From 443f1eecbc4c15c8038920027e62fccbdcb0bbe7 Mon Sep 17 00:00:00 2001
From: Cadence Ember <cadence@disroot.org>
Date: Wed, 23 Oct 2024 22:52:00 +1300
Subject: [PATCH] Add user agent and detect blocked pages

---
 src/fandom-request.rkt | 40 +++++++++++++++++++++++++++++++++-------
 src/page-wiki.rkt      | 28 ++++++++++++++++++++++++++--
 2 files changed, 59 insertions(+), 9 deletions(-)

diff --git a/src/fandom-request.rkt b/src/fandom-request.rkt
index 966eeee..c306b04 100644
--- a/src/fandom-request.rkt
+++ b/src/fandom-request.rkt
@@ -1,12 +1,16 @@
 #lang typed/racket/base
-(require "config.rkt"
+(require racket/format
+         racket/string
+         "config.rkt"
          "../lib/url-utils.rkt")
 (define-type Headers (HashTable Symbol (U Bytes String)))
 (require/typed net/http-easy
   [#:opaque Timeout-Config timeout-config?]
   [#:opaque Response response?]
   [#:opaque Session session?]
+  [response-status-code (Response -> Natural)]
   [current-session (Parameter Session)]
+  [current-user-agent (Parameter (U Bytes String))]
   [make-timeout-config ([#:lease Positive-Real] [#:connect Positive-Real] -> Timeout-Config)]
   [get ((U Bytes String)
         [#:close? Boolean]
@@ -22,19 +26,41 @@
  fandom-get-api
  timeouts)
 
+(unless (string-contains? (~a (current-user-agent)) "BreezeWiki")
+  (current-user-agent
+   (format "BreezeWiki/1.0 (~a) ~a"
+           (if (config-true? 'canonical_origin)
+               (config-get 'canonical_origin)
+               "local")
+           (current-user-agent))))
+
 (define timeouts (make-timeout-config #:lease 5 #:connect 5))
 
+(: last-failure Flonum)
+(define last-failure 0.0)
+(: stored-failure (Option Response))
+(define stored-failure #f)
+(define failure-persist-time 30000)
+
 (: no-headers Headers)
 (define no-headers '#hasheq())
 
 (: fandom-get (String String [#:headers (Option Headers)] -> Response))
 (define (fandom-get wikiname path #:headers [headers #f])
-  (define dest-url (string-append "https://www.fandom.com" path))
-  (define host (string-append wikiname ".fandom.com"))
-  (log-outgoing wikiname path)
-  (get dest-url
-       #:timeouts timeouts
-       #:headers (hash-set (or headers no-headers) 'Host host)))
+  (or
+   (and ((current-inexact-milliseconds) . < . (+ last-failure failure-persist-time)) stored-failure)
+   (let ()
+     (define dest-url (string-append "https://www.fandom.com" path))
+     (define host (string-append wikiname ".fandom.com"))
+     (log-outgoing wikiname path)
+     (define res
+       (get dest-url
+            #:timeouts timeouts
+            #:headers (hash-set (or headers no-headers) 'Host host)))
+     (when (memq (response-status-code res) '(403 406))
+       (set! last-failure (current-inexact-milliseconds))
+       (set! stored-failure res))
+     res)))
 
 (: fandom-get-api (String (Listof (Pair String String)) [#:headers (Option Headers)] -> Response))
 (define (fandom-get-api wikiname params #:headers [headers #f])
diff --git a/src/page-wiki.rkt b/src/page-wiki.rkt
index f16792c..da63617 100644
--- a/src/page-wiki.rkt
+++ b/src/page-wiki.rkt
@@ -18,6 +18,7 @@
          "config.rkt"
          "data.rkt"
          "fandom-request.rkt"
+         "../lib/archive-file-mappings.rkt"
          "../lib/pure-utils.rkt"
          "../lib/syntax.rkt"
          "../lib/thread-utils.rkt"
@@ -37,8 +38,9 @@
 
 (define (page-wiki req)
   (define wikiname (path/param-path (first (url-path (request-uri req)))))
+  (define segments (map path/param-path (cdr (url-path (request-uri req)))))
   (define user-cookies (user-cookies-getter req))
-  (define path (string-join (map path/param-path (cddr (url-path (request-uri req)))) "/"))
+  (define path (string-join (cdr segments) "/"))
   (define source-url (format "https://~a.fandom.com/wiki/~a" wikiname path))
 
   (define-values (dest-res siteinfo)
@@ -101,9 +103,31 @@
                (write-html body out))))))]
     [(eq? 404 (easy:response-status-code dest-res))
      (next-dispatcher)]
+    [(memq (easy:response-status-code dest-res) '(403 406))
+     (response-handler
+      (define body
+        (generate-wiki-page
+         `(div
+           (p "Sorry! Fandom isn't allowing BreezeWiki to show pages right now.")
+           (p "We'll automatically try again in 30 seconds, so please stay on this page and be patient.")
+           (p (small "In a hurry? " (a (@ (href ,source-url)) "Click here to read the page on Fandom."))))
+         #:req req
+         #:source-url source-url
+         #:wikiname wikiname
+         #:title (url-segments->guess-title segments)
+         #:siteinfo siteinfo))
+      (response/output
+       #:code 503
+       #:headers (build-headers
+                  always-headers
+                  (header #"Retry-After" #"30")
+                  (header #"Cache-Control" #"max-age=30, public")
+                  (header #"Refresh" #"35"))
+       (λ (out)
+         (write-html body out))))]
     [else
      (response-handler
-      (error 'page-wiki "Tried to load page ~a/~v~nSadly, the page didn't load because Fandom returned status code ~a with response:~n~a"
+      (error 'page-wiki "Tried to load page ~a/~a~nSadly, the page didn't load because Fandom returned status code ~a with response:~n~a"
              wikiname
              path
              (easy:response-status-code dest-res)