-#lang racket
+#lang typed/racket/no-check
(require openssl/sha1)
(require racket/date)
-(require (prefix-in info: setup/getinfo))
(require
net/http-client
+ net/uri-codec
net/url-string
net/url-structs)
+(require (prefix-in info: "info.rkt"))
+
(module+ test
(require rackunit))
+(define-type Url
+ net/url-structs:url)
+
+(define-type Out-Format
+ (U 'single-line
+ 'multi-line))
+
+(define-type Timeline-Order
+ (U 'old->new
+ 'new->old))
+
(struct msg
- (
- ts_epoch ; Integer
- ts_rfc3339 ; String
- nick ; String
- uri ; net/url-structs:url
- text ; String
- ))
+ ([ts-epoch : Integer]
+ [ts-orig : String]
+ [nick : String]
+ [uri : Url]
+ [text : String])
+ #:type-name Msg)
(struct feed
- (
- nick ; String
- uri ; net/url-structs:url
- ))
+ ([nick : String]
+ [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)
(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
[(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)
[expected (sort ( filter-map f xs) <)])
(check-equal? actual expected "concurrent-filter-map")))
+(: msg-print (-> Out-Format Integer Msg Void))
(define msg-print
(let* ([colors (vector 36 33)]
[n (vector-length colors)])
['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?
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])))))
(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"]
[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)
(msg-text expected)
"str->msg text")))
+(: str->lines (-> String (Listof String)))
(define (str->lines str)
(string-split str (regexp "[\r\n]+")))
(module+ test
(check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi")))
+(: 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))))
-(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"))
-(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))
"")))
+(: str->feed (String (Option Feed)))
(define (str->feed str)
(log-debug "Parsing feed string: ~v" str)
(match (string-split str)
(log-error "Invalid feed string: ~v" str)
#f]))
+(: filter-comments (-> (Listof String) (Listof String)))
(define (filter-comments lines)
(filter-not (λ (line) (string-prefix? line "#")) lines))
+(: str->feeds (-> String (Listof Feed)))
(define (str->feeds str)
(filter-map str->feed (filter-comments (str->lines str))))
+(: file->feeds (-> Path-String (Listof Feed)))
(define (file->feeds filename)
(str->feeds (file->string filename)))
+(: user-agent String)
(define user-agent
(let*
([prog-name "tt"]
- [prog-version ((info:get-info (list prog-name)) 'version)]
+ [prog-version (info:#%info-lookup 'version)]
[prog-uri "https://github.com/xandkar/tt"]
[user-feed-file (expand-user-path "~/twtxt-me.txt")]
[user
(format "+~a" prog-uri))])
(format "~a/~a (~a)" prog-name prog-version user)))
-; uri-download : net/url-structs:url -> Void
+(: uri-download (-> Url Void))
(define (uri-download u)
(define cache-file-path (url->cache-file-path u))
(log-debug "uri-download ~v into ~v" u cache-file-path)
(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)]))
+(: timeline-print (-> Out-Format (Listof Msg) Void))
(define (timeline-print out-format timeline)
(void (foldl (match-lambda**
[((and m (msg _ _ nick _ _)) (cons prev-nick i))
(cons "" 0)
timeline)))
-; feed->msgs : Feed -> (Listof Msg)
+(: 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
+(: 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)
+(: timeline-download (-> Integer (Listof Feed) Void))
+(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 : (U 'old->new 'new->old) -> (Listof Feeds) -> (Listof Msg)
+(: timeline-read (-> Timeline-Order (Listof Feed) (Listof Msg)))
(define (timeline-read order feeds)
(define cmp (match order
['old->new <]
['new->old >]))
(sort (append* (filter-map feed->msgs feeds))
- (λ (a b) (cmp (msg-ts_epoch a) (msg-ts_epoch b)))))
+ (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)])
"and <command> 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)
(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))))
+ ; TODO Sync with logger before exit
+ (eprintf "Timeline downloaded in ~a seconds.~n" (/ real-ms 1000.0))
+ ))]
[(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")