khosra/internal/ext/links/links_test.go

package links

import (
	"bytes"
	"strings"
	"testing"

	"github.com/yuin/goldmark"
	"github.com/yuin/goldmark/parser"

	"khosra/internal/render"
)

// site is the pretend index every case resolves against: three bundles, one of them slugged, one with a
// Bengali variant. Keys in, URLs out — exactly the shape the real resolver has.
func site(key, lang string) (string, bool) {
	routes := map[string]string{
		"posts/day-01":           "/posts/day-01/",
		"posts/first-light":      "/posts/first-light/",
		"writing/notes-on-water": "/writing/notes-on-water/",
		"posts/renamed":          "/posts/a-better-name/", // a slug moved the address, never the key
	}
	url, ok := routes[key]
	if !ok {
		return "", false
	}
	if lang == "bn" {
		return "/bn" + url, true
	}
	return url, true
}

// convert renders body as a bundle at dir would be, and returns the HTML.
func convert(t *testing.T, dir, lang, body string) string {
	t.Helper()
	md := goldmark.New(goldmark.WithExtensions(New()))
	pc := parser.NewContext()
	render.WithOrigin(pc, render.Origin{Dir: dir, Lang: lang, Resolve: site})
	var out bytes.Buffer
	if err := md.Convert([]byte(body), &out, parser.WithContext(pc)); err != nil {
		t.Fatal(err)
	}
	return out.String()
}

// What must be rewritten: a path on disk becomes the address that path is served at.
func TestARelativeLinkBecomesTheServedAddress(t *testing.T) {
	for _, c := range []struct {
		what, from, link, want string
	}{
		{"a sibling single-file bundle, written with its filename", "content/posts/first-light",
			"../day-01.en.md", `href="/posts/day-01/"`},
		{"the same, written without the suffix", "content/posts/first-light",
			"../day-01", `href="/posts/day-01/"`},
		{"the same, written with only .md", "content/posts/first-light",
			"../day-01.md", `href="/posts/day-01/"`},
		{"a directory bundle in another section", "content/posts/first-light",
			"../../writing/notes-on-water/", `href="/writing/notes-on-water/"`},
		{"a directory bundle by its index file", "content/posts/first-light",
			"../../writing/notes-on-water/index.en.md", `href="/writing/notes-on-water/"`},
		{"from a single-file bundle, whose directory is the section", "content/posts",
			"./first-light/", `href="/posts/first-light/"`},
		{"a slug moves the address and never the key", "content/posts/first-light",
			"../renamed.md", `href="/posts/a-better-name/"`},
		{"a fragment survives", "content/posts/first-light",
			"../day-01.md#the-part", `href="/posts/day-01/#the-part"`},
		{"a query survives", "content/posts/first-light",
			"../day-01.md?raw", `href="/posts/day-01/?raw"`},
	} {
		got := convert(t, c.from, "en", "["+c.what+"]("+c.link+")")
		if !strings.Contains(got, c.want) {
			t.Errorf("%s: %q from %s\n  got  %s\n  want %s", c.what, c.link, c.from, strings.TrimSpace(got), c.want)
		}
	}
}

// What prose it eats — the table that decides whether this ships. Every destination here must survive
// **exactly** as written, because rewriting any of them would break something that works today.
func TestWhatItLeavesAlone(t *testing.T) {
	for _, c := range []struct{ what, link string }{
		{"an absolute URL", "https://example.com/posts/day-01/"},
		{"a scheme-relative URL", "//example.com/x"},
		{"a mail address", "mailto:someone@example.com"},
		{"a telephone link", "tel:+8801000000"},
		{"a root-relative path, which is already an address", "/posts/day-01/"},
		{"a bare fragment", "#a-heading"},
		{"a bare query", "?raw"},
		{"a bundle's own asset, which already resolves because the URL mirrors the directory", "cover.jpg"},
		{"the same, written explicitly relative", "./cover.jpg"},
		{"an asset in a subdirectory", "./scans/page-01.png"},
		{"a name climbing out of the content directory", "../../../etc/passwd"},
		{"a relative path naming no bundle at all", "../nothing-here.md"},
		{"a relative path naming no bundle, without a suffix", "../nothing-here"},
		{"an asset one level up, which is not this bundle's to link but is still not a bundle", "../shared.css"},
	} {
		got := convert(t, "content/posts/first-light", "en", "["+c.what+"]("+c.link+")")
		if !strings.Contains(got, `href="`+c.link+`"`) {
			t.Errorf("%s: %q was altered\n  got %s", c.what, c.link, strings.TrimSpace(got))
		}
	}
}

// A Bengali page links the Bengali variant, through the same fallback every lookup uses (ADR-0009).
func TestTheLinkFollowsTheLanguageBeingRendered(t *testing.T) {
	got := convert(t, "content/posts/first-light", "bn", "[চলুন](../day-01.md)")
	if !strings.Contains(got, `href="/bn/posts/day-01/"`) {
		t.Errorf("a Bengali page should link the Bengali variant:\n%s", got)
	}
}

// Without a resolver the feature does nothing at all, rather than guessing an address. That is the state a
// renderer built with no site root is in, and a guess there would emit links to pages that do not exist.
func TestWithoutAResolverNothingIsTouched(t *testing.T) {
	md := goldmark.New(goldmark.WithExtensions(New()))
	pc := parser.NewContext()
	render.WithOrigin(pc, render.Origin{Dir: "content/posts/first-light", Lang: "en"})
	var out bytes.Buffer
	if err := md.Convert([]byte("[x](../day-01.md)"), &out, parser.WithContext(pc)); err != nil {
		t.Fatal(err)
	}
	if !strings.Contains(out.String(), `href="../day-01.md"`) {
		t.Errorf("with no resolver the destination must survive:\n%s", out.String())
	}
}

// Code spans are literal, so a path inside one is text and not a link. Worth asserting rather than assuming:
// every syntax defect this engine has shipped was something eating ordinary prose.
func TestAPathInACodeSpanIsNotALink(t *testing.T) {
	got := convert(t, "content/posts/first-light", "en", "Write `../day-01.md` to link it.")
	if !strings.Contains(got, "<code>../day-01.md</code>") {
		t.Errorf("a code span must stay literal:\n%s", got)
	}
}