Respect codepages.

src/aim1_mod_maker/aim1_mod_maker.h

@@ -85,7 +85,7 @@ struct mod_maker {
             }
         }
         void write() {
-            m.save(fn);
+            m.save(fn, codepage);
             written = true;
         }
         auto &operator[](this auto &&d, const std::string &s) {
@@ -475,7 +475,7 @@ struct mod_maker {
         if (!aim2_available()) {
             throw std::runtime_error{"aim2 is not available, setup it first"};
         }
-        static auto m2 = db2{aim2_game_dir / "data" / "db", 1251}.open().to_map();
+        static auto m2 = db2{aim2_game_dir / "data" / "db"}.open().to_map(1251);
         return m2;
     }
 
@@ -489,7 +489,7 @@ private:
         if (!aim2_available()) {
             throw std::runtime_error{"aim2 is not available, setup it first"};
         }
-        static auto m2 = db2{aim2_game_dir / "data" / "quest", 1251}.open().to_map();
+        static auto m2 = db2{aim2_game_dir / "data" / "quest"}.open().to_map(1251);
         return m2;
     }
     bool aim2_available() const {
@@ -503,16 +503,16 @@ private:
         return backup;
     }
     db_wrapper open_db(auto &&name, int db_codepage) {
-        auto d = db2{get_data_dir() / name, db_codepage};
+        auto d = db2{get_data_dir() / name};
         auto files = d.open().get_files();
         for (auto &&f : files) {
             backup_or_restore_once(f);
             files_to_distribute.insert(f);
         }
         db_wrapper w;
-        w.m = d.open().to_map();
+        w.m = d.open().to_map(db_codepage);
         w.fn = d.fn;
-        w.codepage = d.codepage;
+        w.codepage = db_codepage;
         return w;
     }
     void backup_or_restore_once(const path &fn) {

src/common/common.h

@@ -18,11 +18,25 @@
 
 #pragma once
 
-#include <string>
 #include <iostream>
+#include <map>
+#include <string>
 
-std::string str2utf8(const std::string &codepage_str, int cp = 0);
-std::wstring str2utf16(const std::string &codepage_str, int cp = 0);
+// MultiByteToWideChar: https://msdn.microsoft.com/en-us/library/windows/desktop/dd319072(v=vs.85).aspx
+// code pages: https://msdn.microsoft.com/en-us/library/windows/desktop/dd317756(v=vs.85).aspx
+// https://www.ibm.com/docs/en/rational-soft-arch/9.6.1?topic=overview-locales-code-pages-supported
+static const std::map<std::string, int> code_pages
+{
+    { "en", 0 },
+    { "cz", 1250 },
+    { "ru", 1251 },
+    { "ge", 1252 },
+    { "fr", 1252 },
+    { "et", 1257 },
+};
+
+std::string str2utf8(const std::string &codepage_str, int cp);
+std::wstring str2utf16(const std::string &codepage_str, int cp);
 
 std::string str2str(const std::string &codepage_str, int cp_from, int cp_to);
 

src/common/db.cpp

@@ -147,11 +147,11 @@ void db::open(const path &p)
         v.load_fields(t, b);
 }
 
-polygon4::tools::db::processed_db db::process() const
+polygon4::tools::db::processed_db db::process(int cp) const
 {
-    auto process_string = [](const std::string &s)
+    auto process_string = [&](const std::string &s)
     {
-        return str2utf8(s.c_str());
+        return str2utf8(s.c_str(), cp);
     };
 
     polygon4::tools::db::processed_db pdb;

src/common/db.h

@@ -107,5 +107,5 @@ struct db
 
     void load(const buffer &b);
     void open(const path &basename);
-    polygon4::tools::db::processed_db process() const;
+    polygon4::tools::db::processed_db process(int cp) const;
 };

src/common/db2.h

@@ -25,13 +25,13 @@
 
 #include <variant>
 
-std::string utf8_to_dbstr(const char8_t *s, int codepage = 1251) {
+std::string utf8_to_dbstr(const char8_t *s, int codepage) {
     return str2str((const char *)s, CP_UTF8, codepage);
 }
-std::string utf8_to_dbstr(const char *s, int codepage = 1251) {
+std::string utf8_to_dbstr(const char *s, int codepage) {
     return utf8_to_dbstr((const char8_t *)s, codepage);
 }
-std::string utf8_to_dbstr(const std::string &s, int codepage = 1251) {
+std::string utf8_to_dbstr(const std::string &s, int codepage) {
     return utf8_to_dbstr((const char8_t *)s.c_str(), codepage);
 }
 
@@ -149,7 +149,6 @@ struct db2 {
     };
 
     path fn;
-    int codepage{1251};
 
     template <typename T>
     struct file {
@@ -223,9 +222,9 @@ struct db2 {
                     }
                 }
             }
-            void save(const path &fn, int codepage = 1251) {
-                auto s_to_char20 = [&](char20 &dst, const std::string &in, int codepage = 1251) {
-                    auto s = utf8_to_dbstr(in);
+            void save(const path &fn, int codepage) {
+                auto s_to_char20 = [&](char20 &dst, const std::string &in, int codepage) {
+                    auto s = utf8_to_dbstr(in, codepage);
                     if (s.size() + 1 > sizeof(char20)) {
                         throw std::runtime_error{"too long string"};
                     }
@@ -241,7 +240,7 @@ struct db2 {
                 for (auto &&[tn,td] : m) {
                     tab::table &t = tabv;
                     t.id = table_id;
-                    s_to_char20(t.name, tn, 1251); // always 1251
+                    s_to_char20(t.name, tn, 1251); // always 1251, because latin only letters
 
                     for (auto &&[_,fd] : td) {
                         for (auto &&[fn,fv] : fd) {
@@ -255,7 +254,7 @@ struct db2 {
                         f.table_id = table_id;
                         f.type = ft;
                         ft = (field_type)total_fields;
-                        s_to_char20(f.name, fn, 1251); // always 1251 if we have any field in Russian
+                        s_to_char20(f.name, fn, 1251); // always 1251, because latin only letters
                     }
 
                     ++table_id;
@@ -282,7 +281,7 @@ struct db2 {
                             auto sz = visit(fv,
                                 [&](const int &v) { return datv = v; },
                                 [&](const float &v) { return datv = v; },
-                                [&](const std::string &v) { return datv = utf8_to_dbstr(v); });
+                                [&](const std::string &v) { return datv = utf8_to_dbstr(v, codepage); });
                             auto &v = datv.at<dat::field_value_base>(-(sizeof(dat::field_value_base) + sz));
                             v.field_id = (int)fields[tn].find(fn)->second;
                             v.size = sz;
@@ -300,9 +299,9 @@ struct db2 {
 
         // converts string to utf8
         // filters out values with empty name ""
-        auto to_map() const {
-            auto prepare_string = [](auto &&in) {
-                auto s = str2utf8(in);
+        auto to_map(int cp) const {
+            auto prepare_string = [&](auto &&in) {
+                auto s = str2utf8(in, cp);
                 // we have some erroneous table values (records) with spaces
                 // we can trim only field values, but don't do it at the moment
                 //boost::trim(s);

src/db_add_language/db_add_language.cpp

@@ -41,15 +41,6 @@
 
 #include <math.h>
 
-// MultiByteToWideChar: https://msdn.microsoft.com/en-us/library/windows/desktop/dd319072(v=vs.85).aspx
-// code pages: https://msdn.microsoft.com/en-us/library/windows/desktop/dd317756(v=vs.85).aspx
-const std::map<std::string, int> code_pages
-{
-    { "en", 0 },
-    { "ru", 1251 },
-    { "et", 1257 },
-};
-
 static int get_cp(const std::string &cp)
 {
     auto i = code_pages.find(cp);

src/db_extractor2/db_extractor2.cpp

@@ -32,6 +32,7 @@
 int main(int argc, char *argv[])
 {
     cl::opt<path> db_fn(cl::Positional, cl::desc("<db file or json file to backwards conversion>"), cl::Required);
+    cl::opt<int> codepage(cl::Positional, cl::desc("<codepage>"), cl::Required);
 
     cl::ParseCommandLineOptions(argc, argv);
 
@@ -40,12 +41,12 @@ int main(int argc, char *argv[])
     if (fn.extension() != ".json") {
         db2 db{fn};
         auto f = db.open();
-        auto m = f.to_map();
+        auto m = f.to_map(codepage);
         write_file(path{fn} += ".json", m.to_json().dump(1));
     } else {
         db2::files::db2_internal db;
         db.load_from_json(fn);
-        db.save(fn.parent_path() / fn.stem());
+        db.save(fn.parent_path() / fn.stem(), codepage);
     }
 
     return 0;