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

mapbox · Nov 12, 2016 · f1e7a51 · f1e7a51
1 parent 3c5445e
commit f1e7a51
Show file tree

Hide file tree

Showing 3 changed files with 34 additions and 15 deletions.
diff --git a/src/mbgl/text/glyph_set.cpp b/src/mbgl/text/glyph_set.cpp
@@ -128,8 +128,7 @@ void GlyphSet::lineWrap(Shaping &shaping, const float lineHeight, float maxWidth
                     // 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--;
                     }
 
@@ -144,16 +143,7 @@ void GlyphSet::lineWrap(Shaping &shaping, const float lineHeight, float maxWidth
 
             // Ideographic characters, spaces, and word-breaking punctuation that often appear without surrounding spaces.
             if (useBalancedIdeographicBreaking
-                || 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 */
+                || util::i18n::allowsWordBreaking(shape.glyph)
                 || util::i18n::allowsIdeographicBreaking(shape.glyph)) {
                 lastSafeBreak = i;
             }

diff --git a/src/mbgl/util/i18n.cpp b/src/mbgl/util/i18n.cpp
@@ -4,6 +4,26 @@ namespace mbgl {
 namespace util {
 namespace i18n {
 
+bool isVisible(uint32_t chr) {
+    return (chr == 0x0a /* newline */
+            || chr == 0x20 /* space */
+            || chr == 0x200b /* zero-width space */);
+}
+
+bool allowsWordBreaking(uint32_t chr) {
+    return (chr == 0x0a /* newline */
+            || chr == 0x20 /* space */
+            || chr == 0x26 /* ampersand */
+            || chr == 0x2b /* plus sign */
+            || chr == 0x2d /* hyphen-minus */
+            || chr == 0x2f /* solidus */
+            || chr == 0xad /* soft hyphen */
+            || chr == 0xb7 /* middle dot */
+            || chr == 0x200b /* zero-width space */
+            || chr == 0x2010 /* hyphen */
+            || chr == 0x2013 /* en dash */);
+}
+
 bool allowsIdeographicBreaking(const std::u32string& string) {
     for (uint32_t chr : string) {
         if (!allowsIdeographicBreaking(chr)) {

diff --git a/src/mbgl/util/i18n.hpp b/src/mbgl/util/i18n.hpp
@@ -6,11 +6,20 @@ namespace mbgl {
 namespace util {
 namespace i18n {
 
-/// Returns whether a line break can be inserted after any character in the given string.
-/// If false, line breaking should occur on word boundaries instead.
+/** Returns whether a character is a visible character. */
+bool isVisible(uint32_t chr);
+
+/** Returns whether a line break can be inserted after the character indicated
+    by the given Unicode codepoint due to word breaking. */
+bool allowsWordBreaking(uint32_t chr);
+
+/** Returns whether a line break can be inserted after any character in the
+    given string. If false, line breaking should occur on word boundaries
+    instead. */
 bool allowsIdeographicBreaking(const std::u32string& string);
 
-/// Returns whether a line break can be inserted after the character indicated by the given Unicode codepoint.
+/** Returns whether a line break can be inserted after the character indicated
+    by the given Unicode codepoint due to ideographic breaking. */
 bool allowsIdeographicBreaking(uint32_t chr);
 
 } // namespace i18n