(∀ (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
- ([uri : Url]
+ ([url : Url]
[nick : (Option String)]))
(struct User-Agent
(struct Peer
([nick : (Option String)]
- [uri : Url]
- [uri-str : String]
+ [url : Url]
+ [url-str : String]
[comment : (Option String)])
#:transparent)
(: peers-equal? (-> Peer Peer Boolean))
(define (peers-equal? p1 p2)
- (equal? (Peer-uri-str p1)
- (Peer-uri-str p2)))
+ (equal? (Peer-url-str p1)
+ (Peer-url-str p2)))
(: peer-hash (-> Peer Fixnum))
(define (peer-hash p)
- (equal-hash-code (Peer-uri-str p)))
+ (equal-hash-code (Peer-url-str p)))
(define-custom-set-types peers
#:elem? Peer?
(define groups
(foldl
(λ (p groups)
- (hash-update groups (Peer-uri-str p) (λ (group) (cons p group)) '()))
+ (hash-update groups (Peer-url-str p) (λ (group) (cons p group)) '()))
(hash)
(append* peer-sets)))
(define (merge peers)
(λ (out-format color-i msg)
(let ([color (vector-ref colors (modulo color-i n))]
[nick (Peer-nick (Msg-from msg))]
- [uri (Peer-uri-str (Msg-from msg))]
+ [url (Peer-url-str (Msg-from msg))]
[text (Msg-text msg)])
(match out-format
['single-line
- (let ([nick (if nick nick uri)])
+ (let ([nick (if nick nick url)])
(printf "~a \033[1;37m<~a>\033[0m \033[0;~am~a\033[0m~n"
(parameterize
([date-display-format 'iso-8601])
([date-display-format 'rfc2822])
(date->string (seconds->date (Msg-ts-epoch msg)) #t))
(Msg-ts-orig msg)
- nick uri color text))])))))
+ nick url color text))])))))
(: rfc3339->epoch (-> String (Option Nonnegative-Integer)))
(define rfc3339->epoch
(let ([mentions
(filter-map
(λ (m) (match (regexp-match #px"@<([^>]+)>" m)
- [(list _wholething nick-uri)
- (str->peer nick-uri)]))
+ [(list _wholething nick-url)
+ (str->peer nick-url)]))
(regexp-match* #px"@<[^\\s]+([\\s]+)?[^>]+>" text))])
(Msg ts-epoch ts-orig from text mentions))
(begin
(string-append d h m))]
[tzs (list* "" "Z" tzs)])
(for* ([n '("fake-nick")]
- [u '("http://fake-uri")]
+ [u '("http://fake-url")]
[p (list (Peer n (string->url u) u #f))]
[s '("" ":10")]
[f '("" ".1337")]
[tab " "]
[text "Lorem ipsum"]
[nick "foo"]
- [uri "http://bar/"]
- [peer (Peer nick (string->url uri) uri #f)]
+ [url "http://bar/"]
+ [peer (Peer nick (string->url url) url #f)]
[actual (str->msg peer (string-append ts tab text))]
[expected (Msg 1605756129 ts peer text '())])
(check-equal?
(Peer-nick (Msg-from expected))
"str->msg nick")
(check-equal?
- (Peer-uri (Msg-from actual))
- (Peer-uri (Msg-from expected))
- "str->msg uri")
+ (Peer-url (Msg-from actual))
+ (Peer-url (Msg-from expected))
+ "str->msg url")
(check-equal?
- (Peer-uri-str (Msg-from actual))
- (Peer-uri-str (Msg-from expected))
- "str->msg uri-str")
+ (Peer-url-str (Msg-from actual))
+ (Peer-url-str (Msg-from expected))
+ "str->msg url-str")
(check-equal?
(Msg-text actual)
(Msg-text expected)
(define cache-object-dir (build-path cache-dir "objects"))
(: url->cache-file-path-v1 (-> Url Path-String))
-(define (url->cache-file-path-v1 uri)
+(define (url->cache-file-path-v1 url)
(define (hash-sha1 str) : (-> String String)
(define in (open-input-string str))
(define digest (sha1 in))
(close-input-port in)
digest)
- (build-path cache-object-dir (hash-sha1 (url->string uri))))
+ (build-path cache-object-dir (hash-sha1 (url->string url))))
(: url->cache-file-path-v2 (-> Url Path-String))
-(define (url->cache-file-path-v2 uri)
- (build-path cache-object-dir (uri-encode (url->string uri))))
+(define (url->cache-file-path-v2 url)
+ (build-path cache-object-dir (uri-encode (url->string url))))
(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))))
+(define (url->cache-etag-path url)
+ (build-path cache-dir "etags" (uri-encode (url->string url))))
-(define (url->cache-lmod-path uri)
- (build-path cache-dir "lmods" (uri-encode (url->string uri))))
+(define (url->cache-lmod-path url)
+ (build-path cache-dir "lmods" (uri-encode (url->string url))))
-(: uri-read-cached (-> Url (Option String)))
-(define (uri-read-cached uri)
- (define path-v1 (url->cache-file-path-v1 uri))
- (define path-v2 (url->cache-file-path-v2 uri))
+(: url-read-cached (-> Url (Option String)))
+(define (url-read-cached url)
+ (define path-v1 (url->cache-file-path-v1 url))
+ (define path-v2 (url->cache-file-path-v2 url))
(when (file-exists? path-v1)
(rename-file-or-directory path-v1 path-v2 #t))
(if (file-exists? path-v2)
(file->string path-v2)
(begin
- (log-debug "Cache file not found for URI: ~a" (url->string uri))
+ (log-debug "Cache file not found for URL: ~a" (url->string url))
#f)))
(: str->url (-> String (Option String)))
comment)
(match (str->url url)
[#f
- (log-error "Invalid URI in peer string: ~v" str)
+ (log-error "Invalid URL in peer string: ~v" str)
#f]
[url
(Peer nick url (url->string url) comment)])]
[(list val) val]
[_ #f]))
-(: uri-download-from-port
+(: url-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 (url-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 "url-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)
+(: url-download-http (-> Positive-Float Url Download-Result))
+(define (url-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 . ,(url-download-http-from-port u headers body-input))]
[_
`(error . (http-not-ok . ,status))])])
(close-input-port body-input)
(kill-thread timeout-thread)
result)
+(: url-download (-> Positive-Float Url Download-Result))
+(define (url-download timeout u)
+ (match (url-scheme u)
+ ; TODO Support Gopher.
+ [(or "http" "https")
+ (url-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
(: peer->msgs (-> Peer (Listof Msg)))
(define (peer->msgs peer)
- (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))
+ (match-define (Peer nick url url-str _) peer)
+ (log-debug "Reading peer nick:~v url:~v" nick url-str)
+ (define msgs-data (url-read-cached url))
; TODO Expire cache
(if msgs-data
(str->msgs peer msgs-data)
(Result (U 'skipped-cached 'downloaded-new)
Any)))
(define (peer-download timeout peer)
- (match-define (Peer nick uri u _) peer)
+ (match-define (Peer nick url 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)))
+ (time-apply url-download (list timeout url)))
(define result (car results))
(log-info "Download END in ~a seconds, URL:~a, result:~s"
(/ tm-real-ms 1000.0)
(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-url 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-url 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"