cld3 binding

app/src/main/java/eu/faircode/email/TextHelper.java

@@ -41,11 +41,23 @@ import java.util.Locale;
 import java.util.Set;
 
 public class TextHelper {
+    static {
+        System.loadLibrary("fairemail");
+    }
+
+    private static native String jni_language(byte[] octets);
+
     static Locale detectLanguage(Context context, String text) {
         // Why not ML kit? https://developers.google.com/ml-kit/terms
         if (TextUtils.isEmpty(text))
             return null;
 
+        if (BuildConfig.DEBUG) {
+            // https://github.com/google/cld3
+            String lang = jni_language(text.getBytes());
+            return Locale.forLanguageTag(lang);
+        }
+
         if (Build.VERSION.SDK_INT < Build.VERSION_CODES.Q)
             return null;
 

app/src/main/jni/fairemail.cc

@@ -7,6 +7,7 @@
 #include <netinet/tcp.h>
 
 #include "compact_enc_det/compact_enc_det.h"
+#include "cld_3/src/nnet_language_identifier.h"
 
 void log_android(int prio, const char *fmt, ...) {
     if (prio >= ANDROID_LOG_DEBUG) {
@@ -57,6 +58,22 @@ Java_eu_faircode_email_CharsetHelper_jni_1detect(JNIEnv *env, jclass type, jbyte
             (jboolean) is_reliable);
 }
 
+extern "C"
+JNIEXPORT jstring JNICALL
+Java_eu_faircode_email_TextHelper_jni_1language(JNIEnv *env, jclass clazz, jbyteArray _octets) {
+    int len = env->GetArrayLength(_octets);
+    jbyte *octets = env->GetByteArrayElements(_octets, nullptr);
+
+    std::string text(reinterpret_cast<char const *>(octets), len);
+
+    chrome_lang_id::NNetLanguageIdentifier lang_id(0, 1000);
+    const chrome_lang_id::NNetLanguageIdentifier::Result result = lang_id.FindLanguage(text);
+
+    env->ReleaseByteArrayElements(_octets, octets, JNI_ABORT);
+
+    return env->NewStringUTF(result.language.c_str());
+}
+
 extern "C"
 JNIEXPORT jint JNICALL
 Java_eu_faircode_email_EmailService_jni_1socket_1keep_1alive(