Compare commits

..

6 commits

Author SHA1 Message Date
5ad5326425
Add support for File: pages
Fixes https://lists.sr.ht/~cadence/breezewiki-discuss/%3Cb2835a70-5118-4df0-90c9-4333486a4b69%40nixnetmail.com%3E
2022-10-09 22:05:28 +07:00
5813c49261
Refactor Referrer-Policy header
- Change the variable to always-headers so it can be extended in the future
- New function build-headers that assists combining complex logic headers
  together with less mess
- Also apply headers to the proxy
2022-10-09 23:45:55 +13:00
adc4b47b83 Set Referrer-Policy to no-referrer
Fandom sends a fake 404 to media if there's a Referer header that has an origin
that's not Fandom. However, we can choose not to send the header by setting
Referrer-Policy. See also:
https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Referrer-Policy
2022-10-09 10:45:48 +00:00
ade7878f7b
Redirect to actual wiki main page 2022-10-09 22:54:59 +13:00
59332fd9d1
Pass siteinfo through code; show sitename in title 2022-10-09 22:50:50 +13:00
9aba3ad432
Refactor siteinfo/license fetching 2022-10-09 20:54:49 +13:00
11 changed files with 82 additions and 49 deletions

View file

@ -1,3 +1,3 @@
#lang info #lang info
(define build-deps '("rackunit-lib" "web-server-lib" "http-easy-lib" "html-parsing" "html-writing" "json-pointer" "ini-lib")) (define build-deps '("rackunit-lib" "web-server-lib" "http-easy-lib" "html-parsing" "html-writing" "json-pointer" "ini-lib" "memo"))

View file

@ -9,8 +9,8 @@
"url-utils.rkt") "url-utils.rkt")
(provide (provide
; header to not send referers to fandom ; headers to always send on all http responses
referrer-policy always-headers
; timeout durations for http-easy requests ; timeout durations for http-easy requests
timeouts timeouts
; generates a consistent footer ; generates a consistent footer
@ -24,7 +24,8 @@
(require rackunit (require rackunit
html-writing)) html-writing))
(define referrer-policy (header #"Referrer-Policy" #"no-referrer")) (define always-headers
(list (header #"Referrer-Policy" #"same-origin"))) ; header to not send referers to fandom
(define timeouts (easy:make-timeout-config #:lease 5 #:connect 5)) (define timeouts (easy:make-timeout-config #:lease 5 #:connect 5))
(define (application-footer source-url #:license [license-in #f]) (define (application-footer source-url #:license [license-in #f])
@ -51,8 +52,8 @@
,(if source-url ,(if source-url
`(div (p "This page displays proxied content from " `(div (p "This page displays proxied content from "
(a (@ (href ,source-url) (rel "noreferrer")) ,source-url) (a (@ (href ,source-url) (rel "noreferrer")) ,source-url)
,(format ". Text content is available under the ~a license, " (license-text license)) ,(format ". Text content is available under the ~a license, " (license^-text license))
(a (@ (href ,(license-url license))) "see license info.") (a (@ (href ,(license^-url license))) "see license info.")
" Media files may have different copying restrictions.") " Media files may have different copying restrictions.")
(p ,(format "Fandom is a trademark of Fandom, Inc. ~a is not affiliated with Fandom." (config-get 'application_name)))) (p ,(format "Fandom is a trademark of Fandom, Inc. ~a is not affiliated with Fandom." (config-get 'application_name))))
`(div (p "Text content on wikis run by Fandom is available under the Creative Commons Attribution-Share Alike License 3.0 (Unported), " `(div (p "Text content on wikis run by Fandom is available under the Creative Commons Attribution-Share Alike License 3.0 (Unported), "
@ -66,7 +67,8 @@
#:wikiname wikiname #:wikiname wikiname
#:title title #:title title
#:body-class [body-class-in #f] #:body-class [body-class-in #f]
#:license [license #f]) #:siteinfo [siteinfo-in #f])
(define siteinfo (or siteinfo-in siteinfo-default))
(define body-class (if (not body-class-in) (define body-class (if (not body-class-in)
"skin-fandomdesktop" "skin-fandomdesktop"
body-class-in)) body-class-in))
@ -85,7 +87,10 @@
`(html `(html
(head (head
(meta (@ (name "viewport") (content "width=device-width, initial-scale=1"))) (meta (@ (name "viewport") (content "width=device-width, initial-scale=1")))
(title ,(format "~a | ~a" title (config-get 'application_name))) (title ,(format "~a | ~a+~a"
title
(regexp-replace #rx" ?Wiki$" (siteinfo^-sitename siteinfo) "")
(config-get 'application_name)))
,@(map (λ (url) ,@(map (λ (url)
`(link (@ (rel "stylesheet") (type "text/css") (href ,url)))) `(link (@ (rel "stylesheet") (type "text/css") (href ,url))))
(required-styles (format "https://~a.fandom.com" wikiname))) (required-styles (format "https://~a.fandom.com" wikiname)))
@ -104,7 +109,7 @@
(div (@ (id "content") #;(class "page-content")) (div (@ (id "content") #;(class "page-content"))
(div (@ (id "mw-content-text")) (div (@ (id "mw-content-text"))
,content)) ,content))
,(application-footer source-url #:license license))))))) ,(application-footer source-url #:license (siteinfo^-license siteinfo))))))))
(module+ test (module+ test
(define page (define page
(parameterize ([(config-parameter 'strict_proxy) "true"]) (parameterize ([(config-parameter 'strict_proxy) "true"])

View file

@ -1,33 +1,36 @@
#lang racket/base #lang racket/base
(require (prefix-in easy: net/http-easy) (require racket/list
(prefix-in easy: net/http-easy)
memo
"url-utils.rkt" "url-utils.rkt"
"xexpr-utils.rkt") "xexpr-utils.rkt")
(provide (provide
(struct-out license) (struct-out siteinfo^)
license-default (struct-out license^)
license-auto) siteinfo-fetch
siteinfo-default
license-default)
(struct license (text url) #:transparent) (struct siteinfo^ (sitename basepage license) #:transparent)
(define license-default (license "CC-BY-SA" "https://www.fandom.com/licensing")) (struct license^ (text url) #:transparent)
(define license-hash (make-hash))
(define (license-fetch wikiname) (define license-default (license^ "CC-BY-SA" "https://www.fandom.com/licensing"))
(define siteinfo-default (siteinfo^ "Test Wiki" "Main_Page" license-default))
(define/memoize (siteinfo-fetch wikiname) #:hash hash
(define dest-url (define dest-url
(format "https://~a.fandom.com/api.php?~a" (format "https://~a.fandom.com/api.php?~a"
wikiname wikiname
(params->query '(("action" . "query") (params->query '(("action" . "query")
("meta" . "siteinfo") ("meta" . "siteinfo")
("siprop" . "rightsinfo") ("siprop" . "general|rightsinfo")
("format" . "json") ("format" . "json")
("formatversion" . "2"))))) ("formatversion" . "2")))))
(log-outgoing dest-url) (log-outgoing dest-url)
(define res (easy:get dest-url)) (define res (easy:get dest-url))
(define data (easy:response-json res)) (define data (easy:response-json res))
(license (jp "/query/rightsinfo/text" data) (siteinfo^ (jp "/query/general/sitename" data)
(jp "/query/rightsinfo/url" data))) (second (regexp-match #rx"/wiki/(.*)" (jp "/query/general/base" data)))
(define (license-auto wikiname) (license^ (jp "/query/rightsinfo/text" data)
(if (hash-has-key? license-hash wikiname) (jp "/query/rightsinfo/url" data))))
(hash-ref license-hash wikiname)
(let ([result (license-fetch wikiname)])
(hash-set! license-hash wikiname result)
result)))

View file

@ -35,14 +35,14 @@
#:members-data members-data #:members-data members-data
#:page page #:page page
#:body-class [body-class #f] #:body-class [body-class #f]
#:license [license #f]) #:siteinfo [siteinfo #f])
(define members (jp "/query/categorymembers" members-data)) (define members (jp "/query/categorymembers" members-data))
(generate-wiki-page (generate-wiki-page
#:source-url source-url #:source-url source-url
#:wikiname wikiname #:wikiname wikiname
#:title title #:title title
#:body-class body-class #:body-class body-class
#:license license #:siteinfo siteinfo
`(div `(div
,(update-tree-wiki page wikiname) ,(update-tree-wiki page wikiname)
(hr) (hr)
@ -89,7 +89,7 @@
(log-outgoing dest-url) (log-outgoing dest-url)
(define dest-res (easy:get dest-url #:timeouts timeouts)) (define dest-res (easy:get dest-url #:timeouts timeouts))
(easy:response-json dest-res)] (easy:response-json dest-res)]
[license (license-auto wikiname)]) [siteinfo (siteinfo-fetch wikiname)])
(define title (preprocess-html-wiki (jp "/parse/title" page-data prefixed-category))) (define title (preprocess-html-wiki (jp "/parse/title" page-data prefixed-category)))
(define page-html (preprocess-html-wiki (jp "/parse/text" page-data ""))) (define page-html (preprocess-html-wiki (jp "/parse/text" page-data "")))
@ -105,7 +105,7 @@
#:members-data members-data #:members-data members-data
#:page page #:page page
#:body-class body-class #:body-class body-class
#:license license)) #:siteinfo siteinfo))
(when (config-true? 'debug) (when (config-true? 'debug)
; used for its side effects ; used for its side effects
@ -113,7 +113,7 @@
(xexp->html body)) (xexp->html body))
(response/output (response/output
#:code 200 #:code 200
#:headers (list referrer-policy) #:headers (build-headers always-headers)
(λ (out) (λ (out)
(write-html body out)))))) (write-html body out))))))
(module+ test (module+ test

View file

@ -57,7 +57,7 @@
#:title title #:title title
#:media-detail media-detail #:media-detail media-detail
#:image-content-type image-content-type #:image-content-type image-content-type
#:license [license #f]) #:siteinfo [siteinfo #f])
(define video-embed-code (jp "/videoEmbedCode" media-detail "")) (define video-embed-code (jp "/videoEmbedCode" media-detail ""))
(define raw-image-url (jp "/rawImageUrl" media-detail)) (define raw-image-url (jp "/rawImageUrl" media-detail))
(define image-url (jp "/imageUrl" media-detail raw-image-url)) (define image-url (jp "/imageUrl" media-detail raw-image-url))
@ -72,7 +72,7 @@
#:source-url source-url #:source-url source-url
#:wikiname wikiname #:wikiname wikiname
#:title title #:title title
#:license license #:siteinfo siteinfo
`(div ,(if (non-empty-string? video-embed-code) `(div ,(if (non-empty-string? video-embed-code)
(update-tree-wiki (html->xexp (preprocess-html-wiki video-embed-code)) wikiname) (update-tree-wiki (html->xexp (preprocess-html-wiki video-embed-code)) wikiname)
(get-media-html image-url image-content-type)) (get-media-html image-url image-content-type))
@ -113,7 +113,7 @@
(log-outgoing dest-url) (log-outgoing dest-url)
(define dest-res (easy:get dest-url #:timeouts timeouts)) (define dest-res (easy:get dest-url #:timeouts timeouts))
(easy:response-json dest-res)] (easy:response-json dest-res)]
[license (license-auto wikiname)]) [siteinfo (siteinfo-fetch wikiname)])
(if (not (jp "/exists" media-detail #f)) (if (not (jp "/exists" media-detail #f))
(next-dispatcher) (next-dispatcher)
(response-handler (response-handler
@ -130,7 +130,7 @@
#:title title #:title title
#:media-detail media-detail #:media-detail media-detail
#:image-content-type image-content-type #:image-content-type image-content-type
#:license license)) #:siteinfo siteinfo))
(when (config-true? 'debug) (when (config-true? 'debug)
; used for its side effects ; used for its side effects
; convert to string with error checking, error will be raised if xexp is invalid ; convert to string with error checking, error will be raised if xexp is invalid

View file

@ -4,6 +4,7 @@
html-writing html-writing
web-server/http web-server/http
"application-globals.rkt" "application-globals.rkt"
"url-utils.rkt"
"xexpr-utils.rkt" "xexpr-utils.rkt"
"config.rkt") "config.rkt")
@ -81,6 +82,7 @@
(define (page-home req) (define (page-home req)
(response/output (response/output
#:code 200 #:code 200
#:headers (build-headers always-headers)
(λ (out) (λ (out)
(write-html body out)))) (write-html body out))))

View file

@ -8,6 +8,7 @@
net/url net/url
web-server/http web-server/http
(only-in web-server/dispatchers/dispatch next-dispatcher) (only-in web-server/dispatchers/dispatch next-dispatcher)
"application-globals.rkt"
"url-utils.rkt" "url-utils.rkt"
"xexpr-utils.rkt") "xexpr-utils.rkt")
@ -26,6 +27,7 @@
(response/output (response/output
#:code (easy:response-status-code dest-r) #:code (easy:response-status-code dest-r)
#:mime-type (easy:response-headers-ref dest-r 'content-type) #:mime-type (easy:response-headers-ref dest-r 'content-type)
#:headers (build-headers always-headers)
(λ (out) (λ (out)
(copy-port (easy:response-output dest-r) out) (copy-port (easy:response-output dest-r) out)
(easy:response-close! dest-r)))))) (easy:response-close! dest-r))))))

View file

@ -2,6 +2,7 @@
(require net/url (require net/url
web-server/http web-server/http
"application-globals.rkt" "application-globals.rkt"
"data.rkt"
"url-utils.rkt" "url-utils.rkt"
"xexpr-utils.rkt") "xexpr-utils.rkt")
@ -11,5 +12,6 @@
(define (redirect-wiki-home req) (define (redirect-wiki-home req)
(response-handler (response-handler
(define wikiname (path/param-path (car (url-path (request-uri req))))) (define wikiname (path/param-path (car (url-path (request-uri req)))))
(define dest (format "/~a/wiki/Main_Page" wikiname)) (define siteinfo (siteinfo-fetch wikiname))
(define dest (format "/~a/wiki/~a" wikiname (or (siteinfo^-basepage siteinfo) "Main_Page")))
(generate-redirect dest))) (generate-redirect dest)))

View file

@ -25,13 +25,13 @@
(define search-json-data (define search-json-data
'#hasheq((batchcomplete . #t) (query . #hasheq((search . (#hasheq((ns . 0) (pageid . 219) (size . 1482) (snippet . "") (timestamp . "2022-08-21T08:54:23Z") (title . "Gacha Capsule") (wordcount . 214)) #hasheq((ns . 0) (pageid . 201) (size . 1198) (snippet . "") (timestamp . "2022-07-11T17:52:47Z") (title . "Badges") (wordcount . 181))))))))) '#hasheq((batchcomplete . #t) (query . #hasheq((search . (#hasheq((ns . 0) (pageid . 219) (size . 1482) (snippet . "") (timestamp . "2022-08-21T08:54:23Z") (title . "Gacha Capsule") (wordcount . 214)) #hasheq((ns . 0) (pageid . 201) (size . 1198) (snippet . "") (timestamp . "2022-07-11T17:52:47Z") (title . "Badges") (wordcount . 181)))))))))
(define (generate-results-page dest-url wikiname query data #:license [license #f]) (define (generate-results-page dest-url wikiname query data #:siteinfo [siteinfo #f])
(define search-results (jp "/query/search" data)) (define search-results (jp "/query/search" data))
(generate-wiki-page (generate-wiki-page
#:source-url dest-url #:source-url dest-url
#:wikiname wikiname #:wikiname wikiname
#:title "Search Results" #:title "Search Results"
#:license license #:siteinfo siteinfo
`(div (@ (class "mw-parser-output")) `(div (@ (class "mw-parser-output"))
(p ,(format "~a results found for " (length search-results)) (p ,(format "~a results found for " (length search-results))
(strong ,query)) (strong ,query))
@ -70,18 +70,18 @@
(thread-let (thread-let
([dest-res (log-outgoing dest-url) ([dest-res (log-outgoing dest-url)
(easy:get dest-url #:timeouts timeouts)] (easy:get dest-url #:timeouts timeouts)]
[license (license-auto wikiname)]) [siteinfo (siteinfo-fetch wikiname)])
(define data (easy:response-json dest-res)) (define data (easy:response-json dest-res))
(define body (generate-results-page dest-url wikiname query data #:license license)) (define body (generate-results-page dest-url wikiname query data #:siteinfo siteinfo))
(when (config-true? 'debug) (when (config-true? 'debug)
; used for its side effects ; used for its side effects
; convert to string with error checking, error will be raised if xexp is invalid ; convert to string with error checking, error will be raised if xexp is invalid
(xexp->html body)) (xexp->html body))
(response/output (response/output
#:code 200 #:code 200
#:headers (list referrer-policy) #:headers (build-headers always-headers)
(λ (out) (λ (out)
(write-html body out)))))) (write-html body out))))))
(module+ test (module+ test

View file

@ -252,7 +252,7 @@
("format" . "json"))))) ("format" . "json")))))
(log-outgoing dest-url) (log-outgoing dest-url)
(easy:get dest-url #:timeouts timeouts)] (easy:get dest-url #:timeouts timeouts)]
[license (license-auto wikiname)]) [siteinfo (siteinfo-fetch wikiname)])
(cond (cond
[(eq? 200 (easy:response-status-code dest-res)) [(eq? 200 (easy:response-status-code dest-res))
@ -275,13 +275,15 @@
#:wikiname wikiname #:wikiname wikiname
#:title title #:title title
#:body-class body-class #:body-class body-class
#:license license)) #:siteinfo siteinfo))
(define redirect-msg ((query-selector (attribute-selector 'class "redirectMsg") body))) (define redirect-msg ((query-selector (attribute-selector 'class "redirectMsg") body)))
(define headers (if redirect-msg (define headers
(build-headers
always-headers
(when redirect-msg
(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))])
(list (header #"Refresh" value) referrer-policy)) (header #"Refresh" value)))))
(list referrer-policy)))
(when (config-true? 'debug) (when (config-true? 'debug)
; used for its side effects ; used for its side effects
; convert to string with error checking, error will be raised if xexp is invalid ; convert to string with error checking, error will be raised if xexp is invalid

View file

@ -2,6 +2,8 @@
(require racket/string (require racket/string
"config.rkt" "config.rkt"
"pure-utils.rkt") "pure-utils.rkt")
(require/typed web-server/http/request-structs
[#:opaque Header header?])
(provide (provide
; regex to match wiki names ; regex to match wiki names
@ -13,7 +15,9 @@
; check whether a url is on a domain controlled by fandom ; check whether a url is on a domain controlled by fandom
is-fandom-url? is-fandom-url?
; prints "out: <url>" ; prints "out: <url>"
log-outgoing) log-outgoing
; pass in a header, headers, or something useless. they'll all combine into a list
build-headers)
(module+ test (module+ test
(require "typed-rackunit.rkt")) (require "typed-rackunit.rkt"))
@ -81,3 +85,16 @@
(define (log-outgoing url-string) (define (log-outgoing url-string)
(when (config-true? 'log_outgoing) (when (config-true? 'log_outgoing)
(printf "out: ~a~n" url-string))) (printf "out: ~a~n" url-string)))
(: build-headers ((U Header (Listof Header) False Void) * -> (Listof Header)))
(define (build-headers . fs)
(apply
append
(map (λ ([f : (U Header (Listof Header) False Void)])
(cond
[(not f) null]
[(void? f) null]
[(null? f) null]
[(header? f) (list f)]
[(pair? f) f]))
fs)))