X-Git-Url: https://git.xandkar.net/?a=blobdiff_plain;f=tt.rkt;h=42153ec3bf648ba0fe359f07416580bb4dfd6be5;hb=9c464d95bd3020dec326505b30c5ad1b28dfb655;hp=c505f98a01663ce94f0088ba2856e7da0641a9c6;hpb=b46894648d41fbe9ee516cef054015d686bfe963;p=tt.git diff --git a/tt.rkt b/tt.rkt index c505f98..42153ec 100644 --- a/tt.rkt +++ b/tt.rkt @@ -2,15 +2,31 @@ (require openssl/sha1) (require racket/date) +(require (prefix-in info: setup/getinfo)) +(require + net/http-client + net/url-string + net/url-structs) -(require http-client) (require rfc3339-old) (module+ test (require rackunit)) -(struct msg (ts_epoch ts_rfc3339 nick uri text)) -(struct feed (nick uri)) +(struct msg + ( + ts_epoch ; Integer + ts_rfc3339 ; String + nick ; String + uri ; net/url-structs:url + text ; String + )) + +(struct feed + ( + nick ; String + uri ; net/url-structs:url + )) (define (concurrent-filter-map num_workers f xs) ; TODO preserve order of elements OR communicate that reorder is expected @@ -60,27 +76,31 @@ (λ (out-format color-i msg) (let ([color (vector-ref colors (modulo color-i n))] [nick (msg-nick msg)] - [uri (msg-uri msg)] + [uri (url->string (msg-uri msg))] [text (msg-text msg)]) (match out-format ['single-line (printf "~a \033[1;37m<~a>\033[0m \033[0;~am~a\033[0m~n" (parameterize ([date-display-format 'iso-8601]) - (date->string (seconds->date [msg-ts_epoch msg]) #t)) + (date->string (seconds->date [msg-ts_epoch msg]) #t)) nick color text)] ['multi-line (printf "~a~n\033[1;37m<~a ~a>\033[0m~n\033[0;~am~a\033[0m~n~n" (parameterize ([date-display-format 'rfc2822]) - (date->string (seconds->date [msg-ts_epoch msg]) #t)) + (date->string (seconds->date [msg-ts_epoch msg]) #t)) nick uri color text)]))))) +; TODO Implement rfc3339->epoch and remove dependency on rfc3339-old + (define str->msg (let ([re (pregexp "^([0-9]{4}-[0-9]{2}-[0-9]{2}T[0-9]{2}:[0-9]{2}(:[0-9]{2})?)(\\.[0-9]+)?([^\\s\t]*)[\\s\t]+(.*)$")]) (λ (nick uri str) (with-handlers* ([exn:fail? (λ (e) - (log-error "Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v" str nick uri e) + (log-error + "Failed to parse msg: ~v, from: ~v, at: ~v, because: ~v" + str nick (url->string uri) e) #f)]) (match (regexp-match re str) [(list _wholething ts s _fractional tz text) @@ -168,28 +188,85 @@ (close-input-port in) digest) -(define (uri-fetch use-cache uri) - (define cache-file-path - (expand-user-path (string-append "~/.tt/cache/" (hash-sha1 uri)))) - (if (and use-cache (file-exists? cache-file-path)) - (begin - (log-info "uri-fetch cached ~a" uri) - (file->string cache-file-path)) +(define (url->cache-file-path uri) + ; TODO Replace hashing with encoding + (expand-user-path (string-append "~/.tt/cache/" (hash-sha1 (url->string uri))))) + +(define (uri-read-cached uri) + (define path (url->cache-file-path uri)) + (if (file-exists? path) + (file->string path) (begin - (log-info "uri-fetch new ~a" uri) - ; TODO Timeout. Currently hangs on slow connections. - (let* ([resp (http-get uri)] - [status (http-response-code resp)] - [body (http-response-body resp)]) - (log-debug "finished GET ~a status:~a body length:~a" - uri status (string-length body)) - ; TODO Handle redirects - (if (= status 200) - (begin - (display-to-file body cache-file-path #:exists 'replace) - body) - ; TODO A more-informative exception - (raise status)))))) + (log-warning "Cache file not found for URI: ~a" (url->string uri)) + ""))) + +(define (str->feed str) + (log-debug "Parsing feed string: ~v" str) + (match (string-split str) + [(list nick u) + (with-handlers* + ([exn:fail? + (λ (e) + (log-error "Invalid URI: ~v, exn: ~v" u e) + #f)]) + (feed nick (string->url u)))] + [_ + (log-error "Invalid feed string: ~v" str) + #f])) + +(define (filter-comments lines) + (filter-not (λ (line) (string-prefix? line "#")) lines)) + +(define (str->feeds str) + (filter-map str->feed (filter-comments (str->lines str)))) + +(define (file->feeds filename) + (str->feeds (file->string filename))) + +(define user-agent + (let* + ([prog-name "tt"] + [prog-version ((info:get-info (list prog-name)) 'version)] + [prog-uri "https://github.com/xandkar/tt"] + [user-feed-file (expand-user-path "~/twtxt-me.txt")] + [user + (if (file-exists? user-feed-file) + (let ([user (first (file->feeds user-feed-file))]) + (format "+~a; @~a" (url->string (feed-uri user)) (feed-nick user))) + (format "+~a" prog-uri))]) + (format "~a/~a (~a)" prog-name prog-version user))) + +; uri-download : net/url-structs:url -> Void +(define (uri-download u) + (define cache-file-path (url->cache-file-path u)) + (log-debug "uri-download ~v into ~v" u cache-file-path) + (match* ((url-scheme u) (url-host u) (url-port u)) + [(s h p) + #:when (and s h) + (define ssl? (string=? s "https")) + (define-values (status-line headers body-input) + ; TODO Timeout. Currently hangs on slow connections. + (http-sendrecv + h + (url->string (struct-copy url u [scheme #f] [host #f])) + #:ssl? ssl? + #:port (cond [p p] [ssl? 443] [else 80]) + #:headers (list (format "User-Agent: ~a" user-agent)) + )) + (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 + (if (= 200 status) + (call-with-output-file cache-file-path + (λ (cache-output) + (copy-port body-input cache-output)) + #:exists 'replace) + (raise status))] + [(_ _ _) + (log-error "Invalid URI: ~v" u)])) (define (timeline-print out-format timeline) (void (foldl (match-lambda** @@ -200,58 +277,40 @@ (cons "" 0) timeline))) -(define (feed->msgs use-cache feed) - (log-info "downloading feed nick:~a uri:~a" - (feed-nick feed) - (feed-uri feed)) +; feed->msgs : Feed -> (Listof Msg) +(define (feed->msgs f) + (match-define (feed nick uri) f) + (log-info "Reading feed nick:~a uri:~v" nick uri) + (str->msgs nick uri (uri-read-cached uri))) + +; feed-download : Feed -> Void +(define (feed-download f) + (match-define (feed nick uri) f) + (log-info "Downloading feed nick:~a uri:~a" nick (url->string uri)) (with-handlers - ([exn:fail:network? + ([exn:fail? (λ (e) - (log-error "network error nick:~a uri:~a exn:~a" - (feed-nick feed) - (feed-uri feed) - e) + (log-error "Network error nick:~a uri:~v exn:~v" nick uri e) #f)] [integer? (λ (status) - (log-error "http error nick:~a uri:~a status:~a" - (feed-nick feed) - (feed-uri feed) - status) + (log-error "HTTP error nick:~a uri:~a status:~a" nick uri status) #f)]) - (define uri (feed-uri feed)) - (str->msgs [feed-nick feed] uri [uri-fetch use-cache uri]))) + (uri-download uri))) -; TODO timeline contract : time-sorted list of messages -(define (timeline use-cache num_workers feeds) - (sort (append* (concurrent-filter-map num_workers (curry feed->msgs use-cache) feeds)) - (λ (a b) [< (msg-ts_epoch a) (msg-ts_epoch b)]))) +; timeline-download : Integer -> (Listof Feed) -> Void +(define (timeline-download num_workers feeds) + ; TODO No need for map - can just iter + (void (concurrent-filter-map num_workers feed-download feeds))) -(define (str->feed str) - ; TODO validation - (define toks (string-split str)) - (apply feed toks)) - -(define (filter-comments lines) - (filter-not (λ (line) (string-prefix? line "#")) lines)) - -(define (str->feeds str) - (map str->feed (filter-comments (str->lines str)))) - -(define (file->feeds filename) - (str->feeds (file->string filename))) - -(define (user-agent prog-name prog-version) - (let* - ([prog-uri "https://github.com/xandkar/tt"] - [user-feed-file (expand-user-path "~/twtxt-me.txt")] - [user - (if (file-exists? user-feed-file) - (let ([user (first (file->feeds user-feed-file))]) - (format "+~a; @~a" (feed-uri user) (feed-nick user))) - (format "+~a" prog-uri))] - ) - (format "~a/~a (~a)" prog-name prog-version user))) +; TODO timeline contract : time-sorted list of messages +; timeline-read : (U 'old->new 'new->old) -> (Listof Feeds) -> (Listof Msg) +(define (timeline-read order feeds) + (define cmp (match order + ['old->new <] + ['new->old >])) + (sort (append* (filter-map feed->msgs feeds)) + (λ (a b) (cmp (msg-ts_epoch a) (msg-ts_epoch b))))) (define (start-logger level) (let* ([logger (make-logger #f #f level #f)] @@ -269,47 +328,61 @@ (current-logger logger))) (module+ main - (require setup/getinfo) - - (current-http-client/response-auto #f) - (let* ([prog-name "tt"] - [prog-version ((get-info (list prog-name)) 'version)] - [user-agent (user-agent prog-name prog-version)]) - (current-http-client/user-agent user-agent)) - (let* ([use-cache - #f] - [log-level - 'info] - [out-format - 'multi-line] - [num_workers - 15]) ; 15 was fastest out of the tried 1, 5, 10, 15 and 20. + (let ([log-level 'info]) (command-line + #:program + "tt" #:once-each - [("-c" "--cached") - "Read cached data instead of downloading." - (set! use-cache #t)] - [("-d" "--debug") "Enable debug log level." (set! log-level 'debug)] - - [("-j" "--jobs") - njobs "Number of concurrent jobs." - (set! num_workers (string->number njobs))] - - #:once-any - [("-s" "--short") - "Short output format" - (set! out-format 'single-line)] - - [("-l" "--long") - "Long output format" - (set! out-format 'multi-line)] - - #:args (filename) + #:help-labels + "" + "and is one of" + "r, read i : Read the timeline." + "d, download : Download the timeline." + "u, upload : Upload your twtxt file (alias to execute ~/.tt/upload)." + "" + #:args (command . args) (start-logger log-level) - (timeline-print out-format - (timeline use-cache - num_workers - (file->feeds filename)))))) + (current-command-line-arguments (list->vector args)) + (match command + [(or "d" "download") + (let ([num_workers 15]) ; 15 was fastest out of the tried: 1, 5, 10, 20. + (command-line + #:program + "tt download" + #:once-each + [("-j" "--jobs") + njobs "Number of concurrent jobs." + (set! num_workers (string->number njobs))] + #:args (filename) + (timeline-download num_workers (file->feeds filename))))] + [(or "u" "upload") + (command-line + #:program + "tt upload" + #:args () + (if (system (path->string (expand-user-path "~/.tt/upload"))) + (exit 0) + (exit 1)))] + [(or "r" "read") + (let ([out-format 'multi-line] + [order 'old->new]) + (command-line + #:program + "tt read" + #:once-each + [("-r" "--rev") + "Reverse displayed timeline order." + (set! order 'new->old)] + #:once-any + [("-s" "--short") + "Short output format" + (set! out-format 'single-line)] + [("-l" "--long") + "Long output format" + (set! out-format 'multi-line)] + #:args (filename) + (timeline-print out-format (timeline-read order (file->feeds filename)))))] + ))))