X-Git-Url: https://git.xandkar.net/?a=blobdiff_plain;f=tt.rkt;h=01dbc6b3545a1f6996f58838583bc514609a1198;hb=a78e83b822eb7c4ab7cec791c50a93a80d5ff954;hp=43ec5702790899698f923f50608c3a5d7ea29afe;hpb=98529d3df4ecd4496571916d6abd9c81fd077f94;p=tt.git diff --git a/tt.rkt b/tt.rkt index 43ec570..01dbc6b 100644 --- a/tt.rkt +++ b/tt.rkt @@ -4,6 +4,7 @@ (require racket/date) (require net/http-client + net/uri-codec net/url-string net/url-structs) @@ -24,8 +25,8 @@ 'new->old)) (struct msg - ([ts_epoch : Integer] - [ts_rfc3339 : String] + ([ts-epoch : Integer] + [ts-orig : String] [nick : String] [uri : Url] [text : String]) @@ -36,13 +37,18 @@ [uri : Url]) #:type-name Feed) -(define (concurrent-filter-map num_workers f xs) +(: tt-home-dir Path-String) +(define tt-home-dir (build-path (expand-user-path "~") ".tt")) + +(: concurrent-filter-map (∀ (α β) (-> Natural (-> α β) (Listof α)))) +(define (concurrent-filter-map num-workers f xs) ; TODO preserve order of elements OR communicate that reorder is expected ; TODO switch from mailboxes to channels (define (make-worker id f) (define parent (current-thread)) (λ () - (define self (current-thread)) + (define self : Thread (current-thread)) + (: work (∀ (α) (-> α))) (define (work) (thread-send parent (cons 'next self)) (match (thread-receive) @@ -52,6 +58,7 @@ (when y (thread-send parent (cons 'result y))) (work))])) (work))) + (: dispatch (∀ (α β) (-> (Listof Nonnegative-Integer) (Listof α) (Listof β)))) (define (dispatch ws xs ys) (if (empty? ws) ys @@ -65,7 +72,7 @@ [(cons x xs) (begin (thread-send thd (cons 'unit x)) (dispatch ws xs ys))])]))) - (define workers (range num_workers)) + (define workers (range num-workers)) (define threads (map (λ (id) (thread (make-worker id f))) workers)) (define results (dispatch workers xs '())) (for-each thread-wait threads) @@ -91,18 +98,52 @@ ['single-line (printf "~a \033[1;37m<~a>\033[0m \033[0;~am~a\033[0m~n" (parameterize ([date-display-format 'iso-8601]) - (date->string (seconds->date [msg-ts_epoch msg]) #t)) + (date->string (seconds->date [msg-ts-epoch msg]) #t)) nick color text)] ['multi-line - (printf "~a~n\033[1;37m<~a ~a>\033[0m~n\033[0;~am~a\033[0m~n~n" + (printf "~a (~a)~n\033[1;37m<~a ~a>\033[0m~n\033[0;~am~a\033[0m~n~n" (parameterize ([date-display-format 'rfc2822]) - (date->string (seconds->date [msg-ts_epoch msg]) #t)) + (date->string (seconds->date [msg-ts-epoch msg]) #t)) + (msg-ts-orig msg) nick uri color text)]))))) +(: rfc3339->epoch (-> String (Option Nonnegative-Integer))) +(define rfc3339->epoch + (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}))?$")]) + (λ (ts) + (match (regexp-match re ts) + [(list _wholething yyyy mm dd HH MM _:SS SS _fractional tz-whole tz-sign tz-HH tz-MM) + (let* + ([tz-offset + (match* (tz-whole tz-sign tz-HH tz-MM) + [("Z" #f #f #f) + 0] + [(_ (or "-" "+") (? identity h) (? identity m)) + (let ([h (string->number h)] + [m (string->number m)] + ; Reverse to get back to UTC: + [op (match tz-sign ["+" -] ["-" +])]) + (op 0 (+ (* 60 m) (* 60 (* 60 h)))))] + [(a b c d) + (log-warning "Impossible TZ string: ~v, components: ~v ~v ~v ~v" tz-whole a b c d) + 0])] + [ts-orig ts] + [local-time? #f] + [ts-epoch (find-seconds (if SS (string->number SS) 0) + (string->number MM) + (string->number HH) + (string->number dd) + (string->number mm) + (string->number yyyy) + local-time?)]) + (+ ts-epoch tz-offset))] + [_ + (log-error "Invalid timestamp: ~v" ts) + #f])))) + (: str->msg (-> String Url String (Option Msg))) (define str->msg - ; TODO Split parsing into 2 stages: 1) line->list; 2) rfc3339->epoch. - (let ([re (pregexp "^(([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2})(:([0-9]{2}))?)(\\.[0-9]+)?([^\\s\t]*)[\\s\t]+(.*)$")]) + (let ([re (pregexp "^([^\\s\t]+)[\\s\t]+(.*)$")]) (λ (nick uri str) (with-handlers* ([exn:fail? @@ -112,17 +153,15 @@ str nick (url->string uri) e) #f)]) (match (regexp-match re str) - [(list _wholething ts yyyy mm dd HH MM _:SS SS _f tz text) - (let* - ; TODO handle tz offset - ([ts_rfc3339 (string-append ts (if SS "" ":00") (if tz tz ""))] - [ts_epoch (find-seconds (if SS (string->number SS) 0) - (string->number MM) - (string->number HH) - (string->number dd) - (string->number mm) - (string->number yyyy))]) - (msg ts_epoch ts_rfc3339 nick uri text))] + [(list _wholething ts-orig text) + (let ([ts-epoch (rfc3339->epoch ts-orig)]) + (if ts-epoch + (msg ts-epoch ts-orig nick uri text) + (begin + (log-error + "Msg rejected due to invalid timestamp: ~v, nick:~v, uri:~v" + str nick (url->string uri)) + #f)))] [_ (log-debug "Non-msg line from nick:~a, line:~a" nick str) #f]))))) @@ -148,7 +187,7 @@ (check-equal? (msg-nick m) n) (check-equal? (msg-uri m) u) (check-equal? (msg-text m) txt) - (check-equal? (msg-ts_rfc3339 m) ts (format "Given: ~v" ts)) + (check-equal? (msg-ts-orig m) ts (format "Given: ~v" ts)) ))) (let* ([ts "2020-11-18T22:22:09-0500"] @@ -158,15 +197,14 @@ [uri "bar"] [actual (str->msg nick uri (string-append ts tab text))] [expected (msg 1605756129 ts nick uri text)]) - ; FIXME re-enable after handling tz offset - ;(check-equal? - ; (msg-ts_epoch actual) - ; (msg-ts_epoch expected) - ; "str->msg ts_epoch") (check-equal? - (msg-ts_rfc3339 actual) - (msg-ts_rfc3339 expected) - "str->msg ts_rfc3339") + (msg-ts-epoch actual) + (msg-ts-epoch expected) + "str->msg ts-epoch") + (check-equal? + (msg-ts-orig actual) + (msg-ts-orig expected) + "str->msg ts-orig") (check-equal? (msg-nick actual) (msg-nick expected) @@ -189,26 +227,35 @@ (: str->msgs (-> String Url String (Listof Msg))) (define (str->msgs nick uri str) - (filter-map (λ (line) (str->msg nick uri line)) (str->lines str))) + (filter-map (λ (line) (str->msg nick uri line)) (filter-comments (str->lines str)))) -(: hash-sha1 (-> String String)) -(define (hash-sha1 str) - (define in (open-input-string str)) - (define digest (sha1 in)) - (close-input-port in) - digest) +(: cache-dir Path-String) +(define cache-dir (build-path tt-home-dir "cache")) -(: url->cache-file-path (-> Url Path-String)) -(define (url->cache-file-path uri) - ; TODO Replace hashing with encoding - (expand-user-path (string-append "~/.tt/cache/" (hash-sha1 (url->string uri))))) +(: url->cache-file-path-v1 (-> Url Path-String)) +(define (url->cache-file-path-v1 uri) + (define (hash-sha1 str) : (-> String String) + (define in (open-input-string str)) + (define digest (sha1 in)) + (close-input-port in) + digest) + (build-path cache-dir (hash-sha1 (url->string uri)))) + +(: url->cache-file-path-v2 (-> Url Path-String)) +(define (url->cache-file-path-v2 uri) + (build-path cache-dir (uri-encode (url->string uri)))) + +(define url->cache-file-path url->cache-file-path-v2) ; TODO Return Option (: uri-read-cached (-> Url String)) (define (uri-read-cached uri) - (define path (url->cache-file-path uri)) - (if (file-exists? path) - (file->string path) + (define path-v1 (url->cache-file-path-v1 uri)) + (define path-v2 (url->cache-file-path-v2 uri)) + (when (file-exists? path-v1) + (rename-file-or-directory path-v1 path-v2 #t)) + (if (file-exists? path-v2) + (file->string path-v2) (begin (log-warning "Cache file not found for URI: ~a" (url->string uri)) ""))) @@ -246,7 +293,7 @@ ([prog-name "tt"] [prog-version (info:#%info-lookup 'version)] [prog-uri "https://github.com/xandkar/tt"] - [user-feed-file (expand-user-path "~/twtxt-me.txt")] + [user-feed-file (build-path tt-home-dir "me")] [user (if (file-exists? user-feed-file) (let ([user (first (file->feeds user-feed-file))]) @@ -278,10 +325,11 @@ (log-debug "status: ~v" status) ; TODO Handle redirects (if (= 200 status) - (call-with-output-file cache-file-path - (λ (cache-output) - (copy-port body-input cache-output)) - #:exists 'replace) + (begin + (make-parent-directory* cache-file-path) + (call-with-output-file cache-file-path + (curry copy-port body-input) + #:exists 'replace)) (raise status))] [(_ _ _) (log-error "Invalid URI: ~v" u)])) @@ -299,28 +347,31 @@ (: feed->msgs (-> Feed (Listof Msg))) (define (feed->msgs f) (match-define (feed nick uri) f) - (log-info "Reading feed nick:~a uri:~v" nick uri) + (log-info "Reading feed nick:~a uri:~v" nick (url->string uri)) (str->msgs nick uri (uri-read-cached uri))) (: feed-download (-> Feed Void)) (define (feed-download f) (match-define (feed nick uri) f) - (log-info "Downloading feed nick:~a uri:~a" nick (url->string uri)) + (define u (url->string uri)) + (log-info "Downloading feed uri:~a" u) (with-handlers ([exn:fail? (λ (e) - (log-error "Network error nick:~a uri:~v exn:~v" nick uri e) + (log-error "Network error nick:~a uri:~v exn:~v" nick u e) #f)] [integer? (λ (status) - (log-error "HTTP error nick:~a uri:~a status:~a" nick uri status) + (log-error "HTTP error nick:~a uri:~a status:~a" nick u status) #f)]) - (uri-download uri))) + (define-values (_result _tm-cpu-ms tm-real-ms _tm-gc-ms) + (time-apply uri-download (list uri))) + (log-info "Feed downloaded in ~a seconds, uri: ~a" (/ tm-real-ms 1000.0) u))) (: timeline-download (-> Integer (Listof Feed) Void)) -(define (timeline-download num_workers feeds) +(define (timeline-download num-workers feeds) ; TODO No need for map - can just iter - (void (concurrent-filter-map num_workers feed-download feeds))) + (void (concurrent-filter-map num-workers feed-download feeds))) ; TODO timeline contract : time-sorted list of messages (: timeline-read (-> Timeline-Order (Listof Feed) (Listof Msg))) @@ -329,23 +380,31 @@ ['old->new <] ['new->old >])) (sort (append* (filter-map feed->msgs feeds)) - (λ (a b) (cmp (msg-ts_epoch a) (msg-ts_epoch b))))) - -(: start-logger (-> Log-Level Void)) -(define (start-logger level) - (let* ([logger (make-logger #f #f level #f)] - [log-receiver (make-log-receiver logger level)]) - (void (thread (λ () - (parameterize - ([date-display-format 'iso-8601]) - (let loop () - (define data (sync log-receiver)) - (define level (vector-ref data 0)) - (define msg (vector-ref data 1)) - (define ts (date->string (current-date) #t)) - (eprintf "~a [~a] ~a~n" ts level msg) - (loop)))))) - (current-logger logger))) + (λ (a b) (cmp (msg-ts-epoch a) (msg-ts-epoch b))))) + +(: log-writer-stop (-> Thread Void)) +(define (log-writer-stop log-writer) + (log-message (current-logger) 'fatal 'stop "Exiting." #f) + (thread-wait log-writer)) + +(: logger-start (-> Log-Level Thread)) +(define (logger-start level) + (let* ([logger + (make-logger #f #f level #f)] + [log-receiver + (make-log-receiver logger level)] + [log-writer + (thread + (λ () + (parameterize + ([date-display-format 'iso-8601]) + (let loop () + (match-define (vector level msg _ topic) (sync log-receiver)) + (unless (equal? topic 'stop) + (eprintf "~a [~a] ~a~n" (date->string (current-date) #t) level msg) + (loop))))))]) + (current-logger logger) + log-writer)) (module+ main (let ([log-level 'info]) @@ -361,29 +420,35 @@ "and is one of" "r, read i : Read the timeline." "d, download : Download the timeline." + ; TODO Add path dynamically "u, upload : Upload your twtxt file (alias to execute ~/.tt/upload)." "" #:args (command . args) - (start-logger log-level) + (define log-writer (logger-start log-level)) (current-command-line-arguments (list->vector args)) (match command [(or "d" "download") - (let ([num_workers 15]) ; 15 was fastest out of the tried: 1, 5, 10, 20. + ; Initially, 15 was fastest out of the tried: 1, 5, 10, 20. Then I + ; started noticing significant slowdowns. Reducing to 5 seems to help. + (let ([num-workers 5]) (command-line #:program "tt download" #:once-each [("-j" "--jobs") njobs "Number of concurrent jobs." - (set! num_workers (string->number njobs))] + (set! num-workers (string->number njobs))] #:args (filename) - (timeline-download num_workers (file->feeds filename))))] + (define-values (_res _cpu real-ms _gc) + (time-apply timeline-download (list num-workers (file->feeds filename)))) + (log-info "Timeline downloaded in ~a seconds." (/ real-ms 1000.0)) + (log-writer-stop log-writer)))] [(or "u" "upload") (command-line - #:program - "tt upload" - #:args () - (if (system (path->string (expand-user-path "~/.tt/upload"))) + #:program + "tt upload" + #:args () + (if (system (path->string (build-path tt-home-dir "upload"))) (exit 0) (exit 1)))] [(or "r" "read")