tt.rkt

   1 #lang typed/racket/no-check
   2
   3 (require openssl/sha1)
   4 (require racket/date)
   5 (require
   6   net/http-client
   7   net/url-string
   8   net/url-structs)
   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 (struct msg
  27         ([ts-epoch   : Integer]
  28          [ts-orig    : String]
  29          [nick       : String]
  30          [uri        : Url]
  31          [text       : String])
  32         #:type-name Msg)
  33
  34 (struct feed
  35         ([nick : String]
  36          [uri  : Url])
  37         #:type-name Feed)
  38
  39 (: concurrent-filter-map (∀ (α β) (-> Natural (-> α β) (Listof α))))
  40 (define (concurrent-filter-map num-workers f xs)
  41   ; TODO preserve order of elements OR communicate that reorder is expected
  42   ; TODO switch from mailboxes to channels
  43   (define (make-worker id f)
  44     (define parent (current-thread))
  45     (λ ()
  46        (define self : Thread (current-thread))
  47        (: work (∀ (α) (-> α)))
  48        (define (work)
  49          (thread-send parent (cons 'next self))
  50          (match (thread-receive)
  51            ['done          (thread-send parent (cons 'exit id))]
  52            [(cons 'unit x) (begin
  53                              (define y (f x))
  54                              (when y (thread-send parent (cons 'result y)))
  55                              (work))]))
  56        (work)))
  57   (: dispatch (∀ (α β) (-> (Listof Nonnegative-Integer) (Listof α) (Listof β))))
  58   (define (dispatch ws xs ys)
  59     (if (empty? ws)
  60         ys
  61         (match (thread-receive)
  62           [(cons 'exit w)   (dispatch (remove w ws =) xs ys)]
  63           [(cons 'result y) (dispatch ws xs (cons y ys))]
  64           [(cons 'next thd) (match xs
  65                               ['()         (begin
  66                                              (thread-send thd 'done)
  67                                              (dispatch ws xs ys))]
  68                               [(cons x xs) (begin
  69                                              (thread-send thd (cons 'unit x))
  70                                              (dispatch ws xs ys))])])))
  71   (define workers (range num-workers))
  72   (define threads (map (λ (id) (thread (make-worker id f))) workers))
  73   (define results (dispatch workers xs '()))
  74   (for-each thread-wait threads)
  75   results)
  76
  77 (module+ test
  78   (let* ([f        (λ (x) (if (even? x) x #f))]
  79          [xs       (range 11)]
  80          [actual   (sort (concurrent-filter-map 10 f xs) <)]
  81          [expected (sort (           filter-map    f xs) <)])
  82     (check-equal? actual expected "concurrent-filter-map")))
  83
  84 (: msg-print (-> Out-Format Integer Msg Void))
  85 (define msg-print
  86   (let* ([colors (vector 36 33)]
  87          [n      (vector-length colors)])
  88     (λ (out-format color-i msg)
  89        (let ([color (vector-ref colors (modulo color-i n))]
  90              [nick  (msg-nick msg)]
  91              [uri   (url->string (msg-uri msg))]
  92              [text  (msg-text msg)])
  93          (match out-format
  94            ['single-line
  95             (printf "~a  \033[1;37m<~a>\033[0m  \033[0;~am~a\033[0m~n"
  96                     (parameterize ([date-display-format 'iso-8601])
  97                                   (date->string (seconds->date [msg-ts-epoch msg]) #t))
  98                     nick color text)]
  99            ['multi-line
 100             (printf "~a (~a)~n\033[1;37m<~a ~a>\033[0m~n\033[0;~am~a\033[0m~n~n"
 101                     (parameterize ([date-display-format 'rfc2822])
 102                                   (date->string (seconds->date [msg-ts-epoch msg]) #t))
 103                     (msg-ts-orig msg)
 104                     nick uri color text)])))))
 105
 106 (: rfc3339->epoch (-> String (Option Nonnegative-Integer)))
 107 (define rfc3339->epoch
 108   (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}))?$")])
 109     (λ (ts)
 110        (match (regexp-match re ts)
 111          [(list _wholething yyyy mm dd HH MM _:SS SS _fractional tz-whole tz-sign tz-HH tz-MM)
 112           (let*
 113             ([tz-offset
 114                (match* (tz-whole tz-sign tz-HH tz-MM)
 115                  [("Z" #f #f #f)
 116                   0]
 117                  [(_  (or "-" "+") (? identity h)  (? identity m))
 118                   (let ([h (string->number h)]
 119                         [m (string->number m)]
 120                         ; Reverse to get back to UTC:
 121                         [op (match tz-sign ["+" -] ["-" +])])
 122                     (op 0 (+ (* 60 m) (* 60 (* 60 h)))))]
 123                  [(a b c d)
 124                   (log-warning "Impossible TZ string: ~v, components: ~v ~v ~v ~v" tz-whole a b c d)
 125                   0])]
 126              [ts-orig ts]
 127              [local-time? #f]
 128              [ts-epoch (find-seconds (if SS (string->number SS) 0)
 129                                      (string->number MM)
 130                                      (string->number HH)
 131                                      (string->number dd)
 132                                      (string->number mm)
 133                                      (string->number yyyy)
 134                                      local-time?)])
 135             (+ ts-epoch tz-offset))]
 136          [_
 137            (log-error "Invalid timestamp: ~v" ts)
 138            #f]))))
 139
 140 (: str->msg (-> String Url String (Option Msg)))
 141 (define str->msg
 142   (let ([re (pregexp "^([^\\s\t]+)[\\s\t]+(.*)$")])
 143     (λ (nick uri str)
 144        (with-handlers*
 145          ([exn:fail?
 146             (λ (e)
 147                (log-error
 148                  "Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v"
 149                  str nick (url->string uri) e)
 150                #f)])
 151          (match (regexp-match re str)
 152            [(list _wholething ts-orig text)
 153             (let ([ts-epoch (rfc3339->epoch ts-orig)])
 154               (if ts-epoch
 155                   (msg ts-epoch ts-orig nick uri text)
 156                   (begin
 157                     (log-error
 158                       "Msg rejected due to invalid timestamp: ~v, nick:~v, uri:~v"
 159                       str nick (url->string uri))
 160                     #f)))]
 161            [_
 162              (log-debug "Non-msg line from nick:~a, line:~a" nick str)
 163              #f])))))
 164
 165 (module+ test
 166   (let* ([tzs (for*/list ([d '("-" "+")]
 167                           [h '("5" "05")]
 168                           [m '("00" ":00" "57" ":57")])
 169                          (string-append d h m))]
 170          [tzs (list* "" "Z" tzs)])
 171     (for* ([n   '("fake-nick")]
 172            [u   '("fake-uri")]
 173            [s   '("" ":10")]
 174            [f   '("" ".1337")]
 175            [z   tzs]
 176            [sep (list "\t" " ")]
 177            [txt '("foo bar baz" "'jaz poop bear giraffe / tea" "@*\"``")])
 178           (let* ([ts (string-append "2020-11-18T22:22"
 179                                     (if (non-empty-string? s) s ":00")
 180                                     z)]
 181                  [m  (str->msg n u (string-append ts sep txt))])
 182             (check-not-false m)
 183             (check-equal? (msg-nick m) n)
 184             (check-equal? (msg-uri m) u)
 185             (check-equal? (msg-text m) txt)
 186             (check-equal? (msg-ts-orig m) ts (format "Given: ~v" ts))
 187             )))
 188
 189   (let* ([ts       "2020-11-18T22:22:09-0500"]
 190          [tab      "    "]
 191          [text     "Lorem ipsum"]
 192          [nick     "foo"]
 193          [uri      "bar"]
 194          [actual   (str->msg nick uri (string-append ts tab text))]
 195          [expected (msg 1605756129 ts nick uri text)])
 196     (check-equal?
 197       (msg-ts-epoch actual)
 198       (msg-ts-epoch expected)
 199       "str->msg ts-epoch")
 200     (check-equal?
 201       (msg-ts-orig actual)
 202       (msg-ts-orig expected)
 203       "str->msg ts-orig")
 204     (check-equal?
 205       (msg-nick actual)
 206       (msg-nick expected)
 207       "str->msg nick")
 208     (check-equal?
 209       (msg-uri actual)
 210       (msg-uri expected)
 211       "str->msg uri")
 212     (check-equal?
 213       (msg-text actual)
 214       (msg-text expected)
 215       "str->msg text")))
 216
 217 (: str->lines (-> String (Listof String)))
 218 (define (str->lines str)
 219   (string-split str (regexp "[\r\n]+")))
 220
 221 (module+ test
 222   (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi")))
 223
 224 (: str->msgs (-> String Url String (Listof Msg)))
 225 (define (str->msgs nick uri str)
 226   (filter-map (λ (line) (str->msg nick uri line)) (filter-comments (str->lines str))))
 227
 228 (: hash-sha1 (-> String String))
 229 (define (hash-sha1 str)
 230   (define in (open-input-string str))
 231   (define digest (sha1 in))
 232   (close-input-port in)
 233   digest)
 234
 235 (: url->cache-file-path (-> Url Path-String))
 236 (define (url->cache-file-path uri)
 237   ; TODO Replace hashing with encoding
 238   (expand-user-path (string-append "~/.tt/cache/" (hash-sha1 (url->string uri)))))
 239
 240 ; TODO Return Option
 241 (: uri-read-cached (-> Url String))
 242 (define (uri-read-cached uri)
 243   (define path (url->cache-file-path uri))
 244   (if (file-exists? path)
 245       (file->string path)
 246       (begin
 247         (log-warning "Cache file not found for URI: ~a" (url->string uri))
 248         "")))
 249
 250 (: str->feed (String (Option Feed)))
 251 (define (str->feed str)
 252   (log-debug "Parsing feed string: ~v" str)
 253   (match (string-split str)
 254     [(list nick u)
 255      (with-handlers*
 256        ([exn:fail?
 257           (λ (e)
 258              (log-error "Invalid URI: ~v, exn: ~v" u e)
 259              #f)])
 260        (feed nick (string->url u)))]
 261     [_
 262       (log-error "Invalid feed string: ~v" str)
 263       #f]))
 264
 265 (: filter-comments (-> (Listof String) (Listof String)))
 266 (define (filter-comments lines)
 267   (filter-not (λ (line) (string-prefix? line "#")) lines))
 268
 269 (: str->feeds (-> String (Listof Feed)))
 270 (define (str->feeds str)
 271   (filter-map str->feed (filter-comments (str->lines str))))
 272
 273 (: file->feeds (-> Path-String (Listof Feed)))
 274 (define (file->feeds filename)
 275   (str->feeds (file->string filename)))
 276
 277 (: user-agent String)
 278 (define user-agent
 279   (let*
 280     ([prog-name      "tt"]
 281      [prog-version   (info:#%info-lookup 'version)]
 282      [prog-uri       "https://github.com/xandkar/tt"]
 283      [user-feed-file (expand-user-path "~/twtxt-me.txt")]
 284      [user
 285        (if (file-exists? user-feed-file)
 286            (let ([user (first (file->feeds user-feed-file))])
 287              (format "+~a; @~a" (url->string (feed-uri user)) (feed-nick user)))
 288            (format "+~a" prog-uri))])
 289     (format "~a/~a (~a)" prog-name prog-version user)))
 290
 291 (: uri-download (-> Url Void))
 292 (define (uri-download u)
 293   (define cache-file-path (url->cache-file-path u))
 294   (log-debug "uri-download ~v into ~v" u cache-file-path)
 295   (match* ((url-scheme u) (url-host u) (url-port u))
 296     [(s h p)
 297      #:when (and s h)
 298      (define ssl? (string=? s "https"))
 299      (define-values (status-line headers body-input)
 300        ; TODO Timeout. Currently hangs on slow connections.
 301        (http-sendrecv
 302          h
 303          (url->string (struct-copy url u [scheme #f] [host #f]))
 304          #:ssl? ssl?
 305          #:port (cond [p p] [ssl? 443] [else 80])
 306          #:headers (list (format "User-Agent: ~a" user-agent))
 307          ))
 308      (log-debug "headers: ~v" headers)
 309      (log-debug "status-line: ~v" status-line)
 310      (define status
 311        (string->number (second (string-split (bytes->string/utf-8 status-line)))))
 312      (log-debug "status: ~v" status)
 313      ; TODO Handle redirects
 314      (if (= 200 status)
 315          (call-with-output-file cache-file-path
 316                                 (λ (cache-output)
 317                                    (copy-port body-input cache-output))
 318                                 #:exists 'replace)
 319          (raise status))]
 320     [(_ _ _)
 321      (log-error "Invalid URI: ~v" u)]))
 322
 323 (: timeline-print (-> Out-Format (Listof Msg) Void))
 324 (define (timeline-print out-format timeline)
 325   (void (foldl (match-lambda**
 326                  [((and m (msg _ _ nick _ _)) (cons prev-nick i))
 327                   (let ([i (if (string=? prev-nick nick) i (+ 1 i))])
 328                     (msg-print out-format i m)
 329                     (cons nick i))])
 330                (cons "" 0)
 331                timeline)))
 332
 333 (: feed->msgs (-> Feed (Listof Msg)))
 334 (define (feed->msgs f)
 335   (match-define (feed nick uri) f)
 336   (log-info "Reading feed nick:~a uri:~v" nick (url->string uri))
 337   (str->msgs nick uri (uri-read-cached uri)))
 338
 339 (: feed-download (-> Feed Void))
 340 (define (feed-download f)
 341   (match-define (feed nick uri) f)
 342   (log-info "Downloading feed nick:~a uri:~a" nick (url->string uri))
 343   (with-handlers
 344     ([exn:fail?
 345        (λ (e)
 346           (log-error "Network error nick:~a uri:~v  exn:~v" nick uri e)
 347           #f)]
 348      [integer?
 349        (λ (status)
 350           (log-error "HTTP error nick:~a uri:~a  status:~a" nick uri status)
 351           #f)])
 352     (uri-download uri)))
 353
 354 (: timeline-download (-> Integer (Listof Feed) Void))
 355 (define (timeline-download num-workers feeds)
 356   ; TODO No need for map - can just iter
 357   (void (concurrent-filter-map num-workers feed-download feeds)))
 358
 359 ; TODO timeline contract : time-sorted list of messages
 360 (: timeline-read (-> Timeline-Order (Listof Feed) (Listof Msg)))
 361 (define (timeline-read order feeds)
 362   (define cmp (match order
 363                 ['old->new <]
 364                 ['new->old >]))
 365   (sort (append* (filter-map feed->msgs feeds))
 366         (λ (a b) (cmp (msg-ts-epoch a) (msg-ts-epoch b)))))
 367
 368 (: start-logger (-> Log-Level Void))
 369 (define (start-logger level)
 370   (let* ([logger       (make-logger #f #f level #f)]
 371          [log-receiver (make-log-receiver logger level)])
 372     (void (thread (λ ()
 373                      (parameterize
 374                        ([date-display-format 'iso-8601])
 375                        (let loop ()
 376                          (define data  (sync log-receiver))
 377                          (define level (vector-ref data 0))
 378                          (define msg   (vector-ref data 1))
 379                          (define ts    (date->string (current-date) #t))
 380                          (eprintf "~a [~a] ~a~n" ts level msg)
 381                          (loop))))))
 382     (current-logger logger)))
 383
 384 (module+ main
 385   (let ([log-level 'info])
 386     (command-line
 387       #:program
 388       "tt"
 389       #:once-each
 390       [("-d" "--debug")
 391        "Enable debug log level."
 392        (set! log-level 'debug)]
 393       #:help-labels
 394       ""
 395       "and <command> is one of"
 396       "r, read i   : Read the timeline."
 397       "d, download : Download the timeline."
 398       "u, upload   : Upload your twtxt file (alias to execute ~/.tt/upload)."
 399       ""
 400       #:args (command . args)
 401       (start-logger log-level)
 402       (current-command-line-arguments (list->vector args))
 403       (match command
 404         [(or "d" "download")
 405          (let ([num-workers 15]) ; 15 was fastest out of the tried: 1, 5, 10, 20.
 406            (command-line
 407              #:program
 408              "tt download"
 409              #:once-each
 410              [("-j" "--jobs")
 411               njobs "Number of concurrent jobs."
 412               (set! num-workers (string->number njobs))]
 413              #:args (filename)
 414              (timeline-download num-workers (file->feeds filename))))]
 415         [(or "u" "upload")
 416          (command-line
 417            #:program
 418            "tt upload"
 419            #:args ()
 420            (if (system (path->string (expand-user-path "~/.tt/upload")))
 421                (exit 0)
 422                (exit 1)))]
 423         [(or "r" "read")
 424          (let ([out-format 'multi-line]
 425                [order      'old->new])
 426            (command-line
 427              #:program
 428              "tt read"
 429              #:once-each
 430              [("-r" "--rev")
 431               "Reverse displayed timeline order."
 432               (set! order 'new->old)]
 433              #:once-any
 434              [("-s" "--short")
 435               "Short output format"
 436               (set! out-format 'single-line)]
 437              [("-l" "--long")
 438               "Long output format"
 439               (set! out-format 'multi-line)]
 440              #:args (filename)
 441              (timeline-print out-format (timeline-read order (file->feeds filename)))))]
 442         ))))