From 926ad3ec0da46dc68f2af6b5867f387fa8c35f73 Mon Sep 17 00:00:00 2001
From: Josh Holtrop <jholtrop@gmail.com>
Date: Fri, 27 Nov 2020 20:51:40 -0500
Subject: [PATCH] Add Encoding.encode() and Encoding.decode().

---
 src/jes/core/encoding.d | 156 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 156 insertions(+)

diff --git a/src/jes/core/encoding.d b/src/jes/core/encoding.d
index b8ee7d5..8fe64c4 100644
--- a/src/jes/core/encoding.d
+++ b/src/jes/core/encoding.d
@@ -18,6 +18,162 @@ struct Encoding
         this.type = type;
     }
 
+    uint decode(const(ubyte) * data, size_t * encoded_size)
+    {
+        uint result;
+        size_t size;
+        ubyte following_bytes;
+        ubyte c;
+
+        switch (type)
+        {
+        case UTF8:
+            c = *data;
+            if ((c & 0x80u) == 0u)
+            {
+                result = c;
+                size = 1u;
+            }
+            else
+            {
+                if ((c & 0xE0u) == 0xC0u)
+                {
+                    result = c & 0x1Fu;
+                    following_bytes = 1u;
+                }
+                else if ((c & 0xF0u) == 0xE0u)
+                {
+                    result = c & 0x0Fu;
+                    following_bytes = 2u;
+                }
+                else if ((c & 0xF8u) == 0xF0u)
+                {
+                    result = c & 0x07u;
+                    following_bytes = 3u;
+                }
+                else if ((c & 0xFCu) == 0xF8u)
+                {
+                    result = c & 0x03u;
+                    following_bytes = 4u;
+                }
+                else if ((c & 0xFEu) == 0xFCu)
+                {
+                    result = c & 0x01u;
+                    following_bytes = 5u;
+                }
+                size = following_bytes + 1u;
+                while (following_bytes-- > 0u)
+                {
+                    data++;
+                    result <<= 6u;
+                    result |= *data & 0x3Fu;
+                }
+            }
+            break;
+        case CP1252:
+            /* TODO */
+            break;
+        case UTF16_LE:
+            result = data[0] | (data[1] << 8u);
+            size = 2u;
+            break;
+        case UTF16_BE:
+            result = (data[0] << 8u) | data[1];
+            size = 2u;
+            break;
+        default:
+            break;
+        }
+
+        if (encoded_size != null)
+        {
+            *encoded_size = size;
+        }
+
+        return result;
+    }
+
+    ubyte num_bytes_to_encode_code_point(uint code_point)
+    {
+        switch (type)
+        {
+        case UTF8:
+            if (code_point <= 0x7Fu)
+            {
+                return 1u;
+            }
+            else if (code_point <= 0x7FFu)
+            {
+                return 2u;
+            }
+            else if (code_point <= 0xFFFFu)
+            {
+                return 3u;
+            }
+            else if (code_point <= 0x1FFFFFu)
+            {
+                return 4u;
+            }
+            else if (code_point <= 0x3FFFFFFu)
+            {
+                return 5u;
+            }
+            else
+            {
+                return 6u;
+            }
+        case CP1252:
+            /* TODO */
+            break;
+        case UTF16_LE:
+        case UTF16_BE:
+            return 2u;
+        default:
+            break;
+        }
+
+        return 0u;
+    }
+
+    ubyte encode(uint code_point, ubyte * buffer)
+    {
+        ubyte size;
+        switch (type)
+        {
+        case UTF8:
+            size = num_bytes_to_encode_code_point(code_point);
+            if (size == 1u)
+            {
+                *buffer = cast(ubyte)code_point;
+            }
+            else if (size > 1u)
+            {
+                *buffer = cast(ubyte)((0xFFu << (8u - size)) | (code_point >> (6u * (size - 1u))));
+                const ubyte end = cast(ubyte)(size - 1u);
+                for (ubyte i = 0u; i < end; i++)
+                {
+                    buffer[end - i] = 0x80u | (code_point >> (6u * i) & 0x3Fu);
+                }
+            }
+            return size;
+        case CP1252:
+            /* TODO */
+            break;
+        case UTF16_LE:
+            buffer[0] = cast(ubyte)code_point;
+            buffer[1] = cast(ubyte)(code_point >> 8u);
+            return 2u;
+        case UTF16_BE:
+            buffer[0] = cast(ubyte)(code_point >> 8u);
+            buffer[1] = cast(ubyte)code_point;
+            return 2u;
+        default:
+            break;
+        }
+
+        return 0u;
+    }
+
     static Encoding detect_encoding(const ubyte * data, size_t n, size_t * bom_size)
     {
         *bom_size = 0u;