Fix char count with single regional indicator

CHANGELOG.md

@@ -40,6 +40,7 @@ develop
 - Fix an issue where text view drawing was distorted while resizing window.
 - Fix an issue where line endings of a document that has a line ending chacter at the beginning of the file cannot be interpreted its line ending type correctly.
 - Fix an issue where character inspector returned always `U+000A` (LF) for line ending even the actual line ending of the document is not LF.
+- Fix character count with a single regional indicator symbol.
 - Fix wrong undo action name on encoding conversion via script.
 
 

CotEditor/Sources/NSString+ComposedCharacter.m

@@ -42,26 +42,27 @@
     
     // count composed chars
     __block NSUInteger count = 0;
-    __block BOOL isRegionalIndicator = NO;
+    __block BOOL isLastCharRegionalIndicator = NO;
     NSRange regionalIndicatorRange = NSMakeRange(0xDDE6, 0xDDFF - 0xDDE6 + 1);
     [string enumerateSubstringsInRange:NSMakeRange(0, [string length])
                                options:NSStringEnumerationByComposedCharacterSequences | NSStringEnumerationSubstringNotRequired
                             usingBlock:^(NSString *substring, NSRange substringRange, NSRange enclosingRange, BOOL *stop)
      {
-         // skip if the last composed character was a regional indicator surrogate-pair
-         // 'Cause the so-called national flag emojis consist of two such surrogate pairs
-         // and the first one is already counted in the last loop.
-         // (To simplify the process, we don't check whether this character is also a regional indicator.)
-         if (isRegionalIndicator) {
-             isRegionalIndicator = NO;
-             return;
-         }
+         // detect regional indicator surrogate pair.
+         BOOL isRegionalIndicator = ((substringRange.length == 2) &&
+                                     NSLocationInRange([string characterAtIndex:substringRange.location], regionalIndicatorRange) &&
+                                     NSLocationInRange([string characterAtIndex:substringRange.location + 1], regionalIndicatorRange));
          
-         // detect regional surrogate pair.
-         if ((substringRange.length == 2) &&
-             (NSLocationInRange([string characterAtIndex:substringRange.location + 1], regionalIndicatorRange)))
-         {
-             isRegionalIndicator = YES;
+         // skip if the last composed character was a regional indicator surrogate-pair
+         // -> 'Cause the so-called national flag emojis consist of two such surrogate pairs
+         //    and the first one is already counted in the last loop.
+         if (isLastCharRegionalIndicator) {
+             isLastCharRegionalIndicator = NO;
+             if (isRegionalIndicator) {
+                 return;
+             }
+         } else if (isRegionalIndicator) {
+             isLastCharRegionalIndicator = YES;
          }
          
          count++;

Tests/StringExtensionsTests.swift

@@ -30,19 +30,12 @@ import XCTest
 
 class StringExtensionsTests: XCTestCase {
     
-    override func setUp() {
-        super.setUp()
-    }
-    
-    
-    override func tearDown() {
-        super.tearDown()
-    }
-    
-    
     func testComposedCharactersCount() {
         XCTAssertEqual("foo".numberOfComposedCharacters(), 3)
-        XCTAssertEqual("😀🇯🇵".numberOfComposedCharacters(), 2)
+        XCTAssertEqual("😀🇯🇵a".numberOfComposedCharacters(), 3)
+        
+        // single regional indicator
+        XCTAssertEqual("🇦 ".numberOfComposedCharacters(), 2)
     }
     
     

Tests/TestFiles/characters.md

@@ -26,6 +26,8 @@ combining character:
 
 national indicators:
 	🇯🇵 U+1F1EF (U+D83C U+DDEF)  U+1F1F5 (U+D83C U+DDF5)
+	🇯 U+1F1EF (U+D83C U+DDEF)
+	🇦🇦 U+1F1E6 (U+D83C U+DDE6)  U+1F1E6 (U+D83C U+DDE6)
 
 sound mark:
 	が U+304C