From dd136858f1ea40ad3c94191d647487fa4f31926c Mon Sep 17 00:00:00 2001 From: Daniel Baumann Date: Fri, 18 Oct 2024 20:33:49 +0200 Subject: Adding upstream version 9.0.0. Signed-off-by: Daniel Baumann --- modules/sitemap/sitemap.go | 82 ++++++++++++++++++++ modules/sitemap/sitemap_test.go | 167 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 249 insertions(+) create mode 100644 modules/sitemap/sitemap.go create mode 100644 modules/sitemap/sitemap_test.go (limited to 'modules/sitemap') diff --git a/modules/sitemap/sitemap.go b/modules/sitemap/sitemap.go new file mode 100644 index 0000000..280ca1d --- /dev/null +++ b/modules/sitemap/sitemap.go @@ -0,0 +1,82 @@ +// Copyright 2022 The Gitea Authors. All rights reserved. +// SPDX-License-Identifier: MIT + +package sitemap + +import ( + "bytes" + "encoding/xml" + "fmt" + "io" + "time" +) + +const ( + sitemapFileLimit = 50 * 1024 * 1024 // the maximum size of a sitemap file + urlsLimit = 50000 + + schemaURL = "http://www.sitemaps.org/schemas/sitemap/0.9" + urlsetName = "urlset" + sitemapindexName = "sitemapindex" +) + +// URL represents a single sitemap entry +type URL struct { + URL string `xml:"loc"` + LastMod *time.Time `xml:"lastmod,omitempty"` +} + +// Sitemap represents a sitemap +type Sitemap struct { + XMLName xml.Name + Namespace string `xml:"xmlns,attr"` + + URLs []URL `xml:"url"` + Sitemaps []URL `xml:"sitemap"` +} + +// NewSitemap creates a sitemap +func NewSitemap() *Sitemap { + return &Sitemap{ + XMLName: xml.Name{Local: urlsetName}, + Namespace: schemaURL, + } +} + +// NewSitemapIndex creates a sitemap index. +func NewSitemapIndex() *Sitemap { + return &Sitemap{ + XMLName: xml.Name{Local: sitemapindexName}, + Namespace: schemaURL, + } +} + +// Add adds a URL to the sitemap +func (s *Sitemap) Add(u URL) { + if s.XMLName.Local == sitemapindexName { + s.Sitemaps = append(s.Sitemaps, u) + } else { + s.URLs = append(s.URLs, u) + } +} + +// WriteTo writes the sitemap to a response +func (s *Sitemap) WriteTo(w io.Writer) (int64, error) { + if l := len(s.URLs); l > urlsLimit { + return 0, fmt.Errorf("The sitemap contains %d URLs, but only %d are allowed", l, urlsLimit) + } + if l := len(s.Sitemaps); l > urlsLimit { + return 0, fmt.Errorf("The sitemap contains %d sub-sitemaps, but only %d are allowed", l, urlsLimit) + } + 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 has %d bytes, but only %d are allowed", buf.Len(), sitemapFileLimit) + } + return buf.WriteTo(w) +} diff --git a/modules/sitemap/sitemap_test.go b/modules/sitemap/sitemap_test.go new file mode 100644 index 0000000..39a2178 --- /dev/null +++ b/modules/sitemap/sitemap_test.go @@ -0,0 +1,167 @@ +// Copyright 2022 The Gitea Authors. All rights reserved. +// SPDX-License-Identifier: MIT + +package sitemap + +import ( + "bytes" + "encoding/xml" + "strings" + "testing" + "time" + + "github.com/stretchr/testify/assert" + "github.com/stretchr/testify/require" +) + +func TestNewSitemap(t *testing.T) { + ts := time.Unix(1651322008, 0).UTC() + + tests := []struct { + name string + urls []URL + want string + wantErr string + }{ + { + name: "empty", + urls: []URL{}, + want: xml.Header + `` + + "" + + "\n", + }, + { + name: "regular", + urls: []URL{ + {URL: "https://gitea.io/test1", LastMod: &ts}, + }, + want: xml.Header + `` + + "https://gitea.io/test12022-04-30T12:33:28Z" + + "\n", + }, + { + name: "without lastmod", + urls: []URL{ + {URL: "https://gitea.io/test1"}, + }, + want: xml.Header + `` + + "https://gitea.io/test1" + + "\n", + }, + { + name: "multiple", + urls: []URL{ + {URL: "https://gitea.io/test1", LastMod: &ts}, + {URL: "https://gitea.io/test2", LastMod: nil}, + }, + want: xml.Header + `` + + "https://gitea.io/test12022-04-30T12:33:28Z" + + "https://gitea.io/test2" + + "\n", + }, + { + name: "too many urls", + urls: make([]URL, 50001), + wantErr: "The sitemap contains 50001 URLs, but only 50000 are allowed", + }, + { + name: "too big file", + urls: []URL{ + {URL: strings.Repeat("b", 50*1024*1024+1)}, + }, + wantErr: "The sitemap has 52428932 bytes, but only 52428800 are allowed", + }, + } + for _, tt := range tests { + t.Run(tt.name, func(t *testing.T) { + s := NewSitemap() + for _, url := range tt.urls { + s.Add(url) + } + buf := &bytes.Buffer{} + _, err := s.WriteTo(buf) + if tt.wantErr != "" { + assert.EqualError(t, err, tt.wantErr) + } else { + require.NoError(t, err) + assert.Equalf(t, tt.want, buf.String(), "NewSitemap()") + } + }) + } +} + +func TestNewSitemapIndex(t *testing.T) { + ts := time.Unix(1651322008, 0).UTC() + + tests := []struct { + name string + urls []URL + want string + wantErr string + }{ + { + name: "empty", + urls: []URL{}, + want: xml.Header + `` + + "" + + "\n", + }, + { + name: "regular", + urls: []URL{ + {URL: "https://gitea.io/test1", LastMod: &ts}, + }, + want: xml.Header + `` + + "https://gitea.io/test12022-04-30T12:33:28Z" + + "\n", + }, + { + name: "without lastmod", + urls: []URL{ + {URL: "https://gitea.io/test1"}, + }, + want: xml.Header + `` + + "https://gitea.io/test1" + + "\n", + }, + { + name: "multiple", + urls: []URL{ + {URL: "https://gitea.io/test1", LastMod: &ts}, + {URL: "https://gitea.io/test2", LastMod: nil}, + }, + want: xml.Header + `` + + "https://gitea.io/test12022-04-30T12:33:28Z" + + "https://gitea.io/test2" + + "\n", + }, + { + name: "too many sitemaps", + urls: make([]URL, 50001), + wantErr: "The sitemap contains 50001 sub-sitemaps, but only 50000 are allowed", + }, + { + name: "too big file", + urls: []URL{ + {URL: strings.Repeat("b", 50*1024*1024+1)}, + }, + wantErr: "The sitemap has 52428952 bytes, but only 52428800 are allowed", + }, + } + for _, tt := range tests { + t.Run(tt.name, func(t *testing.T) { + s := NewSitemapIndex() + for _, url := range tt.urls { + s.Add(url) + } + buf := &bytes.Buffer{} + _, err := s.WriteTo(buf) + if tt.wantErr != "" { + assert.EqualError(t, err, tt.wantErr) + } else { + require.NoError(t, err) + assert.Equalf(t, tt.want, buf.String(), "NewSitemapIndex()") + } + }) + } +} -- cgit v1.2.3