c63f35d3bb5068c958a133777ffe7b113bf931b0
[tt.git] / 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 ; Looks like a better CLI parser than "racket/cmdline":
39 ; https://docs.racket-lang.org/natural-cli/
40
41 #lang racket
42
43 (require openssl/sha1)
44 (require racket/date)
45
46 (require http-client)
47 (require rfc3339-old)
48
49 (module+ test
50 (require rackunit))
51
52 (struct msg (ts_epoch ts_rfc3339 nick uri text))
53 (struct feed (nick uri))
54
55 (define (concurrent-filter-map num_workers f xs)
56 ; TODO preserve order of elements OR communicate that reorder is expected
57 ; TODO switch from mailboxes to channels
58 (define (make-worker id f)
59 (define parent (current-thread))
60 (λ ()
61 (define self (current-thread))
62 (define (work)
63 (thread-send parent (cons 'next self))
64 (match (thread-receive)
65 ['done (thread-send parent (cons 'exit id))]
66 [(cons 'unit x) (begin
67 (define y (f x))
68 (when y (thread-send parent (cons 'result y)))
69 (work))]))
70 (work)))
71 (define (dispatch ws xs ys)
72 (if (empty? ws)
73 ys
74 (match (thread-receive)
75 [(cons 'exit w) (dispatch (remove w ws =) xs ys)]
76 [(cons 'result y) (dispatch ws xs (cons y ys))]
77 [(cons 'next thd) (match xs
78 ['() (begin
79 (thread-send thd 'done)
80 (dispatch ws xs ys))]
81 [(cons x xs) (begin
82 (thread-send thd (cons 'unit x))
83 (dispatch ws xs ys))])])))
84 (define workers (range num_workers))
85 (define threads (map (λ (id) (thread (make-worker id f))) workers))
86 (define results (dispatch workers xs '()))
87 (for-each thread-wait threads)
88 results)
89
90 (module+ test
91 (define n-workers 10)
92 (define given (list
93 (λ (x) (if (even? x) x #f))
94 (range 11)))
95 (check-equal?
96 (sort (apply concurrent-filter-map (cons n-workers given)) <)
97 (sort (apply filter-map given ) <)))
98
99 (define (msg-print out-format odd msg)
100 (printf
101 (match out-format
102 ['single-line "~a \033[1;37m<~a ~a>\033[0m \033[0;~am~a\033[0m~n"]
103 ['multi-line "~a~n\033[1;37m<~a ~a>\033[0m~n\033[0;~am~a\033[0m~n~n"]
104 [_ (raise (format "Invalid output format: ~a" out-format))])
105 (date->string (seconds->date [msg-ts_epoch msg]) #t)
106 (msg-nick msg)
107 (msg-uri msg)
108 (if odd 36 33)
109 (msg-text msg)))
110
111 (define re-msg-begin
112 ; TODO Zulu offset. Maybe in several formats. Which ones?
113 (pregexp "^[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2}"))
114
115 (define (str->msg nick uri str)
116 (if (not (regexp-match? re-msg-begin str))
117 (begin
118 (log-debug "Non-msg line from nick:~a, line:~a" nick str)
119 #f)
120 (let ([toks (string-split str (regexp "\t+"))])
121 (if (not (= 2 (length toks)))
122 (begin
123 (log-warning "Invalid msg line from nick:~a, msg:~a" nick str)
124 #f)
125 (let*
126 ([ts_rfc3339 (first toks)]
127 [text (second toks)]
128 [t (string->rfc3339-record ts_rfc3339)]
129 ; TODO handle tz offset
130 [ts_epoch (find-seconds [rfc3339-record:second t]
131 [rfc3339-record:minute t]
132 [rfc3339-record:hour t]
133 [rfc3339-record:mday t]
134 [rfc3339-record:month t]
135 [rfc3339-record:year t])])
136 (msg ts_epoch ts_rfc3339 nick uri text))))))
137
138 (define (str->lines str)
139 (string-split str (regexp "[\r\n]+")))
140
141 (define (str->msgs nick uri str)
142 (filter-map (λ (line) (str->msg nick uri line)) (str->lines str)))
143
144 (define (hash-sha1 str)
145 (define in (open-input-string str))
146 (define digest (sha1 in))
147 (close-input-port in)
148 digest)
149
150 (define (uri-fetch uri)
151 (log-info "GET ~a" uri)
152 (define resp (http-get uri))
153 (define status (http-response-code resp))
154 (define body (http-response-body resp))
155 (log-debug "finished GET ~a status:~a body length:~a"
156 uri status (string-length body))
157 ; TODO Handle redirects
158 (if (= status 200)
159 (let*
160 ([url-digest
161 (hash-sha1 uri)]
162 [cache-file-path
163 (expand-user-path (string-append "~/.tt/cache/" url-digest))])
164 (display-to-file
165 body cache-file-path
166 #:exists 'replace)
167 body)
168 ; TODO A more-informative exception
169 (raise status)))
170
171 (define (timeline-print out-format timeline)
172 (for ([msg timeline]
173 [i (in-naturals)])
174 (msg-print out-format (odd? i) msg)))
175
176 (define (feed->msgs feed)
177 (log-info "downloading feed nick:~a uri:~a"
178 (feed-nick feed)
179 (feed-uri feed))
180 (with-handlers
181 ([exn:fail:network?
182 (λ (e)
183 (log-error "network error nick:~a uri:~a exn:~a"
184 (feed-nick feed)
185 (feed-uri feed)
186 e)
187 #f)]
188 [integer?
189 (λ (status)
190 (log-error "http error nick:~a uri:~a status:~a"
191 (feed-nick feed)
192 (feed-uri feed)
193 status)
194 #f)])
195 (define uri (feed-uri feed))
196 (str->msgs [feed-nick feed] uri [uri-fetch uri])))
197
198 ; TODO timeline contract : time-sorted list of messages
199 (define (timeline num_workers feeds)
200 (sort (append* (concurrent-filter-map num_workers feed->msgs feeds))
201 (λ (a b) [< (msg-ts_epoch a) (msg-ts_epoch b)])))
202
203 (define (str->feed str)
204 ; TODO validation
205 (define toks (string-split str))
206 (apply feed toks))
207
208 (define (str->feeds str)
209 (map str->feed (str->lines str)))
210
211 (define (file->feeds filename)
212 (str->feeds (file->string filename)))
213
214 (define (we-are-twtxt)
215 (define uri
216 "https://raw.githubusercontent.com/mdom/we-are-twtxt/master/we-are-twtxt.txt")
217 (str->feeds (uri-fetch uri)))
218
219 (define (user-agent prog-name prog-version)
220 (let*
221 ([prog-uri "https://github.com/xandkar/tt"]
222 [user-feed-file (expand-user-path "~/twtxt-me.txt")]
223 [user
224 (if (file-exists? user-feed-file)
225 (let ([user (first (file->feeds user-feed-file))])
226 (format "+~a; @~a" (feed-uri user) (feed-nick user)))
227 (format "+~a" prog-uri))]
228 )
229 (format "~a/~a (~a)" prog-name prog-version user)))
230
231 (module+ main
232 (require setup/getinfo)
233 (define (setup-logging)
234 (define logger (make-logger #f #f 'debug #f))
235 (define log-chan (make-log-receiver logger 'debug))
236 (void (thread (λ ()
237 [date-display-format 'iso-8601]
238 [let loop ()
239 (define data (sync log-chan))
240 (define level (vector-ref data 0))
241 (define msg (vector-ref data 1))
242 (define ts (date->string (current-date) #t))
243 (eprintf "~a [~a] ~a~n" ts level msg)
244 (loop)])))
245 (current-logger logger))
246
247 (setup-logging)
248 (current-http-response-auto #f)
249 (let* ([prog-name "tt"]
250 [prog-version ((get-info (list prog-name)) 'version)]
251 [user-agent (user-agent prog-name prog-version)])
252 (current-http-user-agent user-agent))
253 (date-display-format 'rfc2822)
254
255 (define args (current-command-line-arguments))
256 (define feeds
257 (if (= 0 (vector-length args))
258 (we-are-twtxt)
259 (file->feeds (vector-ref args 0))))
260 (define out-format 'multi-line)
261 (define num_workers 15) ; 15 was fastest out of the tried 1, 5, 10, 15 and 20.
262 (timeline-print out-format (timeline num_workers feeds)))
This page took 0.112381 seconds and 4 git commands to generate.