source: code/trunk/msgstore_fs.go@ 593

Last change on this file since 593 was 591, checked in by hubert, 4 years ago

msgstore_fs: escape "." and ".."

File size: 13.8 KB
RevLine 
[439]1package soju
2
3import (
4 "bufio"
5 "fmt"
6 "io"
7 "os"
8 "path/filepath"
[549]9 "sort"
[439]10 "strings"
11 "time"
12
[488]13 "git.sr.ht/~sircmpwn/go-bare"
[439]14 "gopkg.in/irc.v3"
15)
16
17const fsMessageStoreMaxTries = 100
18
[591]19func escapeFilename(unsafe string) (safe string) {
20 if unsafe == "." {
21 return "-"
22 } else if unsafe == ".." {
23 return "--"
24 } else {
25 return strings.NewReplacer("/", "-", "\\", "-").Replace(unsafe)
26 }
27}
[439]28
[488]29type date struct {
30 Year, Month, Day int
31}
32
33func newDate(t time.Time) date {
34 year, month, day := t.Date()
35 return date{year, int(month), day}
36}
37
38func (d date) Time() time.Time {
39 return time.Date(d.Year, time.Month(d.Month), d.Day, 0, 0, 0, 0, time.Local)
40}
41
42type fsMsgID struct {
43 Date date
44 Offset bare.Int
45}
46
47func (fsMsgID) msgIDType() msgIDType {
48 return msgIDFS
49}
50
51func parseFSMsgID(s string) (netID int64, entity string, t time.Time, offset int64, err error) {
52 var id fsMsgID
53 netID, entity, err = parseMsgID(s, &id)
54 if err != nil {
55 return 0, "", time.Time{}, 0, err
56 }
57 return netID, entity, id.Date.Time(), int64(id.Offset), nil
58}
59
60func formatFSMsgID(netID int64, entity string, t time.Time, offset int64) string {
61 id := fsMsgID{
62 Date: newDate(t),
63 Offset: bare.Int(offset),
64 }
65 return formatMsgID(netID, entity, &id)
66}
67
[439]68// fsMessageStore is a per-user on-disk store for IRC messages.
69type fsMessageStore struct {
70 root string
71
72 files map[string]*os.File // indexed by entity
73}
74
[517]75var _ messageStore = (*fsMessageStore)(nil)
76var _ chatHistoryMessageStore = (*fsMessageStore)(nil)
77
[439]78func newFSMessageStore(root, username string) *fsMessageStore {
79 return &fsMessageStore{
[591]80 root: filepath.Join(root, escapeFilename(username)),
[439]81 files: make(map[string]*os.File),
82 }
83}
84
85func (ms *fsMessageStore) logPath(network *network, entity string, t time.Time) string {
86 year, month, day := t.Date()
87 filename := fmt.Sprintf("%04d-%02d-%02d.log", year, month, day)
[591]88 return filepath.Join(ms.root, escapeFilename(network.GetName()), escapeFilename(entity), filename)
[439]89}
90
91// nextMsgID queries the message ID for the next message to be written to f.
[440]92func nextFSMsgID(network *network, entity string, t time.Time, f *os.File) (string, error) {
[439]93 offset, err := f.Seek(0, io.SeekEnd)
94 if err != nil {
[515]95 return "", fmt.Errorf("failed to query next FS message ID: %v", err)
[439]96 }
[440]97 return formatFSMsgID(network.ID, entity, t, offset), nil
[439]98}
99
100func (ms *fsMessageStore) LastMsgID(network *network, entity string, t time.Time) (string, error) {
101 p := ms.logPath(network, entity, t)
102 fi, err := os.Stat(p)
103 if os.IsNotExist(err) {
[440]104 return formatFSMsgID(network.ID, entity, t, -1), nil
[439]105 } else if err != nil {
[515]106 return "", fmt.Errorf("failed to query last FS message ID: %v", err)
[439]107 }
[440]108 return formatFSMsgID(network.ID, entity, t, fi.Size()-1), nil
[439]109}
110
111func (ms *fsMessageStore) Append(network *network, entity string, msg *irc.Message) (string, error) {
112 s := formatMessage(msg)
113 if s == "" {
114 return "", nil
115 }
116
117 var t time.Time
118 if tag, ok := msg.Tags["time"]; ok {
119 var err error
120 t, err = time.Parse(serverTimeLayout, string(tag))
121 if err != nil {
122 return "", fmt.Errorf("failed to parse message time tag: %v", err)
123 }
124 t = t.In(time.Local)
125 } else {
126 t = time.Now()
127 }
128
129 // TODO: enforce maximum open file handles (LRU cache of file handles)
130 f := ms.files[entity]
131
132 // TODO: handle non-monotonic clock behaviour
133 path := ms.logPath(network, entity, t)
134 if f == nil || f.Name() != path {
135 if f != nil {
136 f.Close()
137 }
138
139 dir := filepath.Dir(path)
[558]140 if err := os.MkdirAll(dir, 0750); err != nil {
[439]141 return "", fmt.Errorf("failed to create message logs directory %q: %v", dir, err)
142 }
143
144 var err error
[558]145 f, err = os.OpenFile(path, os.O_RDWR|os.O_CREATE|os.O_APPEND, 0640)
[439]146 if err != nil {
147 return "", fmt.Errorf("failed to open message log file %q: %v", path, err)
148 }
149
150 ms.files[entity] = f
151 }
152
[440]153 msgID, err := nextFSMsgID(network, entity, t, f)
[439]154 if err != nil {
155 return "", fmt.Errorf("failed to generate message ID: %v", err)
156 }
157
158 _, err = fmt.Fprintf(f, "[%02d:%02d:%02d] %s\n", t.Hour(), t.Minute(), t.Second(), s)
159 if err != nil {
160 return "", fmt.Errorf("failed to log message to %q: %v", f.Name(), err)
161 }
162
163 return msgID, nil
164}
165
166func (ms *fsMessageStore) Close() error {
167 var closeErr error
168 for _, f := range ms.files {
169 if err := f.Close(); err != nil {
170 closeErr = fmt.Errorf("failed to close message store: %v", err)
171 }
172 }
173 return closeErr
174}
175
176// formatMessage formats a message log line. It assumes a well-formed IRC
177// message.
178func formatMessage(msg *irc.Message) string {
179 switch strings.ToUpper(msg.Command) {
180 case "NICK":
181 return fmt.Sprintf("*** %s is now known as %s", msg.Prefix.Name, msg.Params[0])
182 case "JOIN":
183 return fmt.Sprintf("*** Joins: %s (%s@%s)", msg.Prefix.Name, msg.Prefix.User, msg.Prefix.Host)
184 case "PART":
185 var reason string
186 if len(msg.Params) > 1 {
187 reason = msg.Params[1]
188 }
189 return fmt.Sprintf("*** Parts: %s (%s@%s) (%s)", msg.Prefix.Name, msg.Prefix.User, msg.Prefix.Host, reason)
190 case "KICK":
191 nick := msg.Params[1]
192 var reason string
193 if len(msg.Params) > 2 {
194 reason = msg.Params[2]
195 }
196 return fmt.Sprintf("*** %s was kicked by %s (%s)", nick, msg.Prefix.Name, reason)
197 case "QUIT":
198 var reason string
199 if len(msg.Params) > 0 {
200 reason = msg.Params[0]
201 }
202 return fmt.Sprintf("*** Quits: %s (%s@%s) (%s)", msg.Prefix.Name, msg.Prefix.User, msg.Prefix.Host, reason)
203 case "TOPIC":
204 var topic string
205 if len(msg.Params) > 1 {
206 topic = msg.Params[1]
207 }
208 return fmt.Sprintf("*** %s changes topic to '%s'", msg.Prefix.Name, topic)
209 case "MODE":
210 return fmt.Sprintf("*** %s sets mode: %s", msg.Prefix.Name, strings.Join(msg.Params[1:], " "))
211 case "NOTICE":
212 return fmt.Sprintf("-%s- %s", msg.Prefix.Name, msg.Params[1])
213 case "PRIVMSG":
214 if cmd, params, ok := parseCTCPMessage(msg); ok && cmd == "ACTION" {
215 return fmt.Sprintf("* %s %s", msg.Prefix.Name, params)
216 } else {
217 return fmt.Sprintf("<%s> %s", msg.Prefix.Name, msg.Params[1])
218 }
219 default:
220 return ""
221 }
222}
223
224func parseMessage(line, entity string, ref time.Time) (*irc.Message, time.Time, error) {
225 var hour, minute, second int
226 _, err := fmt.Sscanf(line, "[%02d:%02d:%02d] ", &hour, &minute, &second)
227 if err != nil {
[515]228 return nil, time.Time{}, fmt.Errorf("malformed timestamp prefix: %v", err)
[439]229 }
230 line = line[11:]
231
232 var cmd, sender, text string
233 if strings.HasPrefix(line, "<") {
234 cmd = "PRIVMSG"
235 parts := strings.SplitN(line[1:], "> ", 2)
236 if len(parts) != 2 {
237 return nil, time.Time{}, nil
238 }
239 sender, text = parts[0], parts[1]
240 } else if strings.HasPrefix(line, "-") {
241 cmd = "NOTICE"
242 parts := strings.SplitN(line[1:], "- ", 2)
243 if len(parts) != 2 {
244 return nil, time.Time{}, nil
245 }
246 sender, text = parts[0], parts[1]
247 } else if strings.HasPrefix(line, "* ") {
248 cmd = "PRIVMSG"
249 parts := strings.SplitN(line[2:], " ", 2)
250 if len(parts) != 2 {
251 return nil, time.Time{}, nil
252 }
253 sender, text = parts[0], "\x01ACTION "+parts[1]+"\x01"
254 } else {
255 return nil, time.Time{}, nil
256 }
257
258 year, month, day := ref.Date()
259 t := time.Date(year, month, day, hour, minute, second, 0, time.Local)
260
261 msg := &irc.Message{
262 Tags: map[string]irc.TagValue{
263 "time": irc.TagValue(t.UTC().Format(serverTimeLayout)),
264 },
265 Prefix: &irc.Prefix{Name: sender},
266 Command: cmd,
267 Params: []string{entity, text},
268 }
269 return msg, t, nil
270}
271
[516]272func (ms *fsMessageStore) parseMessagesBefore(network *network, entity string, ref time.Time, end time.Time, limit int, afterOffset int64) ([]*irc.Message, error) {
[439]273 path := ms.logPath(network, entity, ref)
274 f, err := os.Open(path)
275 if err != nil {
276 if os.IsNotExist(err) {
277 return nil, nil
278 }
[515]279 return nil, fmt.Errorf("failed to parse messages before ref: %v", err)
[439]280 }
281 defer f.Close()
282
283 historyRing := make([]*irc.Message, limit)
284 cur := 0
285
286 sc := bufio.NewScanner(f)
287
288 if afterOffset >= 0 {
289 if _, err := f.Seek(afterOffset, io.SeekStart); err != nil {
290 return nil, nil
291 }
292 sc.Scan() // skip till next newline
293 }
294
295 for sc.Scan() {
296 msg, t, err := parseMessage(sc.Text(), entity, ref)
297 if err != nil {
298 return nil, err
[516]299 } else if msg == nil || !t.After(end) {
[439]300 continue
301 } else if !t.Before(ref) {
302 break
303 }
304
305 historyRing[cur%limit] = msg
306 cur++
307 }
308 if sc.Err() != nil {
[515]309 return nil, fmt.Errorf("failed to parse messages before ref: scanner error: %v", sc.Err())
[439]310 }
311
312 n := limit
313 if cur < limit {
314 n = cur
315 }
316 start := (cur - n + limit) % limit
317
318 if start+n <= limit { // ring doesnt wrap
319 return historyRing[start : start+n], nil
320 } else { // ring wraps
321 history := make([]*irc.Message, n)
322 r := copy(history, historyRing[start:])
323 copy(history[r:], historyRing[:n-r])
324 return history, nil
325 }
326}
327
[516]328func (ms *fsMessageStore) parseMessagesAfter(network *network, entity string, ref time.Time, end time.Time, limit int) ([]*irc.Message, error) {
[439]329 path := ms.logPath(network, entity, ref)
330 f, err := os.Open(path)
331 if err != nil {
332 if os.IsNotExist(err) {
333 return nil, nil
334 }
[515]335 return nil, fmt.Errorf("failed to parse messages after ref: %v", err)
[439]336 }
337 defer f.Close()
338
339 var history []*irc.Message
340 sc := bufio.NewScanner(f)
341 for sc.Scan() && len(history) < limit {
342 msg, t, err := parseMessage(sc.Text(), entity, ref)
343 if err != nil {
344 return nil, err
345 } else if msg == nil || !t.After(ref) {
346 continue
[516]347 } else if !t.Before(end) {
348 break
[439]349 }
350
351 history = append(history, msg)
352 }
353 if sc.Err() != nil {
[515]354 return nil, fmt.Errorf("failed to parse messages after ref: scanner error: %v", sc.Err())
[439]355 }
356
357 return history, nil
358}
359
[516]360func (ms *fsMessageStore) LoadBeforeTime(network *network, entity string, start time.Time, end time.Time, limit int) ([]*irc.Message, error) {
[439]361 history := make([]*irc.Message, limit)
362 remaining := limit
363 tries := 0
[516]364 for remaining > 0 && tries < fsMessageStoreMaxTries && end.Before(start) {
365 buf, err := ms.parseMessagesBefore(network, entity, start, end, remaining, -1)
[439]366 if err != nil {
367 return nil, err
368 }
369 if len(buf) == 0 {
370 tries++
371 } else {
372 tries = 0
373 }
374 copy(history[remaining-len(buf):], buf)
375 remaining -= len(buf)
[516]376 year, month, day := start.Date()
377 start = time.Date(year, month, day, 0, 0, 0, 0, start.Location()).Add(-1)
[439]378 }
379
380 return history[remaining:], nil
381}
382
[516]383func (ms *fsMessageStore) LoadAfterTime(network *network, entity string, start time.Time, end time.Time, limit int) ([]*irc.Message, error) {
[439]384 var history []*irc.Message
385 remaining := limit
386 tries := 0
[516]387 for remaining > 0 && tries < fsMessageStoreMaxTries && start.Before(end) {
388 buf, err := ms.parseMessagesAfter(network, entity, start, end, remaining)
[439]389 if err != nil {
390 return nil, err
391 }
392 if len(buf) == 0 {
393 tries++
394 } else {
395 tries = 0
396 }
397 history = append(history, buf...)
398 remaining -= len(buf)
[516]399 year, month, day := start.Date()
400 start = time.Date(year, month, day+1, 0, 0, 0, 0, start.Location())
[439]401 }
402 return history, nil
403}
404
405func (ms *fsMessageStore) LoadLatestID(network *network, entity, id string, limit int) ([]*irc.Message, error) {
406 var afterTime time.Time
407 var afterOffset int64
408 if id != "" {
[440]409 var idNet int64
410 var idEntity string
[439]411 var err error
[440]412 idNet, idEntity, afterTime, afterOffset, err = parseFSMsgID(id)
[439]413 if err != nil {
414 return nil, err
415 }
[440]416 if idNet != network.ID || idEntity != entity {
[439]417 return nil, fmt.Errorf("cannot find message ID: message ID doesn't match network/entity")
418 }
419 }
420
421 history := make([]*irc.Message, limit)
422 t := time.Now()
423 remaining := limit
424 tries := 0
425 for remaining > 0 && tries < fsMessageStoreMaxTries && !truncateDay(t).Before(afterTime) {
426 var offset int64 = -1
427 if afterOffset >= 0 && truncateDay(t).Equal(afterTime) {
428 offset = afterOffset
429 }
430
[516]431 buf, err := ms.parseMessagesBefore(network, entity, t, time.Time{}, remaining, offset)
[439]432 if err != nil {
433 return nil, err
434 }
435 if len(buf) == 0 {
436 tries++
437 } else {
438 tries = 0
439 }
440 copy(history[remaining-len(buf):], buf)
441 remaining -= len(buf)
442 year, month, day := t.Date()
443 t = time.Date(year, month, day, 0, 0, 0, 0, t.Location()).Add(-1)
444 }
445
446 return history[remaining:], nil
447}
[549]448
449func (ms *fsMessageStore) ListTargets(network *network, start, end time.Time, limit int) ([]chatHistoryTarget, error) {
[591]450 rootPath := filepath.Join(ms.root, escapeFilename(network.GetName()))
[549]451 root, err := os.Open(rootPath)
452 if err != nil {
453 return nil, err
454 }
455
456 // The returned targets are escaped, and there is no way to un-escape
457 // TODO: switch to ReadDir (Go 1.16+)
458 targetNames, err := root.Readdirnames(0)
459 root.Close()
460 if err != nil {
461 return nil, err
462 }
463
464 var targets []chatHistoryTarget
465 for _, target := range targetNames {
466 // target is already escaped here
467 targetPath := filepath.Join(rootPath, target)
468 targetDir, err := os.Open(targetPath)
469 if err != nil {
470 return nil, err
471 }
472
473 entries, err := targetDir.Readdir(0)
474 targetDir.Close()
475 if err != nil {
476 return nil, err
477 }
478
479 // We use mtime here, which may give imprecise or incorrect results
480 var t time.Time
481 for _, entry := range entries {
482 if entry.ModTime().After(t) {
483 t = entry.ModTime()
484 }
485 }
486
487 // The timestamps we get from logs have second granularity
488 t = truncateSecond(t)
489
490 // Filter out targets that don't fullfil the time bounds
491 if !isTimeBetween(t, start, end) {
492 continue
493 }
494
495 targets = append(targets, chatHistoryTarget{
496 Name: target,
497 LatestMessage: t,
498 })
499 }
500
501 // Sort targets by latest message time, backwards or forwards depending on
502 // the order of the time bounds
503 sort.Slice(targets, func(i, j int) bool {
504 t1, t2 := targets[i].LatestMessage, targets[j].LatestMessage
505 if start.Before(end) {
506 return t1.Before(t2)
507 } else {
508 return !t1.Before(t2)
509 }
510 })
511
512 // Truncate the result if necessary
513 if len(targets) > limit {
514 targets = targets[:limit]
515 }
516
517 return targets, nil
518}
519
520func truncateDay(t time.Time) time.Time {
521 year, month, day := t.Date()
522 return time.Date(year, month, day, 0, 0, 0, 0, t.Location())
523}
524
525func truncateSecond(t time.Time) time.Time {
526 year, month, day := t.Date()
527 return time.Date(year, month, day, t.Hour(), t.Minute(), t.Second(), 0, t.Location())
528}
529
530func isTimeBetween(t, start, end time.Time) bool {
531 if end.Before(start) {
532 end, start = start, end
533 }
534 return start.Before(t) && t.Before(end)
535}
Note: See TracBrowser for help on using the repository browser.