[android] Android "collator" implementation

 - Uses java.text.Collator for comparison
 - Uses java.util.Locale for locale loading
 - Uses LanguageTag for BCP 47 parsing
 - Falls back to non-locale-aware nunicode/default comparison for case-sensitive/diacritic-insensitive.
 - Testing these changes depends on running Android render tests
 - "collator" is not yet exposed in the SDK bindings.

4 files changed, 264 insertions, 1 deletions

diff --git a/platform/android/config.cmake b/platform/android/config.cmake
index bb4fa3bfb6..fa26987e5a 100644
--- a/platform/android/config.cmake
+++ b/platform/android/config.cmake
@@ -37,13 +37,17 @@ macro(mbgl_platform_core)
         PRIVATE platform/android/src/timer.cpp
 
         # Misc
-        PRIVATE platform/android/src/text/local_glyph_rasterizer_jni.hpp
+        PRIVATE platform/android/src/text/collator.cpp
+        PRIVATE platform/android/src/text/collator_jni.hpp
         PRIVATE platform/android/src/text/local_glyph_rasterizer.cpp
+        PRIVATE platform/android/src/text/local_glyph_rasterizer_jni.hpp
         PRIVATE platform/android/src/logging_android.cpp
         PRIVATE platform/android/src/thread.cpp
         PRIVATE platform/default/string_stdlib.cpp
         PRIVATE platform/default/bidi.cpp
         PRIVATE platform/default/thread_local.cpp
+        PRIVATE platform/default/unaccent.cpp
+        PRIVATE platform/default/unaccent.hpp
         PRIVATE platform/default/utf.cpp
 
         # Image handling
diff --git a/platform/android/src/jni.cpp b/platform/android/src/jni.cpp
index 2f6ed96ab0..beb2c14eb3 100755
--- a/platform/android/src/jni.cpp
+++ b/platform/android/src/jni.cpp
@@ -46,6 +46,7 @@
 #include "style/light.hpp"
 #include "snapshotter/map_snapshotter.hpp"
 #include "snapshotter/map_snapshot.hpp"
+#include "text/collator_jni.hpp"
 #include "text/local_glyph_rasterizer_jni.hpp"
 #include "java/lang.hpp"
 
@@ -188,6 +189,8 @@ void registerNatives(JavaVM *vm) {
 
     // text
     LocalGlyphRasterizer::registerNative(env);
+    Locale::registerNative(env);
+    Collator::registerNative(env);
 }
 
 } // namespace android
diff --git a/platform/android/src/text/collator.cpp b/platform/android/src/text/collator.cpp
new file mode 100644
index 0000000000..37af0a1c5b
--- /dev/null
+++ b/platform/android/src/text/collator.cpp
@@ -0,0 +1,199 @@
+#include <mbgl/style/expression/collator.hpp>
+#include <mbgl/text/language_tag.hpp>
+#include <mbgl/util/platform.hpp>
+
+#include <unaccent.hpp>
+
+#include <jni/jni.hpp>
+
+#include "../attach_env.hpp"
+#include "collator_jni.hpp"
+
+namespace mbgl {
+namespace android {
+
+void Collator::registerNative(jni::JNIEnv& env) {
+    javaClass = *jni::Class<Collator>::Find(env).NewGlobalRef(env).release();
+}
+
+jni::Class<Collator> Collator::javaClass;
+
+jni::Object<Collator> Collator::getInstance(jni::JNIEnv& env, jni::Object<Locale> locale) {
+    using Signature = jni::Object<Collator>(jni::Object<Locale>);
+    auto method = javaClass.GetStaticMethod<Signature>(env, "getInstance");
+    return javaClass.Call(env, method, locale);
+}
+
+void Collator::setStrength(jni::JNIEnv& env, jni::Object<Collator> collator, jni::jint strength) {
+    using Signature = void(jni::jint);
+    auto static method = javaClass.GetMethod<Signature>(env, "setStrength");
+    collator.Call(env, method, strength);
+}
+
+jni::jint Collator::compare(jni::JNIEnv& env, jni::Object<Collator> collator, jni::String lhs, jni::String rhs) {
+    using Signature = jni::jint(jni::String, jni::String);
+    auto static method = javaClass.GetMethod<Signature>(env, "compare");
+    return collator.Call(env, method, lhs, rhs);
+}
+
+void Locale::registerNative(jni::JNIEnv& env) {
+    javaClass = *jni::Class<Locale>::Find(env).NewGlobalRef(env).release();
+}
+
+    /*
+We would prefer to use for/toLanguageTag, but they're only available in API level 21+
+
+jni::Object<Locale> Locale::forLanguageTag(jni::JNIEnv& env, jni::String languageTag) {
+    using Signature = jni::Object<Locale>(jni::String);
+    auto method = javaClass.GetStaticMethod<Signature>(env, "forLanguageTag");
+    return javaClass.Call(env, method, languageTag);
+}
+
+jni::String Locale::toLanguageTag(jni::JNIEnv& env, jni::Object<Locale> locale) {
+    using Signature = jni::String();
+    auto static method = javaClass.GetMethod<Signature>(env, "toLanguageTag");
+    return locale.Call(env, method);
+}
+     */
+
+
+jni::String Locale::getLanguage(jni::JNIEnv& env, jni::Object<Locale> locale) {
+    using Signature = jni::String();
+    auto static method = javaClass.GetMethod<Signature>(env, "getLanguage");
+    return locale.Call(env, method);
+}
+
+jni::String Locale::getCountry(jni::JNIEnv& env, jni::Object<Locale> locale) {
+    using Signature = jni::String();
+    auto static method = javaClass.GetMethod<Signature>(env, "getCountry");
+    return locale.Call(env, method);
+}
+
+jni::Object<Locale> Locale::getDefault(jni::JNIEnv& env) {
+    using Signature = jni::Object<Locale>();
+    auto method = javaClass.GetStaticMethod<Signature>(env, "getDefault");
+    return javaClass.Call(env, method);
+}
+
+jni::Object<Locale> Locale::New(jni::JNIEnv& env, jni::String language) {
+    static auto constructor = javaClass.GetConstructor<jni::String>(env);
+    return javaClass.New(env, constructor, language);
+}
+
+jni::Object<Locale> Locale::New(jni::JNIEnv& env, jni::String language, jni::String region) {
+    static auto constructor = javaClass.GetConstructor<jni::String, jni::String>(env);
+    return javaClass.New(env, constructor, language, region);
+}
+
+jni::Class<Locale> Locale::javaClass;
+
+} // namespace android
+
+namespace style {
+namespace expression {
+
+class Collator::Impl {
+public:
+    Impl(bool caseSensitive_, bool diacriticSensitive_, optional<std::string> locale_)
+        : caseSensitive(caseSensitive_)
+        , diacriticSensitive(diacriticSensitive_)
+        , env(android::AttachEnv())
+    {
+        LanguageTag languageTag = locale_ ? LanguageTag::fromBCP47(*locale_) : LanguageTag();
+        if (!languageTag.language) {
+            locale = android::Locale::getDefault(*env).NewGlobalRef(*env);
+        } else if (!languageTag.region) {
+            locale = android::Locale::New(*env,
+                                          jni::Make<jni::String>(*env, *(languageTag.language)))
+                    .NewGlobalRef(*env);
+        } else {
+            locale = android::Locale::New(*env,
+                                          jni::Make<jni::String>(*env, *(languageTag.language)),
+                                          jni::Make<jni::String>(*env, *(languageTag.region)))
+                    .NewGlobalRef(*env);
+        }
+        collator = android::Collator::getInstance(*env, *locale).NewGlobalRef(*env);;
+        if (!diacriticSensitive && !caseSensitive) {
+            android::Collator::setStrength(*env, *collator, 0 /*PRIMARY*/);
+        } else if (diacriticSensitive && !caseSensitive) {
+            android::Collator::setStrength(*env, *collator, 1 /*SECONDARY*/);
+        } else if (caseSensitive) {
+            // If we're case-sensitive and diacritic-sensitive, we use a case-sensitive collator
+            // and a fallback implementation of diacritic-insensitivity.
+            android::Collator::setStrength(*env, *collator, 2 /*TERTIARY*/);
+        }
+    }
+
+    bool operator==(const Impl& other) const {
+        return caseSensitive == other.caseSensitive &&
+                diacriticSensitive == other.diacriticSensitive &&
+                resolvedLocale() == other.resolvedLocale();
+    }
+
+    int compare(const std::string& lhs, const std::string& rhs) const {
+        bool useUnaccent = !diacriticSensitive && caseSensitive;
+        // java.text.Collator doesn't support a diacritic-insensitive/case-sensitive collation
+        // order, so we have to compromise here. We use Android's case-sensitive Collator
+        // against strings that have been "unaccented" using non-locale-aware nunicode logic.
+        // Because of the difference in locale-awareness, this means turning on case-sensitivity
+        // can _potentially_ change compare results for strings that don't actually have any case
+        // differences.
+        jni::String jlhs = jni::Make<jni::String>(*env, useUnaccent ?
+                                                        platform::unaccent(lhs) :
+                                                        lhs);
+        jni::String jrhs = jni::Make<jni::String>(*env, useUnaccent ?
+                                                        platform::unaccent(rhs) :
+                                                        rhs);
+
+        jni::jint result = android::Collator::compare(*env, *collator, jlhs, jrhs);;
+
+        jni::DeleteLocalRef(*env, jlhs);
+        jni::DeleteLocalRef(*env, jrhs);
+
+        return result;
+    }
+
+    std::string resolvedLocale() const {
+        jni::String jLanguage = android::Locale::getLanguage(*env, *locale);
+        std::string language = jni::Make<std::string>(*env, jLanguage);
+        jni::DeleteLocalRef(*env, jLanguage);
+        jni::String jRegion = android::Locale::getCountry(*env, *locale);
+        std::string region = jni::Make<std::string>(*env, jRegion);
+        jni::DeleteLocalRef(*env, jRegion);
+
+        optional<std::string> resultLanguage;
+        if (!language.empty()) resultLanguage = language;
+        optional<std::string> resultRegion;
+        if (!region.empty()) resultRegion = region;
+
+        return LanguageTag(resultLanguage, {}, resultRegion).toBCP47();
+    }
+private:
+    bool caseSensitive;
+    bool diacriticSensitive;
+
+    android::UniqueEnv env;
+    jni::UniqueObject<android::Collator> collator;
+    jni::UniqueObject<android::Locale> locale;
+};
+
+
+Collator::Collator(bool caseSensitive, bool diacriticSensitive, optional<std::string> locale_)
+    : impl(std::make_shared<Impl>(caseSensitive, diacriticSensitive, std::move(locale_)))
+{}
+
+bool Collator::operator==(const Collator& other) const {
+    return *impl == *(other.impl);
+}
+
+int Collator::compare(const std::string& lhs, const std::string& rhs) const {
+    return impl->compare(lhs, rhs);
+}
+
+std::string Collator::resolvedLocale() const {
+    return impl->resolvedLocale();
+}
+
+} // namespace expression
+} // namespace style
+} // namespace mbgl
diff --git a/platform/android/src/text/collator_jni.hpp b/platform/android/src/text/collator_jni.hpp
new file mode 100644
index 0000000000..96c381b1c1
--- /dev/null
+++ b/platform/android/src/text/collator_jni.hpp
@@ -0,0 +1,57 @@
+#pragma once
+
+#include <mbgl/util/image.hpp>
+
+#include <jni/jni.hpp>
+
+/*
+    android::Collator and android::Locale are
+    the JNI wrappers of
+    java/text/Collator and java/util/Locale
+
+    mbgl::Collator is the portable interface
+    Both implementations are in collator.cpp
+ */
+
+namespace mbgl {
+namespace android {
+
+class Locale {
+public:
+    static constexpr auto Name() { return "java/util/Locale"; };
+
+    /* Requires API level 21+ in order to support script/variant
+    static jni::Object<Locale> forLanguageTag(jni::JNIEnv&, jni::String);
+    static jni::String toLanguageTag(jni::JNIEnv&, jni::Object<Locale>);
+    */
+    static jni::Object<Locale> getDefault(jni::JNIEnv&);
+    static jni::String getLanguage(jni::JNIEnv&, jni::Object<Locale>);
+    static jni::String getCountry(jni::JNIEnv&, jni::Object<Locale>);
+
+    static jni::Object<Locale> New(jni::JNIEnv&, jni::String);
+    static jni::Object<Locale> New(jni::JNIEnv&, jni::String, jni::String);
+
+    static jni::Class<Locale> javaClass;
+
+    static void registerNative(jni::JNIEnv&);
+
+};
+
+class Collator {
+public:
+    static constexpr auto Name() { return "java/text/Collator"; };
+
+    static jni::Object<Collator> getInstance(jni::JNIEnv&, jni::Object<Locale>);
+
+    static void setStrength(jni::JNIEnv&, jni::Object<Collator>, jni::jint);
+
+    static jni::jint compare(jni::JNIEnv&, jni::Object<Collator>, jni::String, jni::String);
+
+    static jni::Class<Collator> javaClass;
+
+    static void registerNative(jni::JNIEnv&);
+
+};
+
+} // namespace android
+} // namespace mbgl