_impl->entries[k][key];
+
+                if (entry && entry->exists)
+                    break;
+            }
+        }
+    }
+
+    if (!entry || !entry->exists) {
+        // fall back to general fallback
+        if (m_impl->entries.contains(m_impl->fallbackLocale) && m_impl->entries[m_impl->fallbackLocale].size() > key)
+            entry = &m_impl->entries[m_impl->fallbackLocale][key];
+    }
+
+    if (!entry || !entry->exists)
+        return "";
+
+    std::string rawStr = entry->entry;
+
+    if (entry->fn)
+        rawStr = entry->fn(map);
+
+    for (const auto& e : map) {
+        replaceInString(rawStr, "{" + e.first + "}", e.second);
+    }
+
+    return rawStr;
+}
+
+CI18nLocale CI18nEngine::getSystemLocale() {
+    return CI18nLocale(std::locale("").name());
+}

src/i18n/I18nEngine.hpp

@@ -0,0 +1,16 @@
+#pragma once
+
+#include <hyprutils/i18n/I18nEngine.hpp>
+
+namespace Hyprutils::I18n {
+    struct SI18nTranslationEntry {
+        bool          exists = false;
+        std::string   entry  = "";
+        translationFn fn     = nullptr;
+    };
+
+    struct SI18nEngineImpl {
+        std::unordered_map<std::string, std::vector<SI18nTranslationEntry>> entries;
+        std::string                                                         fallbackLocale = "en_US";
+    };
+};

src/i18n/I18nLocale.cpp

@@ -0,0 +1,44 @@
+#include <hyprutils/i18n/I18nEngine.hpp>
+
+using namespace Hyprutils::I18n;
+
+static std::string extractLocale(std::string locale) {
+    // localeStr is very arbitrary... from my testing, it can be:
+    // en_US.UTF-8
+    // LC_CTYPE=en_US
+    // POSIX
+    // *
+    //
+    // We only return e.g. en_US or pl_PL, or pl
+
+    if (locale == "POSIX")
+        return "en_US";
+    if (locale == "*")
+        return "en_US";
+
+    if (locale.contains('='))
+        locale = locale.substr(locale.find('=') + 1);
+
+    if (locale.contains('.'))
+        locale = locale.substr(0, locale.find('.'));
+
+    return locale;
+}
+
+CI18nLocale::CI18nLocale(std::string fullLocale) : m_rawFullLocale(std::move(fullLocale)) {
+    m_locale = extractLocale(m_rawFullLocale);
+}
+
+std::string CI18nLocale::locale() {
+    return m_locale;
+}
+
+std::string CI18nLocale::stem() {
+    if (m_locale.contains('_'))
+        return m_locale.substr(0, m_locale.find('_'));
+    return m_locale;
+}
+
+std::string CI18nLocale::full() {
+    return m_rawFullLocale;
+}

tests/i18n.cpp

@@ -0,0 +1,67 @@
+#include <hyprutils/i18n/I18nEngine.hpp>
+#include "shared.hpp"
+#include <print>
+
+using namespace Hyprutils::I18n;
+
+enum eTxtKeys : uint64_t {
+    TXT_KEY_HELLO,
+    TXT_KEY_I_HAVE_APPLES,
+    TXT_KEY_FALLBACK,
+};
+
+int main(int argc, char** argv, char** envp) {
+    int         ret = 0;
+
+    CI18nEngine engine;
+
+    std::println("System locale: {}, stem: {}", engine.getSystemLocale().locale(), engine.getSystemLocale().stem());
+
+    engine.setFallbackLocale("en_US");
+
+    engine.registerEntry("en_US", TXT_KEY_HELLO, "Hello World!");
+    engine.registerEntry("en_US", TXT_KEY_I_HAVE_APPLES, [](const translationVarMap& m) {
+        if (std::stoi(m.at("count")) == 1)
+            return "I have {count} apple.";
+        else
+            return "I have {count} apples.";
+    });
+    engine.registerEntry("en_US", TXT_KEY_FALLBACK, "Fallback string!");
+
+    engine.registerEntry("pl_PL", TXT_KEY_HELLO, "Witaj świecie!");
+    engine.registerEntry("pl_PL", TXT_KEY_I_HAVE_APPLES, [](const translationVarMap& m) {
+        const auto COUNT = std::stoi(m.at("count"));
+        if (COUNT == 1)
+            return "Mam {count} jabłko.";
+        else if (COUNT < 5)
+            return "Mam {count} jabłka.";
+        else
+            return "Mam {count} jabłek.";
+    });
+
+    engine.registerEntry("es_XX", TXT_KEY_FALLBACK, "I don't speak spanish");
+    engine.registerEntry("es_ES", TXT_KEY_FALLBACK, "I don't speak spanish here either");
+
+    EXPECT(engine.localizeEntry("en_US", TXT_KEY_HELLO, {}), "Hello World!");
+    EXPECT(engine.localizeEntry("pl_PL", TXT_KEY_HELLO, {}), "Witaj świecie!");
+    EXPECT(engine.localizeEntry("de_DE", TXT_KEY_HELLO, {}), "Hello World!");
+
+    EXPECT(engine.localizeEntry("en_US", TXT_KEY_I_HAVE_APPLES, {{"count", "1"}}), "I have 1 apple.");
+    EXPECT(engine.localizeEntry("en_US", TXT_KEY_I_HAVE_APPLES, {{"count", "2"}}), "I have 2 apples.");
+
+    EXPECT(engine.localizeEntry("pl_PL", TXT_KEY_I_HAVE_APPLES, {{"count", "1"}}), "Mam 1 jabłko.");
+    EXPECT(engine.localizeEntry("pl_PL", TXT_KEY_I_HAVE_APPLES, {{"count", "2"}}), "Mam 2 jabłka.");
+    EXPECT(engine.localizeEntry("pl_PL", TXT_KEY_I_HAVE_APPLES, {{"count", "5"}}), "Mam 5 jabłek.");
+
+    EXPECT(engine.localizeEntry("pl", TXT_KEY_I_HAVE_APPLES, {{"count", "5"}}), "Mam 5 jabłek.");
+
+    EXPECT(engine.localizeEntry("pl_XX", TXT_KEY_I_HAVE_APPLES, {{"count", "5"}}), "Mam 5 jabłek.");
+    EXPECT(engine.localizeEntry("en_XX", TXT_KEY_I_HAVE_APPLES, {{"count", "2"}}), "I have 2 apples.");
+
+    EXPECT(engine.localizeEntry("es_YY", TXT_KEY_FALLBACK, {}), "I don't speak spanish here either");
+    EXPECT(engine.localizeEntry("es_XX", TXT_KEY_FALLBACK, {}), "I don't speak spanish");
+
+    EXPECT(engine.localizeEntry("pl_PL", TXT_KEY_FALLBACK, {}), "Fallback string!");
+
+    return ret;
+}