Refactor Unicode General Category lookup

segmenter/unicode14_rules.go

@@ -69,7 +69,7 @@ func (cr *cursor) ruleLB30ab(breakOp *breakOpportunity) {
 		*breakOp = breakProhibited
 	}
 	// [\p{Extended_Pictographic}&\p{Cn}] × EM
-	if unicode.Is(ucd.Extended_Pictographic, cr.prev) && ucd.LookupType(cr.prev) == nil &&
+	if unicode.Is(ucd.Extended_Pictographic, cr.prev) && !ucd.IsAssigned(cr.prev) &&
 		cr.line == ucd.BreakEM {
 		*breakOp = breakProhibited
 	}
@@ -313,8 +313,7 @@ func (cr *cursor) ruleLB1() {
 	case ucd.BreakAI, ucd.BreakSG, ucd.BreakXX:
 		cr.line = ucd.BreakAL
 	case ucd.BreakSA:
-		generalCategory := ucd.LookupType(cr.r)
-		if generalCategory == unicode.Mn || generalCategory == unicode.Mc {
+		if unicode.Is(unicode.Mn, cr.r) || unicode.Is(unicode.Mc, cr.r) {
 			cr.line = ucd.BreakCM
 		} else {
 			cr.line = ucd.BreakAL

unicodedata/unicode.go

@@ -8,25 +8,15 @@ import (
 	"github.com/go-text/typesetting/language"
 )
 
-var categories []*unicode.RangeTable
-
-func init() {
-	for cat, table := range unicode.Categories {
-		if len(cat) == 2 {
-			categories = append(categories, table)
-		}
-	}
-}
-
-// LookupType returns the unicode general categorie of the rune,
-// or nil if not found.
-func LookupType(r rune) *unicode.RangeTable {
-	for _, table := range categories {
+// IsAssigned returns [true] for runes with a general category,
+// other than Cn.
+func IsAssigned(r rune) bool {
+	for _, table := range allCategories {
 		if unicode.Is(table, r) {
-			return table
+			return true
 		}
 	}
-	return nil
+	return false
 }
 
 // LookupCombiningClass returns the class used for the Canonical Ordering Algorithm in the Unicode Standard,

unicodedata/unicode_test.go

@@ -1,7 +1,6 @@
 package unicodedata
 
 import (
-	"reflect"
 	"testing"
 	"unicode"
 
@@ -91,21 +90,23 @@ func TestBreakClass(t *testing.T) {
 	}
 }
 
-func TestLookupType(t *testing.T) {
+func TestIsAssigned(t *testing.T) {
 	// some manual test cases
 	tests := []struct {
 		args rune
-		want *unicode.RangeTable
+		want bool
 	}{
-		{'a', unicode.Ll},
-		{'.', unicode.Po},
-		{'カ', unicode.Lo},
-		{'🦳', unicode.So},
-		{-1, nil},
+		{'a', true},
+		{'.', true},
+		{'カ', true},
+		{'🦳', true},
+		{'\U0001F3FF', true},
+		{'\U0001F02C', false},
+		{-1, false},
 	}
 	for _, tt := range tests {
-		if got := LookupType(tt.args); got != tt.want {
-			t.Errorf("LookupType(%s) = %v, want %v", string(tt.args), got, tt.want)
+		if got := IsAssigned(tt.args); got != tt.want {
+			t.Errorf("IsAssigned(%s) = %v, want %v", string(tt.args), got, tt.want)
 		}
 	}
 }
@@ -508,7 +509,7 @@ func TestLookupLineBreakClass(t *testing.T) {
 
 	}
 	for _, tt := range tests {
-		if got := LookupLineBreakClass(tt.args); !reflect.DeepEqual(got, tt.want) {
+		if got := LookupLineBreakClass(tt.args); got != tt.want {
 			t.Errorf("LookupLineBreakClass(U+%x) = %p, want %p", tt.args, got, tt.want)
 		}
 	}