Add support for spacing combining marks.

A spacing combining mark is a character that combines (somehow) with a preceding base character. For example, TAMIL VOWEL SIGN E. They might combine in funny ways, like to the left of the base character. This commit does more or less what Terminal does, which is to give such a character its own cell but draw it in conjunction with the preceding character, provided the predecessor is drawable non-ascii. It is guarded by the advanced pref 'Detect base unicode characters with lookup table' which is a bit of a misnomer. I am not courageous enough to turn this on for 3.3, but perhaps a later dot release can have it by default.

Author: gnachman

iTerm2XCTests/VT100GridTest.m

@@ -2542,37 +2542,6 @@ - (void)testCoordinateBefore {
     XCTAssert(coord.x == 1);
     XCTAssert(coord.y == 0);
 }
-
-- (void)testAddCombiningCharToCoord {
-    const unichar kCombiningAcuteAccent = 0x301;
-    const unichar kCombiningCedilla = 0x327;
-    const unichar kCombiningEnclosingCircle = 0x20dd;
-
-    VT100Grid *grid = [self gridFromCompactLines:@"abcd"];
-    XCTAssert([grid addCombiningChar:kCombiningEnclosingCircle
-                          toCoord:VT100GridCoordMake(0, 0)]);
-    screen_char_t *line = [grid screenCharsAtLineNumber:0];
-    XCTAssert(line[0].complexChar);
-    NSString *str = ScreenCharToStr(&line[0]);
-    XCTAssert([[str decomposedStringWithCanonicalMapping] isEqualToString:[@"a⃝" decomposedStringWithCanonicalMapping]]);
-
-    // Fail to modify null character
-    grid = [self gridFromCompactLines:@".bcd"];
-    XCTAssert(![grid addCombiningChar:kCombiningAcuteAccent
-                           toCoord:VT100GridCoordMake(0, 0)]);
-
-    // Add two combining marks
-    grid = [self gridFromCompactLines:@"abcd"];
-    XCTAssert([grid addCombiningChar:kCombiningAcuteAccent
-                          toCoord:VT100GridCoordMake(0, 0)]);
-    XCTAssert([grid addCombiningChar:kCombiningCedilla
-                          toCoord:VT100GridCoordMake(0, 0)]);
-    line = [grid screenCharsAtLineNumber:0];
-    XCTAssert(line[0].complexChar);
-    str = ScreenCharToStr(&line[0]);
-    XCTAssert([[str decomposedStringWithCanonicalMapping] isEqualToString:[@"á̧" decomposedStringWithCanonicalMapping]]);
-}
-
 - (void)testDeleteChars {
     // Base case
     VT100Grid *grid = [self gridFromCompactLinesWithContinuationMarks:

iTerm2XCTests/iTermAutomaticProfileSwitcherTest.m

@@ -140,7 +140,7 @@ - (void)testSwitchesOnUserName {
     _profile = self.profileUserX;
     _allProfiles = @[ self.profileUserX, self.profileUserY ];
     [_aps setHostname:@"whatever" username:@"y" path:@"whatever" job:@"whatever"];
-    XCTAssert([_profile isEqualToProfile:self.profileUserY]);CSSMERR_CSP_INVALID_KEY_FORMAT
+    XCTAssert([_profile isEqualToProfile:self.profileUserY]);
 }
 
 - (void)testSwitchesOnPath {

sources/Metal/Infrastructure/GlyphKey.h

@@ -34,6 +34,7 @@ namespace iTerm2 {
 
         inline bool operator==(const GlyphKey &other) const {
             return (_repr.code == other._repr.code &&
+                    _repr.combiningSuccessor == other._repr.combiningSuccessor &&
                     _repr.isComplex == other._repr.isComplex &&
                     _repr.boxDrawing == other._repr.boxDrawing &&
                     _repr.thinStrokes == other._repr.thinStrokes &&
@@ -45,8 +46,9 @@ namespace iTerm2 {
         }
 
         NSString *description() const {
-            return [NSString stringWithFormat:@"[GlyphKey: code=%@ complex=%@ boxdrawing=%@ thinstrokes=%@ drawable=%@ typeface=%@]",
+            return [NSString stringWithFormat:@"[GlyphKey: code=%@ combiningSuccessor=%@ complex=%@ boxdrawing=%@ thinstrokes=%@ drawable=%@ typeface=%@]",
                     @(_repr.code),
+                    @(_repr.combiningSuccessor),
                     @(_repr.isComplex),
                     @(_repr.boxDrawing),
                     @(_repr.thinStrokes),
@@ -59,6 +61,7 @@ namespace iTerm2 {
             std::size_t seed = 0;
 
             hash_combine(seed, _repr.code);
+            hash_combine(seed, _repr.combiningSuccessor);
             hash_combine(seed, _repr.isComplex);
             hash_combine(seed, _repr.boxDrawing);
             hash_combine(seed, _repr.thinStrokes);

sources/Metal/iTermMetalDriver.m

@@ -867,6 +867,8 @@ - (void)updateTextRendererForFrameData:(iTermMetalFrameData *)frameData {
     CGSize glyphSize = frameData.glyphSize;
     CGFloat scale = frameData.scale;
     __weak iTermMetalFrameData *weakFrameData = frameData;
+
+    // Set up the ASCII fast path
     [_textRenderer setASCIICellSize:cellSize
                              offset:frameData.asciiOffset
                          descriptor:[frameData.perFrameState characterSourceDescriptorForASCIIWithGlyphSize:glyphSize
@@ -876,32 +878,34 @@ - (void)updateTextRendererForFrameData:(iTermMetalFrameData *)frameData {
                                __typeof(self) strongSelf = weakSelf;
                                iTermMetalFrameData *strongFrameData = weakFrameData;
                                if (strongSelf && strongFrameData) {
-                                   return [strongSelf dictionaryForCharacter:c
-                                                              withAttributes:attributes
-                                                                   frameData:strongFrameData
-                                                                   glyphSize:glyphSize
-                                                                       scale:scale];
+                                   return [strongSelf dictionaryForASCIICharacter:c
+                                                                   withAttributes:attributes
+                                                                        frameData:strongFrameData
+                                                                        glyphSize:glyphSize
+                                                                            scale:scale];
                                } else {
                                    return nil;
                                }
                            }];
 }
 
-- (NSDictionary<NSNumber *, iTermCharacterBitmap *> *)dictionaryForCharacter:(char)c
-                                                              withAttributes:(iTermASCIITextureAttributes)attributes
-                                                                   frameData:(iTermMetalFrameData *)frameData
-                                                                   glyphSize:(CGSize)glyphSize
-                                                                       scale:(CGFloat)scale {
+- (NSDictionary<NSNumber *, iTermCharacterBitmap *> *)dictionaryForASCIICharacter:(char)c
+                                                                   withAttributes:(iTermASCIITextureAttributes)attributes
+                                                                        frameData:(iTermMetalFrameData *)frameData
+                                                                        glyphSize:(CGSize)glyphSize
+                                                                            scale:(CGFloat)scale {
     static const int typefaceMask = ((1 << iTermMetalGlyphKeyTypefaceNumberOfBitsNeeded) - 1);
     iTermMetalGlyphKey glyphKey = {
         .code = c,
+        .combiningSuccessor = 0,
         .isComplex = NO,
         .boxDrawing = NO,
         .thinStrokes = !!(attributes & iTermASCIITextureAttributesThinStrokes),
         .drawable = YES,
         .typeface = (attributes & typefaceMask),
     };
     BOOL emoji = NO;
+    // Don't need to pass predecessor or successor because ASCII never has combining spacing marks.
     return [frameData.perFrameState metalImagesForGlyphKey:&glyphKey
                                                asciiOffset:frameData.asciiOffset
                                                       size:glyphSize

sources/NSCharacterSet+iTerm.h

@@ -21,7 +21,9 @@
 // Zero-width spaces.
 + (instancetype)zeroWidthSpaceCharacterSetForUnicodeVersion:(NSInteger)version;
 
-+ (instancetype)baseCharactersForUnicodeVersion:(NSInteger)version;
++ (instancetype)spacingCombiningMarksForUnicodeVersion:(int)version;
+
++ (instancetype)codePointsWithOwnCell;
 
 + (NSCharacterSet *)urlCharacterSet;
 + (NSCharacterSet *)filenameCharacterSet;