(∀ (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
[(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
(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
(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