[comment : (Option String)])
#:transparent)
-(struct Resp
- ([status-line : String]
- [headers : (Listof Bytes)]
- [body-input : Input-Port])
- #:transparent)
-
(: prog Prog)
(define prog
(Prog "tt" (info:#%info-lookup 'version)))
(: 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
(-> Url (Listof (U Bytes String)) Input-Port
(U 'skipped-cached 'downloaded-new))) ; TODO 'ok|'error ?
(define (uri-download-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)
(define cached-object-path (url->cache-object-path u))
(channel-put timeout-chan '(error . timeout)))))
(define result-thread
(thread (λ ()
- ; XXX We timeout getting a response, but body download could
- ; also take a long time and we might want to time that out as
- ; well, but then we may end-up with partially downloaded
- ; objects. But that could happen anyway if the server drops the
- ; connection for whatever reason.
- ;
- ; Maybe that is OK once we start treating the
- ; downloaded object as an addition to the stored set of
- ; messages, rather than the final set of messages.
-
- ; TODO message db
- ; - 1st try can just be an in-memory set that gets written-to
- ; and read-from disk as a whole.
(define result
(with-handlers
; TODO Maybe name each known errno? (exn:fail:network:errno-errno e)
(http-sendrecv/url
u
#:headers (list (format "User-Agent: ~a" user-agent-str))))
- `(ok . ,(Resp status-line headers body-input))))
+ (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)
+ (let ([result
+ ; TODO Handle redirects.
+ ; TODO Should a redirect update a peer URL?
+ (match status
+ [200
+ `(ok . ,(uri-download-from-port u headers body-input))]
+ [_
+ `(error . (http-not-ok . ,status))])])
+ (close-input-port body-input)
+ result)))
(channel-put result-chan result))))
- (define result
- (sync timeout-chan
- result-chan))
+ (define result (sync timeout-chan result-chan))
(kill-thread result-thread)
(kill-thread timeout-thread)
- (match result
- [(cons 'error _)
- result]
- [(cons 'ok (Resp status-line headers body-input))
- (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. Should be within same timeout as req and body.
- (let ([result
- (match status
- [200
- `(ok . ,(uri-download-from-port u headers body-input))]
- [_
- `(error . (http . ,status))])])
- (close-input-port body-input)
- result)]))
+ result)
(: timeline-print (-> Out-Format (Listof Msg) Void))
(define (timeline-print out-format 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)
(hash)
msgs))
-(: update-nicks-history (-> (Listof Msg) Void))
-(define (update-nicks-history msgs)
+(: update-nicks-history-files (-> Nick-Hist Void))
+(define (update-nicks-history-files nick-hist)
(hash-for-each
- (msgs->nick-hist msgs)
+ nick-hist
(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 tt-home-dir "nicks" "seen" 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-common nick-hist url)
(nick-hist-most-by nick-hist url Hist-freq))
+(: peers-update-nick-to-common (-> Nick-Hist (Listof Peer) (Listof Peer)))
+(define (peers-update-nick-to-common nick-hist peers)
+ (map
+ (λ (p)
+ (match (nick-hist-common nick-hist (Peer-uri p))
+ [#f p]
+ [n (struct-copy Peer p [nick n])]))
+ peers))
+
(module+ test
(let* ([url-str "http://foo"]
[url (string->url url-str)]
(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
+ (msgs->nick-hist cached-timeline)]
[peers-mentioned-curr
(peers-mentioned cached-timeline)]
[peers-mentioned-prev
[peers-all-prev
(file->peers peers-all-file)]
[peers-mentioned
- (begin
- ; XXX Updating nicks before running peers-merge,
- ; since peers-merge is expected to refer to it in the future.
- (update-nicks-history cached-timeline)
- (peers-merge peers-mentioned-prev
- peers-mentioned-curr))]
+ (peers-merge peers-mentioned-prev
+ peers-mentioned-curr)]
[peers-all
- (peers-merge peers-mentioned
- peers-all-prev
- peers-cached)]
+ (peers-update-nick-to-common
+ nick-hist
+ (peers-merge peers-mentioned
+ peers-all-prev
+ peers-cached))]
[peers-discovered
(set->list (set-subtract (make-immutable-peers peers-all)
(make-immutable-peers peers-all-prev)))]
(match-lambda
[(Peer n _ u c) (list n u c)])
peers-discovered)))
+ (update-nicks-history-files 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))
#: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))))