From fce9f872655f40b9d8ccbcfbb0b9356c6719a3e2 Mon Sep 17 00:00:00 2001
From: Josh Holtrop <jholtrop@gmail.com>
Date: Sat, 21 Jan 2017 13:28:20 -0500
Subject: [PATCH] add EncodedString module

---
 src/core/EncodedString.cc | 44 +++++++++++++++++++++++++++
 src/core/EncodedString.h  | 63 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 107 insertions(+)
 create mode 100644 src/core/EncodedString.cc
 create mode 100644 src/core/EncodedString.h

diff --git a/src/core/EncodedString.cc b/src/core/EncodedString.cc
new file mode 100644
index 0000000..8e75268
--- /dev/null
+++ b/src/core/EncodedString.cc
@@ -0,0 +1,44 @@
+#include "EncodedString.h"
+#include <string.h>
+
+uint32_t EncodedString::iterator::operator*()
+{
+    if (m_offset >= m_encoded_string.size())
+    {
+        return INVALID_CODE_POINT;
+    }
+    return Encoding::decode(m_encoded_string.encoding(), &m_encoded_string[m_offset]);
+}
+
+EncodedString::iterator & EncodedString::iterator::operator++()
+{
+    if (m_offset < m_encoded_string.size())
+    {
+        m_offset += Encoding::num_bytes_in_code_point(m_encoded_string.encoding(), &m_encoded_string[m_offset]);
+    }
+    return *this;
+}
+
+EncodedString::EncodedString(Encoding::Type encoding)
+    : m_encoding(encoding),
+      m_data(nullptr)
+{
+}
+
+EncodedString::EncodedString(const uint8_t * data, size_t size, Encoding::Type encoding)
+    : m_encoding(encoding),
+      m_size(size)
+{
+    m_data = new uint8_t[size];
+    memcpy(m_data, data, size);
+}
+
+EncodedString::iterator EncodedString::begin() const
+{
+    return iterator(*this, 0u);
+}
+
+EncodedString::iterator EncodedString::end() const
+{
+    return iterator(*this, m_size);
+}
diff --git a/src/core/EncodedString.h b/src/core/EncodedString.h
new file mode 100644
index 0000000..f57f8f7
--- /dev/null
+++ b/src/core/EncodedString.h
@@ -0,0 +1,63 @@
+#ifndef ENCODEDSTRING_H
+#define ENCODEDSTRING_H
+
+#include "Encoding.h"
+#include <stdint.h>
+
+class EncodedString
+{
+public:
+    enum : uint32_t
+    {
+        INVALID_CODE_POINT = 0xFFFFFFFFu,
+    };
+
+    class iterator
+    {
+    public:
+        iterator(const EncodedString & encoded_string, size_t offset)
+            : m_encoded_string(encoded_string),
+              m_offset(offset)
+        {
+        }
+        uint32_t operator*();
+        iterator & operator++();
+        bool operator==(const iterator & other) const
+        {
+            return m_offset == other.m_offset;
+        }
+        bool operator!=(const iterator & other) const
+        {
+            return m_offset != other.m_offset;
+        }
+
+    protected:
+        const EncodedString & m_encoded_string;
+        size_t m_offset;
+    };
+
+    EncodedString(Encoding::Type encoding = Encoding::UTF_8);
+    EncodedString(const uint8_t * data, size_t size, Encoding::Type encoding = Encoding::UTF_8);
+
+    iterator begin() const;
+    iterator end() const;
+    size_t size() const
+    {
+        return m_size;
+    }
+    Encoding::Type encoding() const
+    {
+        return m_encoding;
+    }
+    const uint8_t & operator[](size_t index) const
+    {
+        return m_data[index];
+    }
+
+protected:
+    Encoding::Type m_encoding;
+    uint8_t * m_data;
+    size_t m_size;
+};
+
+#endif