mirror of
https://gitflic.ru/project/magnolia1234/bypass-paywalls-clean-filters.git
synced 2025-04-24 08:25:12 +00:00
527 lines
19 KiB
JavaScript
527 lines
19 KiB
JavaScript
// ==UserScript==
|
||
// @name Bypass Paywalls Clean - pl
|
||
// @version 3.9.2.3
|
||
// @description Bypass Paywalls of news sites
|
||
// @author magnolia1234
|
||
// @downloadURL https://gitflic.ru/project/magnolia1234/bypass-paywalls-clean-filters/blob/raw?file=userscript/bpc.pl.user.js
|
||
// @updateURL https://gitflic.ru/project/magnolia1234/bypass-paywalls-clean-filters/blob/raw?file=userscript/bpc.pl.user.js
|
||
// @homepageURL https://gitflic.ru/project/magnolia1234/bypass-paywalls-clean-filters
|
||
// @supportURL https://gitflic.ru/project/magnolia1234/bypass-paywalls-clean-filters
|
||
// @license MIT; https://gitflic.ru/project/magnolia1234/bypass-paywalls-clean-filters/blob/raw?file=LICENSE
|
||
// @match *://*.auto-swiat.pl/*
|
||
// @match *://*.businessinsider.com.pl/*
|
||
// @match *://*.forbes.pl/*
|
||
// @match *://*.komputerswiat.pl/*
|
||
// @match *://*.magazyn-kuchnia.pl/*
|
||
// @match *://*.newsweek.pl/*
|
||
// @match *://*.onet.pl/*
|
||
// @match *://*.parkiet.com/*
|
||
// @match *://*.pb.pl/*
|
||
// @match *://*.rp.pl/*
|
||
// @match *://*.wyborcza.biz/*
|
||
// @match *://*.wyborcza.pl/*
|
||
// @match *://*.wysokieobcasy.pl/*
|
||
// @connect archive.fo
|
||
// @connect archive.is
|
||
// @connect archive.li
|
||
// @connect archive.md
|
||
// @connect archive.ph
|
||
// @connect archive.vn
|
||
// @grant GM.xmlHttpRequest
|
||
// ==/UserScript==
|
||
|
||
(function() {
|
||
//'use strict';
|
||
|
||
var func_post;
|
||
var fetch_headers = {};
|
||
|
||
window.setTimeout(function () {
|
||
|
||
var domain;
|
||
var mobile = window.navigator.userAgent.toLowerCase().includes('mobile');
|
||
var csDoneOnce;
|
||
var cs_param = {};
|
||
|
||
var overlay = document.querySelector('body.didomi-popup-open');
|
||
if (overlay)
|
||
overlay.classList.remove('didomi-popup-open');
|
||
var ads = 'div.OUTBRAIN, div[id^="taboola-"], div.ad-container, div[class*="-ad-container"], div[class*="_ad-container"], div.arc_ad, div[id^="poool-"], amp-ad, amp-embed[type="mgid"], amp-embed[type="outbrain"], amp-embed[type="taboola"]';
|
||
hideDOMStyle(ads, 10);
|
||
|
||
var pl_ringier_domains = ['auto-swiat.pl', 'businessinsider.com.pl', 'forbes.pl', 'komputerswiat.pl', 'newsweek.pl', 'onet.pl'];
|
||
|
||
if (matchDomain('pb.pl')) {
|
||
let paywall = document.querySelector('div.paywall');
|
||
if (paywall) {
|
||
paywall.classList.remove('paywall');
|
||
let article_hidden = paywall.querySelector('section.o-article-content');
|
||
if (article_hidden)
|
||
article_hidden.removeAttribute('class');
|
||
let loader = document.querySelector('div.o-piano-template-loader-box');
|
||
removeDOMElement(loader);
|
||
}
|
||
}
|
||
|
||
else if (matchDomain(pl_ringier_domains)) {
|
||
if (matchDomain('businessinsider.com.pl')) {
|
||
let paywall = document.querySelector('div#content-premium-offer');
|
||
removeDOMElement(paywall);
|
||
}
|
||
let premium = document.querySelector('div.contentPremium[style]');
|
||
if (premium) {
|
||
premium.removeAttribute('class');
|
||
premium.removeAttribute('style');
|
||
}
|
||
let ads = 'div.adPlaceholder , div[class^="Ad"][class*="Placeholder_"], div[data-placeholder-caption], div[data-run-module$=".floatingAd"], aside[data-ad-container], aside.adsContainer, [class^="pwAds"], .hide-for-paying, div.onet-ad, div.bottomBar, ad-default';
|
||
hideDOMStyle(ads);
|
||
}
|
||
|
||
else if (matchDomain(['rp.pl', 'parkiet.com'])) {
|
||
let paywall = document.querySelector('div.paywallComp');
|
||
if (paywall) {
|
||
removeDOMElement(paywall);
|
||
let article = document.querySelector('div.article--content');
|
||
if (article) {
|
||
let url = window.location.href;
|
||
article.firstChild.before(googleSearchToolLink(url));
|
||
}
|
||
}
|
||
}
|
||
|
||
else if (matchDomain(['wyborcza.biz', 'wyborcza.pl', 'wysokieobcasy.pl', 'magazyn-kuchnia.pl'])) {
|
||
let url = window.location.href;
|
||
func_post = function () {
|
||
let empty_spans = document.querySelectorAll('figure > a > span:empty');
|
||
removeDOMElement(...empty_spans);
|
||
}
|
||
if (matchDomain(['wyborcza.biz', 'wyborcza.pl']))
|
||
getArchive(url, 'div.article--content-fadeout', {rm_attrib: 'class'}, 'div.container[class*="pt"]', '', 'div.body > div:not([style*="background-color:"]):not([old-position])');
|
||
else
|
||
getArchive(url, 'section.fade-out-article', {rm_attrib: 'class'}, 'article');
|
||
let ads = 'div[id^="adUnit"], div[id^="ads-"]';
|
||
hideDOMStyle(ads);
|
||
}
|
||
|
||
}, 1000);
|
||
|
||
// General Functions
|
||
|
||
function matchDomain(domains, hostname) {
|
||
var matched_domain = false;
|
||
if (!hostname)
|
||
hostname = window.location.hostname;
|
||
if (typeof domains === 'string')
|
||
domains = [domains];
|
||
domains.some(domain => (hostname === domain || hostname.endsWith('.' + domain)) && (matched_domain = domain));
|
||
return matched_domain;
|
||
}
|
||
|
||
function urlHost(url) {
|
||
if (/^http/.test(url)) {
|
||
try {
|
||
return new URL(url).hostname;
|
||
} catch (e) {
|
||
console.log(`url not valid: ${url} error: ${e}`);
|
||
}
|
||
}
|
||
return url;
|
||
}
|
||
|
||
function matchUrlDomain(domains, url) {
|
||
return matchDomain(domains, urlHost(url));
|
||
}
|
||
|
||
function setCookie(name, value, domain, path, days, localstorage_hold = false) {
|
||
var max_age = days * 24 * 60 * 60;
|
||
document.cookie = name + "=" + (value || "") + "; domain=" + domain + "; path=" + path + "; max-age=" + max_age;
|
||
if (!localstorage_hold)
|
||
window.localStorage.clear();
|
||
}
|
||
|
||
function removeDOMElement(...elements) {
|
||
for (let element of elements) {
|
||
if (element)
|
||
element.remove();
|
||
}
|
||
}
|
||
|
||
function hideDOMElement(...elements) {
|
||
for (let element of elements) {
|
||
if (element)
|
||
element.style = 'display:none !important;';
|
||
}
|
||
}
|
||
|
||
function hideDOMStyle(selector, id = 1) {
|
||
let style = document.querySelector('head > style#ext'+ id);
|
||
if (!style && document.head) {
|
||
let sheet = document.createElement('style');
|
||
sheet.id = 'ext' + id;
|
||
sheet.innerText = selector + ' {display: none !important;}';
|
||
document.head.appendChild(sheet);
|
||
}
|
||
}
|
||
|
||
function addStyle(css, id = 1) {
|
||
let style = document.querySelector('head > style#add'+ id);
|
||
if (!style && document.head) {
|
||
let sheet = document.createElement('style');
|
||
sheet.id = 'add' + id;
|
||
sheet.innerText = css;
|
||
document.head.appendChild(sheet);
|
||
}
|
||
}
|
||
|
||
function clearPaywall(paywall, paywall_action) {
|
||
if (paywall) {
|
||
if (!paywall_action)
|
||
removeDOMElement(...paywall);
|
||
else {
|
||
for (let elem of paywall) {
|
||
if (paywall_action.rm_class)
|
||
elem.classList.remove(paywall_action.rm_class);
|
||
else if (paywall_action.rm_attrib)
|
||
elem.removeAttribute(paywall_action.rm_attrib);
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
function waitDOMElement(selector, tagName = '', callback, multiple = false) {
|
||
new window.MutationObserver(function (mutations) {
|
||
for (let mutation of mutations) {
|
||
for (let node of mutation.addedNodes) {
|
||
if (!tagName || (node.tagName === tagName)) {
|
||
if (node.matches(selector)) {
|
||
callback(node);
|
||
if (!multiple)
|
||
this.disconnect();
|
||
}
|
||
}
|
||
}
|
||
}
|
||
}).observe(document, {
|
||
subtree: true,
|
||
childList: true
|
||
});
|
||
}
|
||
|
||
function waitDOMAttribute(selector, tagName = '', attributeName = '', callback, multiple = false) {
|
||
let targetNode = document.querySelector(selector);
|
||
if (!targetNode)
|
||
return;
|
||
new window.MutationObserver(function (mutations) {
|
||
for (let mutation of mutations) {
|
||
if (mutation.target.attributes[attributeName]) {
|
||
callback(mutation.target);
|
||
if (!multiple)
|
||
this.disconnect();
|
||
}
|
||
}
|
||
}).observe(targetNode, {
|
||
attributes: true,
|
||
attributeFilter: [attributeName]
|
||
});
|
||
}
|
||
|
||
function getSelectorLevel(selector) {
|
||
if (selector.replace(/,\s+/g, ',').match(/[>\s]+/))
|
||
selector = selector.replace(/,\s+/g, ',').split(',').map(x => x.match(/[>\s]+/) ? x + ', ' + x.split(/[>\s]+/).pop() : x).join(', ');
|
||
return selector;
|
||
}
|
||
|
||
function getArticleSrc(url, url_src, proxy, base64, selector, text_fail = '', selector_source = selector, selector_archive = selector) {
|
||
let url_fetch = url_src || url;
|
||
GM.xmlHttpRequest({
|
||
method: "GET",
|
||
url: url_fetch,
|
||
headers: fetch_headers,
|
||
onload: function (response) {
|
||
let html = response.responseText;
|
||
if (proxy && base64) {
|
||
html = decode_utf8(atob(html));
|
||
selector_source = 'body';
|
||
}
|
||
let recursive;
|
||
if (url.startsWith('https://archive.')) {
|
||
if (url_fetch.includes('/https')) {
|
||
if (html.includes('<div class="TEXT-BLOCK"')) {
|
||
url_src = html.split('<div class="TEXT-BLOCK"')[1].split('</div>')[0].split('href="')[1].split('"')[0];
|
||
getArticleSrc(url, url_src, proxy, base64, selector, text_fail, selector_source, selector_archive);
|
||
recursive = true;
|
||
} else
|
||
html = '';
|
||
}
|
||
}
|
||
if (!recursive)
|
||
replaceDomElementExtSrc(url, url_src, html, proxy, base64, selector, text_fail, selector_source, selector_archive);
|
||
}
|
||
});
|
||
}
|
||
|
||
function replaceDomElementExt(url, proxy, base64, selector, text_fail = '', selector_source = selector, selector_archive = selector) {
|
||
let article = document.querySelector(selector);
|
||
if (!article)
|
||
return;
|
||
if (proxy) {
|
||
if (!text_fail) {
|
||
if (url.startsWith('https://archive.'))
|
||
text_fail = 'BPC > Try for full article text (no need to report issue for external site):\r\n';
|
||
else if (!matchUrlDomain(window.location.hostname, url))
|
||
text_fail = 'BPC > failed to load from external site:\r\n';
|
||
}
|
||
getArticleSrc(url, '', proxy, base64, selector, text_fail, selector_source, selector_archive);
|
||
} else {
|
||
fetch(url, {headers: fetch_headers})
|
||
.then(response => {
|
||
let article = document.querySelector(selector);
|
||
if (response.ok) {
|
||
response.text().then(html => {
|
||
replaceDomElementExtSrc(url, '', html, false, base64, selector, text_fail, selector_source);
|
||
});
|
||
} else {
|
||
replaceTextFail(url, article, proxy, text_fail);
|
||
}
|
||
}).catch(function (err) {
|
||
replaceTextFail(url, article, proxy, text_fail);
|
||
});
|
||
}
|
||
}
|
||
|
||
var selector_level = false;
|
||
function replaceDomElementExtSrc(url, url_src, html, proxy, base64, selector, text_fail = '', selector_source = selector, selector_archive = selector) {
|
||
let article = document.querySelector(selector);
|
||
let article_link = document.querySelector(selector_archive);
|
||
let no_content_msg = ' | no article content found! | :';
|
||
if (html) {
|
||
if (!proxy && base64) {
|
||
html = decode_utf8(atob(html));
|
||
selector_source = 'body';
|
||
}
|
||
let parser = new DOMParser();
|
||
window.setTimeout(function () {
|
||
if (url.startsWith('https://archive.') && url_src) {
|
||
let domain_archive = url.match(/^https:\/\/(archive\.\w{2})/)[1];
|
||
let pathname = new URL(url_src).pathname;
|
||
html = html.replace(new RegExp('https:\\/\\/' + domain_archive.replace('.', '\\.') + '\\/o\\/\\w+\\/', 'g'), '').replace(new RegExp("(src=\"|background-image:url\\(')" + pathname.replace('/', '\\/'), 'g'), "$1" + 'https://' + domain_archive + pathname);
|
||
}
|
||
let doc = parser.parseFromString(html, 'text/html');
|
||
if (selector_level)
|
||
selector_source = getSelectorLevel(selector_source);
|
||
let article_new = doc.querySelector(selector_source);
|
||
if (article_new) {
|
||
if (article && article.parentNode) {
|
||
if (url.startsWith('https://archive.')) {
|
||
let arch_dom = (selector_archive !== selector) ? (article_new.querySelector(selector_archive) || document.querySelector(selector_archive)) : article_new;
|
||
if (arch_dom) {
|
||
if (arch_dom.firstChild)
|
||
arch_dom = arch_dom.firstChild;
|
||
let arch_div = document.createElement('div');
|
||
arch_div.appendChild(archiveLink_renew(url_src));
|
||
arch_div.appendChild(archiveLink(window.location.href, 'BPC > Full article text fetched from (no need to report issue for external site):\r\n'));
|
||
arch_div.style = 'margin: 0px 0px 50px;';
|
||
arch_dom.before(arch_div);
|
||
}
|
||
let targets = article_new.querySelectorAll('a[target="_blank"][href^="' + window.location.origin + '"]');
|
||
for (let elem of targets)
|
||
elem.removeAttribute('target');
|
||
let invalid_links = article_new.querySelectorAll('link[rel*="preload"]:not([href])');
|
||
removeDOMElement(...invalid_links);
|
||
}
|
||
window.setTimeout(function () {
|
||
if (article.parentNode) {
|
||
article.parentNode.replaceChild(article_new, article);
|
||
if (func_post)
|
||
func_post();
|
||
}
|
||
}, 200);
|
||
}
|
||
} else
|
||
replaceTextFail(url, article_link, proxy, text_fail.replace(':', no_content_msg));
|
||
}, 200);
|
||
} else {
|
||
replaceTextFail(url, article_link, proxy, url_src ? text_fail.replace(':', no_content_msg) : text_fail);
|
||
}
|
||
}
|
||
|
||
function replaceTextFail(url, article, proxy, text_fail) {
|
||
if (text_fail && article) {
|
||
let text_fail_div = document.createElement('div');
|
||
text_fail_div.id = 'bpc_fail';
|
||
text_fail_div.setAttribute('style', 'margin: 0px 50px; font-weight: bold; color: red;');
|
||
text_fail_div.appendChild(document.createTextNode(text_fail));
|
||
if (proxy) {
|
||
if (url.startsWith('https://archive.')) {
|
||
text_fail_div = archiveLink(url.replace(/^https:\/\/archive\.\w{2}\//, ''), text_fail);
|
||
} else {
|
||
let a_link = document.createElement('a');
|
||
a_link.innerText = url;
|
||
a_link.href = url;
|
||
a_link.target = '_blank';
|
||
text_fail_div.appendChild(a_link);
|
||
}
|
||
}
|
||
if (article.firstChild)
|
||
article.firstChild.before(text_fail_div);
|
||
else
|
||
article.appendChild(text_fail_div);
|
||
}
|
||
}
|
||
|
||
function randomInt(max) {
|
||
return Math.floor(Math.random() * Math.floor(max));
|
||
}
|
||
|
||
function archiveRandomDomain() {
|
||
let tld_array = ['fo', 'is', 'li', 'md', 'ph', 'vn'];
|
||
let tld = tld_array[randomInt(6)];
|
||
return 'archive.' + tld;
|
||
}
|
||
|
||
function getArchive(url, paywall_sel, paywall_action = '', selector, text_fail = '', selector_source = selector, selector_archive = selector) {
|
||
let url_archive = 'https://' + archiveRandomDomain() + '/' + url.split(/[#\?]/)[0];
|
||
let paywall = document.querySelectorAll(paywall_sel);
|
||
if (paywall.length) {
|
||
clearPaywall(paywall, paywall_action);
|
||
replaceDomElementExt(url_archive, true, false, selector, text_fail, selector_source, selector_archive);
|
||
}
|
||
}
|
||
|
||
function archiveLink(url, text_fail = 'BPC > Try for full article text (no need to report issue for external site):\r\n') {
|
||
return externalLink(['archive.today', archiveRandomDomain()], 'https://{domain}?run=1&url={url}', url, text_fail);
|
||
}
|
||
|
||
function archiveLink_renew(url, text_fail = 'BPC > Only use to renew if text is incomplete or updated:\r\n') {
|
||
return externalLink([new URL(url).hostname], '{url}/again?url=' + window.location.href, url, text_fail);
|
||
}
|
||
|
||
function googleSearchToolLink(url, text_fail = 'BPC > Try for full article text (test url & copy html (tab) code to [https://codebeautify.org/htmlviewer]):\r\n') {
|
||
return externalLink(['search.google.com'], 'https://search.google.com/test/rich-results?url={url}', encodeURIComponent(url), text_fail);
|
||
}
|
||
|
||
function externalLink(domains, ext_url_templ, url, text_fail = 'BPC > Full article text:\r\n') {
|
||
let text_fail_div = document.createElement('div');
|
||
text_fail_div.id = 'bpc_archive';
|
||
text_fail_div.setAttribute('style', 'margin: 20px; font-size: 20px; font-weight: bold; color: red;');
|
||
let parser = new DOMParser();
|
||
text_fail = text_fail.replace(/\[(?<url>[^\]]+)\]/g, function (match, url) {
|
||
return "<a href='" + url + "' target='_blank' style='color: red'>" + new URL(url).hostname + "</a>";
|
||
});
|
||
let doc = parser.parseFromString('<span>' + text_fail + '</span>', 'text/html');
|
||
let elem = doc.querySelector('span');
|
||
text_fail_div.appendChild(elem);
|
||
for (let domain of domains) {
|
||
let ext_url = ext_url_templ.replace('{domain}', domain).replace('{url}', url.split('?')[0]);
|
||
let a_link = document.createElement('a');
|
||
a_link.innerText = domain;
|
||
a_link.href = ext_url;
|
||
a_link.target = '_blank';
|
||
text_fail_div.appendChild(document.createTextNode(' | '));
|
||
text_fail_div.appendChild(a_link);
|
||
}
|
||
return text_fail_div;
|
||
}
|
||
|
||
function getArticleJsonScript() {
|
||
let scripts = document.querySelectorAll('script[type="application/ld+json"]');
|
||
let json_script;
|
||
for (let script of scripts) {
|
||
if (script.innerText.match(/"(articlebody|text)":/i)) {
|
||
json_script = script;
|
||
break;
|
||
}
|
||
}
|
||
return json_script;
|
||
}
|
||
|
||
function getNestedKeys(obj, key) {
|
||
if (key in obj)
|
||
return obj[key];
|
||
let keys = key.split('.');
|
||
let value = obj;
|
||
for (let i = 0; i < keys.length; i++) {
|
||
value = value[keys[i]];
|
||
if (value === undefined)
|
||
break;
|
||
}
|
||
return value;
|
||
}
|
||
|
||
function header_nofix(header, cond_sel = '', msg = 'BPC > no fix') {
|
||
if (header && typeof header === 'string')
|
||
header = document.querySelector(header);
|
||
if (header && !document.querySelector('div#bpc_nofix')) {
|
||
if (cond_sel) {
|
||
let elem = document.querySelectorAll(cond_sel);
|
||
if (elem.length)
|
||
removeDOMElement(...elem);
|
||
else
|
||
return false;
|
||
}
|
||
let nofix_div = document.createElement('div');
|
||
nofix_div.id = 'bpc_nofix';
|
||
nofix_div.style = 'margin: 20px; font-size: 20px; font-weight: bold; color: red;';
|
||
nofix_div.innerText = msg;
|
||
header.before(nofix_div);
|
||
}
|
||
}
|
||
|
||
function pageContains(selector, text) {
|
||
let elements = document.querySelectorAll(selector);
|
||
return Array.prototype.filter.call(elements, function (element) {
|
||
return RegExp(text).test(element.textContent);
|
||
});
|
||
}
|
||
|
||
function parseHtmlEntities(encodedString) {
|
||
let translate_re = /&(nbsp|amp|quot|lt|gt|deg|hellip|laquo|raquo|ldquo|rdquo|lsquo|rsquo|mdash);/g;
|
||
let translate = {"nbsp": " ", "amp": "&", "quot": "\"", "lt": "<", "gt": ">", "deg": "°", "hellip": "…",
|
||
"laquo": "«", "raquo": "»", "ldquo": "“", "rdquo": "”", "lsquo": "‘", "rsquo": "’", "mdash": "—"};
|
||
return encodedString.replace(translate_re, function (match, entity) {
|
||
return translate[entity];
|
||
}).replace(/&#(\d+);/gi, function (match, numStr) {
|
||
let num = parseInt(numStr, 10);
|
||
return String.fromCharCode(num);
|
||
});
|
||
}
|
||
|
||
function encode_utf8(str) {
|
||
return unescape(encodeURIComponent(str));
|
||
}
|
||
|
||
function decode_utf8(str) {
|
||
return decodeURIComponent(escape(str));
|
||
}
|
||
|
||
function ampToHtml() {
|
||
window.setTimeout(function () {
|
||
let canonical = document.querySelector('head > link[rel="canonical"][href]');
|
||
if (canonical)
|
||
window.location.href = canonical.href;
|
||
}, 1000);
|
||
}
|
||
|
||
function check_loop(interval = 2000) {
|
||
let loop = true;
|
||
let loop_date = Number(sessionStorage.getItem('###_loop'));
|
||
if (!(loop_date && (Date.now() - loop_date < interval))) {
|
||
sessionStorage.setItem('###_loop', Date.now());
|
||
loop = false;
|
||
}
|
||
return loop;
|
||
}
|
||
|
||
function refreshCurrentTab(not_loop = true) {
|
||
if (!not_loop || !check_loop(5000)) {
|
||
window.setTimeout(function () {
|
||
window.location.reload(true);
|
||
}, 500);
|
||
} else {
|
||
let header = (document.body && document.body.firstChild) || document.documentElement;
|
||
header_nofix(header, '', 'BPC > refresh loop stopped');
|
||
}
|
||
}
|
||
|
||
})();
|