Add sitemap support (#18407)

custom/conf/app.example.ini

@@ -1097,6 +1097,9 @@ PATH =
 ;; Number of items that are displayed in home feed
 ;FEED_PAGING_NUM = 20
 ;;
+;; Number of items that are displayed in a single subsitemap
+;SITEMAP_PAGING_NUM = 20
+;;
 ;; Number of maximum commits displayed in commit graph.
 ;GRAPH_MAX_COMMIT_NUM = 100
 ;;

docs/content/doc/advanced/config-cheat-sheet.en-us.md

@@ -174,6 +174,7 @@ The following configuration set `Content-Type: application/vnd.android.package-a
 - `MEMBERS_PAGING_NUM`: **20**: Number of members that are shown in organization members.
 - `FEED_MAX_COMMIT_NUM`: **5**: Number of maximum commits shown in one activity feed.
 - `FEED_PAGING_NUM`: **20**: Number of items that are displayed in home feed.
+- `SITEMAP_PAGING_NUM`: **20**: Number of items that are displayed in a single subsitemap.
 - `GRAPH_MAX_COMMIT_NUM`: **100**: Number of maximum commits shown in the commit graph.
 - `CODE_COMMENT_LINES`: **4**: Number of line of codes shown for a code comment.
 - `DEFAULT_THEME`: **auto**: \[auto, gitea, arc-green\]: Set the default theme for the Gitea install.

modules/setting/setting.go

@@ -207,6 +207,7 @@ var (
 	// UI settings
 	UI = struct {
 		ExplorePagingNum      int
+		SitemapPagingNum      int
 		IssuePagingNum        int
 		RepoSearchPagingNum   int
 		MembersPagingNum      int
@@ -260,6 +261,7 @@ var (
 		} `ini:"ui.meta"`
 	}{
 		ExplorePagingNum:    20,
+		SitemapPagingNum:    20,
 		IssuePagingNum:      10,
 		RepoSearchPagingNum: 10,
 		MembersPagingNum:    20,

modules/sitemap/sitemap.go

@@ -0,0 +1,69 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package sitemap
+
+import (
+	"bytes"
+	"encoding/xml"
+	"fmt"
+	"io"
+	"time"
+)
+
+// sitemapFileLimit contains the maximum size of a sitemap file
+const sitemapFileLimit = 50 * 1024 * 1024
+
+// Url represents a single sitemap entry
+type URL struct {
+	URL     string     `xml:"loc"`
+	LastMod *time.Time `xml:"lastmod,omitempty"`
+}
+
+// SitemapUrl represents a sitemap
+type Sitemap struct {
+	XMLName   xml.Name
+	Namespace string `xml:"xmlns,attr"`
+
+	URLs []URL `xml:"url"`
+}
+
+// NewSitemap creates a sitemap
+func NewSitemap() *Sitemap {
+	return &Sitemap{
+		XMLName:   xml.Name{Local: "urlset"},
+		Namespace: "http://www.sitemaps.org/schemas/sitemap/0.9",
+	}
+}
+
+// NewSitemap creates a sitemap index.
+func NewSitemapIndex() *Sitemap {
+	return &Sitemap{
+		XMLName:   xml.Name{Local: "sitemapindex"},
+		Namespace: "http://www.sitemaps.org/schemas/sitemap/0.9",
+	}
+}
+
+// Add adds a URL to the sitemap
+func (s *Sitemap) Add(u URL) {
+	s.URLs = append(s.URLs, u)
+}
+
+// Write writes the sitemap to a response
+func (s *Sitemap) WriteTo(w io.Writer) (int64, error) {
+	if len(s.URLs) > 50000 {
+		return 0, fmt.Errorf("The sitemap contains too many URLs: %d", len(s.URLs))
+	}
+	buf := bytes.NewBufferString(xml.Header)
+	if err := xml.NewEncoder(buf).Encode(s); err != nil {
+		return 0, err
+	}
+	if err := buf.WriteByte('\n'); err != nil {
+		return 0, err
+	}
+	if buf.Len() > sitemapFileLimit {
+		return 0, fmt.Errorf("The sitemap is too big: %d", buf.Len())
+	}
+	return buf.WriteTo(w)
+}

modules/sitemap/sitemap_test.go

@@ -0,0 +1,77 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// Use of this source code is governed by a MIT-style
+// license that can be found in the LICENSE file.
+
+package sitemap
+
+import (
+	"bytes"
+	"encoding/xml"
+	"fmt"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestOk(t *testing.T) {
+	testReal := func(s *Sitemap, name string, urls []URL, expected string) {
+		for _, url := range urls {
+			s.Add(url)
+		}
+		buf := &bytes.Buffer{}
+		_, err := s.WriteTo(buf)
+		assert.NoError(t, nil, err)
+		assert.Equal(t, xml.Header+"<"+name+" xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\">"+expected+"</"+name+">\n", buf.String())
+	}
+	test := func(urls []URL, expected string) {
+		testReal(NewSitemap(), "urlset", urls, expected)
+		testReal(NewSitemapIndex(), "sitemapindex", urls, expected)
+	}
+
+	ts := time.Unix(1651322008, 0).UTC()
+
+	test(
+		[]URL{},
+		"",
+	)
+	test(
+		[]URL{
+			{URL: "https://gitea.io/test1", LastMod: &ts},
+		},
+		"<url><loc>https://gitea.io/test1</loc><lastmod>2022-04-30T12:33:28Z</lastmod></url>",
+	)
+	test(
+		[]URL{
+			{URL: "https://gitea.io/test2", LastMod: nil},
+		},
+		"<url><loc>https://gitea.io/test2</loc></url>",
+	)
+	test(
+		[]URL{
+			{URL: "https://gitea.io/test1", LastMod: &ts},
+			{URL: "https://gitea.io/test2", LastMod: nil},
+		},
+		"<url><loc>https://gitea.io/test1</loc><lastmod>2022-04-30T12:33:28Z</lastmod></url>"+
+			"<url><loc>https://gitea.io/test2</loc></url>",
+	)
+}
+
+func TestTooManyURLs(t *testing.T) {
+	s := NewSitemap()
+	for i := 0; i < 50001; i++ {
+		s.Add(URL{URL: fmt.Sprintf("https://gitea.io/test%d", i)})
+	}
+	buf := &bytes.Buffer{}
+	_, err := s.WriteTo(buf)
+	assert.EqualError(t, err, "The sitemap contains too many URLs: 50001")
+}
+
+func TestSitemapTooBig(t *testing.T) {
+	s := NewSitemap()
+	s.Add(URL{URL: strings.Repeat("b", sitemapFileLimit)})
+	buf := &bytes.Buffer{}
+	_, err := s.WriteTo(buf)
+	assert.EqualError(t, err, "The sitemap is too big: 52428931")
+}

routers/web/explore/repo.go

@@ -11,7 +11,9 @@ import (
 	repo_model "code.gitea.io/gitea/models/repo"
 	"code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/context"
+	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/sitemap"
 )
 
 const (
@@ -30,11 +32,21 @@ type RepoSearchOptions struct {
 
 // RenderRepoSearch render repositories search page
 func RenderRepoSearch(ctx *context.Context, opts *RepoSearchOptions) {
-	page := ctx.FormInt("page")
+	// Sitemap index for sitemap paths
+	page := int(ctx.ParamsInt64("idx"))
+	isSitemap := ctx.Params("idx") != ""
+	if page <= 1 {
+		page = ctx.FormInt("page")
+	}
+
 	if page <= 0 {
 		page = 1
 	}
 
+	if isSitemap {
+		opts.PageSize = setting.UI.SitemapPagingNum
+	}
+
 	var (
 		repos   []*repo_model.Repository
 		count   int64
@@ -100,6 +112,18 @@ func RenderRepoSearch(ctx *context.Context, opts *RepoSearchOptions) {
 		ctx.ServerError("SearchRepository", err)
 		return
 	}
+	if isSitemap {
+		m := sitemap.NewSitemap()
+		for _, item := range repos {
+			m.Add(sitemap.URL{URL: item.HTMLURL(), LastMod: item.UpdatedUnix.AsTimePtr()})
+		}
+		ctx.Resp.Header().Set("Content-Type", "text/xml")
+		if _, err := m.WriteTo(ctx.Resp); err != nil {
+			log.Error("Failed writing sitemap: %v", err)
+		}
+		return
+	}
+
 	ctx.Data["Keyword"] = keyword
 	ctx.Data["Total"] = count
 	ctx.Data["Repos"] = repos

routers/web/explore/user.go

@@ -12,7 +12,9 @@ import (
 	user_model "code.gitea.io/gitea/models/user"
 	"code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/context"
+	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/sitemap"
 	"code.gitea.io/gitea/modules/structs"
 	"code.gitea.io/gitea/modules/util"
 )
@@ -33,11 +35,20 @@ func isKeywordValid(keyword string) bool {
 
 // RenderUserSearch render user search page
 func RenderUserSearch(ctx *context.Context, opts *user_model.SearchUserOptions, tplName base.TplName) {
+	// Sitemap index for sitemap paths
+	opts.Page = int(ctx.ParamsInt64("idx"))
+	isSitemap := ctx.Params("idx") != ""
+	if opts.Page <= 1 {
 		opts.Page = ctx.FormInt("page")
+	}
 	if opts.Page <= 1 {
 		opts.Page = 1
 	}
 
+	if isSitemap {
+		opts.PageSize = setting.UI.SitemapPagingNum
+	}
+
 	var (
 		users   []*user_model.User
 		count   int64
@@ -73,6 +84,18 @@ func RenderUserSearch(ctx *context.Context, opts *user_model.SearchUserOptions,
 			return
 		}
 	}
+	if isSitemap {
+		m := sitemap.NewSitemap()
+		for _, item := range users {
+			m.Add(sitemap.URL{URL: item.HTMLURL(), LastMod: item.UpdatedUnix.AsTimePtr()})
+		}
+		ctx.Resp.Header().Set("Content-Type", "text/xml")
+		if _, err := m.WriteTo(ctx.Resp); err != nil {
+			log.Error("Failed writing sitemap: %v", err)
+		}
+		return
+	}
+
 	ctx.Data["Keyword"] = opts.Keyword
 	ctx.Data["Total"] = count
 	ctx.Data["Users"] = users

routers/web/home.go

@@ -7,11 +7,18 @@ package web
 
 import (
 	"net/http"
+	"strconv"
 
+	"code.gitea.io/gitea/models/db"
+	repo_model "code.gitea.io/gitea/models/repo"
+	user_model "code.gitea.io/gitea/models/user"
 	"code.gitea.io/gitea/modules/base"
 	"code.gitea.io/gitea/modules/context"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
+	"code.gitea.io/gitea/modules/sitemap"
+	"code.gitea.io/gitea/modules/structs"
+	"code.gitea.io/gitea/modules/util"
 	"code.gitea.io/gitea/modules/web/middleware"
 	"code.gitea.io/gitea/routers/web/auth"
 	"code.gitea.io/gitea/routers/web/user"
@@ -59,6 +66,52 @@ func Home(ctx *context.Context) {
 	ctx.HTML(http.StatusOK, tplHome)
 }
 
+// HomeSitemap renders the main sitemap
+func HomeSitemap(ctx *context.Context) {
+	m := sitemap.NewSitemapIndex()
+	if !setting.Service.Explore.DisableUsersPage {
+		_, cnt, err := user_model.SearchUsers(&user_model.SearchUserOptions{
+			Type:        user_model.UserTypeIndividual,
+			ListOptions: db.ListOptions{PageSize: 1},
+			IsActive:    util.OptionalBoolTrue,
+			Visible:     []structs.VisibleType{structs.VisibleTypePublic},
+		})
+		if err != nil {
+			ctx.ServerError("SearchUsers", err)
+			return
+		}
+		count := int(cnt)
+		idx := 1
+		for i := 0; i < count; i += setting.UI.SitemapPagingNum {
+			m.Add(sitemap.URL{URL: setting.AppURL + "explore/users/sitemap-" + strconv.Itoa(idx) + ".xml"})
+			idx++
+		}
+	}
+
+	_, cnt, err := repo_model.SearchRepository(&repo_model.SearchRepoOptions{
+		ListOptions: db.ListOptions{
+			PageSize: 1,
+		},
+		Actor:     ctx.Doer,
+		AllPublic: true,
+	})
+	if err != nil {
+		ctx.ServerError("SearchRepository", err)
+		return
+	}
+	count := int(cnt)
+	idx := 1
+	for i := 0; i < count; i += setting.UI.SitemapPagingNum {
+		m.Add(sitemap.URL{URL: setting.AppURL + "explore/repos/sitemap-" + strconv.Itoa(idx) + ".xml"})
+		idx++
+	}
+
+	ctx.Resp.Header().Set("Content-Type", "text/xml")
+	if _, err := m.WriteTo(ctx.Resp); err != nil {
+		log.Error("Failed writing sitemap: %v", err)
+	}
+}
+
 // NotFound render 404 page
 func NotFound(ctx *context.Context) {
 	ctx.Data["Title"] = "Page Not Found"

routers/web/web.go

@@ -294,6 +294,7 @@ func RegisterRoutes(m *web.Route) {
 	// Routers.
 	// for health check
 	m.Get("/", Home)
+	m.Get("/sitemap.xml", ignExploreSignIn, HomeSitemap)
 	m.Group("/.well-known", func() {
 		m.Get("/openid-configuration", auth.OIDCWellKnown)
 		m.Group("", func() {
@@ -310,7 +311,9 @@ func RegisterRoutes(m *web.Route) {
 			ctx.Redirect(setting.AppSubURL + "/explore/repos")
 		})
 		m.Get("/repos", explore.Repos)
+		m.Get("/repos/sitemap-{idx}.xml", explore.Repos)
 		m.Get("/users", explore.Users)
+		m.Get("/users/sitemap-{idx}.xml", explore.Users)
 		m.Get("/organizations", explore.Organizations)
 		m.Get("/code", explore.Code)
 		m.Get("/topics/search", explore.TopicSearch)