X-Git-Url: https://git.xandkar.net/?a=blobdiff_plain;f=tt.rkt;h=bc8e8bc5c134b92f596b08b59bfc37ce3a8865a8;hb=refs%2Ftags%2F0.18.0;hp=358161600a8afc3d4060baf87d7d2c7666e04314;hpb=b0ff061a507b47a19115cf6838ceb400cb6a4962;p=tt.git

diff --git a/tt.rkt b/tt.rkt
index 3581616..bc8e8bc 100644
--- a/tt.rkt
+++ b/tt.rkt
@@ -3,10 +3,9 @@
 (require openssl/sha1)
 (require racket/date)
 (require
-  net/http-client
+  net/head
   net/uri-codec
-  net/url-string
-  net/url-structs)
+  net/url)
 
 (require (prefix-in info: "info.rkt"))
 
@@ -34,12 +33,13 @@
 
 (struct Peer
         ([nick : (Option String)]
-         [uri  : Url]))
+         [uri  : Url])
+        #:transparent)
 
 (: tt-home-dir Path-String)
 (define tt-home-dir (build-path (expand-user-path "~") ".tt"))
 
-(: concurrent-filter-map (∀ (α β) (-> Natural (-> α β) (Listof α))))
+(: 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
   ; TODO switch from mailboxes to channels
@@ -243,6 +243,8 @@
 (: cache-dir Path-String)
 (define cache-dir (build-path tt-home-dir "cache"))
 
+(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 (hash-sha1 str) : (-> String String)
@@ -250,13 +252,19 @@
     (define digest (sha1 in))
     (close-input-port in)
     digest)
-  (build-path cache-dir (hash-sha1 (url->string uri))))
+  (build-path cache-object-dir (hash-sha1 (url->string uri))))
 
 (: url->cache-file-path-v2 (-> Url Path-String))
 (define (url->cache-file-path-v2 uri)
-  (build-path cache-dir (uri-encode (url->string uri))))
+  (build-path cache-object-dir (uri-encode (url->string uri))))
+
+(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-file-path url->cache-file-path-v2)
+(define (url->cache-lmod-path uri)
+  (build-path cache-dir "lmods" (uri-encode (url->string uri))))
 
 ; TODO Return Option
 (: uri-read-cached (-> Url String))
@@ -271,6 +279,10 @@
         (log-warning "Cache file not found for URI: ~a" (url->string uri))
         "")))
 
+(: uri? (-> String Boolean))
+(define (uri? str)
+  (regexp-match? #rx"^[a-z]+://.*" (string-downcase str)))
+
 (: str->peer (String (Option Peer)))
 (define (str->peer str)
   (log-debug "Parsing peer string: ~v" str)
@@ -280,8 +292,8 @@
           (log-error "Invalid URI in string: ~v, exn: ~v" str e)
           #f)])
     (match (string-split str)
-      [(list u)   (Peer #f  (string->url u))]
-      [(list n u) (Peer  n  (string->url u))]
+      [(list u)   #:when (uri? u) (Peer #f  (string->url u))]
+      [(list n u) #:when (uri? u) (Peer  n  (string->url u))]
       [_
         (log-error "Invalid peer string: ~v" str)
         #f])))
@@ -295,9 +307,60 @@
 (define (str->peers str)
   (filter-map str->peer (filter-comments (str->lines str))))
 
+(: peers->file (-> (Listof Peers) Path-String Void))
+(define (peers->file peers path)
+  (display-lines-to-file
+    (map (match-lambda
+           [(Peer n u)
+            (format "~a~a" (if n (format "~a " n) "") (url->string u))])
+         peers)
+    path
+    #:exists 'replace))
+
 (: file->peers (-> Path-String (Listof Peer)))
-(define (file->peers filename)
-  (str->peers (file->string filename)))
+(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))
+        '())))
+
+(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")
+
+(: b->n (-> Bytes (Option Number)))
+(define (b->n b)
+  (string->number (bytes->string/utf-8 b)))
+
+(: mon->num (-> Bytes Natural))
+(define/match (mon->num mon)
+  [(#"Jan")  1]
+  [(#"Feb")  2]
+  [(#"Mar")  3]
+  [(#"Apr")  4]
+  [(#"May")  5]
+  [(#"Jun")  6]
+  [(#"Jul")  7]
+  [(#"Aug")  8]
+  [(#"Sep")  9]
+  [(#"Oct") 10]
+  [(#"Nov") 11]
+  [(#"Dec") 12])
+
+(: rfc2822->epoch (-> Bytes (Option Nonnegative-Integer)))
+(define (rfc2822->epoch timestamp)
+  (match (regexp-match re-rfc2822 timestamp)
+    [(list _ _ dd mo yyyy HH MM SS)
+     #:when (and dd mo yyyy HH MM SS)
+     (find-seconds (b->n SS)
+                   (b->n MM)
+                   (b->n HH)
+                   (b->n dd)
+                   (mon->num mo)
+                   (b->n yyyy)
+                   #f)]
+    [_
+      #f]))
 
 (: user-agent String)
 (define user-agent
@@ -314,38 +377,65 @@
            (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)
+    [(list val) val]
+    [_           #f]))
+
 (: uri-download (-> 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)
-         (begin
-           (make-parent-directory* cache-file-path)
-           (call-with-output-file cache-file-path
-                                  (curry copy-port body-input)
-                                  #:exists 'replace))
-         (raise status))]
-    [(_ _ _)
-     (log-error "Invalid URI: ~v" u)]))
+  (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))
+  (log-debug "uri-download ~v into ~v" u cached-object-path)
+  (define-values (status-line headers body-input)
+    ; TODO Timeout. Currently hangs on slow connections.
+    (http-sendrecv/url u #: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
+  (match status
+    [200
+      (let* ([etag      (header-get headers #"ETag")]
+             [lmod      (header-get headers #"Last-Modified")]
+             [lmod-curr (if lmod (rfc2822->epoch lmod) #f)]
+             [lmod-prev (if (file-exists? cached-lmod-path)
+                            (rfc2822->epoch (file->bytes cached-lmod-path))
+                            #f)])
+        (log-debug "lmod-curr:~v lmod-prev:~v" lmod-curr lmod-prev)
+        (unless (or (and etag
+                         (file-exists? cached-etag-path)
+                         (bytes=? etag (file->bytes cached-etag-path))
+                         (begin
+                           (log-info "ETags match, skipping the rest of ~v" (url->string u))
+                           #t))
+                    (and lmod-curr
+                         lmod-prev
+                         (<= lmod-curr lmod-prev)
+                         (begin
+                           (log-info "Last-Modified <= current skipping the rest of ~v" (url->string u))
+                           #t)))
+          (begin
+            (log-info
+              "Downloading the rest of ~v. ETag: ~a, Last-Modified: ~v"
+              (url->string u) 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))))
+        (close-input-port body-input))]
+    [_
+      (raise status)]))
 
 (: timeline-print (-> Out-Format (Listof Msg) Void))
 (define (timeline-print out-format timeline)
@@ -386,22 +476,44 @@
   ; TODO No need for map - can just iter
   (void (concurrent-filter-map num-workers peer-download peers)))
 
-; TODO timeline contract : time-sorted list of messages
-(: timeline-read (-> Timeline-Order (Listof Peer) (Listof Msg)))
-(define (timeline-read order peers)
+(define (uniq xs)
+  (set->list (list->set xs)))
+
+(: peers->timeline (-> (listof Peer) (listof Msg)))
+(define (peers->timeline peers)
+  (append* (filter-map peer->msgs peers)))
+
+(: timeline-sort (-> (listof Msg) timeline-order (Listof Msgs)))
+(define (timeline-sort msgs order)
   (define cmp (match order
                 ['old->new <]
                 ['new->old >]))
-  (sort (append* (filter-map peer->msgs peers))
-        (λ (a b) (cmp (Msg-ts-epoch a) (Msg-ts-epoch b)))))
+  (sort msgs (λ (a b) (cmp (Msg-ts-epoch a)
+                           (Msg-ts-epoch b)))))
+
+(: paths->peers (-> (Listof String) (Listof Peer)))
+(define (paths->peers paths)
+  (let* ([paths (match paths
+                  ['()
+                   (let ([peer-refs-file (build-path tt-home-dir "peers")])
+                     (log-debug
+                       "No peer ref file paths provided, defaulting to ~v"
+                       (path->string peer-refs-file))
+                     (list peer-refs-file))]
+                  [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))
+    peers))
 
 (: log-writer-stop (-> Thread Void))
 (define (log-writer-stop log-writer)
   (log-message (current-logger) 'fatal 'stop "Exiting." #f)
   (thread-wait log-writer))
 
-(: logger-start (-> Log-Level Thread))
-(define (logger-start level)
+(: log-writer-start (-> Log-Level Thread))
+(define (log-writer-start level)
   (let* ([logger
            (make-logger #f #f level #f)]
          [log-receiver
@@ -431,13 +543,14 @@
       #:help-labels
       ""
       "and <command> is one of"
-      "r, read i   : Read the timeline."
+      "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)."
+      "c, crawl    : Discover new peers mentioned by known peers (offline operation)."
       ""
       #:args (command . args)
-      (define log-writer (logger-start log-level))
+      (define log-writer (log-writer-start log-level))
       (current-command-line-arguments (list->vector args))
       (match command
         [(or "d" "download")
@@ -451,11 +564,13 @@
              [("-j" "--jobs")
               njobs "Number of concurrent jobs."
               (set! num-workers (string->number njobs))]
-             #:args (filename)
-             (define-values (_res _cpu real-ms _gc)
-               (time-apply timeline-download (list num-workers (file->peers filename))))
-             (log-info "Timeline downloaded in ~a seconds." (/ real-ms 1000.0))
-             (log-writer-stop log-writer)))]
+             #:args file-paths
+             (let ([peers (paths->peers file-paths)])
+               (define-values (_res _cpu real-ms _gc)
+                 (time-apply timeline-download (list num-workers peers)))
+               (log-info "Downloaded timelines from ~a peers in ~a seconds."
+                         (length peers)
+                         (/ real-ms 1000.0)))))]
         [(or "u" "upload")
          (command-line
            #:program
@@ -481,6 +596,48 @@
              [("-l" "--long")
               "Long output format"
               (set! out-format 'multi-line)]
-             #:args (filename)
-             (timeline-print out-format (timeline-read order (file->peers filename)))))]
-        ))))
+             #:args file-paths
+             (let* ([peers
+                      (paths->peers file-paths)]
+                    [timeline
+                      (timeline-sort (peers->timeline peers) order)])
+               (timeline-print out-format timeline))))]
+        [(or "c" "crawl")
+         (command-line
+           #:program
+           "tt crawl"
+           #:args file-paths
+           (let* ([peers-all-file
+                    (build-path tt-home-dir "peers-all")]
+                  [peers-mentioned-file
+                    (build-path tt-home-dir "peers-mentioned")]
+                  [peers
+                    (paths->peers
+                      (match file-paths
+                        ; TODO Refactor such that path->string not needed
+                        ['() (list (path->string peers-all-file))]
+                        [_   file-paths]))]
+                  [timeline
+                    (peers->timeline peers)]
+                  [peers-mentioned-curr
+                    (uniq (append* (map Msg-mentions timeline)))]
+                  [peers-mentioned-prev
+                    (file->peers peers-mentioned-file)]
+                  [peers-mentioned
+                    (uniq (append peers-mentioned-prev
+                                  peers-mentioned-curr))]
+                  [peers-all-prev
+                    (file->peers peers-all-file)]
+                  [peers-all
+                    (uniq (append peers
+                                  peers-mentioned
+                                  peers-all-prev))])
+             (peers->file peers-mentioned
+                          peers-mentioned-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)])
+      (log-writer-stop log-writer))))