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 |
f1493e49 | 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) | |
f1493e49 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) | |
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 |
4d958983 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) <)]) | |
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 | |
f1493e49 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 | |
b201e854 | 112 | (define (str->msg nick uri str) |
88d50b3e | 113 | (if (not (regexp-match? re-msg-begin str)) |
f1493e49 SK |
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)))))) | |
88d50b3e | 134 | |
63afa259 SK |
135 | (module+ test |
136 | (let* ([ts "2020-11-18T22:22:09-0500"] | |
137 | [tab " "] | |
138 | [text "Lorem ipsum"] | |
139 | [nick "foo"] | |
140 | [uri "bar"] | |
141 | [actual (str->msg nick uri (string-join (list ts text) tab))] | |
142 | [expected (msg 1605756129 ts nick uri text)]) | |
2822e4ff SK |
143 | ; FIXME re-enable after handling tz offset |
144 | ;(check-equal? | |
145 | ; (msg-ts_epoch actual) | |
146 | ; (msg-ts_epoch expected) | |
147 | ; "str->msg ts_epoch") | |
63afa259 SK |
148 | (check-equal? |
149 | (msg-ts_rfc3339 actual) | |
150 | (msg-ts_rfc3339 expected) | |
151 | "str->msg ts_rfc3339") | |
152 | (check-equal? | |
153 | (msg-nick actual) | |
154 | (msg-nick expected) | |
155 | "str->msg nick") | |
156 | (check-equal? | |
157 | (msg-uri actual) | |
158 | (msg-uri expected) | |
159 | "str->msg uri") | |
160 | (check-equal? | |
161 | (msg-text actual) | |
162 | (msg-text expected) | |
163 | "str->msg text"))) | |
164 | ||
e96264cc SK |
165 | (define (str->lines str) |
166 | (string-split str (regexp "[\r\n]+"))) | |
167 | ||
63afa259 SK |
168 | (module+ test |
169 | (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi"))) | |
170 | ||
b201e854 SK |
171 | (define (str->msgs nick uri str) |
172 | (filter-map (λ (line) (str->msg nick uri line)) (str->lines str))) | |
4764ff89 | 173 | |
1d753430 SK |
174 | (define (hash-sha1 str) |
175 | (define in (open-input-string str)) | |
176 | (define digest (sha1 in)) | |
177 | (close-input-port in) | |
178 | digest) | |
179 | ||
0e16a46c SK |
180 | (define (uri-fetch use-cache uri) |
181 | (define cache-file-path | |
182 | (expand-user-path (string-append "~/.tt/cache/" (hash-sha1 uri)))) | |
183 | (if (and use-cache (file-exists? cache-file-path)) | |
184 | (begin | |
185 | (log-info "uri-fetch cached ~a" uri) | |
186 | (file->string cache-file-path)) | |
187 | (begin | |
188 | (log-info "uri-fetch new ~a" uri) | |
189 | (let* ([resp (http-get uri)] | |
190 | [status (http-response-code resp)] | |
191 | [body (http-response-body resp)]) | |
192 | (log-debug "finished GET ~a status:~a body length:~a" | |
193 | uri status (string-length body)) | |
194 | ; TODO Handle redirects | |
195 | (if (= status 200) | |
196 | (begin | |
197 | (display-to-file body cache-file-path #:exists 'replace) | |
198 | body) | |
199 | ; TODO A more-informative exception | |
200 | (raise status)))))) | |
4764ff89 | 201 | |
b201e854 | 202 | (define (timeline-print out-format timeline) |
4764ff89 SK |
203 | (for ([msg timeline] |
204 | [i (in-naturals)]) | |
f1493e49 | 205 | (msg-print out-format (odd? i) msg))) |
4764ff89 | 206 | |
0e16a46c | 207 | (define (feed->msgs use-cache feed) |
9a6a9f9a SK |
208 | (log-info "downloading feed nick:~a uri:~a" |
209 | (feed-nick feed) | |
210 | (feed-uri feed)) | |
211 | (with-handlers | |
212 | ([exn:fail:network? | |
213 | (λ (e) | |
214 | (log-error "network error nick:~a uri:~a exn:~a" | |
215 | (feed-nick feed) | |
216 | (feed-uri feed) | |
217 | e) | |
218 | #f)] | |
219 | [integer? | |
220 | (λ (status) | |
221 | (log-error "http error nick:~a uri:~a status:~a" | |
222 | (feed-nick feed) | |
223 | (feed-uri feed) | |
224 | status) | |
225 | #f)]) | |
b201e854 | 226 | (define uri (feed-uri feed)) |
0e16a46c | 227 | (str->msgs [feed-nick feed] uri [uri-fetch use-cache uri]))) |
9a6a9f9a SK |
228 | |
229 | ; TODO timeline contract : time-sorted list of messages | |
0e16a46c SK |
230 | (define (timeline use-cache num_workers feeds) |
231 | (sort (append* (concurrent-filter-map num_workers (curry feed->msgs use-cache) feeds)) | |
87d0eb92 | 232 | (λ (a b) [< (msg-ts_epoch a) (msg-ts_epoch b)]))) |
4764ff89 | 233 | |
be96956c SK |
234 | (define (str->feed str) |
235 | ; TODO validation | |
236 | (define toks (string-split str)) | |
87d0eb92 | 237 | (apply feed toks)) |
be96956c SK |
238 | |
239 | (define (str->feeds str) | |
240 | (map str->feed (str->lines str))) | |
241 | ||
242 | (define (file->feeds filename) | |
243 | (str->feeds (file->string filename))) | |
244 | ||
4f3dbb24 | 245 | (define (user-agent prog-name prog-version) |
24c6a76b | 246 | (let* |
4f3dbb24 | 247 | ([prog-uri "https://github.com/xandkar/tt"] |
24c6a76b SK |
248 | [user-feed-file (expand-user-path "~/twtxt-me.txt")] |
249 | [user | |
250 | (if (file-exists? user-feed-file) | |
f1493e49 SK |
251 | (let ([user (first (file->feeds user-feed-file))]) |
252 | (format "+~a; @~a" (feed-uri user) (feed-nick user))) | |
253 | (format "+~a" prog-uri))] | |
24c6a76b SK |
254 | ) |
255 | (format "~a/~a (~a)" prog-name prog-version user))) | |
256 | ||
257 | (module+ main | |
f1493e49 | 258 | (require setup/getinfo) |
26d172a9 | 259 | |
0e16a46c SK |
260 | (let* ([level 'info] |
261 | [logger (make-logger #f #f level #f)] | |
262 | [log-receiver (make-log-receiver logger level)]) | |
f1493e49 SK |
263 | (void (thread (λ () |
264 | [date-display-format 'iso-8601] | |
265 | [let loop () | |
266 | (define data (sync log-receiver)) | |
267 | (define level (vector-ref data 0)) | |
268 | (define msg (vector-ref data 1)) | |
269 | (define ts (date->string (current-date) #t)) | |
270 | (eprintf "~a [~a] ~a~n" ts level msg) | |
271 | (loop)]))) | |
272 | (current-logger logger)) | |
273 | (current-http-response-auto #f) | |
274 | (let* ([prog-name "tt"] | |
275 | [prog-version ((get-info (list prog-name)) 'version)] | |
276 | [user-agent (user-agent prog-name prog-version)]) | |
277 | (current-http-user-agent user-agent)) | |
278 | (date-display-format 'rfc2822) | |
0e16a46c SK |
279 | (let* ([use-cache |
280 | #f] | |
281 | [out-format | |
282 | 'multi-line] | |
283 | [num_workers | |
284 | 15]) ; 15 was fastest out of the tried 1, 5, 10, 15 and 20. | |
285 | (command-line | |
286 | #:once-any | |
287 | [("-c" "--cached") | |
288 | "Read cached data instead of downloading." | |
289 | (set! use-cache #t)] | |
290 | ||
291 | [("-j" "--jobs") | |
292 | njobs "Number of concurrent jobs." | |
293 | (set! num_workers (string->number njobs))] | |
294 | ||
295 | #:args (filename) | |
296 | ||
297 | (timeline-print out-format | |
298 | (timeline use-cache | |
299 | num_workers | |
300 | (file->feeds filename)))))) |