From 4bef36fcae1897ac8d0f4826575b8d5ef7e2ce0a Mon Sep 17 00:00:00 2001
From: fallenoak <git@fallenoak.me>
Date: Sun, 22 Nov 2020 00:12:24 -0600
Subject: [PATCH] feat(unicode): add SUniSGetUTF8

---
 storm/Unicode.cpp | 59 +++++++++++++++++++++++++++++++++++++++++++++++
 storm/Unicode.hpp |  8 +++++++
 test/Unicode.cpp  | 32 +++++++++++++++++++++++++
 3 files changed, 99 insertions(+)
 create mode 100644 storm/Unicode.cpp
 create mode 100644 storm/Unicode.hpp
 create mode 100644 test/Unicode.cpp

diff --git a/storm/Unicode.cpp b/storm/Unicode.cpp
new file mode 100644
index 0000000..f5aa864
--- /dev/null
+++ b/storm/Unicode.cpp
@@ -0,0 +1,59 @@
+#include "storm/Unicode.hpp"
+
+uint32_t SUniSGetUTF8(const uint8_t* strptr, int32_t* chars) {
+    if (chars) {
+        *chars = 0;
+    }
+
+    if (!strptr || !*strptr) {
+        return -1;
+    }
+
+    if (chars) {
+        *chars = *chars + 1;
+    }
+
+    uint32_t value = *strptr;
+    int32_t extra;
+
+    if ((value & 0xFE) == 0xFC) {
+        value &= 0x01;
+        extra = 5;
+    } else if ((value & 0xFC) == 0xF8) {
+        value &= 0x03;
+        extra = 4;
+    } else if ((value & 0xF8) == 0xF0) {
+        value &= 0x07;
+        extra = 3;
+    } else if ((value & 0xF0) == 0xE0) {
+        value &= 0x0F;
+        extra = 2;
+    } else if ((value & 0xE0) == 0xC0) {
+        value &= 0x1F;
+        extra = 1;
+    } else if ((value & 0x80) == 0x00) {
+        return value;
+    } else {
+        return 0x80000000;
+    }
+
+    const uint8_t* next = strptr + 1;
+
+    for (int32_t i = 0; i < extra; i++, next++) {
+        if (!*next) {
+            return -1;
+        }
+
+        if (chars) {
+            *chars = *chars + 1;
+        }
+
+        if ((*next & 0xC0) != 0x80) {
+            return 0x80000000;
+        }
+
+        value = (value << 6) | (*next & 0x3F);
+    }
+
+    return value;
+}
diff --git a/storm/Unicode.hpp b/storm/Unicode.hpp
new file mode 100644
index 0000000..cdccd1d
--- /dev/null
+++ b/storm/Unicode.hpp
@@ -0,0 +1,8 @@
+#ifndef STORM_UNICODE_HPP
+#define STORM_UNICODE_HPP
+
+#include <cstdint>
+
+uint32_t SUniSGetUTF8(const uint8_t* strptr, int32_t* chars);
+
+#endif
diff --git a/test/Unicode.cpp b/test/Unicode.cpp
new file mode 100644
index 0000000..e951c5c
--- /dev/null
+++ b/test/Unicode.cpp
@@ -0,0 +1,32 @@
+#include "storm/Unicode.hpp"
+#include "test/Test.hpp"
+
+TEST_CASE("SUniSGetUTF8", "[unicode]") {
+    SECTION("returns ascii-range utf-8 first character") {
+        auto string = "foobar";
+        int32_t chars = 0;
+        auto code = SUniSGetUTF8(reinterpret_cast<const uint8_t*>(string), &chars);
+
+        REQUIRE(code == 'f');
+        REQUIRE(chars == 1);
+    }
+
+    SECTION("returns non-ascii-range utf-8 first character") {
+        auto string = "\xF0\x9F\x99\x82"
+                      "foobar";
+        int32_t chars = 0;
+        auto code = SUniSGetUTF8(reinterpret_cast<const uint8_t*>(string), &chars);
+
+        REQUIRE(code == 0x1F642);
+        REQUIRE(chars == 4);
+    }
+
+    SECTION("returns null first character") {
+        auto string = "";
+        int32_t chars = 0;
+        auto code = SUniSGetUTF8(reinterpret_cast<const uint8_t*>(string), &chars);
+
+        REQUIRE(code == -1u);
+        REQUIRE(chars == 0);
+    }
+}