From c489549b56861520b0d120034150f82a67faf953 Mon Sep 17 00:00:00 2001
From: Dane Springmeyer <dane@mapbox.com>
Date: Wed, 4 Jun 2014 12:09:42 -0700
Subject: [PATCH] add benchmark of utf 32->8 encoding speed

---
 benchmark/test_utf_encoding.cpp | 100 ++++++++++++++++++++++++++++++++
 1 file changed, 100 insertions(+)
 create mode 100644 benchmark/test_utf_encoding.cpp

diff --git a/benchmark/test_utf_encoding.cpp b/benchmark/test_utf_encoding.cpp
new file mode 100644
index 000000000..ad7fcc361
--- /dev/null
+++ b/benchmark/test_utf_encoding.cpp
@@ -0,0 +1,100 @@
+#include "bench_framework.hpp"
+#include <mapnik/unicode.hpp>
+#include <mapnik/value.hpp>
+#include <codecvt>
+#include <boost/locale.hpp>
+
+class test : public benchmark::test_case
+{
+    std::string utf8_;
+public:
+    test(mapnik::parameters const& params)
+     : test_case(params),
+       utf8_(u8"שלום") {}
+    bool validate() const
+    {
+        std::wstring_convert<std::codecvt_utf8<char32_t>, char32_t> utf32conv;
+        std::u32string utf32 = utf32conv.from_bytes(utf8_);
+        if (utf32.size() != 4) return false;
+        if (utf32[0] != 0x5e9 &&
+            utf32[1] != 0x5dc &&
+            utf32[2] != 0x5d5 &&
+            utf32[3] != 0x5dd) return false;
+        return true;
+    }
+    void operator()() const
+    {
+        std::u32string utf32;
+        std::wstring_convert<std::codecvt_utf8<char32_t>, char32_t> utf32conv;
+        for (std::size_t i=0;i<iterations_;++i) {
+             utf32 = utf32conv.from_bytes(utf8_);
+        }
+    }
+};
+
+class test2 : public benchmark::test_case
+{
+    std::string utf8_;
+public:
+    test2(mapnik::parameters const& params)
+     : test_case(params),
+       utf8_(u8"שלום") {}
+    bool validate() const
+    {
+        std::u32string utf32 = boost::locale::conv::utf_to_utf<char32_t>(utf8_);
+        if (utf32.size() != 4) return false;
+        if (utf32[0] != 0x5e9 &&
+            utf32[1] != 0x5dc &&
+            utf32[2] != 0x5d5 &&
+            utf32[3] != 0x5dd) return false;
+        return true;
+    }
+    void operator()() const
+    {
+         std::u32string utf32;
+         for (std::size_t i=0;i<iterations_;++i) {
+             utf32 = boost::locale::conv::utf_to_utf<char32_t>(utf8_);
+         }
+    }
+};
+
+class test3 : public benchmark::test_case
+{
+    std::string utf8_;
+public:
+    test3(mapnik::parameters const& params)
+     : test_case(params),
+       utf8_(u8"שלום") {}
+    bool validate() const
+    {
+        mapnik::transcoder tr_("utf-8");
+        mapnik::value_unicode_string utf32 = tr_.transcode(utf8_.data(),utf8_.size());
+        //std::u32string utf32 = boost::locale::conv::utf_to_utf<char32_t>(utf8_);
+        if (utf32.length() != 4) return false;
+        if (utf32[0] != 0x5e9 &&
+            utf32[1] != 0x5dc &&
+            utf32[2] != 0x5d5 &&
+            utf32[3] != 0x5dd) return false;
+        return true;
+    }
+    void operator()() const
+    {
+        mapnik::transcoder tr_("utf-8");
+        mapnik::value_unicode_string utf32;
+        for (std::size_t i=0;i<iterations_;++i) {
+            utf32 = tr_.transcode(utf8_.data(),utf8_.size());
+        }
+    }
+};
+
+int main(int argc, char** argv)
+{
+    mapnik::parameters params;
+    benchmark::handle_args(argc,argv,params);
+    test test_runner(params);
+    run(test_runner,"utf encode std::codecvt");
+    test2 test_runner2(params);
+    run(test_runner2,"utf encode boost::locale");
+    test3 test_runner3(params);
+    return run(test_runner3,"utf encode ICU");
+}