sm-sitemap-plugin/classes/serialize_test.go

package classes_test

import (
	"context"
	"strings"
	"testing"

	"git.golem15.com/golem15/sm-sitemap-plugin/classes"
	"git.golem15.com/golem15/sm-sitemap-plugin/models"
)

// TestXMLEscaping ports the PHP XmlEscapingTest (SITE-002) as behaviour:
// every value the sitemap writes is escaped exactly as libxml escapes it,
// text nodes (loc, lastmod, changefreq, priority) and attribute values
// (hreflang, href), and no raw value reaches the output.
func TestXMLEscaping(t *testing.T) {
	const raw = "a&b<c>d\"e'f\rg\th\ni"
	if got := classes.EscapeText(raw); got != "a&amp;b&lt;c&gt;d\"e'f&#13;g\th\ni" {
		t.Fatalf("EscapeText = %q", got)
	}
	if got := classes.EscapeAttr(raw); got != "a&amp;b&lt;c&gt;d&quot;e'f&#13;g&#9;h&#10;i" {
		t.Fatalf("EscapeAttr = %q", got)
	}
	if got := classes.EscapeText("zażółć — &amp;"); got != "zażółć — &amp;amp;" {
		t.Fatalf("EscapeText keeps UTF-8 and escapes an existing entity again: %q", got)
	}

	xml := classes.Serialize("https://h/sitemap.xsl", []classes.URLEntry{{
		Loc:        "https://h/a?x=1&y=<2>",
		Lastmod:    "2026-01-02T03:04:05+00:00",
		Changefreq: "weekly&<>",
		Priority:   "0.5\r",
		Alternates: []classes.Alternate{
			{Hreflang: "pl\"x", Href: "https://h/pl?a=1&b=\"2\"\t\n"},
			{Hreflang: "en", Href: "https://h/en'x"},
		},
	}})
	want := "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" +
		"<?xml-stylesheet type=\"text/xsl\" href=\"https://h/sitemap.xsl\"?>\n" +
		`<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" xmlns:xhtml="http://www.w3.org/1999/xhtml" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd">` +
		"<url><loc>https://h/a?x=1&amp;y=&lt;2&gt;</loc><lastmod>2026-01-02T03:04:05+00:00</lastmod><changefreq>weekly&amp;&lt;&gt;</changefreq><priority>0.5&#13;</priority>" +
		`<xhtml:link rel="alternate" hreflang="pl&quot;x" href="https://h/pl?a=1&amp;b=&quot;2&quot;&#9;&#10;"/>` +
		`<xhtml:link rel="alternate" hreflang="en" href="https://h/en'x"/>` +
		"</url></urlset>\n"
	if xml != want {
		t.Fatalf("serialized:\n%q\nwant:\n%q", xml, want)
	}

	// One alternate is not written; an empty urlset is self-closed.
	single := classes.Serialize("s", []classes.URLEntry{{Loc: "l", Changefreq: "c", Priority: "p", Alternates: []classes.Alternate{{Hreflang: "pl", Href: "x"}}}})
	if strings.Contains(single, "xhtml:link") || strings.Contains(single, "<lastmod>") {
		t.Fatalf("single alternate or empty lastmod written: %q", single)
	}
	if empty := classes.Serialize("s", nil); !strings.HasSuffix(empty, `sitemap.xsd"/>`+"\n") {
		t.Fatalf("empty urlset = %q", empty)
	}

	// Through the generator: stored changefreq and priority, a resolved
	// alternate's locale and a url item with every special character.
	url := "/q?a=1&b=<x>&c=\"y\""
	changefreq := "daily&<weekly>"
	priority := "0.9\"'"
	gen := &classes.Generator{Root: "https://api.example.test", Settings: models.Settings{}, Resolver: fakeResolver{t: t, types: map[string]map[string]goldenResolved{
		"acme-post": {"1": {URL: "/p?x=1&y=2", Alternates: [][2]string{{"pl&<", "/p?x=1&y=2"}, {"en\"", "/en/p?z=<1>"}}}},
	}}}
	out, err := gen.Generate(context.Background(), []models.DefinitionItem{
		{Type: models.ItemTypeURL, URL: &url, Changefreq: &changefreq, Priority: &priority},
		{Type: "acme-post", Reference: text("1"), Changefreq: &changefreq, Priority: &priority},
	})
	if err != nil {
		t.Fatal(err)
	}
	body := out[strings.Index(out, "<url>"):]
	for _, bad := range []string{"<x>", "&b", "&y", `"y"`, "<weekly>", "daily&<", `hreflang="pl&<"`, `en""`, "<1>"} {
		if strings.Contains(body, bad) {
			t.Fatalf("raw value %q reached the output:\n%s", bad, body)
		}
	}
	for _, good := range []string{
		"<changefreq>daily&amp;&lt;weekly&gt;</changefreq>",
		"<priority>0.9\"'</priority>",
		`hreflang="pl&amp;&lt;"`,
		`hreflang="en&quot;"`,
		"y=2",
	} {
		if !strings.Contains(body, good) {
			t.Fatalf("output lacks %q:\n%s", good, body)
		}
	}
	if strings.Count(body, "<url>") != 2 {
		t.Fatalf("urls = %d:\n%s", strings.Count(body, "<url>"), body)
	}
}

func text(s string) *string { return &s }