Merge pull request #657 from mapbox/nunicode

Nunicode

3 files changed, 51 insertions, 30 deletions

diff --git a/configure b/configure
index b0a5e5332c..08b972b056 100755
--- a/configure
+++ b/configure
@@ -33,7 +33,7 @@ case ${MASON_PLATFORM} in
         LIBUV_VERSION=0.10.28
         ZLIB_VERSION=system
         BOOST_VERSION=system
-        NUNICODE_VERSION=1.4
+        NUNICODE_VERSION=1.5-dev
         ;;
 esac
 
diff --git a/platform/default/string_stdlib.cpp b/platform/default/string_stdlib.cpp
index 9639cd7b99..ac4c26234c 100644
--- a/platform/default/string_stdlib.cpp
+++ b/platform/default/string_stdlib.cpp
@@ -6,48 +6,71 @@
 #include <libnu/libnu.h>
 #include <cstring>
 
-namespace mbgl {
-namespace platform {
+namespace mbgl { namespace platform {
 
 std::string uppercase(const std::string& str)
 {
-    boost::u8_to_u32_iterator<std::string::const_iterator> itr(str.begin());
-    boost::u8_to_u32_iterator<std::string::const_iterator> end(str.end());
-    std::string output;
-    char buf[5];
-    for (; itr!=end; ++itr)
+    std::stringstream output;
+    char const *itr = str.c_str(), *nitr = itr;
+    char const *end = itr + str.length();
+    char lo[5] = { 0 };
+
+    for (; itr < end; itr = nitr)
     {
-        char const* up = nu_toupper(*itr);
-        if (up != 0) output.append(up);
+        uint32_t code_point = 0;
+        char const* buf = 0;
+
+        nitr = _nu_toupper(itr, end, nu_utf8_read, &code_point, &buf, 0);
+        if (buf != 0)
+        {
+            do
+            {
+                buf = NU_CASEMAP_DECODING_FUNCTION(buf, &code_point);
+                if (code_point == 0) break;
+                output.write(lo, nu_utf8_write(code_point, lo) - lo);
+            }
+            while (code_point != 0);
+        }
         else
         {
-            std::memset(buf, 0, 5);
-            nu_utf8_write(*itr, buf);
-            output.append(buf);
+            output.write(itr, nitr - itr);
         }
     }
-    return output;
+
+    return output.str();
+
 }
 
 std::string lowercase(const std::string& str)
 {
-    boost::u8_to_u32_iterator<std::string::const_iterator> itr(str.begin());
-    boost::u8_to_u32_iterator<std::string::const_iterator> end(str.end());
-    std::string output;
-    char buf[5];
-    for (; itr!=end; ++itr)
+    std::stringstream output;
+    char const *itr = str.c_str(), *nitr = itr;
+    char const *end = itr + str.length();
+    char lo[5] = { 0 };
+
+    for (; itr < end; itr = nitr)
     {
-        char const* lo = nu_tolower(*itr);
-        if (lo != 0)  output.append(lo);
+        uint32_t code_point = 0;
+        char const* buf = 0;
+
+        nitr = _nu_tolower(itr, end, nu_utf8_read, &code_point, &buf, 0);
+        if (buf != 0)
+        {
+            do
+            {
+                buf = NU_CASEMAP_DECODING_FUNCTION(buf, &code_point);
+                if (code_point == 0) break;
+                output.write(lo, nu_utf8_write(code_point, lo) - lo);
+            }
+            while (code_point != 0);
+        }
         else
         {
-            std::memset(buf, 0, 5);
-            nu_utf8_write(*itr, buf);
-            output.append(buf);
+            output.write(itr, nitr - itr);
         }
     }
-    return output;
-}
 
+    return output.str();
 }
-}
+
+}}
diff --git a/test/text_conversions.cpp b/test/text_conversions.cpp
index 4f71d6a61d..756bc4db1f 100644
--- a/test/text_conversions.cpp
+++ b/test/text_conversions.cpp
@@ -30,8 +30,6 @@ TEST(TextConversions, to_lower) {
     EXPECT_EQ(std::string("weisskopfseeadler"), platform::lowercase("weiSSkopfseeadler")); // DE
 
     EXPECT_EQ(std::string("azərbaycan"), platform::lowercase("AZƏRBAYCAN")); // AZ
-
-    //EXPECT_EQ(std::string("ὀδυσσεύσ"), platform::lowercase("ὈΔΥΣΣΕΎΣ")); // GR
-    // nunicode can't map last Σ to word-ending ς
+    EXPECT_EQ(std::string("ὀδυσσεύς"), platform::lowercase("ὈΔΥΣΣΕΎΣ")); // GR
 
 }