[core] Add ICU package for Bidirectional text support and arabic text shaping.

Apply bidi and shaping in symbol_layout.
Add utility functions for converting to and from UTF-16.

CMakeLists.txt

@@ -52,6 +52,7 @@ mason_use(earcut VERSION 0.12.1 HEADER_ONLY)
 mason_use(protozero VERSION 1.4.2 HEADER_ONLY)
 mason_use(pixelmatch VERSION 0.10.0 HEADER_ONLY)
 mason_use(geojson VERSION 0.3.2 HEADER_ONLY)
+mason_use(icu VERSION 58.1)
 
 if(WITH_COVERAGE)
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} --coverage")

cmake/core-files.cmake

@@ -360,6 +360,8 @@ set(MBGL_CORE_FILES
     src/mbgl/text/quads.hpp
     src/mbgl/text/shaping.cpp
     src/mbgl/text/shaping.hpp
+    src/mbgl/text/bidi.cpp
+    src/mbgl/text/bidi.hpp
 
     # tile
     src/mbgl/tile/geojson_tile.cpp

cmake/core.cmake

@@ -45,6 +45,7 @@ target_add_mason_package(mbgl-core PRIVATE supercluster)
 target_add_mason_package(mbgl-core PRIVATE kdbush)
 target_add_mason_package(mbgl-core PRIVATE earcut)
 target_add_mason_package(mbgl-core PRIVATE protozero)
+target_add_mason_package(mbgl-core PRIVATE icu)
 
 mbgl_platform_core()
 

package.json

@@ -24,7 +24,7 @@
     "lodash": "^4.16.4",
     "mapbox-gl-shaders": "mapbox/mapbox-gl-shaders#597115a1e1bd982944b068f8accde34eada74fc2",
     "mapbox-gl-style-spec": "mapbox/mapbox-gl-style-spec#7f62a4fc9f21e619824d68abbc4b03cbc1685572",
-    "mapbox-gl-test-suite": "mapbox/mapbox-gl-test-suite#87192085b3c1ebe668524511bfba28381e5eb627",
+    "mapbox-gl-test-suite": "mapbox/mapbox-gl-test-suite#c32d0c5ac80e3b7393bc17b8944e64fa5cffd90a",
     "mkdirp": "^0.5.1",
     "node-cmake": "^1.2.1",
     "request": "^2.72.0",

src/mbgl/layout/symbol_layout.cpp

@@ -20,6 +20,8 @@
 #include <mbgl/platform/platform.hpp>
 #include <mbgl/platform/log.hpp>
 
+#include <mbgl/text/bidi.hpp>
+
 namespace mbgl {
 
 using namespace style;
@@ -90,7 +92,7 @@ SymbolLayout::SymbolLayout(std::string bucketName_,
                 u8string = platform::lowercase(u8string);
             }
 
-            ft.text = util::utf8_to_utf16::convert( u8string );
+            ft.text = bidi.bidiTransform(util::utf8_to_utf16::convert(u8string));
 
             // Loop through all characters of this text and collect unique codepoints.
             for (char16_t chr : *ft.text) {

src/mbgl/layout/symbol_layout.hpp

@@ -4,6 +4,7 @@
 #include <mbgl/style/layers/symbol_layer_properties.hpp>
 #include <mbgl/layout/symbol_feature.hpp>
 #include <mbgl/layout/symbol_instance.hpp>
+#include <mbgl/text/bidi.hpp>
 
 #include <memory>
 #include <map>
@@ -91,6 +92,8 @@ private:
     GlyphRangeSet ranges;
     std::vector<SymbolInstance> symbolInstances;
     std::vector<SymbolFeature> features;
+    
+    BiDi bidi; // Consider moving this up to geometry tile worker to reduce reinstantiation costs; use of BiDi/ubiditransform object must be constrained to one thread
 };
 
 } // namespace mbgl

src/mbgl/text/bidi.cpp

@@ -0,0 +1,53 @@
+#include <memory>
+
+#include <mbgl/text/bidi.hpp>
+#include <unicode/ubidi.h>
+#include <unicode/ushape.h>
+
+namespace mbgl {
+
+BiDi::BiDi() {
+    UErrorCode errorCode = U_ZERO_ERROR;
+    transform = ubiditransform_open(&errorCode); // Only error is failure to allocate memory, in
+                                                 // that case ubidi_transform would fall back to
+                                                 // creating transform object on the fly
+}
+
+BiDi::~BiDi() {
+    if (transform)
+        ubiditransform_close(transform);
+}
+
+std::u16string BiDi::bidiTransform(const std::u16string& input) {
+    UErrorCode errorCode = U_ZERO_ERROR;
+
+    std::unique_ptr<UChar[]> outputText =
+        std::make_unique<UChar[]>(input.size() * 2); // Maximum output of ubidi_transform is twice
+                                                     // the size of input according to
+                                                     // ubidi_transform.h
+    uint32_t outputLength = ubiditransform_transform(
+        transform, input.c_str(), static_cast<int32_t>(input.size()), outputText.get(),
+        static_cast<int32_t>(input.size()) * 2,
+        UBIDI_DEFAULT_LTR,     // Assume input is LTR unless strong RTL characters are found
+        UBIDI_LOGICAL,         // Input is in logical order
+        UBIDI_LTR,             // Output is in "visual LTR" order
+        UBIDI_VISUAL,          //  ''
+        UBIDI_MIRRORING_ON,    // Use mirroring lookups for things like parentheses that need mirroring
+                               // in RTL text
+        U_SHAPE_LETTERS_SHAPE, // Add options here for handling numbers in bidirectional text
+        &errorCode);
+
+    // If the algorithm fails for any reason, fall back to non-transformed text
+    if (U_FAILURE(errorCode))
+        return input;
+
+    return std::u16string(outputText.get(), outputLength);
+}
+
+bool BiDi::baseDirectionRightToLeft(const std::u16string& input) {
+    // This just looks for the first character with a strong direction property, it does not perform
+    // the BiDi algorithm
+    return ubidi_getBaseDirection(input.c_str(), static_cast<int32_t>(input.size())) == UBIDI_RTL;
+}
+
+} // end namespace mbgl

src/mbgl/text/bidi.hpp

@@ -0,0 +1,22 @@
+#pragma once
+
+#include <string>
+
+#include <mbgl/util/noncopyable.hpp>
+#include <unicode/ubiditransform.h>
+
+namespace mbgl {
+
+class BiDi : private util::noncopyable {
+public:
+    BiDi();
+    ~BiDi();
+
+    std::u16string bidiTransform(const std::u16string&);
+    bool baseDirectionRightToLeft(const std::u16string&);
+
+private:
+    UBiDiTransform* transform;
+};
+
+} // end namespace mbgl

src/mbgl/util/utf.hpp

@@ -10,12 +10,10 @@ namespace util {
 
 class utf8_to_utf16 {
 public:
-    static std::u16string convert(std::string const& utf8)
-    {
-        std::wstring_convert<std::codecvt_utf8_utf16<char16_t>,char16_t> converter;
-        return converter.from_bytes( utf8 );
+    static std::u16string convert(std::string const& utf8) {
+        std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> converter;
+        return converter.from_bytes(utf8);
     }
-    
 };
 
 } // namespace util