rss-tools/app/atom.go (view raw)
| 1 | package app |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "crypto/sha1" |
| 6 | "encoding/xml" |
| 7 | "fmt" |
| 8 | "io" |
| 9 | "net/http" |
| 10 | "strings" |
| 11 | "time" |
| 12 | ) |
| 13 | |
| 14 | const ( |
| 15 | atomNamespace = "http://www.w3.org/2005/Atom" |
| 16 | xhtmlNamespace = "http://www.w3.org/1999/xhtml" |
| 17 | defaultAuthor = "rss-tools" |
| 18 | ) |
| 19 | |
| 20 | type AtomFeed struct { |
| 21 | XMLName xml.Name `xml:"feed"` |
| 22 | XMLNS string `xml:"xmlns,attr"` |
| 23 | Title string `xml:"title"` |
| 24 | ID string `xml:"id"` |
| 25 | Updated string `xml:"updated"` |
| 26 | Authors []AtomPerson `xml:"author,omitempty"` |
| 27 | Subtitle string `xml:"subtitle,omitempty"` |
| 28 | Entries []AtomEntry `xml:"entry"` |
| 29 | } |
| 30 | |
| 31 | type AtomEntry struct { |
| 32 | Title string `xml:"title"` |
| 33 | ID string `xml:"id"` |
| 34 | Updated string `xml:"updated"` |
| 35 | Links []AtomLink `xml:"link,omitempty"` |
| 36 | Content AtomContent `xml:"content"` |
| 37 | } |
| 38 | |
| 39 | type AtomContent struct { |
| 40 | XMLName xml.Name `xml:"content"` |
| 41 | Type string `xml:"type,attr,omitempty"` |
| 42 | Value string `xml:",chardata"` |
| 43 | } |
| 44 | |
| 45 | func (c AtomContent) MarshalXML(e *xml.Encoder, start xml.StartElement) error { |
| 46 | contentType := c.Type |
| 47 | if contentType == "" { |
| 48 | contentType = "text" |
| 49 | } |
| 50 | |
| 51 | start.Name = xml.Name{Local: "content"} |
| 52 | start.Attr = append(start.Attr, xml.Attr{ |
| 53 | Name: xml.Name{Local: "type"}, |
| 54 | Value: contentType, |
| 55 | }) |
| 56 | |
| 57 | if err := e.EncodeToken(start); err != nil { |
| 58 | return err |
| 59 | } |
| 60 | |
| 61 | if contentType == "xhtml" { |
| 62 | if err := validateXHTMLFragment(c.Value); err != nil { |
| 63 | return err |
| 64 | } |
| 65 | |
| 66 | if err := e.Encode(xhtmlDiv{ |
| 67 | XMLNS: xhtmlNamespace, |
| 68 | Inner: c.Value, |
| 69 | }); err != nil { |
| 70 | return err |
| 71 | } |
| 72 | } else { |
| 73 | if err := e.EncodeToken(xml.CharData([]byte(c.Value))); err != nil { |
| 74 | return err |
| 75 | } |
| 76 | } |
| 77 | |
| 78 | if err := e.EncodeToken(start.End()); err != nil { |
| 79 | return err |
| 80 | } |
| 81 | return e.Flush() |
| 82 | } |
| 83 | |
| 84 | type xhtmlDiv struct { |
| 85 | XMLName xml.Name `xml:"div"` |
| 86 | XMLNS string `xml:"xmlns,attr"` |
| 87 | Inner string `xml:",innerxml"` |
| 88 | } |
| 89 | |
| 90 | func validateXHTMLFragment(fragment string) error { |
| 91 | wrapped := fmt.Sprintf(`<div xmlns="%s">%s</div>`, xhtmlNamespace, fragment) |
| 92 | dec := xml.NewDecoder(strings.NewReader(wrapped)) |
| 93 | for { |
| 94 | _, err := dec.Token() |
| 95 | if err == io.EOF { |
| 96 | return nil |
| 97 | } |
| 98 | if err != nil { |
| 99 | return fmt.Errorf("invalid xhtml content: %w", err) |
| 100 | } |
| 101 | } |
| 102 | } |
| 103 | |
| 104 | type AtomPerson struct { |
| 105 | Name string `xml:"name"` |
| 106 | } |
| 107 | |
| 108 | type AtomLink struct { |
| 109 | Rel string `xml:"rel,attr,omitempty"` |
| 110 | Type string `xml:"type,attr,omitempty"` |
| 111 | Length string `xml:"length,attr,omitempty"` |
| 112 | Href string `xml:"href,attr"` |
| 113 | } |
| 114 | |
| 115 | type FeedEntry struct { |
| 116 | Title string |
| 117 | ID string |
| 118 | Links []FeedLink |
| 119 | Content string |
| 120 | ContentType string // "text", "html", or "xhtml"; defaults to "text" |
| 121 | Updated time.Time |
| 122 | } |
| 123 | |
| 124 | type FeedLink struct { |
| 125 | Rel string |
| 126 | Type string |
| 127 | Length string |
| 128 | Href string |
| 129 | } |
| 130 | |
| 131 | type FeedBuilder struct{ f AtomFeed } |
| 132 | |
| 133 | func NewFeed(title, id string) *FeedBuilder { |
| 134 | return &FeedBuilder{f: AtomFeed{ |
| 135 | XMLNS: atomNamespace, |
| 136 | Title: title, |
| 137 | ID: id, |
| 138 | Updated: time.Now().Format(time.RFC3339), |
| 139 | Authors: []AtomPerson{{Name: defaultAuthor}}, |
| 140 | }} |
| 141 | } |
| 142 | |
| 143 | func (f *FeedBuilder) WithSubtitle(subtitle string) *FeedBuilder { |
| 144 | f.f.Subtitle = subtitle |
| 145 | return f |
| 146 | } |
| 147 | |
| 148 | func (f *FeedBuilder) WithAuthor(name string) *FeedBuilder { |
| 149 | name = strings.TrimSpace(name) |
| 150 | if name == "" { |
| 151 | return f |
| 152 | } |
| 153 | f.f.Authors = []AtomPerson{{Name: name}} |
| 154 | return f |
| 155 | } |
| 156 | |
| 157 | func (f *FeedBuilder) WithUpdated(updated time.Time) *FeedBuilder { |
| 158 | if !updated.IsZero() { |
| 159 | f.f.Updated = updated.Format(time.RFC3339) |
| 160 | } |
| 161 | return f |
| 162 | } |
| 163 | |
| 164 | func (f *FeedBuilder) Add(entry FeedEntry) *FeedBuilder { |
| 165 | if entry.Updated.IsZero() { |
| 166 | entry.Updated = time.Now() |
| 167 | } |
| 168 | if entry.ID == "" { |
| 169 | hash := sha1.Sum(fmt.Appendf(nil, "%s|%s|%s", entry.Title, entry.Content, entry.Updated.Format(time.RFC3339Nano))) |
| 170 | entry.ID = fmt.Sprintf("urn:sha1:%x", hash) |
| 171 | } |
| 172 | |
| 173 | contentType := entry.ContentType |
| 174 | if contentType == "" { |
| 175 | contentType = "text" |
| 176 | } |
| 177 | |
| 178 | links := make([]AtomLink, 0, len(entry.Links)) |
| 179 | for _, link := range entry.Links { |
| 180 | if link.Href == "" { |
| 181 | continue |
| 182 | } |
| 183 | links = append(links, AtomLink(link)) |
| 184 | } |
| 185 | |
| 186 | f.f.Entries = append(f.f.Entries, AtomEntry{ |
| 187 | Title: entry.Title, |
| 188 | ID: entry.ID, |
| 189 | Updated: entry.Updated.Format(time.RFC3339), |
| 190 | Links: links, |
| 191 | Content: AtomContent{ |
| 192 | Type: contentType, |
| 193 | Value: entry.Content, |
| 194 | }, |
| 195 | }) |
| 196 | |
| 197 | feedUpdated, err := time.Parse(time.RFC3339, f.f.Updated) |
| 198 | if err != nil || entry.Updated.After(feedUpdated) { |
| 199 | f.f.Updated = entry.Updated.Format(time.RFC3339) |
| 200 | } |
| 201 | return f |
| 202 | } |
| 203 | |
| 204 | func (f *FeedBuilder) SetUpdated(updated time.Time) *FeedBuilder { |
| 205 | if !updated.IsZero() { |
| 206 | f.f.Updated = updated.Format(time.RFC3339) |
| 207 | } |
| 208 | return f |
| 209 | } |
| 210 | |
| 211 | func (f *FeedBuilder) WriteTo(w io.Writer) error { |
| 212 | enc := xml.NewEncoder(w) |
| 213 | enc.Indent("", " ") |
| 214 | return enc.Encode(f.f) |
| 215 | } |
| 216 | |
| 217 | func (f *FeedBuilder) Bytes() ([]byte, error) { |
| 218 | var buf bytes.Buffer |
| 219 | if err := f.WriteTo(&buf); err != nil { |
| 220 | return nil, err |
| 221 | } |
| 222 | return buf.Bytes(), nil |
| 223 | } |
| 224 | |
| 225 | func (f *FeedBuilder) Render(w http.ResponseWriter) error { |
| 226 | w.Header().Set("Content-Type", "application/atom+xml; charset=utf-8") |
| 227 | return f.WriteTo(w) |
| 228 | } |