Add a test suite for the shipped extensions

1 file changed, 248 insertions, 0 deletions

diff --git a/tests/extensions/test-about-render.lua b/tests/extensions/test-about-render.lua
new file mode 100644
index 0000000..aa7e6ad
--- /dev/null
+++ b/tests/extensions/test-about-render.lua
@@ -0,0 +1,248 @@
+-- Unit checks for custom/extensions/about-render.lua, run under a standalone
+-- Lua by t0501-about-render.sh with the script path as the first argument.
+-- The second argument may be nolpeg, which makes the lpeg module fail to
+-- load so the escaped plain text fallback is proven even on a machine where
+-- lpeg is installed. Without it the markdown and man checks run when lpeg is
+-- present and are skipped with a note when it is not, the same way the
+-- filter itself degrades.
+
+local test_directory = arg[0]:match("^(.*)/") or "."
+local h = dofile(test_directory .. "/harness.lua")
+
+local script = arg[1]
+local mode = arg[2] or ""
+
+local has_lpeg = false
+if mode == "nolpeg" then
+	h.preload_failure("lpeg")
+else
+	has_lpeg = pcall(require, "lpeg")
+	if not has_lpeg then
+		io.write("lpeg not installed, markdown and man checks skipped\n")
+	end
+end
+
+h.load(script)
+
+local function render(filename, text)
+	return h.run({ filename }, { text })
+end
+
+-- Dispatch and the plain text rendering need nothing but the script itself,
+-- so they run whatever became of lpeg.
+
+local out, ret = render("README", "plain <text> & more")
+h.contains("no extension renders as plain text", out,
+	"<pre class='plaintext'>plain &lt;text&gt; &amp; more</pre>")
+h.equals("close answers zero", ret, 0)
+
+out = render("notes.txt", "just text")
+h.contains("an unknown extension renders as plain text", out,
+	"<pre class='plaintext'>just text</pre>")
+
+out = render(nil, "no name at all")
+h.contains("a missing filename renders as plain text", out,
+	"<pre class='plaintext'>no name at all</pre>")
+
+-- cgit's C sink stops at the first NUL byte, which the script documents, so
+-- the text past one is lost while the wrapper written separately survives.
+out = render("README", "before\0after")
+h.contains("text before a nul byte survives", out, "before")
+h.excludes("text after a nul byte is dropped", out, "after")
+h.contains("the wrapper written after the text survives", out, "</pre>")
+
+out = h.run({ "README" }, { "two ", "writes" })
+h.contains("writes are joined before rendering", out, "two writes")
+
+if mode == "nolpeg" then
+	out = render("README.md", "# Title")
+	h.contains("markdown without lpeg falls back to plain text", out,
+		"<pre class='plaintext'># Title</pre>")
+	out = render("page.1", ".SH NAME")
+	h.contains("man without lpeg falls back to plain text", out,
+		"<pre class='plaintext'>.SH NAME</pre>")
+	h.finish()
+end
+
+if not has_lpeg then
+	h.finish()
+end
+
+-- Markdown.
+
+out = render("README.md", "# Title")
+h.contains("a heading renders inside the wrapper",
+	out, "<div class='markdown'><h1>Title</h1></div>")
+
+out = render("README.md", "## Sub ##")
+h.contains("trailing hashes are stripped from a heading", out,
+	"<h2>Sub</h2>")
+
+out = render("README.MD", "# Upper")
+h.contains("the extension matches whatever its case", out, "<h1>Upper</h1>")
+
+out = render("readme.markdown", "# Long")
+h.contains("the long markdown extension dispatches too", out,
+	"<h1>Long</h1>")
+
+out = render("README.md", "line one\nline two")
+h.contains("a paragraph keeps its line break", out,
+	"<p>line one<br>line two</p>")
+
+out = render("README.md", "first para\r\nsecond line")
+h.contains("crlf line endings are normalised", out,
+	"<p>first para<br>second line</p>")
+
+out = render("README.md", "a **bold** and *leaning* word")
+h.contains("double stars embolden", out, "<strong>bold</strong>")
+h.contains("single stars lean", out, "<em>leaning</em>")
+
+out = render("README.md", "an __up__ and _down_ word")
+h.contains("double underscores embolden", out, "<strong>up</strong>")
+h.contains("single underscores lean", out, "<em>down</em>")
+
+out = render("README.md", "run `x < y` here")
+h.contains("inline code is escaped", out, "<code>x &lt; y</code>")
+
+out = render("README.md", "raw <script>alert(1)</script> here")
+h.contains("markup in text reaches the page escaped", out,
+	"&lt;script&gt;alert(1)&lt;/script&gt;")
+h.excludes("no live tag slips through", out, "<script>")
+
+out = render("README.md", "[docs](https://example.com/docs)")
+h.contains("an https link renders as an anchor", out,
+	"<a href='https://example.com/docs'>docs</a>")
+
+out = render("README.md", "[docs](docs/page.html)")
+h.contains("a relative link renders as an anchor", out,
+	"<a href='docs/page.html'>docs</a>")
+
+out = render("README.md", "[mail](mailto:a@example.com)")
+h.contains("a mailto link renders as an anchor", out,
+	"<a href='mailto:a@example.com'>mail</a>")
+
+out = render("README.md", "[bad](javascript:alert(1))")
+h.excludes("a javascript link renders no anchor", out, "<a ")
+h.contains("its text still reaches the page", out, "bad")
+
+out = render("README.md", "[bad](java\nscript:alert(1))")
+h.excludes("a split scheme is still caught", out, "<a ")
+
+out = render("README.md", "[bad](//evil.example/x)")
+h.excludes("a scheme relative link renders no anchor", out, "<a ")
+
+out = render("README.md", "[q](https://example.com/a'b)")
+h.contains("a quote in a link target is escaped", out,
+	"href='https://example.com/a&#x27;b'")
+
+out = render("README.md", "![a & b](https://example.com/i.png)")
+h.contains("an image renders with its alt text escaped", out,
+	"<img src='https://example.com/i.png' alt='a &amp; b'>")
+
+out = render("README.md", "![alt](data:image/png;base64,x)")
+h.excludes("an unsafe image renders no tag", out, "<img")
+h.contains("its alt text is kept as text", out, "![alt]")
+
+out = render("README.md", "```lua\nlocal x = 1 < 2\n```\nafter")
+h.contains("a fenced block carries its language", out,
+	"<pre><code data-lang='lua'>local x = 1 &lt; 2\n</code></pre>")
+h.contains("text after the fence renders on", out, "<p>after</p>")
+
+out = render("README.md", "~~~\n**not bold**\n~~~")
+h.contains("no inline parsing happens inside a fence", out,
+	"<code>**not bold**\n</code>")
+
+out = render("README.md", "---")
+h.contains("a rule of dashes renders a break", out, "<hr>")
+
+out = render("README.md", "- - -")
+h.contains("a spaced rule renders a break too", out, "<hr>")
+
+out = render("README.md", "> quoted words")
+h.contains("a quote renders as a blockquote", out,
+	"<blockquote><p>quoted words</p></blockquote>")
+
+out = render("README.md", "> > deep")
+h.contains("quotes nest", out,
+	"<blockquote><blockquote><p>deep</p></blockquote></blockquote>")
+
+-- A hostile readme of stacked markers has to hit the depth ceiling instead
+-- of the interpreter's stack.
+out = render("README.md", string.rep("> ", 30) .. "x")
+local _, quote_count = out:gsub("<blockquote>", "")
+h.equals("quote nesting stops at the ceiling", quote_count, 24)
+h.contains("the innermost text still renders", out, "x")
+
+out = render("README.md", string.rep("*a", 200))
+h.check("an emphasis bomb still renders something", #out > 0)
+
+out = render("README.md", "| a | b |\n| --- | --- |\n| 1 | 2 |")
+h.contains("a pipe table renders its head",
+	out, "<thead><tr><th>a</th><th>b</th></tr></thead>")
+h.contains("a pipe table renders its body",
+	out, "<tbody><tr><td>1</td><td>2</td></tr></tbody>")
+
+out = render("README.md", "a | b in prose")
+h.excludes("a pipe without a delimiter row stays prose", out, "<table")
+
+out = render("README.md", "- first\n- second")
+h.contains("dashes render an unordered list", out,
+	"<ul><li>first</li><li>second</li></ul>")
+
+out = render("README.md", "1. first\n2. second")
+h.contains("numbers render an ordered list", out,
+	"<ol><li>first</li><li>second</li></ol>")
+
+out = render("README.md", "- plain\n1. numbered")
+h.contains("a change of marker splits the list", out, "</ul><ol>")
+
+out = render("README.md", "- a **bold** item")
+h.contains("list items parse their text", out,
+	"<li>a <strong>bold</strong> item</li>")
+
+out = render("README.md", string.rep("a", 512 * 1024 + 1))
+h.contains("an oversized readme renders as plain text", out,
+	"<pre class='plaintext'>")
+
+-- Man pages.
+
+out = render("page.1", ".SH NAME")
+h.contains("a section heading renders", out,
+	"<div class='markdown manpage'><h2>NAME</h2></div>")
+
+out = render("page.man", '.SH "TWO WORDS"')
+h.contains("a quoted heading argument is unquoted", out,
+	"<h2>TWO WORDS</h2>")
+
+out = render("page.3", ".SS Details")
+h.contains("a subsection renders one level down", out, "<h3>Details</h3>")
+
+out = render("page.1", ".B bold words\nplain after")
+h.contains("the bold macro renders strong", out,
+	"<p><strong>bold words</strong><br>plain after</p>")
+
+out = render("page.1", "a \\fBbold\\fR c")
+h.contains("font escapes carry across a run", out,
+	"<p>a <strong>bold</strong> c</p>")
+
+out = render("page.1", "x \\(em y \\- z")
+h.contains("character escapes are translated", out, "<p>x - y - z</p>")
+
+out = render("page.1", ".nf\ncode <x>\n.fi\nafter")
+h.contains("a no fill block renders as code", out,
+	"<pre><code>code &lt;x&gt;\n</code></pre>")
+h.contains("filling resumes after it", out, "<p>after</p>")
+
+out = render("page.1", '.\\" a comment\nvisible')
+h.excludes("a roff comment is dropped", out, "comment")
+h.contains("the line after it renders", out, "visible")
+
+out = render("page.1", "first\n\nsecond")
+h.contains("a blank line splits paragraphs", out,
+	"<p>first</p><p>second</p>")
+
+out = render("page.1", ".TH title 1\nbody")
+h.excludes("the title macro renders nothing itself", out, "title 1")
+h.contains("the body after it renders", out, "<p>body</p>")
+
+h.finish()