1 #lang typed/racket/no-check
10 (require (prefix-in info: "info.rkt"))
18 (define-type Out-Format
22 (define-type Timeline-Order
29 [nick : (Option String)]
32 [mentions : (Listof Peer)]))
35 ([nick : (Option String)]
40 ([status-line : String]
41 [headers : (Listof Bytes)]
42 [body-input : Input-Port]))
44 (: tt-home-dir Path-String)
45 (define tt-home-dir (build-path (expand-user-path "~") ".tt"))
47 (: concurrent-filter-map (∀ (α β) (-> Natural (-> α β) (Listof α) (Listof β))))
48 (define (concurrent-filter-map num-workers f xs)
49 ; TODO preserve order of elements OR communicate that reorder is expected
50 ; TODO switch from mailboxes to channels
51 (define (make-worker id f)
52 (define parent (current-thread))
54 (define self : Thread (current-thread))
55 (: work (∀ (α) (-> α)))
57 (thread-send parent (cons 'next self))
58 (match (thread-receive)
59 ['done (thread-send parent (cons 'exit id))]
60 [(cons 'unit x) (begin
62 (when y (thread-send parent (cons 'result y)))
65 (: dispatch (∀ (α β) (-> (Listof Nonnegative-Integer) (Listof α) (Listof β))))
66 (define (dispatch ws xs ys)
69 (match (thread-receive)
70 [(cons 'exit w) (dispatch (remove w ws =) xs ys)]
71 [(cons 'result y) (dispatch ws xs (cons y ys))]
72 [(cons 'next thd) (match xs
74 (thread-send thd 'done)
77 (thread-send thd (cons 'unit x))
78 (dispatch ws xs ys))])])))
79 (define workers (range num-workers))
80 (define threads (map (λ (id) (thread (make-worker id f))) workers))
81 (define results (dispatch workers xs '()))
82 (for-each thread-wait threads)
86 (let* ([f (λ (x) (if (even? x) x #f))]
88 [actual (sort (concurrent-filter-map 10 f xs) <)]
89 [expected (sort ( filter-map f xs) <)])
90 (check-equal? actual expected "concurrent-filter-map")))
92 (: msg-print (-> Out-Format Integer Msg Void))
94 (let* ([colors (vector 36 33)]
95 [n (vector-length colors)])
96 (λ (out-format color-i msg)
97 (let ([color (vector-ref colors (modulo color-i n))]
99 [uri (url->string (Msg-uri msg))]
100 [text (Msg-text msg)]
101 [mentions (Msg-mentions msg)])
104 (let ([nick (if nick nick uri)])
105 (printf "~a \033[1;37m<~a>\033[0m \033[0;~am~a\033[0m~n"
107 ([date-display-format 'iso-8601])
108 (date->string (seconds->date (Msg-ts-epoch msg)) #t))
111 (let ([nick (if nick (string-append nick " ") "")])
112 (printf "~a (~a)~n\033[1;37m<~a~a>\033[0m~n\033[0;~am~a\033[0m~n~n"
114 ([date-display-format 'rfc2822])
115 (date->string (seconds->date (Msg-ts-epoch msg)) #t))
117 nick uri color text))])))))
119 (: rfc3339->epoch (-> String (Option Nonnegative-Integer)))
120 (define rfc3339->epoch
121 (let ([re (pregexp "^([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2})(:([0-9]{2}))?(\\.[0-9]+)?(Z|([+-])([0-9]{1,2}):?([0-9]{2}))?$")])
123 (match (regexp-match re ts)
124 [(list _wholething yyyy mm dd HH MM _:SS SS _fractional tz-whole tz-sign tz-HH tz-MM)
127 (match* (tz-whole tz-sign tz-HH tz-MM)
130 [(_ (or "-" "+") (? identity h) (? identity m))
131 (let ([h (string->number h)]
132 [m (string->number m)]
133 ; Reverse to get back to UTC:
134 [op (match tz-sign ["+" -] ["-" +])])
135 (op 0 (+ (* 60 m) (* 60 (* 60 h)))))]
137 (log-warning "Impossible TZ string: ~v, components: ~v ~v ~v ~v" tz-whole a b c d)
141 [ts-epoch (find-seconds (if SS (string->number SS) 0)
146 (string->number yyyy)
148 (+ ts-epoch tz-offset))]
150 (log-error "Invalid timestamp: ~v" ts)
153 (: str->msg (-> (Option String) Url String (Option Msg)))
155 (let ([re (pregexp "^([^\\s\t]+)[\\s\t]+(.*)$")])
157 (define str-head (substring str 0 (min 100 (string-length str))))
162 "Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v"
163 str-head nick (url->string uri) e)
165 (match (regexp-match re str)
166 [(list _wholething ts-orig text)
167 (let ([ts-epoch (rfc3339->epoch ts-orig)])
171 (λ (m) (match (regexp-match #px"@<([^>]+)>" m)
172 [(list _wholething nick-uri)
173 (str->peer nick-uri)]))
174 (regexp-match* #px"@<[^\\s]+([\\s]+)?[^>]+>" text))])
175 (Msg ts-epoch ts-orig nick uri text mentions))
178 "Msg rejected due to invalid timestamp: ~v, nick:~v, uri:~v"
179 str-head nick (url->string uri))
182 (log-debug "Non-msg line from nick:~v, line:~a" nick str-head)
186 ; TODO Test for when missing-nick case
187 (let* ([tzs (for*/list ([d '("-" "+")]
189 [m '("00" ":00" "57" ":57")])
190 (string-append d h m))]
191 [tzs (list* "" "Z" tzs)])
192 (for* ([n '("fake-nick")]
197 [sep (list "\t" " ")]
198 [txt '("foo bar baz" "'jaz poop bear giraffe / tea" "@*\"``")])
199 (let* ([ts (string-append "2020-11-18T22:22"
200 (if (non-empty-string? s) s ":00")
202 [m (str->msg n u (string-append ts sep txt))])
204 (check-equal? (Msg-nick m) n)
205 (check-equal? (Msg-uri m) u)
206 (check-equal? (Msg-text m) txt)
207 (check-equal? (Msg-ts-orig m) ts (format "Given: ~v" ts))
210 (let* ([ts "2020-11-18T22:22:09-0500"]
215 [actual (str->msg nick uri (string-append ts tab text))]
216 [expected (Msg 1605756129 ts nick uri text '())])
218 (Msg-ts-epoch actual)
219 (Msg-ts-epoch expected)
223 (Msg-ts-orig expected)
238 (: str->lines (-> String (Listof String)))
239 (define (str->lines str)
240 (string-split str (regexp "[\r\n]+")))
243 (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi")))
245 (: str->msgs (-> (Option String) Url String (Listof Msg)))
246 (define (str->msgs nick uri str)
247 (filter-map (λ (line) (str->msg nick uri line)) (filter-comments (str->lines str))))
249 (: cache-dir Path-String)
250 (define cache-dir (build-path tt-home-dir "cache"))
252 (define cache-object-dir (build-path cache-dir "objects"))
254 (: url->cache-file-path-v1 (-> Url Path-String))
255 (define (url->cache-file-path-v1 uri)
256 (define (hash-sha1 str) : (-> String String)
257 (define in (open-input-string str))
258 (define digest (sha1 in))
259 (close-input-port in)
261 (build-path cache-object-dir (hash-sha1 (url->string uri))))
263 (: url->cache-file-path-v2 (-> Url Path-String))
264 (define (url->cache-file-path-v2 uri)
265 (build-path cache-object-dir (uri-encode (url->string uri))))
267 (define url->cache-object-path url->cache-file-path-v2)
269 (: cache-object-filename->url (-> Path-String Url))
270 (define (cache-object-filename->url name)
271 (string->url (uri-decode (path->string name))))
273 (define (url->cache-etag-path uri)
274 (build-path cache-dir "etags" (uri-encode (url->string uri))))
276 (define (url->cache-lmod-path uri)
277 (build-path cache-dir "lmods" (uri-encode (url->string uri))))
280 (: uri-read-cached (-> Url String))
281 (define (uri-read-cached uri)
282 (define path-v1 (url->cache-file-path-v1 uri))
283 (define path-v2 (url->cache-file-path-v2 uri))
284 (when (file-exists? path-v1)
285 (rename-file-or-directory path-v1 path-v2 #t))
286 (if (file-exists? path-v2)
287 (file->string path-v2)
289 (log-warning "Cache file not found for URI: ~a" (url->string uri))
292 (: uri? (-> String Boolean))
294 (regexp-match? #rx"^[a-z]+://.*" (string-downcase str)))
296 (: str->peer (String (Option Peer)))
297 (define (str->peer str)
298 (log-debug "Parsing peer string: ~v" str)
302 (log-error "Invalid URI in string: ~v, exn: ~v" str e)
304 (match (string-split str)
305 [(list u) #:when (uri? u) (Peer #f (string->url u))]
306 [(list n u) #:when (uri? u) (Peer n (string->url u))]
308 (log-error "Invalid peer string: ~v" str)
312 (: filter-comments (-> (Listof String) (Listof String)))
313 (define (filter-comments lines)
314 (filter-not (λ (line) (string-prefix? line "#")) lines))
316 (: str->peers (-> String (Listof Peer)))
317 (define (str->peers str)
318 (filter-map str->peer (filter-comments (str->lines str))))
320 (: peers->file (-> (Listof Peers) Path-String Void))
321 (define (peers->file peers path)
322 (display-lines-to-file
325 (format "~a~a" (if n (format "~a " n) "") (url->string u))])
330 (: file->peers (-> Path-String (Listof Peer)))
331 (define (file->peers file-path)
332 (if (file-exists? file-path)
333 (str->peers (file->string file-path))
335 (log-warning "File does not exist: ~v" (path->string file-path))
339 #px"^(Mon|Tue|Wed|Thu|Fri|Sat|Sun), ([0-9]{2}) (Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec) ([0-9]{4}) ([0-2][0-9]):([0-6][0-9]):([0-6][0-9]) GMT")
341 (: b->n (-> Bytes (Option Number)))
343 (string->number (bytes->string/utf-8 b)))
345 (: mon->num (-> Bytes Natural))
346 (define/match (mon->num mon)
360 (: rfc2822->epoch (-> Bytes (Option Nonnegative-Integer)))
361 (define (rfc2822->epoch timestamp)
362 (match (regexp-match re-rfc2822 timestamp)
363 [(list _ _ dd mo yyyy HH MM SS)
364 #:when (and dd mo yyyy HH MM SS)
365 (find-seconds (b->n SS)
375 (: user-agent String)
379 [prog-version (info:#%info-lookup 'version)]
380 [prog-uri "https://github.com/xandkar/tt"]
381 [user-peer-file (build-path tt-home-dir "me")]
383 (if (file-exists? user-peer-file)
384 (match (first (file->peers user-peer-file))
385 [(Peer #f u) (format "+~a" (url->string u) )]
386 [(Peer n u) (format "+~a; @~a" (url->string u) n)])
387 (format "+~a" prog-uri))])
388 (format "~a/~a (~a)" prog-name prog-version user)))
390 (: header-get (-> (Listof Bytes) Bytes (Option Bytes)))
391 (define (header-get headers name)
392 (match (filter-map (curry extract-field name) headers)
396 (: uri-download (-> Positive-Float Url Void))
397 (define (uri-download timeout u)
398 (define cached-object-path (url->cache-object-path u))
399 (define cached-etag-path (url->cache-etag-path u))
400 (define cached-lmod-path (url->cache-lmod-path u))
401 (define u-str (url->string u))
402 (log-debug "uri-download ~v into ~v" u-str cached-object-path)
403 (define timeout-chan (make-channel))
404 (define result-chan (make-channel))
405 (define timeout-thread
407 ; Doing this instead of sync/timeout to distinguish error values,
408 ; rather than just have #f to work with.
410 (channel-put timeout-chan (cons 'error 'timeout)))))
411 (define result-thread
413 ; XXX We timeout getting a response, but body download could
414 ; also take a long time and we might want to time that out as
415 ; well, but then we may end-up with partially downloaded
416 ; objects. But that could happen anyway if the server drops the
417 ; connection for whatever reason.
419 ; Maybe that is OK once we start treating the
420 ; downloaded object as an addition to the stored set of
421 ; messages, rather than the final set of messages.
424 ; - 1st try can just be an in-memory set that gets written-to
425 ; and read-from disk as a whole.
428 ([exn:fail? (λ (e) (cons 'error (cons 'net-error e)))])
429 (define-values (status-line headers body-input)
432 #:headers (list (format "User-Agent: ~a" user-agent))))
433 (cons 'ok (Resp status-line headers body-input))))
434 (channel-put result-chan result))))
438 (kill-thread result-thread)
439 (kill-thread timeout-thread)
441 [(cons 'error 'timeout)
442 (log-error "Download failed: timeout. URL:~v" u-str)]
443 [(cons 'error (cons 'net-error e))
444 (log-error "Download failed. Network error. URL:~v EXN:~v" u-str e)]
445 [(cons 'ok (Resp status-line headers body-input))
446 (log-debug "headers: ~v" headers)
447 (log-debug "status-line: ~v" status-line)
449 (string->number (second (string-split (bytes->string/utf-8 status-line)))))
450 (log-debug "status: ~v" status)
451 ; TODO Handle redirects
454 (let* ([etag (header-get headers #"ETag")]
455 [lmod (header-get headers #"Last-Modified")]
456 [lmod-curr (if lmod (rfc2822->epoch lmod) #f)]
457 [lmod-prev (if (file-exists? cached-lmod-path)
458 (rfc2822->epoch (file->bytes cached-lmod-path))
460 (log-debug "lmod-curr:~v lmod-prev:~v" lmod-curr lmod-prev)
461 (unless (or (and etag
462 (file-exists? cached-etag-path)
463 (bytes=? etag (file->bytes cached-etag-path))
465 (log-info "ETags match, skipping the rest of ~v" u-str)
469 (<= lmod-curr lmod-prev)
471 (log-info "Last-Modified <= current skipping the rest of ~v" u-str)
475 "Downloading the rest of ~v. ETag: ~a, Last-Modified: ~v"
477 (make-parent-directory* cached-object-path)
478 (make-parent-directory* cached-etag-path)
479 (make-parent-directory* cached-lmod-path)
480 (call-with-output-file cached-object-path
481 (curry copy-port body-input)
484 (display-to-file etag cached-etag-path #:exists 'replace))
486 (display-to-file lmod cached-lmod-path #:exists 'replace))))
487 (close-input-port body-input))]
489 (log-error "HTTP error URL:~a status:~a" u-str status)])]))
491 (: timeline-print (-> Out-Format (Listof Msg) Void))
492 (define (timeline-print out-format timeline)
493 (void (foldl (match-lambda**
494 [((and m (Msg _ _ nick _ _ _)) (cons prev-nick i))
495 (let ([i (if (equal? prev-nick nick) i (+ 1 i))])
496 (msg-print out-format i m)
501 (: peer->msgs (-> Peer (Listof Msg)))
502 (define (peer->msgs f)
503 (match-define (Peer nick uri) f)
504 (log-info "Reading peer nick:~v uri:~v" nick (url->string uri))
505 (str->msgs nick uri (uri-read-cached uri)))
507 (: peer-download (-> Positive-Float Peer Void))
508 (define (peer-download timeout peer)
509 (match-define (Peer nick uri) peer)
510 (define u (url->string uri))
511 (log-info "Download BEGIN uri:~a" u)
512 (define-values (_result _tm-cpu-ms tm-real-ms _tm-gc-ms)
513 (time-apply uri-download (list timeout uri)))
514 (log-info "Download END in ~a seconds, uri: ~a" (/ tm-real-ms 1000.0) u))
516 (: timeline-download (-> Integer Positive-Float (Listof Peer) Void))
517 (define (timeline-download num-workers timeout peers)
518 ; TODO No need for map - can just iter
519 (void (concurrent-filter-map num-workers (λ (p) (peer-download timeout p)) peers)))
521 (: uniq (∀ (α) (-> (Listof α) (Listof α))))
523 (set->list (list->set xs)))
525 (: peers->timeline (-> (listof Peer) (listof Msg)))
526 (define (peers->timeline peers)
527 (append* (filter-map peer->msgs peers)))
529 (: timeline-sort (-> (listof Msg) timeline-order (Listof Msgs)))
530 (define (timeline-sort msgs order)
531 (define cmp (match order
534 (sort msgs (λ (a b) (cmp (Msg-ts-epoch a)
537 (: paths->peers (-> (Listof String) (Listof Peer)))
538 (define (paths->peers paths)
539 (let* ([paths (match paths
541 (let ([peer-refs-file (build-path tt-home-dir "peers")])
543 "No peer ref file paths provided, defaulting to ~v"
544 (path->string peer-refs-file))
545 (list peer-refs-file))]
547 (log-debug "Peer ref file paths provided: ~v" paths)
548 (map string->path paths)])]
549 [peers (append* (map file->peers paths))])
550 (log-info "Read-in ~a peers." (length peers))
553 (: mentioned-peers-in-cache (-> (Listof Peer)))
554 (define (mentioned-peers-in-cache)
556 (append* (map (λ (filename)
557 (define path (build-path cache-object-dir filename))
558 (define size (/ (file-size path) 1000000.0))
559 (log-info "BEGIN parsing ~a MB from file: ~v"
562 (define t0 (current-inexact-milliseconds))
563 (define m (filter-map
565 (str->msg #f (cache-object-filename->url filename) line))
567 (file->lines path))))
568 (define t1 (current-inexact-milliseconds))
569 (log-info "END parsing ~a MB in ~a seconds from file: ~v."
574 (log-warning "No messages found in ~a" (path->string path)))
576 (directory-list cache-object-dir))))
577 (uniq (append* (map Msg-mentions msgs))))
579 (: log-writer-stop (-> Thread Void))
580 (define (log-writer-stop log-writer)
581 (log-message (current-logger) 'fatal 'stop "Exiting." #f)
582 (thread-wait log-writer))
584 (: log-writer-start (-> Log-Level Thread))
585 (define (log-writer-start level)
587 (make-logger #f #f level #f)]
589 (make-log-receiver logger level)]
594 ([date-display-format 'iso-8601])
596 (match-define (vector level msg _ topic) (sync log-receiver))
597 (unless (equal? topic 'stop)
598 (eprintf "~a [~a] ~a~n" (date->string (current-date) #t) level msg)
600 (current-logger logger)
604 (let ([log-level 'info])
610 "Enable debug log level."
611 (set! log-level 'debug)]
614 "and <command> is one of"
615 "r, read : Read the timeline (offline operation)."
616 "d, download : Download the timeline."
617 ; TODO Add path dynamically
618 "u, upload : Upload your twtxt file (alias to execute ~/.tt/upload)."
619 "c, crawl : Discover new peers mentioned by known peers (offline operation)."
621 #:args (command . args)
622 (define log-writer (log-writer-start log-level))
623 (current-command-line-arguments (list->vector args))
626 ; Initially, 15 was fastest out of the tried: 1, 5, 10, 20. Then I
627 ; started noticing significant slowdowns. Reducing to 5 seems to help.
628 (let ([num-workers 5]
635 njobs "Number of concurrent jobs."
636 (set! num-workers (string->number njobs))]
638 seconds "Timeout seconds per request."
639 (set! timeout (string->number seconds))]
641 (let ([peers (paths->peers file-paths)])
642 (define-values (_res _cpu real-ms _gc)
643 (time-apply timeline-download (list num-workers timeout peers)))
644 (log-info "Downloaded timelines from ~a peers in ~a seconds."
646 (/ real-ms 1000.0)))))]
652 (if (system (path->string (build-path tt-home-dir "upload")))
656 (let ([out-format 'multi-line]
665 "Reverse displayed timeline order."
666 (set! order 'new->old)]
668 m "Earliest time to display (ignore anything before it)."
669 (set! ts-min (rfc3339->epoch m))]
671 x "Latest time to display (ignore anything after it)."
672 (set! ts-max (rfc3339->epoch x))]
675 "Short output format"
676 (set! out-format 'single-line)]
679 (set! out-format 'multi-line)]
682 (paths->peers file-paths)]
684 (timeline-sort (peers->timeline peers) order)]
686 (filter (λ (m) (and (if ts-min (>= (Msg-ts-epoch m)
689 (if ts-max (<= (Msg-ts-epoch m)
693 (timeline-print out-format timeline))))]
700 (λ (peers) (sort peers (match-lambda**
701 [((Peer n1 _) (Peer n2 _))
702 (string<? (if n1 n1 "")
705 (build-path tt-home-dir "peers-all")]
706 [peers-mentioned-file
707 (build-path tt-home-dir "peers-mentioned")]
708 [peers-mentioned-curr
709 (mentioned-peers-in-cache)]
710 [peers-mentioned-prev
711 (file->peers peers-mentioned-file)]
713 (peers-sort (uniq (append peers-mentioned-prev
714 peers-mentioned-curr)))]
716 (file->peers peers-all-file)]
718 (list->set (append peers-mentioned
721 (set-subtract peers-all (list->set peers-all-prev))]
723 (peers-sort (set->list peers-all))])
724 (log-info "Known peers mentioned: ~a" (length peers-mentioned))
725 (log-info "Known peers total: ~a" (length peers-all))
726 (log-info "Discovered ~a new peers:~n~a"
727 (set-count peers-discovered)
729 (λ (p) (cons (Peer-nick p)
730 (url->string (Peer-uri p))))
731 (set->list peers-discovered))))
732 (peers->file peers-mentioned
733 peers-mentioned-file)
734 (peers->file peers-all
737 (eprintf "Error: invalid command: ~v\n" command)
738 (eprintf "Please use the \"--help\" option to see a list of available commands.\n")
740 (log-writer-stop log-writer))))