[core] Line-break ideographic text by character (#6828)

* [core] Line-break ideographic text by character

Allow a line break to be inserted after any supported Chinese, Japanese, or Yi character in a point-placed label. Balance the lines unless non-ideographic text such as Latin letters are present.

Fixes #1223.

* [core] Moved more character classing into util::i18n

* [core] Detect character properties by Unicode block

* [test] Reenabled ideographic breaking tests

cmake/core-files.cmake

@@ -439,6 +439,8 @@ set(MBGL_CORE_FILES
     src/mbgl/util/http_header.hpp
     src/mbgl/util/http_timeout.cpp
     src/mbgl/util/http_timeout.hpp
+    src/mbgl/util/i18n.cpp
+    src/mbgl/util/i18n.hpp
     src/mbgl/util/interpolate.hpp
     src/mbgl/util/intersection_tests.cpp
     src/mbgl/util/intersection_tests.hpp

package.json

@@ -23,7 +23,7 @@
     "lodash": "^4.16.4",
     "mapbox-gl-shaders": "mapbox/mapbox-gl-shaders#ec891ce5360e488d81f60991f95d2038b83c4e3c",
     "mapbox-gl-style-spec": "mapbox/mapbox-gl-style-spec#7f62a4fc9f21e619824d68abbc4b03cbc1685572",
-    "mapbox-gl-test-suite": "mapbox/mapbox-gl-test-suite#af9ee275f19e81f839a2733e6906c3fac272620e",
+    "mapbox-gl-test-suite": "mapbox/mapbox-gl-test-suite#2f6453e17cd3d17ab2ff677056f65b9cab70f2e8",
     "mkdirp": "^0.5.1",
     "node-cmake": "^1.2.1",
     "request": "^2.72.0",

platform/ios/CHANGELOG.md

@@ -21,6 +21,7 @@ Mapbox welcomes participation and contributions from everyone. Please read [CONT
 * TileJSON manifests can now specify `"scheme": "tms"` to indicate the use of [TMS](https://en.wikipedia.org/wiki/Tile_Map_Service) coordinates. ([#2270](https://github.com/mapbox/mapbox-gl-native/pull/2270))
 * Fixed an issue causing abstract MGLMultiPointFeature objects to be returned in feature query results. Now concrete MGLPointCollectionFeature objects are returned. ([#6742](https://github.com/mapbox/mapbox-gl-native/pull/6742))
 * Fixed rendering artifacts and missing glyphs that occurred after viewing a large number of CJK characters on the map. ([#5908](https://github.com/mapbox/mapbox-gl-native/pull/5908))
+* Improved the line wrapping behavior of point-placed labels written in Chinese, Japanese, and Yi. ([#6828](https://github.com/mapbox/mapbox-gl-native/pull/6828))
 * `-[MGLMapView resetPosition]` now resets to the current style’s default center coordinates, zoom level, direction, and pitch, if specified. ([#6127](https://github.com/mapbox/mapbox-gl-native/pull/6127))
 * Fixed an issue where feature querying sometimes failed to return the expected features when the map was tilted. ([#6773](https://github.com/mapbox/mapbox-gl-native/pull/6773))
 * MGLFeature’s `attributes` and `identifier` properties are now writable. ([#6728](https://github.com/mapbox/mapbox-gl-native/pull/6728))

platform/macos/CHANGELOG.md

@@ -17,6 +17,7 @@
 * TileJSON manifests can now specify `"scheme": "tms"` to indicate the use of [TMS](https://en.wikipedia.org/wiki/Tile_Map_Service) coordinates. ([#2270](https://github.com/mapbox/mapbox-gl-native/pull/2270))
 * Fixed an issue causing abstract `MGLMultiPointFeature` objects to be returned in feature query results. Now concrete `MGLPointCollectionFeature` objects are returned. ([#6742](https://github.com/mapbox/mapbox-gl-native/pull/6742))
 * Fixed rendering artifacts and missing glyphs that occurred after viewing a large number of CJK characters on the map. ([#5908](https://github.com/mapbox/mapbox-gl-native/pull/5908))
+* Improved the line wrapping behavior of point-placed labels written in Chinese, Japanese, and Yi. ([#6828](https://github.com/mapbox/mapbox-gl-native/pull/6828))
 * Fixed an issue where the style zoom levels were not respected when deciding when to render a layer. ([#5811](https://github.com/mapbox/mapbox-gl-native/issues/5811))
 * Fixed an issue where feature querying sometimes failed to return the expected features when the map was tilted. ([#6773](https://github.com/mapbox/mapbox-gl-native/pull/6773))
 * MGLFeature’s `attributes` and `identifier` properties are now writable. ([#6728](https://github.com/mapbox/mapbox-gl-native/pull/6728))

src/mbgl/text/glyph_set.cpp

@@ -1,6 +1,7 @@
 #include <mbgl/text/glyph_set.hpp>
 #include <mbgl/platform/log.hpp>
 #include <mbgl/math/minmax.hpp>
+#include <mbgl/util/i18n.hpp>
 
 #include <cassert>
 
@@ -54,7 +55,8 @@ const Shaping GlyphSet::getShaping(const std::u32string &string, const float max
     if (shaping.positionedGlyphs.empty())
         return shaping;
 
-    lineWrap(shaping, lineHeight, maxWidth, horizontalAlign, verticalAlign, justify, translate);
+    lineWrap(shaping, lineHeight, maxWidth, horizontalAlign, verticalAlign, justify, translate,
+             util::i18n::allowsIdeographicBreaking(string));
 
     return shaping;
 }
@@ -85,9 +87,10 @@ void justifyLine(std::vector<PositionedGlyph> &positionedGlyphs, const std::map<
     }
 }
 
-void GlyphSet::lineWrap(Shaping &shaping, const float lineHeight, const float maxWidth,
-                         const float horizontalAlign, const float verticalAlign,
-                         const float justify, const Point<float> &translate) const {
+void GlyphSet::lineWrap(Shaping &shaping, const float lineHeight, float maxWidth,
+                        const float horizontalAlign, const float verticalAlign,
+                        const float justify, const Point<float> &translate,
+                        bool useBalancedIdeographicBreaking) const {
     uint32_t lastSafeBreak = 0;
 
     uint32_t lengthBeforeCurrentLine = 0;
@@ -99,6 +102,12 @@ void GlyphSet::lineWrap(Shaping &shaping, const float lineHeight, const float ma
     std::vector<PositionedGlyph> &positionedGlyphs = shaping.positionedGlyphs;
 
     if (maxWidth) {
+        if (useBalancedIdeographicBreaking) {
+            auto lastPositionedGlyph = positionedGlyphs[positionedGlyphs.size() - 1];
+            uint32_t estimatedLineCount = std::fmax(1, std::ceil(lastPositionedGlyph.x / maxWidth));
+            maxWidth = lastPositionedGlyph.x / estimatedLineCount;
+        }
+
         for (uint32_t i = 0; i < positionedGlyphs.size(); i++) {
             PositionedGlyph &shape = positionedGlyphs[i];
 
@@ -119,8 +128,7 @@ void GlyphSet::lineWrap(Shaping &shaping, const float lineHeight, const float ma
                     // Collapse invisible characters.
                     uint32_t breakGlyph = positionedGlyphs[lastSafeBreak].glyph;
                     uint32_t lineEnd = lastSafeBreak;
-                    if (breakGlyph == 0x20 /* space */
-                        || breakGlyph == 0x200b /* zero-width space */) {
+                    if (util::i18n::isVisible(breakGlyph)) {
                         lineEnd--;
                     }
 
@@ -133,17 +141,10 @@ void GlyphSet::lineWrap(Shaping &shaping, const float lineHeight, const float ma
                 line++;
             }
 
-            // Spaces, plus word-breaking punctuation that often appears without surrounding spaces.
-            if (shape.glyph == 0x20 /* space */
-                || shape.glyph == 0x26 /* ampersand */
-                || shape.glyph == 0x2b /* plus sign */
-                || shape.glyph == 0x2d /* hyphen-minus */
-                || shape.glyph == 0x2f /* solidus */
-                || shape.glyph == 0xad /* soft hyphen */
-                || shape.glyph == 0xb7 /* middle dot */
-                || shape.glyph == 0x200b /* zero-width space */
-                || shape.glyph == 0x2010 /* hyphen */
-                || shape.glyph == 0x2013 /* en dash */) {
+            // Ideographic characters, spaces, and word-breaking punctuation that often appear without surrounding spaces.
+            if (useBalancedIdeographicBreaking
+                || util::i18n::allowsWordBreaking(shape.glyph)
+                || util::i18n::allowsIdeographicBreaking(shape.glyph)) {
                 lastSafeBreak = i;
             }
         }

src/mbgl/text/glyph_set.hpp

@@ -13,7 +13,8 @@ class GlyphSet {
                              float horizontalAlign, float verticalAlign, float justify,
                              float spacing, const Point<float> &translate) const;
     void lineWrap(Shaping &shaping, float lineHeight, float maxWidth, float horizontalAlign,
-                  float verticalAlign, float justify, const Point<float> &translate) const;
+                  float verticalAlign, float justify, const Point<float> &translate,
+                  bool useBalancedIdeographicBreaking) const;
 
 private:
     std::map<uint32_t, SDFGlyph> sdfs;