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 | 43 | (require racket/date) |
0016b0ad | 44 | (require (prefix-in srfi/19: srfi/19)) |
4764ff89 SK |
45 | |
46 | (require http-client) | |
4764ff89 | 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 | |
b201e854 SK |
96 | (define (msg-print out-format odd msg) |
97 | (printf | |
98 | (match out-format | |
c562bea3 SK |
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))]) | |
87d0eb92 | 102 | (date->string (seconds->date [msg-ts_epoch msg]) #t) |
b201e854 SK |
103 | (msg-nick msg) |
104 | (msg-uri msg) | |
105 | (if odd 36 33) | |
106 | (msg-text msg))) | |
e96264cc SK |
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}")) | |
88d50b3e | 111 | |
0016b0ad SK |
112 | (define (rfc3339->epoch str) |
113 | (with-handlers | |
114 | ([exn? (λ (e) (log-error "Timestamp parse failure of ~v with ~v" str e))]) | |
115 | ((srfi/19:time-second | |
116 | (srfi/19:date->time-utc | |
117 | (srfi/19:string->date | |
118 | str | |
119 | "~Y-~m-~dT~2")))))) | |
120 | ||
b201e854 | 121 | (define (str->msg nick uri str) |
88d50b3e | 122 | (if (not (regexp-match? re-msg-begin str)) |
f1493e49 SK |
123 | (begin |
124 | (log-debug "Non-msg line from nick:~a, line:~a" nick str) | |
125 | #f) | |
126 | (let ([toks (string-split str (regexp "\t+"))]) | |
c562bea3 SK |
127 | (if (not (= 2 (length toks))) |
128 | (begin | |
129 | (log-warning "Invalid msg line from nick:~a, msg:~a" nick str) | |
130 | #f) | |
131 | (let* | |
132 | ([ts_rfc3339 (first toks)] | |
133 | [text (second toks)] | |
0016b0ad | 134 | [ts_epoch (rfc3339->epoch ts_rfc3339)]) |
c562bea3 | 135 | (msg ts_epoch ts_rfc3339 nick uri text)))))) |
88d50b3e | 136 | |
63afa259 | 137 | (module+ test |
de3ff448 SK |
138 | (let* ([ts "2020-11-18T22:22:09-0500"] |
139 | [tab " "] | |
140 | [text "Lorem ipsum"] | |
141 | [nick "foo"] | |
142 | [uri "bar"] | |
143 | [actual (str->msg nick uri (string-append ts tab text))] | |
144 | [expected (msg 1605756129 ts nick uri text)]) | |
0016b0ad SK |
145 | (check-equal? |
146 | (msg-ts_epoch actual) | |
147 | (msg-ts_epoch expected) | |
148 | "str->msg ts_epoch") | |
c562bea3 SK |
149 | (check-equal? |
150 | (msg-ts_rfc3339 actual) | |
151 | (msg-ts_rfc3339 expected) | |
152 | "str->msg ts_rfc3339") | |
153 | (check-equal? | |
154 | (msg-nick actual) | |
155 | (msg-nick expected) | |
156 | "str->msg nick") | |
157 | (check-equal? | |
158 | (msg-uri actual) | |
159 | (msg-uri expected) | |
160 | "str->msg uri") | |
161 | (check-equal? | |
162 | (msg-text actual) | |
163 | (msg-text expected) | |
164 | "str->msg text"))) | |
63afa259 | 165 | |
e96264cc SK |
166 | (define (str->lines str) |
167 | (string-split str (regexp "[\r\n]+"))) | |
168 | ||
63afa259 | 169 | (module+ test |
de3ff448 | 170 | (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi"))) |
63afa259 | 171 | |
b201e854 SK |
172 | (define (str->msgs nick uri str) |
173 | (filter-map (λ (line) (str->msg nick uri line)) (str->lines str))) | |
4764ff89 | 174 | |
1d753430 SK |
175 | (define (hash-sha1 str) |
176 | (define in (open-input-string str)) | |
177 | (define digest (sha1 in)) | |
178 | (close-input-port in) | |
179 | digest) | |
180 | ||
0e16a46c SK |
181 | (define (uri-fetch use-cache uri) |
182 | (define cache-file-path | |
183 | (expand-user-path (string-append "~/.tt/cache/" (hash-sha1 uri)))) | |
184 | (if (and use-cache (file-exists? cache-file-path)) | |
185 | (begin | |
186 | (log-info "uri-fetch cached ~a" uri) | |
187 | (file->string cache-file-path)) | |
188 | (begin | |
189 | (log-info "uri-fetch new ~a" uri) | |
190 | (let* ([resp (http-get uri)] | |
191 | [status (http-response-code resp)] | |
192 | [body (http-response-body resp)]) | |
c562bea3 SK |
193 | (log-debug "finished GET ~a status:~a body length:~a" |
194 | uri status (string-length body)) | |
195 | ; TODO Handle redirects | |
196 | (if (= status 200) | |
197 | (begin | |
198 | (display-to-file body cache-file-path #:exists 'replace) | |
199 | body) | |
200 | ; TODO A more-informative exception | |
201 | (raise status)))))) | |
4764ff89 | 202 | |
b201e854 | 203 | (define (timeline-print out-format timeline) |
4764ff89 SK |
204 | (for ([msg timeline] |
205 | [i (in-naturals)]) | |
f1493e49 | 206 | (msg-print out-format (odd? i) msg))) |
4764ff89 | 207 | |
0e16a46c | 208 | (define (feed->msgs use-cache feed) |
9a6a9f9a SK |
209 | (log-info "downloading feed nick:~a uri:~a" |
210 | (feed-nick feed) | |
211 | (feed-uri feed)) | |
212 | (with-handlers | |
213 | ([exn:fail:network? | |
214 | (λ (e) | |
215 | (log-error "network error nick:~a uri:~a exn:~a" | |
216 | (feed-nick feed) | |
217 | (feed-uri feed) | |
218 | e) | |
219 | #f)] | |
220 | [integer? | |
221 | (λ (status) | |
222 | (log-error "http error nick:~a uri:~a status:~a" | |
223 | (feed-nick feed) | |
224 | (feed-uri feed) | |
225 | status) | |
226 | #f)]) | |
b201e854 | 227 | (define uri (feed-uri feed)) |
0e16a46c | 228 | (str->msgs [feed-nick feed] uri [uri-fetch use-cache uri]))) |
9a6a9f9a SK |
229 | |
230 | ; TODO timeline contract : time-sorted list of messages | |
0e16a46c SK |
231 | (define (timeline use-cache num_workers feeds) |
232 | (sort (append* (concurrent-filter-map num_workers (curry feed->msgs use-cache) feeds)) | |
87d0eb92 | 233 | (λ (a b) [< (msg-ts_epoch a) (msg-ts_epoch b)]))) |
4764ff89 | 234 | |
be96956c SK |
235 | (define (str->feed str) |
236 | ; TODO validation | |
237 | (define toks (string-split str)) | |
87d0eb92 | 238 | (apply feed toks)) |
be96956c SK |
239 | |
240 | (define (str->feeds str) | |
241 | (map str->feed (str->lines str))) | |
242 | ||
243 | (define (file->feeds filename) | |
244 | (str->feeds (file->string filename))) | |
245 | ||
4f3dbb24 | 246 | (define (user-agent prog-name prog-version) |
24c6a76b | 247 | (let* |
4f3dbb24 | 248 | ([prog-uri "https://github.com/xandkar/tt"] |
24c6a76b SK |
249 | [user-feed-file (expand-user-path "~/twtxt-me.txt")] |
250 | [user | |
251 | (if (file-exists? user-feed-file) | |
f1493e49 | 252 | (let ([user (first (file->feeds user-feed-file))]) |
c562bea3 | 253 | (format "+~a; @~a" (feed-uri user) (feed-nick user))) |
f1493e49 | 254 | (format "+~a" prog-uri))] |
24c6a76b SK |
255 | ) |
256 | (format "~a/~a (~a)" prog-name prog-version user))) | |
257 | ||
258 | (module+ main | |
de3ff448 | 259 | (require setup/getinfo) |
26d172a9 | 260 | |
de3ff448 SK |
261 | (let* ([level 'info] |
262 | [logger (make-logger #f #f level #f)] | |
263 | [log-receiver (make-log-receiver logger level)]) | |
c562bea3 SK |
264 | (void (thread (λ () |
265 | [date-display-format 'iso-8601] | |
266 | [let loop () | |
267 | (define data (sync log-receiver)) | |
268 | (define level (vector-ref data 0)) | |
269 | (define msg (vector-ref data 1)) | |
270 | (define ts (date->string (current-date) #t)) | |
271 | (eprintf "~a [~a] ~a~n" ts level msg) | |
272 | (loop)]))) | |
273 | (current-logger logger)) | |
de3ff448 SK |
274 | (current-http-response-auto #f) |
275 | (let* ([prog-name "tt"] | |
276 | [prog-version ((get-info (list prog-name)) 'version)] | |
277 | [user-agent (user-agent prog-name prog-version)]) | |
c562bea3 | 278 | (current-http-user-agent user-agent)) |
de3ff448 SK |
279 | (date-display-format 'rfc2822) |
280 | (let* ([use-cache | |
281 | #f] | |
282 | [out-format | |
283 | 'multi-line] | |
284 | [num_workers | |
285 | 15]) ; 15 was fastest out of the tried 1, 5, 10, 15 and 20. | |
c562bea3 SK |
286 | (command-line |
287 | #:once-each | |
288 | [("-c" "--cached") | |
289 | "Read cached data instead of downloading." | |
290 | (set! use-cache #t)] | |
0e16a46c | 291 | |
c562bea3 SK |
292 | [("-j" "--jobs") |
293 | njobs "Number of concurrent jobs." | |
294 | (set! num_workers (string->number njobs))] | |
0e16a46c | 295 | |
c562bea3 | 296 | #:args (filename) |
0e16a46c | 297 | |
c562bea3 SK |
298 | (timeline-print out-format |
299 | (timeline use-cache | |
300 | num_workers | |
301 | (file->feeds filename)))))) |