path: root/vrsite/markdown_test.go

package main

import (
	"net/http/httptest"
	"net/url"
	"strings"
	"testing"
)

func testMarkdown(t *testing.T) *markdown {
	t.Helper()
	return newMarkdown(newHighlighter(loadTheme("", "", "dark"), loadTheme("", "", "light")))
}

func TestMarkdownNotes(t *testing.T) {
	m := testMarkdown(t)
	cases := []struct {
		name string
		in   string
		want []string
		not  []string
	}{
		{
			name: "inline marks",
			in:   "`detect()` prefers **direct env** and only walks the tree as *fallback*",
			want: []string{"<code>detect()</code>", "<strong>direct env</strong>", "<em>fallback</em>"},
		},
		{
			name: "list after a paragraph",
			in:   "three classes:\n\n- persisted RPC narrowed\n- ingress bounds change",
			want: []string{"<p>three classes:</p>", "<ul>", "<li>persisted RPC narrowed</li>"},
		},
		{
			name: "bare url",
			in:   "see https://example.com/x for the rest",
			want: []string{`<a href="https://example.com/x">`},
		},
		{
			name: "single newlines are kept",
			in:   "first line\nsecond line",
			want: []string{"first line<br>", "second line"},
			not:  []string{"<ul>"},
		},
		// the identifiers real notes are full of must survive untouched:
		// underscores are not emphasis, "4.2" is not a list, "@" is not a
		// mail address
		{
			name: "code prose is left alone",
			in:   "4.2 drops stake_minimum_delegation_for_rewards and remove_simple_vote, unlike @",
			want: []string{"stake_minimum_delegation_for_rewards", "remove_simple_vote", "4.2 drops"},
			not:  []string{"<em>", "<ol>", "<a href"},
		},
	}
	for _, c := range cases {
		t.Run(c.name, func(t *testing.T) {
			got := string(m.render(c.in))
			for _, w := range c.want {
				if !strings.Contains(got, w) {
					t.Errorf("missing %q in:\n%s", w, got)
				}
			}
			for _, n := range c.not {
				if strings.Contains(got, n) {
					t.Errorf("unwanted %q in:\n%s", n, got)
				}
			}
		})
	}
}

func TestNoteLinksFindOnlyLocalMarkdownLinks(t *testing.T) {
	p := newNoteLinkParser()
	got := noteLinks(p, strings.Join([]string{
		"[root](/note/2)",
		"[same target](note/2#discussion)",
		"[static relative](../note/3.html?scope=all)",
		"`[code](note/4)` and plain note/5 are not links",
		"[remote](https://example.com/note/6)",
		"[protocol relative](//example.com/note/7)",
		"[reference][n]",
		"[n]: ./note/8",
	}, "\n\n"))
	if len(got) != 3 || got[0] != 2 || got[1] != 3 || got[2] != 8 {
		t.Errorf("note links = %v, want [2 3 8]", got)
	}

	m := testMarkdown(t)
	text := "[note](../note/3.html) and [remote](https://example.com/note/4)"
	if body := string(m.renderAt(text, &page{Live: true, Root: "/"})); !strings.Contains(body, `href="/note/3"`) || !strings.Contains(body, `href="https://example.com/note/4"`) {
		t.Errorf("live note links were not canonicalized selectively:\n%s", body)
	}
	if body := string(m.renderAt(text, &page{Root: "../"})); !strings.Contains(body, `href="../note/3.html"`) {
		t.Errorf("static note link does not follow the page root:\n%s", body)
	}
}

// Angle brackets in a note are code, never markup: they must come through as
// text, and must not stop the markdown around them from being parsed. That
// second half is what goldmark's HTML *block* parser would break — a note
// opening with a quoted tag would render the rest of its paragraph as source.
func TestMarkdownShowsAngleBracketsAsText(t *testing.T) {
	m := testMarkdown(t)
	cases := []struct{ name, in string }{
		{"inline", "sysvar_cache: RwLock<SysvarCache> — hoisted out of the loop"},
		{"block, tag alone on the line", "<div>\nthe **wrapper** in [page.go](https://go.dev)"},
		{"script", "<script>alert(1)</script> is quoted from [the file](https://go.dev)"},
	}
	for _, c := range cases {
		t.Run(c.name, func(t *testing.T) {
			got := string(m.render(c.in))
			if strings.Contains(got, "<div>") || strings.Contains(got, "<script>") {
				t.Errorf("markup reached the page:\n%s", got)
			}
			if strings.Contains(got, "raw HTML omitted") {
				t.Errorf("text was dropped:\n%s", got)
			}
			for _, w := range []string{"&lt;", "&gt;"} {
				if !strings.Contains(got, w) {
					t.Errorf("missing %q in:\n%s", w, got)
				}
			}
			if strings.Contains(c.in, "[") && !strings.Contains(got, `<a href="https://go.dev">`) {
				t.Errorf("markdown stopped being parsed:\n%s", got)
			}
		})
	}
}

// A note is written by whoever ran vr, so the renderer is the last thing
// standing between the log and the page.
func TestMarkdownDropsDangerousLinks(t *testing.T) {
	m := testMarkdown(t)
	for _, in := range []string{
		"[click](javascript:alert(1))",
		"![pic](javascript:alert(1))",
		"[click](vbscript:alert(1))",
	} {
		if got := string(m.render(in)); strings.Contains(got, "javascript:") || strings.Contains(got, "vbscript:") {
			t.Errorf("%q rendered as %s", in, got)
		}
	}
}

// A fenced block is painted by the same grammars as the code view, so a
// snippet in a note wears the reader's theme.
func TestMarkdownHighlightsFences(t *testing.T) {
	m := testMarkdown(t)
	got := string(m.render("```go\nfunc load() error { return nil }\n```"))
	for _, w := range []string{`<pre class="mdcode" data-lang="go">`, `<span class="s-keyword">func</span>`, "</code></pre>"} {
		if !strings.Contains(got, w) {
			t.Errorf("missing %q in:\n%s", w, got)
		}
	}
	// a language no grammar claims still renders, just unpainted
	got = string(m.render("```brainfuck\n+[-]<>\n```"))
	if !strings.Contains(got, "+[-]&lt;&gt;") {
		t.Errorf("unclaimed fence lost its text:\n%s", got)
	}
	if strings.Contains(got, "<span") {
		t.Errorf("unclaimed fence was painted:\n%s", got)
	}
}

// Fence languages are mostly extensions already, so the grammar table does the
// lookup; fenceAliases only has to cover the spelled-out names. Each pair here
// is a snippet the claimed grammar must find something to paint in.
func TestMarkdownFenceLanguages(t *testing.T) {
	hl := newHighlighter(loadTheme("", "", "dark"), loadTheme("", "", "light"))
	for _, c := range []struct{ lang, src string }{
		{"go", "func f() {}"}, {"GO", "func f() {}"}, {"golang", "func f() {}"},
		{"rs", "fn f() {}"}, {"rust", "fn f() {}"},
		{"py", "def f(): pass"}, {"python", "def f(): pass"}, {"python3", "def f(): pass"},
		{"js", "function f() {}"}, {"javascript", "function f() {}"}, {"node", "function f() {}"},
		{"jsx", "function f() {}"}, {"mjs", "function f() {}"},
		{"ts", "function f(): void {}"}, {"typescript", "function f(): void {}"},
		{"tsx", "function f(): void {}"},
		{"c", "int f(void) { return 0; }"},
		{"cpp", "int f() { return 0; }"}, {"c++", "int f() { return 0; }"}, {"h", "int f();"},
		{"json", `{"a": 1}`}, {"jsonc", `{"a": 1}`}, {"json5", `{"a": 1}`},
		{"sh", "for x in a; do echo $x; done"}, {"bash", "for x in a; do echo $x; done"},
		{"shell", "for x in a; do echo $x; done"}, {"zsh", "for x in a; do echo $x; done"},
		{"console", "for x in a; do echo $x; done"},
	} {
		if !painted(hl.classifyLang(c.lang, []byte(c.src))) {
			t.Errorf("fence language %q painted nothing in %q", c.lang, c.src)
		}
	}
	for _, lang := range []string{"", "text", "brainfuck", "  ", "."} {
		src := []byte("func f() {}")
		if got := hl.classifyLang(lang, src); len(got) != len(src) || painted(got) {
			t.Errorf("classifyLang(%q) claimed a grammar it should not have", lang)
		}
	}
}

func painted(classes []int16) bool {
	for _, c := range classes {
		if c != 0 {
			return true
		}
	}
	return false
}

// The note card is the one place all of this has to arrive: the notes page,
// the aside on a file, and the conversation view all render it.
func TestNotePageRendersMarkdown(t *testing.T) {
	sv, _, _ := testServer(t)
	h := sv.handler()

	form := url.Values{"file": {"hello.txt"}, "text": {
		"`cover()` widens the **mask**\n\n- one\n- two\n\n```go\nfunc f() {}\n```"}, "rev": {"@"}}
	r := httptest.NewRequest("POST", "/notes", strings.NewReader(form.Encode()))
	r.Header.Set("Content-Type", "application/x-www-form-urlencoded")
	r.Header.Set("HX-Request", "true")
	w := httptest.NewRecorder()
	h.ServeHTTP(w, r)
	if w.Code != 200 {
		t.Fatalf("POST /notes = %d: %s", w.Code, w.Body)
	}
	for _, want := range []string{
		`<div class="body">`,
		"<code>cover()</code>", "<strong>mask</strong>",
		"<li>one</li>",
		`<pre class="mdcode" data-lang="go">`, `<span class="s-keyword">func</span>`,
	} {
		if !strings.Contains(w.Body.String(), want) {
			t.Errorf("missing %q in the posted note card:\n%s", want, w.Body)
		}
	}
	// and again off the page itself, not just the htmx fragment
	if b := get(t, h, "/notes", false).Body.String(); !strings.Contains(b, "<code>cover()</code>") {
		t.Error("notes page did not render the note as markdown")
	}
}

func TestMarkdownEmpty(t *testing.T) {
	m := testMarkdown(t)
	for _, in := range []string{"", "   ", "\n\n"} {
		if got := m.render(in); got != "" {
			t.Errorf("render(%q) = %q, want empty", in, got)
		}
	}
}