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)]
39 (: tt-home-dir Path-String)
40 (define tt-home-dir (build-path (expand-user-path "~") ".tt"))
42 (: concurrent-filter-map (∀ (α β) (-> Natural (-> α β) (Listof α) (Listof β))))
43 (define (concurrent-filter-map num-workers f xs)
44 ; TODO preserve order of elements OR communicate that reorder is expected
45 ; TODO switch from mailboxes to channels
46 (define (make-worker id f)
47 (define parent (current-thread))
49 (define self : Thread (current-thread))
50 (: work (∀ (α) (-> α)))
52 (thread-send parent (cons 'next self))
53 (match (thread-receive)
54 ['done (thread-send parent (cons 'exit id))]
55 [(cons 'unit x) (begin
57 (when y (thread-send parent (cons 'result y)))
60 (: dispatch (∀ (α β) (-> (Listof Nonnegative-Integer) (Listof α) (Listof β))))
61 (define (dispatch ws xs ys)
64 (match (thread-receive)
65 [(cons 'exit w) (dispatch (remove w ws =) xs ys)]
66 [(cons 'result y) (dispatch ws xs (cons y ys))]
67 [(cons 'next thd) (match xs
69 (thread-send thd 'done)
72 (thread-send thd (cons 'unit x))
73 (dispatch ws xs ys))])])))
74 (define workers (range num-workers))
75 (define threads (map (λ (id) (thread (make-worker id f))) workers))
76 (define results (dispatch workers xs '()))
77 (for-each thread-wait threads)
81 (let* ([f (λ (x) (if (even? x) x #f))]
83 [actual (sort (concurrent-filter-map 10 f xs) <)]
84 [expected (sort ( filter-map f xs) <)])
85 (check-equal? actual expected "concurrent-filter-map")))
87 (: msg-print (-> Out-Format Integer Msg Void))
89 (let* ([colors (vector 36 33)]
90 [n (vector-length colors)])
91 (λ (out-format color-i msg)
92 (let ([color (vector-ref colors (modulo color-i n))]
94 [uri (url->string (Msg-uri msg))]
96 [mentions (Msg-mentions msg)])
99 (let ([nick (if nick nick uri)])
100 (printf "~a \033[1;37m<~a>\033[0m \033[0;~am~a\033[0m~n"
102 ([date-display-format 'iso-8601])
103 (date->string (seconds->date (Msg-ts-epoch msg)) #t))
106 (let ([nick (if nick (string-append nick " ") "")])
107 (printf "~a (~a)~n\033[1;37m<~a~a>\033[0m~n\033[0;~am~a\033[0m~n~n"
109 ([date-display-format 'rfc2822])
110 (date->string (seconds->date (Msg-ts-epoch msg)) #t))
112 nick uri color text))])))))
114 (: rfc3339->epoch (-> String (Option Nonnegative-Integer)))
115 (define rfc3339->epoch
116 (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}))?$")])
118 (match (regexp-match re ts)
119 [(list _wholething yyyy mm dd HH MM _:SS SS _fractional tz-whole tz-sign tz-HH tz-MM)
122 (match* (tz-whole tz-sign tz-HH tz-MM)
125 [(_ (or "-" "+") (? identity h) (? identity m))
126 (let ([h (string->number h)]
127 [m (string->number m)]
128 ; Reverse to get back to UTC:
129 [op (match tz-sign ["+" -] ["-" +])])
130 (op 0 (+ (* 60 m) (* 60 (* 60 h)))))]
132 (log-warning "Impossible TZ string: ~v, components: ~v ~v ~v ~v" tz-whole a b c d)
136 [ts-epoch (find-seconds (if SS (string->number SS) 0)
141 (string->number yyyy)
143 (+ ts-epoch tz-offset))]
145 (log-error "Invalid timestamp: ~v" ts)
148 (: str->msg (-> (Option String) Url String (Option Msg)))
150 (let ([re (pregexp "^([^\\s\t]+)[\\s\t]+(.*)$")])
156 "Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v"
157 str nick (url->string uri) e)
159 (match (regexp-match re str)
160 [(list _wholething ts-orig text)
161 (let ([ts-epoch (rfc3339->epoch ts-orig)])
165 (λ (m) (match (regexp-match #px"@<([^>]+)>" m)
166 [(list _wholething nick-uri)
167 (str->peer nick-uri)]))
168 (regexp-match* #px"@<[^\\s]+([\\s]+)?[^>]+>" text))])
169 (Msg ts-epoch ts-orig nick uri text mentions))
172 "Msg rejected due to invalid timestamp: ~v, nick:~v, uri:~v"
173 str nick (url->string uri))
176 (log-debug "Non-msg line from nick:~v, line:~a" nick str)
180 ; TODO Test for when missing-nick case
181 (let* ([tzs (for*/list ([d '("-" "+")]
183 [m '("00" ":00" "57" ":57")])
184 (string-append d h m))]
185 [tzs (list* "" "Z" tzs)])
186 (for* ([n '("fake-nick")]
191 [sep (list "\t" " ")]
192 [txt '("foo bar baz" "'jaz poop bear giraffe / tea" "@*\"``")])
193 (let* ([ts (string-append "2020-11-18T22:22"
194 (if (non-empty-string? s) s ":00")
196 [m (str->msg n u (string-append ts sep txt))])
198 (check-equal? (Msg-nick m) n)
199 (check-equal? (Msg-uri m) u)
200 (check-equal? (Msg-text m) txt)
201 (check-equal? (Msg-ts-orig m) ts (format "Given: ~v" ts))
204 (let* ([ts "2020-11-18T22:22:09-0500"]
209 [actual (str->msg nick uri (string-append ts tab text))]
210 [expected (Msg 1605756129 ts nick uri text '())])
212 (Msg-ts-epoch actual)
213 (Msg-ts-epoch expected)
217 (Msg-ts-orig expected)
232 (: str->lines (-> String (Listof String)))
233 (define (str->lines str)
234 (string-split str (regexp "[\r\n]+")))
237 (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi")))
239 (: str->msgs (-> (Option String) Url String (Listof Msg)))
240 (define (str->msgs nick uri str)
241 (filter-map (λ (line) (str->msg nick uri line)) (filter-comments (str->lines str))))
243 (: cache-dir Path-String)
244 (define cache-dir (build-path tt-home-dir "cache"))
246 (define cache-object-dir (build-path cache-dir "objects"))
248 (: url->cache-file-path-v1 (-> Url Path-String))
249 (define (url->cache-file-path-v1 uri)
250 (define (hash-sha1 str) : (-> String String)
251 (define in (open-input-string str))
252 (define digest (sha1 in))
253 (close-input-port in)
255 (build-path cache-object-dir (hash-sha1 (url->string uri))))
257 (: url->cache-file-path-v2 (-> Url Path-String))
258 (define (url->cache-file-path-v2 uri)
259 (build-path cache-object-dir (uri-encode (url->string uri))))
261 (define url->cache-object-path url->cache-file-path-v2)
263 (define (url->cache-etag-path uri)
264 (build-path cache-dir "etags" (uri-encode (url->string uri))))
266 (define (url->cache-lmod-path uri)
267 (build-path cache-dir "lmods" (uri-encode (url->string uri))))
270 (: uri-read-cached (-> Url String))
271 (define (uri-read-cached uri)
272 (define path-v1 (url->cache-file-path-v1 uri))
273 (define path-v2 (url->cache-file-path-v2 uri))
274 (when (file-exists? path-v1)
275 (rename-file-or-directory path-v1 path-v2 #t))
276 (if (file-exists? path-v2)
277 (file->string path-v2)
279 (log-warning "Cache file not found for URI: ~a" (url->string uri))
282 (: uri? (-> String Boolean))
284 (regexp-match? #rx"^[a-z]+://.*" (string-downcase str)))
286 (: str->peer (String (Option Peer)))
287 (define (str->peer str)
288 (log-debug "Parsing peer string: ~v" str)
292 (log-error "Invalid URI in string: ~v, exn: ~v" str e)
294 (match (string-split str)
295 [(list u) #:when (uri? u) (Peer #f (string->url u))]
296 [(list n u) #:when (uri? u) (Peer n (string->url u))]
298 (log-error "Invalid peer string: ~v" str)
302 (: filter-comments (-> (Listof String) (Listof String)))
303 (define (filter-comments lines)
304 (filter-not (λ (line) (string-prefix? line "#")) lines))
306 (: str->peers (-> String (Listof Peer)))
307 (define (str->peers str)
308 (filter-map str->peer (filter-comments (str->lines str))))
310 (: peers->file (-> (Listof Peers) Path-String Void))
311 (define (peers->file peers path)
312 (display-lines-to-file
315 (format "~a~a" (if n (format "~a " n) "") (url->string u))])
320 (: file->peers (-> Path-String (Listof Peer)))
321 (define (file->peers file-path)
322 (if (file-exists? file-path)
323 (str->peers (file->string file-path))
325 (log-warning "File does not exist: ~v" (path->string file-path))
329 #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")
331 (: b->n (-> Bytes (Option Number)))
333 (string->number (bytes->string/utf-8 b)))
335 (: mon->num (-> Bytes Natural))
336 (define/match (mon->num mon)
350 (: rfc2822->epoch (-> Bytes (Option Nonnegative-Integer)))
351 (define (rfc2822->epoch timestamp)
352 (match (regexp-match re-rfc2822 timestamp)
353 [(list _ _ dd mo yyyy HH MM SS)
354 #:when (and dd mo yyyy HH MM SS)
355 (find-seconds (b->n SS)
365 (: user-agent String)
369 [prog-version (info:#%info-lookup 'version)]
370 [prog-uri "https://github.com/xandkar/tt"]
371 [user-peer-file (build-path tt-home-dir "me")]
373 (if (file-exists? user-peer-file)
374 (match (first (file->peers user-peer-file))
375 [(Peer #f u) (format "+~a" (url->string u) )]
376 [(Peer n u) (format "+~a; @~a" (url->string u) n)])
377 (format "+~a" prog-uri))])
378 (format "~a/~a (~a)" prog-name prog-version user)))
380 (: header-get (-> (Listof Bytes) Bytes (Option Bytes)))
381 (define (header-get headers name)
382 (match (filter-map (curry extract-field name) headers)
386 (: uri-download (-> Url Void))
387 (define (uri-download u)
388 (define cached-object-path (url->cache-object-path u))
389 (define cached-etag-path (url->cache-etag-path u))
390 (define cached-lmod-path (url->cache-lmod-path u))
391 (log-debug "uri-download ~v into ~v" u cached-object-path)
392 (define-values (status-line headers body-input)
393 ; TODO Timeout. Currently hangs on slow connections.
394 (http-sendrecv/url u #:headers (list (format "User-Agent: ~a" user-agent))))
395 (log-debug "headers: ~v" headers)
396 (log-debug "status-line: ~v" status-line)
398 (string->number (second (string-split (bytes->string/utf-8 status-line)))))
399 (log-debug "status: ~v" status)
400 ; TODO Handle redirects
403 (let* ([etag (header-get headers #"ETag")]
404 [lmod (header-get headers #"Last-Modified")]
405 [lmod-curr (if lmod (rfc2822->epoch lmod) #f)]
406 [lmod-prev (if (file-exists? cached-lmod-path)
407 (rfc2822->epoch (file->bytes cached-lmod-path))
409 (log-debug "lmod-curr:~v lmod-prev:~v" lmod-curr lmod-prev)
410 (unless (or (and etag
411 (file-exists? cached-etag-path)
412 (bytes=? etag (file->bytes cached-etag-path))
414 (log-info "ETags match, skipping the rest of ~v" (url->string u))
418 (<= lmod-curr lmod-prev)
420 (log-info "Last-Modified <= current skipping the rest of ~v" (url->string u))
424 "Downloading the rest of ~v. ETag: ~a, Last-Modified: ~v"
425 (url->string u) etag lmod)
426 (make-parent-directory* cached-object-path)
427 (make-parent-directory* cached-etag-path)
428 (make-parent-directory* cached-lmod-path)
429 (call-with-output-file cached-object-path
430 (curry copy-port body-input)
433 (display-to-file etag cached-etag-path #:exists 'replace))
435 (display-to-file lmod cached-lmod-path #:exists 'replace))))
436 (close-input-port body-input))]
440 (: timeline-print (-> Out-Format (Listof Msg) Void))
441 (define (timeline-print out-format timeline)
442 (void (foldl (match-lambda**
443 [((and m (Msg _ _ nick _ _ _)) (cons prev-nick i))
444 (let ([i (if (equal? prev-nick nick) i (+ 1 i))])
445 (msg-print out-format i m)
450 (: peer->msgs (-> Peer (Listof Msg)))
451 (define (peer->msgs f)
452 (match-define (Peer nick uri) f)
453 (log-info "Reading peer nick:~v uri:~v" nick (url->string uri))
454 (str->msgs nick uri (uri-read-cached uri)))
456 (: peer-download (-> Peer Void))
457 (define (peer-download f)
458 (match-define (Peer nick uri) f)
459 (define u (url->string uri))
460 (log-info "Downloading peer uri:~a" u)
464 (log-error "Network error nick:~v uri:~v exn:~v" nick u e)
468 (log-error "HTTP error nick:~v uri:~a status:~a" nick u status)
470 (define-values (_result _tm-cpu-ms tm-real-ms _tm-gc-ms)
471 (time-apply uri-download (list uri)))
472 (log-info "Peer downloaded in ~a seconds, uri: ~a" (/ tm-real-ms 1000.0) u)))
474 (: timeline-download (-> Integer (Listof Peer) Void))
475 (define (timeline-download num-workers peers)
476 ; TODO No need for map - can just iter
477 (void (concurrent-filter-map num-workers peer-download peers)))
479 (: uniq (∀ (α) (-> (Listof α) (Listof α))))
481 (set->list (list->set xs)))
483 (: peers->timeline (-> (listof Peer) (listof Msg)))
484 (define (peers->timeline peers)
485 (append* (filter-map peer->msgs peers)))
487 (: timeline-sort (-> (listof Msg) timeline-order (Listof Msgs)))
488 (define (timeline-sort msgs order)
489 (define cmp (match order
492 (sort msgs (λ (a b) (cmp (Msg-ts-epoch a)
495 (: paths->peers (-> (Listof String) (Listof Peer)))
496 (define (paths->peers paths)
497 (let* ([paths (match paths
499 (let ([peer-refs-file (build-path tt-home-dir "peers")])
501 "No peer ref file paths provided, defaulting to ~v"
502 (path->string peer-refs-file))
503 (list peer-refs-file))]
505 (log-debug "Peer ref file paths provided: ~v" paths)
506 (map string->path paths)])]
507 [peers (append* (map file->peers paths))])
508 (log-info "Read-in ~a peers." (length peers))
511 (: log-writer-stop (-> Thread Void))
512 (define (log-writer-stop log-writer)
513 (log-message (current-logger) 'fatal 'stop "Exiting." #f)
514 (thread-wait log-writer))
516 (: log-writer-start (-> Log-Level Thread))
517 (define (log-writer-start level)
519 (make-logger #f #f level #f)]
521 (make-log-receiver logger level)]
526 ([date-display-format 'iso-8601])
528 (match-define (vector level msg _ topic) (sync log-receiver))
529 (unless (equal? topic 'stop)
530 (eprintf "~a [~a] ~a~n" (date->string (current-date) #t) level msg)
532 (current-logger logger)
536 (let ([log-level 'info])
542 "Enable debug log level."
543 (set! log-level 'debug)]
546 "and <command> is one of"
547 "r, read : Read the timeline (offline operation)."
548 "d, download : Download the timeline."
549 ; TODO Add path dynamically
550 "u, upload : Upload your twtxt file (alias to execute ~/.tt/upload)."
551 "c, crawl : Discover new peers mentioned by known peers (offline operation)."
553 #:args (command . args)
554 (define log-writer (log-writer-start log-level))
555 (current-command-line-arguments (list->vector args))
558 ; Initially, 15 was fastest out of the tried: 1, 5, 10, 20. Then I
559 ; started noticing significant slowdowns. Reducing to 5 seems to help.
560 (let ([num-workers 5])
566 njobs "Number of concurrent jobs."
567 (set! num-workers (string->number njobs))]
569 (let ([peers (paths->peers file-paths)])
570 (define-values (_res _cpu real-ms _gc)
571 (time-apply timeline-download (list num-workers peers)))
572 (log-info "Downloaded timelines from ~a peers in ~a seconds."
574 (/ real-ms 1000.0)))))]
580 (if (system (path->string (build-path tt-home-dir "upload")))
584 (let ([out-format 'multi-line]
591 "Reverse displayed timeline order."
592 (set! order 'new->old)]
595 "Short output format"
596 (set! out-format 'single-line)]
599 (set! out-format 'multi-line)]
602 (paths->peers file-paths)]
604 (timeline-sort (peers->timeline peers) order)])
605 (timeline-print out-format timeline))))]
612 (λ (peers) (sort peers (match-lambda**
613 [((Peer n1 _) (Peer n2 _))
614 (string<? (if n1 n1 "")
617 (build-path tt-home-dir "peers-all")]
618 [peers-mentioned-file
619 (build-path tt-home-dir "peers-mentioned")]
623 ; TODO Refactor such that path->string not needed
624 ['() (list (path->string peers-all-file))]
627 (peers->timeline peers)]
628 [peers-mentioned-curr
629 (uniq (append* (map Msg-mentions timeline)))]
630 [peers-mentioned-prev
631 (file->peers peers-mentioned-file)]
633 (peers-sort (uniq (append peers-mentioned-prev
634 peers-mentioned-curr)))]
636 (file->peers peers-all-file)]
638 (list->set (append peers
642 (set-count (set-subtract peers-all
643 (list->set peers-all-prev)))]
645 (peers-sort (set->list peers-all))])
646 (log-info "Discovered ~a new peers." n-peers-discovered)
647 (peers->file peers-mentioned
648 peers-mentioned-file)
649 (peers->file peers-all
652 (eprintf "Error: invalid command: ~v\n" command)
653 (eprintf "Please use the \"--help\" option to see a list of available commands.\n")
655 (log-writer-stop log-writer))))