Introduce library/cpp/charset/lite which excludes functionality with dependency on libiconv

Introduce library/cpp/charset/lite which excludes functionality with dependency on libiconv

Задача:
- выпилить зависимость от libiconv в библиотеке клавиатуры  для андроид

libiconv приходит из двух мест:
- dict/dictutil
- library/cpp/charset

В обоих случаях libiconv используется в вариантах CharToWide/WideToChar для перекодировки.

Первый кейс dictutil - клавиатура для релиза строится с ключом DICTUTIL_WITHOUT_ICONV_AND_UNICODE, который  исключает зависимость. Следующий шаг сделать флаг по умолчанию для мобильных билдов.

Второй кейс library/cpp/charset:
1. Шаг - выделяем библиотеку library/cpp/charset/lite без части функциональности, зависящей от libiconv (этот ПР)
2. Шаг - переводим библиотеки, используемые клавиатурой на lite версию (следующий ПР)

16 files changed, 273 insertions, 149 deletions

diff --git a/library/cpp/charset/CMakeLists.darwin-x86_64.txt b/library/cpp/charset/CMakeLists.darwin-x86_64.txt
index 3dd7934c11..04a8e24de1 100644
--- a/library/cpp/charset/CMakeLists.darwin-x86_64.txt
+++ b/library/cpp/charset/CMakeLists.darwin-x86_64.txt
@@ -6,19 +6,16 @@
 # original buildsystem will not be accepted.
 
 
+add_subdirectory(lite)
 
 add_library(library-cpp-charset)
 target_link_libraries(library-cpp-charset PUBLIC
   contrib-libs-cxxsupp
   yutil
+  cpp-charset-lite
 )
 target_sources(library-cpp-charset PRIVATE
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/cp_data.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/encrec_data.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/codepage.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/cp_encrec.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/doccodes.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/decodeunknownplane.cpp
   ${CMAKE_SOURCE_DIR}/library/cpp/charset/iconv.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/ci_string.cpp
   ${CMAKE_SOURCE_DIR}/library/cpp/charset/wide.cpp
 )
diff --git a/library/cpp/charset/CMakeLists.linux-aarch64.txt b/library/cpp/charset/CMakeLists.linux-aarch64.txt
index ff528662af..9d456b690b 100644
--- a/library/cpp/charset/CMakeLists.linux-aarch64.txt
+++ b/library/cpp/charset/CMakeLists.linux-aarch64.txt
@@ -6,20 +6,17 @@
 # original buildsystem will not be accepted.
 
 
+add_subdirectory(lite)
 
 add_library(library-cpp-charset)
 target_link_libraries(library-cpp-charset PUBLIC
   contrib-libs-linux-headers
   contrib-libs-cxxsupp
   yutil
+  cpp-charset-lite
 )
 target_sources(library-cpp-charset PRIVATE
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/cp_data.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/encrec_data.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/codepage.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/cp_encrec.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/doccodes.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/decodeunknownplane.cpp
   ${CMAKE_SOURCE_DIR}/library/cpp/charset/iconv.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/ci_string.cpp
   ${CMAKE_SOURCE_DIR}/library/cpp/charset/wide.cpp
 )
diff --git a/library/cpp/charset/CMakeLists.linux-x86_64.txt b/library/cpp/charset/CMakeLists.linux-x86_64.txt
index ff528662af..9d456b690b 100644
--- a/library/cpp/charset/CMakeLists.linux-x86_64.txt
+++ b/library/cpp/charset/CMakeLists.linux-x86_64.txt
@@ -6,20 +6,17 @@
 # original buildsystem will not be accepted.
 
 
+add_subdirectory(lite)
 
 add_library(library-cpp-charset)
 target_link_libraries(library-cpp-charset PUBLIC
   contrib-libs-linux-headers
   contrib-libs-cxxsupp
   yutil
+  cpp-charset-lite
 )
 target_sources(library-cpp-charset PRIVATE
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/cp_data.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/encrec_data.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/codepage.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/cp_encrec.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/doccodes.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/decodeunknownplane.cpp
   ${CMAKE_SOURCE_DIR}/library/cpp/charset/iconv.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/ci_string.cpp
   ${CMAKE_SOURCE_DIR}/library/cpp/charset/wide.cpp
 )
diff --git a/library/cpp/charset/CMakeLists.windows-x86_64.txt b/library/cpp/charset/CMakeLists.windows-x86_64.txt
index 2be179087c..3756311ac1 100644
--- a/library/cpp/charset/CMakeLists.windows-x86_64.txt
+++ b/library/cpp/charset/CMakeLists.windows-x86_64.txt
@@ -7,20 +7,17 @@
 
 
 find_package(Iconv REQUIRED)
+add_subdirectory(lite)
 
 add_library(library-cpp-charset)
 target_link_libraries(library-cpp-charset PUBLIC
   contrib-libs-cxxsupp
   yutil
+  cpp-charset-lite
   Iconv::Iconv
 )
 target_sources(library-cpp-charset PRIVATE
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/cp_data.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/encrec_data.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/codepage.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/cp_encrec.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/doccodes.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/decodeunknownplane.cpp
   ${CMAKE_SOURCE_DIR}/library/cpp/charset/iconv.cpp
-  ${CMAKE_SOURCE_DIR}/library/cpp/charset/ci_string.cpp
   ${CMAKE_SOURCE_DIR}/library/cpp/charset/wide.cpp
 )
diff --git a/library/cpp/charset/README.md b/library/cpp/charset/README.md
index 9a257e3b4d..6878cbf4f5 100644
--- a/library/cpp/charset/README.md
+++ b/library/cpp/charset/README.md
@@ -8,3 +8,5 @@
 3. [Класс `TCiString`](https://a.yandex-team.ru/arc/trunk/arcadia/library/cpp/charset/ci_string.h) - аналог `TString`, но использующий case-insensitive-компаратор и хеш и поддерживающий разные кодировки.
 
 В комплекте есть ещё много функций для работы со старой однобайтной Yandex-кодировкой. Не рекомендуется к использованию. Для преобразования из UTF-8 в `TUtf16String` и для работы с Unicode используйте функции из [arcadia/util/charset](https://a.yandex-team.ru/arc/trunk/arcadia/util/charset).
+
+Библиотека library/cpp/charset/lite - содержит часть функциональности, не зависящей от libiconv (e.g. пункты 2 и 3 из основной библиотеки)
diff --git a/library/cpp/charset/codepage.cpp b/library/cpp/charset/codepage.cpp
index 0431bef31b..d75dbe6b6a 100644
--- a/library/cpp/charset/codepage.cpp
+++ b/library/cpp/charset/codepage.cpp
@@ -1,6 +1,4 @@
 #include "ci_string.h"
-#include "wide.h"
-#include "recyr.hh"
 #include "codepage.h"
 
 #include <util/string/cast.h>
@@ -203,121 +201,6 @@ ECharset CharsetByNameOrDie(TStringBuf name) {
     return result;
 }
 
-template <typename TxChar>
-static inline RECODE_RESULT utf8_read_rune_from_unknown_plane(TxChar& rune, size_t& rune_len, const TxChar* s, const TxChar* end) {
-    if ((*s & 0xFF00) != 0xF000) {
-        rune_len = 1;
-        rune = *s;
-        return RECODE_OK;
-    }
-
-    rune_len = 0;
-
-    size_t _len = UTF8RuneLen((unsigned char)(*s));
-    if (s + _len > end)
-        return RECODE_EOINPUT; //[EOINPUT]
-    if (_len == 0)
-        return RECODE_BROKENSYMBOL; //[BROKENSYMBOL] in first byte
-
-    wchar32 _rune = (ui8)(*s++); //[00000000 0XXXXXXX]
-    if (_len > 1) {
-        _rune &= UTF8LeadByteMask(_len);
-        wchar32 ch = *s++;
-        if ((ch & 0xFFC0) != 0xF080)
-            return RECODE_BROKENSYMBOL; //[BROKENSYMBOL] in second byte
-        _rune <<= 6;
-        _rune |= ch & 0x3F; //[00000XXX XXYYYYYY]
-        if (_len > 2) {
-            ch = *s++;
-            if ((ch & 0xFFC0) != 0xF080)
-                return RECODE_BROKENSYMBOL; //[BROKENSYMBOL] in third byte
-            _rune <<= 6;
-            _rune |= ch & 0x3F; //[XXXXYYYY YYZZZZZZ]
-            if (_len > 3) {
-                ch = *s;
-                if ((ch & 0xFFC0) != 0xF080)
-                    return RECODE_BROKENSYMBOL; //[BROKENSYMBOL] in fourth byte
-                _rune <<= 6;
-                _rune |= ch & 0x3F; //[XXXYY YYYYZZZZ ZZQQQQQQ]
-            }
-        }
-    }
-    rune_len = _len;
-    if (_rune > Max<TxChar>())
-        rune = ' '; // maybe put sequence
-    else
-        rune = TxChar(_rune);
-    return RECODE_OK;
-}
-
-template <typename TxChar>
-void DoDecodeUnknownPlane(TxChar* str, TxChar*& ee, const ECharset enc) {
-    TxChar* e = ee;
-    if (SingleByteCodepage(enc)) {
-        const CodePage* cp = CodePageByCharset(enc);
-        for (TxChar* s = str; s < e; s++) {
-            if (Hi8(Lo16(*s)) == 0xF0)
-                *s = (TxChar)cp->unicode[Lo8(Lo16(*s))]; // NOT mb compliant
-        }
-    } else if (enc == CODES_UTF8) {
-        TxChar* s;
-        TxChar* d;
-
-        for (s = d = str; s < e;) {
-            size_t l = 0;
-
-            if (utf8_read_rune_from_unknown_plane(*d, l, s, e) == RECODE_OK) {
-                d++, s += l;
-            } else {
-                *d++ = BROKEN_RUNE;
-                ++s;
-            }
-        }
-        e = d;
-    } else if (enc == CODES_UNKNOWN) {
-        for (TxChar* s = str; s < e; s++) {
-            if (Hi8(Lo16(*s)) == 0xF0)
-                *s = Lo8(Lo16(*s));
-        }
-    } else {
-        Y_ASSERT(!SingleByteCodepage(enc));
-
-        TxChar* s = str;
-        TxChar* d = str;
-
-        TVector<char> buf;
-
-        size_t read = 0;
-        size_t written = 0;
-        for (; s < e; ++s) {
-            if (Hi8(Lo16(*s)) == 0xF0) {
-                buf.push_back(Lo8(Lo16(*s)));
-            } else {
-                if (!buf.empty()) {
-                    if (RecodeToUnicode(enc, buf.data(), d, buf.size(), e - d, read, written) == RECODE_OK) {
-                        Y_ASSERT(read == buf.size());
-                        d += written;
-                    } else { // just copying broken symbols
-                        Y_ASSERT(buf.size() <= static_cast<size_t>(e - d));
-                        Copy(buf.data(), buf.size(), d);
-                        d += buf.size();
-                    }
-                    buf.clear();
-                }
-                *d++ = *s;
-            }
-        }
-    }
-    ee = e;
-}
-
-void DecodeUnknownPlane(wchar16* str, wchar16*& ee, const ECharset enc) {
-    DoDecodeUnknownPlane(str, ee, enc);
-}
-void DecodeUnknownPlane(wchar32* str, wchar32*& ee, const ECharset enc) {
-    DoDecodeUnknownPlane(str, ee, enc);
-}
-
 namespace {
     class THashSetType: public THashSet<TString> {
     public:
diff --git a/library/cpp/charset/decodeunknownplane.cpp b/library/cpp/charset/decodeunknownplane.cpp
new file mode 100644
index 0000000000..201a25cd54
--- /dev/null
+++ b/library/cpp/charset/decodeunknownplane.cpp
@@ -0,0 +1,121 @@
+#include "ci_string.h"
+#include "codepage.h"
+#include "recyr.hh"
+
+#include <util/system/hi_lo.h>
+#include <util/generic/vector.h>
+
+template <typename TxChar>
+static inline RECODE_RESULT utf8_read_rune_from_unknown_plane(TxChar& rune, size_t& rune_len, const TxChar* s, const TxChar* end) {
+    if ((*s & 0xFF00) != 0xF000) {
+        rune_len = 1;
+        rune = *s;
+        return RECODE_OK;
+    }
+
+    rune_len = 0;
+
+    size_t _len = UTF8RuneLen((unsigned char)(*s));
+    if (s + _len > end)
+        return RECODE_EOINPUT; //[EOINPUT]
+    if (_len == 0)
+        return RECODE_BROKENSYMBOL; //[BROKENSYMBOL] in first byte
+
+    wchar32 _rune = (ui8)(*s++); //[00000000 0XXXXXXX]
+    if (_len > 1) {
+        _rune &= UTF8LeadByteMask(_len);
+        wchar32 ch = *s++;
+        if ((ch & 0xFFC0) != 0xF080)
+            return RECODE_BROKENSYMBOL; //[BROKENSYMBOL] in second byte
+        _rune <<= 6;
+        _rune |= ch & 0x3F; //[00000XXX XXYYYYYY]
+        if (_len > 2) {
+            ch = *s++;
+            if ((ch & 0xFFC0) != 0xF080)
+                return RECODE_BROKENSYMBOL; //[BROKENSYMBOL] in third byte
+            _rune <<= 6;
+            _rune |= ch & 0x3F; //[XXXXYYYY YYZZZZZZ]
+            if (_len > 3) {
+                ch = *s;
+                if ((ch & 0xFFC0) != 0xF080)
+                    return RECODE_BROKENSYMBOL; //[BROKENSYMBOL] in fourth byte
+                _rune <<= 6;
+                _rune |= ch & 0x3F; //[XXXYY YYYYZZZZ ZZQQQQQQ]
+            }
+        }
+    }
+    rune_len = _len;
+    if (_rune > Max<TxChar>())
+        rune = ' '; // maybe put sequence
+    else
+        rune = TxChar(_rune);
+    return RECODE_OK;
+}
+
+template <typename TxChar>
+void DoDecodeUnknownPlane(TxChar* str, TxChar*& ee, const ECharset enc) {
+    TxChar* e = ee;
+    if (SingleByteCodepage(enc)) {
+        const CodePage* cp = CodePageByCharset(enc);
+        for (TxChar* s = str; s < e; s++) {
+            if (Hi8(Lo16(*s)) == 0xF0)
+                *s = (TxChar)cp->unicode[Lo8(Lo16(*s))]; // NOT mb compliant
+        }
+    } else if (enc == CODES_UTF8) {
+        TxChar* s;
+        TxChar* d;
+
+        for (s = d = str; s < e;) {
+            size_t l = 0;
+
+            if (utf8_read_rune_from_unknown_plane(*d, l, s, e) == RECODE_OK) {
+                d++, s += l;
+            } else {
+                *d++ = BROKEN_RUNE;
+                ++s;
+            }
+        }
+        e = d;
+    } else if (enc == CODES_UNKNOWN) {
+        for (TxChar* s = str; s < e; s++) {
+            if (Hi8(Lo16(*s)) == 0xF0)
+                *s = Lo8(Lo16(*s));
+        }
+    } else {
+        Y_ASSERT(!SingleByteCodepage(enc));
+
+        TxChar* s = str;
+        TxChar* d = str;
+
+        TVector<char> buf;
+
+        size_t read = 0;
+        size_t written = 0;
+        for (; s < e; ++s) {
+            if (Hi8(Lo16(*s)) == 0xF0) {
+                buf.push_back(Lo8(Lo16(*s)));
+            } else {
+                if (!buf.empty()) {
+                    if (RecodeToUnicode(enc, buf.data(), d, buf.size(), e - d, read, written) == RECODE_OK) {
+                        Y_ASSERT(read == buf.size());
+                        d += written;
+                    } else { // just copying broken symbols
+                        Y_ASSERT(buf.size() <= static_cast<size_t>(e - d));
+                        Copy(buf.data(), buf.size(), d);
+                        d += buf.size();
+                    }
+                    buf.clear();
+                }
+                *d++ = *s;
+            }
+        }
+    }
+    ee = e;
+}
+
+void DecodeUnknownPlane(wchar16* str, wchar16*& ee, const ECharset enc) {
+    DoDecodeUnknownPlane(str, ee, enc);
+}
+void DecodeUnknownPlane(wchar32* str, wchar32*& ee, const ECharset enc) {
+    DoDecodeUnknownPlane(str, ee, enc);
+}
diff --git a/library/cpp/charset/lite/CMakeLists.darwin-x86_64.txt b/library/cpp/charset/lite/CMakeLists.darwin-x86_64.txt
new file mode 100644
index 0000000000..0a3337929c
--- /dev/null
+++ b/library/cpp/charset/lite/CMakeLists.darwin-x86_64.txt
@@ -0,0 +1,22 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_library(cpp-charset-lite)
+target_link_libraries(cpp-charset-lite PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+)
+target_sources(cpp-charset-lite PRIVATE
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/cp_data.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/encrec_data.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/codepage.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/cp_encrec.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/doccodes.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/ci_string.cpp
+)
diff --git a/library/cpp/charset/lite/CMakeLists.linux-aarch64.txt b/library/cpp/charset/lite/CMakeLists.linux-aarch64.txt
new file mode 100644
index 0000000000..ff0a9b367a
--- /dev/null
+++ b/library/cpp/charset/lite/CMakeLists.linux-aarch64.txt
@@ -0,0 +1,23 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_library(cpp-charset-lite)
+target_link_libraries(cpp-charset-lite PUBLIC
+  contrib-libs-linux-headers
+  contrib-libs-cxxsupp
+  yutil
+)
+target_sources(cpp-charset-lite PRIVATE
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/cp_data.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/encrec_data.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/codepage.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/cp_encrec.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/doccodes.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/ci_string.cpp
+)
diff --git a/library/cpp/charset/lite/CMakeLists.linux-x86_64.txt b/library/cpp/charset/lite/CMakeLists.linux-x86_64.txt
new file mode 100644
index 0000000000..ff0a9b367a
--- /dev/null
+++ b/library/cpp/charset/lite/CMakeLists.linux-x86_64.txt
@@ -0,0 +1,23 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_library(cpp-charset-lite)
+target_link_libraries(cpp-charset-lite PUBLIC
+  contrib-libs-linux-headers
+  contrib-libs-cxxsupp
+  yutil
+)
+target_sources(cpp-charset-lite PRIVATE
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/cp_data.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/encrec_data.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/codepage.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/cp_encrec.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/doccodes.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/ci_string.cpp
+)
diff --git a/library/cpp/charset/lite/CMakeLists.txt b/library/cpp/charset/lite/CMakeLists.txt
new file mode 100644
index 0000000000..f8b31df0c1
--- /dev/null
+++ b/library/cpp/charset/lite/CMakeLists.txt
@@ -0,0 +1,17 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+if (CMAKE_SYSTEM_NAME STREQUAL "Linux" AND CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64" AND NOT HAVE_CUDA)
+  include(CMakeLists.linux-aarch64.txt)
+elseif (CMAKE_SYSTEM_NAME STREQUAL "Darwin" AND CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64")
+  include(CMakeLists.darwin-x86_64.txt)
+elseif (WIN32 AND CMAKE_SYSTEM_PROCESSOR STREQUAL "AMD64" AND NOT HAVE_CUDA)
+  include(CMakeLists.windows-x86_64.txt)
+elseif (CMAKE_SYSTEM_NAME STREQUAL "Linux" AND CMAKE_SYSTEM_PROCESSOR STREQUAL "x86_64" AND NOT HAVE_CUDA)
+  include(CMakeLists.linux-x86_64.txt)
+endif()
diff --git a/library/cpp/charset/lite/CMakeLists.windows-x86_64.txt b/library/cpp/charset/lite/CMakeLists.windows-x86_64.txt
new file mode 100644
index 0000000000..0a3337929c
--- /dev/null
+++ b/library/cpp/charset/lite/CMakeLists.windows-x86_64.txt
@@ -0,0 +1,22 @@
+
+# This file was generated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_library(cpp-charset-lite)
+target_link_libraries(cpp-charset-lite PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+)
+target_sources(cpp-charset-lite PRIVATE
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/cp_data.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/generated/encrec_data.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/codepage.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/cp_encrec.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/doccodes.cpp
+  ${CMAKE_SOURCE_DIR}/library/cpp/charset/ci_string.cpp
+)
diff --git a/library/cpp/charset/lite/ut/ya.make b/library/cpp/charset/lite/ut/ya.make
new file mode 100644
index 0000000000..53c85fae98
--- /dev/null
+++ b/library/cpp/charset/lite/ut/ya.make
@@ -0,0 +1,10 @@
+UNITTEST_FOR(library/cpp/charset/lite)
+
+SRCDIR(library/cpp/charset)
+
+SRCS(
+    ci_string_ut.cpp
+    codepage_ut.cpp
+)
+
+END()
diff --git a/library/cpp/charset/lite/ya.make b/library/cpp/charset/lite/ya.make
new file mode 100644
index 0000000000..12c6c9975c
--- /dev/null
+++ b/library/cpp/charset/lite/ya.make
@@ -0,0 +1,18 @@
+LIBRARY()
+
+SRCDIR(library/cpp/charset)
+
+SRCS(
+    generated/cp_data.cpp
+    generated/encrec_data.cpp
+    codepage.cpp
+    cp_encrec.cpp
+    doccodes.cpp
+    ci_string.cpp
+)
+
+END()
+
+RECURSE_FOR_TESTS(
+    ut
+)
diff --git a/library/cpp/charset/ut/ya.make b/library/cpp/charset/ut/ya.make
index 42f7ca54e2..38f029cafc 100644
--- a/library/cpp/charset/ut/ya.make
+++ b/library/cpp/charset/ut/ya.make
@@ -1,8 +1,6 @@
 UNITTEST_FOR(library/cpp/charset)
 
 SRCS(
-    ci_string_ut.cpp
-    codepage_ut.cpp
     iconv_ut.cpp
     wide_ut.cpp
 )
diff --git a/library/cpp/charset/ya.make b/library/cpp/charset/ya.make
index f9fbbf9d09..6a52da56bb 100644
--- a/library/cpp/charset/ya.make
+++ b/library/cpp/charset/ya.make
@@ -5,19 +5,15 @@ IF (NOT OPENSOURCE)
 ENDIF()
 
 SRCS(
-    generated/cp_data.cpp
-    generated/encrec_data.cpp
-    codepage.cpp
-    cp_encrec.cpp
-    doccodes.cpp
+    decodeunknownplane.cpp
     iconv.cpp
     recyr.hh
     recyr_int.hh
-    ci_string.cpp
     wide.cpp
 )
 
 PEERDIR(
+    library/cpp/charset/lite
     contrib/libs/libiconv
 )
 
@@ -25,4 +21,5 @@ END()
 
 RECURSE_FOR_TESTS(
     ut
+    lite/ut
 )