(struct Peer
([nick : (Option String)]
[uri : Url]
+ [uri-str : String]
[comment : (Option String)])
#:transparent)
[body-input : Input-Port])
#:transparent)
+(define-custom-set-types peers
+ #:elem? Peer?
+ (λ (p1 p2)
+ (equal? (Peer-uri-str p1)
+ (Peer-uri-str p2)))
+ (λ (p)
+ (equal-hash-code (Peer-uri-str p))))
+; XXX Without supplying above explicit hash procedure, we INTERMITTENTLY get
+; the following contract violations:
+;
+; custom-elem-contents: contract violation
+; expected: custom-elem?
+; given: #f
+; context...:
+; /usr/share/racket/collects/racket/private/set-types.rkt:104:0: custom-set->list
+; /home/siraaj/proj/pub/tt/tt.rkt:716:0: crawl
+; /usr/share/racket/collects/racket/cmdline.rkt:191:51
+; body of (submod "/home/siraaj/proj/pub/tt/tt.rkt" main)
+;
+; TODO Investigate why and make a minimal reproducible test case.
+
(: tt-home-dir Path-String)
(define tt-home-dir (build-path (expand-user-path "~") ".tt"))
local-time?)])
(+ ts-epoch tz-offset))]
[_
- (log-error "Invalid timestamp: ~v" ts)
+ (log-debug "Invalid timestamp: ~v" ts)
#f]))))
(: str->msg (-> (Option String) Url String (Option Msg)))
(with-handlers*
([exn:fail?
(λ (e)
- (log-error
+ (log-debug
"Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v"
str-head nick (url->string uri) e)
#f)])
(regexp-match* #px"@<[^\\s]+([\\s]+)?[^>]+>" text))])
(Msg ts-epoch ts-orig nick uri text mentions))
(begin
- (log-error
+ (log-debug
"Msg rejected due to invalid timestamp: ~v, nick:~v, uri:~v"
str-head nick (url->string uri))
#f)))]
(if (file-exists? path-v2)
(file->string path-v2)
(begin
- (log-warning "Cache file not found for URI: ~a" (url->string uri))
+ (log-debug "Cache file not found for URI: ~a" (url->string uri))
#f)))
(: str->url (-> String (Option String)))
[#f
(log-error "Invalid URI in peer string: ~v" str)
#f]
- [url (Peer nick url comment)])]
+ [url
+ (Peer nick url (url->string url) comment)])]
[_
- (log-error "Invalid peer string: ~v" str)
+ (log-debug "Invalid peer string: ~v" str)
#f]))
(module+ test
(check-equal?
(str->peer "foo http://bar/file.txt # some rando")
- (Peer "foo" (str->url "http://bar/file.txt") "some rando"))
+ (Peer "foo" (str->url "http://bar/file.txt") "http://bar/file.txt" "some rando"))
(check-equal?
(str->peer "http://bar/file.txt # some rando")
- (Peer #f (str->url "http://bar/file.txt") "some rando"))
+ (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" "some rando"))
(check-equal?
(str->peer "http://bar/file.txt #")
- (Peer #f (str->url "http://bar/file.txt") ""))
+ (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" ""))
(check-equal?
(str->peer "http://bar/file.txt#") ; XXX URLs can have #s
- (Peer #f (str->url "http://bar/file.txt#") #f))
+ (Peer #f (str->url "http://bar/file.txt#") "http://bar/file.txt#" #f))
(check-equal?
(str->peer "http://bar/file.txt")
- (Peer #f (str->url "http://bar/file.txt") #f))
+ (Peer #f (str->url "http://bar/file.txt") "http://bar/file.txt" #f))
(check-equal?
(str->peer "foo http://bar/file.txt")
- (Peer "foo" (str->url "http://bar/file.txt") #f))
+ (Peer "foo" (str->url "http://bar/file.txt") "http://bar/file.txt" #f))
(check-equal?
(str->peer "foo bar # baz")
#f)
(check-equal?
(str->peer "foo bar://baz # quux")
- (Peer "foo" (str->url "bar://baz") "quux"))
+ (Peer "foo" (str->url "bar://baz") "bar://baz" "quux"))
(check-equal?
(str->peer "foo bar//baz # quux")
#f))
(define (filter-comments lines)
(filter-not (λ (line) (string-prefix? line "#")) lines))
-(: str->peers (-> String (Listof Peer)))
+(: str->peers (-> String (Setof Peer)))
(define (str->peers str)
- (filter-map str->peer (filter-comments (str->lines str))))
+ (make-immutable-peers (filter-map str->peer (filter-comments (str->lines str)))))
-(: peers->file (-> (Listof Peers) Path-String Void))
+(: peers->file (-> (Setof Peers) Path-String Void))
(define (peers->file peers path)
(display-lines-to-file
(map (match-lambda
- [(Peer n u c)
+ [(Peer n _ u c)
(format "~a~a~a"
(if n (format "~a " n) "")
- (url->string u)
+ u
(if c (format " # ~a" c) ""))])
- peers)
+ (sort (set->list peers)
+ (match-lambda**
+ [((Peer n1 _ _ _) (Peer n2 _ _ _))
+ (string<? (if n1 n1 "")
+ (if n2 n2 ""))])))
path
#:exists 'replace))
-(: file->peers (-> Path-String (Listof Peer)))
+(: file->peers (-> Path-String (Setof Peer)))
(define (file->peers file-path)
(if (file-exists? file-path)
(str->peers (file->string file-path))
(begin
(log-warning "File does not exist: ~v" (path->string file-path))
- '())))
+ (make-immutable-peers))))
(define re-rfc2822
#px"^(Mon|Tue|Wed|Thu|Fri|Sat|Sun), ([0-9]{2}) (Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec) ([0-9]{4}) ([0-2][0-9]):([0-6][0-9]):([0-6][0-9]) GMT")
[user-peer-file (build-path tt-home-dir "me")]
[user
(if (file-exists? user-peer-file)
- (match (first (file->peers user-peer-file))
- [(Peer #f u _) (format "+~a" (url->string u) )]
- [(Peer n u _) (format "+~a; @~a" (url->string u) n)])
+ (match (set-first (file->peers user-peer-file))
+ [(Peer #f _ u _) (format "+~a" u )]
+ [(Peer n _ u _) (format "+~a; @~a" u n)])
(format "+~a" prog-uri))])
(format "~a/~a (~a)" prog-name prog-version user)))
(log-debug "Last-Modified <= current skipping the rest of ~v" u-str)
#t))))
(if (not cached?)
- (begin
- (log-debug
- "Downloading the rest of ~v. ETag: ~a, Last-Modified: ~v"
- u-str etag lmod)
- (make-parent-directory* cached-object-path)
- (make-parent-directory* cached-etag-path)
- (make-parent-directory* cached-lmod-path)
- (call-with-output-file cached-object-path
- (curry copy-port body-input)
- #:exists 'replace)
- (when etag
- (display-to-file etag cached-etag-path #:exists 'replace))
- (when lmod
- (display-to-file lmod cached-lmod-path #:exists 'replace))
- 'downloaded-new)
- 'skipped-cached))
+ (begin
+ (log-debug
+ "Downloading the rest of ~v. ETag: ~a, Last-Modified: ~v"
+ u-str etag lmod)
+ (make-parent-directory* cached-object-path)
+ (make-parent-directory* cached-etag-path)
+ (make-parent-directory* cached-lmod-path)
+ (call-with-output-file cached-object-path
+ (curry copy-port body-input)
+ #:exists 'replace)
+ (when etag
+ (display-to-file etag cached-etag-path #:exists 'replace))
+ (when lmod
+ (display-to-file lmod cached-lmod-path #:exists 'replace))
+ 'downloaded-new)
+ 'skipped-cached))
(: uri-download
(-> Positive-Float Url
(: peer->msgs (-> Peer (Listof Msg)))
(define (peer->msgs peer)
- (match-define (Peer nick uri _) peer)
- (log-info "Reading peer nick:~v uri:~v" nick (url->string uri))
+ (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))
+ ; TODO Expire cache
(if msgs-data
(str->msgs nick uri msgs-data)
'()))
(Result (U 'skipped-cached 'downloaded-new)
Any)))
(define (peer-download timeout peer)
- (match-define (Peer nick uri _) peer)
- (define u (url->string uri))
+ (match-define (Peer nick uri 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)))
result)
result)
-(: timeline-download (-> Integer Positive-Float (Listof Peer) Void))
+(: timeline-download (-> Integer Positive-Float (Setof Peer) Void))
(define (timeline-download num-workers timeout peers)
(define results
(concurrent-filter-map num-workers
(λ (p) (cons p (peer-download timeout p)))
- peers))
+ (set->list peers)))
(define peers-ok
(filter-map (match-lambda
[(cons p (cons 'ok _)) p]
(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")))
-(: uniq (∀ (α) (-> (Listof α) (Listof α))))
-(define (uniq xs)
- (set->list (list->set xs)))
-
-(: peers->timeline (-> (listof Peer) (listof Msg)))
+(: peers->timeline (-> (Listof Peer) (Listof Msg)))
(define (peers->timeline peers)
- (append* (filter-map peer->msgs peers)))
+ (append* (filter-map peer->msgs (set->list peers))))
-(: timeline-sort (-> (listof Msg) timeline-order (Listof Msgs)))
+(: timeline-sort (-> (Listof Msg) timeline-order (Listof Msgs)))
(define (timeline-sort msgs order)
(define cmp (match order
['old->new <]
(sort msgs (λ (a b) (cmp (Msg-ts-epoch a)
(Msg-ts-epoch b)))))
-(: paths->peers (-> (Listof String) (Listof Peer)))
+(: paths->peers (-> (Listof String) (Setof Peer)))
(define (paths->peers paths)
(let* ([paths (match paths
['()
[paths
(log-debug "Peer ref file paths provided: ~v" paths)
(map string->path paths)])]
- [peers (append* (map file->peers paths))])
- (log-info "Read-in ~a peers." (length peers))
- (uniq peers)))
+ [peers (apply set-union (map file->peers paths))])
+ (log-info "Read-in ~a peers." (set-count peers))
+ peers))
-(: mentioned-peers-in-cache (-> (Listof Peer)))
+(: mentioned-peers-in-cache (-> (Setof Peer)))
(define (mentioned-peers-in-cache)
+ ; TODO Expire cache
(define msgs
(append* (map (λ (filename)
(define path (build-path cache-object-dir filename))
(define size (/ (file-size path) 1000000.0))
- (log-info "BEGIN parsing ~a MB from file: ~v"
- size
- (path->string path))
+ (log-debug "BEGIN parsing ~a MB from file: ~v"
+ size
+ (path->string path))
(define t0 (current-inexact-milliseconds))
(define m (filter-map
(λ (line)
(filter-comments
(file->lines path))))
(define t1 (current-inexact-milliseconds))
- (log-info "END parsing ~a MB in ~a seconds from file: ~v."
- size
- (* 0.001 (- t1 t0))
- (path->string path))
+ (log-debug "END parsing ~a MB in ~a seconds from file: ~v."
+ size
+ (* 0.001 (- t1 t0))
+ (path->string path))
(when (empty? m)
- (log-warning "No messages found in ~a" (path->string path)))
+ (log-debug "No messages found in ~a" (path->string path)))
m)
(directory-list cache-object-dir))))
- (uniq (append* (map Msg-mentions msgs))))
+ (make-immutable-peers (append* (map Msg-mentions msgs))))
(: log-writer-stop (-> Thread Void))
(define (log-writer-stop log-writer)
(current-logger logger)
log-writer))
+(: crawl (-> Void))
+(define (crawl)
+ (let* ([peers-all-file
+ (build-path tt-home-dir "peers-all")]
+ [peers-mentioned-file
+ (build-path tt-home-dir "peers-mentioned")]
+ [peers-parsed-file
+ (build-path tt-home-dir "peers-parsed")]
+ [peers-mentioned-curr
+ (mentioned-peers-in-cache)]
+ [peers-mentioned-prev
+ (file->peers peers-mentioned-file)]
+ [peers-mentioned
+ (set-union peers-mentioned-prev
+ peers-mentioned-curr)]
+ [peers-all-prev
+ (file->peers peers-all-file)]
+ [peers-all
+ (set-union peers-mentioned
+ peers-all-prev)]
+ [peers-discovered
+ (set-subtract peers-all
+ peers-all-prev)]
+ [peers-parsed
+ (for/set ([p peers-all] #:when (> (length (peer->msgs p)) 0)) p)])
+ ; TODO Deeper de-duping
+ (log-info "Known peers mentioned: ~a" (set-count peers-mentioned))
+ (log-info "Known peers parsed ~a" (set-count peers-parsed))
+ (log-info "Known peers total: ~a" (set-count peers-all))
+ (log-info "Discovered ~a new peers:~n~a"
+ (set-count peers-discovered)
+ (pretty-format (map
+ (match-lambda
+ [(Peer n _ u c) (list n u c)])
+ (set->list peers-discovered))))
+ (peers->file peers-mentioned
+ peers-mentioned-file)
+ (peers->file peers-parsed
+ peers-parsed-file)
+ (peers->file peers-all
+ peers-all-file)))
+
+(: read (-> (Listof String) Number Number Timeline-Order Out-Format Void))
+(define (read file-paths ts-min ts-max order out-format)
+ (let* ([peers
+ (paths->peers file-paths)]
+ [msgs
+ (timeline-sort (peers->timeline peers) order)]
+ [include?
+ (λ (m)
+ (and (or (not ts-min) (>= (Msg-ts-epoch m) ts-min))
+ (or (not ts-max) (<= (Msg-ts-epoch m) ts-max))))])
+ (timeline-print out-format (filter include? msgs))))
+
+(: upload (-> Void))
+(define (upload)
+ ; FIXME Should not exit from here, but only after cleanup/logger-stoppage.
+ (if (system (path->string (build-path tt-home-dir "hooks" "upload")))
+ (exit 0)
+ (exit 1)))
+
+(: download (-> (Listof String) Positive-Integer Positive-Float Void))
+(define (download file-paths num-workers timeout)
+ (let ([peers (paths->peers file-paths)])
+ (define-values (_res _cpu real-ms _gc)
+ (time-apply timeline-download (list num-workers timeout peers)))
+ (log-info "Downloaded timelines from ~a peers in ~a seconds."
+ (set-count peers)
+ (/ 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]
+ [timeout 10.0])
+ (command-line
+ #:program "tt download"
+ #:once-each
+ [("-j" "--jobs")
+ njobs "Number of concurrent jobs."
+ (set! num-workers (string->number njobs))]
+ [("-t" "--timeout")
+ seconds "Timeout seconds per request."
+ (set! timeout (string->number seconds))]
+ #:args file-paths
+ (download file-paths num-workers timeout)))]
+ [(or "u" "upload")
+ (command-line
+ #:program "tt upload" #:args () (upload))]
+ [(or "r" "read")
+ (let ([out-format 'multi-line]
+ [order 'old->new]
+ [ts-min #f]
+ [ts-max #f])
+ (command-line
+ #:program "tt read"
+ #:once-each
+ [("-r" "--rev")
+ "Reverse displayed timeline order."
+ (set! order 'new->old)]
+ [("-m" "--min")
+ m "Earliest time to display (ignore anything before it)."
+ (set! ts-min (rfc3339->epoch m))]
+ [("-x" "--max")
+ x "Latest time to display (ignore anything after it)."
+ (set! ts-max (rfc3339->epoch x))]
+ #:once-any
+ [("-s" "--short")
+ "Short output format"
+ (set! out-format 'single-line)]
+ [("-l" "--long")
+ "Long output format"
+ (set! out-format 'multi-line)]
+ #:args file-paths
+ (read file-paths ts-min ts-max order out-format)))]
+ [(or "c" "crawl")
+ (command-line
+ #:program "tt crawl" #:args () (crawl))]
+ [command
+ (eprintf "Error: invalid command: ~v\n" command)
+ (eprintf "Please use the \"--help\" option to see a list of available commands.\n")
+ (exit 1)]))
+
(module+ main
(let ([log-level 'info])
(command-line
"r, read : Read the timeline (offline operation)."
"d, download : Download the timeline."
; TODO Add path dynamically
- "u, upload : Upload your twtxt file (alias to execute ~/.tt/upload)."
+ "u, upload : Upload your twtxt file (alias to execute ~/.tt/hooks/upload)."
"c, crawl : Discover new peers mentioned by known peers (offline operation)."
""
#:args (command . args)
(define log-writer (log-writer-start log-level))
(current-command-line-arguments (list->vector args))
- (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]
- [timeout 10.0])
- (command-line
- #:program
- "tt download"
- #:once-each
- [("-j" "--jobs")
- njobs "Number of concurrent jobs."
- (set! num-workers (string->number njobs))]
- [("-t" "--timeout")
- seconds "Timeout seconds per request."
- (set! timeout (string->number seconds))]
- #:args file-paths
- (let ([peers (paths->peers file-paths)])
- (define-values (_res _cpu real-ms _gc)
- (time-apply timeline-download (list num-workers timeout peers)))
- (log-info "Downloaded timelines from ~a peers in ~a seconds."
- (length peers)
- (/ real-ms 1000.0)))))]
- [(or "u" "upload")
- (command-line
- #:program
- "tt upload"
- #:args ()
- (if (system (path->string (build-path tt-home-dir "upload")))
- (exit 0)
- (exit 1)))]
- [(or "r" "read")
- (let ([out-format 'multi-line]
- [order 'old->new]
- [ts-min #f]
- [ts-max #f])
- (command-line
- #:program
- "tt read"
- #:once-each
- [("-r" "--rev")
- "Reverse displayed timeline order."
- (set! order 'new->old)]
- [("-m" "--min")
- m "Earliest time to display (ignore anything before it)."
- (set! ts-min (rfc3339->epoch m))]
- [("-x" "--max")
- x "Latest time to display (ignore anything after it)."
- (set! ts-max (rfc3339->epoch x))]
- #:once-any
- [("-s" "--short")
- "Short output format"
- (set! out-format 'single-line)]
- [("-l" "--long")
- "Long output format"
- (set! out-format 'multi-line)]
- #:args file-paths
- (let* ([peers
- (paths->peers file-paths)]
- [timeline
- (timeline-sort (peers->timeline peers) order)]
- [timeline
- (filter (λ (m) (and (if ts-min (>= (Msg-ts-epoch m)
- ts-min)
- #t)
- (if ts-max (<= (Msg-ts-epoch m)
- ts-max)
- #t)))
- timeline)])
- (timeline-print out-format timeline))))]
- [(or "c" "crawl")
- (command-line
- #:program
- "tt crawl"
- #:args ()
- (let* ([peers-sort
- (λ (peers) (sort peers (match-lambda**
- [((Peer n1 _ _) (Peer n2 _ _))
- (string<? (if n1 n1 "")
- (if n2 n2 ""))])))]
- [peers-all-file
- (build-path tt-home-dir "peers-all")]
- [peers-mentioned-file
- (build-path tt-home-dir "peers-mentioned")]
- [peers-parsed-file
- (build-path tt-home-dir "peers-parsed")]
- [peers-mentioned-curr
- (mentioned-peers-in-cache)]
- [peers-mentioned-prev
- (file->peers peers-mentioned-file)]
- [peers-mentioned
- (peers-sort (uniq (append peers-mentioned-prev
- peers-mentioned-curr)))]
- [peers-all-prev
- (file->peers peers-all-file)]
- [peers-all
- (list->set (append peers-mentioned
- peers-all-prev))]
- [peers-discovered
- (set-subtract peers-all (list->set peers-all-prev))]
- [peers-all
- (peers-sort (set->list peers-all))]
- [peers-parsed
- (filter
- (λ (p) (< 0 (length (peer->msgs p))))
- peers-all)])
- (log-info "Known peers mentioned: ~a" (length peers-mentioned))
- (log-info "Known peers parsed ~a" (length peers-parsed))
- (log-info "Known peers total: ~a" (length peers-all))
- (log-info "Discovered ~a new peers:~n~a"
- (set-count peers-discovered)
- (pretty-format (map
- (λ (p) (cons (Peer-nick p)
- (url->string (Peer-uri p))))
- (set->list peers-discovered))))
- (peers->file peers-mentioned
- peers-mentioned-file)
- (peers->file peers-parsed
- peers-parsed-file)
- (peers->file peers-all
- peers-all-file)))]
- [command
- (eprintf "Error: invalid command: ~v\n" command)
- (eprintf "Please use the \"--help\" option to see a list of available commands.\n")
- (exit 1)])
+ ; TODO dispatch should return status with which we should exit after cleanups
+ (dispatch command)
(log-writer-stop log-writer))))
home / code / tt.git / blobdiff