From c47ff998214b8729bb0586beab5a4f8217defe8e Mon Sep 17 00:00:00 2001
From: Gregory Lirent <opensource@lirent.ru>
Date: Tue, 31 May 2022 15:27:46 +0300
Subject: [PATCH] Add charsize

---
 char16size.c | 20 ++++++++++++++++
 char32size.c | 11 +++++++++
 charsize.c   | 68 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 99 insertions(+)
 create mode 100644 char16size.c
 create mode 100644 char32size.c
 create mode 100644 charsize.c

diff --git a/char16size.c b/char16size.c
new file mode 100644
index 0000000..bbaa2c8
--- /dev/null
+++ b/char16size.c
@@ -0,0 +1,20 @@
+/* This software is licensed by the MIT License, see LICENSE file */
+/*                                Copyright © 2022 Gregory Lirent */
+
+#include "build.h"
+
+int char16size(const char16_t* s) {
+    unsigned int v;
+
+    memcpy(&v, s, 4);
+
+    if (IS_LITTLE_ENDIAN) {
+        if ((v&0xfc00fc00) == 0xdc00d800) return 2;
+        if ((v&0x0000f800) != 0x0000d800) return 1;
+    } else {
+        if ((v&0xfc00fc00) == 0xd800dc00) return 2;
+        if ((v&0xf8000000) != 0xd8000000) return 1;
+    }
+
+    return 0;
+}
diff --git a/char32size.c b/char32size.c
new file mode 100644
index 0000000..9d29516
--- /dev/null
+++ b/char32size.c
@@ -0,0 +1,11 @@
+/* This software is licensed by the MIT License, see LICENSE file */
+/*                                Copyright © 2022 Gregory Lirent */
+
+#include "build.h"
+
+int char32size(const char32_t* s) {
+    if (*(unsigned int*)s <= 0x10ffff && (*(unsigned int*)s < 0xd800 || *(unsigned int*)s > 0xdfff))
+        return 1;
+
+    return 0;
+}
diff --git a/charsize.c b/charsize.c
new file mode 100644
index 0000000..a175bb3
--- /dev/null
+++ b/charsize.c
@@ -0,0 +1,68 @@
+/* This software is licensed by the MIT License, see LICENSE file */
+/*                                Copyright © 2022 Gregory Lirent */
+
+#include "build.h"
+
+int charsize(const char* s) {
+
+    unsigned int v;
+  #ifndef UNICODE_CHARSIZE_CHECK_DISABLE
+    unsigned int c;
+  #endif
+    if (*(unsigned char*)s <= 0x7f)
+        return 1;
+
+    memcpy(&v, s, 4);
+
+    if (IS_LITTLE_ENDIAN) {
+        if ((v&0x0000c0e0) == 0x000080c0) {
+  #ifndef UNICODE_CHARSIZE_CHECK_DISABLE
+            if (v&0x001f) // > 0x7f
+  #endif
+                return 2;
+        } else {
+
+  #ifndef UNICODE_CHARSIZE_CHECK_DISABLE
+            c  = (v&0x000f) << 6;
+            c |= (v&0x3f00) >> 8;
+
+            if (c >= 0x20) {
+  #endif
+                if ((v&0xc0c0c0f8) == 0x808080f0)
+                    return 4;
+                if ((v&0x00c0c0f0) == 0x008080e0)
+  #ifndef UNICODE_CHARSIZE_CHECK_DISABLE
+                if (c >= (0xd8<<2) && c <= (0xdf<<2)) // != (0xd800 <-> 0xdfff)
+  #endif
+                    return 3;
+            }
+  #ifndef UNICODE_CHARSIZE_CHECK_DISABLE
+        }
+  #endif
+    } else {
+        if ((v&0xe0c00000) == 0xc0800000) {
+  #ifndef UNICODE_CHARSIZE_CHECK_DISABLE
+            if (v&0x1f000000) // > 0x7f
+  #endif
+                return 2;
+        } else {
+  #ifndef UNICODE_CHARSIZE_CHECK_DISABLE
+            c  = (v&0x0f<<24) >> 2;
+            c |= (v&(0x3f<<16));
+
+            if (c >= (0x20<<16)) {
+  #endif
+                if ((v&0xf8c0c0c0) == 0xf0808080)
+                    return 4;
+                if ((v&0xf0c0c000) == 0xe0808000)
+  #ifndef UNICODE_CHARSIZE_CHECK_DISABLE
+                if (c >= (0xd8<<22) && c <= (0xdf<<22)) // != (0xd800 <-> 0xdfff)
+  #endif
+                    return 3;
+  #ifndef UNICODE_CHARSIZE_CHECK_DISABLE
+            }
+  #endif
+        }
+    }
+    return 0;
+}