cleanup

6 files changed, 123 insertions, 26 deletions

diff --git a/script/encoder/ansi.lua b/script/encoder/ansi.lua
new file mode 100644
index 00000000..1016e668
--- /dev/null
+++ b/script/encoder/ansi.lua
@@ -0,0 +1,24 @@
+local platform = require 'bee.platform'
+local unicode
+
+if platform.OS == 'Windows' then
+    unicode = require 'bee.unicode'
+end
+
+local m = {}
+
+function m.decode(text)
+    if not unicode then
+        return text
+    end
+    return unicode.a2u(text)
+end
+
+function m.encode(text)
+    if not unicode then
+        return text
+    end
+    return unicode.u2a(text)
+end
+
+return m
diff --git a/script/encoder/init.lua b/script/encoder/init.lua
index a8f9a8ac..152193e7 100644
--- a/script/encoder/init.lua
+++ b/script/encoder/init.lua
@@ -1,24 +1,5 @@
-local platform = require 'bee.platform'
-local unicode
-
-if platform.OS == 'Windows' then
-    unicode = require 'bee.unicode'
-end
-
-local m = {}
-
-function m.ansi2utf8(text)
-    if not unicode then
-        return text
-    end
-    return unicode.a2u(text)
-end
-
-function m.utf82ansi(text)
-    if not unicode then
-        return text
-    end
-    return unicode.u2a(text)
-end
-
-return m
+return {
+    ansi    = require 'encoder.ansi',
+    utf16le = require 'encoder.utf16le',
+    utf16be = require 'encoder.utf16be'
+}
diff --git a/script/encoder/utf16be.lua b/script/encoder/utf16be.lua
new file mode 100644
index 00000000..179a676d
--- /dev/null
+++ b/script/encoder/utf16be.lua
@@ -0,0 +1,46 @@
+local function tochar(code)
+    return string.char((code >> 8) & 0xFF, code & 0xFF)
+end
+
+local function tobyte(s, i)
+    local h, l = string.byte(s, i, i+1)
+    return (h << 8) | l
+end
+
+local function char(code)
+    if code <= 0xffff then
+        return tochar(code)
+    end
+    code = code - 0x10000
+    return tochar(0xD800 + (code >> 10))..tochar(0xDC00 + (code & 0x3FF))
+end
+
+local m = {}
+
+function m.encode(s)
+    local r = {}
+    for _, c in utf8.codes(s) do
+        r[#r+1] = char(c)
+    end
+    return table.concat(r)
+end
+
+function m.decode(s)
+    local r = {}
+    local i = 1
+    while i < #s do
+        local code1 = tobyte(s, i)
+        if code1 < 0xD800 then
+            r[#r+1] = utf8.char(code1)
+        else
+            i = i + 2
+            local code2 = tobyte(s, i)
+            local code = 0x10000 + ((code1 - 0xD800) << 10) + ((code2 - 0xDC00) & 0x3FF)
+            r[#r+1] = utf8.char(code)
+        end
+        i = i + 2
+    end
+    return table.concat(r)
+end
+
+return m
diff --git a/script/encoder/utf16le.lua b/script/encoder/utf16le.lua
new file mode 100644
index 00000000..cfbc33c2
--- /dev/null
+++ b/script/encoder/utf16le.lua
@@ -0,0 +1,46 @@
+local function tochar(code)
+    return string.char(code & 0xFF, (code >> 8) & 0xFF)
+end
+
+local function tobyte(s, i)
+    local l, h = string.byte(s, i, i+1)
+    return (h << 8) | l
+end
+
+local function char(code)
+    if code <= 0xffff then
+        return tochar(code)
+    end
+    code = code - 0x10000
+    return tochar(0xD800 + (code >> 10))..tochar(0xDC00 + (code & 0x3FF))
+end
+
+local m = {}
+
+function m.encode(s)
+    local r = {}
+    for _, c in utf8.codes(s) do
+        r[#r+1] = char(c)
+    end
+    return table.concat(r)
+end
+
+function m.decode(s)
+    local r = {}
+    local i = 1
+    while i < #s do
+        local code1 = tobyte(s, i)
+        if code1 < 0xD800 then
+            r[#r+1] = utf8.char(code1)
+        else
+            i = i + 2
+            local code2 = tobyte(s, i)
+            local code = 0x10000 + ((code1 - 0xD800) << 10) + ((code2 - 0xDC00) & 0x3FF)
+            r[#r+1] = utf8.char(code)
+        end
+        i = i + 2
+    end
+    return table.concat(r)
+end
+
+return m
diff --git a/script/files.lua b/script/files.lua
index dac23b86..566b012c 100644
--- a/script/files.lua
+++ b/script/files.lua
@@ -141,7 +141,7 @@ function m.setText(uri, text, isTrust, instance)
     end
     if not isTrust then
         if config.get 'Lua.runtime.fileEncoding' == 'ansi' then
-            text = encoder.ansi2utf8(text)
+            text = encoder.ansi.decode(text)
         end
     end
     if file.originText == text then
diff --git a/script/library.lua b/script/library.lua
index ba7dc657..66b26b0d 100644
--- a/script/library.lua
+++ b/script/library.lua
@@ -237,7 +237,7 @@ local function initBuiltIn()
         if metaDoc then
             local outPath = metaPath / libName
             if encoding == 'ansi' then
-                metaDoc = encoder.utf82ansi(metaDoc)
+                metaDoc = encoder.ansi.encode(metaDoc)
             end
             out:saveFile(libName, metaDoc)
             m.metaPaths[#m.metaPaths+1] = outPath:string()