Implement language detection by POSIX locale

ui_i18n.c

@@ -1,4 +1,5 @@
 #include <stdint.h>
+#include <string.h>
 
 #include "sized_string.h"
 #include "ui_i18n_decls.h"
@@ -12,12 +13,23 @@
 #ifdef msgstr
   #error "msgstr is already defined"
 #endif
+#ifdef LANG_POSIX_LOCALE
+  #error "LANG_POSIX_LOCALE is already defined"
+#endif
+#ifdef LANG_PRIORITY
+  #error "LANG_PRIORITY is already defined"
+#endif
 
 #define STR(x) { .str = (uint8_t*)x, .length = sizeof(x) - 1 }
+
+/***** Parsing localized strings *****/
+
 #define msgid(x) curr_id = (STR_##x);
 #define msgstr(x) \
     localized_strings[_LANG_ID][curr_id].str = (uint8_t*)(x); \
     localized_strings[_LANG_ID][curr_id].length = sizeof(x) - 1;
+#define LANG_POSIX_LOCALE(x)
+#define LANG_PRIORITY(x)
 
 static STRING canary = STR("BUG. PLEASE REPORT.");
 
@@ -36,9 +48,10 @@ static void init_strings(STRING (*localized_strings)[STRS_MAX+1]) {
 
 }
 
+#undef LANG_PRIORITY
+#undef LANG_POSIX_LOCALE
 #undef msgstr
 #undef msgid
-#undef STR
 
 STRING* ui_gettext(UI_LANG_ID lang, UI_STRING_ID string_id) {
     static STRING localized_strings[LANGS_MAX+1][STRS_MAX+1];
@@ -55,3 +68,72 @@ STRING* ui_gettext(UI_LANG_ID lang, UI_STRING_ID string_id) {
 
     return &localized_strings[lang][string_id];
 }
+
+/***** Parsing detection by POSIX locale *****/
+
+#define msgid(x)
+#define msgstr(x)
+#define LANG_POSIX_LOCALE(x) posix_locales[_LANG_ID] = (x);
+#define LANG_PRIORITY(x) priorities[_LANG_ID] = (x);
+
+static void init_posix_locales(const char* posix_locales[], int8_t priorities[]) {
+
+#include "ui_i18n.h"
+
+}
+
+#undef LANG_PRIORITY
+#undef LANG_POSIX_LOCALE
+#undef msgstr
+#undef msgid
+
+UI_LANG_ID ui_guess_lang_by_posix_locale(const char* locale, UI_LANG_ID deflt) {
+    static const char* posix_locales[LANGS_MAX+1];
+    static int8_t priorities[LANGS_MAX+1];
+    static int ready = 0;
+
+    if(!ready) {
+        init_posix_locales(posix_locales, priorities);
+        ready = 1;
+    }
+
+    UI_LANG_ID i, found_lang;
+    int8_t found_prio = INT8_MAX;
+
+    // Try detecting by full prefix match first.
+    for(i = 0; i <= LANGS_MAX; i++) {
+        const char* l = posix_locales[i];
+        if(!l) continue;
+
+        if(strstr(locale, l)) {
+            if(found_prio > priorities[i]) {
+                found_lang = i;
+                found_prio = priorities[i];
+            }
+        }
+    }
+
+    if(found_prio < INT8_MAX) {
+        return found_lang;
+    }
+
+    // It appears we haven't found exact language_territory 
+    // match (e.g. zh_TW) for given locale. ,
+    // Try stripping territory off and search only by language part.
+    for(i = 0; i <= LANGS_MAX; i++) {
+        const char* l = posix_locales[i];
+        if(!l) continue;
+
+        char* sep = strchr(l, '_');
+        if(!sep) continue;
+
+        if(!strncmp(locale, l, sep-l)) {
+            if(found_prio > priorities[i]) {
+                found_lang = i;
+                found_prio = priorities[i];
+            }
+        }
+    }
+
+    return found_prio < INT8_MAX ? found_lang : deflt;
+}

ui_i18n.h

@@ -1,113 +1,137 @@
 
 //"CZECH" "Čeština"
 #define _LANG_ID LANG_CS
+LANG_POSIX_LOCALE("cs_CZ")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/cs.h"
 #undef _LANG_ID
 
 //"BULGARIAN" "Български"
 #define _LANG_ID LANG_BG
+LANG_POSIX_LOCALE("bg_BG")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/bg.h"
 #undef _LANG_ID
 
 //"GERMAN" "Deutsch"
 #define _LANG_ID LANG_DE
+LANG_POSIX_LOCALE("de_DE")
+LANG_PRIORITY(-1) // Ensure this lang gets chosen for unknown de locales.
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/de.h"
 #undef _LANG_ID
 
 //"ENGLISH" "English"
 #define _LANG_ID LANG_EN
+LANG_POSIX_LOCALE("en_US")
+LANG_PRIORITY(-1) // Ensure this lang gets chosen for unknown en locales.
 #include "langs/en.h"
 #undef _LANG_ID
 
 //"SPANISH" "Spanish"
 #define _LANG_ID LANG_ES
+LANG_POSIX_LOCALE("es_ES")
+LANG_PRIORITY(-1) // Ensure this lang gets chosen for unknown es locales.
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/es.h"
 #undef _LANG_ID
 
 //"FRENCH" "Français"
 #define _LANG_ID LANG_FR
+LANG_POSIX_LOCALE("fr_FR")
+LANG_PRIORITY(-1) // Ensure this lang gets chosen for unknown fr locales.
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/fr.h"
 #undef _LANG_ID
 
 //"HINDI" "Hindi"
 #define _LANG_ID LANG_HI
+LANG_POSIX_LOCALE("hi_IN")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/hi.h"
 #undef _LANG_ID
 
 //"JAPANESE" "日本語"
 #define _LANG_ID LANG_JA
+LANG_POSIX_LOCALE("ja_JP")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/ja.h"
 #undef _LANG_ID
 
 //"ITALIAN" "Italiano"
 #define _LANG_ID LANG_IT
+LANG_POSIX_LOCALE("it_IT")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/it.h"
 #undef _LANG_ID
 
 //"LATVIAN" "Latviešu"
 #define _LANG_ID LANG_LV
+LANG_POSIX_LOCALE("lv_LV")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/lv.h"
 #undef _LANG_ID
 
 //"DUTCH" "Nederlands"
 #define _LANG_ID LANG_NL
+LANG_POSIX_LOCALE("nl_NL")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/nl.h"
 #undef _LANG_ID
 
 //"NORWEGIAN" "Norsk"
 #define _LANG_ID LANG_NO
+LANG_POSIX_LOCALE("no_NO")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/no.h"
 #undef _LANG_ID
 
 //"POLISH" "Polski"
 #define _LANG_ID LANG_PL
+LANG_POSIX_LOCALE("pl_PL")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/pl.h"
 #undef _LANG_ID
 
 //"ROMANIAN" "Română"
 #define _LANG_ID LANG_RO
+LANG_POSIX_LOCALE("ro_RO")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/ro.h"
 #undef _LANG_ID
 
 //"RUSSIAN" "Русский"
 #define _LANG_ID LANG_RU
+LANG_POSIX_LOCALE("ru_RU")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/ru.h"
 #undef _LANG_ID
 
 //"TURKISH" "Türk"
 #define _LANG_ID LANG_TR
+LANG_POSIX_LOCALE("tr_TR")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/tr.h"
 #undef _LANG_ID
 
 //"UKRAINIAN" "Українська"
 #define _LANG_ID LANG_UA
+LANG_POSIX_LOCALE("uk_UA")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/ua.h"
 #undef _LANG_ID
 
 //"SIMPLIFIED CHINESE" "简体中文"
 #define _LANG_ID LANG_CN
+LANG_POSIX_LOCALE("zh_CN")
+LANG_PRIORITY(-1) // Ensure this lang gets chosen for unknown zh locales.
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/cn.h"
 #undef _LANG_ID
 
 //"TRADITIONAL CHINESE" "繁體中文"
 #define _LANG_ID LANG_TW
+LANG_POSIX_LOCALE("zh_TW")
 #include "langs/en.h" //fallback to English for untranslated things
 #include "langs/tw.h"
 #undef _LANG_ID

ui_i18n_decls.h

@@ -126,3 +126,5 @@ typedef uint8_t UI_LANG_ID;
 typedef uint16_t UI_STRING_ID;
 
 STRING* ui_gettext(UI_LANG_ID lang, UI_STRING_ID string_id);
+
+UI_LANG_ID ui_guess_lang_by_posix_locale(const char* locale, UI_LANG_ID deflt);