rss-tools/sources/telegram/telegram.go (view raw)
| 1 | package telegram |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "context" |
| 6 | "encoding/binary" |
| 7 | "encoding/gob" |
| 8 | "fmt" |
| 9 | "html" |
| 10 | "log/slog" |
| 11 | "net/http" |
| 12 | "strings" |
| 13 | "time" |
| 14 | |
| 15 | "olexsmir.xyz/rss-tools/app" |
| 16 | "olexsmir.xyz/rss-tools/app/atom" |
| 17 | ) |
| 18 | |
| 19 | type telegram struct { |
| 20 | db *app.Bucket |
| 21 | messages *app.Bucket |
| 22 | client *http.Client |
| 23 | get func(context.Context, string) (*http.Response, error) |
| 24 | tg *TelegramSDK |
| 25 | allowedID int64 |
| 26 | logger *slog.Logger |
| 27 | } |
| 28 | |
| 29 | func Register(a *app.App) error { |
| 30 | db, err := a.Bucket("telegram") |
| 31 | if err != nil { |
| 32 | return err |
| 33 | } |
| 34 | |
| 35 | messages, err := a.Bucket("telegram:messages") |
| 36 | if err != nil { |
| 37 | return err |
| 38 | } |
| 39 | |
| 40 | t := &telegram{ |
| 41 | db: db, |
| 42 | messages: messages, |
| 43 | client: a.Client, |
| 44 | get: a.Get, |
| 45 | tg: NewSDK(a.Client, a.Config.TGToken), |
| 46 | allowedID: a.Config.TGUserID, |
| 47 | logger: a.Logger, |
| 48 | } |
| 49 | |
| 50 | a.AddWorker(t.worker) |
| 51 | a.Route("GET /telegram", t.handler) |
| 52 | a.Logger.Info("telegram source registered") |
| 53 | return nil |
| 54 | } |
| 55 | |
| 56 | func (t *telegram) handler(w http.ResponseWriter, r *http.Request) { |
| 57 | // todo: cache feed contruction |
| 58 | // todo: dont include messages older than N days |
| 59 | |
| 60 | messages, err := t.loadMessages(r.Context()) |
| 61 | if err != nil { |
| 62 | http.Error(w, "failed to load messages", http.StatusInternalServerError) |
| 63 | return |
| 64 | } |
| 65 | |
| 66 | feed := atom.NewFeed("Telegram feed", "telegram-feed") |
| 67 | for _, m := range messages { |
| 68 | if changed := t.enrichMessageWithLinkTitles(r.Context(), m); changed { |
| 69 | if err := t.saveMessage(m); err != nil { |
| 70 | http.Error(w, "failed to update cached titles", http.StatusInternalServerError) |
| 71 | return |
| 72 | } |
| 73 | } |
| 74 | } |
| 75 | |
| 76 | for _, m := range groupMessages(messages) { |
| 77 | feed.Add(feedEntryFromMessage(m)) |
| 78 | } |
| 79 | |
| 80 | if err := feed.Render(w); err != nil { |
| 81 | http.Error(w, "failed to render feed", http.StatusInternalServerError) |
| 82 | return |
| 83 | } |
| 84 | } |
| 85 | |
| 86 | func (t *telegram) worker(ctx context.Context) error { |
| 87 | t.logger.Info("starting telegram bot") |
| 88 | |
| 89 | offset, err := t.loadOffset() |
| 90 | if err != nil { |
| 91 | return err |
| 92 | } |
| 93 | |
| 94 | for { |
| 95 | updates, err := t.tg.GetUpdates(ctx, offset) |
| 96 | if err != nil { |
| 97 | t.logger.ErrorContext(ctx, "getUpdates failed", "err", err) |
| 98 | select { |
| 99 | case <-ctx.Done(): |
| 100 | return nil |
| 101 | case <-time.After(5 * time.Second): |
| 102 | continue |
| 103 | } |
| 104 | } |
| 105 | |
| 106 | for _, u := range updates { |
| 107 | if u.Message != nil && u.Message.From != nil { |
| 108 | t.logger.InfoContext(ctx, "message from", "user_id", u.Message.From.ID, "username", u.Message.From.Username, "msg", messageText(u.Message)) |
| 109 | } |
| 110 | |
| 111 | if u.Message == nil || u.Message.From == nil || u.Message.From.ID != t.allowedID { |
| 112 | offset = u.UpdateID + 1 |
| 113 | continue |
| 114 | } |
| 115 | |
| 116 | _ = t.enrichMessageWithLinkTitles(ctx, u.Message) |
| 117 | |
| 118 | if err := t.saveMessage(u.Message); err != nil { |
| 119 | t.logger.ErrorContext(ctx, "failed to save message", "err", err) |
| 120 | } |
| 121 | |
| 122 | if err := t.tg.SetReaction(ctx, u.Message.From.ID, u.Message.MessageID, "👍"); err != nil { |
| 123 | slog.ErrorContext(ctx, "failed to set reaction", "err", err) |
| 124 | } |
| 125 | |
| 126 | offset = u.UpdateID + 1 |
| 127 | } |
| 128 | |
| 129 | if err := t.saveOffset(offset); err != nil { |
| 130 | slog.ErrorContext(ctx, "failed to save offset", "err", err) |
| 131 | } |
| 132 | |
| 133 | select { |
| 134 | case <-ctx.Done(): |
| 135 | return nil |
| 136 | case <-time.After(time.Second): |
| 137 | } |
| 138 | } |
| 139 | } |
| 140 | |
| 141 | func (t *telegram) saveOffset(offset int64) error { |
| 142 | return t.db.Set([]byte("offset"), binary.BigEndian.AppendUint64(nil, uint64(offset))) |
| 143 | } |
| 144 | |
| 145 | func (t *telegram) loadOffset() (int64, error) { |
| 146 | val, err := t.db.Get([]byte("offset")) |
| 147 | if err != nil || val == nil { |
| 148 | return 0, err |
| 149 | } |
| 150 | return int64(binary.BigEndian.Uint64(val)), nil |
| 151 | } |
| 152 | |
| 153 | func (t *telegram) saveMessage(m *Message) error { |
| 154 | var buf bytes.Buffer |
| 155 | if err := gob.NewEncoder(&buf).Encode(m); err != nil { |
| 156 | return err |
| 157 | } |
| 158 | key := binary.BigEndian.AppendUint64(nil, uint64(m.MessageID)) |
| 159 | return t.messages.Set(key, buf.Bytes()) |
| 160 | } |
| 161 | |
| 162 | func (t *telegram) loadMessages(ctx context.Context) ([]*Message, error) { |
| 163 | var messages []*Message |
| 164 | err := t.messages.ForEach(func(k, v []byte) error { |
| 165 | var m Message |
| 166 | if err := gob.NewDecoder(bytes.NewReader(v)).Decode(&m); err != nil { |
| 167 | t.logger.WarnContext(ctx, "failed to decode telegram message, skipping", "key", fmt.Sprintf("%x", k), "err", err) |
| 168 | return nil |
| 169 | } |
| 170 | messages = append(messages, &m) |
| 171 | return nil |
| 172 | }) |
| 173 | return messages, err |
| 174 | } |
| 175 | |
| 176 | func groupMessages(messages []*Message) []*Message { |
| 177 | if len(messages) == 0 { |
| 178 | return messages |
| 179 | } |
| 180 | |
| 181 | groups := make(map[string]*Message) |
| 182 | out := make([]*Message, 0, len(messages)) |
| 183 | for _, m := range messages { |
| 184 | if m == nil || strings.TrimSpace(m.MediaGroupID) == "" { |
| 185 | out = append(out, m) |
| 186 | continue |
| 187 | } |
| 188 | |
| 189 | group, ok := groups[m.MediaGroupID] |
| 190 | if !ok { |
| 191 | group = &Message{ |
| 192 | MessageID: m.MessageID, |
| 193 | From: m.From, |
| 194 | Chat: m.Chat, |
| 195 | Text: m.Text, |
| 196 | Caption: m.Caption, |
| 197 | Date: m.Date, |
| 198 | MediaGroupID: m.MediaGroupID, |
| 199 | LinkTitles: m.LinkTitles, |
| 200 | } |
| 201 | groups[m.MediaGroupID] = group |
| 202 | out = append(out, group) |
| 203 | } |
| 204 | |
| 205 | if m.MessageID != 0 && (group.MessageID == 0 || m.MessageID < group.MessageID) { |
| 206 | group.MessageID = m.MessageID |
| 207 | } |
| 208 | if m.Date != 0 && (group.Date == 0 || m.Date < group.Date) { |
| 209 | group.Date = m.Date |
| 210 | } |
| 211 | if strings.TrimSpace(messageText(group)) == "" && strings.TrimSpace(messageText(m)) != "" { |
| 212 | group.Caption = m.Caption |
| 213 | group.Text = m.Text |
| 214 | if len(m.LinkTitles) > 0 { |
| 215 | group.LinkTitles = m.LinkTitles |
| 216 | } |
| 217 | } else if len(group.LinkTitles) == 0 && len(m.LinkTitles) > 0 { |
| 218 | group.LinkTitles = m.LinkTitles |
| 219 | } |
| 220 | |
| 221 | group.PhotoAttachments = append(group.PhotoAttachments, messagePhotos(m)...) |
| 222 | if group.PhotoBase64 == "" && m.PhotoBase64 != "" { |
| 223 | group.PhotoBase64 = m.PhotoBase64 |
| 224 | group.PhotoMIMEType = m.PhotoMIMEType |
| 225 | } |
| 226 | } |
| 227 | return out |
| 228 | } |
| 229 | |
| 230 | func (t *telegram) enrichMessageWithLinkTitles(ctx context.Context, m *Message) bool { |
| 231 | text := messageText(m) |
| 232 | if !isSingleLinkMessage(text) { |
| 233 | return false |
| 234 | } |
| 235 | |
| 236 | links := normalizeLinks(messageLinks(text)) |
| 237 | if len(links) == 0 { |
| 238 | return false |
| 239 | } |
| 240 | if m.LinkTitles == nil { |
| 241 | m.LinkTitles = make(map[string]string, len(links)) |
| 242 | } |
| 243 | |
| 244 | changed := false |
| 245 | for _, link := range links { |
| 246 | cachedTitle := normalizePageTitle(m.LinkTitles[link]) |
| 247 | if isMeaningfulPageTitle(cachedTitle) { |
| 248 | continue |
| 249 | } |
| 250 | if cachedTitle != "" { |
| 251 | delete(m.LinkTitles, link) |
| 252 | changed = true |
| 253 | } |
| 254 | title, err := fetchPageTitle(ctx, t.get, link) |
| 255 | if err != nil { |
| 256 | t.logger.WarnContext(ctx, "failed to lookup page title", "url", link, "err", err) |
| 257 | continue |
| 258 | } |
| 259 | m.LinkTitles[link] = title |
| 260 | changed = true |
| 261 | } |
| 262 | return changed |
| 263 | } |
| 264 | |
| 265 | func feedEntryFromMessage(m *Message) *atom.Entry { |
| 266 | updated := time.Unix(m.Date, 0) |
| 267 | text := normalizeMessageText(messageText(m)) |
| 268 | normalizedLinks := normalizeLinks(messageLinks(text)) |
| 269 | entryID := fmt.Sprintf("telegram-%d", m.MessageID) |
| 270 | if videoID, ok := firstYouTubeVideoID(normalizedLinks); ok { |
| 271 | entryID = "yt:video:" + videoID |
| 272 | } |
| 273 | |
| 274 | photos := messagePhotos(m) |
| 275 | if len(photos) == 0 { |
| 276 | title := text |
| 277 | if isSingleLinkMessage(text) { |
| 278 | for _, link := range normalizedLinks { |
| 279 | if t := strings.TrimSpace(m.LinkTitles[link]); t != "" { |
| 280 | title = t |
| 281 | break |
| 282 | } |
| 283 | } |
| 284 | } |
| 285 | if len(title) > 64 { |
| 286 | title = title[:64] + "..." |
| 287 | } |
| 288 | |
| 289 | content := text |
| 290 | contentType := "" |
| 291 | if len(normalizedLinks) > 0 { |
| 292 | content, _ = linkifyMessageText(text) |
| 293 | content = preserveLineBreaks(content) |
| 294 | contentType = "html" |
| 295 | } else if strings.Contains(text, "\n") { |
| 296 | content = preserveLineBreaks(html.EscapeString(text)) |
| 297 | contentType = "html" |
| 298 | } |
| 299 | |
| 300 | return &atom.Entry{ |
| 301 | Title: title, |
| 302 | ID: entryID, |
| 303 | Link: feedLinks(normalizedLinks), |
| 304 | Content: atom.NewText(content, contentType), |
| 305 | Updated: atom.Time(updated), |
| 306 | } |
| 307 | } |
| 308 | |
| 309 | parts := make([]string, 0, 1+len(photos)) |
| 310 | if t := strings.TrimSpace(text); t != "" { |
| 311 | linkified, _ := linkifyMessageText(text) |
| 312 | linkified = preserveLineBreaks(linkified) |
| 313 | parts = append(parts, "<p>"+linkified+"</p>") |
| 314 | } |
| 315 | for _, photo := range photos { |
| 316 | if photo.Base64 == "" { |
| 317 | continue |
| 318 | } |
| 319 | mimeType := photo.MIMEType |
| 320 | if mimeType == "" { |
| 321 | mimeType = "image/jpeg" |
| 322 | } |
| 323 | parts = append(parts, fmt.Sprintf(`<p><img src="data:%s;base64,%s" alt="telegram image"/></p>`, mimeType, photo.Base64)) |
| 324 | } |
| 325 | |
| 326 | return &atom.Entry{ |
| 327 | Title: fmt.Sprintf("🖼️ [%s]", updated.Format("2006-01-02")), |
| 328 | ID: entryID, |
| 329 | Link: feedLinks(normalizedLinks), |
| 330 | Content: atom.NewText(strings.Join(parts, ""), "html"), |
| 331 | Updated: atom.Time(updated), |
| 332 | } |
| 333 | } |
| 334 | |
| 335 | func isSingleLinkMessage(text string) bool { |
| 336 | links := findLinks(text) |
| 337 | if len(links) != 1 { |
| 338 | return false |
| 339 | } |
| 340 | link := links[0] |
| 341 | if strings.TrimSpace(text[:link.start]) != "" { |
| 342 | return false |
| 343 | } |
| 344 | after := strings.TrimSpace(text[link.end:]) |
| 345 | return trailingPunctRe.ReplaceAllString(after, "") == "" |
| 346 | } |
| 347 | |
| 348 | func messageText(m *Message) string { |
| 349 | if m == nil { |
| 350 | return "" |
| 351 | } |
| 352 | if caption := strings.TrimSpace(m.Caption); caption != "" { |
| 353 | return m.Caption |
| 354 | } |
| 355 | return m.Text |
| 356 | } |
| 357 | |
| 358 | func normalizeMessageText(text string) string { |
| 359 | text = strings.ReplaceAll(text, "\r\n", "\n") |
| 360 | return strings.ReplaceAll(text, "\r", "\n") |
| 361 | } |
| 362 | |
| 363 | func preserveLineBreaks(text string) string { |
| 364 | if !strings.Contains(text, "\n") { |
| 365 | return text |
| 366 | } |
| 367 | return strings.ReplaceAll(text, "\n", "<br/>") |
| 368 | } |
| 369 | |
| 370 | func messagePhotos(m *Message) []PhotoAttachment { |
| 371 | if m == nil { |
| 372 | return nil |
| 373 | } |
| 374 | if len(m.PhotoAttachments) > 0 { |
| 375 | out := make([]PhotoAttachment, len(m.PhotoAttachments)) |
| 376 | copy(out, m.PhotoAttachments) |
| 377 | return out |
| 378 | } |
| 379 | if m.PhotoBase64 == "" { |
| 380 | return nil |
| 381 | } |
| 382 | mimeType := m.PhotoMIMEType |
| 383 | if mimeType == "" { |
| 384 | mimeType = "image/jpeg" |
| 385 | } |
| 386 | return []PhotoAttachment{{Base64: m.PhotoBase64, MIMEType: mimeType}} |
| 387 | } |