vojo/apps/ai-bot/markdown_test.go

package main

import (
	"strings"
	"testing"
)

// TestMarkdownToHTML asserts the rich constructs render and plain text stays
// plain. It checks for the meaningful tags/escaping (Contains), not goldmark's
// exact byte output — the converter's precise formatting is its own contract, not
// ours to pin.
func TestMarkdownToHTML(t *testing.T) {
	rich := []struct {
		name     string
		in       string
		contains []string
	}{
		{"bold", "a **bold** b", []string{"<strong>bold</strong>"}},
		{"italic star", "a *it* b", []string{"<em>it</em>"}},
		{"italic underscore", "a _it_ b", []string{"<em>it</em>"}},
		{"bold italic", "***x***", []string{"<strong>", "<em>", "x"}},
		{"strikethrough", "~~gone~~", []string{"gone"}}, // <del> or <s>; both rich
		{"inline code", "use `npm i`", []string{"<code>npm i</code>"}},
		{"inline code keeps stars literal", "`a*b*c`", []string{"<code>a*b*c</code>"}},
		{"heading h1", "# Title", []string{"<h1>", "Title", "</h1>"}},
		{"hr", "---", []string{"<hr"}},
		{"unordered list", "- one\n- two", []string{"<ul>", "<li>", "one", "two"}},
		{"ordered list", "1. one\n2. two", []string{"<ol>", "<li>", "one"}},
		{"blockquote", "> quoted", []string{"<blockquote>", "quoted"}},
		{"link", "see [xAI](https://x.ai)", []string{`href="https://x.ai"`, "xAI"}},
		{"fenced code", "```go\nfmt.Println()\n```", []string{"<pre>", "<code", "fmt.Println"}},
		{"table", "| a | b |\n| - | - |\n| 1 | 2 |", []string{"<table>", "<th>", "a", "<td>", "1"}},
		{"image as link", "![logo](https://x.ai/logo.png)", []string{`href="https://x.ai/logo.png"`, "logo"}},
		{"autolink bare url", "visit https://x.ai now", []string{`href="https://x.ai"`}},
	}
	for _, c := range rich {
		t.Run("rich/"+c.name, func(t *testing.T) {
			got, formatted := markdownToHTML(c.in)
			if !formatted {
				t.Fatalf("markdownToHTML(%q) formatted=false, want true (got %q)", c.in, got)
			}
			for _, sub := range c.contains {
				if !strings.Contains(got, sub) {
					t.Fatalf("markdownToHTML(%q) = %q, missing %q", c.in, got, sub)
				}
			}
		})
	}

	// Plain text (even multi-line or with stray punctuation) carries no
	// formatting, so the bot sends only the bare body.
	plain := []string{
		"just a sentence",
		"line one\nline two",
		"a < b & c > d",
		"2 * 3 * 4",
		"snake_case_name",
		"файл_имя_тут",
		"text with ! bang",
		`path c:\users`,
		"",
	}
	for _, in := range plain {
		t.Run("plain", func(t *testing.T) {
			if got, formatted := markdownToHTML(in); formatted {
				t.Fatalf("markdownToHTML(%q) formatted=true, want false (got %q)", in, got)
			}
		})
	}
}

func TestMarkdownNeverEmitsUnsafeScheme(t *testing.T) {
	for _, bad := range []string{
		"[a](javascript:x)", "[a](data:text/html,x)", "[a](vbscript:x)", "[a](file:///etc)",
		"[a](JaVaScRiPt:x)", "[a](java\tscript:x)",
	} {
		if html, _ := markdownToHTML(bad); strings.Contains(html, "href=") {
			t.Fatalf("emitted a link for unsafe scheme: %q -> %q", bad, html)
		}
	}
}

func TestMarkdownOversizeFallsBackToPlain(t *testing.T) {
	// A formatted reply whose HTML exceeds maxFormattedBytes must return ("", false)
	// so the bot sends only the plain body.
	big := strings.Repeat("- item\n", 8000)
	if html, formatted := markdownToHTML(big); formatted || html != "" {
		t.Fatalf("oversize formatted output should fall back to plain: formatted=%v len=%d", formatted, len(html))
	}
	// Implausibly large input is rejected outright.
	huge := strings.Repeat("a", maxInputBytes+1)
	if html, formatted := markdownToHTML(huge); formatted || html != "" {
		t.Fatalf("oversize input should fall back to plain: formatted=%v len=%d", formatted, len(html))
	}
}

func TestMarkdownAdversarialNoPanicNoInjection(t *testing.T) {
	inputs := []string{
		strings.Repeat("[", 20000) + "x",
		"x" + strings.Repeat("](https://a)", 20000),
		strings.Repeat("*", 5000) + "x" + strings.Repeat("*", 5000),
		strings.Repeat("> ", 5000) + "deep",
		strings.Repeat("  ", 50) + "- nested",
		strings.Repeat("`", 4000) + "code",
		"| " + strings.Repeat("a |", 2000) + "\n| " + strings.Repeat("- |", 2000) + "\n| x |",
		"<script>alert(1)</script>\n**`<b>`**\n[x](\"><svg onload=alert(1)>)",
		strings.Repeat("***nest ", 200) + "x" + strings.Repeat(" nest***", 200),
	}
	// Every model '<' is escaped to &lt;, so a dangerous element can only exist if
	// the converter emitted it — and it emits none of these tag names. (Attribute
	// vectors like onload= can appear only as escaped literal text, which is safe;
	// the safe-href guarantee is covered by the unit + scheme tests.)
	for i, in := range inputs {
		html, _ := markdownToHTML(in) // must not panic
		for _, tag := range []string{"<script", "<svg", "<img", "<iframe", "<style", "<object", "<embed"} {
			if strings.Contains(strings.ToLower(html), tag) {
				t.Fatalf("case %d emitted a dangerous tag %q: %.160q", i, tag, html)
			}
		}
	}
}

func TestBuildNoticeContentAttachesFormatted(t *testing.T) {
	c := buildNoticeContent("$evt", "@u:vojo.chat", "", "Here is **bold**.")
	if c["format"] != matrixHTMLFormat {
		t.Fatalf("format = %v, want %v", c["format"], matrixHTMLFormat)
	}
	fb, _ := c["formatted_body"].(string)
	if !strings.Contains(fb, "<strong>bold</strong>") {
		t.Fatalf("formatted_body missing bold: %q", fb)
	}
	if c["body"] != "Here is **bold**." {
		t.Fatalf("plain body must be preserved, got %v", c["body"])
	}
}

func TestBuildNoticeContentSkipsFormattedForPlain(t *testing.T) {
	c := buildNoticeContent("$evt", "@u:vojo.chat", "", "no markdown here")
	if _, ok := c["format"]; ok {
		t.Fatalf("format must be absent for plain text")
	}
	if _, ok := c["formatted_body"]; ok {
		t.Fatalf("formatted_body must be absent for plain text")
	}
}

// TestMarkdownNoHangOnBangAndBackslash guards the inline-parser infinite loop: a
// '!' not starting an image, or a backslash not before ASCII punctuation
// (trailing, or before a letter/space/Cyrillic), used to fall through to a
// non-advancing default branch and spin forever — freezing the whole bot under
// the transaction mutex. These must all RETURN; if the bug returns this test
// hangs and `go test` times out instead of passing.
func TestMarkdownNoHangOnBangAndBackslash(t *testing.T) {
	for _, in := range []string{
		"Привет!",
		"Hello! How are you?",
		`path c:\users`,
		`trailing backslash \`,
		`что-то \ или вот это`,
		`\` + "д",
		"!",
		"!!! wow",
		"text with ! bang",
		strings.Repeat("a! ", 2000),
		strings.Repeat(`\`, 2000),
	} {
		_, _ = markdownToHTML(in) // a hang fails via test timeout
	}
}