RenewCMS/test/seo_output_sitemap_http_test.go

package test

import (
	"RenewCMS/api"
	"RenewCMS/api/pkg/apperr"
	"RenewCMS/internal/domain/article"
	"encoding/json"
	"net/http"
	"net/http/httptest"
	"reflect"
	"strconv"
	"strings"
	"testing"
)

func (e *seoOutEnv) sitemap(t *testing.T, who string, query string) (*httptest.ResponseRecorder, []map[string]any) {
	t.Helper()
	rec := e.call(t, who, http.MethodGet, "/v1/site/renew/sitemap"+query, "")
	var items []map[string]any
	if rec.Code == http.StatusOK {
		if err := json.Unmarshal(rec.Body.Bytes(), &items); err != nil {
			t.Fatalf("decode %q: %v", rec.Body.String(), err)
		}
	}
	return rec, items
}

func (e *seoOutEnv) seedOnline(count int) {
	e.cleanArticles()
	for id := 1; id <= count; id++ {
		e.articles.articles[uint32(id)] = article.Article{
			ID: uint32(id), Title: "Article " + strconv.Itoa(id), IsOnline: true, SiteID: siteMainID, Slug: "slug-" + strconv.Itoa(id),
			CreatedAt: seoOutCreated, UpdatedAt: seoOutModified,
		}
	}
}

func TestSeoOutputSitemap(t *testing.T) {
	t.Run("The sitemap lists the online articles of one site", func(t *testing.T) {
		e := newSeoOutEnv(t)
		e.articles.articles[16] = article.Article{ID: 16, Title: "Sixteen", IsOnline: true, SiteID: siteMainID, Slug: "sixteen", CreatedAt: seoOutCreated, UpdatedAt: seoOutModified}
		e.articles.articles[20] = article.Article{ID: 20, Title: "Twenty", IsOnline: true, SiteID: seoOutJournalID, Slug: "twenty", CreatedAt: seoOutCreated, UpdatedAt: seoOutModified}
		rec, items := e.sitemap(t, "", "")
		want := []map[string]any{
			{"url": seoOutBase + "/" + seoOutArticleSlug, "modified_at": "2026-09-10T09:30:00Z", "noindex": false},
			{"url": seoOutBase + "/sixteen", "modified_at": "2026-09-10T09:30:00Z", "noindex": false},
		}
		if rec.Code != http.StatusOK || !reflect.DeepEqual(items, want) || rec.Header().Get("X-Total-Count") != "2" {
			t.Errorf("status %d total %q items %v", rec.Code, rec.Header().Get("X-Total-Count"), items)
		}
	})

	t.Run("Noindex articles are included with the flag", func(t *testing.T) {
		e := newSeoOutEnv(t)
		e.change(12, func(a *article.Article) { a.Noindex = true })
		_, items := e.sitemap(t, "", "")
		if len(items) != 1 || items[0]["noindex"] != true {
			t.Errorf("items = %v", items)
		}
	})

	t.Run("The url uses the slug and not the canonical override", func(t *testing.T) {
		e := newSeoOutEnv(t)
		e.change(12, func(a *article.Article) { a.CanonicalURL = "https://other.example/a" })
		_, items := e.sitemap(t, "", "")
		if items[0]["url"] != seoOutBase+"/"+seoOutArticleSlug {
			t.Errorf("items = %v", items)
		}
	})

	t.Run("Drafts are never listed", func(t *testing.T) {
		e := newSeoOutEnv(t)
		rec, items := e.sitemap(t, "owner", "")
		if rec.Header().Get("X-Total-Count") != "1" || len(items) != 1 {
			t.Errorf("items = %v", items)
		}
	})

	t.Run("An unpublished article leaves the sitemap", func(t *testing.T) {
		e := newSeoOutEnv(t)
		e.call(t, "owner", http.MethodPatch, "/v1/article/12/unpublish", "")
		rec, items := e.sitemap(t, "", "")
		if rec.Header().Get("X-Total-Count") != "0" || len(items) != 0 {
			t.Errorf("items = %v", items)
		}
	})

	t.Run("An empty site answers an empty array", func(t *testing.T) {
		e := newSeoOutEnv(t)
		e.change(12, func(a *article.Article) { a.IsOnline = false })
		rec, _ := e.sitemap(t, "", "")
		if rec.Code != http.StatusOK || strings.TrimSpace(rec.Body.String()) != "[]" || rec.Header().Get("X-Total-Count") != "0" {
			t.Errorf("%d %q", rec.Code, rec.Body.String())
		}
	})

	t.Run("Each item has exactly the documented keys", func(t *testing.T) {
		e := newSeoOutEnv(t)
		_, items := e.sitemap(t, "", "")
		if sortedKeys(items[0]) != "modified_at,noindex,url" {
			t.Errorf("keys = %s", sortedKeys(items[0]))
		}
	})

	t.Run("Paging and the total header", func(t *testing.T) {
		cases := []struct {
			query string
			count int
		}{{"", 100}, {"?size=25", 25}, {"?page=3", 50}, {"?page=2&size=100", 100}, {"?size=500", 250}, {"?page=99&size=25", 0}}
		e := newSeoOutEnv(t)
		e.seedOnline(250)
		for _, c := range cases {
			rec, items := e.sitemap(t, "", c.query)
			if len(items) != c.count || rec.Header().Get("X-Total-Count") != "250" {
				t.Errorf("%q: %d items, total %q", c.query, len(items), rec.Header().Get("X-Total-Count"))
			}
		}
	})

	t.Run("Pages partition the sitemap in a stable order", func(t *testing.T) {
		e := newSeoOutEnv(t)
		e.seedOnline(60)
		seen := []string{}
		for page, want := range map[int]int{1: 25, 2: 25, 3: 10} {
			_, items := e.sitemap(t, "", "?size=25&page="+strconv.Itoa(page))
			if len(items) != want {
				t.Errorf("page %d holds %d", page, len(items))
			}
			for index, item := range items {
				id := (page-1)*25 + index + 1
				if item["url"] != seoOutBase+"/slug-"+strconv.Itoa(id) {
					t.Errorf("page %d item %d = %v", page, index, item["url"])
				}
				seen = append(seen, item["url"].(string))
			}
		}
		if len(seen) != 60 {
			t.Errorf("seen %d", len(seen))
		}
	})

	t.Run("Invalid paging answers 400", func(t *testing.T) {
		for _, query := range []string{"page=0", "page=-1", "page=abc", "page=", "size=0", "size=10", "size=1000", "size=abc"} {
			e := newSeoOutEnv(t)
			before := e.counted.readCalls()
			expectStatusAndText(t, e.call(t, "", http.MethodGet, "/v1/site/renew/sitemap?"+query, ""), 400, apperr.BodyErrorMessage)
			if e.counted.readCalls() != before {
				t.Errorf("%s: the repository was called", query)
			}
		}
	})

	t.Run("An unknown key is not found", func(t *testing.T) {
		e := newSeoOutEnv(t)
		expectStatusAndText(t, e.call(t, "", http.MethodGet, "/v1/site/nope/sitemap", ""), 404, apperr.SiteNotFoundErrorMessage)
	})

	t.Run("The key is not an id", func(t *testing.T) {
		e := newSeoOutEnv(t)
		if rec := e.call(t, "", http.MethodGet, "/v1/site/1/sitemap", ""); rec.Code != http.StatusNotFound {
			t.Errorf("status %d", rec.Code)
		}
	})

	t.Run("The route does not disturb the owner-only site routes", func(t *testing.T) {
		e := newSeoOutEnv(t)
		expectStatusAndText(t, e.call(t, "admin", http.MethodPatch, "/v1/site/2", `{"site_name": "x"}`), 403, apperr.OwnerRequiredErrorMessage)
	})

	t.Run("The browser can read the total header", func(t *testing.T) {
		e := newSeoOutEnv(t)
		settings := e.settings
		settings.CorsAllowedOrigins = []string{"https://app.example"}
		req := httptest.NewRequest(http.MethodGet, "/v1/site/renew/sitemap", nil)
		req.Header.Set("Origin", "https://app.example")
		rec := httptest.NewRecorder()
		api.InitRoutes(e.container, settings).ServeHTTP(rec, req)
		exposed := rec.Header().Get("Access-Control-Expose-Headers")
		if rec.Code != http.StatusOK || !strings.Contains(exposed, "X-Total-Count") || !strings.Contains(exposed, "Link") {
			t.Errorf("status %d exposed %q", rec.Code, exposed)
		}
	})

	t.Run("A repository failure answers 500", func(t *testing.T) {
		e := newSeoOutEnv(t)
		e.articles.fail.failNext("FindPage")
		if rec := e.call(t, "", http.MethodGet, "/v1/site/renew/sitemap", ""); rec.Code != http.StatusInternalServerError {
			t.Errorf("status %d", rec.Code)
		}
	})
}