[core] Default "collator" implementation

 - Based on nunicode
 - Not locale-aware
 - Used by linux and Qt builds

6 files changed, 143 insertions, 2 deletions

diff --git a/platform/default/collator.cpp b/platform/default/collator.cpp
new file mode 100644
index 0000000000..b7f256756e
--- /dev/null
+++ b/platform/default/collator.cpp
@@ -0,0 +1,79 @@
+#include <mbgl/style/expression/collator.hpp>
+#include <mbgl/util/platform.hpp>
+#include <libnu/strcoll.h>
+#include <unaccent.hpp>
+
+/*
+    The default implementation of Collator ignores locale.
+    Case sensitivity and collation order are based on
+    Default Unicode Collation Element Table (DUCET).
+
+    Diacritic-insensitivity is implemented with nunicode's
+    non-standard "unaccent" functionality, which is tailored
+    to European languages.
+
+    It would be possible to implement locale awareness using ICU,
+    but would require bundling locale data.
+*/
+
+namespace mbgl {
+namespace style {
+namespace expression {
+
+class Collator::Impl {
+public:
+    Impl(bool caseSensitive_, bool diacriticSensitive_, optional<std::string>)
+        : caseSensitive(caseSensitive_)
+        , diacriticSensitive(diacriticSensitive_)
+    {}
+
+    bool operator==(const Impl& other) const {
+        return caseSensitive == other.caseSensitive &&
+            diacriticSensitive == other.diacriticSensitive;
+    }
+
+    int compare(const std::string& lhs, const std::string& rhs) const {
+        if (caseSensitive && diacriticSensitive) {
+            return nu_strcoll(lhs.c_str(), rhs.c_str(),
+                              nu_utf8_read, nu_utf8_read);
+        } else if (!caseSensitive && diacriticSensitive) {
+            return nu_strcasecoll(lhs.c_str(), rhs.c_str(),
+                                  nu_utf8_read, nu_utf8_read);
+        } else if (caseSensitive && !diacriticSensitive) {
+            return nu_strcoll(platform::unaccent(lhs).c_str(), platform::unaccent(rhs).c_str(),
+                              nu_utf8_read, nu_utf8_read);
+        } else {
+            return nu_strcasecoll(platform::unaccent(lhs).c_str(), platform::unaccent(rhs).c_str(),
+                                  nu_utf8_read, nu_utf8_read);
+        }
+    }
+
+    std::string resolvedLocale() const {
+        return "";
+    }
+private:
+    bool caseSensitive;
+    bool diacriticSensitive;
+};
+
+
+Collator::Collator(bool caseSensitive, bool diacriticSensitive, optional<std::string> locale_)
+    : impl(std::make_shared<Impl>(caseSensitive, diacriticSensitive, std::move(locale_)))
+{}
+
+bool Collator::operator==(const Collator& other) const {
+    return *impl == *(other.impl);
+}
+
+int Collator::compare(const std::string& lhs, const std::string& rhs) const {
+    return impl->compare(lhs, rhs);
+}
+
+std::string Collator::resolvedLocale() const {
+    return impl->resolvedLocale();
+}
+
+
+} // namespace expression
+} // namespace style
+} // namespace mbgl
diff --git a/platform/default/unaccent.cpp b/platform/default/unaccent.cpp
new file mode 100644
index 0000000000..faefb4b4cd
--- /dev/null
+++ b/platform/default/unaccent.cpp
@@ -0,0 +1,43 @@
+#include <mbgl/util/platform.hpp>
+#include <libnu/unaccent.h>
+#include <unaccent.hpp>
+
+#include <cstring>
+#include <sstream>
+
+namespace mbgl { namespace platform {
+
+std::string unaccent(const std::string& str)
+{
+    std::stringstream output;
+    char const *itr = str.c_str(), *nitr;
+    char const *end = itr + str.length();
+    char lo[5] = { 0 };
+
+    for (; itr < end; itr = nitr)
+    {
+        uint32_t code_point = 0;
+        char const* buf = nullptr;
+
+        nitr = _nu_tounaccent(itr, end, nu_utf8_read, &code_point, &buf, nullptr);
+        if (buf != nullptr)
+        {
+            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
+        {
+            output.write(itr, nitr - itr);
+        }
+    }
+
+    return output.str();
+}
+
+} // namespace platform
+} // namespace mbgl
diff --git a/platform/default/unaccent.hpp b/platform/default/unaccent.hpp
new file mode 100644
index 0000000000..85ac37a7de
--- /dev/null
+++ b/platform/default/unaccent.hpp
@@ -0,0 +1,13 @@
+#pragma once
+
+#include <string>
+
+namespace mbgl {
+namespace platform {
+
+// Non-locale-aware diacritic folding based on nunicode
+// Used as a fallback when locale-aware comparisons aren't available
+std::string unaccent(const std::string &string);
+
+} // namespace platform
+} // namespace mbgl
diff --git a/platform/linux/config.cmake b/platform/linux/config.cmake
index fe4cd5fcc3..08376cbfa3 100644
--- a/platform/linux/config.cmake
+++ b/platform/linux/config.cmake
@@ -58,8 +58,11 @@ macro(mbgl_platform_core)
         PRIVATE platform/default/string_stdlib.cpp
         PRIVATE platform/default/thread.cpp
         PRIVATE platform/default/bidi.cpp
+        PRIVATE platform/default/collator.cpp
         PRIVATE platform/default/local_glyph_rasterizer.cpp
         PRIVATE platform/default/thread_local.cpp
+        PRIVATE platform/default/unaccent.cpp
+        PRIVATE platform/default/unaccent.hpp
         PRIVATE platform/default/utf.cpp
 
         # Image handling
diff --git a/platform/qt/config.cmake b/platform/qt/config.cmake
index fdbc463601..757c0ac44a 100644
--- a/platform/qt/config.cmake
+++ b/platform/qt/config.cmake
@@ -51,8 +51,6 @@ macro(mbgl_platform_core)
         target_sources(mbgl-core PRIVATE platform/qt/src/bidi.cpp)
     endif()
 
-    target_sources(mbgl-core PRIVATE platform/default/local_glyph_rasterizer.cpp)
-
     if (CMAKE_HOST_SYSTEM_NAME STREQUAL "Windows")
         target_add_mason_package(mbgl-core PRIVATE optional)
         target_add_mason_package(mbgl-core PRIVATE tao_tuple)
diff --git a/platform/qt/qt.cmake b/platform/qt/qt.cmake
index dda15174fb..fafbb895dc 100644
--- a/platform/qt/qt.cmake
+++ b/platform/qt/qt.cmake
@@ -38,6 +38,11 @@ set(MBGL_QT_CORE_FILES
     PRIVATE platform/qt/src/timer.cpp
     PRIVATE platform/qt/src/timer_impl.hpp
     PRIVATE platform/qt/src/utf.cpp
+
+    PRIVATE platform/default/local_glyph_rasterizer.cpp
+    PRIVATE platform/default/collator.cpp
+    PRIVATE platform/default/unaccent.cpp
+    PRIVATE platform/default/unaccent.hpp
 )
 
 set(MBGL_QT_FILESOURCE_FILES