fix sentence rule SB3

grapheme_test.go

@@ -104,7 +104,7 @@ func TestGraphemesClass(t *testing.T) {
 // class.
 func TestGraphemesClassWord(t *testing.T) {
 	for testNum, testCase := range wordBreakTestCases {
-		if testNum == 1700 {
+		if testNum == 1703 {
 			// This test case reveals an inconsistency in the Unicode rule set,
 			// namely the handling of ZWJ within two RI graphemes. (Grapheme
 			// rules will restart the RI count, word rules will ignore the ZWJ.)

sentencerules.go

@@ -20,7 +20,13 @@ const (
 	sbSTerm
 	sbSB8aClose
 	sbSB8aSp
-	sbMax = iota
+	sbMax
+)
+
+const (
+	sbStateMask SentenceBreakState = 0x0F
+	sbSB3Mask   SentenceBreakState = 0x10
+	sbSB3       SentenceBreakState = 0x10
 )
 
 type sbTransitionResult struct {
@@ -133,6 +139,9 @@ func transitionSentenceBreakState[T bytes](state SentenceBreakState, r rune, str
 	// Determine the property of the next character.
 	nextProperty := sentenceBreakCodePoints.search(r)
 
+	sb3state := state & sbSB3Mask
+	state &= sbStateMask
+
 	// SB5 (Replacing Ignore Rules).
 	if nextProperty == sbprExtend || nextProperty == sbprFormat {
 		if state == sbParaSep || state == sbCR {
@@ -176,6 +185,15 @@ func transitionSentenceBreakState[T bytes](state SentenceBreakState, r rune, str
 		}
 	}
 
+	// SB3.
+	if rule > 30 && sb3state != 0 && nextProperty == sbprLF {
+		sentenceBreak = false
+		rule = 300
+	}
+	if nextProperty == sbprCR {
+		newState |= sbSB3
+	}
+
 	// SB8.
 	if rule > 80 && (state == sbATerm || state == sbSB8Close || state == sbSB8Sp || state == sbSB7) {
 		// Check the right side of the rule.

step_test.go

@@ -296,7 +296,7 @@ func TestStepStringGrapheme(t *testing.T) {
 // the StepString() function.
 func TestStepStringWord(t *testing.T) {
 	for testNum, testCase := range wordBreakTestCases {
-		if testNum == 1700 {
+		if testNum == 1703 {
 			// This test case reveals an inconsistency in the Unicode rule set,
 			// namely the handling of ZWJ within two RI graphemes. (Grapheme
 			// rules will restart the RI count, word rules will ignore the ZWJ.)