1package webmail

2

3import (

4 "bufio"

5 "fmt"

6 "io"

7 "log/slog"

8 "mime"

9 "net/url"

10 "strings"

11

12 "golang.org/x/text/encoding/ianaindex"

13

14 "github.com/mjl-/mox/dns"

15 "github.com/mjl-/mox/message"

16 "github.com/mjl-/mox/mlog"

17 "github.com/mjl-/mox/mox-"

18 "github.com/mjl-/mox/moxio"

19 "github.com/mjl-/mox/smtp"

20 "github.com/mjl-/mox/store"

21)

22

23// todo: we should have all needed information for messageItem in store.Message (perhaps some data in message.Part) for fast access, not having to parse the on-disk message file.

24

25var wordDecoder = mime.WordDecoder{

26 CharsetReader: func(charset string, r io.Reader) (io.Reader, error) {

27 switch strings.ToLower(charset) {

28 case "", "us-ascii", "utf-8":

29 return r, nil

30 }

31 enc, _ := ianaindex.MIME.Encoding(charset)

32 if enc == nil {

33 enc, _ = ianaindex.IANA.Encoding(charset)

34 }

35 if enc == nil {

36 return r, fmt.Errorf("unknown charset %q", charset)

37 }

38 return enc.NewDecoder().Reader(r), nil

39 },

40}

41

42// Attempt q/b-word-decode name, coming from Content-Type "name" field or

43// Content-Disposition "filename" field.

44//

45// RFC 2231 specify an encoding for non-ascii values in mime header parameters. But

46// it appears common practice to instead just q/b-word encode the values.

47// Thunderbird and gmail.com do this for the Content-Type "name" parameter.

48// gmail.com also does that for the Content-Disposition "filename" parameter, where

49// Thunderbird uses the RFC 2231-defined encoding. Go's mime.ParseMediaType parses

50// the mechanism specified in RFC 2231 only. The value for "name" we get here would

51// already be decoded properly for standards-compliant headers, like

52// "filename*0*=UTF-8”%...; filename*1*=%.... We'll look for Q/B-word encoding

53// markers ("=?"-prefix or "?="-suffix) and try to decode if present. This would

54// only cause trouble for filenames having this prefix/suffix.

55func tryDecodeParam(log mlog.Log, name string) string {

56 if name == "" || !strings.HasPrefix(name, "=?") && !strings.HasSuffix(name, "?=") {

57 return name

58 }

59 // todo: find where this is allowed. it seems quite common. perhaps we should remove the pedantic check?

60 if mox.Pedantic {

61 log.Debug("attachment contains rfc2047 q/b-word-encoded mime parameter instead of rfc2231-encoded", slog.String("name", name))

62 return name

63 }

64 s, err := wordDecoder.DecodeHeader(name)

65 if err != nil {

66 log.Debugx("q/b-word decoding mime parameter", err, slog.String("name", name))

67 return name

68 }

69 return s

70}

71

72// todo: mime.FormatMediaType does not wrap long lines. should do it ourselves, and split header into several parts (if commonly supported).

73

74func messageItem(log mlog.Log, m store.Message, state *msgState) (MessageItem, error) {

75 pm, err := parsedMessage(log, m, state, false, true)

76 if err != nil {

77 return MessageItem{}, fmt.Errorf("parsing message %d for item: %v", m.ID, err)

78 }

79 // Clear largish unused data.

80 m.MsgPrefix = nil

81 m.ParsedBuf = nil

82 return MessageItem{m, pm.envelope, pm.attachments, pm.isSigned, pm.isEncrypted, pm.firstLine, true}, nil

83}

84

85// formatFirstLine returns a line the client can display next to the subject line

86// in a mailbox. It will replace quoted text, and any prefixing "On ... write:"

87// line with "[...]" so only new and useful information will be displayed.

88// Trailing signatures are not included.

89func formatFirstLine(r io.Reader) (string, error) {

90 // We look quite a bit of lines ahead for trailing signatures with trailing empty lines.

91 var lines []string

92 scanner := bufio.NewScanner(r)

93 ensureLines := func() {

94 for len(lines) < 10 && scanner.Scan() {

95 lines = append(lines, strings.TrimSpace(scanner.Text()))

96 }

97 }

98 ensureLines()

99

100 isSnipped := func(s string) bool {

101 return s == "[...]" || s == "[…]" || s == "..."

102 }

103

104 nextLineQuoted := func(i int) bool {

105 if i+1 < len(lines) && lines[i+1] == "" {

106 i++

107 }

108 return i+1 < len(lines) && (strings.HasPrefix(lines[i+1], ">") || isSnipped(lines[i+1]))

109 }

110

111 // Remainder is signature if we see a line with only and minimum 2 dashes, and

112 // there are no more empty lines, and there aren't more than 5 lines left.

113 isSignature := func() bool {

114 if len(lines) == 0 || !strings.HasPrefix(lines[0], "--") || strings.Trim(strings.TrimSpace(lines[0]), "-") != "" {

115 return false

116 }

117 l := lines[1:]

118 for len(l) > 0 && l[len(l)-1] == "" {

119 l = l[:len(l)-1]

120 }

121 if len(l) >= 5 {

122 return false

123 }

124 for _, line := range l {

125 if line == "" {

126 return false

127 }

128 }

129 return true

130 }

131

132 result := ""

133

134 resultSnipped := func() bool {

135 return strings.HasSuffix(result, "[...]\n") || strings.HasSuffix(result, "[…]")

136 }

137

138 // Quick check for initial wrapped "On ... wrote:" line.

139 if len(lines) > 3 && strings.HasPrefix(lines[0], "On ") && !strings.HasSuffix(lines[0], "wrote:") && strings.HasSuffix(lines[1], ":") && nextLineQuoted(1) {

140 result = "[...]\n"

141 lines = lines[3:]

142 ensureLines()

143 }

144

145 for ; len(lines) > 0 && !isSignature(); ensureLines() {

146 line := lines[0]

147 if strings.HasPrefix(line, ">") {

148 if !resultSnipped() {

149 result += "[...]\n"

150 }

151 lines = lines[1:]

152 continue

153 }

154 if line == "" {

155 lines = lines[1:]

156 continue

157 }

158 // Check for a "On <date>, <person> wrote:", we require digits before a quoted

159 // line, with an optional empty line in between. If we don't have any text yet, we

160 // don't require the digits.

161 if strings.HasSuffix(line, ":") && (strings.ContainsAny(line, "0123456789") || result == "") && nextLineQuoted(0) {

162 if !resultSnipped() {

163 result += "[...]\n"

164 }

165 lines = lines[1:]

166 continue

167 }

168 // Skip possibly duplicate snipping by author.

169 if !isSnipped(line) || !resultSnipped() {

170 result += line + "\n"

171 }

172 lines = lines[1:]

173 if len(result) > 250 {

174 break

175 }

176 }

177 if len(result) > 250 {

178 result = result[:230] + "..."

179 }

180 return result, scanner.Err()

181}

182

183func parsedMessage(log mlog.Log, m store.Message, state *msgState, full, msgitem bool) (pm ParsedMessage, rerr error) {

184 if full || msgitem {

185 if !state.ensurePart(m, true) {

186 return pm, state.err

187 }

188 if full {

189 pm.Part = *state.part

190 }

191 } else {

192 if !state.ensurePart(m, false) {

193 return pm, state.err

194 }

195 }

196

197 // todo: we should store this form in message.Part, requires a data structure update.

198

199 convertAddrs := func(l []message.Address) []MessageAddress {

200 r := make([]MessageAddress, len(l))

201 for i, a := range l {

202 d, err := dns.ParseDomain(a.Host)

203 log.Check(err, "parsing domain")

204 if err != nil {

205 d = dns.Domain{ASCII: a.Host}

206 }

207 r[i] = MessageAddress{a.Name, a.User, d}

208 }

209 return r

210 }

211

212 if full || msgitem {

213 env := MessageEnvelope{}

214 if state.part.Envelope != nil {

215 e := *state.part.Envelope

216 env.Date = e.Date

217 env.Subject = e.Subject

218 env.InReplyTo = e.InReplyTo

219 env.MessageID = e.MessageID

220 env.From = convertAddrs(e.From)

221 env.Sender = convertAddrs(e.Sender)

222 env.ReplyTo = convertAddrs(e.ReplyTo)

223 env.To = convertAddrs(e.To)

224 env.CC = convertAddrs(e.CC)

225 env.BCC = convertAddrs(e.BCC)

226 }

227 pm.envelope = env

228 }

229

230 if full && state.part.BodyOffset > 0 {

231 hdrs, err := state.part.Header()

232 if err != nil {

233 return ParsedMessage{}, fmt.Errorf("parsing headers: %v", err)

234 }

235 pm.Headers = hdrs

236

237 pm.ListReplyAddress = parseListPostAddress(hdrs.Get("List-Post"))

238 } else {

239 pm.Headers = map[string][]string{}

240 }

241

242 pm.Texts = []string{}

243

244 // We track attachments from multipart/mixed differently from other attachments.

245 // The others are often inline, sometimes just some logo's in HTML alternative

246 // messages. We want to have our mixed attachments at the start of the list, but

247 // our descent-first parsing would result in inline messages first in the typical

248 // message.

249 var attachmentsMixed []Attachment

250 var attachmentsOther []Attachment

251

252 addAttachment := func(a Attachment, isMixed bool) {

253 if isMixed {

254 attachmentsMixed = append(attachmentsMixed, a)

255 } else {

256 attachmentsOther = append(attachmentsOther, a)

257 }

258 }

259

260 // todo: how should we handle messages where a user prefers html, and we want to show it, but it's a DSN that also has textual-only parts? e.g. gmail's dsn where the first part is multipart/related with multipart/alternative, and second part is the regular message/delivery-status. we want to display both the html and the text.

261

262 var usePart func(p message.Part, index int, parent *message.Part, path []int, parentMixed bool)

263 usePart = func(p message.Part, index int, parent *message.Part, path []int, parentMixed bool) {

264 mt := p.MediaType + "/" + p.MediaSubType

265 newParentMixed := mt == "MULTIPART/MIXED"

266 for i, sp := range p.Parts {

267 if mt == "MULTIPART/SIGNED" && i >= 1 {

268 continue

269 }

270 usePart(sp, i, &p, append(append([]int{}, path...), i), newParentMixed)

271 }

272 switch mt {

273 case "TEXT/PLAIN", "/":

274 // Don't include if Content-Disposition attachment.

275 if full || msgitem {

276 // todo: should have this, and perhaps all content-* headers, preparsed in message.Part?

277 h, err := p.Header()

278 log.Check(err, "parsing attachment headers", slog.Int64("msgid", m.ID))

279 cp := h.Get("Content-Disposition")

280 if cp != "" {

281 disp, params, err := mime.ParseMediaType(cp)

282 log.Check(err, "parsing content-disposition", slog.String("cp", cp))

283 if strings.EqualFold(disp, "attachment") {

284 name := tryDecodeParam(log, p.ContentTypeParams["name"])

285 if name == "" {

286 name = tryDecodeParam(log, params["filename"])

287 }

288 addAttachment(Attachment{path, name, p}, parentMixed)

289 return

290 }

291 }

292 }

293

294 if full {

295 buf, err := io.ReadAll(&moxio.LimitReader{R: p.ReaderUTF8OrBinary(), Limit: 2 * 1024 * 1024})

296 if err != nil {

297 rerr = fmt.Errorf("reading text part: %v", err)

298 return

299 }

300 pm.Texts = append(pm.Texts, string(buf))

301 }

302 if msgitem && pm.firstLine == "" {

303 pm.firstLine, rerr = formatFirstLine(p.ReaderUTF8OrBinary())

304 if rerr != nil {

305 rerr = fmt.Errorf("reading text for first line snippet: %v", rerr)

306 return

307 }

308 }

309

310 case "TEXT/HTML":

311 pm.HasHTML = true

312

313 default:

314 // todo: see if there is a common nesting messages that are both signed and encrypted.

315 if parent == nil && mt == "MULTIPART/SIGNED" {

316 pm.isSigned = true

317 }

318 if parent == nil && mt == "MULTIPART/ENCRYPTED" {

319 pm.isEncrypted = true

320 }

321 // todo: possibly do not include anything below multipart/alternative that starts with text/html, they may be cids. perhaps have a separate list of attachments for the text vs html version?

322 if p.MediaType != "MULTIPART" {

323 var parentct string

324 if parent != nil {

325 parentct = parent.MediaType + "/" + parent.MediaSubType

326 }

327

328 // Recognize DSNs.

329 if parentct == "MULTIPART/REPORT" && index == 1 && (mt == "MESSAGE/GLOBAL-DELIVERY-STATUS" || mt == "MESSAGE/DELIVERY-STATUS") {

330 if full {

331 buf, err := io.ReadAll(&moxio.LimitReader{R: p.ReaderUTF8OrBinary(), Limit: 1024 * 1024})

332 if err != nil {

333 rerr = fmt.Errorf("reading text part: %v", err)

334 return

335 }

336 pm.Texts = append(pm.Texts, string(buf))

337 }

338 return

339 }

340 if parentct == "MULTIPART/REPORT" && index == 2 && (mt == "MESSAGE/GLOBAL-HEADERS" || mt == "TEXT/RFC822-HEADERS") {

341 if full {

342 buf, err := io.ReadAll(&moxio.LimitReader{R: p.ReaderUTF8OrBinary(), Limit: 1024 * 1024})

343 if err != nil {

344 rerr = fmt.Errorf("reading text part: %v", err)

345 return

346 }

347 pm.Texts = append(pm.Texts, string(buf))

348 }

349 return

350 }

351 if parentct == "MULTIPART/REPORT" && index == 2 && (mt == "MESSAGE/GLOBAL" || mt == "TEXT/RFC822") {

352 addAttachment(Attachment{path, "original.eml", p}, parentMixed)

353 return

354 }

355

356 name := tryDecodeParam(log, p.ContentTypeParams["name"])

357 if name == "" && (full || msgitem) {

358 // todo: should have this, and perhaps all content-* headers, preparsed in message.Part?

359 h, err := p.Header()

360 log.Check(err, "parsing attachment headers", slog.Int64("msgid", m.ID))

361 cp := h.Get("Content-Disposition")

362 if cp != "" {

363 _, params, err := mime.ParseMediaType(cp)

364 log.Check(err, "parsing content-disposition", slog.String("cp", cp))

365 name = tryDecodeParam(log, params["filename"])

366 }

367 }

368 addAttachment(Attachment{path, name, p}, parentMixed)

369 }

370 }

371 }

372 usePart(*state.part, -1, nil, []int{}, false)

373

374 pm.attachments = []Attachment{}

375 pm.attachments = append(pm.attachments, attachmentsMixed...)

376 pm.attachments = append(pm.attachments, attachmentsOther...)

377

378 if rerr == nil {

379 pm.ID = m.ID

380 }

381 return

382}

383

384// parses List-Post header, returning an address if it could be found, and nil otherwise.

385func parseListPostAddress(s string) *MessageAddress {

386 /*

387 Examples:

388 List-Post: <mailto:list@host.com>

389 List-Post: <mailto:moderator@host.com> (Postings are Moderated)

390 List-Post: <mailto:moderator@host.com?subject=list%20posting>

391 List-Post: NO (posting not allowed on this list)

392 List-Post: <https://groups.google.com/group/golang-dev/post>, <mailto:golang-dev@googlegroups.com>

393 */

394 s = strings.TrimSpace(s)

395 for s != "" {

396 if !strings.HasPrefix(s, "<") {

397 return nil

398 }

399 addr, ns, found := strings.Cut(s[1:], ">")

400 if !found {

401 return nil

402 }

403 if strings.HasPrefix(addr, "mailto:") {

404 u, err := url.Parse(addr)

405 if err != nil {

406 return nil

407 }

408 addr, err := smtp.ParseAddress(u.Opaque)

409 if err != nil {

410 return nil

411 }

412 return &MessageAddress{User: addr.Localpart.String(), Domain: addr.Domain}

413 }

414 s = strings.TrimSpace(ns)

415 s = strings.TrimPrefix(s, ",")

416 s = strings.TrimSpace(s)

417 }

418 return nil

419}

420