tt.rkt

   1 ; TODO optional text wrap
   2 ; TODO write
   3 ; TODO caching (use cache by default, unless explicitly asked for update)
   4 ; - [x] value --> cache
   5 ; - [ ] value <-- cache
   6 ;   requires: commands
   7 ; TODO timeline limits
   8 ; TODO feed set operations (perhaps better done externally?)
   9 ; TODO timeline as a result of a query (feed set op + filter expressions)
  10 ; TODO named timelines
  11 ; TODO CLI params
  12 ; TODO config files
  13 ; TODO parse "following" from feed
  14 ; - following = <nick> <uri>
  15 ; TODO parse mentions:
  16 ; - @<source.nick source.url> | @<source.url>
  17 ; TODO highlight mentions
  18 ; TODO filter on mentions
  19 ; TODO highlight hashtags
  20 ; TODO filter on hashtags
  21 ; TODO hashtags as channels? initial hashtag special?
  22 ; TODO query language
  23 ; TODO console logger colors by level ('error)
  24 ; TODO file logger ('debug)
  25 ; TODO commands:
  26 ; - r | read
  27 ;   - see timeline ops above
  28 ; - w | write
  29 ;   - arg or stdin
  30 ;   - nick expand to URI
  31 ; - q | query
  32 ;   - see timeline ops above
  33 ;   - see hashtag and channels above
  34 ; - d | download
  35 ; - u | upload
  36 ;   - calls user-configured command to upload user's own feed file to their server
  37
  38 #lang racket
  39
  40 (require openssl/sha1)
  41 (require racket/date)
  42
  43 (require http-client)
  44 (require rfc3339-old)
  45
  46 (module+ test
  47   (require rackunit))
  48
  49 (struct msg  (ts_epoch ts_rfc3339 nick uri text))
  50 (struct feed (nick uri))
  51
  52 (define (concurrent-filter-map num_workers f xs)
  53   ; TODO preserve order of elements OR communicate that reorder is expected
  54   ; TODO switch from mailboxes to channels
  55   (define (make-worker id f)
  56     (define parent (current-thread))
  57     (λ ()
  58        (define self (current-thread))
  59        (define (work)
  60          (thread-send parent (cons 'next self))
  61          (match (thread-receive)
  62            ['done          (thread-send parent (cons 'exit id))]
  63            [(cons 'unit x) (begin
  64                              (define y (f x))
  65                              (when y (thread-send parent (cons 'result y)))
  66                              (work))]))
  67        (work)))
  68   (define (dispatch ws xs ys)
  69     (if (empty? ws)
  70       ys
  71       (match (thread-receive)
  72         [(cons 'exit w)   (dispatch (remove w ws =) xs ys)]
  73         [(cons 'result y) (dispatch ws xs (cons y ys))]
  74         [(cons 'next thd) (match xs
  75                             ['()         (begin
  76                                            (thread-send thd 'done)
  77                                            (dispatch ws xs ys))]
  78                             [(cons x xs) (begin
  79                                            (thread-send thd (cons 'unit x))
  80                                            (dispatch ws xs ys))])])))
  81   (define workers (range num_workers))
  82   (define threads (map (λ (id) (thread (make-worker id f))) workers))
  83   (define results (dispatch workers xs '()))
  84   (for-each thread-wait threads)
  85   results)
  86
  87 (module+ test
  88   (define n-workers 10)
  89   (define given (list
  90                   (λ (x) (if (even? x) x #f))
  91                   (range 11)))
  92   (check-equal?
  93     (sort (apply concurrent-filter-map (cons n-workers given)) <)
  94     (sort (apply            filter-map                 given ) <)))
  95
  96 (define (msg-print out-format odd msg)
  97   (printf
  98     (match out-format
  99       ['single-line "~a  \033[1;37m<~a ~a>\033[0m  \033[0;~am~a\033[0m~n"]
 100       ['multi-line  "~a~n\033[1;37m<~a ~a>\033[0m~n\033[0;~am~a\033[0m~n~n"]
 101       [_           (raise (format "Invalid output format: ~a" out-format))])
 102     (date->string (seconds->date [msg-ts_epoch msg]) #t)
 103     (msg-nick msg)
 104     (msg-uri  msg)
 105     (if odd 36 33)
 106     (msg-text msg)))
 107
 108 (define re-msg-begin
 109   ; TODO Zulu offset. Maybe in several formats. Which ones?
 110   (pregexp "^[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2}"))
 111
 112 (define (str->msg nick uri str)
 113   (if (not (regexp-match? re-msg-begin str))
 114     (begin
 115       (log-debug "Non-msg line from nick:~a, line:~a" nick str)
 116       #f)
 117     (let ([toks (string-split str (regexp "\t+"))])
 118       (if (not (= 2 (length toks)))
 119         (begin
 120           (log-warning "Invalid msg line from nick:~a, msg:~a" nick str)
 121           #f)
 122         (let*
 123           ([ts_rfc3339 (first  toks)]
 124            [text       (second toks)]
 125            [t          (string->rfc3339-record ts_rfc3339)]
 126            ; TODO handle tz offset
 127            [ts_epoch (find-seconds [rfc3339-record:second t]
 128                                    [rfc3339-record:minute t]
 129                                    [rfc3339-record:hour   t]
 130                                    [rfc3339-record:mday   t]
 131                                    [rfc3339-record:month  t]
 132                                    [rfc3339-record:year   t])])
 133           (msg ts_epoch ts_rfc3339 nick uri text))))))
 134
 135 (define (str->lines str)
 136   (string-split str (regexp "[\r\n]+")))
 137
 138 (define (str->msgs nick uri str)
 139   (filter-map (λ (line) (str->msg nick uri line)) (str->lines str)))
 140
 141 (define (hash-sha1 str)
 142   (define in (open-input-string str))
 143   (define digest (sha1 in))
 144   (close-input-port in)
 145   digest)
 146
 147 (define (uri-fetch uri)
 148   (log-info "GET ~a" uri)
 149   (define resp (http-get uri))
 150   (define status (http-response-code resp))
 151   (define body (http-response-body resp))
 152   (log-debug "finished GET ~a status:~a  body length:~a"
 153              uri status (string-length body))
 154   ; TODO Handle redirects
 155   (if (= status 200)
 156     (let*
 157       ([url-digest
 158          (hash-sha1 uri)]
 159        [cache-file-path
 160          (expand-user-path (string-append "~/.tt/cache/" url-digest))])
 161       (display-to-file
 162         body cache-file-path
 163         #:exists 'replace)
 164       body)
 165     ; TODO A more-informative exception
 166     (raise status)))
 167
 168 (define (timeline-print out-format timeline)
 169   (for ([msg timeline]
 170         [i   (in-naturals)])
 171     (msg-print out-format (odd? i) msg)))
 172
 173 (define (feed->msgs feed)
 174   (log-info "downloading feed nick:~a uri:~a"
 175             (feed-nick feed)
 176             (feed-uri feed))
 177   (with-handlers
 178     ([exn:fail:network?
 179        (λ (e)
 180           (log-error "network error nick:~a uri:~a  exn:~a"
 181                      (feed-nick feed)
 182                      (feed-uri feed)
 183                      e)
 184           #f)]
 185      [integer?
 186        (λ (status)
 187           (log-error "http error nick:~a uri:~a  status:~a"
 188                      (feed-nick feed)
 189                      (feed-uri feed)
 190                      status)
 191           #f)])
 192     (define uri (feed-uri feed))
 193     (str->msgs [feed-nick feed] uri [uri-fetch uri])))
 194
 195 ; TODO timeline contract : time-sorted list of messages
 196 (define (timeline num_workers feeds)
 197   (sort (append* (concurrent-filter-map num_workers feed->msgs feeds))
 198         (λ (a b) [< (msg-ts_epoch a) (msg-ts_epoch b)])))
 199
 200 (define (str->feed str)
 201   ; TODO validation
 202   (define toks (string-split str))
 203   (apply feed toks))
 204
 205 (define (str->feeds str)
 206   (map str->feed (str->lines str)))
 207
 208 (define (file->feeds filename)
 209   (str->feeds (file->string filename)))
 210
 211 (define (we-are-twtxt)
 212   (define uri
 213     "https://raw.githubusercontent.com/mdom/we-are-twtxt/master/we-are-twtxt.txt")
 214   (str->feeds (uri-fetch uri)))
 215
 216 (define user-agent
 217   (let*
 218     ([prog-name      "tt"]
 219      [prog-version   "0.3.4"]
 220      [prog-uri       "https://github.com/xandkar/tt"]
 221      [user-feed-file (expand-user-path "~/twtxt-me.txt")]
 222      [user
 223        (if (file-exists? user-feed-file)
 224          (let ([user (first (file->feeds user-feed-file))])
 225            (format "+~a; @~a" (feed-uri user) (feed-nick user)))
 226          (format "+~a" prog-uri))]
 227      )
 228     (format "~a/~a (~a)" prog-name prog-version user)))
 229
 230 (module+ main
 231   (define (setup-logging)
 232     (define logger (make-logger #f #f 'debug #f))
 233     (define log-chan (make-log-receiver logger 'debug))
 234     (void (thread (λ ()
 235                      [date-display-format 'iso-8601]
 236                      [let loop ()
 237                        (define data  (sync log-chan))
 238                        (define level (vector-ref data 0))
 239                        (define msg   (vector-ref data 1))
 240                        (define ts    (date->string (current-date) #t))
 241                        (eprintf "~a [~a] ~a~n" ts level msg)
 242                        (loop)])))
 243     (current-logger logger))
 244
 245   (setup-logging)
 246   (current-http-response-auto #f)
 247   (current-http-user-agent user-agent)
 248   (date-display-format 'rfc2822)
 249
 250   (define args (current-command-line-arguments))
 251   (define feeds
 252     (if (vector-empty? args)
 253       (we-are-twtxt)
 254       (file->feeds (vector-ref args 0))))
 255   (define out-format 'multi-line)
 256   (define num_workers 15) ; 15 was fastest out of the tried 1, 5, 10, 15 and 20.
 257   (timeline-print out-format (timeline num_workers feeds)))