Commit | Line | Data |
---|---|---|
e5c3ae92 | 1 | ; TODO optional text wrap |
d6a187b7 SK |
2 | ; TODO write |
3 | ; TODO caching (use cache by default, unless explicitly asked for update) | |
1d753430 | 4 | ; - [x] value --> cache |
0e16a46c | 5 | ; - [x] value <-- cache |
1d753430 | 6 | ; requires: commands |
d6a187b7 | 7 | ; TODO timeline limits |
be96956c | 8 | ; TODO feed set operations (perhaps better done externally?) |
d6a187b7 SK |
9 | ; TODO timeline as a result of a query (feed set op + filter expressions) |
10 | ; TODO named timelines | |
d6a187b7 | 11 | ; TODO config files |
895a32cf SK |
12 | ; TODO parse "following" from feed |
13 | ; - following = <nick> <uri> | |
93ebe03e SK |
14 | ; TODO parse mentions: |
15 | ; - @<source.nick source.url> | @<source.url> | |
d6a187b7 SK |
16 | ; TODO highlight mentions |
17 | ; TODO filter on mentions | |
18 | ; TODO highlight hashtags | |
19 | ; TODO filter on hashtags | |
e0e0fb20 | 20 | ; TODO hashtags as channels? initial hashtag special? |
d6a187b7 | 21 | ; TODO query language |
e0e0fb20 SK |
22 | ; TODO console logger colors by level ('error) |
23 | ; TODO file logger ('debug) | |
dbccadf9 SK |
24 | ; TODO commands: |
25 | ; - r | read | |
26 | ; - see timeline ops above | |
27 | ; - w | write | |
93ebe03e | 28 | ; - arg or stdin |
dbccadf9 SK |
29 | ; - nick expand to URI |
30 | ; - q | query | |
31 | ; - see timeline ops above | |
32 | ; - see hashtag and channels above | |
93ebe03e SK |
33 | ; - d | download |
34 | ; - u | upload | |
35 | ; - calls user-configured command to upload user's own feed file to their server | |
812f0b04 SK |
36 | ; |
37 | ; Looks like a better CLI parser than "racket/cmdline": | |
38 | ; https://docs.racket-lang.org/natural-cli/ | |
d6a187b7 | 39 | |
4764ff89 SK |
40 | #lang racket |
41 | ||
1d753430 | 42 | (require openssl/sha1) |
4764ff89 SK |
43 | (require racket/date) |
44 | ||
45 | (require http-client) | |
46 | (require rfc3339-old) | |
47 | ||
78398948 | 48 | (module+ test |
de3ff448 | 49 | (require rackunit)) |
78398948 | 50 | |
87d0eb92 | 51 | (struct msg (ts_epoch ts_rfc3339 nick uri text)) |
4764ff89 SK |
52 | (struct feed (nick uri)) |
53 | ||
895a32cf | 54 | (define (concurrent-filter-map num_workers f xs) |
dad4504d | 55 | ; TODO preserve order of elements OR communicate that reorder is expected |
a239a233 | 56 | ; TODO switch from mailboxes to channels |
895a32cf SK |
57 | (define (make-worker id f) |
58 | (define parent (current-thread)) | |
59 | (λ () | |
60 | (define self (current-thread)) | |
61 | (define (work) | |
62 | (thread-send parent (cons 'next self)) | |
63 | (match (thread-receive) | |
c562bea3 SK |
64 | ['done (thread-send parent (cons 'exit id))] |
65 | [(cons 'unit x) (begin | |
66 | (define y (f x)) | |
67 | (when y (thread-send parent (cons 'result y))) | |
68 | (work))])) | |
895a32cf SK |
69 | (work))) |
70 | (define (dispatch ws xs ys) | |
71 | (if (empty? ws) | |
f1493e49 SK |
72 | ys |
73 | (match (thread-receive) | |
c562bea3 SK |
74 | [(cons 'exit w) (dispatch (remove w ws =) xs ys)] |
75 | [(cons 'result y) (dispatch ws xs (cons y ys))] | |
76 | [(cons 'next thd) (match xs | |
77 | ['() (begin | |
78 | (thread-send thd 'done) | |
79 | (dispatch ws xs ys))] | |
80 | [(cons x xs) (begin | |
81 | (thread-send thd (cons 'unit x)) | |
82 | (dispatch ws xs ys))])]))) | |
9926c9a9 SK |
83 | (define workers (range num_workers)) |
84 | (define threads (map (λ (id) (thread (make-worker id f))) workers)) | |
85 | (define results (dispatch workers xs '())) | |
895a32cf SK |
86 | (for-each thread-wait threads) |
87 | results) | |
88 | ||
dad4504d | 89 | (module+ test |
de3ff448 SK |
90 | (let* ([f (λ (x) (if (even? x) x #f))] |
91 | [xs (range 11)] | |
92 | [actual (sort (concurrent-filter-map 10 f xs) <)] | |
93 | [expected (sort ( filter-map f xs) <)]) | |
c562bea3 | 94 | (check-equal? actual expected "concurrent-filter-map"))) |
dad4504d | 95 | |
3d042e75 SK |
96 | (define msg-print |
97 | (let* ([colors (vector 36 33)] | |
98 | [n (vector-length colors)]) | |
99 | (λ (out-format color-i msg) | |
01e4c499 SK |
100 | (let ([color (vector-ref colors (modulo color-i n))] |
101 | [nick (msg-nick msg)] | |
102 | [uri (msg-uri msg)] | |
103 | [text (msg-text msg)]) | |
3d042e75 | 104 | (match out-format |
01e4c499 SK |
105 | ['single-line |
106 | (printf "~a \033[1;37m<~a>\033[0m \033[0;~am~a\033[0m~n" | |
107 | (parameterize ([date-display-format 'iso-8601]) | |
108 | (date->string (seconds->date [msg-ts_epoch msg]) #t)) | |
109 | nick color text)] | |
110 | ['multi-line | |
111 | (printf "~a~n\033[1;37m<~a ~a>\033[0m~n\033[0;~am~a\033[0m~n~n" | |
112 | (parameterize ([date-display-format 'rfc2822]) | |
113 | (date->string (seconds->date [msg-ts_epoch msg]) #t)) | |
114 | nick uri color text)]))))) | |
e96264cc SK |
115 | |
116 | (define re-msg-begin | |
117 | ; TODO Zulu offset. Maybe in several formats. Which ones? | |
118 | (pregexp "^[0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}:[0-9]{2}")) | |
88d50b3e | 119 | |
b201e854 | 120 | (define (str->msg nick uri str) |
88d50b3e | 121 | (if (not (regexp-match? re-msg-begin str)) |
f1493e49 SK |
122 | (begin |
123 | (log-debug "Non-msg line from nick:~a, line:~a" nick str) | |
124 | #f) | |
125 | (let ([toks (string-split str (regexp "\t+"))]) | |
c562bea3 SK |
126 | (if (not (= 2 (length toks))) |
127 | (begin | |
128 | (log-warning "Invalid msg line from nick:~a, msg:~a" nick str) | |
129 | #f) | |
130 | (let* | |
131 | ([ts_rfc3339 (first toks)] | |
132 | [text (second toks)] | |
133 | [t (string->rfc3339-record ts_rfc3339)] | |
134 | ; TODO handle tz offset | |
135 | [ts_epoch (find-seconds [rfc3339-record:second t] | |
136 | [rfc3339-record:minute t] | |
137 | [rfc3339-record:hour t] | |
138 | [rfc3339-record:mday t] | |
139 | [rfc3339-record:month t] | |
140 | [rfc3339-record:year t])]) | |
141 | (msg ts_epoch ts_rfc3339 nick uri text)))))) | |
88d50b3e | 142 | |
63afa259 | 143 | (module+ test |
de3ff448 SK |
144 | (let* ([ts "2020-11-18T22:22:09-0500"] |
145 | [tab " "] | |
146 | [text "Lorem ipsum"] | |
147 | [nick "foo"] | |
148 | [uri "bar"] | |
149 | [actual (str->msg nick uri (string-append ts tab text))] | |
150 | [expected (msg 1605756129 ts nick uri text)]) | |
c562bea3 SK |
151 | ; FIXME re-enable after handling tz offset |
152 | ;(check-equal? | |
153 | ; (msg-ts_epoch actual) | |
154 | ; (msg-ts_epoch expected) | |
155 | ; "str->msg ts_epoch") | |
156 | (check-equal? | |
157 | (msg-ts_rfc3339 actual) | |
158 | (msg-ts_rfc3339 expected) | |
159 | "str->msg ts_rfc3339") | |
160 | (check-equal? | |
161 | (msg-nick actual) | |
162 | (msg-nick expected) | |
163 | "str->msg nick") | |
164 | (check-equal? | |
165 | (msg-uri actual) | |
166 | (msg-uri expected) | |
167 | "str->msg uri") | |
168 | (check-equal? | |
169 | (msg-text actual) | |
170 | (msg-text expected) | |
171 | "str->msg text"))) | |
63afa259 | 172 | |
e96264cc SK |
173 | (define (str->lines str) |
174 | (string-split str (regexp "[\r\n]+"))) | |
175 | ||
63afa259 | 176 | (module+ test |
de3ff448 | 177 | (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi"))) |
63afa259 | 178 | |
b201e854 SK |
179 | (define (str->msgs nick uri str) |
180 | (filter-map (λ (line) (str->msg nick uri line)) (str->lines str))) | |
4764ff89 | 181 | |
1d753430 SK |
182 | (define (hash-sha1 str) |
183 | (define in (open-input-string str)) | |
184 | (define digest (sha1 in)) | |
185 | (close-input-port in) | |
186 | digest) | |
187 | ||
0e16a46c SK |
188 | (define (uri-fetch use-cache uri) |
189 | (define cache-file-path | |
190 | (expand-user-path (string-append "~/.tt/cache/" (hash-sha1 uri)))) | |
191 | (if (and use-cache (file-exists? cache-file-path)) | |
192 | (begin | |
193 | (log-info "uri-fetch cached ~a" uri) | |
194 | (file->string cache-file-path)) | |
195 | (begin | |
196 | (log-info "uri-fetch new ~a" uri) | |
01e4c499 | 197 | ; TODO Timeout. Currently hangs on slow connections. |
0e16a46c SK |
198 | (let* ([resp (http-get uri)] |
199 | [status (http-response-code resp)] | |
200 | [body (http-response-body resp)]) | |
c562bea3 SK |
201 | (log-debug "finished GET ~a status:~a body length:~a" |
202 | uri status (string-length body)) | |
203 | ; TODO Handle redirects | |
204 | (if (= status 200) | |
205 | (begin | |
206 | (display-to-file body cache-file-path #:exists 'replace) | |
207 | body) | |
208 | ; TODO A more-informative exception | |
209 | (raise status)))))) | |
4764ff89 | 210 | |
b201e854 | 211 | (define (timeline-print out-format timeline) |
3d042e75 SK |
212 | (void (foldl (match-lambda** |
213 | [((and m (msg _ _ nick _ _)) (cons prev-nick i)) | |
214 | (let ([i (if (string=? prev-nick nick) i (+ 1 i))]) | |
215 | (msg-print out-format i m) | |
216 | (cons nick i))]) | |
217 | (cons "" 0) | |
218 | timeline))) | |
4764ff89 | 219 | |
0e16a46c | 220 | (define (feed->msgs use-cache feed) |
9a6a9f9a SK |
221 | (log-info "downloading feed nick:~a uri:~a" |
222 | (feed-nick feed) | |
223 | (feed-uri feed)) | |
224 | (with-handlers | |
225 | ([exn:fail:network? | |
226 | (λ (e) | |
227 | (log-error "network error nick:~a uri:~a exn:~a" | |
228 | (feed-nick feed) | |
229 | (feed-uri feed) | |
230 | e) | |
231 | #f)] | |
232 | [integer? | |
233 | (λ (status) | |
234 | (log-error "http error nick:~a uri:~a status:~a" | |
235 | (feed-nick feed) | |
236 | (feed-uri feed) | |
237 | status) | |
238 | #f)]) | |
b201e854 | 239 | (define uri (feed-uri feed)) |
0e16a46c | 240 | (str->msgs [feed-nick feed] uri [uri-fetch use-cache uri]))) |
9a6a9f9a SK |
241 | |
242 | ; TODO timeline contract : time-sorted list of messages | |
0e16a46c SK |
243 | (define (timeline use-cache num_workers feeds) |
244 | (sort (append* (concurrent-filter-map num_workers (curry feed->msgs use-cache) feeds)) | |
87d0eb92 | 245 | (λ (a b) [< (msg-ts_epoch a) (msg-ts_epoch b)]))) |
4764ff89 | 246 | |
be96956c SK |
247 | (define (str->feed str) |
248 | ; TODO validation | |
249 | (define toks (string-split str)) | |
87d0eb92 | 250 | (apply feed toks)) |
be96956c SK |
251 | |
252 | (define (str->feeds str) | |
253 | (map str->feed (str->lines str))) | |
254 | ||
255 | (define (file->feeds filename) | |
256 | (str->feeds (file->string filename))) | |
257 | ||
4f3dbb24 | 258 | (define (user-agent prog-name prog-version) |
24c6a76b | 259 | (let* |
4f3dbb24 | 260 | ([prog-uri "https://github.com/xandkar/tt"] |
24c6a76b SK |
261 | [user-feed-file (expand-user-path "~/twtxt-me.txt")] |
262 | [user | |
263 | (if (file-exists? user-feed-file) | |
f1493e49 | 264 | (let ([user (first (file->feeds user-feed-file))]) |
c562bea3 | 265 | (format "+~a; @~a" (feed-uri user) (feed-nick user))) |
f1493e49 | 266 | (format "+~a" prog-uri))] |
24c6a76b SK |
267 | ) |
268 | (format "~a/~a (~a)" prog-name prog-version user))) | |
269 | ||
01e4c499 SK |
270 | (define (start-logger level) |
271 | (let* ([logger (make-logger #f #f level #f)] | |
272 | [log-receiver (make-log-receiver logger level)]) | |
273 | (void (thread (λ () | |
274 | (parameterize | |
275 | ([date-display-format 'iso-8601]) | |
276 | (let loop () | |
277 | (define data (sync log-receiver)) | |
278 | (define level (vector-ref data 0)) | |
279 | (define msg (vector-ref data 1)) | |
280 | (define ts (date->string (current-date) #t)) | |
281 | (eprintf "~a [~a] ~a~n" ts level msg) | |
282 | (loop)))))) | |
283 | (current-logger logger))) | |
284 | ||
24c6a76b | 285 | (module+ main |
de3ff448 | 286 | (require setup/getinfo) |
26d172a9 | 287 | |
de3ff448 SK |
288 | (current-http-response-auto #f) |
289 | (let* ([prog-name "tt"] | |
290 | [prog-version ((get-info (list prog-name)) 'version)] | |
291 | [user-agent (user-agent prog-name prog-version)]) | |
c562bea3 | 292 | (current-http-user-agent user-agent)) |
de3ff448 SK |
293 | (let* ([use-cache |
294 | #f] | |
01e4c499 SK |
295 | [log-level |
296 | 'info] | |
de3ff448 SK |
297 | [out-format |
298 | 'multi-line] | |
299 | [num_workers | |
300 | 15]) ; 15 was fastest out of the tried 1, 5, 10, 15 and 20. | |
c562bea3 SK |
301 | (command-line |
302 | #:once-each | |
303 | [("-c" "--cached") | |
304 | "Read cached data instead of downloading." | |
305 | (set! use-cache #t)] | |
0e16a46c | 306 | |
01e4c499 SK |
307 | [("-d" "--debug") |
308 | "Enable debug log level." | |
309 | (set! log-level 'debug)] | |
310 | ||
c562bea3 SK |
311 | [("-j" "--jobs") |
312 | njobs "Number of concurrent jobs." | |
313 | (set! num_workers (string->number njobs))] | |
0e16a46c | 314 | |
01e4c499 SK |
315 | #:once-any |
316 | [("-s" "--short") | |
317 | "Short output format" | |
318 | (set! out-format 'single-line)] | |
0e16a46c | 319 | |
01e4c499 SK |
320 | [("-l" "--long") |
321 | "Long output format" | |
322 | (set! out-format 'multi-line)] | |
323 | ||
324 | #:args (filename) | |
325 | (start-logger log-level) | |
c562bea3 SK |
326 | (timeline-print out-format |
327 | (timeline use-cache | |
328 | num_workers | |
329 | (file->feeds filename)))))) |