X-Git-Url: https://git.xandkar.net/?a=blobdiff_plain;f=tt.rkt;h=6e227283e08694972a02baf4a703f3661e680076;hb=5fef9856f51272827b23cbf47f3b61d3a51abef0;hp=6ce0bd17e2ce2b4b3d2218af16cf38e916c4b2d7;hpb=50f0609d3fd9ea1ab8c37907c7d7ee7a80546e0a;p=tt.git diff --git a/tt.rkt b/tt.rkt index 6ce0bd1..6e22728 100644 --- a/tt.rkt +++ b/tt.rkt @@ -27,6 +27,26 @@ (∀ (α β) (U (cons 'ok α) (cons 'error β)))) +(struct Hist + ([freq : Nonnegative-Integer] + [last : Nonnegative-Integer]) + #:transparent) + +(define-type Url-Nick-Hist + (Immutable-HashTable Url (Immutable-HashTable (Option String) Hist))) + +(struct User + ([uri : Url] + [nick : (Option String)])) + +(struct User-Agent + ([user : User] + [prog : Prog])) + +(struct Prog + ([name : String] + [version : String])) + (struct Msg ([ts-epoch : Integer] [ts-orig : String] @@ -41,11 +61,73 @@ [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))) + +(: user-default User) +(define user-default + (User (string->url "https://github.com/xandkar/tt") #f)) + +(: user->str (-> User String)) +(define (user->str user) + (match-define (User u0 n) user) + (define u (url->string u0)) + (if n + (format "+~a; @~a" u n) + (format "+~a" u ))) + +(: user-agent->str (-> User-Agent String)) +(define (user-agent->str ua) + (match-define (User-Agent u p) ua) + (format "~a/~a (~a)" (Prog-name p) (Prog-version p) (user->str u))) + +(: user->user-agent User) +(define (user->user-agent user) + (User-Agent user prog)) + +(: user-agent-str String) +(define user-agent-str + (user-agent->str (user->user-agent user-default))) + +(: set-user-agent-str (-> Path-String Void)) +(define (set-user-agent-str filename) + (set! user-agent-str (user-agent->str (user->user-agent (file->user filename)))) + (log-info "User-Agent string is now set to: ~v" user-agent-str)) + +(: file->user (-> Path-String User)) +(define (file->user filename) + (if (file-exists? filename) + (match (file->peers filename) + [(list p) + (log-info + "User-Agent. Found one peer in file: ~v. Using the found peer: ~a" + filename + (peer->str p)) + (peer->user p)] + [(list* p _) + (log-warning + "User-Agent. Multiple peers in file: ~v. Picking arbitrary: ~a" + filename + (peer->str p)) + (peer->user p)] + ['() + (log-warning + "User-Agent. No peers found in file: ~v. Using the default user: ~a" + filename + user-default) + user-default]) + (begin + (log-warning + "User-Agent. File doesn't exist: ~v. Using the default user: ~a" + filename + user-default) + user-default))) + +(: peer->user (-> Peer User)) +(define (peer->user p) + (match-define (Peer n u _ _) p) + (User u n)) (: peers-equal? (-> Peer Peer Boolean)) (define (peers-equal? p1 p2) @@ -74,13 +156,14 @@ ; ; TODO Investigate why and make a minimal reproducible test case. -(define (peers-union . peer-sets) +(: peers-merge (-> (Listof Peer) * (Listof Peer))) +(define (peers-merge . peer-sets) (define groups (foldl (λ (p groups) (hash-update groups (Peer-uri-str p) (λ (group) (cons p group)) '())) (hash) - (append* (map set->list peer-sets)))) + (append* peer-sets))) (define (merge peers) (match peers ['() (raise 'impossible)] @@ -89,14 +172,17 @@ (let* ([n1 (Peer-nick p1)] [n2 (Peer-nick p2)] [p (cond - [(and (not n1) (not n2)) p1] - [(and n1 n2 ) p1] + ; TODO Try to pick from nicks db: preferred, otherwise seen + [(and (not n1) (not n2)) p1] ; TODO update with most-common nick + [(and n1 n2 ) p1] ; TODO compare which is more-common [(and n1 (not n2)) p1] [(and (not n1) n2) p2] [else (raise 'impossible)])]) (merge (cons p ps)))])) - (make-immutable-peers (map merge (hash-values groups)))) + (sort (map merge (hash-values groups)) + (match-lambda** + [((Peer _ _ u1 _) (Peer _ _ u2 _)) (stringurl u1) u1 #f)] [p2 (Peer "a" (string->url u1) u1 #f)] [p3 (Peer "b" (string->url u2) u2 #f)] - [s1 (make-immutable-peers (list p1))] - [s2 (make-immutable-peers (list p2 p3))]) - (check-true (peers? (peers-union s1 s2))) - (check-true (peers? (peers-union s2 s1))) - (check-equal? (list p3 p2) (set->list (peers-union s1 s2))) - (check-equal? (list p3 p2) (set->list (peers-union s2 s1))))) + [s1 (list p1)] + [s2 (list p2 p3)]) + (check-equal? (list p3 p2) (peers-merge s1 s2)) + (check-equal? (list p3 p2) (peers-merge s2 s1)))) (: 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 @@ -317,6 +404,8 @@ (module+ test (check-equal? (str->lines "abc\ndef\n\nghi") '("abc" "def" "ghi"))) +; TODO Should return 2 things: 1) msgs; 2) metadata parsed from comments +; TODO Update peer nick based on metadata? (: str->msgs (-> Peer String (Listof Msg))) (define (str->msgs peer str) (filter-map (λ (line) (str->msg peer line)) @@ -431,15 +520,16 @@ (define (filter-comments lines) (filter-not (λ (line) (string-prefix? line "#")) lines)) -(: str->peers (-> String (Setof Peer))) +(: str->peers (-> String (Listof Peer))) (define (str->peers str) - (make-immutable-peers (filter-map str->peer (filter-comments (str->lines str))))) + (filter-map str->peer (filter-comments (str->lines str)))) -(: peers->file (-> (Setof Peers) Path-String Void)) +(: peers->file (-> (Listof Peers) Path-String Void)) (define (peers->file peers path) + (make-parent-directory* path) (display-lines-to-file (map peer->str - (sort (set->list peers) + (sort peers (match-lambda** [((Peer n1 _ _ _) (Peer n2 _ _ _)) (stringpeers (-> Path-String (Setof Peer))) +(: file->peers (-> Path-String (Listof 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") @@ -492,21 +582,6 @@ [_ #f])) -(: user-agent String) -(define user-agent - (let* - ([prog-name "tt"] - [prog-version (info:#%info-lookup 'version)] - [prog-uri "https://github.com/xandkar/tt"] - [user-peer-file (build-path tt-home-dir "me")] - [user - (if (file-exists? user-peer-file) - (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))) - (: header-get (-> (Listof Bytes) Bytes (Option Bytes))) (define (header-get headers name) (match (filter-map (curry extract-field name) headers) @@ -517,6 +592,9 @@ (-> 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)) @@ -576,19 +654,6 @@ (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) @@ -599,32 +664,27 @@ (define-values (status-line headers body-input) (http-sendrecv/url u - #:headers (list (format "User-Agent: ~a" user-agent)))) - `(ok . ,(Resp status-line headers body-input)))) + #:headers (list (format "User-Agent: ~a" user-agent-str)))) + (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) @@ -666,12 +726,12 @@ result) result) -(: timeline-download (-> Integer Positive-Float (Setof Peer) Void)) +(: timeline-download (-> Integer Positive-Float (Listof Peer) Void)) (define (timeline-download num-workers timeout peers) (define results (concurrent-filter-map num-workers (λ (p) (cons p (peer-download timeout p))) - (set->list peers))) + peers)) (define peers-ok (filter-map (match-lambda [(cons p (cons 'ok _)) p] @@ -684,12 +744,12 @@ [(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 (-> (Setof Peer) (Listof Msg))) +(: peers->timeline (-> (Listof Peer) (Listof Msg))) (define (peers->timeline peers) - (append* (filter-map peer->msgs (set->list peers)))) + (append* (filter-map peer->msgs peers))) (: timeline-sort (-> (Listof Msg) timeline-order (Listof Msgs))) (define (timeline-sort msgs order) @@ -699,11 +759,11 @@ (sort msgs (λ (a b) (cmp (Msg-ts-epoch a) (Msg-ts-epoch b))))) -(: paths->peers (-> (Listof String) (Setof Peer))) +(: paths->peers (-> (Listof String) (Listof Peer))) (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)) @@ -711,8 +771,8 @@ [paths (log-debug "Peer ref file paths provided: ~v" paths) (map string->path paths)])] - [peers (apply peers-union (map file->peers paths))]) - (log-info "Read-in ~a peers." (set-count peers)) + [peers (apply peers-merge (map file->peers paths))]) + (log-info "Read-in ~a peers." (length peers)) peers)) (: cache-filename->peer (-> Path-String (Option Peer))) @@ -723,16 +783,28 @@ [#f #f] [url (Peer nick url url-str #f)])) -(: peers-cached (-> (Setof Peer))) +(: peers-cached (-> (Listof Peer))) (define (peers-cached) ; TODO Expire cache? - (make-immutable-peers - (filter-map cache-filename->peer - (directory-list cache-object-dir)))) + (filter-map cache-filename->peer (directory-list cache-object-dir))) -(: peers-mentioned (-> (Listof Msg) (Setof Peer))) +(: peers-mentioned (-> (Listof Msg) (Listof Peer))) (define (peers-mentioned msgs) - (make-immutable-peers (append* (map Msg-mentions 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) @@ -758,49 +830,182 @@ (current-logger logger) log-writer)) +(: msgs->nick-hist (-> (Listof Msg) Url-Nick-Hist)) +(define (msgs->nick-hist msgs) + (foldl + (λ (msg url->nick->hist) + (match-define (Msg curr _ from _ mentions) msg) + (foldl + (λ (peer url->nick->hist) + (match-define (Peer nick url _ _) peer) + (if nick + (hash-update url->nick->hist + url + (λ (nick->hist) + (hash-update nick->hist + nick + (match-lambda + [(Hist freq prev) + (Hist (+ 1 freq) (max prev curr))]) + (Hist 0 0))) + (hash)) + url->nick->hist)) + url->nick->hist + (cons from mentions))) + (hash) + msgs)) + +(: 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) (stringdir (-> Url-Nick-Hist Path-String Void)) +(define (url-nick-hist->dir unh dirpath) + (hash-for-each + unh + (λ (url nick->hist) + (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)) + (format "~a ~a ~a" nick freq last)]) + (sort (hash->list nick->hist) + (match-lambda** + [((cons _ (Hist a _)) (cons _ (Hist b _))) + (> a b)]))) + filepath + #:exists 'replace)))) + +(: 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 + (match (sort (hash->list nick->hist) + (λ (a b) (> (by (cdr a)) + (by (cdr b))))) + ['() #f] + [(cons (cons nick _) _) nick])])) + +(: 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)) + +(: 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 (-> Url-Nick-Hist (Listof Peer) (Listof Peer))) +(define (peers-update-nick-to-common unh peers) + (map + (λ (p) + (match (url-nick-hist-common unh (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)] + [nick1 "a"] + [nick2 "b"] + [nick3 "c"] + [ts-str-1 "2021-11-29T23:29:08-0500"] + [ts-str-2 "2021-11-29T23:30:00-0500"] + [ts-1 (rfc3339->epoch ts-str-1)] + [ts-2 (rfc3339->epoch ts-str-2)] + [msgs + (map (match-lambda + [(cons ts-str nick) + (str->msg (str->peer "test http://test") + (string-append ts-str " Hi @<" nick " " url-str ">"))]) + (list (cons ts-str-2 nick1) + (cons ts-str-1 nick2) + (cons ts-str-1 nick2) + (cons ts-str-1 nick3) + (cons ts-str-1 nick3) + (cons ts-str-1 nick3)))] + [hist + (msgs->nick-hist msgs)]) + (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? (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)] + [url-nick-hist + (msgs->nick-hist cached-timeline)] [peers-mentioned-curr (peers-mentioned cached-timeline)] [peers-mentioned-prev (file->peers peers-mentioned-file)] - [peers-mentioned - (peers-union peers-mentioned-prev - peers-mentioned-curr)] [peers-all-prev (file->peers peers-all-file)] + [peers-mentioned + (peers-merge peers-mentioned-prev + peers-mentioned-curr)] [peers-all - (peers-union peers-mentioned - peers-all-prev - peers-cached)] + (peers-update-nick-to-common + url-nick-hist + (peers-merge peers-mentioned + peers-all-prev + peers-cached))] [peers-discovered - (set-subtract peers-all - peers-all-prev)] + (set->list (set-subtract (make-immutable-peers peers-all) + (make-immutable-peers peers-all-prev)))] [peers-parsed - (for/set ([p peers-all] #:when (> (length (peer->msgs p)) 0)) p)]) + (filter (λ (p) (> (length (peer->msgs p)) 0)) peers-all)]) ; TODO Deeper de-duping - (log-info "Known peers cached ~a" (set-count peers-cached)) - (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 "Known peers cached ~a" (length peers-cached)) + (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) + (length peers-discovered) (pretty-format (map (match-lambda [(Peer n _ u c) (list n u c)]) - (set->list peers-discovered)))) + peers-discovered))) + (update-nicks-history-files url-nick-hist) (peers->file peers-cached peers-cached-file) (peers->file peers-mentioned @@ -831,20 +1036,21 @@ (: 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." - (set-count 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" @@ -915,6 +1121,7 @@ #: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 "user.txt")) ; TODO dispatch should return status with which we should exit after cleanups (dispatch command) (log-writer-stop log-writer))))