Fix TextLength to count characters instead of bytes

Use utf8.RuneCountInString so multi-byte characters like é count as one
character. Add a unicode test case to cover this.

Signed-off-by: Émile Ré <emile@getprobo.com>

pkg/prosemirror/node.go

@@ -17,6 +17,7 @@ package prosemirror
 import (
 	"encoding/json"
 	"fmt"
+	"unicode/utf8"
 )
 
 type (
@@ -168,13 +169,12 @@ func (n Node) TableCellAttrs() (TableCellAttrs, error) {
 	return a, nil
 }
 
-// TextLength returns the total length of all text content in the node tree,
+// TextLength returns the total number of characters (Unicode code points)
-// measured in bytes (consistent with Go's len on strings). Only text carried
+// across all text nodes in the tree. Structural markup is excluded.
-// by leaf text nodes is counted; structural markup is excluded.
 func (n Node) TextLength() int {
 	length := 0
 	if n.Text != nil {
-		length += len(*n.Text)
+		length += utf8.RuneCountInString(*n.Text)
 	}
 	for _, child := range n.Content {
 		length += child.TextLength()

pkg/prosemirror/node_test.go

@@ -487,6 +487,17 @@ func TestTextLength(t *testing.T) {
 		},
 	)
 
+	t.Run(
+		"multi-byte unicode characters",
+		func(t *testing.T) {
+			t.Parallel()
+			raw := `{"type":"doc","content":[{"type":"paragraph","content":[{"type":"text","text":"café résumé"}]}]}`
+			doc, err := Parse(raw)
+			require.NoError(t, err)
+			assert.Equal(t, 11, doc.TextLength())
+		},
+	)
+
 	t.Run(
 		"testdata document",
 		func(t *testing.T) {