Convert the unicode lua module over to using the ffi

automation/include/aegisub/unicode.moon

@@ -29,6 +29,26 @@
 -- http://www.ietf.org/rfc/rfc2279.txt
 
 impl = require 'aegisub.__unicode_impl'
+ffi = require 'ffi'
+ffi.cdef[[
+  void free(void *ptr);
+]]
+
+transfer_string = (cdata) ->
+  return nil if cdata == nil
+  str = ffi.string cdata
+  ffi.C.free cdata
+  str
+
+conv_func = (f) ->
+  err = ffi.new 'char *[1]'
+  (str) ->
+    err[0] = nil
+    result = f str, err
+    errmsg = transfer_string err[0]
+    if errmsg
+      error errmsg, 2
+    transfer_string result
 
 local unicode
 unicode =
@@ -86,8 +106,8 @@ unicode =
       res = res*64 + s\byte(i) - 128
     res
 
-  to_upper_case: impl.to_upper_case
-  to_lower_case: impl.to_lower_case
-  to_fold_case: impl.to_fold_case
+  to_upper_case: conv_func impl.to_upper_case
+  to_lower_case: conv_func impl.to_lower_case
+  to_fold_case: conv_func impl.to_fold_case
 
 return unicode

automation/tests/aegisub.cpp

@@ -21,6 +21,7 @@
 #include <libaegisub/dispatch.h>
 #include <libaegisub/log.h>
 
+#include <boost/locale/generator.hpp>
 #include <cstdio>
 #include <cstdlib>
 
@@ -47,6 +48,7 @@ int main(int argc, char **argv) {
 		return 1;
 	}
 
+	std::locale::global(boost::locale::generator().generate(""));
 	agi::dispatch::Init([](agi::dispatch::Thunk f) { });
 	agi::log::log = new agi::log::LogSink;
 

automation/tests/modules/unicode.moon

@@ -45,3 +45,45 @@ describe 'codepoint', ->
     assert.is.equal 0x1F113, unicode.codepoint '🄓'
   it 'should give ignore codepoints after the first', ->
     assert.is.equal 97, unicode.codepoint 'abc'
+
+describe 'to_upper_case', ->
+  it 'should support plain ASCII', ->
+    assert.is.equal 'ABC', unicode.to_upper_case 'abc'
+  it 'should support accents', ->
+    assert.is.equal 'ÀÈÌ', unicode.to_upper_case 'àèì'
+  it 'should support fullwidth letters', ->
+    assert.is.equal 'ＡＢＣ', unicode.to_upper_case 'ａｂｃ'
+  it 'should support greek', ->
+    assert.is.equal 'ΑΒΓ', unicode.to_upper_case 'αβγ'
+  it 'should support sharp-s', ->
+    assert.is.equal 'SS', unicode.to_upper_case 'ß'
+  it 'should support ligatures', ->
+    assert.is.equal 'FFI', unicode.to_upper_case 'ffi'
+
+describe 'to_lower_case', ->
+  it 'should support plain ASCII', ->
+    assert.is.equal 'abc', unicode.to_lower_case 'ABC'
+  it 'should support accents', ->
+    assert.is.equal 'àèì', unicode.to_lower_case 'ÀÈÌ'
+  it 'should support fullwidth letters', ->
+    assert.is.equal 'ａｂｃ', unicode.to_lower_case 'ＡＢＣ'
+  it 'should support greek', ->
+    assert.is.equal 'αβγ', unicode.to_lower_case 'ΑΒΓ'
+  it 'should support sharp-s', ->
+    assert.is.equal 'ß', unicode.to_lower_case 'ẞ'
+  -- note: Unicode doesn't have any uppercase precomposed ligatures
+
+describe 'to_fold_case', ->
+  it 'should support plain ASCII', ->
+    assert.is.equal 'abc', unicode.to_fold_case 'ABC'
+  it 'should support accents', ->
+    assert.is.equal 'àèì', unicode.to_fold_case 'ÀÈÌ'
+  it 'should support fullwidth letters', ->
+    assert.is.equal 'ａｂｃ', unicode.to_fold_case 'ＡＢＣ'
+  it 'should support greek', ->
+    assert.is.equal 'αβγ', unicode.to_fold_case 'ΑΒΓ'
+  it 'should support sharp-s', ->
+    assert.is.equal 'ss', unicode.to_fold_case 'ẞ'
+  it 'should support ligatures', ->
+    assert.is.equal 'ffi', unicode.to_fold_case 'ffi'
+

libaegisub/lua/modules/unicode.cpp

@@ -14,33 +14,45 @@
 //
 // Aegisub Project http://www.aegisub.org/
 
-#include "libaegisub/lua/utils.h"
+#include <libaegisub/type_name.h>
 
 #include <boost/locale/conversion.hpp>
+#include <lua.hpp>
 
 namespace {
-using namespace agi::lua;
-
-int unicode_upper(lua_State *L) {
-	push_value(L, boost::locale::to_upper(check_string(L, 1)));
-	return 1;
+template<typename T>
+void push_ffi_function(lua_State *L, const char *name, T *func) {
+	lua_pushvalue(L, -2); // push cast function
+	lua_pushstring(L, agi::type_name<T*>::name().c_str());
+	// This cast isn't legal, but LuaJIT internally requires that it work
+	lua_pushlightuserdata(L, (void *)func);
+	lua_call(L, 2, 1);
+	lua_setfield(L, -2, name);
 }
 
-int unicode_lower(lua_State *L) {
-	push_value(L, boost::locale::to_lower(check_string(L, 1)));
-	return 1;
-}
-
-int unicode_fold(lua_State *L) {
-	push_value(L, boost::locale::fold_case(check_string(L, 1)));
-	return 1;
+template<std::string (*func)(const char *, std::locale const&)>
+char *wrap(const char *str, char **err) {
+	try {
+		return strdup(func(str, std::locale()).c_str());
+	} catch (std::exception const& e) {
+		*err = strdup(e.what());
+		return nullptr;
+	}
 }
 }
 
 extern "C" int luaopen_unicode_impl(lua_State *L) {
+	lua_getglobal(L, "require");
+	lua_pushstring(L, "ffi");
+	lua_call(L, 1, 1);
+	lua_getfield(L, -1, "cast");
+	lua_remove(L, -2); // ffi table
+
 	lua_createtable(L, 0, 3);
-	set_field<unicode_upper>(L, "to_upper_case");
-	set_field<unicode_lower>(L, "to_lower_case");
-	set_field<unicode_fold>(L, "to_fold_case");
+	push_ffi_function(L, "to_upper_case", wrap<boost::locale::to_upper<char>>);
+	push_ffi_function(L, "to_lower_case", wrap<boost::locale::to_lower<char>>);
+	push_ffi_function(L, "to_fold_case", wrap<boost::locale::fold_case<char>>);
+
+	lua_remove(L, -2); // ffi.cast function
 	return 1;
 }