width_test.go

// ABOUTME: Readable spot checks for charWidth/stringWidth (unicode-width 0.2.0
// ABOUTME: semantics) and the identifier-character predicates.
// Derived from crates/codegen/xai-grok-markdown/src/mermaid.rs in
// https://github.com/xai-org/grok-build at commit
// b189869b7755d2b482969acf6c92da3ecfeffd36 (Apache-2.0). See THIRD_PARTY_NOTICES.md.
// Exhaustive parity with unicode-width 0.2.0 is asserted in width_parity_test.go;
// these cases document the classes the renderer depends on.

package mermaidtext

import "testing"

func TestCharWidthSpotChecks(t *testing.T) {
	cases := []struct {
		r    rune
		want int
	}{
		{'a', 1},
		{'Z', 1},
		{'0', 1},
		{' ', 1},
		{'日', 2},
		{'本', 2},
		{'語', 2},
		{'́', 0},    // combining acute accent
		{'֑', 0},    // Hebrew accent etnahta
		{'ְ', 0},    // Hebrew point sheva (niqqud)
		{'ً', 0},    // Arabic fathatan (haraka)
		{'्', 0},    // Devanagari sign virama
		{'‍', 0},    // zero-width joiner
		{'\x00', 0}, // NUL (the CONT sentinel)
		{'\x1b', 0}, // ESC
		{'\x7f', 0}, // DEL
		{'\t', 0},   // control
		{'៘', 3},    // Khmer sign beyyal
		{'…', 1},
		{'—', 1}, // em dash (ambiguous width -> narrow)
		{'─', 1},
		{'│', 1},
		{'╭', 1},
		{'×', 1},
		{'●', 1},
		{'▼', 1},
		{'«', 1},
		{'»', 1},
	}
	for _, c := range cases {
		if got := charWidth(c.r); got != c.want {
			t.Errorf("charWidth(%q) = %d, want %d", c.r, got, c.want)
		}
	}
}

func TestStringWidthSpotChecks(t *testing.T) {
	cases := []struct {
		s    string
		want int
	}{
		{"", 0},
		{"abc", 3},
		{"日本語ab", 8},
		{"café", 4},  // combining mark contributes 0
		{"a\x00b", 3}, // control chars count 1 at string level (charWidth 0)
		{"\t", 1},
		{"\r\n", 1}, // CRLF clusters to 1
		{"╭─╮", 3},
		{" mermaid: gantt ", 16},
		{"בְרֵא", 3}, // Hebrew word with niqqud
		{"لا", 1},    // Arabic lam-alef ligature
		{"❤️", 2},    // VS16 makes the base wide
		{"0️⃣", 2},   // keycap sequence
		{"\U0001F468‍\U0001F469‍\U0001F467", 2}, // ZWJ family clusters
	}
	for _, c := range cases {
		if got := stringWidth(c.s); got != c.want {
			t.Errorf("stringWidth(%q) = %d, want %d", c.s, got, c.want)
		}
	}
}

func TestIsAlphanumericMatchesRustSemantics(t *testing.T) {
	trues := []rune{'a', 'Z', '0', '9', 'é', '日', 'Ⓐ', '¼'}
	for _, r := range trues {
		if !isAlphanumeric(r) {
			t.Errorf("isAlphanumeric(%q) = false, want true", r)
		}
	}
	falses := []rune{'_', '-', '.', '/', ' ', '*', '<', '\x00'}
	for _, r := range falses {
		if isAlphanumeric(r) {
			t.Errorf("isAlphanumeric(%q) = true, want false", r)
		}
	}
}

func TestIsIDChar(t *testing.T) {
	trues := []rune{'a', 'Z', '0', '_', 'é', '日', 'Ⓐ'}
	for _, r := range trues {
		if !isIDChar(r) {
			t.Errorf("isIDChar(%q) = false, want true", r)
		}
	}
	falses := []rune{'-', '.', '/', ' ', '[', '>', '&'}
	for _, r := range falses {
		if isIDChar(r) {
			t.Errorf("isIDChar(%q) = true, want false", r)
		}
	}
}