Remove Gamespot reviews/ads #14

Merged
cadence merged 1 commit from blankie/breezewiki:gamespot-reviews into main 2023-05-24 20:07:47 +00:00

View file

@ -58,7 +58,9 @@
(data-src "https://static.wikia.nocookie.net/nice-image-thumbnail.png") (data-src "https://static.wikia.nocookie.net/nice-image-thumbnail.png")
(class "thumbimage"))))) (class "thumbimage")))))
(figcaption "Test figure!")) (figcaption "Test figure!"))
(iframe (@ (src "https://example.com/iframe-src"))))))) (iframe (@ (src "https://example.com/iframe-src")))
(div (@ (class "reviews"))
(header "GameSpot Expert Reviews"))))))
(define (updater wikiname #:strict-proxy? [strict-proxy? #f]) (define (updater wikiname #:strict-proxy? [strict-proxy? #f])
;; precompute wikiurl regex for efficency ;; precompute wikiurl regex for efficency
@ -238,6 +240,9 @@
[(list (list 'img _)) #t] [(list (list 'img _)) #t]
[_ #f])) [_ #f]))
return-no-element] return-no-element]
; remove gamespot reviews/ads
[(has-class? "reviews" attributes)
return-no-element]
[#t [#t
(list element-type (list element-type
;; attributes ;; attributes
@ -297,6 +302,8 @@
"/proxy?dest=https%3A%2F%2Fstatic.wikia.nocookie.net%2Fnice-image.png") "/proxy?dest=https%3A%2F%2Fstatic.wikia.nocookie.net%2Fnice-image.png")
; check that noscript images are removed ; check that noscript images are removed
(check-equal? ((query-selector (λ (t a c) (eq? t 'noscript)) transformed)) #f) (check-equal? ((query-selector (λ (t a c) (eq? t 'noscript)) transformed)) #f)
; check that gamespot reviews/ads are removed
(check-equal? ((query-selector (λ (t a c) (has-class? "reviews" a)) transformed)) #f)
; benchmark ; benchmark
(when (file-exists? "../storage/Frog.html") (when (file-exists? "../storage/Frog.html")
(with-input-from-file "../storage/Frog.html" (with-input-from-file "../storage/Frog.html"