From 971583b1c78653fb75085a7fc4d551f95b12e05b Mon Sep 17 00:00:00 2001
From: ValKmjolnir <lhk101lhk101@qq.com>
Date: Wed, 22 May 2024 19:23:09 +0800
Subject: [PATCH] :sparkles: support utf8 ast dump on windows

---
 src/ast_dumper.h              | 16 ++++++++++++++++
 src/natives/nasal_builtin.cpp | 14 ++++++++++++++
 src/natives/nasal_builtin.h   |  3 +++
 std/runtime.nas               |  6 ++++++
 test/utf8chk.nas              | 21 +++++++++++++--------
 5 files changed, 52 insertions(+), 8 deletions(-)

diff --git a/src/ast_dumper.h b/src/ast_dumper.h
index 5825ed7..f95f6bd 100644
--- a/src/ast_dumper.h
+++ b/src/ast_dumper.h
@@ -2,6 +2,10 @@
 
 #include "ast_visitor.h"
 
+#ifdef _WIN32
+#include <windows.h>
+#endif
+
 #include <iostream>
 #include <cstring>
 #include <sstream>
@@ -89,7 +93,19 @@ public:
 
 public:
     void dump(code_block* root) {
+#ifdef _WIN32
+        // store previous code page
+        auto cp = GetConsoleOutputCP();
+        // allow 65001 code page
+        SetConsoleOutputCP(CP_UTF8);
+#endif
+
         root->accept(this);
+
+#ifdef _WIN32
+        // restore previous code page
+        SetConsoleOutputCP(cp);
+#endif
     }
 };
 
diff --git a/src/natives/nasal_builtin.cpp b/src/natives/nasal_builtin.cpp
index 628c711..d9beaba 100644
--- a/src/natives/nasal_builtin.cpp
+++ b/src/natives/nasal_builtin.cpp
@@ -1,6 +1,10 @@
 #include "natives/nasal_builtin.h"
 #include <chrono>
 
+#ifdef _WIN32
+#include <windows.h>
+#endif
+
 namespace nasal {
 
 var builtin_unsafe(context* ctx, gc* ngc) {
@@ -717,6 +721,15 @@ var builtin_ghosttype(context* ctx, gc* ngc) {
     return ngc->newstr(name);
 }
 
+var builtin_set_utf8_output(context* ctx, gc* ngc) {
+#ifdef _WIN32
+    // allow 65001 code page
+    SetConsoleOutputCP(CP_UTF8);
+#endif
+    // do nothing on other platforms
+    return nil;
+}
+
 nasal_builtin_table builtin[] = {
     {"__print", builtin_print},
     {"__println", builtin_println},
@@ -763,6 +776,7 @@ nasal_builtin_table builtin[] = {
     {"__gcinfo", builtin_gcinfo},
     {"__logtime", builtin_logtime},
     {"__ghosttype", builtin_ghosttype},
+    {"__set_utf8_output", builtin_set_utf8_output},
     {nullptr, nullptr}
 };
 
diff --git a/src/natives/nasal_builtin.h b/src/natives/nasal_builtin.h
index 11f096d..2239a17 100644
--- a/src/natives/nasal_builtin.h
+++ b/src/natives/nasal_builtin.h
@@ -81,6 +81,9 @@ var builtin_gcinfo(context*, gc*);
 var builtin_logtime(context*, gc*);
 var builtin_ghosttype(context*, gc*);
 
+// only useful on windows platform
+var builtin_set_utf8_output(context*, gc*);
+
 // register builtin function's name and it's address here in this table below
 // this table must end with {nullptr, nullptr}
 struct nasal_builtin_table {
diff --git a/std/runtime.nas b/std/runtime.nas
index bad3203..2cd57a1 100644
--- a/std/runtime.nas
+++ b/std/runtime.nas
@@ -27,3 +27,9 @@ var gc = {
     },
     info: func() {return __gcinfo;}
 };
+
+var windows = {
+    set_utf8_output: func() {
+        return __set_utf8_output;
+    }
+};
diff --git a/test/utf8chk.nas b/test/utf8chk.nas
index ff2b445..63c5cfe 100644
--- a/test/utf8chk.nas
+++ b/test/utf8chk.nas
@@ -1,11 +1,16 @@
+use std.runtime;
+
+# do nothing if platform is not windows
+runtime.windows.set_utf8_output();
+
 var unicode测试 = func() {
     var 输出=print;
     var 测试成功=[
-        "unicode: utf-8支持测试成功",
-        "目前仅支持utf-8以及ascii格式文件",
-        "注意: windows系统请开启chcp 65001代码页"
+        "unicode: utf-8 支持测试成功",
+        "目前仅支持 utf-8 以及 ascii 格式文件",
+        "注意: windows 系统请开启 chcp 65001 代码页"
     ];
-    foreach(var 内容;测试成功)
+    foreach(var 内容; 测试成功)
         输出(内容~"\n");
 }
 
@@ -41,10 +46,10 @@ var emoji测试 = func() {
         🍾:🍾,
         🐘:🐘
     };
-    foreach(var 📄;📁)
-        💻(📄,🎤);
-    foreach(var 📄;keys(🗄️))
-        💻(📄,🗄️[📄],🎤);
+    foreach(var 📄; 📁)
+        💻(📄, 🎤);
+    foreach(var 📄; keys(🗄️))
+        💻(📄, 🗄️[📄], 🎤);
 }
 
 var dotsgen = func() {