(∀ (α β) (U (cons 'ok α)
(cons 'error β))))
+(define-type Download-Result
+ (Result (U 'skipped-cached 'downloaded-new)
+ (U 'timeout
+ (Pair 'unsupported-url-scheme String)
+ (Pair 'http-not-ok Positive-Integer)
+ (Pair 'net-error Any)
+ (Pair 'other Any))))
+
(struct Hist
([freq : Nonnegative-Integer]
[last : Nonnegative-Integer])
#:transparent)
-(define-type Nick-Hist
+(define-type Url-Nick-Hist
(Immutable-HashTable Url (Immutable-HashTable (Option String) Hist)))
(struct User
(: tt-home-dir Path-String)
(define tt-home-dir (build-path (expand-user-path "~") ".tt"))
+(: pub-peers-dir Path-String)
+(define pub-peers-dir (build-path tt-home-dir "peers"))
+
(: 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
(: peers->file (-> (Listof Peers) Path-String Void))
(define (peers->file peers path)
+ (make-parent-directory* path)
(display-lines-to-file
(map peer->str
(sort peers
[(list val) val]
[_ #f]))
-(: uri-download-from-port
+(: uri-download-http-from-port
(-> Url (Listof (U Bytes String)) Input-Port
(U 'skipped-cached 'downloaded-new))) ; TODO 'ok|'error ?
-(define (uri-download-from-port u headers body-input)
+(define (uri-download-http-from-port u headers body-input)
; TODO Update message db from here? or where?
; - 1st try can just be an in-memory set that gets written-to
; and read-from disk as a whole.
(define u-str (url->string u))
- (log-debug "uri-download-from-port ~v into ~v" u-str cached-object-path)
+ (log-debug "uri-download-http-from-port ~v into ~v" u-str cached-object-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))
'downloaded-new)
'skipped-cached))
-(: uri-download
- (-> Positive-Float Url
- (Result (U 'skipped-cached 'downloaded-new)
- Any))) ; TODO Maybe more-precise error type?
-(define (uri-download timeout u)
+(: uri-download-http (-> Positive-Float Url Download-Result))
+(define (uri-download-http timeout u)
(define u-str (url->string u))
(define timeout-chan (make-channel))
(define result-chan (make-channel))
; TODO Should a redirect update a peer URL?
(match status
[200
- `(ok . ,(uri-download-from-port u headers body-input))]
+ `(ok . ,(uri-download-http-from-port u headers body-input))]
[_
`(error . (http-not-ok . ,status))])])
(close-input-port body-input)
(kill-thread timeout-thread)
result)
+(: uri-download (-> Positive-Float Url Download-Result))
+(define (uri-download timeout u)
+ (match (url-scheme u)
+ ; TODO Support Gopher.
+ [(or "http" "https")
+ (uri-download-http timeout u)]
+ [scheme
+ `(error . (unsupported-url-scheme . ,scheme))]))
+
(: timeline-print (-> Out-Format (Listof Msg) Void))
(define (timeline-print out-format timeline)
(match timeline
[(cons p (cons 'error e))
(struct-copy Peer p [comment (format "~s" e)])])
results))
- (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")))
+ (peers->file peers-ok (build-path tt-home-dir "peers-last-downloaded-ok.txt"))
+ (peers->file peers-err (build-path tt-home-dir "peers-last-downloaded-err.txt")))
(: peers->timeline (-> (Listof Peer) (Listof Msg)))
(define (peers->timeline peers)
(define (paths->peers paths)
(let* ([paths (match paths
['()
- (let ([peer-refs-file (build-path tt-home-dir "peers")])
+ (let ([peer-refs-file (build-path tt-home-dir "following.txt")])
(log-debug
"No peer ref file paths provided, defaulting to ~v"
(path->string peer-refs-file))
(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 path (build-path tt-home-dir "nicks" "seen" (uri-encode (url->string url))))
- (make-parent-directory* path)
+ (define filename (string-append (uri-encode (url->string url)) ".txt"))
+ (define filepath (build-path dirpath filename))
+ (make-parent-directory* filepath)
(display-lines-to-file
(map (match-lambda
[(cons nick (Hist freq last))
(match-lambda**
[((cons _ (Hist a _)) (cons _ (Hist b _)))
(> a b)])))
- path
+ 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)
; TODO Test the non-io parts of crawling
(let* ([peers-all-file
- (build-path tt-home-dir "peers-all")]
+ (build-path pub-peers-dir "all.txt")]
[peers-mentioned-file
- (build-path tt-home-dir "peers-mentioned")]
+ (build-path pub-peers-dir "mentioned.txt")]
[peers-parsed-file
- (build-path tt-home-dir "peers-parsed")]
+ (build-path pub-peers-dir "downloaded-and-parsed.txt")]
[peers-cached-file
- (build-path tt-home-dir "peers-cached")]
+ (build-path pub-peers-dir "downloaded.txt")]
[peers-cached
(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"
#:args (command . args)
(define log-writer (log-writer-start log-level))
(current-command-line-arguments (list->vector args))
- (set-user-agent-str (build-path tt-home-dir "me"))
+ (set-user-agent-str (build-path tt-home-dir "user.txt"))
; TODO dispatch should return status with which we should exit after cleanups
(dispatch command)
(log-writer-stop log-writer))))