Use the simpler http-client API
[tt.git] / tt.rkt
CommitLineData
cd541405 1#lang typed/racket/no-check
4764ff89 2
1d753430 3(require openssl/sha1)
4764ff89 4(require racket/date)
9c464d95 5(require
d718efc4 6 net/head
edadb804 7 net/uri-codec
8da029e4 8 net/url)
4764ff89 9
3c9c8266
SK
10(require (prefix-in info: "info.rkt"))
11
78398948 12(module+ test
de3ff448 13 (require rackunit))
78398948 14
98529d3d
SK
15(define-type Url
16 net/url-structs:url)
17
18(define-type Out-Format
19 (U 'single-line
20 'multi-line))
21
22(define-type Timeline-Order
23 (U 'old->new
24 'new->old))
25
b0ff061a 26(struct Msg
78142acb 27 ([ts-epoch : Integer]
3877a0c4 28 [ts-orig : String]
13c11724 29 [nick : (Option String)]
98529d3d 30 [uri : Url]
13c11724 31 [text : String]
b0ff061a 32 [mentions : (Listof Peer)]))
9c464d95 33
dbc26280 34(struct Peer
13c11724 35 ([nick : (Option String)]
d0a0e073
SK
36 [uri : Url])
37 #:transparent)
4764ff89 38
edadb804
SK
39(: tt-home-dir Path-String)
40(define tt-home-dir (build-path (expand-user-path "~") ".tt"))
41
9a346534 42(: concurrent-filter-map (∀ (α β) (-> Natural (-> α β) (Listof α) (Listof β))))
78142acb 43(define (concurrent-filter-map num-workers f xs)
dad4504d 44 ; TODO preserve order of elements OR communicate that reorder is expected
a239a233 45 ; TODO switch from mailboxes to channels
895a32cf
SK
46 (define (make-worker id f)
47 (define parent (current-thread))
48 (λ ()
a9511f7c
SK
49 (define self : Thread (current-thread))
50 (: work (∀ (α) (-> α)))
895a32cf
SK
51 (define (work)
52 (thread-send parent (cons 'next self))
53 (match (thread-receive)
c562bea3
SK
54 ['done (thread-send parent (cons 'exit id))]
55 [(cons 'unit x) (begin
56 (define y (f x))
57 (when y (thread-send parent (cons 'result y)))
58 (work))]))
895a32cf 59 (work)))
a9511f7c 60 (: dispatch (∀ (α β) (-> (Listof Nonnegative-Integer) (Listof α) (Listof β))))
895a32cf
SK
61 (define (dispatch ws xs ys)
62 (if (empty? ws)
f1493e49
SK
63 ys
64 (match (thread-receive)
c562bea3
SK
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
68 ['() (begin
69 (thread-send thd 'done)
70 (dispatch ws xs ys))]
71 [(cons x xs) (begin
72 (thread-send thd (cons 'unit x))
73 (dispatch ws xs ys))])])))
78142acb 74 (define workers (range num-workers))
9926c9a9
SK
75 (define threads (map (λ (id) (thread (make-worker id f))) workers))
76 (define results (dispatch workers xs '()))
895a32cf
SK
77 (for-each thread-wait threads)
78 results)
79
dad4504d 80(module+ test
de3ff448
SK
81 (let* ([f (λ (x) (if (even? x) x #f))]
82 [xs (range 11)]
83 [actual (sort (concurrent-filter-map 10 f xs) <)]
84 [expected (sort ( filter-map f xs) <)])
c562bea3 85 (check-equal? actual expected "concurrent-filter-map")))
dad4504d 86
98529d3d 87(: msg-print (-> Out-Format Integer Msg Void))
3d042e75
SK
88(define msg-print
89 (let* ([colors (vector 36 33)]
90 [n (vector-length colors)])
91 (λ (out-format color-i msg)
01e4c499 92 (let ([color (vector-ref colors (modulo color-i n))]
b0ff061a
SK
93 [nick (Msg-nick msg)]
94 [uri (url->string (Msg-uri msg))]
95 [text (Msg-text msg)]
96 [mentions (Msg-mentions msg)])
3d042e75 97 (match out-format
01e4c499 98 ['single-line
13c11724
SK
99 (let ([nick (if nick nick uri)])
100 (printf "~a \033[1;37m<~a>\033[0m \033[0;~am~a\033[0m~n"
101 (parameterize
102 ([date-display-format 'iso-8601])
b0ff061a 103 (date->string (seconds->date (Msg-ts-epoch msg)) #t))
13c11724 104 nick color text))]
01e4c499 105 ['multi-line
13c11724
SK
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"
108 (parameterize
109 ([date-display-format 'rfc2822])
b0ff061a
SK
110 (date->string (seconds->date (Msg-ts-epoch msg)) #t))
111 (Msg-ts-orig msg)
13c11724 112 nick uri color text))])))))
e96264cc 113
3877a0c4
SK
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}))?$")])
117 (λ (ts)
118 (match (regexp-match re ts)
119 [(list _wholething yyyy mm dd HH MM _:SS SS _fractional tz-whole tz-sign tz-HH tz-MM)
120 (let*
121 ([tz-offset
122 (match* (tz-whole tz-sign tz-HH tz-MM)
123 [("Z" #f #f #f)
124 0]
125 [(_ (or "-" "+") (? identity h) (? identity m))
126 (let ([h (string->number h)]
127 [m (string->number m)]
128 ; Reverse to get back to UTC:
129 [op (match tz-sign ["+" -] ["-" +])])
130 (op 0 (+ (* 60 m) (* 60 (* 60 h)))))]
131 [(a b c d)
132 (log-warning "Impossible TZ string: ~v, components: ~v ~v ~v ~v" tz-whole a b c d)
133 0])]
134 [ts-orig ts]
135 [local-time? #f]
136 [ts-epoch (find-seconds (if SS (string->number SS) 0)
137 (string->number MM)
138 (string->number HH)
139 (string->number dd)
140 (string->number mm)
141 (string->number yyyy)
142 local-time?)])
143 (+ ts-epoch tz-offset))]
144 [_
145 (log-error "Invalid timestamp: ~v" ts)
146 #f]))))
147
13c11724 148(: str->msg (-> (Option String) Url String (Option Msg)))
b4689464 149(define str->msg
3877a0c4 150 (let ([re (pregexp "^([^\\s\t]+)[\\s\t]+(.*)$")])
b4689464
SK
151 (λ (nick uri str)
152 (with-handlers*
153 ([exn:fail?
154 (λ (e)
9c464d95
SK
155 (log-error
156 "Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v"
157 str nick (url->string uri) e)
b4689464
SK
158 #f)])
159 (match (regexp-match re str)
3877a0c4
SK
160 [(list _wholething ts-orig text)
161 (let ([ts-epoch (rfc3339->epoch ts-orig)])
162 (if ts-epoch
13c11724
SK
163 (let ([mentions
164 (filter-map
165 (λ (m) (match (regexp-match #px"@<([^>]+)>" m)
166 [(list _wholething nick-uri)
dbc26280 167 (str->peer nick-uri)]))
13c11724 168 (regexp-match* #px"@<[^\\s]+([\\s]+)?[^>]+>" text))])
b0ff061a 169 (Msg ts-epoch ts-orig nick uri text mentions))
3877a0c4
SK
170 (begin
171 (log-error
172 "Msg rejected due to invalid timestamp: ~v, nick:~v, uri:~v"
173 str nick (url->string uri))
174 #f)))]
b4689464 175 [_
13c11724 176 (log-debug "Non-msg line from nick:~v, line:~a" nick str)
b4689464 177 #f])))))
88d50b3e 178
63afa259 179(module+ test
13c11724 180 ; TODO Test for when missing-nick case
b4689464
SK
181 (let* ([tzs (for*/list ([d '("-" "+")]
182 [h '("5" "05")]
183 [m '("00" ":00" "57" ":57")])
184 (string-append d h m))]
185 [tzs (list* "" "Z" tzs)])
186 (for* ([n '("fake-nick")]
187 [u '("fake-uri")]
188 [s '("" ":10")]
189 [f '("" ".1337")]
190 [z tzs]
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")
195 z)]
196 [m (str->msg n u (string-append ts sep txt))])
197 (check-not-false m)
b0ff061a
SK
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))
b4689464
SK
202 )))
203
de3ff448
SK
204 (let* ([ts "2020-11-18T22:22:09-0500"]
205 [tab " "]
206 [text "Lorem ipsum"]
207 [nick "foo"]
208 [uri "bar"]
209 [actual (str->msg nick uri (string-append ts tab text))]
b0ff061a 210 [expected (Msg 1605756129 ts nick uri text)])
c562bea3 211 (check-equal?
b0ff061a
SK
212 (Msg-ts-epoch actual)
213 (Msg-ts-epoch expected)
78142acb 214 "str->msg ts-epoch")
3877a0c4 215 (check-equal?
b0ff061a
SK
216 (Msg-ts-orig actual)
217 (Msg-ts-orig expected)
3877a0c4 218 "str->msg ts-orig")
c562bea3 219 (check-equal?
b0ff061a
SK
220 (Msg-nick actual)
221 (Msg-nick expected)
c562bea3
SK
222 "str->msg nick")
223 (check-equal?
b0ff061a
SK
224 (Msg-uri actual)
225 (Msg-uri expected)
c562bea3
SK
226 "str->msg uri")
227 (check-equal?
b0ff061a
SK
228 (Msg-text actual)
229 (Msg-text expected)
c562bea3 230 "str->msg text")))
63afa259 231
98529d3d 232(: str->lines (-> String (Listof String)))
e96264cc
SK
233(define (str->lines str)
234 (string-split str (regexp "[\r\n]+")))
235
63afa259 236(module+ test
de3ff448 237 (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi")))
63afa259 238
13c11724 239(: str->msgs (-> (Option String) Url String (Listof Msg)))
b201e854 240(define (str->msgs nick uri str)
3877a0c4 241 (filter-map (λ (line) (str->msg nick uri line)) (filter-comments (str->lines str))))
4764ff89 242
edadb804
SK
243(: cache-dir Path-String)
244(define cache-dir (build-path tt-home-dir "cache"))
245
d718efc4
SK
246(define cache-object-dir (build-path cache-dir "objects"))
247
edadb804
SK
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)
254 digest)
d718efc4 255 (build-path cache-object-dir (hash-sha1 (url->string uri))))
edadb804
SK
256
257(: url->cache-file-path-v2 (-> Url Path-String))
258(define (url->cache-file-path-v2 uri)
d718efc4
SK
259 (build-path cache-object-dir (uri-encode (url->string uri))))
260
261(define url->cache-object-path url->cache-file-path-v2)
1d753430 262
d718efc4
SK
263(define (url->cache-etag-path uri)
264 (build-path cache-dir "etags" (uri-encode (url->string uri))))
265
266(define (url->cache-lmod-path uri)
267 (build-path cache-dir "lmods" (uri-encode (url->string uri))))
9c464d95 268
98529d3d
SK
269; TODO Return Option
270(: uri-read-cached (-> Url String))
4214c0f3 271(define (uri-read-cached uri)
edadb804
SK
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)
0e16a46c 278 (begin
9c464d95 279 (log-warning "Cache file not found for URI: ~a" (url->string uri))
4214c0f3
SK
280 "")))
281
dbc26280
SK
282(: str->peer (String (Option Peer)))
283(define (str->peer str)
284 (log-debug "Parsing peer string: ~v" str)
13c11724
SK
285 (with-handlers*
286 ([exn:fail?
287 (λ (e)
288 (log-error "Invalid URI in string: ~v, exn: ~v" str e)
289 #f)])
290 (match (string-split str)
dbc26280
SK
291 [(list u) (Peer #f (string->url u))]
292 [(list n u) (Peer n (string->url u))]
13c11724 293 [_
dbc26280 294 (log-error "Invalid peer string: ~v" str)
13c11724
SK
295 #f])))
296
9c464d95 297
98529d3d 298(: filter-comments (-> (Listof String) (Listof String)))
9c464d95
SK
299(define (filter-comments lines)
300 (filter-not (λ (line) (string-prefix? line "#")) lines))
301
dbc26280
SK
302(: str->peers (-> String (Listof Peer)))
303(define (str->peers str)
304 (filter-map str->peer (filter-comments (str->lines str))))
9c464d95 305
dbc26280 306(: file->peers (-> Path-String (Listof Peer)))
d0a0e073
SK
307(define (file->peers file-path)
308 (if (file-exists? file-path)
309 (str->peers (file->string file-path))
310 (begin
311 (log-error "File does not exist: ~v" (path->string file-path))
312 '())))
9c464d95 313
98529d3d 314(: user-agent String)
9c464d95
SK
315(define user-agent
316 (let*
317 ([prog-name "tt"]
3c9c8266 318 [prog-version (info:#%info-lookup 'version)]
9c464d95 319 [prog-uri "https://github.com/xandkar/tt"]
dbc26280 320 [user-peer-file (build-path tt-home-dir "me")]
9c464d95 321 [user
dbc26280
SK
322 (if (file-exists? user-peer-file)
323 (match (first (file->peers user-peer-file))
324 [(Peer #f u) (format "+~a" (url->string u) )]
325 [(Peer n u) (format "+~a; @~a" (url->string u) n)])
9c464d95
SK
326 (format "+~a" prog-uri))])
327 (format "~a/~a (~a)" prog-name prog-version user)))
328
d718efc4
SK
329(: header-get (-> (Listof Bytes) Bytes (Option Bytes)))
330(define (header-get headers name)
331 (match (filter-map (curry extract-field name) headers)
332 [(list val) val]
333 [_ #f]))
334
98529d3d 335(: uri-download (-> Url Void))
9c464d95 336(define (uri-download u)
d718efc4
SK
337 (define cached-object-path (url->cache-object-path u))
338 (define cached-etag-path (url->cache-etag-path u))
339 (define cached-lmod-path (url->cache-lmod-path u))
340 (log-debug "uri-download ~v into ~v" u cached-object-path)
8da029e4
SK
341 (define-values (status-line headers body-input)
342 ; TODO Timeout. Currently hangs on slow connections.
343 (http-sendrecv/url u #:headers (list (format "User-Agent: ~a" user-agent))))
344 (log-debug "headers: ~v" headers)
345 (log-debug "status-line: ~v" status-line)
346 (define status
347 (string->number (second (string-split (bytes->string/utf-8 status-line)))))
348 (log-debug "status: ~v" status)
349 ; TODO Handle redirects
350 (match status
351 [200
352 (let ([etag (header-get headers #"ETag")]
353 [lmod (header-get headers #"Last-Modified")])
354 (if (and etag
355 (file-exists? cached-etag-path)
356 (bytes=? etag (file->bytes cached-etag-path)))
357 (log-info "ETags match, skipping the rest of ~v" (url->string u))
358 (begin
359 (log-info
360 "Downloading the rest of ~v. ETag: ~a, Last-Modified: ~v"
361 (url->string u) etag lmod)
362 (make-parent-directory* cached-object-path)
363 (make-parent-directory* cached-etag-path)
364 (make-parent-directory* cached-lmod-path)
365 (call-with-output-file cached-object-path
366 (curry copy-port body-input)
367 #:exists 'replace)
368 (when etag
369 (display-to-file etag cached-etag-path #:exists 'replace))
370 (when lmod
371 (display-to-file etag cached-lmod-path #:exists 'replace))))
372 (close-input-port body-input))]
373 [_
374 (raise status)]))
4764ff89 375
98529d3d 376(: timeline-print (-> Out-Format (Listof Msg) Void))
b201e854 377(define (timeline-print out-format timeline)
3d042e75 378 (void (foldl (match-lambda**
b0ff061a 379 [((and m (Msg _ _ nick _ _ _)) (cons prev-nick i))
13c11724 380 (let ([i (if (equal? prev-nick nick) i (+ 1 i))])
3d042e75
SK
381 (msg-print out-format i m)
382 (cons nick i))])
383 (cons "" 0)
384 timeline)))
4764ff89 385
dbc26280
SK
386(: peer->msgs (-> Peer (Listof Msg)))
387(define (peer->msgs f)
388 (match-define (Peer nick uri) f)
389 (log-info "Reading peer nick:~v uri:~v" nick (url->string uri))
9c464d95 390 (str->msgs nick uri (uri-read-cached uri)))
4214c0f3 391
dbc26280
SK
392(: peer-download (-> Peer Void))
393(define (peer-download f)
394 (match-define (Peer nick uri) f)
2db1b40f 395 (define u (url->string uri))
dbc26280 396 (log-info "Downloading peer uri:~a" u)
9a6a9f9a 397 (with-handlers
9c464d95 398 ([exn:fail?
9a6a9f9a 399 (λ (e)
13c11724 400 (log-error "Network error nick:~v uri:~v exn:~v" nick u e)
9a6a9f9a
SK
401 #f)]
402 [integer?
403 (λ (status)
13c11724 404 (log-error "HTTP error nick:~v uri:~a status:~a" nick u status)
9a6a9f9a 405 #f)])
2db1b40f
SK
406 (define-values (_result _tm-cpu-ms tm-real-ms _tm-gc-ms)
407 (time-apply uri-download (list uri)))
dbc26280 408 (log-info "Peer downloaded in ~a seconds, uri: ~a" (/ tm-real-ms 1000.0) u)))
4214c0f3 409
dbc26280
SK
410(: timeline-download (-> Integer (Listof Peer) Void))
411(define (timeline-download num-workers peers)
4214c0f3 412 ; TODO No need for map - can just iter
dbc26280 413 (void (concurrent-filter-map num-workers peer-download peers)))
9a6a9f9a
SK
414
415; TODO timeline contract : time-sorted list of messages
dbc26280
SK
416(: timeline-read (-> Timeline-Order (Listof Peer) (Listof Msg)))
417(define (timeline-read order peers)
a4899240
SK
418 (define cmp (match order
419 ['old->new <]
420 ['new->old >]))
dbc26280 421 (sort (append* (filter-map peer->msgs peers))
b0ff061a 422 (λ (a b) (cmp (Msg-ts-epoch a) (Msg-ts-epoch b)))))
4764ff89 423
d0a0e073
SK
424(: paths->peers (-> (Listof String) (Listof Peer)))
425(define (paths->peers paths)
426 (let* ([paths (match paths
427 ['()
428 (let ([peer-refs-file (build-path tt-home-dir "peers")])
429 (log-debug
430 "No peer ref file paths provided, defaulting to ~v"
431 (path->string peer-refs-file))
432 (list peer-refs-file))]
433 [paths
434 (log-debug "Peer ref file paths provided: ~v" paths)
435 (map string->path paths)])]
436 [peers (append* (map file->peers paths))])
437 (log-info "Read-in ~a peers." (length peers))
438 peers))
439
56de6228
SK
440(: log-writer-stop (-> Thread Void))
441(define (log-writer-stop log-writer)
442 (log-message (current-logger) 'fatal 'stop "Exiting." #f)
443 (thread-wait log-writer))
444
0d3f753c
SK
445(: log-writer-start (-> Log-Level Thread))
446(define (log-writer-start level)
56de6228
SK
447 (let* ([logger
448 (make-logger #f #f level #f)]
449 [log-receiver
450 (make-log-receiver logger level)]
451 [log-writer
452 (thread
453 (λ ()
454 (parameterize
455 ([date-display-format 'iso-8601])
456 (let loop ()
457 (match-define (vector level msg _ topic) (sync log-receiver))
458 (unless (equal? topic 'stop)
459 (eprintf "~a [~a] ~a~n" (date->string (current-date) #t) level msg)
460 (loop))))))])
461 (current-logger logger)
462 log-writer))
01e4c499 463
24c6a76b 464(module+ main
24f1f64b 465 (let ([log-level 'info])
c562bea3 466 (command-line
24f1f64b
SK
467 #:program
468 "tt"
c562bea3 469 #:once-each
01e4c499
SK
470 [("-d" "--debug")
471 "Enable debug log level."
472 (set! log-level 'debug)]
24f1f64b
SK
473 #:help-labels
474 ""
475 "and <command> is one of"
4214c0f3
SK
476 "r, read i : Read the timeline."
477 "d, download : Download the timeline."
edadb804 478 ; TODO Add path dynamically
3a4b2233 479 "u, upload : Upload your twtxt file (alias to execute ~/.tt/upload)."
24f1f64b
SK
480 ""
481 #:args (command . args)
0d3f753c 482 (define log-writer (log-writer-start log-level))
4214c0f3 483 (current-command-line-arguments (list->vector args))
24f1f64b 484 (match command
4214c0f3 485 [(or "d" "download")
2db1b40f 486 ; Initially, 15 was fastest out of the tried: 1, 5, 10, 20. Then I
edadb804 487 ; started noticing significant slowdowns. Reducing to 5 seems to help.
2db1b40f 488 (let ([num-workers 5])
24f1f64b
SK
489 (command-line
490 #:program
4214c0f3 491 "tt download"
24f1f64b
SK
492 #:once-each
493 [("-j" "--jobs")
494 njobs "Number of concurrent jobs."
78142acb 495 (set! num-workers (string->number njobs))]
d0a0e073 496 #:args file-paths
32dcc0b0 497 (define-values (_res _cpu real-ms _gc)
d0a0e073 498 (time-apply timeline-download (list num-workers (paths->peers file-paths))))
0d3f753c 499 (log-info "Timeline downloaded in ~a seconds." (/ real-ms 1000.0))))]
3a4b2233
SK
500 [(or "u" "upload")
501 (command-line
3877a0c4
SK
502 #:program
503 "tt upload"
504 #:args ()
edadb804 505 (if (system (path->string (build-path tt-home-dir "upload")))
3a4b2233
SK
506 (exit 0)
507 (exit 1)))]
4214c0f3 508 [(or "r" "read")
a4899240
SK
509 (let ([out-format 'multi-line]
510 [order 'old->new])
4214c0f3
SK
511 (command-line
512 #:program
513 "tt read"
a4899240
SK
514 #:once-each
515 [("-r" "--rev")
516 "Reverse displayed timeline order."
517 (set! order 'new->old)]
24f1f64b
SK
518 #:once-any
519 [("-s" "--short")
520 "Short output format"
521 (set! out-format 'single-line)]
522 [("-l" "--long")
523 "Long output format"
524 (set! out-format 'multi-line)]
d0a0e073
SK
525 #:args file-paths
526 (timeline-print out-format (timeline-read order (paths->peers file-paths)))))])
0d3f753c 527 (log-writer-stop log-writer))))
This page took 0.126616 seconds and 4 git commands to generate.