(require openssl/sha1)
(require racket/date)
(require
+ net/head
net/http-client
net/uri-codec
net/url-string
(U 'old->new
'new->old))
-(struct msg
+(struct Msg
([ts-epoch : Integer]
[ts-orig : String]
- [nick : String]
+ [nick : (Option String)]
[uri : Url]
- [text : String])
- #:type-name Msg)
+ [text : String]
+ [mentions : (Listof Peer)]))
-(struct feed
- ([nick : String]
+(struct Peer
+ ([nick : (Option String)]
[uri : Url])
- #:type-name Feed)
+ #:transparent)
(: 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))))
+(: concurrent-filter-map (∀ (α β) (-> Natural (-> α β) (Listof α) (Listof β))))
(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
[n (vector-length colors)])
(λ (out-format color-i msg)
(let ([color (vector-ref colors (modulo color-i n))]
- [nick (msg-nick msg)]
- [uri (url->string (msg-uri msg))]
- [text (msg-text msg)])
+ [nick (Msg-nick msg)]
+ [uri (url->string (Msg-uri msg))]
+ [text (Msg-text msg)]
+ [mentions (Msg-mentions msg)])
(match out-format
['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))
- nick color text)]
+ (let ([nick (if nick nick uri)])
+ (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))
+ nick color text))]
['multi-line
- (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))
- (msg-ts-orig msg)
- nick uri color text)])))))
+ (let ([nick (if nick (string-append nick " ") "")])
+ (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))
+ (Msg-ts-orig msg)
+ nick uri color text))])))))
(: rfc3339->epoch (-> String (Option Nonnegative-Integer)))
(define rfc3339->epoch
(log-error "Invalid timestamp: ~v" ts)
#f]))))
-(: str->msg (-> String Url String (Option Msg)))
+(: str->msg (-> (Option String) Url String (Option Msg)))
(define str->msg
(let ([re (pregexp "^([^\\s\t]+)[\\s\t]+(.*)$")])
(λ (nick uri str)
[(list _wholething ts-orig text)
(let ([ts-epoch (rfc3339->epoch ts-orig)])
(if ts-epoch
- (msg ts-epoch ts-orig nick uri text)
+ (let ([mentions
+ (filter-map
+ (λ (m) (match (regexp-match #px"@<([^>]+)>" m)
+ [(list _wholething nick-uri)
+ (str->peer nick-uri)]))
+ (regexp-match* #px"@<[^\\s]+([\\s]+)?[^>]+>" text))])
+ (Msg ts-epoch ts-orig nick uri text mentions))
(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)
+ (log-debug "Non-msg line from nick:~v, line:~a" nick str)
#f])))))
(module+ test
+ ; TODO Test for when missing-nick case
(let* ([tzs (for*/list ([d '("-" "+")]
[h '("5" "05")]
[m '("00" ":00" "57" ":57")])
z)]
[m (str->msg n u (string-append ts sep txt))])
(check-not-false m)
- (check-equal? (msg-nick m) n)
- (check-equal? (msg-uri m) u)
- (check-equal? (msg-text m) txt)
- (check-equal? (msg-ts-orig m) ts (format "Given: ~v" ts))
+ (check-equal? (Msg-nick m) n)
+ (check-equal? (Msg-uri m) u)
+ (check-equal? (Msg-text m) txt)
+ (check-equal? (Msg-ts-orig m) ts (format "Given: ~v" ts))
)))
(let* ([ts "2020-11-18T22:22:09-0500"]
[nick "foo"]
[uri "bar"]
[actual (str->msg nick uri (string-append ts tab text))]
- [expected (msg 1605756129 ts nick uri text)])
+ [expected (Msg 1605756129 ts nick uri text)])
(check-equal?
- (msg-ts-epoch actual)
- (msg-ts-epoch expected)
+ (Msg-ts-epoch actual)
+ (Msg-ts-epoch expected)
"str->msg ts-epoch")
(check-equal?
- (msg-ts-orig actual)
- (msg-ts-orig expected)
+ (Msg-ts-orig actual)
+ (Msg-ts-orig expected)
"str->msg ts-orig")
(check-equal?
- (msg-nick actual)
- (msg-nick expected)
+ (Msg-nick actual)
+ (Msg-nick expected)
"str->msg nick")
(check-equal?
- (msg-uri actual)
- (msg-uri expected)
+ (Msg-uri actual)
+ (Msg-uri expected)
"str->msg uri")
(check-equal?
- (msg-text actual)
- (msg-text expected)
+ (Msg-text actual)
+ (Msg-text expected)
"str->msg text")))
(: str->lines (-> String (Listof String)))
(module+ test
(check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi")))
-(: str->msgs (-> String Url String (Listof Msg)))
+(: str->msgs (-> (Option String) Url String (Listof Msg)))
(define (str->msgs nick uri str)
(filter-map (λ (line) (str->msg nick uri line)) (filter-comments (str->lines str))))
(: cache-dir Path-String)
(define cache-dir (build-path tt-home-dir "cache"))
+(define cache-object-dir (build-path cache-dir "objects"))
+
(: url->cache-file-path-v1 (-> Url Path-String))
(define (url->cache-file-path-v1 uri)
(define (hash-sha1 str) : (-> String String)
(define digest (sha1 in))
(close-input-port in)
digest)
- (build-path cache-dir (hash-sha1 (url->string uri))))
+ (build-path cache-object-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))))
+ (build-path cache-object-dir (uri-encode (url->string uri))))
+
+(define url->cache-object-path url->cache-file-path-v2)
+
+(define (url->cache-etag-path uri)
+ (build-path cache-dir "etags" (uri-encode (url->string uri))))
-(define url->cache-file-path url->cache-file-path-v2)
+(define (url->cache-lmod-path uri)
+ (build-path cache-dir "lmods" (uri-encode (url->string uri))))
; TODO Return Option
(: uri-read-cached (-> Url String))
(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)
- [(list nick u)
- (with-handlers*
- ([exn:fail?
- (λ (e)
- (log-error "Invalid URI: ~v, exn: ~v" u e)
- #f)])
- (feed nick (string->url u)))]
- [_
- (log-error "Invalid feed string: ~v" str)
- #f]))
+(: str->peer (String (Option Peer)))
+(define (str->peer str)
+ (log-debug "Parsing peer string: ~v" str)
+ (with-handlers*
+ ([exn:fail?
+ (λ (e)
+ (log-error "Invalid URI in string: ~v, exn: ~v" str e)
+ #f)])
+ (match (string-split str)
+ [(list u) (Peer #f (string->url u))]
+ [(list n u) (Peer n (string->url u))]
+ [_
+ (log-error "Invalid peer 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))))
+(: str->peers (-> String (Listof Peer)))
+(define (str->peers str)
+ (filter-map str->peer (filter-comments (str->lines str))))
-(: file->feeds (-> Path-String (Listof Feed)))
-(define (file->feeds filename)
- (str->feeds (file->string filename)))
+(: file->peers (-> Path-String (Listof Peer)))
+(define (file->peers file-path)
+ (if (file-exists? file-path)
+ (str->peers (file->string file-path))
+ (begin
+ (log-error "File does not exist: ~v" (path->string file-path))
+ '())))
(: user-agent String)
(define user-agent
([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-peer-file (build-path tt-home-dir "me")]
[user
- (if (file-exists? user-feed-file)
- (let ([user (first (file->feeds user-feed-file))])
- (format "+~a; @~a" (url->string (feed-uri user)) (feed-nick user)))
+ (if (file-exists? user-peer-file)
+ (match (first (file->peers user-peer-file))
+ [(Peer #f u) (format "+~a" (url->string u) )]
+ [(Peer n u) (format "+~a; @~a" (url->string u) n)])
(format "+~a" prog-uri))])
(format "~a/~a (~a)" prog-name prog-version user)))
+(: header-get (-> (Listof Bytes) Bytes (Option Bytes)))
+(define (header-get headers name)
+ (match (filter-map (curry extract-field name) headers)
+ [(list val) val]
+ [_ #f]))
+
(: 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)
+ (define cached-object-path (url->cache-object-path u))
+ (define cached-etag-path (url->cache-etag-path u))
+ (define cached-lmod-path (url->cache-lmod-path u))
+ (log-debug "uri-download ~v into ~v" u cached-object-path)
(match* ((url-scheme u) (url-host u) (url-port u))
[(s h p)
#:when (and s h)
(url->string (struct-copy url u [scheme #f] [host #f]))
#:ssl? ssl?
#:port (cond [p p] [ssl? 443] [else 80])
- #:headers (list (format "User-Agent: ~a" user-agent))
- ))
+ #:headers (list (format "User-Agent: ~a" user-agent))))
(log-debug "headers: ~v" headers)
(log-debug "status-line: ~v" status-line)
(define status
(string->number (second (string-split (bytes->string/utf-8 status-line)))))
(log-debug "status: ~v" status)
; TODO Handle redirects
- (if (= 200 status)
- (begin
- (make-parent-directory* cache-file-path)
- (call-with-output-file cache-file-path
- (curry copy-port body-input)
- #:exists 'replace))
- (raise status))]
+ (match status
+ [200
+ (let ([etag (header-get headers #"ETag")]
+ [lmod (header-get headers #"Last-Modified")])
+ (if (and etag
+ (file-exists? cached-etag-path)
+ (bytes=? etag (file->bytes cached-etag-path)))
+ (log-info "ETags match, skipping the rest of ~v" (url->string u))
+ (begin
+ (log-info
+ "Downloading the rest of ~v. ETag: ~a, Last-Modified: ~v"
+ (url->string u) etag lmod)
+ (make-parent-directory* cached-object-path)
+ (make-parent-directory* cached-etag-path)
+ (make-parent-directory* cached-lmod-path)
+ (call-with-output-file cached-object-path
+ (curry copy-port body-input)
+ #:exists 'replace)
+ (when etag
+ (display-to-file etag cached-etag-path #:exists 'replace))
+ (when lmod
+ (display-to-file etag cached-lmod-path #:exists 'replace))))
+ (close-input-port body-input))]
+ [_
+ (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))
- (let ([i (if (string=? prev-nick nick) i (+ 1 i))])
+ [((and m (Msg _ _ nick _ _ _)) (cons prev-nick i))
+ (let ([i (if (equal? prev-nick nick) i (+ 1 i))])
(msg-print out-format i m)
(cons nick i))])
(cons "" 0)
timeline)))
-(: feed->msgs (-> Feed (Listof Msg)))
-(define (feed->msgs f)
- (match-define (feed nick uri) f)
- (log-info "Reading feed nick:~a uri:~v" nick (url->string uri))
+(: peer->msgs (-> Peer (Listof Msg)))
+(define (peer->msgs f)
+ (match-define (Peer nick uri) f)
+ (log-info "Reading peer nick:~v 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)
+(: peer-download (-> Peer Void))
+(define (peer-download f)
+ (match-define (Peer nick uri) f)
(define u (url->string uri))
- (log-info "Downloading feed uri:~a" u)
+ (log-info "Downloading peer uri:~a" u)
(with-handlers
([exn:fail?
(λ (e)
- (log-error "Network error nick:~a uri:~v exn:~v" nick u e)
+ (log-error "Network error nick:~v uri:~v exn:~v" nick u e)
#f)]
[integer?
(λ (status)
- (log-error "HTTP error nick:~a uri:~a status:~a" nick u status)
+ (log-error "HTTP error nick:~v uri:~a status:~a" nick u status)
#f)])
(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)))
+ (log-info "Peer 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 Peer) Void))
+(define (timeline-download num-workers peers)
; TODO No need for map - can just iter
- (void (concurrent-filter-map num-workers feed-download feeds)))
+ (void (concurrent-filter-map num-workers peer-download peers)))
; TODO timeline contract : time-sorted list of messages
-(: timeline-read (-> Timeline-Order (Listof Feed) (Listof Msg)))
-(define (timeline-read order feeds)
+(: timeline-read (-> Timeline-Order (Listof Peer) (Listof Msg)))
+(define (timeline-read order peers)
(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)))))
+ (sort (append* (filter-map peer->msgs peers))
+ (λ (a b) (cmp (Msg-ts-epoch a) (Msg-ts-epoch b)))))
+
+(: paths->peers (-> (Listof String) (Listof Peer)))
+(define (paths->peers paths)
+ (let* ([paths (match paths
+ ['()
+ (let ([peer-refs-file (build-path tt-home-dir "peers")])
+ (log-debug
+ "No peer ref file paths provided, defaulting to ~v"
+ (path->string peer-refs-file))
+ (list peer-refs-file))]
+ [paths
+ (log-debug "Peer ref file paths provided: ~v" paths)
+ (map string->path paths)])]
+ [peers (append* (map file->peers paths))])
+ (log-info "Read-in ~a peers." (length peers))
+ peers))
(: 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)
+(: log-writer-start (-> Log-Level Thread))
+(define (log-writer-start level)
(let* ([logger
(make-logger #f #f level #f)]
[log-receiver
"u, upload : Upload your twtxt file (alias to execute ~/.tt/upload)."
""
#:args (command . args)
- (define log-writer (logger-start log-level))
+ (define log-writer (log-writer-start log-level))
(current-command-line-arguments (list->vector args))
(match command
[(or "d" "download")
[("-j" "--jobs")
njobs "Number of concurrent jobs."
(set! num-workers (string->number njobs))]
- #:args (filename)
+ #:args file-paths
(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)))]
+ (time-apply timeline-download (list num-workers (paths->peers file-paths))))
+ (log-info "Timeline downloaded in ~a seconds." (/ real-ms 1000.0))))]
[(or "u" "upload")
(command-line
#:program
[("-l" "--long")
"Long output format"
(set! out-format 'multi-line)]
- #:args (filename)
- (timeline-print out-format (timeline-read order (file->feeds filename)))))]
- ))))
+ #:args file-paths
+ (timeline-print out-format (timeline-read order (paths->peers file-paths)))))])
+ (log-writer-stop log-writer))))
home / code / tt.git / blobdiff