Update custom sites (load text from Google webcache)

README.md

@@ -816,7 +816,7 @@ Check 'Options'-link in popup-menu and go to custom sites.
 \*  by default BPC has limited permissions, but you can opt-in to enable custom sites (and also clear cookies/block general paywall-scripts for non-listed sites). You can also just request permissions for the custom sites you added yourself (or `clear cookies` to ask for permission for current site).
 
 By default sites' cookies/local storage are removed after page loads (to bypass article limit).  
-Also you can enable Googlebot/Bingbot user-agent, set referer (to Facebook, Google or Twitter; ignored when Googlebot is set), set random ip-address, disable Javascript for (sub)domain(s) and/or external domains, block regular expression, unhide text on (or when paywall(selector) redirect to) amp-page and/or load text from json (paywall|article selector).
+Also you can enable Googlebot/Bingbot user-agent, set referer (to Facebook, Google or Twitter; ignored when Googlebot is set), set random ip-address, disable Javascript for (sub)domain(s) and/or external domains, block regular expression, unhide text on (or when paywall(selector) redirect to) amp-page and/or load text from json or Google webcache (paywall|article selector).
 
 [Example list of custom sites](https://gitlab.com/magnolia1234/bypass-paywalls-chrome-clean/-/blob/master/custom/sites_custom.json) or [download list (json)](https://gitlab.com/magnolia1234/bypass-paywalls-chrome-clean/-/raw/master/custom/sites_custom.json)
 

background.js

@@ -70,6 +70,8 @@ var amp_redirect;
 var cs_code;
 // load text from json
 var ld_json;
+// load text from Google webcache
+var ld_google_webcache;
 
 // custom: block javascript
 var block_js_custom = [];
@@ -91,6 +93,7 @@ function initSetRules() {
   amp_redirect = {};
   cs_code = {};
   ld_json = {};
+  ld_google_webcache = {};
   block_js_custom = [];
   block_js_custom_ext = [];
   blockedRegexes = {};
@@ -278,6 +281,11 @@ function set_rules(sites, sites_updated, sites_custom) {
           if (!dompurify_sites.includes(domain))
             dompurify_sites.push(domain);
         }
+        if (rule.ld_google_webcache) {
+          ld_google_webcache[domain] = rule.ld_google_webcache;
+          if (!dompurify_sites.includes(domain))
+            dompurify_sites.push(domain);
+        }
       }
     }
   }
@@ -1103,6 +1111,9 @@ if (matchUrlDomain(change_headers, details.url) && !['font', 'image', 'styleshee
           let ld_json_domain = '';
           if (ld_json_domain = matchUrlDomain(Object.keys(ld_json), currentTabUrl))
             bg2csData.ld_json = ld_json[ld_json_domain];
+          let ld_google_webcache_domain = '';
+          if (ld_google_webcache_domain = matchUrlDomain(Object.keys(ld_google_webcache), currentTabUrl))
+            bg2csData.ld_google_webcache = ld_google_webcache[ld_google_webcache_domain];
           ext_api.tabs.executeScript(tabId, {
             code: 'var bg2csData = ' + JSON.stringify(bg2csData) + ';'
           }, function () {

changelog.txt

@@ -5,6 +5,7 @@ Post-release
 Add Tagesspiegel.de
 Redirect Google AMP cache (opt-in to custom sites)
 Update block general paywall script (limit Evolok WordPress)
+Update custom sites (load text from Google webcache)
 
 * v2.7.8.0 (2022-07-31)
 Add Crusoe (Brazil)

contentScript.js

@@ -60,24 +60,44 @@ var bg2csData;
 // custom/updated sites: load text from json
 if ((bg2csData !== undefined) && bg2csData.ld_json && dompurify_loaded) {
   if (bg2csData.ld_json.includes('|')) {
-    let ld_json_split = bg2csData.ld_json.split('|');
-    let paywall_sel = ld_json_split[0];
-    let article_sel = ld_json_split[1];
-    let paywall = document.querySelector(paywall_sel);
-    if (paywall) {
-      removeDOMElement(paywall);
-      let json_script = getArticleJsonScript();
-      if (json_script) {
-        let json_text = parseHtmlEntities(JSON.parse(json_script.text).articleBody);
-        let content = document.querySelector(article_sel);
-        if (json_text && content) {
-          let parser = new DOMParser();
-          let doc = parser.parseFromString('<div>' + DOMPurify.sanitize(json_text) + '</div>', 'text/html');
-          let content_new = doc.querySelector('div');
-          content.parentNode.replaceChild(content_new, content);
+    window.setTimeout(function () {
+      let ld_json_split = bg2csData.ld_json.split('|');
+      let paywall_sel = ld_json_split[0];
+      let article_sel = ld_json_split[1];
+      let paywall = document.querySelector(paywall_sel);
+      if (paywall) {
+        removeDOMElement(paywall);
+        let json_script = getArticleJsonScript();
+        if (json_script) {
+          let json_text = parseHtmlEntities(JSON.parse(json_script.text).articleBody);
+          let content = document.querySelector(article_sel);
+          if (json_text && content) {
+            let parser = new DOMParser();
+            let doc = parser.parseFromString('<div>' + DOMPurify.sanitize(json_text) + '</div>', 'text/html');
+            let content_new = doc.querySelector('div');
+            content.parentNode.replaceChild(content_new, content);
+          }
         }
       }
-    }
+    }, 1000);
+  }
+}
+
+// custom/updated sites: load text from Google webcache
+if ((bg2csData !== undefined) && bg2csData.ld_google_webcache && dompurify_loaded) {
+  if (bg2csData.ld_google_webcache.includes('|')) {
+    window.setTimeout(function () {
+      let url = window.location.href;
+      let ld_google_webcache_split = bg2csData.ld_google_webcache.split('|');
+      let paywall_sel = ld_google_webcache_split[0];
+      let article_sel = ld_google_webcache_split[1];
+      let paywall = document.querySelector(paywall_sel);
+      if (paywall) {
+        removeDOMElement(paywall);
+        let url_cache = 'https://webcache.googleusercontent.com/search?q=cache:' + url;
+        replaceDomElementExt(url_cache, true, false, article_sel);
+      }
+    }, 1000);
   }
 }
 

custom/manifest.json

@@ -43,5 +43,5 @@
         "webRequestBlocking",
         "<all_urls>"
     ],
-    "version": "2.7.8.1"
+    "version": "2.7.8.2"
 }

manifest.json

@@ -655,7 +655,6 @@
     "*://*.weborama.fr/*",
     "*://*.zephr.com/*",
     "*://*.amazonaws.com/*",
-    "*://*.apnarm.net.au/*",
     "*://*.bntech.io/*",
     "*://*.bwbx.io/*",
     "*://*.cedsdigital.it/*",
@@ -686,5 +685,5 @@
     "*://gcm.omerlocdn.com/*",
     "*://webcache.googleusercontent.com/*"
   ],
-  "version": "2.7.8.1"
+  "version": "2.7.8.2"
 }

options/options_custom.html

@@ -20,7 +20,7 @@
   <h2>Custom Sites</h2>
   <div style="width:90%;">
     To add a new site, enter an unique title/domain (without www.).<br>
-    Select options for useragent (like Googlebot), set referer (ignored when Googlebot is set), set random ip-address, block Javascript, block regular expression, unhide text on (or when paywall(selector) redirect to) amp-page and/or load text from json (paywall|article selector).<br>
+    Select options for useragent (like Googlebot), set referer (ignored when Googlebot is set), set random ip-address, block Javascript, block regular expression, unhide text on (or when paywall(selector) redirect to) amp-page and/or load text from json or Google webcache (paywall|article selector).<br>
     Custom sites (new) are enabled automatically in <small><button><a href="options.html" style="text-decoration:none;color:inherit">Options</a></button></small> (cookies will be removed by default unless you enable allow_cookies).<br>
     If you want to use custom sites (for non-listed sites) enable it in <small><button><a href="optin/opt-in.html" style="text-decoration:none;color:inherit">Opt-in</a></button></small>
     <strong>Custom sites enabled: <span id="custom-enabled"></span></strong><br>

options/options_custom.js

@@ -215,6 +215,7 @@ function edit_options() {
     document.querySelector('input[data-key="amp_unhide"]').checked = (edit_site.amp_unhide > 0);
     document.querySelector('input[data-key="amp_redirect"]').value = edit_site.amp_redirect ? edit_site.amp_redirect : '';
     document.querySelector('input[data-key="ld_json"]').value = edit_site.ld_json ? edit_site.ld_json : '';
+    document.querySelector('input[data-key="ld_google_webcache"]').value = edit_site.ld_google_webcache ? edit_site.ld_google_webcache : '';
     document.querySelector('select[data-key="referer"]').selectedIndex = referer_options.indexOf(edit_site.referer);
     document.querySelector('select[data-key="random_ip"]').selectedIndex = random_ip_options.indexOf(edit_site.random_ip);
   });
@@ -279,7 +280,8 @@ function renderOptions() {
       'block_regex': 0,
       'amp_unhide': 1,
       'amp_redirect': 0,
-      'ld_json': 0
+      'ld_json': 0,
+      'ld_google_webcache': 0
     };
     for (var key in add_checkboxes) {
       labelEl = document.createElement('label');
@@ -295,7 +297,8 @@ function renderOptions() {
           domain: 'example.com',
           block_regex: '\\.example\\.com\\/js\\/',
           amp_redirect: 'div.paywall',
-          ld_json: 'div.paywall|div.article'
+          ld_json: 'div.paywall|div.article',
+          ld_google_webcache: 'div.paywall|div.article',
         };
         if (placeholders[key])
           inputEl.placeholder = placeholders[key];
@@ -351,7 +354,8 @@ function renderOptions() {
       (sites_custom[key]['random_ip'] ? ' | random_ip: ' + sites_custom[key]['random_ip'] : '') +
       (sites_custom[key]['amp_unhide'] > 0 ? ' | amp_unhide' : '') +
       (sites_custom[key]['amp_redirect'] ? ' | amp_redirect' : '') +
-      (sites_custom[key]['ld_json'] ? ' | ld_json' : '');
+      (sites_custom[key]['ld_json'] ? ' | ld_json' : '') +
+      (sites_custom[key]['ld_google_webcache'] ? ' | ld_google_webcache' : '');	  
       optionEl.value = key;
       selectEl.add(optionEl);
     }