| 1 | #lang typed/racket/no-check |
| 2 | |
| 3 | (require openssl/sha1) |
| 4 | (require racket/date) |
| 5 | (require |
| 6 | net/head |
| 7 | net/uri-codec |
| 8 | net/url) |
| 9 | |
| 10 | (require (prefix-in info: "info.rkt")) |
| 11 | |
| 12 | (module+ test |
| 13 | (require rackunit)) |
| 14 | |
| 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 | |
| 26 | (define-type Result |
| 27 | (∀ (α β) (U (cons 'ok α) |
| 28 | (cons 'error β)))) |
| 29 | |
| 30 | (struct User |
| 31 | ([uri : String] |
| 32 | [nick : (Option String)])) |
| 33 | |
| 34 | (struct User-Agent |
| 35 | ([user : User] |
| 36 | [prog : Prog])) |
| 37 | |
| 38 | (struct Prog |
| 39 | ([name : String] |
| 40 | [version : String])) |
| 41 | |
| 42 | (struct Msg |
| 43 | ([ts-epoch : Integer] |
| 44 | [ts-orig : String] |
| 45 | [from : Peer] |
| 46 | [text : String] |
| 47 | [mentions : (Listof Peer)])) |
| 48 | |
| 49 | (struct Peer |
| 50 | ([nick : (Option String)] |
| 51 | [uri : Url] |
| 52 | [uri-str : String] |
| 53 | [comment : (Option String)]) |
| 54 | #:transparent) |
| 55 | |
| 56 | (struct Resp |
| 57 | ([status-line : String] |
| 58 | [headers : (Listof Bytes)] |
| 59 | [body-input : Input-Port]) |
| 60 | #:transparent) |
| 61 | |
| 62 | (: prog Prog) |
| 63 | (define prog |
| 64 | (Prog "tt" (info:#%info-lookup 'version))) |
| 65 | |
| 66 | (: user-default User) |
| 67 | (define user-default |
| 68 | (User "https://github.com/xandkar/tt" #f)) |
| 69 | |
| 70 | (: user->str (-> User String)) |
| 71 | (define (user->str user) |
| 72 | (match-define (User u n) user) |
| 73 | (if n |
| 74 | (format "+~a; @~a" u n) |
| 75 | (format "+~a" u ))) |
| 76 | |
| 77 | (: user-agent->str (-> User-Agent String)) |
| 78 | (define (user-agent->str ua) |
| 79 | (match-define (User-Agent u p) ua) |
| 80 | (format "~a/~a (~a)" (Prog-name p) (Prog-version p) (user->str u))) |
| 81 | |
| 82 | (: user->user-agent User) |
| 83 | (define (user->user-agent user) |
| 84 | (User-Agent user prog)) |
| 85 | |
| 86 | (: user-agent-str String) |
| 87 | (define user-agent-str |
| 88 | (user-agent->str (user->user-agent user-default))) |
| 89 | |
| 90 | (: set-user-agent-str (-> Path-String Void)) |
| 91 | (define (set-user-agent-str filename) |
| 92 | (set! user-agent-str (user-agent->str (user->user-agent (file->user filename)))) |
| 93 | (log-info "User-Agent string is now set to: ~v" user-agent-str)) |
| 94 | |
| 95 | (: file->user (-> Path-String User)) |
| 96 | (define (file->user filename) |
| 97 | (if (file-exists? filename) |
| 98 | (match (set->list (file->peers filename)) |
| 99 | [(list p) |
| 100 | (log-info |
| 101 | "User-Agent. Found one peer in file: ~v. Using the found peer: ~a" |
| 102 | filename |
| 103 | (peer->str p)) |
| 104 | (peer->user p)] |
| 105 | [(list* p _) |
| 106 | (log-warning |
| 107 | "User-Agent. Multiple peers in file: ~v. Picking arbitrary: ~a" |
| 108 | filename |
| 109 | (peer->str p)) |
| 110 | (peer->user p)] |
| 111 | ['() |
| 112 | (log-warning |
| 113 | "User-Agent. No peers found in file: ~v. Using the default user: ~a" |
| 114 | filename |
| 115 | user-default) |
| 116 | user-default]) |
| 117 | (begin |
| 118 | (log-warning |
| 119 | "User-Agent. File doesn't exist: ~v. Using the default user: ~a" |
| 120 | filename |
| 121 | user-default) |
| 122 | user-default))) |
| 123 | |
| 124 | (define (peer->user p) |
| 125 | (match-define (Peer n _ u _) p) |
| 126 | (User u n)) |
| 127 | |
| 128 | (: peers-equal? (-> Peer Peer Boolean)) |
| 129 | (define (peers-equal? p1 p2) |
| 130 | (equal? (Peer-uri-str p1) |
| 131 | (Peer-uri-str p2))) |
| 132 | |
| 133 | (: peer-hash (-> Peer Fixnum)) |
| 134 | (define (peer-hash p) |
| 135 | (equal-hash-code (Peer-uri-str p))) |
| 136 | |
| 137 | (define-custom-set-types peers |
| 138 | #:elem? Peer? |
| 139 | peers-equal? |
| 140 | peer-hash) |
| 141 | ; XXX Without supplying above explicit hash procedure, we INTERMITTENTLY get |
| 142 | ; the following contract violations: |
| 143 | ; |
| 144 | ; custom-elem-contents: contract violation |
| 145 | ; expected: custom-elem? |
| 146 | ; given: #f |
| 147 | ; context...: |
| 148 | ; /usr/share/racket/collects/racket/private/set-types.rkt:104:0: custom-set->list |
| 149 | ; /home/siraaj/proj/pub/tt/tt.rkt:716:0: crawl |
| 150 | ; /usr/share/racket/collects/racket/cmdline.rkt:191:51 |
| 151 | ; body of (submod "/home/siraaj/proj/pub/tt/tt.rkt" main) |
| 152 | ; |
| 153 | ; TODO Investigate why and make a minimal reproducible test case. |
| 154 | |
| 155 | (define (peers-union . peer-sets) |
| 156 | (define groups |
| 157 | (foldl |
| 158 | (λ (p groups) |
| 159 | (hash-update groups (Peer-uri-str p) (λ (group) (cons p group)) '())) |
| 160 | (hash) |
| 161 | (append* (map set->list peer-sets)))) |
| 162 | (define (merge peers) |
| 163 | (match peers |
| 164 | ['() (raise 'impossible)] |
| 165 | [(list p) p] |
| 166 | [(list* p1 p2 ps) |
| 167 | (let* ([n1 (Peer-nick p1)] |
| 168 | [n2 (Peer-nick p2)] |
| 169 | [p (cond |
| 170 | [(and (not n1) (not n2)) p1] |
| 171 | [(and n1 n2 ) p1] |
| 172 | [(and n1 (not n2)) p1] |
| 173 | [(and (not n1) n2) p2] |
| 174 | [else |
| 175 | (raise 'impossible)])]) |
| 176 | (merge (cons p ps)))])) |
| 177 | (make-immutable-peers (map merge (hash-values groups)))) |
| 178 | |
| 179 | (module+ test |
| 180 | (let* ([u1 "http://foo/bar"] |
| 181 | [u2 "http://baz/quux"] |
| 182 | [p1 (Peer #f (string->url u1) u1 #f)] |
| 183 | [p2 (Peer "a" (string->url u1) u1 #f)] |
| 184 | [p3 (Peer "b" (string->url u2) u2 #f)] |
| 185 | [s1 (make-immutable-peers (list p1))] |
| 186 | [s2 (make-immutable-peers (list p2 p3))]) |
| 187 | (check-true (peers? (peers-union s1 s2))) |
| 188 | (check-true (peers? (peers-union s2 s1))) |
| 189 | (check-equal? (list p3 p2) (set->list (peers-union s1 s2))) |
| 190 | (check-equal? (list p3 p2) (set->list (peers-union s2 s1))))) |
| 191 | |
| 192 | (: tt-home-dir Path-String) |
| 193 | (define tt-home-dir (build-path (expand-user-path "~") ".tt")) |
| 194 | |
| 195 | (: concurrent-filter-map (∀ (α β) (-> Natural (-> α β) (Listof α) (Listof β)))) |
| 196 | (define (concurrent-filter-map num-workers f xs) |
| 197 | ; TODO preserve order of elements OR communicate that reorder is expected |
| 198 | ; TODO switch from mailboxes to channels |
| 199 | (define (make-worker id f) |
| 200 | (define parent (current-thread)) |
| 201 | (λ () |
| 202 | (define self : Thread (current-thread)) |
| 203 | (: work (∀ (α) (-> α))) |
| 204 | (define (work) |
| 205 | (thread-send parent (cons 'next self)) |
| 206 | (match (thread-receive) |
| 207 | ['done (thread-send parent (cons 'exit id))] |
| 208 | [(cons 'unit x) (begin |
| 209 | (define y (f x)) |
| 210 | (when y (thread-send parent (cons 'result y))) |
| 211 | (work))])) |
| 212 | (work))) |
| 213 | (: dispatch (∀ (α β) (-> (Listof Nonnegative-Integer) (Listof α) (Listof β)))) |
| 214 | (define (dispatch ws xs ys) |
| 215 | (if (empty? ws) |
| 216 | ys |
| 217 | (match (thread-receive) |
| 218 | [(cons 'exit w) (dispatch (remove w ws =) xs ys)] |
| 219 | [(cons 'result y) (dispatch ws xs (cons y ys))] |
| 220 | [(cons 'next thd) (match xs |
| 221 | ['() (begin |
| 222 | (thread-send thd 'done) |
| 223 | (dispatch ws xs ys))] |
| 224 | [(cons x xs) (begin |
| 225 | (thread-send thd (cons 'unit x)) |
| 226 | (dispatch ws xs ys))])]))) |
| 227 | (define workers (range num-workers)) |
| 228 | (define threads (map (λ (id) (thread (make-worker id f))) workers)) |
| 229 | (define results (dispatch workers xs '())) |
| 230 | (for-each thread-wait threads) |
| 231 | results) |
| 232 | |
| 233 | (module+ test |
| 234 | (let* ([f (λ (x) (if (even? x) x #f))] |
| 235 | [xs (range 11)] |
| 236 | [actual (sort (concurrent-filter-map 10 f xs) <)] |
| 237 | [expected (sort ( filter-map f xs) <)]) |
| 238 | (check-equal? actual expected "concurrent-filter-map"))) |
| 239 | |
| 240 | (: msg-print (-> Out-Format Integer Msg Void)) |
| 241 | (define msg-print |
| 242 | (let* ([colors (vector 36 33)] |
| 243 | [n (vector-length colors)]) |
| 244 | (λ (out-format color-i msg) |
| 245 | (let ([color (vector-ref colors (modulo color-i n))] |
| 246 | [nick (Peer-nick (Msg-from msg))] |
| 247 | [uri (Peer-uri-str (Msg-from msg))] |
| 248 | [text (Msg-text msg)]) |
| 249 | (match out-format |
| 250 | ['single-line |
| 251 | (let ([nick (if nick nick uri)]) |
| 252 | (printf "~a \033[1;37m<~a>\033[0m \033[0;~am~a\033[0m~n" |
| 253 | (parameterize |
| 254 | ([date-display-format 'iso-8601]) |
| 255 | (date->string (seconds->date (Msg-ts-epoch msg)) #t)) |
| 256 | nick color text))] |
| 257 | ['multi-line |
| 258 | (let ([nick (if nick (string-append nick " ") "")]) |
| 259 | (printf "~a (~a)~n\033[1;37m<~a~a>\033[0m~n\033[0;~am~a\033[0m~n~n" |
| 260 | (parameterize |
| 261 | ([date-display-format 'rfc2822]) |
| 262 | (date->string (seconds->date (Msg-ts-epoch msg)) #t)) |
| 263 | (Msg-ts-orig msg) |
| 264 | nick uri color text))]))))) |
| 265 | |
| 266 | (: rfc3339->epoch (-> String (Option Nonnegative-Integer))) |
| 267 | (define rfc3339->epoch |
| 268 | (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}))?$")]) |
| 269 | (λ (ts) |
| 270 | (match (regexp-match re ts) |
| 271 | [(list _wholething yyyy mm dd HH MM _:SS SS _fractional tz-whole tz-sign tz-HH tz-MM) |
| 272 | (let* |
| 273 | ([tz-offset |
| 274 | (match* (tz-whole tz-sign tz-HH tz-MM) |
| 275 | [("Z" #f #f #f) |
| 276 | 0] |
| 277 | [(_ (or "-" "+") (? identity h) (? identity m)) |
| 278 | (let ([h (string->number h)] |
| 279 | [m (string->number m)] |
| 280 | ; Reverse to get back to UTC: |
| 281 | [op (match tz-sign ["+" -] ["-" +])]) |
| 282 | (op 0 (+ (* 60 m) (* 60 (* 60 h)))))] |
| 283 | [(a b c d) |
| 284 | (log-warning "Impossible TZ string: ~v, components: ~v ~v ~v ~v" tz-whole a b c d) |
| 285 | 0])] |
| 286 | [ts-orig ts] |
| 287 | [local-time? #f] |
| 288 | [ts-epoch (find-seconds (if SS (string->number SS) 0) |
| 289 | (string->number MM) |
| 290 | (string->number HH) |
| 291 | (string->number dd) |
| 292 | (string->number mm) |
| 293 | (string->number yyyy) |
| 294 | local-time?)]) |
| 295 | (+ ts-epoch tz-offset))] |
| 296 | [_ |
| 297 | (log-debug "Invalid timestamp: ~v" ts) |
| 298 | #f])))) |
| 299 | |
| 300 | (: str->msg (-> Peer String (Option Msg))) |
| 301 | (define str->msg |
| 302 | (let ([re (pregexp "^([^\\s\t]+)[\\s\t]+(.*)$")]) |
| 303 | (λ (from str) |
| 304 | (define from-str (peer->str from)) |
| 305 | (define str-head (substring str 0 (min 100 (string-length str)))) |
| 306 | (with-handlers* |
| 307 | ([exn:fail? |
| 308 | (λ (e) |
| 309 | (log-debug |
| 310 | "Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v" |
| 311 | str-head from-str e) |
| 312 | #f)]) |
| 313 | (match (regexp-match re str) |
| 314 | [(list _wholething ts-orig text) |
| 315 | (let ([ts-epoch (rfc3339->epoch ts-orig)]) |
| 316 | (if ts-epoch |
| 317 | (let ([mentions |
| 318 | (filter-map |
| 319 | (λ (m) (match (regexp-match #px"@<([^>]+)>" m) |
| 320 | [(list _wholething nick-uri) |
| 321 | (str->peer nick-uri)])) |
| 322 | (regexp-match* #px"@<[^\\s]+([\\s]+)?[^>]+>" text))]) |
| 323 | (Msg ts-epoch ts-orig from text mentions)) |
| 324 | (begin |
| 325 | (log-debug |
| 326 | "Msg rejected due to invalid timestamp. From:~v. Line:~v" |
| 327 | from-str str-head) |
| 328 | #f)))] |
| 329 | [_ |
| 330 | (log-debug "Non-msg line. From:~v. Line:~v" from-str str-head) |
| 331 | #f]))))) |
| 332 | |
| 333 | (module+ test |
| 334 | ; TODO Test for when missing-nick case |
| 335 | (let* ([tzs (for*/list ([d '("-" "+")] |
| 336 | [h '("5" "05")] |
| 337 | [m '("00" ":00" "57" ":57")]) |
| 338 | (string-append d h m))] |
| 339 | [tzs (list* "" "Z" tzs)]) |
| 340 | (for* ([n '("fake-nick")] |
| 341 | [u '("http://fake-uri")] |
| 342 | [p (list (Peer n (string->url u) u #f))] |
| 343 | [s '("" ":10")] |
| 344 | [f '("" ".1337")] |
| 345 | [z tzs] |
| 346 | [sep (list "\t" " ")] |
| 347 | [txt '("foo bar baz" "'jaz poop bear giraffe / tea" "@*\"``")]) |
| 348 | (let* ([ts (string-append "2020-11-18T22:22" |
| 349 | (if (non-empty-string? s) s ":00") |
| 350 | z)] |
| 351 | [m (str->msg p (string-append ts sep txt))]) |
| 352 | (check-not-false m) |
| 353 | (check-equal? (Msg-from m) p) |
| 354 | (check-equal? (Msg-text m) txt) |
| 355 | (check-equal? (Msg-ts-orig m) ts (format "Given: ~v" ts)) |
| 356 | ))) |
| 357 | |
| 358 | (let* ([ts "2020-11-18T22:22:09-0500"] |
| 359 | [tab " "] |
| 360 | [text "Lorem ipsum"] |
| 361 | [nick "foo"] |
| 362 | [uri "http://bar/"] |
| 363 | [peer (Peer nick (string->url uri) uri #f)] |
| 364 | [actual (str->msg peer (string-append ts tab text))] |
| 365 | [expected (Msg 1605756129 ts peer text '())]) |
| 366 | (check-equal? |
| 367 | (Msg-ts-epoch actual) |
| 368 | (Msg-ts-epoch expected) |
| 369 | "str->msg ts-epoch") |
| 370 | (check-equal? |
| 371 | (Msg-ts-orig actual) |
| 372 | (Msg-ts-orig expected) |
| 373 | "str->msg ts-orig") |
| 374 | (check-equal? |
| 375 | (Peer-nick (Msg-from actual)) |
| 376 | (Peer-nick (Msg-from expected)) |
| 377 | "str->msg nick") |
| 378 | (check-equal? |
| 379 | (Peer-uri (Msg-from actual)) |
| 380 | (Peer-uri (Msg-from expected)) |
| 381 | "str->msg uri") |
| 382 | (check-equal? |
| 383 | (Peer-uri-str (Msg-from actual)) |
| 384 | (Peer-uri-str (Msg-from expected)) |
| 385 | "str->msg uri-str") |
| 386 | (check-equal? |
| 387 | (Msg-text actual) |
| 388 | (Msg-text expected) |
| 389 | "str->msg text"))) |
| 390 | |
| 391 | (: str->lines (-> String (Listof String))) |
| 392 | (define (str->lines str) |
| 393 | (string-split str (regexp "[\r\n]+"))) |
| 394 | |
| 395 | (module+ test |
| 396 | (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi"))) |
| 397 | |
| 398 | (: str->msgs (-> Peer String (Listof Msg))) |
| 399 | (define (str->msgs peer str) |
| 400 | (filter-map (λ (line) (str->msg peer line)) |
| 401 | (filter-comments (str->lines str)))) |
| 402 | |
| 403 | (: cache-dir Path-String) |
| 404 | (define cache-dir (build-path tt-home-dir "cache")) |
| 405 | |
| 406 | (define cache-object-dir (build-path cache-dir "objects")) |
| 407 | |
| 408 | (: url->cache-file-path-v1 (-> Url Path-String)) |
| 409 | (define (url->cache-file-path-v1 uri) |
| 410 | (define (hash-sha1 str) : (-> String String) |
| 411 | (define in (open-input-string str)) |
| 412 | (define digest (sha1 in)) |
| 413 | (close-input-port in) |
| 414 | digest) |
| 415 | (build-path cache-object-dir (hash-sha1 (url->string uri)))) |
| 416 | |
| 417 | (: url->cache-file-path-v2 (-> Url Path-String)) |
| 418 | (define (url->cache-file-path-v2 uri) |
| 419 | (build-path cache-object-dir (uri-encode (url->string uri)))) |
| 420 | |
| 421 | (define url->cache-object-path |
| 422 | url->cache-file-path-v2) |
| 423 | |
| 424 | (define (url->cache-etag-path uri) |
| 425 | (build-path cache-dir "etags" (uri-encode (url->string uri)))) |
| 426 | |
| 427 | (define (url->cache-lmod-path uri) |
| 428 | (build-path cache-dir "lmods" (uri-encode (url->string uri)))) |
| 429 | |
| 430 | (: uri-read-cached (-> Url (Option String))) |
| 431 | (define (uri-read-cached uri) |
| 432 | (define path-v1 (url->cache-file-path-v1 uri)) |
| 433 | (define path-v2 (url->cache-file-path-v2 uri)) |
| 434 | (when (file-exists? path-v1) |
| 435 | (rename-file-or-directory path-v1 path-v2 #t)) |
| 436 | (if (file-exists? path-v2) |
| 437 | (file->string path-v2) |
| 438 | (begin |
| 439 | (log-debug "Cache file not found for URI: ~a" (url->string uri)) |
| 440 | #f))) |
| 441 | |
| 442 | (: str->url (-> String (Option String))) |
| 443 | (define (str->url s) |
| 444 | (with-handlers* |
| 445 | ([exn:fail? (λ (e) #f)]) |
| 446 | (string->url s))) |
| 447 | |
| 448 | (: peer->str (-> Peer String)) |
| 449 | (define (peer->str peer) |
| 450 | (match-define (Peer n _ u c) peer) |
| 451 | (format "~a~a~a" |
| 452 | (if n (format "~a " n) "") |
| 453 | u |
| 454 | (if c (format " # ~a" c) ""))) |
| 455 | |
| 456 | (: str->peer (-> String (Option Peer))) |
| 457 | (define (str->peer str) |
| 458 | (log-debug "Parsing peer string: ~v" str) |
| 459 | (match |
| 460 | (regexp-match |
| 461 | #px"(([^\\s\t]+)[\\s\t]+)?([a-zA-Z]+://[^\\s\t]*)[\\s\t]*(#\\s*(.*))?" |
| 462 | str) |
| 463 | [(list _wholething |
| 464 | _nick-with-space |
| 465 | nick |
| 466 | url |
| 467 | _comment-with-hash |
| 468 | comment) |
| 469 | (match (str->url url) |
| 470 | [#f |
| 471 | (log-error "Invalid URI in peer string: ~v" str) |
| 472 | #f] |
| 473 | [url |
| 474 | (Peer nick url (url->string url) comment)])] |
| 475 | [_ |
| 476 | (log-debug "Invalid peer string: ~v" str) |
| 477 | #f])) |
| 478 | |
| 479 | (module+ test |
| 480 | (check-equal? |
| 481 | (str->peer "foo http://bar/file.txt # some rando") |
| 482 | (Peer "foo" (str->url "http://bar/file.txt") "http://bar/file.txt" "some rando")) |
| 483 | (check-equal? |
| 484 | (str->peer "http://bar/file.txt # some rando") |
| 485 | (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" "some rando")) |
| 486 | (check-equal? |
| 487 | (str->peer "http://bar/file.txt #") |
| 488 | (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" "")) |
| 489 | (check-equal? |
| 490 | (str->peer "http://bar/file.txt#") ; XXX URLs can have #s |
| 491 | (Peer #f (str->url "http://bar/file.txt#") "http://bar/file.txt#" #f)) |
| 492 | (check-equal? |
| 493 | (str->peer "http://bar/file.txt") |
| 494 | (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" #f)) |
| 495 | (check-equal? |
| 496 | (str->peer "foo http://bar/file.txt") |
| 497 | (Peer "foo" (str->url "http://bar/file.txt") "http://bar/file.txt" #f)) |
| 498 | (check-equal? |
| 499 | (str->peer "foo bar # baz") |
| 500 | #f) |
| 501 | (check-equal? |
| 502 | (str->peer "foo bar://baz # quux") |
| 503 | (Peer "foo" (str->url "bar://baz") "bar://baz" "quux")) |
| 504 | (check-equal? |
| 505 | (str->peer "foo bar//baz # quux") |
| 506 | #f)) |
| 507 | |
| 508 | (: filter-comments (-> (Listof String) (Listof String))) |
| 509 | (define (filter-comments lines) |
| 510 | (filter-not (λ (line) (string-prefix? line "#")) lines)) |
| 511 | |
| 512 | (: str->peers (-> String (Setof Peer))) |
| 513 | (define (str->peers str) |
| 514 | (make-immutable-peers (filter-map str->peer (filter-comments (str->lines str))))) |
| 515 | |
| 516 | (: peers->file (-> (Setof Peers) Path-String Void)) |
| 517 | (define (peers->file peers path) |
| 518 | (display-lines-to-file |
| 519 | (map peer->str |
| 520 | (sort (set->list peers) |
| 521 | (match-lambda** |
| 522 | [((Peer n1 _ _ _) (Peer n2 _ _ _)) |
| 523 | (string<? (if n1 n1 "") |
| 524 | (if n2 n2 ""))]))) |
| 525 | path |
| 526 | #:exists 'replace)) |
| 527 | |
| 528 | (: file->peers (-> Path-String (Setof Peer))) |
| 529 | (define (file->peers file-path) |
| 530 | (if (file-exists? file-path) |
| 531 | (str->peers (file->string file-path)) |
| 532 | (begin |
| 533 | (log-warning "File does not exist: ~v" (path->string file-path)) |
| 534 | (make-immutable-peers)))) |
| 535 | |
| 536 | (define re-rfc2822 |
| 537 | #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") |
| 538 | |
| 539 | (: b->n (-> Bytes (Option Number))) |
| 540 | (define (b->n b) |
| 541 | (string->number (bytes->string/utf-8 b))) |
| 542 | |
| 543 | (: mon->num (-> Bytes Natural)) |
| 544 | (define/match (mon->num mon) |
| 545 | [(#"Jan") 1] |
| 546 | [(#"Feb") 2] |
| 547 | [(#"Mar") 3] |
| 548 | [(#"Apr") 4] |
| 549 | [(#"May") 5] |
| 550 | [(#"Jun") 6] |
| 551 | [(#"Jul") 7] |
| 552 | [(#"Aug") 8] |
| 553 | [(#"Sep") 9] |
| 554 | [(#"Oct") 10] |
| 555 | [(#"Nov") 11] |
| 556 | [(#"Dec") 12]) |
| 557 | |
| 558 | (: rfc2822->epoch (-> Bytes (Option Nonnegative-Integer))) |
| 559 | (define (rfc2822->epoch timestamp) |
| 560 | (match (regexp-match re-rfc2822 timestamp) |
| 561 | [(list _ _ dd mo yyyy HH MM SS) |
| 562 | #:when (and dd mo yyyy HH MM SS) |
| 563 | (find-seconds (b->n SS) |
| 564 | (b->n MM) |
| 565 | (b->n HH) |
| 566 | (b->n dd) |
| 567 | (mon->num mo) |
| 568 | (b->n yyyy) |
| 569 | #f)] |
| 570 | [_ |
| 571 | #f])) |
| 572 | |
| 573 | (: header-get (-> (Listof Bytes) Bytes (Option Bytes))) |
| 574 | (define (header-get headers name) |
| 575 | (match (filter-map (curry extract-field name) headers) |
| 576 | [(list val) val] |
| 577 | [_ #f])) |
| 578 | |
| 579 | (: uri-download-from-port |
| 580 | (-> Url (Listof (U Bytes String)) Input-Port |
| 581 | (U 'skipped-cached 'downloaded-new))) ; TODO 'ok|'error ? |
| 582 | (define (uri-download-from-port u headers body-input) |
| 583 | (define u-str (url->string u)) |
| 584 | (log-debug "uri-download-from-port ~v into ~v" u-str cached-object-path) |
| 585 | (define cached-object-path (url->cache-object-path u)) |
| 586 | (define cached-etag-path (url->cache-etag-path u)) |
| 587 | (define cached-lmod-path (url->cache-lmod-path u)) |
| 588 | (define etag (header-get headers #"ETag")) |
| 589 | (define lmod (header-get headers #"Last-Modified")) |
| 590 | (define lmod-curr (if lmod (rfc2822->epoch lmod) #f)) |
| 591 | (define lmod-prev (if (file-exists? cached-lmod-path) |
| 592 | (rfc2822->epoch (file->bytes cached-lmod-path)) |
| 593 | #f)) |
| 594 | (log-debug "lmod-curr:~v lmod-prev:~v" lmod-curr lmod-prev) |
| 595 | (define cached? |
| 596 | (or (and etag |
| 597 | (file-exists? cached-etag-path) |
| 598 | (bytes=? etag (file->bytes cached-etag-path)) |
| 599 | (begin |
| 600 | (log-debug "ETags match, skipping the rest of ~v" u-str) |
| 601 | #t)) |
| 602 | (and lmod-curr |
| 603 | lmod-prev |
| 604 | (<= lmod-curr lmod-prev) |
| 605 | (begin |
| 606 | (log-debug "Last-Modified <= current skipping the rest of ~v" u-str) |
| 607 | #t)))) |
| 608 | (if (not cached?) |
| 609 | (begin |
| 610 | (log-debug |
| 611 | "Downloading the rest of ~v. ETag: ~a, Last-Modified: ~v" |
| 612 | u-str etag lmod) |
| 613 | (make-parent-directory* cached-object-path) |
| 614 | (make-parent-directory* cached-etag-path) |
| 615 | (make-parent-directory* cached-lmod-path) |
| 616 | (call-with-output-file cached-object-path |
| 617 | (curry copy-port body-input) |
| 618 | #:exists 'replace) |
| 619 | (when etag |
| 620 | (display-to-file etag cached-etag-path #:exists 'replace)) |
| 621 | (when lmod |
| 622 | (display-to-file lmod cached-lmod-path #:exists 'replace)) |
| 623 | 'downloaded-new) |
| 624 | 'skipped-cached)) |
| 625 | |
| 626 | (: uri-download |
| 627 | (-> Positive-Float Url |
| 628 | (Result (U 'skipped-cached 'downloaded-new) |
| 629 | Any))) ; TODO Maybe more-precise error type? |
| 630 | (define (uri-download timeout u) |
| 631 | (define u-str (url->string u)) |
| 632 | (define timeout-chan (make-channel)) |
| 633 | (define result-chan (make-channel)) |
| 634 | (define timeout-thread |
| 635 | (thread (λ () |
| 636 | ; Doing this instead of sync/timeout to distinguish error values, |
| 637 | ; rather than just have #f to work with. |
| 638 | (sleep timeout) |
| 639 | (channel-put timeout-chan '(error . timeout))))) |
| 640 | (define result-thread |
| 641 | (thread (λ () |
| 642 | ; XXX We timeout getting a response, but body download could |
| 643 | ; also take a long time and we might want to time that out as |
| 644 | ; well, but then we may end-up with partially downloaded |
| 645 | ; objects. But that could happen anyway if the server drops the |
| 646 | ; connection for whatever reason. |
| 647 | ; |
| 648 | ; Maybe that is OK once we start treating the |
| 649 | ; downloaded object as an addition to the stored set of |
| 650 | ; messages, rather than the final set of messages. |
| 651 | |
| 652 | ; TODO message db |
| 653 | ; - 1st try can just be an in-memory set that gets written-to |
| 654 | ; and read-from disk as a whole. |
| 655 | (define result |
| 656 | (with-handlers |
| 657 | ; TODO Maybe name each known errno? (exn:fail:network:errno-errno e) |
| 658 | ([exn:fail:network? |
| 659 | (λ (e) `(error . (net-error . ,e)))] |
| 660 | [exn? |
| 661 | (λ (e) `(error . (other . ,e)))]) |
| 662 | (define-values (status-line headers body-input) |
| 663 | (http-sendrecv/url |
| 664 | u |
| 665 | #:headers (list (format "User-Agent: ~a" user-agent-str)))) |
| 666 | `(ok . ,(Resp status-line headers body-input)))) |
| 667 | (channel-put result-chan result)))) |
| 668 | (define result |
| 669 | (sync timeout-chan |
| 670 | result-chan)) |
| 671 | (kill-thread result-thread) |
| 672 | (kill-thread timeout-thread) |
| 673 | (match result |
| 674 | [(cons 'error _) |
| 675 | result] |
| 676 | [(cons 'ok (Resp status-line headers body-input)) |
| 677 | (log-debug "headers: ~v" headers) |
| 678 | (log-debug "status-line: ~v" status-line) |
| 679 | (define status |
| 680 | (string->number (second (string-split (bytes->string/utf-8 status-line))))) |
| 681 | (log-debug "status: ~v" status) |
| 682 | ; TODO Handle redirects. Should be within same timeout as req and body. |
| 683 | (let ([result |
| 684 | (match status |
| 685 | [200 |
| 686 | `(ok . ,(uri-download-from-port u headers body-input))] |
| 687 | [_ |
| 688 | `(error . (http . ,status))])]) |
| 689 | (close-input-port body-input) |
| 690 | result)])) |
| 691 | |
| 692 | (: timeline-print (-> Out-Format (Listof Msg) Void)) |
| 693 | (define (timeline-print out-format timeline) |
| 694 | (match timeline |
| 695 | ['() |
| 696 | (void)] |
| 697 | [(cons first-msg _) |
| 698 | (void (foldl (match-lambda** |
| 699 | [((and m (Msg _ _ from _ _)) (cons prev-from i)) |
| 700 | (let ([i (if (peers-equal? prev-from from) i (+ 1 i))]) |
| 701 | (msg-print out-format i m) |
| 702 | (cons from i))]) |
| 703 | (cons (Msg-from first-msg) 0) |
| 704 | timeline))])) |
| 705 | |
| 706 | (: peer->msgs (-> Peer (Listof Msg))) |
| 707 | (define (peer->msgs peer) |
| 708 | (match-define (Peer nick uri uri-str _) peer) |
| 709 | (log-debug "Reading peer nick:~v uri:~v" nick uri-str) |
| 710 | (define msgs-data (uri-read-cached uri)) |
| 711 | ; TODO Expire cache |
| 712 | (if msgs-data |
| 713 | (str->msgs peer msgs-data) |
| 714 | '())) |
| 715 | |
| 716 | (: peer-download |
| 717 | (-> Positive-Float Peer |
| 718 | (Result (U 'skipped-cached 'downloaded-new) |
| 719 | Any))) |
| 720 | (define (peer-download timeout peer) |
| 721 | (match-define (Peer nick uri u _) peer) |
| 722 | (log-info "Download BEGIN URL:~a" u) |
| 723 | (define-values (results _tm-cpu-ms tm-real-ms _tm-gc-ms) |
| 724 | (time-apply uri-download (list timeout uri))) |
| 725 | (define result (car results)) |
| 726 | (log-info "Download END in ~a seconds, URL:~a, result:~s" |
| 727 | (/ tm-real-ms 1000.0) |
| 728 | u |
| 729 | result) |
| 730 | result) |
| 731 | |
| 732 | (: timeline-download (-> Integer Positive-Float (Setof Peer) Void)) |
| 733 | (define (timeline-download num-workers timeout peers) |
| 734 | (define results |
| 735 | (concurrent-filter-map num-workers |
| 736 | (λ (p) (cons p (peer-download timeout p))) |
| 737 | (set->list peers))) |
| 738 | (define peers-ok |
| 739 | (filter-map (match-lambda |
| 740 | [(cons p (cons 'ok _)) p] |
| 741 | [(cons _ (cons 'error e)) #f]) |
| 742 | results)) |
| 743 | (define peers-err |
| 744 | (filter-map (match-lambda |
| 745 | [(cons _ (cons 'ok _)) |
| 746 | #f] |
| 747 | [(cons p (cons 'error e)) |
| 748 | (struct-copy Peer p [comment (format "~s" e)])]) |
| 749 | results)) |
| 750 | (peers->file peers-ok (build-path tt-home-dir "peers-last-downloaded-ok")) |
| 751 | (peers->file peers-err (build-path tt-home-dir "peers-last-downloaded-err"))) |
| 752 | |
| 753 | (: peers->timeline (-> (Setof Peer) (Listof Msg))) |
| 754 | (define (peers->timeline peers) |
| 755 | (append* (filter-map peer->msgs (set->list peers)))) |
| 756 | |
| 757 | (: timeline-sort (-> (Listof Msg) timeline-order (Listof Msgs))) |
| 758 | (define (timeline-sort msgs order) |
| 759 | (define cmp (match order |
| 760 | ['old->new <] |
| 761 | ['new->old >])) |
| 762 | (sort msgs (λ (a b) (cmp (Msg-ts-epoch a) |
| 763 | (Msg-ts-epoch b))))) |
| 764 | |
| 765 | (: paths->peers (-> (Listof String) (Setof Peer))) |
| 766 | (define (paths->peers paths) |
| 767 | (let* ([paths (match paths |
| 768 | ['() |
| 769 | (let ([peer-refs-file (build-path tt-home-dir "peers")]) |
| 770 | (log-debug |
| 771 | "No peer ref file paths provided, defaulting to ~v" |
| 772 | (path->string peer-refs-file)) |
| 773 | (list peer-refs-file))] |
| 774 | [paths |
| 775 | (log-debug "Peer ref file paths provided: ~v" paths) |
| 776 | (map string->path paths)])] |
| 777 | [peers (apply peers-union (map file->peers paths))]) |
| 778 | (log-info "Read-in ~a peers." (set-count peers)) |
| 779 | peers)) |
| 780 | |
| 781 | (: cache-filename->peer (-> Path-String (Option Peer))) |
| 782 | (define (cache-filename->peer filename) |
| 783 | (define nick #f) ; TODO Look it up in the nick-db when it exists. |
| 784 | (define url-str (uri-decode (path->string filename))) ; TODO Can these crash? |
| 785 | (match (str->url url-str) |
| 786 | [#f #f] |
| 787 | [url (Peer nick url url-str #f)])) |
| 788 | |
| 789 | (: peers-cached (-> (Setof Peer))) |
| 790 | (define (peers-cached) |
| 791 | ; TODO Expire cache? |
| 792 | (make-immutable-peers |
| 793 | (filter-map cache-filename->peer |
| 794 | (directory-list cache-object-dir)))) |
| 795 | |
| 796 | (: peers-mentioned (-> (Listof Msg) (Setof Peer))) |
| 797 | (define (peers-mentioned msgs) |
| 798 | (make-immutable-peers (append* (map Msg-mentions msgs)))) |
| 799 | |
| 800 | (: log-writer-stop (-> Thread Void)) |
| 801 | (define (log-writer-stop log-writer) |
| 802 | (log-message (current-logger) 'fatal 'stop "Exiting." #f) |
| 803 | (thread-wait log-writer)) |
| 804 | |
| 805 | (: log-writer-start (-> Log-Level Thread)) |
| 806 | (define (log-writer-start level) |
| 807 | (let* ([logger |
| 808 | (make-logger #f #f level #f)] |
| 809 | [log-receiver |
| 810 | (make-log-receiver logger level)] |
| 811 | [log-writer |
| 812 | (thread |
| 813 | (λ () |
| 814 | (parameterize |
| 815 | ([date-display-format 'iso-8601]) |
| 816 | (let loop () |
| 817 | (match-define (vector level msg _ topic) (sync log-receiver)) |
| 818 | (unless (equal? topic 'stop) |
| 819 | (eprintf "~a [~a] ~a~n" (date->string (current-date) #t) level msg) |
| 820 | (loop))))))]) |
| 821 | (current-logger logger) |
| 822 | log-writer)) |
| 823 | |
| 824 | (: crawl (-> Void)) |
| 825 | (define (crawl) |
| 826 | (let* ([peers-all-file |
| 827 | (build-path tt-home-dir "peers-all")] |
| 828 | [peers-mentioned-file |
| 829 | (build-path tt-home-dir "peers-mentioned")] |
| 830 | [peers-parsed-file |
| 831 | (build-path tt-home-dir "peers-parsed")] |
| 832 | [peers-cached-file |
| 833 | (build-path tt-home-dir "peers-cached")] |
| 834 | [peers-cached |
| 835 | (peers-cached)] |
| 836 | [cached-timeline |
| 837 | (peers->timeline peers-cached)] |
| 838 | [peers-mentioned-curr |
| 839 | (peers-mentioned cached-timeline)] |
| 840 | [peers-mentioned-prev |
| 841 | (file->peers peers-mentioned-file)] |
| 842 | [peers-mentioned |
| 843 | (peers-union peers-mentioned-prev |
| 844 | peers-mentioned-curr)] |
| 845 | [peers-all-prev |
| 846 | (file->peers peers-all-file)] |
| 847 | [peers-all |
| 848 | (peers-union peers-mentioned |
| 849 | peers-all-prev |
| 850 | peers-cached)] |
| 851 | [peers-discovered |
| 852 | (set-subtract peers-all |
| 853 | peers-all-prev)] |
| 854 | [peers-parsed |
| 855 | (for/set ([p peers-all] #:when (> (length (peer->msgs p)) 0)) p)]) |
| 856 | ; TODO Deeper de-duping |
| 857 | (log-info "Known peers cached ~a" (set-count peers-cached)) |
| 858 | (log-info "Known peers mentioned: ~a" (set-count peers-mentioned)) |
| 859 | (log-info "Known peers parsed ~a" (set-count peers-parsed)) |
| 860 | (log-info "Known peers total: ~a" (set-count peers-all)) |
| 861 | (log-info "Discovered ~a new peers:~n~a" |
| 862 | (set-count peers-discovered) |
| 863 | (pretty-format (map |
| 864 | (match-lambda |
| 865 | [(Peer n _ u c) (list n u c)]) |
| 866 | (set->list peers-discovered)))) |
| 867 | (peers->file peers-cached |
| 868 | peers-cached-file) |
| 869 | (peers->file peers-mentioned |
| 870 | peers-mentioned-file) |
| 871 | (peers->file peers-parsed |
| 872 | peers-parsed-file) |
| 873 | (peers->file peers-all |
| 874 | peers-all-file))) |
| 875 | |
| 876 | (: read (-> (Listof String) Number Number Timeline-Order Out-Format Void)) |
| 877 | (define (read file-paths ts-min ts-max order out-format) |
| 878 | (let* ([peers |
| 879 | (paths->peers file-paths)] |
| 880 | [msgs |
| 881 | (timeline-sort (peers->timeline peers) order)] |
| 882 | [include? |
| 883 | (λ (m) |
| 884 | (and (or (not ts-min) (>= (Msg-ts-epoch m) ts-min)) |
| 885 | (or (not ts-max) (<= (Msg-ts-epoch m) ts-max))))]) |
| 886 | (timeline-print out-format (filter include? msgs)))) |
| 887 | |
| 888 | (: upload (-> Void)) |
| 889 | (define (upload) |
| 890 | ; FIXME Should not exit from here, but only after cleanup/logger-stoppage. |
| 891 | (if (system (path->string (build-path tt-home-dir "hooks" "upload"))) |
| 892 | (exit 0) |
| 893 | (exit 1))) |
| 894 | |
| 895 | (: download (-> (Listof String) Positive-Integer Positive-Float Void)) |
| 896 | (define (download file-paths num-workers timeout) |
| 897 | (let ([peers (paths->peers file-paths)]) |
| 898 | (define-values (_res _cpu real-ms _gc) |
| 899 | (time-apply timeline-download (list num-workers timeout peers))) |
| 900 | (log-info "Downloaded timelines from ~a peers in ~a seconds." |
| 901 | (set-count peers) |
| 902 | (/ real-ms 1000.0)))) |
| 903 | |
| 904 | (: dispatch (-> String Void)) |
| 905 | (define (dispatch command) |
| 906 | (match command |
| 907 | [(or "d" "download") |
| 908 | ; Initially, 15 was fastest out of the tried: 1, 5, 10, 20. Then I |
| 909 | ; started noticing significant slowdowns. Reducing to 5 seems to help. |
| 910 | (let ([num-workers 5] |
| 911 | [timeout 10.0]) |
| 912 | (command-line |
| 913 | #:program "tt download" |
| 914 | #:once-each |
| 915 | [("-j" "--jobs") |
| 916 | njobs "Number of concurrent jobs." |
| 917 | (set! num-workers (string->number njobs))] |
| 918 | [("-t" "--timeout") |
| 919 | seconds "Timeout seconds per request." |
| 920 | (set! timeout (string->number seconds))] |
| 921 | #:args file-paths |
| 922 | (download file-paths num-workers timeout)))] |
| 923 | [(or "u" "upload") |
| 924 | (command-line |
| 925 | #:program "tt upload" #:args () (upload))] |
| 926 | [(or "r" "read") |
| 927 | (let ([out-format 'multi-line] |
| 928 | [order 'old->new] |
| 929 | [ts-min #f] |
| 930 | [ts-max #f]) |
| 931 | (command-line |
| 932 | #:program "tt read" |
| 933 | #:once-each |
| 934 | [("-r" "--rev") |
| 935 | "Reverse displayed timeline order." |
| 936 | (set! order 'new->old)] |
| 937 | [("-m" "--min") |
| 938 | m "Earliest time to display (ignore anything before it)." |
| 939 | (set! ts-min (rfc3339->epoch m))] |
| 940 | [("-x" "--max") |
| 941 | x "Latest time to display (ignore anything after it)." |
| 942 | (set! ts-max (rfc3339->epoch x))] |
| 943 | #:once-any |
| 944 | [("-s" "--short") |
| 945 | "Short output format" |
| 946 | (set! out-format 'single-line)] |
| 947 | [("-l" "--long") |
| 948 | "Long output format" |
| 949 | (set! out-format 'multi-line)] |
| 950 | #:args file-paths |
| 951 | (read file-paths ts-min ts-max order out-format)))] |
| 952 | [(or "c" "crawl") |
| 953 | (command-line |
| 954 | #:program "tt crawl" #:args () (crawl))] |
| 955 | [command |
| 956 | (eprintf "Error: invalid command: ~v\n" command) |
| 957 | (eprintf "Please use the \"--help\" option to see a list of available commands.\n") |
| 958 | (exit 1)])) |
| 959 | |
| 960 | (module+ main |
| 961 | (let ([log-level 'info]) |
| 962 | (command-line |
| 963 | #:program |
| 964 | "tt" |
| 965 | #:once-each |
| 966 | [("-d" "--debug") |
| 967 | "Enable debug log level." |
| 968 | (set! log-level 'debug)] |
| 969 | #:help-labels |
| 970 | "" |
| 971 | "and <command> is one of" |
| 972 | "r, read : Read the timeline (offline operation)." |
| 973 | "d, download : Download the timeline." |
| 974 | ; TODO Add path dynamically |
| 975 | "u, upload : Upload your twtxt file (alias to execute ~/.tt/hooks/upload)." |
| 976 | "c, crawl : Discover new peers mentioned by known peers (offline operation)." |
| 977 | "" |
| 978 | #:args (command . args) |
| 979 | (define log-writer (log-writer-start log-level)) |
| 980 | (current-command-line-arguments (list->vector args)) |
| 981 | (set-user-agent-str (build-path tt-home-dir "me")) |
| 982 | ; TODO dispatch should return status with which we should exit after cleanups |
| 983 | (dispatch command) |
| 984 | (log-writer-stop log-writer)))) |