(struct Msg
([ts-epoch : Integer]
[ts-orig : String]
- [nick : (Option String)]
- [uri : Url]
+ [from : Peer]
[text : String]
[mentions : (Listof Peer)]))
(struct Peer
([nick : (Option String)]
[uri : Url]
+ [uri-str : String]
[comment : (Option String)])
#:transparent)
[body-input : Input-Port])
#:transparent)
+(: peers-equal? (-> Peer Peer Boolean))
+(define (peers-equal? p1 p2)
+ (equal? (Peer-uri-str p1)
+ (Peer-uri-str p2)))
+
+(: peer-hash (-> Peer Fixnum))
+(define (peer-hash p)
+ (equal-hash-code (Peer-uri-str p)))
+
+(define-custom-set-types peers
+ #:elem? Peer?
+ peers-equal?
+ peer-hash)
+; XXX Without supplying above explicit hash procedure, we INTERMITTENTLY get
+; the following contract violations:
+;
+; custom-elem-contents: contract violation
+; expected: custom-elem?
+; given: #f
+; context...:
+; /usr/share/racket/collects/racket/private/set-types.rkt:104:0: custom-set->list
+; /home/siraaj/proj/pub/tt/tt.rkt:716:0: crawl
+; /usr/share/racket/collects/racket/cmdline.rkt:191:51
+; body of (submod "/home/siraaj/proj/pub/tt/tt.rkt" main)
+;
+; TODO Investigate why and make a minimal reproducible test case.
+
+(define (peers-union . peer-sets)
+ (define groups
+ (foldl
+ (λ (p groups)
+ (hash-update groups (Peer-uri-str p) (λ (group) (cons p group)) '()))
+ (hash)
+ (append* (map set->list peer-sets))))
+ (define (merge peers)
+ (match peers
+ ['() (raise 'impossible)]
+ [(list p) p]
+ [(list* p1 p2 ps)
+ (let* ([n1 (Peer-nick p1)]
+ [n2 (Peer-nick p2)]
+ [p (cond
+ [(and (not n1) (not n2)) p1]
+ [(and n1 n2 ) p1]
+ [(and n1 (not n2)) p1]
+ [(and (not n1) n2) p2]
+ [else
+ (raise 'impossible)])])
+ (merge (cons p ps)))]))
+ (make-immutable-peers (map merge (hash-values groups))))
+
+(module+ test
+ (let* ([u1 "http://foo/bar"]
+ [u2 "http://baz/quux"]
+ [p1 (Peer #f (string->url u1) u1 #f)]
+ [p2 (Peer "a" (string->url u1) u1 #f)]
+ [p3 (Peer "b" (string->url u2) u2 #f)]
+ [s1 (make-immutable-peers (list p1))]
+ [s2 (make-immutable-peers (list p2 p3))])
+ (check-true (peers? (peers-union s1 s2)))
+ (check-true (peers? (peers-union s2 s1)))
+ (check-equal? (list p3 p2) (set->list (peers-union s1 s2)))
+ (check-equal? (list p3 p2) (set->list (peers-union s2 s1)))))
+
(: tt-home-dir Path-String)
(define tt-home-dir (build-path (expand-user-path "~") ".tt"))
[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)]
- [mentions (Msg-mentions msg)])
+ [nick (Peer-nick (Msg-from msg))]
+ [uri (Peer-uri-str (Msg-from msg))]
+ [text (Msg-text msg)])
(match out-format
['single-line
(let ([nick (if nick nick uri)])
(log-debug "Invalid timestamp: ~v" ts)
#f]))))
-(: str->msg (-> (Option String) Url String (Option Msg)))
+(: str->msg (-> Peer String (Option Msg)))
(define str->msg
(let ([re (pregexp "^([^\\s\t]+)[\\s\t]+(.*)$")])
- (λ (nick uri str)
+ (λ (from str)
+ (define from-str (peer->str from))
(define str-head (substring str 0 (min 100 (string-length str))))
(with-handlers*
([exn:fail?
(λ (e)
(log-debug
"Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v"
- str-head nick (url->string uri) e)
+ str-head from-str e)
#f)])
(match (regexp-match re str)
[(list _wholething ts-orig text)
[(list _wholething nick-uri)
(str->peer nick-uri)]))
(regexp-match* #px"@<[^\\s]+([\\s]+)?[^>]+>" text))])
- (Msg ts-epoch ts-orig nick uri text mentions))
+ (Msg ts-epoch ts-orig from text mentions))
(begin
(log-debug
- "Msg rejected due to invalid timestamp: ~v, nick:~v, uri:~v"
- str-head nick (url->string uri))
+ "Msg rejected due to invalid timestamp. From:~v. Line:~v"
+ from-str str-head)
#f)))]
[_
- (log-debug "Non-msg line from nick:~v, line:~a" nick str-head)
+ (log-debug "Non-msg line. From:~v. Line:~v" from-str str-head)
#f])))))
(module+ test
(string-append d h m))]
[tzs (list* "" "Z" tzs)])
(for* ([n '("fake-nick")]
- [u '("fake-uri")]
+ [u '("http://fake-uri")]
+ [p (list (Peer n (string->url u) u #f))]
[s '("" ":10")]
[f '("" ".1337")]
[z tzs]
(let* ([ts (string-append "2020-11-18T22:22"
(if (non-empty-string? s) s ":00")
z)]
- [m (str->msg n u (string-append ts sep txt))])
+ [m (str->msg p (string-append ts sep txt))])
(check-not-false m)
- (check-equal? (Msg-nick m) n)
- (check-equal? (Msg-uri m) u)
+ (check-equal? (Msg-from m) p)
(check-equal? (Msg-text m) txt)
(check-equal? (Msg-ts-orig m) ts (format "Given: ~v" ts))
)))
[tab " "]
[text "Lorem ipsum"]
[nick "foo"]
- [uri "bar"]
- [actual (str->msg nick uri (string-append ts tab text))]
- [expected (Msg 1605756129 ts nick uri text '())])
+ [uri "http://bar/"]
+ [peer (Peer nick (string->url uri) uri #f)]
+ [actual (str->msg peer (string-append ts tab text))]
+ [expected (Msg 1605756129 ts peer text '())])
(check-equal?
(Msg-ts-epoch actual)
(Msg-ts-epoch expected)
(Msg-ts-orig expected)
"str->msg ts-orig")
(check-equal?
- (Msg-nick actual)
- (Msg-nick expected)
+ (Peer-nick (Msg-from actual))
+ (Peer-nick (Msg-from expected))
"str->msg nick")
(check-equal?
- (Msg-uri actual)
- (Msg-uri expected)
+ (Peer-uri (Msg-from actual))
+ (Peer-uri (Msg-from expected))
"str->msg uri")
+ (check-equal?
+ (Peer-uri-str (Msg-from actual))
+ (Peer-uri-str (Msg-from expected))
+ "str->msg uri-str")
(check-equal?
(Msg-text actual)
(Msg-text expected)
(module+ test
(check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi")))
-(: 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))))
+(: str->msgs (-> Peer String (Listof Msg)))
+(define (str->msgs peer str)
+ (filter-map (λ (line) (str->msg peer line))
+ (filter-comments (str->lines str))))
(: cache-dir Path-String)
(define cache-dir (build-path tt-home-dir "cache"))
(define (url->cache-file-path-v2 uri)
(build-path cache-object-dir (uri-encode (url->string uri))))
-(define url->cache-object-path url->cache-file-path-v2)
-
-(: cache-object-filename->url (-> Path-String Url))
-(define (cache-object-filename->url name)
- (string->url (uri-decode (path->string name))))
+(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))))
([exn:fail? (λ (e) #f)])
(string->url s)))
-(: str->peer (String (Option Peer)))
+(: peer->str (-> Peer String))
+(define (peer->str peer)
+ (match-define (Peer n _ u c) peer)
+ (format "~a~a~a"
+ (if n (format "~a " n) "")
+ u
+ (if c (format " # ~a" c) "")))
+
+(: str->peer (-> String (Option Peer)))
(define (str->peer str)
(log-debug "Parsing peer string: ~v" str)
(match
[#f
(log-error "Invalid URI in peer string: ~v" str)
#f]
- [url (Peer nick url comment)])]
+ [url
+ (Peer nick url (url->string url) comment)])]
[_
(log-debug "Invalid peer string: ~v" str)
#f]))
(module+ test
(check-equal?
(str->peer "foo http://bar/file.txt # some rando")
- (Peer "foo" (str->url "http://bar/file.txt") "some rando"))
+ (Peer "foo" (str->url "http://bar/file.txt") "http://bar/file.txt" "some rando"))
(check-equal?
(str->peer "http://bar/file.txt # some rando")
- (Peer #f (str->url "http://bar/file.txt") "some rando"))
+ (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" "some rando"))
(check-equal?
(str->peer "http://bar/file.txt #")
- (Peer #f (str->url "http://bar/file.txt") ""))
+ (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" ""))
(check-equal?
(str->peer "http://bar/file.txt#") ; XXX URLs can have #s
- (Peer #f (str->url "http://bar/file.txt#") #f))
+ (Peer #f (str->url "http://bar/file.txt#") "http://bar/file.txt#" #f))
(check-equal?
(str->peer "http://bar/file.txt")
- (Peer #f (str->url "http://bar/file.txt") #f))
+ (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" #f))
(check-equal?
(str->peer "foo http://bar/file.txt")
- (Peer "foo" (str->url "http://bar/file.txt") #f))
+ (Peer "foo" (str->url "http://bar/file.txt") "http://bar/file.txt" #f))
(check-equal?
(str->peer "foo bar # baz")
#f)
(check-equal?
(str->peer "foo bar://baz # quux")
- (Peer "foo" (str->url "bar://baz") "quux"))
+ (Peer "foo" (str->url "bar://baz") "bar://baz" "quux"))
(check-equal?
(str->peer "foo bar//baz # quux")
#f))
(define (filter-comments lines)
(filter-not (λ (line) (string-prefix? line "#")) lines))
-(: str->peers (-> String (Listof Peer)))
+(: str->peers (-> String (Setof Peer)))
(define (str->peers str)
- (filter-map str->peer (filter-comments (str->lines str))))
+ (make-immutable-peers (filter-map str->peer (filter-comments (str->lines str)))))
-(: peers->file (-> (Listof Peers) Path-String Void))
+(: peers->file (-> (Setof Peers) Path-String Void))
(define (peers->file peers path)
(display-lines-to-file
- (map (match-lambda
- [(Peer n u c)
- (format "~a~a~a"
- (if n (format "~a " n) "")
- (url->string u)
- (if c (format " # ~a" c) ""))])
- peers)
+ (map peer->str
+ (sort (set->list peers)
+ (match-lambda**
+ [((Peer n1 _ _ _) (Peer n2 _ _ _))
+ (string<? (if n1 n1 "")
+ (if n2 n2 ""))])))
path
#:exists 'replace))
-(: file->peers (-> Path-String (Listof Peer)))
+(: file->peers (-> Path-String (Setof Peer)))
(define (file->peers file-path)
(if (file-exists? file-path)
(str->peers (file->string file-path))
(begin
(log-warning "File does not exist: ~v" (path->string file-path))
- '())))
+ (make-immutable-peers))))
(define re-rfc2822
#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")
[user-peer-file (build-path tt-home-dir "me")]
[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)])
+ (match (set-first (file->peers user-peer-file))
+ [(Peer #f _ u _) (format "+~a" u )]
+ [(Peer n _ u _) (format "+~a; @~a" u n)])
(format "+~a" prog-uri))])
(format "~a/~a (~a)" prog-name prog-version user)))
(: 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 (equal? prev-nick nick) i (+ 1 i))])
- (msg-print out-format i m)
- (cons nick i))])
- (cons "" 0)
- timeline)))
+ (match timeline
+ ['()
+ (void)]
+ [(cons first-msg _)
+ (void (foldl (match-lambda**
+ [((and m (Msg _ _ from _ _)) (cons prev-from i))
+ (let ([i (if (peers-equal? prev-from from) i (+ 1 i))])
+ (msg-print out-format i m)
+ (cons from i))])
+ (cons (Msg-from first-msg) 0)
+ timeline))]))
(: peer->msgs (-> Peer (Listof Msg)))
(define (peer->msgs peer)
- (match-define (Peer nick uri _) peer)
- (log-debug "Reading peer nick:~v uri:~v" nick (url->string uri))
+ (match-define (Peer nick uri uri-str _) peer)
+ (log-debug "Reading peer nick:~v uri:~v" nick uri-str)
(define msgs-data (uri-read-cached uri))
; TODO Expire cache
(if msgs-data
- (str->msgs nick uri msgs-data)
+ (str->msgs peer msgs-data)
'()))
(: peer-download
(Result (U 'skipped-cached 'downloaded-new)
Any)))
(define (peer-download timeout peer)
- (match-define (Peer nick uri _) peer)
- (define u (url->string uri))
+ (match-define (Peer nick uri u _) peer)
(log-info "Download BEGIN URL:~a" u)
(define-values (results _tm-cpu-ms tm-real-ms _tm-gc-ms)
(time-apply uri-download (list timeout uri)))
result)
result)
-(: timeline-download (-> Integer Positive-Float (Listof Peer) Void))
+(: timeline-download (-> Integer Positive-Float (Setof Peer) Void))
(define (timeline-download num-workers timeout peers)
(define results
(concurrent-filter-map num-workers
(λ (p) (cons p (peer-download timeout p)))
- peers))
+ (set->list peers)))
(define peers-ok
(filter-map (match-lambda
[(cons p (cons 'ok _)) p]
(peers->file peers-ok (build-path tt-home-dir "peers-last-downloaded-ok"))
(peers->file peers-err (build-path tt-home-dir "peers-last-downloaded-err")))
-(: uniq (∀ (α) (-> (Listof α) (Listof α))))
-(define (uniq xs)
- (set->list (list->set xs)))
-
-(: peers->timeline (-> (listof Peer) (listof Msg)))
+(: peers->timeline (-> (Setof Peer) (Listof Msg)))
(define (peers->timeline peers)
- (append* (filter-map peer->msgs peers)))
+ (append* (filter-map peer->msgs (set->list peers))))
-(: timeline-sort (-> (listof Msg) timeline-order (Listof Msgs)))
+(: timeline-sort (-> (Listof Msg) timeline-order (Listof Msgs)))
(define (timeline-sort msgs order)
(define cmp (match order
['old->new <]
(sort msgs (λ (a b) (cmp (Msg-ts-epoch a)
(Msg-ts-epoch b)))))
-(: paths->peers (-> (Listof String) (Listof Peer)))
+(: paths->peers (-> (Listof String) (Setof Peer)))
(define (paths->peers paths)
(let* ([paths (match paths
['()
[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))
- (uniq peers)))
-
-(: mentioned-peers-in-cache (-> (Listof Peer)))
-(define (mentioned-peers-in-cache)
- ; TODO Expire cache
- (define msgs
- (append* (map (λ (filename)
- (define path (build-path cache-object-dir filename))
- (define size (/ (file-size path) 1000000.0))
- (log-debug "BEGIN parsing ~a MB from file: ~v"
- size
- (path->string path))
- (define t0 (current-inexact-milliseconds))
- (define m (filter-map
- (λ (line)
- (str->msg #f (cache-object-filename->url filename) line))
- (filter-comments
- (file->lines path))))
- (define t1 (current-inexact-milliseconds))
- (log-debug "END parsing ~a MB in ~a seconds from file: ~v."
- size
- (* 0.001 (- t1 t0))
- (path->string path))
- (when (empty? m)
- (log-debug "No messages found in ~a" (path->string path)))
- m)
- (directory-list cache-object-dir))))
- (uniq (append* (map Msg-mentions msgs))))
+ [peers (apply peers-union (map file->peers paths))])
+ (log-info "Read-in ~a peers." (set-count peers))
+ peers))
+
+(: cache-filename->peer (-> Path-String (Option Peer)))
+(define (cache-filename->peer filename)
+ (define nick #f) ; TODO Look it up in the nick-db when it exists.
+ (define url-str (uri-decode (path->string filename))) ; TODO Can these crash?
+ (match (str->url url-str)
+ [#f #f]
+ [url (Peer nick url url-str #f)]))
+
+(: peers-cached (-> (Setof Peer)))
+(define (peers-cached)
+ ; TODO Expire cache?
+ (make-immutable-peers
+ (filter-map cache-filename->peer
+ (directory-list cache-object-dir))))
+
+(: peers-mentioned (-> (Listof Msg) (Setof Peer)))
+(define (peers-mentioned msgs)
+ (make-immutable-peers (append* (map Msg-mentions msgs))))
(: log-writer-stop (-> Thread Void))
(define (log-writer-stop log-writer)
(: crawl (-> Void))
(define (crawl)
- (let* ([peers-sort
- (λ (peers) (sort peers (match-lambda**
- [((Peer n1 _ _) (Peer n2 _ _))
- (string<? (if n1 n1 "")
- (if n2 n2 ""))])))]
- [peers-all-file
+ (let* ([peers-all-file
(build-path tt-home-dir "peers-all")]
[peers-mentioned-file
(build-path tt-home-dir "peers-mentioned")]
[peers-parsed-file
(build-path tt-home-dir "peers-parsed")]
+ [peers-cached-file
+ (build-path tt-home-dir "peers-cached")]
+ [peers-cached
+ (peers-cached)]
+ [cached-timeline
+ (peers->timeline peers-cached)]
[peers-mentioned-curr
- (mentioned-peers-in-cache)]
+ (peers-mentioned cached-timeline)]
[peers-mentioned-prev
(file->peers peers-mentioned-file)]
[peers-mentioned
- (peers-sort (uniq (append peers-mentioned-prev
- peers-mentioned-curr)))]
+ (peers-union peers-mentioned-prev
+ peers-mentioned-curr)]
[peers-all-prev
(file->peers peers-all-file)]
[peers-all
- (list->set (append peers-mentioned
- peers-all-prev))]
+ (peers-union peers-mentioned
+ peers-all-prev
+ peers-cached)]
[peers-discovered
- (set-subtract peers-all (list->set peers-all-prev))]
- [peers-all
- (peers-sort (set->list peers-all))]
+ (set-subtract peers-all
+ peers-all-prev)]
[peers-parsed
- (filter
- (λ (p) (< 0 (length (peer->msgs p))))
- peers-all)])
+ (for/set ([p peers-all] #:when (> (length (peer->msgs p)) 0)) p)])
; TODO Deeper de-duping
- (log-info "Known peers mentioned: ~a" (length peers-mentioned))
- (log-info "Known peers parsed ~a" (length peers-parsed))
- (log-info "Known peers total: ~a" (length peers-all))
+ (log-info "Known peers cached ~a" (set-count peers-cached))
+ (log-info "Known peers mentioned: ~a" (set-count peers-mentioned))
+ (log-info "Known peers parsed ~a" (set-count peers-parsed))
+ (log-info "Known peers total: ~a" (set-count peers-all))
(log-info "Discovered ~a new peers:~n~a"
(set-count peers-discovered)
(pretty-format (map
- (λ (p) (cons (Peer-nick p)
- (url->string (Peer-uri p))))
+ (match-lambda
+ [(Peer n _ u c) (list n u c)])
(set->list peers-discovered))))
+ (peers->file peers-cached
+ peers-cached-file)
(peers->file peers-mentioned
peers-mentioned-file)
(peers->file peers-parsed
(define-values (_res _cpu real-ms _gc)
(time-apply timeline-download (list num-workers timeout peers)))
(log-info "Downloaded timelines from ~a peers in ~a seconds."
- (length peers)
+ (set-count peers)
(/ real-ms 1000.0))))
(: dispatch (-> String Void))
home / code / tt.git / blobdiff