From e4ecc217e31c15cec5c777d6729b09df2f742687 Mon Sep 17 00:00:00 2001
From: jacob <jacob@cagori.com>
Date: Thu, 4 Apr 2024 14:51:32 -0500
Subject: [PATCH] process unicode input in win32

---
 src/common.h    | 11 ++---------
 src/sys.h       |  2 +-
 src/sys_win32.c | 51 ++++++++++++++++++++++++++++++++++++++-----------
 src/utf.c       | 10 ++++++++++
 src/utf.h       |  2 ++
 5 files changed, 55 insertions(+), 21 deletions(-)

diff --git a/src/common.h b/src/common.h
index 4845f096..58d58c2a 100644
--- a/src/common.h
+++ b/src/common.h
@@ -402,16 +402,9 @@ struct buffer {
     .text = (u8 *)(cstr_lit)                    \
 }
 
-#define STRING_FROM_ARRAY(a) \
-    ( \
-        /* Must be array */ \
-        ASSERT(IS_ARRAY(a)), \
-        /* Must be array of bytes */ \
-        ASSERT(sizeof(a[0]) == sizeof(u8)), \
-        ((struct string) { .len = ARRAY_COUNT(a), .text = (u8 *)(a) }) \
-    )
+#define STRING_FROM_BUFFER(buff) ((struct string) { buff.size, buff.data })
 
-#define STRING_FROM_BUFFER(buff) ((struct string) { buff.size, buff.data})
+#define STRING_FROM_ARRAY(a) STRING_FROM_BUFFER(BUFFER_FROM_ARRAY(a))
 
 /* ========================== *
  * Math types
diff --git a/src/sys.h b/src/sys.h
index c8c5b56c..50cec0cd 100644
--- a/src/sys.h
+++ b/src/sys.h
@@ -133,7 +133,7 @@ struct sys_event {
     b32 is_repeat;
 
     /* SYS_EVENT_KIND_TEXT */
-    u32 text_character;
+    u32 text_codepoint;
 
     /* SYS_EVENT_KIND_CURSOR_MOVE */
     struct v2 cursor_position;
diff --git a/src/sys_win32.c b/src/sys_win32.c
index 9e9b29ac..0d574c37 100644
--- a/src/sys_win32.c
+++ b/src/sys_win32.c
@@ -59,6 +59,8 @@ struct win32_window {
     u32 tid;
     struct sync_flag ready_sf;
 
+    u16 utf16_high_surrogate_last_input;
+
     struct sys_rw_mutex settings_rw_mutex;
     struct sys_window_settings settings;
 
@@ -914,19 +916,46 @@ INTERNAL LRESULT CALLBACK win32_window_proc(HWND hwnd, UINT msg, WPARAM wparam,
         /* Text */
         case WM_SYSCHAR:
         case WM_CHAR: {
-            u32 character = (u32)wparam;
-            if (character == '\r') {
-                character = '\n';  /* Just treat all \r as newline */
-            }
-            if((character >= 32 && character != 127) || character == '\t' || character == '\n') {
-                win32_window_process_event(
-                    window,
-                    (struct sys_event) {
-                        .kind = SYS_EVENT_KIND_TEXT,
-                        .text_character = character
+            u16 utf16_char = (u32)wparam;
+
+            /* Decode */
+            u32 codepoint = 0;
+            if (utf16_is_high_surrogate(utf16_char)) {
+                window->utf16_high_surrogate_last_input = utf16_char;
+            } else if (utf16_is_low_surrogate(utf16_char)) {
+                u16 high = window->utf16_high_surrogate_last_input;
+                u16 low = utf16_char;
+                if (high) {
+                    u16 utf16_pair_bytes[2] = { high, low };
+                    struct utf16_decode_result decoded = utf16_decode((struct string16) { .len = ARRAY_COUNT(utf16_pair_bytes), .text = utf16_pair_bytes });
+                    if (decoded.advance16 == 2 && decoded.codepoint < U32_MAX) {
+                        codepoint = decoded.codepoint;
+                    } else {
+                        codepoint = '?';
                     }
-                );
+                }
+                window->utf16_high_surrogate_last_input = 0;
+            } else {
+                window->utf16_high_surrogate_last_input = 0;
+                codepoint = utf16_char;
             }
+
+            if (codepoint) {
+                if (codepoint == '\r') {
+                    codepoint = '\n';  /* Just treat all \r as newline */
+                }
+                if((codepoint >= 32 && codepoint != 127) || codepoint == '\t' || codepoint == '\n') {
+                    win32_window_process_event(
+                        window,
+                        (struct sys_event) {
+                        .kind = SYS_EVENT_KIND_TEXT,
+                        .text_codepoint = codepoint
+                    }
+                    );
+                }
+            }
+
+
         } break;
 
         /* Mouse buttons */
diff --git a/src/utf.c b/src/utf.c
index b6fe488f..17a09da9 100644
--- a/src/utf.c
+++ b/src/utf.c
@@ -154,6 +154,16 @@ struct utf16_encode_result utf16_encode(u32 codepoint)
     return res;
 }
 
+b32 utf16_is_high_surrogate(u16 c)
+{
+    return 0xD800 <= c && c < 0xDC00;
+}
+
+b32 utf16_is_low_surrogate(u16 c)
+{
+    return 0xDC00 <= c && c < 0xE000;
+}
+
 /* ========================== *
  * utf32
  * ========================== */
diff --git a/src/utf.h b/src/utf.h
index 6c3a33ef..119a1a93 100644
--- a/src/utf.h
+++ b/src/utf.h
@@ -34,6 +34,8 @@ struct utf16_encode_result {
 
 struct utf16_decode_result utf16_decode(struct string16 str);
 struct utf16_encode_result utf16_encode(u32 codepoint);
+b32 utf16_is_high_surrogate(u16 c);
+b32 utf16_is_low_surrogate(u16 c);
 
 /* ========================== *
  * utf32