[last : Nonnegative-Integer])
#:transparent)
-(define-type Nick-Hist
+(define-type Url-Nick-Hist
(Immutable-HashTable Url (Immutable-HashTable (Option String) Hist)))
(struct User
(define (peers-mentioned msgs)
(append* (map Msg-mentions msgs)))
+(: peers-filter-denied-domains (-> (Listof Peer) (Listof Peer)))
+(define (peers-filter-denied-domains peers)
+ (define deny-file (build-path tt-home-dir "domains-deny.txt"))
+ (define denied-hosts
+ (list->set (map string-trim (filter-comments (file->lines deny-file)))))
+ (define denied-domain-patterns
+ (set-map denied-hosts (λ (h) (pregexp (string-append "\\." h "$")))))
+ (filter
+ (λ (p)
+ (define host (url-host (Peer-uri p)))
+ (not (or (set-member? denied-hosts host)
+ (ormap (λ (d) (regexp-match? d host)) denied-domain-patterns))))
+ peers))
+
(: log-writer-stop (-> Thread Void))
(define (log-writer-stop log-writer)
(log-message (current-logger) 'fatal 'stop "Exiting." #f)
(current-logger logger)
log-writer))
-(: msgs->nick-hist (-> (Listof Msg) Nick-Hist))
+(: msgs->nick-hist (-> (Listof Msg) Url-Nick-Hist))
(define (msgs->nick-hist msgs)
(foldl
(λ (msg url->nick->hist)
(hash)
msgs))
-(: update-nicks-history-files (-> Nick-Hist Void))
-(define (update-nicks-history-files nick-hist)
+(: url-nick-hist->file (-> Url-Nick-Hist Path-String Void))
+(define (url-nick-hist->file unh filepath)
+ (define out (open-output-file filepath #:exists 'replace))
+ (for-each
+ (match-lambda
+ [(cons url nick->hist)
+ (displayln (url->string url) out)
+ (for-each (match-lambda
+ [(cons nick (Hist freq last))
+ (displayln (format " ~a ~a ~a" nick freq last) out)])
+ (sort (hash->list nick->hist)
+ (match-lambda**
+ [((cons _ (Hist a _)) (cons _ (Hist b _)))
+ (> a b)])))])
+ (sort
+ (hash->list unh)
+ (λ (a b) (string<? (url-host (car a))
+ (url-host (car b))))))
+ (close-output-port out))
+
+(: url-nick-hist->dir (-> Url-Nick-Hist Path-String Void))
+(define (url-nick-hist->dir unh dirpath)
(hash-for-each
- nick-hist
+ unh
(λ (url nick->hist)
(define filename (string-append (uri-encode (url->string url)) ".txt"))
- (define filepath (build-path tt-home-dir "nicks" "seen" filename))
+ (define filepath (build-path dirpath filename))
(make-parent-directory* filepath)
(display-lines-to-file
(map (match-lambda
filepath
#:exists 'replace))))
-(: nick-hist-most-by (-> Nick-Hist Url (-> Hist Nonnegative-Integer) (Option String)))
-(define (nick-hist-most-by url->nick->hist url by)
+(: update-nicks-history-files (-> Url-Nick-Hist Void))
+(define (update-nicks-history-files unh)
+ (define nicks-dir (build-path tt-home-dir "nicks"))
+ (url-nick-hist->file unh (build-path nicks-dir "seen.txt"))
+ (url-nick-hist->dir unh (build-path nicks-dir "seen")))
+
+(: url-nick-hist-most-by (-> Url-Nick-Hist Url (-> Hist Nonnegative-Integer) (Option String)))
+(define (url-nick-hist-most-by url->nick->hist url by)
(match (hash-ref url->nick->hist url #f)
[#f #f]
[nick->hist
['() #f]
[(cons (cons nick _) _) nick])]))
-(: nick-hist-latest (-> Nick-Hist Url (Option String)))
-(define (nick-hist-latest nick-hist url)
- (nick-hist-most-by nick-hist url Hist-last))
+(: url-nick-hist-latest (-> Url-Nick-Hist Url (Option String)))
+(define (url-nick-hist-latest unh url)
+ (url-nick-hist-most-by unh url Hist-last))
-(: nick-hist-common (-> Nick-Hist Url (Option String)))
-(define (nick-hist-common nick-hist url)
- (nick-hist-most-by nick-hist url Hist-freq))
+(: url-nick-hist-common (-> Url-Nick-Hist Url (Option String)))
+(define (url-nick-hist-common unh url)
+ (url-nick-hist-most-by unh url Hist-freq))
-(: peers-update-nick-to-common (-> Nick-Hist (Listof Peer) (Listof Peer)))
-(define (peers-update-nick-to-common nick-hist peers)
+(: peers-update-nick-to-common (-> Url-Nick-Hist (Listof Peer) (Listof Peer)))
+(define (peers-update-nick-to-common unh peers)
(map
(λ (p)
- (match (nick-hist-common nick-hist (Peer-uri p))
+ (match (url-nick-hist-common unh (Peer-uri p))
[#f p]
[n (struct-copy Peer p [nick n])]))
peers))
(check-equal? (hash-ref (hash-ref hist url) nick1) (Hist 1 ts-2))
(check-equal? (hash-ref (hash-ref hist url) nick2) (Hist 2 ts-1))
(check-equal? (hash-ref (hash-ref hist url) nick3) (Hist 3 ts-1))
- (check-equal? (nick-hist-common hist url) nick3)
- (check-equal? (nick-hist-latest hist url) nick1)))
+ (check-equal? (url-nick-hist-common hist url) nick3)
+ (check-equal? (url-nick-hist-latest hist url) nick1)))
(: crawl (-> Void))
(define (crawl)
(peers-cached)]
[cached-timeline
(peers->timeline peers-cached)]
- [nick-hist
+ [url-nick-hist
(msgs->nick-hist cached-timeline)]
[peers-mentioned-curr
(peers-mentioned cached-timeline)]
peers-mentioned-curr)]
[peers-all
(peers-update-nick-to-common
- nick-hist
+ url-nick-hist
(peers-merge peers-mentioned
peers-all-prev
peers-cached))]
(match-lambda
[(Peer n _ u c) (list n u c)])
peers-discovered)))
- (update-nicks-history-files nick-hist)
+ (update-nicks-history-files url-nick-hist)
(peers->file peers-cached
peers-cached-file)
(peers->file peers-mentioned
(: download (-> (Listof String) Positive-Integer Positive-Float Void))
(define (download file-paths num-workers timeout)
- (let ([peers (paths->peers file-paths)])
+ (let* ([peers-given (paths->peers file-paths)]
+ [peers-kept (peers-filter-denied-domains peers-given)]
+ [peers-denied (set-subtract peers-given peers-kept)])
+ (log-info "Denied ~a peers" (length peers-denied))
(define-values (_res _cpu real-ms _gc)
- (time-apply timeline-download (list num-workers timeout peers)))
+ (time-apply timeline-download (list num-workers timeout peers-kept)))
(log-info "Downloaded timelines from ~a peers in ~a seconds."
- (length peers)
+ (length peers-kept)
(/ real-ms 1000.0))))
(: dispatch (-> String Void))
(define (dispatch command)
(match command
[(or "d" "download")
- ; 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]
+ (let ([num-workers 20] ; 20 was fastest out of the tried: 1, 5, 10, 20, 25, 30.
[timeout 10.0])
(command-line
#:program "tt download"
home / code / tt.git / blobdiff