forked from cadence/breezewiki
Compare commits
4 commits
a661ddb313
...
f216a1996a
| Author | SHA1 | Date | |
|---|---|---|---|
f216a1996a
|
|||
|
99b5d8d8f6
|
|||
|
9fd2b4699d
|
|||
|
ba3b39242c
|
5 changed files with 281 additions and 232 deletions
|
|
@ -91,5 +91,5 @@
|
||||||
#:max-age (* 60 60 24 365 10)))))
|
#:max-age (* 60 60 24 365 10)))))
|
||||||
|
|
||||||
(define (user-cookies-setter-url req new-settings)
|
(define (user-cookies-setter-url req new-settings)
|
||||||
(format "/set-user-settings?~a" (params->query `(("ref" . ,(url->string (request-uri req)))
|
(format "/set-user-settings?~a" (params->query `(("next_location" . ,(url->string (request-uri req)))
|
||||||
("new_settings" . ,(format "~a" new-settings))))))
|
("new_settings" . ,(format "~a" new-settings))))))
|
||||||
|
|
|
||||||
|
|
@ -12,7 +12,7 @@
|
||||||
|
|
||||||
(define (page-set-user-settings req)
|
(define (page-set-user-settings req)
|
||||||
(response-handler
|
(response-handler
|
||||||
(define ref (dict-ref (url-query (request-uri req)) 'ref))
|
(define next-location (dict-ref (url-query (request-uri req)) 'next_location))
|
||||||
(define new-settings (read (open-input-string (dict-ref (url-query (request-uri req)) 'new_settings))))
|
(define new-settings (read (open-input-string (dict-ref (url-query (request-uri req)) 'new_settings))))
|
||||||
(define headers (user-cookies-setter new-settings))
|
(define headers (user-cookies-setter new-settings))
|
||||||
(generate-redirect ref #:headers headers)))
|
(generate-redirect next-location #:headers headers)))
|
||||||
|
|
|
||||||
|
|
@ -19,6 +19,7 @@
|
||||||
"data.rkt"
|
"data.rkt"
|
||||||
"pure-utils.rkt"
|
"pure-utils.rkt"
|
||||||
"syntax.rkt"
|
"syntax.rkt"
|
||||||
|
"tree-updater.rkt"
|
||||||
"xexpr-utils.rkt"
|
"xexpr-utils.rkt"
|
||||||
"url-utils.rkt")
|
"url-utils.rkt")
|
||||||
|
|
||||||
|
|
@ -30,51 +31,19 @@
|
||||||
preprocess-html-wiki)
|
preprocess-html-wiki)
|
||||||
|
|
||||||
(module+ test
|
(module+ test
|
||||||
(require rackunit)
|
(require rackunit))
|
||||||
(define wiki-document
|
|
||||||
'(*TOP*
|
|
||||||
(div (@ (class "mw-parser-output"))
|
|
||||||
(aside (@ (role "region") (class "portable-infobox pi-theme-wikia pi-layout-default"))
|
|
||||||
(h2 (@ (class "pi-item pi-title") (data-source "title"))
|
|
||||||
"Infobox Title")
|
|
||||||
(figure (@ (class "pi-item pi-image") (data-source "image"))
|
|
||||||
(a (@ (href "https://static.wikia.nocookie.net/nice-image.png") (class "image image-thumbnail") (title ""))
|
|
||||||
(img (@ (src "https://static.wikia.nocookie.net/nice-image-thumbnail.png") (class "pi-image-thumbnail")))))
|
|
||||||
(div (@ (class "pi-item pi-data") (data-source "description"))
|
|
||||||
(h3 (@ (class "pi-data-label"))
|
|
||||||
"Description")
|
|
||||||
(div (@ (class "pi-data-value"))
|
|
||||||
"Mystery infobox!")))
|
|
||||||
(div (@ (data-test-collapsesection) (class "collapsible collapsetoggle-inline collapsed"))
|
|
||||||
(i (b "This section is hidden for dramatic effect."))
|
|
||||||
(div (@ (class "collapsible-content"))
|
|
||||||
(p "Another page link: "
|
|
||||||
(a (@ (data-test-wikilink) (href "https://test.fandom.com/wiki/Another_Page") (title "Another Page"))
|
|
||||||
"Another Page"))))
|
|
||||||
(figure (@ (class "thumb tnone"))
|
|
||||||
(a (@ (href "https://static.wikia.nocookie.net/nice-image.png") (class "image") (data-test-figure-a))
|
|
||||||
(img (@ (src "data:image/gif;base64,R0lGODlhAQABAIABAAAAAP///yH5BAEAAAEALAAAAAABAAEAQAICTAEAOw%3D%3D")
|
|
||||||
(data-src "https://static.wikia.nocookie.net/nice-image-thumbnail.png")
|
|
||||||
(class "thumbimage lazyload"))))
|
|
||||||
(noscript
|
|
||||||
(a (@ (href "https://static.wikia.nocookie.net/nice-image.png") (class "image"))
|
|
||||||
(img (@ (src "https://static.wikia.nocookie.net/nice-image-thumbnail.png")
|
|
||||||
(data-src "https://static.wikia.nocookie.net/nice-image-thumbnail.png")
|
|
||||||
(class "thumbimage")))))
|
|
||||||
(figcaption "Test figure!"))
|
|
||||||
(iframe (@ (src "https://example.com/iframe-src")))))))
|
|
||||||
|
|
||||||
(define (preprocess-html-wiki html)
|
(define (preprocess-html-wiki html)
|
||||||
(define (rr* find replace contents)
|
(define ((rr* find replace) contents)
|
||||||
(regexp-replace* find contents replace))
|
(regexp-replace* find contents replace))
|
||||||
((compose1
|
((compose1
|
||||||
; fix navbox list nesting
|
; fix navbox list nesting
|
||||||
; navbox on right of page has incorrect html "<td ...><li>" and the xexpr parser puts the <li> much further up the tree
|
; navbox on right of page has incorrect html "<td ...><li>" and the xexpr parser puts the <li> much further up the tree
|
||||||
; add a <ul> to make the parser happy
|
; add a <ul> to make the parser happy
|
||||||
; usage: /fallout/wiki/Fallout:_New_Vegas_achievements_and_trophies
|
; usage: /fallout/wiki/Fallout:_New_Vegas_achievements_and_trophies
|
||||||
(curry rr* #rx"(<td[^>]*>\n?)(<li>)" "\\1<ul>\\2")
|
(rr* #rx"(<td[^>]*>\n?)(<li>)" "\\1<ul>\\2")
|
||||||
; change <figcaption><p> to <figcaption><span> to make the parser happy
|
; change <figcaption><p> to <figcaption><span> to make the parser happy
|
||||||
(curry rr* #rx"(<figcaption[^>]*>)[ \t]*<p class=\"caption\">([^<]*)</p>" "\\1<span class=\"caption\">\\2</span>"))
|
(rr* #rx"(<figcaption[^>]*>)[ \t]*<p class=\"caption\">([^<]*)</p>" "\\1<span class=\"caption\">\\2</span>"))
|
||||||
html))
|
html))
|
||||||
(module+ test
|
(module+ test
|
||||||
(check-equal? (preprocess-html-wiki "<td class=\"va-navbox-column\" style=\"width: 33%\">\n<li>Hey</li>")
|
(check-equal? (preprocess-html-wiki "<td class=\"va-navbox-column\" style=\"width: 33%\">\n<li>Hey</li>")
|
||||||
|
|
@ -82,198 +51,6 @@
|
||||||
(check-equal? (preprocess-html-wiki "<figure class=\"thumb tright\" style=\"width: 150px\"><a class=\"image\"><img></a><noscript><a><img></a></noscript><figcaption class=\"thumbcaption\"> <p class=\"caption\">Caption text.</p></figcaption></figure>")
|
(check-equal? (preprocess-html-wiki "<figure class=\"thumb tright\" style=\"width: 150px\"><a class=\"image\"><img></a><noscript><a><img></a></noscript><figcaption class=\"thumbcaption\"> <p class=\"caption\">Caption text.</p></figcaption></figure>")
|
||||||
"<figure class=\"thumb tright\" style=\"width: 150px\"><a class=\"image\"><img></a><noscript><a><img></a></noscript><figcaption class=\"thumbcaption\"><span class=\"caption\">Caption text.</span></figcaption></figure>"))
|
"<figure class=\"thumb tright\" style=\"width: 150px\"><a class=\"image\"><img></a><noscript><a><img></a></noscript><figcaption class=\"thumbcaption\"><span class=\"caption\">Caption text.</span></figcaption></figure>"))
|
||||||
|
|
||||||
(define (update-tree-wiki tree wikiname)
|
|
||||||
(update-tree
|
|
||||||
(λ (element element-type attributes children)
|
|
||||||
;; replace whole element?
|
|
||||||
(cond
|
|
||||||
; wrap tables in a div.table-scroller
|
|
||||||
[(and (eq? element-type 'table)
|
|
||||||
(has-class? "wikitable" attributes)
|
|
||||||
(not (dict-has-key? attributes 'data-scrolling)))
|
|
||||||
`(div
|
|
||||||
((class "table-scroller"))
|
|
||||||
((,element-type (@ (data-scrolling) ,@attributes)
|
|
||||||
,@children)))]
|
|
||||||
; exclude empty figcaptions
|
|
||||||
[(and (eq? element-type 'figcaption)
|
|
||||||
(or (eq? (length (filter element-is-element? children)) 0)
|
|
||||||
((query-selector (λ (element-type attributes children)
|
|
||||||
(eq? element-type 'use))
|
|
||||||
element))))
|
|
||||||
return-no-element]
|
|
||||||
; exclude infobox items that are videos, and gallery items that are videos
|
|
||||||
[(and (or (has-class? "pi-item" attributes)
|
|
||||||
(has-class? "wikia-gallery-item" attributes))
|
|
||||||
((query-selector (λ (element-type attributes children)
|
|
||||||
(has-class? "video-thumbnail" attributes))
|
|
||||||
element)))
|
|
||||||
return-no-element]
|
|
||||||
; exclude the invisible brackets after headings
|
|
||||||
[(and (eq? element-type 'span)
|
|
||||||
(has-class? "mw-editsection" attributes))
|
|
||||||
return-no-element]
|
|
||||||
; display a link instead of an iframe
|
|
||||||
[(eq? element-type 'iframe)
|
|
||||||
(define src (car (dict-ref attributes 'src null)))
|
|
||||||
`(a
|
|
||||||
((class "iframe-alternative") (href ,src))
|
|
||||||
(,(format "Embedded media: ~a" src)))]
|
|
||||||
; remove noscript versions of images because they are likely lower quality than the script versions
|
|
||||||
[(and (eq? element-type 'noscript)
|
|
||||||
(match children
|
|
||||||
; either the noscript has a.image as a first child...
|
|
||||||
[(list (list 'a (list '@ a-att ...) _)) (has-class? "image" a-att)]
|
|
||||||
; or the noscript has img as a first child
|
|
||||||
[(list (list 'img _)) #t]
|
|
||||||
[_ #f]))
|
|
||||||
return-no-element]
|
|
||||||
[#t
|
|
||||||
(list element-type
|
|
||||||
;; attributes
|
|
||||||
((compose1
|
|
||||||
; uncollapsing
|
|
||||||
(curry attribute-maybe-update 'class
|
|
||||||
(λ (class)
|
|
||||||
(string-join
|
|
||||||
((compose1
|
|
||||||
; uncollapse all navbox items (bottom of page mass navigation)
|
|
||||||
(curry u
|
|
||||||
(λ (classlist) (and (eq? element-type 'table)
|
|
||||||
(member "navbox" classlist)
|
|
||||||
(member "collapsed" classlist)))
|
|
||||||
(λ (classlist) (filter (curry (negate equal?) "collapsed") classlist)))
|
|
||||||
; uncollapse portable-infobox sections
|
|
||||||
(curry u
|
|
||||||
(λ (classlist) (and (eq? element-type 'section)
|
|
||||||
(member "pi-collapse" classlist)))
|
|
||||||
(λ (classlist) (filter (λ (v)
|
|
||||||
(and (not (equal? v "pi-collapse-closed"))
|
|
||||||
(not (equal? v "pi-collapse"))))
|
|
||||||
classlist)))
|
|
||||||
; generic: includes article sections and tables, probably more
|
|
||||||
(curry u
|
|
||||||
(λ (classlist) (and (member "collapsible" classlist)
|
|
||||||
(member "collapsed" classlist)))
|
|
||||||
(λ (classlist) (filter (curry (negate equal?) "collapsed") classlist))))
|
|
||||||
(string-split class " "))
|
|
||||||
" ")))
|
|
||||||
; change links to stay on the same wiki
|
|
||||||
(curry attribute-maybe-update 'href
|
|
||||||
(λ (href)
|
|
||||||
((compose1
|
|
||||||
(λ (href) (regexp-replace #rx"^(/wiki/.*)" href (format "/~a\\1" wikiname)))
|
|
||||||
(λ (href) (regexp-replace (pregexp (format "^https://(~a)\\.fandom\\.com(/wiki/.*)" px-wikiname)) href "/\\1\\2")))
|
|
||||||
href)))
|
|
||||||
; add noreferrer to a.image
|
|
||||||
(curry u
|
|
||||||
(λ (v) (and (eq? element-type 'a)
|
|
||||||
(has-class? "image" v)))
|
|
||||||
(λ (v) (dict-update v 'rel (λ (s)
|
|
||||||
(list (string-append (car s) " noreferrer")))
|
|
||||||
'(""))))
|
|
||||||
; proxy images from inline styles, if strict_proxy is set
|
|
||||||
(curry u
|
|
||||||
(λ (v) (config-true? 'strict_proxy))
|
|
||||||
(λ (v) (attribute-maybe-update
|
|
||||||
'style
|
|
||||||
(λ (style)
|
|
||||||
(regexp-replace #rx"url\\(['\"]?(.*?)['\"]?\\)" style
|
|
||||||
(λ (whole url)
|
|
||||||
(string-append
|
|
||||||
"url("
|
|
||||||
(u-proxy-url url)
|
|
||||||
")")))) v)))
|
|
||||||
; and also their links, if strict_proxy is set
|
|
||||||
(curry u
|
|
||||||
(λ (v)
|
|
||||||
(and (config-true? 'strict_proxy)
|
|
||||||
(eq? element-type 'a)
|
|
||||||
(or (has-class? "image-thumbnail" v)
|
|
||||||
(has-class? "image" v))))
|
|
||||||
(λ (v) (attribute-maybe-update 'href u-proxy-url v)))
|
|
||||||
; proxy images from src attributes, if strict_proxy is set
|
|
||||||
(curry u
|
|
||||||
(λ (v) (config-true? 'strict_proxy))
|
|
||||||
(λ (v) (attribute-maybe-update 'src u-proxy-url v)))
|
|
||||||
; don't lazyload images
|
|
||||||
(curry u
|
|
||||||
(λ (v) (dict-has-key? v 'data-src))
|
|
||||||
(λ (v) (attribute-maybe-update 'src (λ (_) (car (dict-ref v 'data-src))) v)))
|
|
||||||
; don't use srcset - TODO: use srcset?
|
|
||||||
(λ (v) (dict-remove v 'srcset)))
|
|
||||||
attributes)
|
|
||||||
;; children
|
|
||||||
((compose1
|
|
||||||
; more uncollapsing - sample: bandori/wiki/BanG_Dream!_Wikia
|
|
||||||
(curry u
|
|
||||||
(λ (v) (has-class? "mw-collapsible-content" attributes))
|
|
||||||
(λ (v) (for/list ([element v])
|
|
||||||
(u (λ (element) (pair? element))
|
|
||||||
(λ (element)
|
|
||||||
`(,(car element)
|
|
||||||
(@ ,@(attribute-maybe-update 'style (λ (a) (regexp-replace #rx"display: *none" a "display:inline")) (bits->attributes element)))
|
|
||||||
,@(filter element-is-content? (cdr element))))
|
|
||||||
element))))
|
|
||||||
; wrap blinking animated images in a slot so they can be animated with CSS
|
|
||||||
(curry u
|
|
||||||
(λ (v) (and (has-class? "animated" attributes)
|
|
||||||
((length v) . > . 1)))
|
|
||||||
(λ (v)
|
|
||||||
`((span (@ (class "animated-slot__outer") (style ,(format "--steps: ~a" (length v))))
|
|
||||||
(span (@ (class "animated-slot__inner"))
|
|
||||||
,@v))))))
|
|
||||||
children))]))
|
|
||||||
tree))
|
|
||||||
(module+ test
|
|
||||||
(define transformed
|
|
||||||
(parameterize ([(config-parameter 'strict_proxy) "true"])
|
|
||||||
(update-tree-wiki wiki-document "test")))
|
|
||||||
; check that wikilinks are changed to be local
|
|
||||||
(check-equal? (get-attribute 'href (bits->attributes
|
|
||||||
((query-selector
|
|
||||||
(λ (t a c) (dict-has-key? a 'data-test-wikilink))
|
|
||||||
transformed))))
|
|
||||||
"/test/wiki/Another_Page")
|
|
||||||
; check that a.image has noreferrer
|
|
||||||
(check-equal? (get-attribute 'rel (bits->attributes
|
|
||||||
((query-selector
|
|
||||||
(λ (t a c) (and (eq? t 'a)
|
|
||||||
(has-class? "image" a)))
|
|
||||||
transformed))))
|
|
||||||
" noreferrer")
|
|
||||||
; check that article collapse sections become uncollapsed
|
|
||||||
(check-equal? (get-attribute 'class (bits->attributes
|
|
||||||
((query-selector
|
|
||||||
(λ (t a c) (dict-has-key? a 'data-test-collapsesection))
|
|
||||||
transformed))))
|
|
||||||
"collapsible collapsetoggle-inline")
|
|
||||||
; check that iframes are gone
|
|
||||||
(check-false ((query-selector (λ (t a c) (eq? t 'iframe)) transformed)))
|
|
||||||
(check-equal? (let* ([alternative ((query-selector (λ (t a c) (has-class? "iframe-alternative" a)) transformed))]
|
|
||||||
[link ((query-selector (λ (t a c) (eq? t 'a)) alternative))])
|
|
||||||
(get-attribute 'href (bits->attributes link)))
|
|
||||||
"https://example.com/iframe-src")
|
|
||||||
; check that images are proxied
|
|
||||||
(check-equal? (get-attribute 'src (bits->attributes
|
|
||||||
((query-selector
|
|
||||||
(λ (t a c) (eq? t 'img))
|
|
||||||
transformed))))
|
|
||||||
"/proxy?dest=https%3A%2F%2Fstatic.wikia.nocookie.net%2Fnice-image-thumbnail.png")
|
|
||||||
; check that links to images are proxied
|
|
||||||
(check-equal? (get-attribute 'href (bits->attributes
|
|
||||||
((query-selector
|
|
||||||
(λ (t a c) (and (eq? t 'a) (has-class? "image-thumbnail" a)))
|
|
||||||
transformed))))
|
|
||||||
"/proxy?dest=https%3A%2F%2Fstatic.wikia.nocookie.net%2Fnice-image.png")
|
|
||||||
(check-equal? (get-attribute 'href (bits->attributes
|
|
||||||
((query-selector
|
|
||||||
(λ (t a c) (member '(data-test-figure-a) a))
|
|
||||||
transformed))))
|
|
||||||
"/proxy?dest=https%3A%2F%2Fstatic.wikia.nocookie.net%2Fnice-image.png")
|
|
||||||
; check that noscript images are removed
|
|
||||||
(check-equal? ((query-selector (λ (t a c) (eq? t 'noscript)) transformed)) #f))
|
|
||||||
|
|
||||||
(define (page-wiki req)
|
(define (page-wiki req)
|
||||||
(define wikiname (path/param-path (first (url-path (request-uri req)))))
|
(define wikiname (path/param-path (first (url-path (request-uri req)))))
|
||||||
(define user-cookies (user-cookies-getter req))
|
(define user-cookies (user-cookies-getter req))
|
||||||
|
|
@ -317,10 +94,14 @@
|
||||||
#:head-data head-data
|
#:head-data head-data
|
||||||
#:siteinfo siteinfo))
|
#:siteinfo siteinfo))
|
||||||
(define redirect-msg ((query-selector (attribute-selector 'class "redirectMsg") body)))
|
(define redirect-msg ((query-selector (attribute-selector 'class "redirectMsg") body)))
|
||||||
|
(define redirect-query-parameter (dict-ref (url-query (request-uri req)) 'redirect "yes"))
|
||||||
(define headers
|
(define headers
|
||||||
(build-headers
|
(build-headers
|
||||||
always-headers
|
always-headers
|
||||||
(when redirect-msg
|
; redirect-query-parameter: only the string "no" is significant:
|
||||||
|
; https://github.com/Wikia/app/blob/fe60579a53f16816d65dad1644363160a63206a6/includes/Wiki.php#L367
|
||||||
|
(when (and redirect-msg
|
||||||
|
(not (equal? redirect-query-parameter "no")))
|
||||||
(let* ([dest (get-attribute 'href (bits->attributes ((query-selector (λ (t a c) (eq? t 'a)) redirect-msg))))]
|
(let* ([dest (get-attribute 'href (bits->attributes ((query-selector (λ (t a c) (eq? t 'a)) redirect-msg))))]
|
||||||
[value (bytes-append #"0;url=" (string->bytes/utf-8 dest))])
|
[value (bytes-append #"0;url=" (string->bytes/utf-8 dest))])
|
||||||
(header #"Refresh" value)))))
|
(header #"Refresh" value)))))
|
||||||
|
|
|
||||||
267
src/tree-updater.rkt
Normal file
267
src/tree-updater.rkt
Normal file
|
|
@ -0,0 +1,267 @@
|
||||||
|
#lang racket/base
|
||||||
|
(require racket/dict
|
||||||
|
racket/function
|
||||||
|
racket/match
|
||||||
|
racket/string
|
||||||
|
"config.rkt"
|
||||||
|
"pure-utils.rkt"
|
||||||
|
"url-utils.rkt"
|
||||||
|
"xexpr-utils.rkt")
|
||||||
|
|
||||||
|
(provide
|
||||||
|
update-tree-wiki)
|
||||||
|
|
||||||
|
(module+ test
|
||||||
|
(require rackunit
|
||||||
|
html-parsing)
|
||||||
|
(define wiki-document
|
||||||
|
'(*TOP*
|
||||||
|
(div (@ (class "mw-parser-output"))
|
||||||
|
(aside (@ (role "region") (class "portable-infobox pi-theme-wikia pi-layout-default"))
|
||||||
|
(h2 (@ (class "pi-item pi-title") (data-source "title"))
|
||||||
|
"Infobox Title")
|
||||||
|
(figure (@ (class "pi-item pi-image") (data-source "image"))
|
||||||
|
(a (@ (href "https://static.wikia.nocookie.net/nice-image.png") (class "image image-thumbnail") (title ""))
|
||||||
|
(img (@ (src "https://static.wikia.nocookie.net/nice-image-thumbnail.png") (class "pi-image-thumbnail")))))
|
||||||
|
(div (@ (class "pi-item pi-data") (data-source "description"))
|
||||||
|
(h3 (@ (class "pi-data-label"))
|
||||||
|
"Description")
|
||||||
|
(div (@ (class "pi-data-value"))
|
||||||
|
"Mystery infobox!")))
|
||||||
|
(div (@ (data-test-collapsesection) (class "collapsible collapsetoggle-inline collapsed"))
|
||||||
|
(i (b "This section is hidden for dramatic effect."))
|
||||||
|
(div (@ (class "collapsible-content"))
|
||||||
|
(p "Another page link: "
|
||||||
|
(a (@ (data-test-wikilink) (href "https://test.fandom.com/wiki/Another_Page") (title "Another Page"))
|
||||||
|
"Another Page"))))
|
||||||
|
(figure (@ (class "thumb tnone"))
|
||||||
|
(a (@ (href "https://static.wikia.nocookie.net/nice-image.png") (class "image") (data-test-figure-a))
|
||||||
|
(img (@ (src "data:image/gif;base64,R0lGODlhAQABAIABAAAAAP///yH5BAEAAAEALAAAAAABAAEAQAICTAEAOw%3D%3D")
|
||||||
|
(data-src "https://static.wikia.nocookie.net/nice-image-thumbnail.png")
|
||||||
|
(class "thumbimage lazyload"))))
|
||||||
|
(noscript
|
||||||
|
(a (@ (href "https://static.wikia.nocookie.net/nice-image.png") (class "image"))
|
||||||
|
(img (@ (src "https://static.wikia.nocookie.net/nice-image-thumbnail.png")
|
||||||
|
(data-src "https://static.wikia.nocookie.net/nice-image-thumbnail.png")
|
||||||
|
(class "thumbimage")))))
|
||||||
|
(figcaption "Test figure!"))
|
||||||
|
(iframe (@ (src "https://example.com/iframe-src")))))))
|
||||||
|
|
||||||
|
(define (updater wikiname)
|
||||||
|
(define classlist-updater
|
||||||
|
(compose1
|
||||||
|
; uncollapse all navbox items (bottom of page mass navigation)
|
||||||
|
(curry u
|
||||||
|
(λ (classlist) (and ; removed due to scoping, would improve peformance (eq? element-type 'table)
|
||||||
|
(member "navbox" classlist)
|
||||||
|
(member "collapsed" classlist)))
|
||||||
|
(λ (classlist) (filter (curry (negate equal?) "collapsed") classlist)))
|
||||||
|
; uncollapse portable-infobox sections
|
||||||
|
(curry u
|
||||||
|
(λ (classlist) (and ; removed due to scoping, would improve performance (eq? element-type 'section)
|
||||||
|
(member "pi-collapse" classlist)))
|
||||||
|
(λ (classlist) (filter (λ (v)
|
||||||
|
(and (not (equal? v "pi-collapse-closed"))
|
||||||
|
(not (equal? v "pi-collapse"))))
|
||||||
|
classlist)))
|
||||||
|
; generic: includes article sections and tables, probably more
|
||||||
|
(curry u
|
||||||
|
(λ (classlist) (and (member "collapsible" classlist)
|
||||||
|
(member "collapsed" classlist)))
|
||||||
|
(λ (classlist) (filter (curry (negate equal?) "collapsed") classlist)))))
|
||||||
|
|
||||||
|
(define ((string-replace-curried from to) str)
|
||||||
|
(string-replace str from to))
|
||||||
|
|
||||||
|
(define class-updater
|
||||||
|
(compose1
|
||||||
|
(string-replace-curried " collapsed" "")
|
||||||
|
(string-replace-curried "pi-collapse-closed" "")
|
||||||
|
(string-replace-curried "pi-collapse" "")))
|
||||||
|
|
||||||
|
(define attributes-updater
|
||||||
|
(compose1
|
||||||
|
; uncollapsing
|
||||||
|
#;(curry attribute-maybe-update 'class
|
||||||
|
(λ (class) (string-join (classlist-updater (string-split class " ")) " ")))
|
||||||
|
(curry attribute-maybe-update 'class class-updater)
|
||||||
|
; change links to stay on the same wiki
|
||||||
|
(curry attribute-maybe-update 'href
|
||||||
|
(λ (href)
|
||||||
|
((compose1
|
||||||
|
(λ (href) (regexp-replace #rx"^(/wiki/.*)" href (format "/~a\\1" wikiname)))
|
||||||
|
(λ (href) (regexp-replace (pregexp (format "^https://(~a)\\.fandom\\.com(/wiki/.*)" px-wikiname)) href "/\\1\\2")))
|
||||||
|
href)))
|
||||||
|
; add noreferrer to a.image
|
||||||
|
(curry u
|
||||||
|
(λ (v) (and #;(eq? element-type 'a)
|
||||||
|
(has-class? "image" v)))
|
||||||
|
(λ (v) (dict-update v 'rel (λ (s)
|
||||||
|
(list (string-append (car s) " noreferrer")))
|
||||||
|
'(""))))
|
||||||
|
; proxy images from inline styles, if strict_proxy is set
|
||||||
|
(curry u
|
||||||
|
(λ (v) (config-true? 'strict_proxy))
|
||||||
|
(λ (v) (attribute-maybe-update
|
||||||
|
'style
|
||||||
|
(λ (style)
|
||||||
|
(regexp-replace #rx"url\\(['\"]?(.*?)['\"]?\\)" style
|
||||||
|
(λ (whole url)
|
||||||
|
(string-append
|
||||||
|
"url("
|
||||||
|
(u-proxy-url url)
|
||||||
|
")")))) v)))
|
||||||
|
; and also their links, if strict_proxy is set
|
||||||
|
(curry u
|
||||||
|
(λ (v)
|
||||||
|
(and (config-true? 'strict_proxy)
|
||||||
|
#;(eq? element-type 'a)
|
||||||
|
(or (has-class? "image-thumbnail" v)
|
||||||
|
(has-class? "image" v))))
|
||||||
|
(λ (v) (attribute-maybe-update 'href u-proxy-url v)))
|
||||||
|
; proxy images from src attributes, if strict_proxy is set
|
||||||
|
(curry u
|
||||||
|
(λ (v) (config-true? 'strict_proxy))
|
||||||
|
(λ (v) (attribute-maybe-update 'src u-proxy-url v)))
|
||||||
|
; don't lazyload images
|
||||||
|
(curry u
|
||||||
|
(λ (v) (dict-has-key? v 'data-src))
|
||||||
|
(λ (v) (attribute-maybe-update 'src (λ (_) (car (dict-ref v 'data-src))) v)))
|
||||||
|
; don't use srcset - TODO: use srcset?
|
||||||
|
(λ (v) (dict-remove v 'srcset))))
|
||||||
|
|
||||||
|
(define (children-updater attributes children)
|
||||||
|
; more uncollapsing - sample: bandori/wiki/BanG_Dream!_Wikia
|
||||||
|
((λ (children)
|
||||||
|
(u
|
||||||
|
(λ (v) (has-class? "mw-collapsible-content" attributes))
|
||||||
|
(λ (v) (for/list ([element v])
|
||||||
|
(u (λ (element) (pair? element))
|
||||||
|
(λ (element)
|
||||||
|
`(,(car element)
|
||||||
|
(@ ,@(attribute-maybe-update 'style (λ (a) (regexp-replace #rx"display: *none" a "display:inline")) (bits->attributes element)))
|
||||||
|
,@(filter element-is-content? (cdr element))))
|
||||||
|
element)))
|
||||||
|
children))
|
||||||
|
; wrap blinking animated images in a slot so they can be animated with CSS
|
||||||
|
((λ (children)
|
||||||
|
(u
|
||||||
|
(λ (v) (and (has-class? "animated" attributes)
|
||||||
|
((length v) . > . 1)))
|
||||||
|
(λ (v)
|
||||||
|
`((span (@ (class "animated-slot__outer") (style ,(format "--steps: ~a" (length v))))
|
||||||
|
(span (@ (class "animated-slot__inner"))
|
||||||
|
,@v))))
|
||||||
|
children))
|
||||||
|
children)))
|
||||||
|
|
||||||
|
(define (updater element element-type attributes children)
|
||||||
|
;; replace whole element?
|
||||||
|
(cond
|
||||||
|
; wrap tables in a div.table-scroller
|
||||||
|
[(and (eq? element-type 'table)
|
||||||
|
(has-class? "wikitable" attributes)
|
||||||
|
(not (dict-has-key? attributes 'data-scrolling)))
|
||||||
|
`(div
|
||||||
|
((class "table-scroller"))
|
||||||
|
((,element-type (@ (data-scrolling) ,@attributes)
|
||||||
|
,@children)))]
|
||||||
|
; exclude empty figcaptions
|
||||||
|
[(and (eq? element-type 'figcaption)
|
||||||
|
(or (eq? (length (filter element-is-element? children)) 0)
|
||||||
|
((query-selector (λ (element-type attributes children)
|
||||||
|
(eq? element-type 'use))
|
||||||
|
element))))
|
||||||
|
return-no-element]
|
||||||
|
; exclude infobox items that are videos, and gallery items that are videos
|
||||||
|
[(and (or (has-class? "pi-item" attributes)
|
||||||
|
(has-class? "wikia-gallery-item" attributes))
|
||||||
|
((query-selector (λ (element-type attributes children)
|
||||||
|
(has-class? "video-thumbnail" attributes))
|
||||||
|
element)))
|
||||||
|
return-no-element]
|
||||||
|
; exclude the invisible brackets after headings
|
||||||
|
[(and (eq? element-type 'span)
|
||||||
|
(has-class? "mw-editsection" attributes))
|
||||||
|
return-no-element]
|
||||||
|
; display a link instead of an iframe
|
||||||
|
[(eq? element-type 'iframe)
|
||||||
|
(define src (car (dict-ref attributes 'src null)))
|
||||||
|
`(a
|
||||||
|
((class "iframe-alternative") (href ,src))
|
||||||
|
(,(format "Embedded media: ~a" src)))]
|
||||||
|
; remove noscript versions of images because they are likely lower quality than the script versions
|
||||||
|
[(and (eq? element-type 'noscript)
|
||||||
|
(match children
|
||||||
|
; either the noscript has a.image as a first child...
|
||||||
|
[(list (list 'a (list '@ a-att ...) _)) (has-class? "image" a-att)]
|
||||||
|
; or the noscript has img as a first child
|
||||||
|
[(list (list 'img _)) #t]
|
||||||
|
[_ #f]))
|
||||||
|
return-no-element]
|
||||||
|
[#t
|
||||||
|
(list element-type
|
||||||
|
;; attributes
|
||||||
|
(attributes-updater #; element-type attributes)
|
||||||
|
;; children
|
||||||
|
(children-updater attributes children))]))
|
||||||
|
|
||||||
|
updater)
|
||||||
|
|
||||||
|
(define (update-tree-wiki tree wikiname)
|
||||||
|
(update-tree (updater wikiname) tree))
|
||||||
|
|
||||||
|
(module+ test
|
||||||
|
(define transformed
|
||||||
|
(parameterize ([(config-parameter 'strict_proxy) "true"])
|
||||||
|
(update-tree-wiki wiki-document "test")))
|
||||||
|
; check that wikilinks are changed to be local
|
||||||
|
(check-equal? (get-attribute 'href (bits->attributes
|
||||||
|
((query-selector
|
||||||
|
(λ (t a c) (dict-has-key? a 'data-test-wikilink))
|
||||||
|
transformed))))
|
||||||
|
"/test/wiki/Another_Page")
|
||||||
|
; check that a.image has noreferrer
|
||||||
|
(check-equal? (get-attribute 'rel (bits->attributes
|
||||||
|
((query-selector
|
||||||
|
(λ (t a c) (and (eq? t 'a)
|
||||||
|
(has-class? "image" a)))
|
||||||
|
transformed))))
|
||||||
|
" noreferrer")
|
||||||
|
; check that article collapse sections become uncollapsed
|
||||||
|
(check-equal? (get-attribute 'class (bits->attributes
|
||||||
|
((query-selector
|
||||||
|
(λ (t a c) (dict-has-key? a 'data-test-collapsesection))
|
||||||
|
transformed))))
|
||||||
|
"collapsible collapsetoggle-inline")
|
||||||
|
; check that iframes are gone
|
||||||
|
(check-false ((query-selector (λ (t a c) (eq? t 'iframe)) transformed)))
|
||||||
|
(check-equal? (let* ([alternative ((query-selector (λ (t a c) (has-class? "iframe-alternative" a)) transformed))]
|
||||||
|
[link ((query-selector (λ (t a c) (eq? t 'a)) alternative))])
|
||||||
|
(get-attribute 'href (bits->attributes link)))
|
||||||
|
"https://example.com/iframe-src")
|
||||||
|
; check that images are proxied
|
||||||
|
(check-equal? (get-attribute 'src (bits->attributes
|
||||||
|
((query-selector
|
||||||
|
(λ (t a c) (eq? t 'img))
|
||||||
|
transformed))))
|
||||||
|
"/proxy?dest=https%3A%2F%2Fstatic.wikia.nocookie.net%2Fnice-image-thumbnail.png")
|
||||||
|
; check that links to images are proxied
|
||||||
|
(check-equal? (get-attribute 'href (bits->attributes
|
||||||
|
((query-selector
|
||||||
|
(λ (t a c) (and (eq? t 'a) (has-class? "image-thumbnail" a)))
|
||||||
|
transformed))))
|
||||||
|
"/proxy?dest=https%3A%2F%2Fstatic.wikia.nocookie.net%2Fnice-image.png")
|
||||||
|
(check-equal? (get-attribute 'href (bits->attributes
|
||||||
|
((query-selector
|
||||||
|
(λ (t a c) (member '(data-test-figure-a) a))
|
||||||
|
transformed))))
|
||||||
|
"/proxy?dest=https%3A%2F%2Fstatic.wikia.nocookie.net%2Fnice-image.png")
|
||||||
|
; check that noscript images are removed
|
||||||
|
(check-equal? ((query-selector (λ (t a c) (eq? t 'noscript)) transformed)) #f)
|
||||||
|
; benchmark
|
||||||
|
(when (file-exists? "Frog.html2")
|
||||||
|
(with-input-from-file "Frog.html2"
|
||||||
|
(λ ()
|
||||||
|
(define tree (html->xexp (current-input-port)))
|
||||||
|
(time (length (update-tree-wiki tree "minecraft")))))))
|
||||||
|
|
@ -74,6 +74,7 @@ p {
|
||||||
margin-top: 30px;
|
margin-top: 30px;
|
||||||
padding-top: 20px;
|
padding-top: 20px;
|
||||||
border-top: 1px solid var(--theme-page-text-color);
|
border-top: 1px solid var(--theme-page-text-color);
|
||||||
|
word-break: break-word;
|
||||||
}
|
}
|
||||||
.custom-footer__cols {
|
.custom-footer__cols {
|
||||||
display: grid;
|
display: grid;
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue