[ion/unicode] StringGlyphLength method

ion/Makefile

@@ -34,8 +34,9 @@ tests += $(addprefix ion/test/,\
   keyboard.cpp\
   storage.cpp\
   utf8_decoder.cpp\
+  utf8_helper.cpp\
 )
 
 ifdef ION_STORAGE_LOG
 SFLAGS += -DION_STORAGE_LOG=1
-endif
+endif

ion/include/ion/unicode/utf8_helper.h

@@ -68,6 +68,10 @@ int RemovePreviousCodePoint(const char * text, char * location, CodePoint * c);
 const char * CodePointAtGlyphOffset(const char * buffer, int position);
 size_t GlyphOffsetAtCodePoint(const char * buffer, const char * position);
 
+/* Return the number of glyphs in a string.
+ * For instance, strlen("∑") = 3 but StringGlyphLength("∑") = 1 */
+size_t StringGlyphLength(const char * s, int maxSize = -1);
+
 };
 
 #endif

ion/src/shared/unicode/utf8_helper.cpp

@@ -326,5 +326,20 @@ size_t GlyphOffsetAtCodePoint(const char * buffer, const char * position) {
   return glyphIndex;
 }
 
+size_t StringGlyphLength(const char * s, int maxSize) {
+  if (maxSize == 0) {
+    return 0;
+  }
+  UTF8Decoder decoder(s);
+  CodePoint codePoint = decoder.nextCodePoint();
+  size_t glyphIndex = 0;
+  while (codePoint != UCodePointNull && (maxSize < 0 || ((decoder.stringPosition() - s) <= maxSize))) {
+    if (!codePoint.isCombining()) {
+      glyphIndex++;
+    }
+    codePoint = decoder.nextCodePoint();
+  }
+  return glyphIndex;
+}
 
-};
+}

ion/test/utf8_helper.cpp

@@ -0,0 +1,14 @@
+#include <quiz.h>
+#include <ion/unicode/utf8_helper.h>
+
+void assert_string_glyph_length_is(const char * string, int maxSize, size_t result) {
+  quiz_assert(UTF8Helper::StringGlyphLength(string, maxSize) == result);
+}
+
+QUIZ_CASE(ion_utf8_helper_glyph_length) {
+  assert_string_glyph_length_is("123", -1, 3);
+  assert_string_glyph_length_is("1ᴇ3", -1, 3);
+  assert_string_glyph_length_is("∑∫𝐢", -1, 3);
+  assert_string_glyph_length_is("123", 2, 2);
+  assert_string_glyph_length_is("1ᴇ3", 2, 1);
+}