summercms/internal/docsite/highlight_test.go

package docsite

import (
	"html"
	"regexp"
	"strings"
	"testing"
)

// spanText matches one tok-* span.
var spanText = regexp.MustCompile(`<span class="(tok-[a-z]+)">([^<]*)</span>`)

// tokens returns the class and unescaped text of every tok-* span.
func tokens(out string) map[string][]string {
	got := map[string][]string{}
	for _, m := range spanText.FindAllStringSubmatch(out, -1) {
		got[m[1]] = append(got[m[1]], html.UnescapeString(m[2]))
	}
	return got
}

func assertToken(t *testing.T, lang, out, class, text string) {
	t.Helper()
	for _, v := range tokens(out)[class] {
		if strings.TrimSpace(v) == text {
			return
		}
	}
	t.Errorf("%s: no %s span for %q in\n%s", lang, class, text, out)
}

// assertNoTags fails when output has markup other than tok-* spans.
func assertNoTags(t *testing.T, lang, out string) {
	t.Helper()
	rest := spanText.ReplaceAllString(out, "")
	if strings.ContainsAny(rest, "<>") {
		t.Errorf("%s: unexpected markup in %q", lang, rest)
	}
	if strings.Contains(out, "style=") {
		t.Errorf("%s: inline style in %q", lang, out)
	}
}

func TestHighlightGo(t *testing.T) {
	src := "package main\n\n// Say says.\nfunc Say() (int, string) {\n\treturn 42, \"a<b & \\\"c\\\"\"\n}\n"
	out := highlight("go", src)
	assertToken(t, "go", out, "tok-kw", "package")
	assertToken(t, "go", out, "tok-kw", "func")
	assertToken(t, "go", out, "tok-kw", "return")
	assertToken(t, "go", out, "tok-com", "// Say says.")
	assertToken(t, "go", out, "tok-num", "42")
	assertToken(t, "go", out, "tok-str", `"a<b & \"c\""`)
	assertNoTags(t, "go", out)
	for _, want := range []string{"&lt;", "&amp;", "&#34;"} {
		if !strings.Contains(out, want) {
			t.Errorf("go output does not escape to %s: %s", want, out)
		}
	}
	if html.UnescapeString(spanText.ReplaceAllString(out, "$2")) != src {
		t.Errorf("go output does not round-trip to the source:\n%s", out)
	}
}

func TestHighlightYAML(t *testing.T) {
	src := "# comment\nname: \"quoted <x>\"\nport: 5432\nenabled: true\nplain: text value\n"
	for _, lang := range []string{"yaml", "yml"} {
		out := highlight(lang, src)
		assertToken(t, lang, out, "tok-com", "# comment")
		assertToken(t, lang, out, "tok-key", "name")
		assertToken(t, lang, out, "tok-str", `"quoted <x>"`)
		assertToken(t, lang, out, "tok-num", "5432")
		assertToken(t, lang, out, "tok-num", "true")
		assertToken(t, lang, out, "tok-str", "text value")
		assertNoTags(t, lang, out)
	}
	out := highlight("json", `{"a": null, "b": 1}`)
	assertToken(t, "json", out, "tok-num", "null")
	assertToken(t, "json", out, "tok-num", "1")
	assertNoTags(t, "json", out)
}

func TestHighlightShell(t *testing.T) {
	src := "$ summer docs:build --out \"site\"  # build\n# a comment line\n  ./bin/acme migrate \\\n    --force\n\nsummer\n"
	for _, lang := range []string{"sh", "shell", "bash", "console"} {
		out := highlight(lang, src)
		assertToken(t, lang, out, "tok-prompt", "$")
		assertToken(t, lang, out, "tok-kw", "summer")
		assertToken(t, lang, out, "tok-kw", "./bin/acme")
		assertToken(t, lang, out, "tok-com", "# a comment line")
		assertToken(t, lang, out, "tok-com", "# build")
		assertToken(t, lang, out, "tok-str", `"site"`)
		assertNoTags(t, lang, out)
		// The continuation line has no command word.
		for _, kw := range tokens(out)["tok-kw"] {
			if strings.Contains(kw, "--force") {
				t.Errorf("%s: continuation line highlighted as a command: %q", lang, kw)
			}
		}
		if !strings.HasPrefix(out, `<span class="tok-prompt">$ </span><span class="tok-kw">summer</span> `) {
			t.Errorf("%s: output starts %q", lang, out[:min(len(out), 80)])
		}
		if !strings.Contains(out, "\n  <span class=\"tok-kw\">./bin/acme</span>") {
			t.Errorf("%s: indentation before a command was not kept:\n%s", lang, out)
		}
		if html.UnescapeString(spanText.ReplaceAllString(out, "$2")) != src {
			t.Errorf("%s: output does not round-trip to the source:\n%s", lang, out)
		}
	}
	// A line fragment without a trailing newline stays one line.
	if out := highlight("sh", "summer serve"); strings.Contains(out, "\n") {
		t.Errorf("sh fragment gained a newline: %q", out)
	}
}

func TestHighlightFallback(t *testing.T) {
	src := "a < b && c > \"d\" 'e'"
	want := html.EscapeString(src)
	for _, lang := range []string{"", "text", "txt", "plain", "plaintext", "no-such-language"} {
		if got := highlight(lang, src); got != want {
			t.Errorf("highlight(%q) = %q, want escaped plain text %q", lang, got, want)
		}
	}
	if lexerFor("go") == nil || lexerFor("text") != nil || lexerFor("no-such-language") != nil {
		t.Error("lexerFor")
	}
	var b strings.Builder
	writeSpan(&b, "tok-kw", "")
	writeSpan(&b, "", "<x>")
	if b.String() != "&lt;x&gt;" {
		t.Errorf("writeSpan = %q", b.String())
	}
}

func TestHighlightFenceMarkup(t *testing.T) {
	s, pages := renderFixture(t, map[string]string{
		"docs/setup/start.md": page("Start", "setup", 10, "```\nno <lang>\n```\n\n```go title=x\nx := 1\n```\n"),
	})
	out := string(renderHTML(t, s, pages["setup/start"]).html)
	for _, want := range []string{
		`<figure class="code"><button type="button" class="copy-button" aria-label="Copy code" hidden><svg class="icon icon-copy"`,
		`<svg class="icon icon-check"`,
		"</button><pre><code>no &lt;lang&gt;\n</code></pre></figure>",
		`<pre><code class="language-go">x := <span class="tok-num">1</span>`,
	} {
		if !strings.Contains(out, want) {
			t.Errorf("missing %s in\n%s", want, out)
		}
	}
	if strings.Contains(out, "<figcaption>") {
		t.Error("a fence without src= has a caption")
	}
}

func TestGoLang(t *testing.T) {
	for _, lang := range []string{"go", "Go", "GO", "golang", "Golang", "main.go"} {
		if !goLang(lang) {
			t.Errorf("goLang(%q) = false, want true", lang)
		}
	}
	for _, lang := range []string{"go-html-template", "go-text-template", "text", "yaml", "sh", ""} {
		if goLang(lang) {
			t.Errorf("goLang(%q) = true, want false", lang)
		}
	}
}