Commit | Line | Data |
---|---|---|
cd541405 | 1 | #lang typed/racket/no-check |
4764ff89 | 2 | |
1d753430 | 3 | (require openssl/sha1) |
4764ff89 | 4 | (require racket/date) |
9c464d95 | 5 | (require |
d718efc4 | 6 | net/head |
edadb804 | 7 | net/uri-codec |
8da029e4 | 8 | net/url) |
4764ff89 | 9 | |
3c9c8266 SK |
10 | (require (prefix-in info: "info.rkt")) |
11 | ||
78398948 | 12 | (module+ test |
de3ff448 | 13 | (require rackunit)) |
78398948 | 14 | |
98529d3d SK |
15 | (define-type Url |
16 | net/url-structs:url) | |
17 | ||
18 | (define-type Out-Format | |
19 | (U 'single-line | |
20 | 'multi-line)) | |
21 | ||
22 | (define-type Timeline-Order | |
23 | (U 'old->new | |
24 | 'new->old)) | |
25 | ||
7fd20778 SK |
26 | (define-type Result |
27 | (∀ (α β) (U (cons 'ok α) | |
28 | (cons 'error β)))) | |
29 | ||
b0ff061a | 30 | (struct Msg |
78142acb | 31 | ([ts-epoch : Integer] |
3877a0c4 | 32 | [ts-orig : String] |
0cb1ae9c | 33 | [from : Peer] |
13c11724 | 34 | [text : String] |
b0ff061a | 35 | [mentions : (Listof Peer)])) |
9c464d95 | 36 | |
dbc26280 | 37 | (struct Peer |
b056019b SK |
38 | ([nick : (Option String)] |
39 | [uri : Url] | |
e2840743 | 40 | [uri-str : String] |
b056019b | 41 | [comment : (Option String)]) |
d0a0e073 | 42 | #:transparent) |
4764ff89 | 43 | |
f65d6338 SK |
44 | (struct Resp |
45 | ([status-line : String] | |
46 | [headers : (Listof Bytes)] | |
7fd20778 SK |
47 | [body-input : Input-Port]) |
48 | #:transparent) | |
f65d6338 | 49 | |
0cb1ae9c SK |
50 | (: peers-equal? (-> Peer Peer Boolean)) |
51 | (define (peers-equal? p1 p2) | |
52 | (equal? (Peer-uri-str p1) | |
53 | (Peer-uri-str p2))) | |
54 | ||
55 | (: peer-hash (-> Peer Fixnum)) | |
56 | (define (peer-hash p) | |
57 | (equal-hash-code (Peer-uri-str p))) | |
58 | ||
e2840743 SK |
59 | (define-custom-set-types peers |
60 | #:elem? Peer? | |
0cb1ae9c SK |
61 | peers-equal? |
62 | peer-hash) | |
e2840743 SK |
63 | ; XXX Without supplying above explicit hash procedure, we INTERMITTENTLY get |
64 | ; the following contract violations: | |
65 | ; | |
66 | ; custom-elem-contents: contract violation | |
67 | ; expected: custom-elem? | |
68 | ; given: #f | |
69 | ; context...: | |
70 | ; /usr/share/racket/collects/racket/private/set-types.rkt:104:0: custom-set->list | |
71 | ; /home/siraaj/proj/pub/tt/tt.rkt:716:0: crawl | |
72 | ; /usr/share/racket/collects/racket/cmdline.rkt:191:51 | |
73 | ; body of (submod "/home/siraaj/proj/pub/tt/tt.rkt" main) | |
74 | ; | |
75 | ; TODO Investigate why and make a minimal reproducible test case. | |
76 | ||
50f0609d SK |
77 | (define (peers-union . peer-sets) |
78 | (define groups | |
79 | (foldl | |
80 | (λ (p groups) | |
81 | (hash-update groups (Peer-uri-str p) (λ (group) (cons p group)) '())) | |
82 | (hash) | |
83 | (append* (map set->list peer-sets)))) | |
84 | (define (merge peers) | |
85 | (match peers | |
86 | ['() (raise 'impossible)] | |
87 | [(list p) p] | |
88 | [(list* p1 p2 ps) | |
89 | (let* ([n1 (Peer-nick p1)] | |
90 | [n2 (Peer-nick p2)] | |
91 | [p (cond | |
92 | [(and (not n1) (not n2)) p1] | |
93 | [(and n1 n2 ) p1] | |
94 | [(and n1 (not n2)) p1] | |
95 | [(and (not n1) n2) p2] | |
96 | [else | |
97 | (raise 'impossible)])]) | |
98 | (merge (cons p ps)))])) | |
99 | (make-immutable-peers (map merge (hash-values groups)))) | |
100 | ||
101 | (module+ test | |
102 | (let* ([u1 "http://foo/bar"] | |
103 | [u2 "http://baz/quux"] | |
104 | [p1 (Peer #f (string->url u1) u1 #f)] | |
105 | [p2 (Peer "a" (string->url u1) u1 #f)] | |
106 | [p3 (Peer "b" (string->url u2) u2 #f)] | |
107 | [s1 (make-immutable-peers (list p1))] | |
108 | [s2 (make-immutable-peers (list p2 p3))]) | |
109 | (check-true (peers? (peers-union s1 s2))) | |
110 | (check-true (peers? (peers-union s2 s1))) | |
111 | (check-equal? (list p3 p2) (set->list (peers-union s1 s2))) | |
112 | (check-equal? (list p3 p2) (set->list (peers-union s2 s1))))) | |
113 | ||
edadb804 SK |
114 | (: tt-home-dir Path-String) |
115 | (define tt-home-dir (build-path (expand-user-path "~") ".tt")) | |
116 | ||
9a346534 | 117 | (: concurrent-filter-map (∀ (α β) (-> Natural (-> α β) (Listof α) (Listof β)))) |
78142acb | 118 | (define (concurrent-filter-map num-workers f xs) |
dad4504d | 119 | ; TODO preserve order of elements OR communicate that reorder is expected |
a239a233 | 120 | ; TODO switch from mailboxes to channels |
895a32cf SK |
121 | (define (make-worker id f) |
122 | (define parent (current-thread)) | |
123 | (λ () | |
a9511f7c SK |
124 | (define self : Thread (current-thread)) |
125 | (: work (∀ (α) (-> α))) | |
895a32cf SK |
126 | (define (work) |
127 | (thread-send parent (cons 'next self)) | |
128 | (match (thread-receive) | |
c562bea3 SK |
129 | ['done (thread-send parent (cons 'exit id))] |
130 | [(cons 'unit x) (begin | |
131 | (define y (f x)) | |
132 | (when y (thread-send parent (cons 'result y))) | |
133 | (work))])) | |
895a32cf | 134 | (work))) |
a9511f7c | 135 | (: dispatch (∀ (α β) (-> (Listof Nonnegative-Integer) (Listof α) (Listof β)))) |
895a32cf SK |
136 | (define (dispatch ws xs ys) |
137 | (if (empty? ws) | |
f1493e49 SK |
138 | ys |
139 | (match (thread-receive) | |
c562bea3 SK |
140 | [(cons 'exit w) (dispatch (remove w ws =) xs ys)] |
141 | [(cons 'result y) (dispatch ws xs (cons y ys))] | |
142 | [(cons 'next thd) (match xs | |
143 | ['() (begin | |
144 | (thread-send thd 'done) | |
145 | (dispatch ws xs ys))] | |
146 | [(cons x xs) (begin | |
147 | (thread-send thd (cons 'unit x)) | |
148 | (dispatch ws xs ys))])]))) | |
78142acb | 149 | (define workers (range num-workers)) |
9926c9a9 SK |
150 | (define threads (map (λ (id) (thread (make-worker id f))) workers)) |
151 | (define results (dispatch workers xs '())) | |
895a32cf SK |
152 | (for-each thread-wait threads) |
153 | results) | |
154 | ||
dad4504d | 155 | (module+ test |
de3ff448 SK |
156 | (let* ([f (λ (x) (if (even? x) x #f))] |
157 | [xs (range 11)] | |
158 | [actual (sort (concurrent-filter-map 10 f xs) <)] | |
159 | [expected (sort ( filter-map f xs) <)]) | |
c562bea3 | 160 | (check-equal? actual expected "concurrent-filter-map"))) |
dad4504d | 161 | |
98529d3d | 162 | (: msg-print (-> Out-Format Integer Msg Void)) |
3d042e75 SK |
163 | (define msg-print |
164 | (let* ([colors (vector 36 33)] | |
165 | [n (vector-length colors)]) | |
166 | (λ (out-format color-i msg) | |
01e4c499 | 167 | (let ([color (vector-ref colors (modulo color-i n))] |
0cb1ae9c SK |
168 | [nick (Peer-nick (Msg-from msg))] |
169 | [uri (Peer-uri-str (Msg-from msg))] | |
170 | [text (Msg-text msg)]) | |
3d042e75 | 171 | (match out-format |
01e4c499 | 172 | ['single-line |
13c11724 SK |
173 | (let ([nick (if nick nick uri)]) |
174 | (printf "~a \033[1;37m<~a>\033[0m \033[0;~am~a\033[0m~n" | |
175 | (parameterize | |
176 | ([date-display-format 'iso-8601]) | |
b0ff061a | 177 | (date->string (seconds->date (Msg-ts-epoch msg)) #t)) |
13c11724 | 178 | nick color text))] |
01e4c499 | 179 | ['multi-line |
13c11724 SK |
180 | (let ([nick (if nick (string-append nick " ") "")]) |
181 | (printf "~a (~a)~n\033[1;37m<~a~a>\033[0m~n\033[0;~am~a\033[0m~n~n" | |
182 | (parameterize | |
183 | ([date-display-format 'rfc2822]) | |
b0ff061a SK |
184 | (date->string (seconds->date (Msg-ts-epoch msg)) #t)) |
185 | (Msg-ts-orig msg) | |
13c11724 | 186 | nick uri color text))]))))) |
e96264cc | 187 | |
3877a0c4 SK |
188 | (: rfc3339->epoch (-> String (Option Nonnegative-Integer))) |
189 | (define rfc3339->epoch | |
190 | (let ([re (pregexp "^([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2})(:([0-9]{2}))?(\\.[0-9]+)?(Z|([+-])([0-9]{1,2}):?([0-9]{2}))?$")]) | |
191 | (λ (ts) | |
192 | (match (regexp-match re ts) | |
193 | [(list _wholething yyyy mm dd HH MM _:SS SS _fractional tz-whole tz-sign tz-HH tz-MM) | |
194 | (let* | |
195 | ([tz-offset | |
196 | (match* (tz-whole tz-sign tz-HH tz-MM) | |
197 | [("Z" #f #f #f) | |
198 | 0] | |
199 | [(_ (or "-" "+") (? identity h) (? identity m)) | |
200 | (let ([h (string->number h)] | |
201 | [m (string->number m)] | |
202 | ; Reverse to get back to UTC: | |
203 | [op (match tz-sign ["+" -] ["-" +])]) | |
204 | (op 0 (+ (* 60 m) (* 60 (* 60 h)))))] | |
205 | [(a b c d) | |
206 | (log-warning "Impossible TZ string: ~v, components: ~v ~v ~v ~v" tz-whole a b c d) | |
207 | 0])] | |
208 | [ts-orig ts] | |
209 | [local-time? #f] | |
210 | [ts-epoch (find-seconds (if SS (string->number SS) 0) | |
211 | (string->number MM) | |
212 | (string->number HH) | |
213 | (string->number dd) | |
214 | (string->number mm) | |
215 | (string->number yyyy) | |
216 | local-time?)]) | |
217 | (+ ts-epoch tz-offset))] | |
218 | [_ | |
b8b29fbb | 219 | (log-debug "Invalid timestamp: ~v" ts) |
3877a0c4 SK |
220 | #f])))) |
221 | ||
0cb1ae9c | 222 | (: str->msg (-> Peer String (Option Msg))) |
b4689464 | 223 | (define str->msg |
3877a0c4 | 224 | (let ([re (pregexp "^([^\\s\t]+)[\\s\t]+(.*)$")]) |
0cb1ae9c SK |
225 | (λ (from str) |
226 | (define from-str (peer->str from)) | |
d3ac9e11 | 227 | (define str-head (substring str 0 (min 100 (string-length str)))) |
b4689464 SK |
228 | (with-handlers* |
229 | ([exn:fail? | |
230 | (λ (e) | |
b8b29fbb | 231 | (log-debug |
9c464d95 | 232 | "Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v" |
0cb1ae9c | 233 | str-head from-str e) |
b4689464 SK |
234 | #f)]) |
235 | (match (regexp-match re str) | |
3877a0c4 SK |
236 | [(list _wholething ts-orig text) |
237 | (let ([ts-epoch (rfc3339->epoch ts-orig)]) | |
238 | (if ts-epoch | |
13c11724 SK |
239 | (let ([mentions |
240 | (filter-map | |
241 | (λ (m) (match (regexp-match #px"@<([^>]+)>" m) | |
242 | [(list _wholething nick-uri) | |
dbc26280 | 243 | (str->peer nick-uri)])) |
13c11724 | 244 | (regexp-match* #px"@<[^\\s]+([\\s]+)?[^>]+>" text))]) |
0cb1ae9c | 245 | (Msg ts-epoch ts-orig from text mentions)) |
3877a0c4 | 246 | (begin |
b8b29fbb | 247 | (log-debug |
0cb1ae9c SK |
248 | "Msg rejected due to invalid timestamp. From:~v. Line:~v" |
249 | from-str str-head) | |
3877a0c4 | 250 | #f)))] |
b4689464 | 251 | [_ |
0cb1ae9c | 252 | (log-debug "Non-msg line. From:~v. Line:~v" from-str str-head) |
b4689464 | 253 | #f]))))) |
88d50b3e | 254 | |
63afa259 | 255 | (module+ test |
13c11724 | 256 | ; TODO Test for when missing-nick case |
b4689464 SK |
257 | (let* ([tzs (for*/list ([d '("-" "+")] |
258 | [h '("5" "05")] | |
259 | [m '("00" ":00" "57" ":57")]) | |
260 | (string-append d h m))] | |
261 | [tzs (list* "" "Z" tzs)]) | |
262 | (for* ([n '("fake-nick")] | |
0cb1ae9c | 263 | [u '("http://fake-uri")] |
50f0609d | 264 | [p (list (Peer n (string->url u) u #f))] |
b4689464 SK |
265 | [s '("" ":10")] |
266 | [f '("" ".1337")] | |
267 | [z tzs] | |
268 | [sep (list "\t" " ")] | |
269 | [txt '("foo bar baz" "'jaz poop bear giraffe / tea" "@*\"``")]) | |
270 | (let* ([ts (string-append "2020-11-18T22:22" | |
271 | (if (non-empty-string? s) s ":00") | |
272 | z)] | |
0cb1ae9c | 273 | [m (str->msg p (string-append ts sep txt))]) |
b4689464 | 274 | (check-not-false m) |
0cb1ae9c | 275 | (check-equal? (Msg-from m) p) |
b0ff061a SK |
276 | (check-equal? (Msg-text m) txt) |
277 | (check-equal? (Msg-ts-orig m) ts (format "Given: ~v" ts)) | |
b4689464 SK |
278 | ))) |
279 | ||
de3ff448 SK |
280 | (let* ([ts "2020-11-18T22:22:09-0500"] |
281 | [tab " "] | |
282 | [text "Lorem ipsum"] | |
283 | [nick "foo"] | |
0cb1ae9c | 284 | [uri "http://bar/"] |
50f0609d | 285 | [peer (Peer nick (string->url uri) uri #f)] |
0cb1ae9c SK |
286 | [actual (str->msg peer (string-append ts tab text))] |
287 | [expected (Msg 1605756129 ts peer text '())]) | |
c562bea3 | 288 | (check-equal? |
b0ff061a SK |
289 | (Msg-ts-epoch actual) |
290 | (Msg-ts-epoch expected) | |
78142acb | 291 | "str->msg ts-epoch") |
3877a0c4 | 292 | (check-equal? |
b0ff061a SK |
293 | (Msg-ts-orig actual) |
294 | (Msg-ts-orig expected) | |
3877a0c4 | 295 | "str->msg ts-orig") |
c562bea3 | 296 | (check-equal? |
0cb1ae9c SK |
297 | (Peer-nick (Msg-from actual)) |
298 | (Peer-nick (Msg-from expected)) | |
c562bea3 SK |
299 | "str->msg nick") |
300 | (check-equal? | |
0cb1ae9c SK |
301 | (Peer-uri (Msg-from actual)) |
302 | (Peer-uri (Msg-from expected)) | |
c562bea3 | 303 | "str->msg uri") |
0cb1ae9c SK |
304 | (check-equal? |
305 | (Peer-uri-str (Msg-from actual)) | |
306 | (Peer-uri-str (Msg-from expected)) | |
307 | "str->msg uri-str") | |
c562bea3 | 308 | (check-equal? |
b0ff061a SK |
309 | (Msg-text actual) |
310 | (Msg-text expected) | |
c562bea3 | 311 | "str->msg text"))) |
63afa259 | 312 | |
98529d3d | 313 | (: str->lines (-> String (Listof String))) |
e96264cc SK |
314 | (define (str->lines str) |
315 | (string-split str (regexp "[\r\n]+"))) | |
316 | ||
63afa259 | 317 | (module+ test |
de3ff448 | 318 | (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi"))) |
63afa259 | 319 | |
0cb1ae9c SK |
320 | (: str->msgs (-> Peer String (Listof Msg))) |
321 | (define (str->msgs peer str) | |
322 | (filter-map (λ (line) (str->msg peer line)) | |
323 | (filter-comments (str->lines str)))) | |
4764ff89 | 324 | |
edadb804 SK |
325 | (: cache-dir Path-String) |
326 | (define cache-dir (build-path tt-home-dir "cache")) | |
327 | ||
d718efc4 SK |
328 | (define cache-object-dir (build-path cache-dir "objects")) |
329 | ||
edadb804 SK |
330 | (: url->cache-file-path-v1 (-> Url Path-String)) |
331 | (define (url->cache-file-path-v1 uri) | |
332 | (define (hash-sha1 str) : (-> String String) | |
333 | (define in (open-input-string str)) | |
334 | (define digest (sha1 in)) | |
335 | (close-input-port in) | |
336 | digest) | |
d718efc4 | 337 | (build-path cache-object-dir (hash-sha1 (url->string uri)))) |
edadb804 SK |
338 | |
339 | (: url->cache-file-path-v2 (-> Url Path-String)) | |
340 | (define (url->cache-file-path-v2 uri) | |
d718efc4 SK |
341 | (build-path cache-object-dir (uri-encode (url->string uri)))) |
342 | ||
0cb1ae9c SK |
343 | (define url->cache-object-path |
344 | url->cache-file-path-v2) | |
d3ac9e11 | 345 | |
d718efc4 SK |
346 | (define (url->cache-etag-path uri) |
347 | (build-path cache-dir "etags" (uri-encode (url->string uri)))) | |
348 | ||
349 | (define (url->cache-lmod-path uri) | |
350 | (build-path cache-dir "lmods" (uri-encode (url->string uri)))) | |
9c464d95 | 351 | |
fee11be9 | 352 | (: uri-read-cached (-> Url (Option String))) |
4214c0f3 | 353 | (define (uri-read-cached uri) |
edadb804 SK |
354 | (define path-v1 (url->cache-file-path-v1 uri)) |
355 | (define path-v2 (url->cache-file-path-v2 uri)) | |
356 | (when (file-exists? path-v1) | |
357 | (rename-file-or-directory path-v1 path-v2 #t)) | |
358 | (if (file-exists? path-v2) | |
359 | (file->string path-v2) | |
0e16a46c | 360 | (begin |
b8b29fbb | 361 | (log-debug "Cache file not found for URI: ~a" (url->string uri)) |
fee11be9 | 362 | #f))) |
4214c0f3 | 363 | |
b056019b SK |
364 | (: str->url (-> String (Option String))) |
365 | (define (str->url s) | |
366 | (with-handlers* | |
367 | ([exn:fail? (λ (e) #f)]) | |
368 | (string->url s))) | |
a60c484e | 369 | |
0cb1ae9c SK |
370 | (: peer->str (-> Peer String)) |
371 | (define (peer->str peer) | |
372 | (match-define (Peer n _ u c) peer) | |
373 | (format "~a~a~a" | |
374 | (if n (format "~a " n) "") | |
375 | u | |
376 | (if c (format " # ~a" c) ""))) | |
377 | ||
378 | (: str->peer (-> String (Option Peer))) | |
dbc26280 SK |
379 | (define (str->peer str) |
380 | (log-debug "Parsing peer string: ~v" str) | |
b056019b SK |
381 | (match |
382 | (regexp-match | |
383 | #px"(([^\\s\t]+)[\\s\t]+)?([a-zA-Z]+://[^\\s\t]*)[\\s\t]*(#\\s*(.*))?" | |
384 | str) | |
385 | [(list _wholething | |
386 | _nick-with-space | |
387 | nick | |
388 | url | |
389 | _comment-with-hash | |
390 | comment) | |
391 | (match (str->url url) | |
392 | [#f | |
393 | (log-error "Invalid URI in peer string: ~v" str) | |
394 | #f] | |
e2840743 SK |
395 | [url |
396 | (Peer nick url (url->string url) comment)])] | |
b056019b | 397 | [_ |
b8b29fbb | 398 | (log-debug "Invalid peer string: ~v" str) |
b056019b | 399 | #f])) |
13c11724 | 400 | |
b056019b SK |
401 | (module+ test |
402 | (check-equal? | |
403 | (str->peer "foo http://bar/file.txt # some rando") | |
e2840743 | 404 | (Peer "foo" (str->url "http://bar/file.txt") "http://bar/file.txt" "some rando")) |
b056019b SK |
405 | (check-equal? |
406 | (str->peer "http://bar/file.txt # some rando") | |
e2840743 | 407 | (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" "some rando")) |
b056019b SK |
408 | (check-equal? |
409 | (str->peer "http://bar/file.txt #") | |
e2840743 | 410 | (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" "")) |
b056019b SK |
411 | (check-equal? |
412 | (str->peer "http://bar/file.txt#") ; XXX URLs can have #s | |
e2840743 | 413 | (Peer #f (str->url "http://bar/file.txt#") "http://bar/file.txt#" #f)) |
b056019b SK |
414 | (check-equal? |
415 | (str->peer "http://bar/file.txt") | |
e2840743 | 416 | (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" #f)) |
b056019b SK |
417 | (check-equal? |
418 | (str->peer "foo http://bar/file.txt") | |
e2840743 | 419 | (Peer "foo" (str->url "http://bar/file.txt") "http://bar/file.txt" #f)) |
b056019b SK |
420 | (check-equal? |
421 | (str->peer "foo bar # baz") | |
422 | #f) | |
423 | (check-equal? | |
424 | (str->peer "foo bar://baz # quux") | |
e2840743 | 425 | (Peer "foo" (str->url "bar://baz") "bar://baz" "quux")) |
b056019b SK |
426 | (check-equal? |
427 | (str->peer "foo bar//baz # quux") | |
428 | #f)) | |
9c464d95 | 429 | |
98529d3d | 430 | (: filter-comments (-> (Listof String) (Listof String))) |
9c464d95 SK |
431 | (define (filter-comments lines) |
432 | (filter-not (λ (line) (string-prefix? line "#")) lines)) | |
433 | ||
e2840743 | 434 | (: str->peers (-> String (Setof Peer))) |
dbc26280 | 435 | (define (str->peers str) |
e2840743 | 436 | (make-immutable-peers (filter-map str->peer (filter-comments (str->lines str))))) |
9c464d95 | 437 | |
e2840743 | 438 | (: peers->file (-> (Setof Peers) Path-String Void)) |
a60c484e SK |
439 | (define (peers->file peers path) |
440 | (display-lines-to-file | |
0cb1ae9c | 441 | (map peer->str |
e2840743 SK |
442 | (sort (set->list peers) |
443 | (match-lambda** | |
444 | [((Peer n1 _ _ _) (Peer n2 _ _ _)) | |
445 | (string<? (if n1 n1 "") | |
446 | (if n2 n2 ""))]))) | |
a60c484e SK |
447 | path |
448 | #:exists 'replace)) | |
449 | ||
e2840743 | 450 | (: file->peers (-> Path-String (Setof Peer))) |
d0a0e073 SK |
451 | (define (file->peers file-path) |
452 | (if (file-exists? file-path) | |
453 | (str->peers (file->string file-path)) | |
454 | (begin | |
a60c484e | 455 | (log-warning "File does not exist: ~v" (path->string file-path)) |
e2840743 | 456 | (make-immutable-peers)))) |
9c464d95 | 457 | |
9c5e4499 SK |
458 | (define re-rfc2822 |
459 | #px"^(Mon|Tue|Wed|Thu|Fri|Sat|Sun), ([0-9]{2}) (Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec) ([0-9]{4}) ([0-2][0-9]):([0-6][0-9]):([0-6][0-9]) GMT") | |
460 | ||
461 | (: b->n (-> Bytes (Option Number))) | |
462 | (define (b->n b) | |
463 | (string->number (bytes->string/utf-8 b))) | |
464 | ||
465 | (: mon->num (-> Bytes Natural)) | |
466 | (define/match (mon->num mon) | |
467 | [(#"Jan") 1] | |
468 | [(#"Feb") 2] | |
469 | [(#"Mar") 3] | |
470 | [(#"Apr") 4] | |
471 | [(#"May") 5] | |
472 | [(#"Jun") 6] | |
473 | [(#"Jul") 7] | |
474 | [(#"Aug") 8] | |
475 | [(#"Sep") 9] | |
476 | [(#"Oct") 10] | |
477 | [(#"Nov") 11] | |
478 | [(#"Dec") 12]) | |
479 | ||
480 | (: rfc2822->epoch (-> Bytes (Option Nonnegative-Integer))) | |
481 | (define (rfc2822->epoch timestamp) | |
482 | (match (regexp-match re-rfc2822 timestamp) | |
483 | [(list _ _ dd mo yyyy HH MM SS) | |
484 | #:when (and dd mo yyyy HH MM SS) | |
485 | (find-seconds (b->n SS) | |
486 | (b->n MM) | |
487 | (b->n HH) | |
488 | (b->n dd) | |
489 | (mon->num mo) | |
490 | (b->n yyyy) | |
491 | #f)] | |
492 | [_ | |
493 | #f])) | |
494 | ||
98529d3d | 495 | (: user-agent String) |
9c464d95 SK |
496 | (define user-agent |
497 | (let* | |
498 | ([prog-name "tt"] | |
3c9c8266 | 499 | [prog-version (info:#%info-lookup 'version)] |
9c464d95 | 500 | [prog-uri "https://github.com/xandkar/tt"] |
dbc26280 | 501 | [user-peer-file (build-path tt-home-dir "me")] |
9c464d95 | 502 | [user |
dbc26280 | 503 | (if (file-exists? user-peer-file) |
e2840743 SK |
504 | (match (set-first (file->peers user-peer-file)) |
505 | [(Peer #f _ u _) (format "+~a" u )] | |
506 | [(Peer n _ u _) (format "+~a; @~a" u n)]) | |
9c464d95 SK |
507 | (format "+~a" prog-uri))]) |
508 | (format "~a/~a (~a)" prog-name prog-version user))) | |
509 | ||
d718efc4 SK |
510 | (: header-get (-> (Listof Bytes) Bytes (Option Bytes))) |
511 | (define (header-get headers name) | |
512 | (match (filter-map (curry extract-field name) headers) | |
513 | [(list val) val] | |
514 | [_ #f])) | |
515 | ||
7fd20778 SK |
516 | (: uri-download-from-port |
517 | (-> Url (Listof (U Bytes String)) Input-Port | |
518 | (U 'skipped-cached 'downloaded-new))) ; TODO 'ok|'error ? | |
519 | (define (uri-download-from-port u headers body-input) | |
520 | (define u-str (url->string u)) | |
521 | (log-debug "uri-download-from-port ~v into ~v" u-str cached-object-path) | |
d718efc4 SK |
522 | (define cached-object-path (url->cache-object-path u)) |
523 | (define cached-etag-path (url->cache-etag-path u)) | |
524 | (define cached-lmod-path (url->cache-lmod-path u)) | |
7fd20778 SK |
525 | (define etag (header-get headers #"ETag")) |
526 | (define lmod (header-get headers #"Last-Modified")) | |
527 | (define lmod-curr (if lmod (rfc2822->epoch lmod) #f)) | |
528 | (define lmod-prev (if (file-exists? cached-lmod-path) | |
529 | (rfc2822->epoch (file->bytes cached-lmod-path)) | |
530 | #f)) | |
531 | (log-debug "lmod-curr:~v lmod-prev:~v" lmod-curr lmod-prev) | |
532 | (define cached? | |
533 | (or (and etag | |
534 | (file-exists? cached-etag-path) | |
535 | (bytes=? etag (file->bytes cached-etag-path)) | |
536 | (begin | |
537 | (log-debug "ETags match, skipping the rest of ~v" u-str) | |
538 | #t)) | |
539 | (and lmod-curr | |
540 | lmod-prev | |
541 | (<= lmod-curr lmod-prev) | |
542 | (begin | |
543 | (log-debug "Last-Modified <= current skipping the rest of ~v" u-str) | |
544 | #t)))) | |
545 | (if (not cached?) | |
68bbd2e9 SK |
546 | (begin |
547 | (log-debug | |
548 | "Downloading the rest of ~v. ETag: ~a, Last-Modified: ~v" | |
549 | u-str etag lmod) | |
550 | (make-parent-directory* cached-object-path) | |
551 | (make-parent-directory* cached-etag-path) | |
552 | (make-parent-directory* cached-lmod-path) | |
553 | (call-with-output-file cached-object-path | |
554 | (curry copy-port body-input) | |
555 | #:exists 'replace) | |
556 | (when etag | |
557 | (display-to-file etag cached-etag-path #:exists 'replace)) | |
558 | (when lmod | |
559 | (display-to-file lmod cached-lmod-path #:exists 'replace)) | |
560 | 'downloaded-new) | |
561 | 'skipped-cached)) | |
7fd20778 SK |
562 | |
563 | (: uri-download | |
564 | (-> Positive-Float Url | |
565 | (Result (U 'skipped-cached 'downloaded-new) | |
566 | Any))) ; TODO Maybe more-precise error type? | |
567 | (define (uri-download timeout u) | |
f65d6338 | 568 | (define u-str (url->string u)) |
f65d6338 SK |
569 | (define timeout-chan (make-channel)) |
570 | (define result-chan (make-channel)) | |
571 | (define timeout-thread | |
572 | (thread (λ () | |
573 | ; Doing this instead of sync/timeout to distinguish error values, | |
574 | ; rather than just have #f to work with. | |
575 | (sleep timeout) | |
7fd20778 | 576 | (channel-put timeout-chan '(error . timeout))))) |
f65d6338 SK |
577 | (define result-thread |
578 | (thread (λ () | |
579 | ; XXX We timeout getting a response, but body download could | |
580 | ; also take a long time and we might want to time that out as | |
581 | ; well, but then we may end-up with partially downloaded | |
582 | ; objects. But that could happen anyway if the server drops the | |
583 | ; connection for whatever reason. | |
584 | ; | |
585 | ; Maybe that is OK once we start treating the | |
586 | ; downloaded object as an addition to the stored set of | |
587 | ; messages, rather than the final set of messages. | |
588 | ||
589 | ; TODO message db | |
590 | ; - 1st try can just be an in-memory set that gets written-to | |
591 | ; and read-from disk as a whole. | |
592 | (define result | |
593 | (with-handlers | |
7fd20778 SK |
594 | ; TODO Maybe name each known errno? (exn:fail:network:errno-errno e) |
595 | ([exn:fail:network? | |
596 | (λ (e) `(error . (net-error . ,e)))] | |
597 | [exn? | |
598 | (λ (e) `(error . (other . ,e)))]) | |
f65d6338 SK |
599 | (define-values (status-line headers body-input) |
600 | (http-sendrecv/url | |
601 | u | |
602 | #:headers (list (format "User-Agent: ~a" user-agent)))) | |
7fd20778 | 603 | `(ok . ,(Resp status-line headers body-input)))) |
f65d6338 SK |
604 | (channel-put result-chan result)))) |
605 | (define result | |
606 | (sync timeout-chan | |
607 | result-chan)) | |
608 | (kill-thread result-thread) | |
609 | (kill-thread timeout-thread) | |
610 | (match result | |
7fd20778 SK |
611 | [(cons 'error _) |
612 | result] | |
f65d6338 SK |
613 | [(cons 'ok (Resp status-line headers body-input)) |
614 | (log-debug "headers: ~v" headers) | |
615 | (log-debug "status-line: ~v" status-line) | |
616 | (define status | |
617 | (string->number (second (string-split (bytes->string/utf-8 status-line))))) | |
618 | (log-debug "status: ~v" status) | |
7fd20778 SK |
619 | ; TODO Handle redirects. Should be within same timeout as req and body. |
620 | (let ([result | |
621 | (match status | |
622 | [200 | |
623 | `(ok . ,(uri-download-from-port u headers body-input))] | |
624 | [_ | |
625 | `(error . (http . ,status))])]) | |
626 | (close-input-port body-input) | |
627 | result)])) | |
4764ff89 | 628 | |
98529d3d | 629 | (: timeline-print (-> Out-Format (Listof Msg) Void)) |
b201e854 | 630 | (define (timeline-print out-format timeline) |
0cb1ae9c SK |
631 | (match timeline |
632 | ['() | |
633 | (void)] | |
634 | [(cons first-msg _) | |
635 | (void (foldl (match-lambda** | |
636 | [((and m (Msg _ _ from _ _)) (cons prev-from i)) | |
637 | (let ([i (if (peers-equal? prev-from from) i (+ 1 i))]) | |
638 | (msg-print out-format i m) | |
639 | (cons from i))]) | |
640 | (cons (Msg-from first-msg) 0) | |
641 | timeline))])) | |
4764ff89 | 642 | |
dbc26280 | 643 | (: peer->msgs (-> Peer (Listof Msg))) |
b056019b | 644 | (define (peer->msgs peer) |
e2840743 SK |
645 | (match-define (Peer nick uri uri-str _) peer) |
646 | (log-debug "Reading peer nick:~v uri:~v" nick uri-str) | |
fee11be9 | 647 | (define msgs-data (uri-read-cached uri)) |
4a23fd99 | 648 | ; TODO Expire cache |
fee11be9 | 649 | (if msgs-data |
0cb1ae9c | 650 | (str->msgs peer msgs-data) |
fee11be9 | 651 | '())) |
4214c0f3 | 652 | |
7fd20778 SK |
653 | (: peer-download |
654 | (-> Positive-Float Peer | |
655 | (Result (U 'skipped-cached 'downloaded-new) | |
656 | Any))) | |
f65d6338 | 657 | (define (peer-download timeout peer) |
e2840743 | 658 | (match-define (Peer nick uri u _) peer) |
7fd20778 SK |
659 | (log-info "Download BEGIN URL:~a" u) |
660 | (define-values (results _tm-cpu-ms tm-real-ms _tm-gc-ms) | |
f65d6338 | 661 | (time-apply uri-download (list timeout uri))) |
7fd20778 SK |
662 | (define result (car results)) |
663 | (log-info "Download END in ~a seconds, URL:~a, result:~s" | |
664 | (/ tm-real-ms 1000.0) | |
665 | u | |
666 | result) | |
667 | result) | |
4214c0f3 | 668 | |
e2840743 | 669 | (: timeline-download (-> Integer Positive-Float (Setof Peer) Void)) |
f65d6338 | 670 | (define (timeline-download num-workers timeout peers) |
7fd20778 SK |
671 | (define results |
672 | (concurrent-filter-map num-workers | |
673 | (λ (p) (cons p (peer-download timeout p))) | |
e2840743 | 674 | (set->list peers))) |
d54812ea SK |
675 | (define peers-ok |
676 | (filter-map (match-lambda | |
677 | [(cons p (cons 'ok _)) p] | |
678 | [(cons _ (cons 'error e)) #f]) | |
679 | results)) | |
680 | (define peers-err | |
681 | (filter-map (match-lambda | |
682 | [(cons _ (cons 'ok _)) | |
683 | #f] | |
684 | [(cons p (cons 'error e)) | |
685 | (struct-copy Peer p [comment (format "~s" e)])]) | |
686 | results)) | |
7fd20778 | 687 | (peers->file peers-ok (build-path tt-home-dir "peers-last-downloaded-ok")) |
7fd20778 | 688 | (peers->file peers-err (build-path tt-home-dir "peers-last-downloaded-err"))) |
9a6a9f9a | 689 | |
50f0609d | 690 | (: peers->timeline (-> (Setof Peer) (Listof Msg))) |
a60c484e | 691 | (define (peers->timeline peers) |
e2840743 | 692 | (append* (filter-map peer->msgs (set->list peers)))) |
a60c484e | 693 | |
e2840743 | 694 | (: timeline-sort (-> (Listof Msg) timeline-order (Listof Msgs))) |
a60c484e | 695 | (define (timeline-sort msgs order) |
a4899240 SK |
696 | (define cmp (match order |
697 | ['old->new <] | |
698 | ['new->old >])) | |
a60c484e SK |
699 | (sort msgs (λ (a b) (cmp (Msg-ts-epoch a) |
700 | (Msg-ts-epoch b))))) | |
4764ff89 | 701 | |
e2840743 | 702 | (: paths->peers (-> (Listof String) (Setof Peer))) |
d0a0e073 SK |
703 | (define (paths->peers paths) |
704 | (let* ([paths (match paths | |
705 | ['() | |
706 | (let ([peer-refs-file (build-path tt-home-dir "peers")]) | |
707 | (log-debug | |
708 | "No peer ref file paths provided, defaulting to ~v" | |
709 | (path->string peer-refs-file)) | |
710 | (list peer-refs-file))] | |
711 | [paths | |
712 | (log-debug "Peer ref file paths provided: ~v" paths) | |
713 | (map string->path paths)])] | |
50f0609d | 714 | [peers (apply peers-union (map file->peers paths))]) |
e2840743 SK |
715 | (log-info "Read-in ~a peers." (set-count peers)) |
716 | peers)) | |
d0a0e073 | 717 | |
50f0609d SK |
718 | (: cache-filename->peer (-> Path-String (Option Peer))) |
719 | (define (cache-filename->peer filename) | |
720 | (define nick #f) ; TODO Look it up in the nick-db when it exists. | |
721 | (define url-str (uri-decode (path->string filename))) ; TODO Can these crash? | |
722 | (match (str->url url-str) | |
723 | [#f #f] | |
724 | [url (Peer nick url url-str #f)])) | |
725 | ||
726 | (: peers-cached (-> (Setof Peer))) | |
727 | (define (peers-cached) | |
728 | ; TODO Expire cache? | |
729 | (make-immutable-peers | |
730 | (filter-map cache-filename->peer | |
731 | (directory-list cache-object-dir)))) | |
732 | ||
733 | (: peers-mentioned (-> (Listof Msg) (Setof Peer))) | |
734 | (define (peers-mentioned msgs) | |
e2840743 | 735 | (make-immutable-peers (append* (map Msg-mentions msgs)))) |
d3ac9e11 | 736 | |
56de6228 SK |
737 | (: log-writer-stop (-> Thread Void)) |
738 | (define (log-writer-stop log-writer) | |
739 | (log-message (current-logger) 'fatal 'stop "Exiting." #f) | |
740 | (thread-wait log-writer)) | |
741 | ||
0d3f753c SK |
742 | (: log-writer-start (-> Log-Level Thread)) |
743 | (define (log-writer-start level) | |
56de6228 SK |
744 | (let* ([logger |
745 | (make-logger #f #f level #f)] | |
746 | [log-receiver | |
747 | (make-log-receiver logger level)] | |
748 | [log-writer | |
749 | (thread | |
750 | (λ () | |
751 | (parameterize | |
752 | ([date-display-format 'iso-8601]) | |
753 | (let loop () | |
754 | (match-define (vector level msg _ topic) (sync log-receiver)) | |
755 | (unless (equal? topic 'stop) | |
756 | (eprintf "~a [~a] ~a~n" (date->string (current-date) #t) level msg) | |
757 | (loop))))))]) | |
758 | (current-logger logger) | |
759 | log-writer)) | |
01e4c499 | 760 | |
e8856d5c SK |
761 | (: crawl (-> Void)) |
762 | (define (crawl) | |
e2840743 | 763 | (let* ([peers-all-file |
e8856d5c SK |
764 | (build-path tt-home-dir "peers-all")] |
765 | [peers-mentioned-file | |
766 | (build-path tt-home-dir "peers-mentioned")] | |
767 | [peers-parsed-file | |
768 | (build-path tt-home-dir "peers-parsed")] | |
50f0609d SK |
769 | [peers-cached-file |
770 | (build-path tt-home-dir "peers-cached")] | |
771 | [peers-cached | |
772 | (peers-cached)] | |
773 | [cached-timeline | |
774 | (peers->timeline peers-cached)] | |
e8856d5c | 775 | [peers-mentioned-curr |
50f0609d | 776 | (peers-mentioned cached-timeline)] |
e8856d5c SK |
777 | [peers-mentioned-prev |
778 | (file->peers peers-mentioned-file)] | |
779 | [peers-mentioned | |
50f0609d SK |
780 | (peers-union peers-mentioned-prev |
781 | peers-mentioned-curr)] | |
e8856d5c SK |
782 | [peers-all-prev |
783 | (file->peers peers-all-file)] | |
784 | [peers-all | |
50f0609d SK |
785 | (peers-union peers-mentioned |
786 | peers-all-prev | |
787 | peers-cached)] | |
e8856d5c | 788 | [peers-discovered |
e2840743 SK |
789 | (set-subtract peers-all |
790 | peers-all-prev)] | |
e8856d5c | 791 | [peers-parsed |
e2840743 | 792 | (for/set ([p peers-all] #:when (> (length (peer->msgs p)) 0)) p)]) |
e8856d5c | 793 | ; TODO Deeper de-duping |
50f0609d | 794 | (log-info "Known peers cached ~a" (set-count peers-cached)) |
e2840743 SK |
795 | (log-info "Known peers mentioned: ~a" (set-count peers-mentioned)) |
796 | (log-info "Known peers parsed ~a" (set-count peers-parsed)) | |
797 | (log-info "Known peers total: ~a" (set-count peers-all)) | |
e8856d5c SK |
798 | (log-info "Discovered ~a new peers:~n~a" |
799 | (set-count peers-discovered) | |
800 | (pretty-format (map | |
e2840743 SK |
801 | (match-lambda |
802 | [(Peer n _ u c) (list n u c)]) | |
e8856d5c | 803 | (set->list peers-discovered)))) |
50f0609d SK |
804 | (peers->file peers-cached |
805 | peers-cached-file) | |
e8856d5c SK |
806 | (peers->file peers-mentioned |
807 | peers-mentioned-file) | |
808 | (peers->file peers-parsed | |
809 | peers-parsed-file) | |
810 | (peers->file peers-all | |
811 | peers-all-file))) | |
812 | ||
813 | (: read (-> (Listof String) Number Number Timeline-Order Out-Format Void)) | |
814 | (define (read file-paths ts-min ts-max order out-format) | |
815 | (let* ([peers | |
816 | (paths->peers file-paths)] | |
817 | [msgs | |
818 | (timeline-sort (peers->timeline peers) order)] | |
819 | [include? | |
820 | (λ (m) | |
821 | (and (or (not ts-min) (>= (Msg-ts-epoch m) ts-min)) | |
822 | (or (not ts-max) (<= (Msg-ts-epoch m) ts-max))))]) | |
823 | (timeline-print out-format (filter include? msgs)))) | |
824 | ||
825 | (: upload (-> Void)) | |
826 | (define (upload) | |
827 | ; FIXME Should not exit from here, but only after cleanup/logger-stoppage. | |
828 | (if (system (path->string (build-path tt-home-dir "hooks" "upload"))) | |
829 | (exit 0) | |
830 | (exit 1))) | |
831 | ||
832 | (: download (-> (Listof String) Positive-Integer Positive-Float Void)) | |
833 | (define (download file-paths num-workers timeout) | |
834 | (let ([peers (paths->peers file-paths)]) | |
835 | (define-values (_res _cpu real-ms _gc) | |
836 | (time-apply timeline-download (list num-workers timeout peers))) | |
837 | (log-info "Downloaded timelines from ~a peers in ~a seconds." | |
e2840743 | 838 | (set-count peers) |
e8856d5c SK |
839 | (/ real-ms 1000.0)))) |
840 | ||
841 | (: dispatch (-> String Void)) | |
842 | (define (dispatch command) | |
843 | (match command | |
844 | [(or "d" "download") | |
845 | ; Initially, 15 was fastest out of the tried: 1, 5, 10, 20. Then I | |
846 | ; started noticing significant slowdowns. Reducing to 5 seems to help. | |
847 | (let ([num-workers 5] | |
848 | [timeout 10.0]) | |
849 | (command-line | |
850 | #:program "tt download" | |
851 | #:once-each | |
852 | [("-j" "--jobs") | |
853 | njobs "Number of concurrent jobs." | |
854 | (set! num-workers (string->number njobs))] | |
855 | [("-t" "--timeout") | |
856 | seconds "Timeout seconds per request." | |
857 | (set! timeout (string->number seconds))] | |
858 | #:args file-paths | |
859 | (download file-paths num-workers timeout)))] | |
860 | [(or "u" "upload") | |
861 | (command-line | |
862 | #:program "tt upload" #:args () (upload))] | |
863 | [(or "r" "read") | |
864 | (let ([out-format 'multi-line] | |
865 | [order 'old->new] | |
866 | [ts-min #f] | |
867 | [ts-max #f]) | |
868 | (command-line | |
869 | #:program "tt read" | |
870 | #:once-each | |
871 | [("-r" "--rev") | |
872 | "Reverse displayed timeline order." | |
873 | (set! order 'new->old)] | |
874 | [("-m" "--min") | |
875 | m "Earliest time to display (ignore anything before it)." | |
876 | (set! ts-min (rfc3339->epoch m))] | |
877 | [("-x" "--max") | |
878 | x "Latest time to display (ignore anything after it)." | |
879 | (set! ts-max (rfc3339->epoch x))] | |
880 | #:once-any | |
881 | [("-s" "--short") | |
882 | "Short output format" | |
883 | (set! out-format 'single-line)] | |
884 | [("-l" "--long") | |
885 | "Long output format" | |
886 | (set! out-format 'multi-line)] | |
887 | #:args file-paths | |
888 | (read file-paths ts-min ts-max order out-format)))] | |
889 | [(or "c" "crawl") | |
890 | (command-line | |
891 | #:program "tt crawl" #:args () (crawl))] | |
892 | [command | |
893 | (eprintf "Error: invalid command: ~v\n" command) | |
894 | (eprintf "Please use the \"--help\" option to see a list of available commands.\n") | |
895 | (exit 1)])) | |
896 | ||
24c6a76b | 897 | (module+ main |
24f1f64b | 898 | (let ([log-level 'info]) |
c562bea3 | 899 | (command-line |
24f1f64b SK |
900 | #:program |
901 | "tt" | |
c562bea3 | 902 | #:once-each |
01e4c499 SK |
903 | [("-d" "--debug") |
904 | "Enable debug log level." | |
905 | (set! log-level 'debug)] | |
24f1f64b SK |
906 | #:help-labels |
907 | "" | |
908 | "and <command> is one of" | |
a60c484e | 909 | "r, read : Read the timeline (offline operation)." |
4214c0f3 | 910 | "d, download : Download the timeline." |
edadb804 | 911 | ; TODO Add path dynamically |
0868c39a | 912 | "u, upload : Upload your twtxt file (alias to execute ~/.tt/hooks/upload)." |
a60c484e | 913 | "c, crawl : Discover new peers mentioned by known peers (offline operation)." |
24f1f64b SK |
914 | "" |
915 | #:args (command . args) | |
0d3f753c | 916 | (define log-writer (log-writer-start log-level)) |
4214c0f3 | 917 | (current-command-line-arguments (list->vector args)) |
e8856d5c SK |
918 | ; TODO dispatch should return status with which we should exit after cleanups |
919 | (dispatch command) | |
0d3f753c | 920 | (log-writer-stop log-writer)))) |