Merge pull request #67 from igor-chernikov/fix/cyrillic-rune-handling

Author: AlexanderGrooff

cmd/draw.go

@@ -48,11 +48,21 @@ func (g *graph) drawEdge(e *edge) (*drawing, *drawing, *drawing, *drawing, *draw
 }
 
 func (d *drawing) drawText(start drawingCoord, text string) {
-	// Increase dimensions if necessary.
-	d.increaseSize(start.x+len(text), start.y)
-	log.Debug("Drawing '", text, "' from ", start, " to ", drawingCoord{x: start.x + len(text), y: start.y})
-	for x := 0; x < len(text); x++ {
-		(*d)[x+start.x][start.y] = string(text[x])
+	// Increase dimensions if necessary. Use the visual width so multibyte
+	// (e.g. Cyrillic, CJK) runes reserve the correct number of cells.
+	textWidth := runewidth.StringWidth(text)
+	d.increaseSize(start.x+textWidth, start.y)
+	log.Debug("Drawing '", text, "' from ", start, " to ", drawingCoord{x: start.x + textWidth, y: start.y})
+	// Iterate over runes (not bytes) so multibyte characters are placed in a
+	// single cell instead of being split into invalid byte fragments.
+	textX := start.x
+	for _, r := range text {
+		runeWidth := Max(runewidth.RuneWidth(r), 1)
+		(*d)[textX][start.y] = string(r)
+		for offset := 1; offset < runeWidth; offset++ {
+			(*d)[textX+offset][start.y] = ""
+		}
+		textX += runeWidth
 	}
 }
 

cmd/graph_test.go

@@ -64,6 +64,25 @@ func TestExtendedChars(t *testing.T) {
 	}
 }
 
+// TestMultibyte verifies that multibyte UTF-8 labels (Cyrillic, Greek, accented
+// Latin, etc.) render correctly without splitting runes into invalid byte
+// fragments. Test cases are loaded from testdata/multibyte/*.txt.
+func TestMultibyte(t *testing.T) {
+	dir := "testdata/multibyte"
+	files, err := os.ReadDir(dir)
+	if err != nil {
+		t.Fatalf("Failed to read directory %s: %v", dir, err)
+	}
+
+	for _, file := range files {
+		if !file.IsDir() && strings.HasSuffix(file.Name(), ".txt") {
+			t.Run(file.Name(), func(t *testing.T) {
+				verifyMap(t, filepath.Join(dir, file.Name()), true)
+			})
+		}
+	}
+}
+
 // TestGraphUseAsciiConfig tests that RenderDiagram respects config.UseAscii for graphs
 func TestGraphUseAsciiConfig(t *testing.T) {
 	mermaidInput := `graph LR

cmd/testdata/multibyte/accented_latin_node_and_edge_labels.txt

@@ -0,0 +1,8 @@
+graph LR
+A[Café]-->|résumé|B[Über]
+---
++------+          +------+
+|      |          |      |
+| Café |résumé--->| Über |
+|      |          |      |
++------+          +------+

cmd/testdata/multibyte/cyrillic_node_and_edge_labels.txt

@@ -0,0 +1,8 @@
+graph LR
+A[Привет]-->|метка|B[Мир]
+---
++--------+            +-----+
+|        |            |     |
+| Привет |метка------>| Мир |
+|        |            |     |
++--------+            +-----+

cmd/testdata/multibyte/greek_node_and_edge_labels.txt

@@ -0,0 +1,8 @@
+graph LR
+A[Γειά]-->|ετικέτα|B[Κόσμος]
+---
++------+                +--------+
+|      |                |        |
+| Γειά |ετικέτα-------->| Κόσμος |
+|      |                |        |
++------+                +--------+