Fix WSJ (amp 404 & restore option to disable Googlebot)

2024-11-10 03:41:57 +01:00 · 2023-08-22 19:46:22 +02:00 · 2023-08-22 19:46:22 +02:00 · 3ade708dea
commit 3ade708dea
parent 3ed2cd2601
9 changed files with 38 additions and 58 deletions
--- a/README.md
+++ b/README.md
@ -936,7 +936,7 @@ Grouped in options:\
 ##### Mexico
 [Mexico News Daily](https://mexiconewsdaily.com)
 ##### Peru
-*Grupo El Comercio* sites like
+*Grupo El Comercio* sites like\
 [Diario Correo](https://diariocorreo.pe) -
 [El Comercio](https://elcomercio.pe) -
 [Gestión](https://gestion.pe)
--- a/background.js
+++ b/background.js
@ -427,7 +427,7 @@ ext_api.storage.local.get({
  optIn: false,
  optInUpdate: true
 }, function (items) {
-  var sites = ext_name.includes('Clean') ? items.sites : {};
+  var sites = items.sites;
  optionSites = sites;
  var sites_default = items.sites_default;
  customSites = items.sites_custom;
@ -496,7 +496,7 @@ ext_api.storage.local.get({
  check_sites_custom_ext();
  if (optin_update)
    check_update();
-  if (!Object.keys(items.sites).length)
+  if (!Object.keys(sites).length)
    ext_api.runtime.openOptionsPage();
 });

@ -906,7 +906,7 @@ ext_api.tabs.onUpdated.addListener((tabId, changeInfo, tab) => {
    if ((tab_status && tab_status === 'complete') || (changeInfo.url)) {
      let timeout = changeInfo.url ? 500 : 0;
      setTimeout(function () {
-        if (matchUrlDomain(enabledSites, tab.url)) {
+        if (isSiteEnabled(tab)) {
          runOnTab(tab);
        }
        runOnTab_once(tab);
@ -999,7 +999,7 @@ if (matchUrlDomain(change_headers, details.url) && !ignore_types.includes(detail
    !(matchUrlDomain(['economictimes.com', 'economictimes.indiatimes.com'], details.url) && !details.url.split(/\?|#/)[0].endsWith('.cms')) &&
    !(matchUrlDomain(au_news_corp_domains, details.url) && (details.url.includes('?amp') || (!matchUrlDomain(au_news_corp_no_amp_fix, details.url) && enabledSites.includes('#options_disable_gb_au_news_corp')))) &&
    !(matchUrlDomain('uol.com.br', details.url) && !matchUrlDomain('folha.uol.com.br', details.url)) &&
-    !(matchUrlDomain('wsj.com', details.url) && !details.url.match(/((\w)+(\-)+){3,}/) && details.type === 'main_frame' && mobile);
+    !(matchUrlDomain('wsj.com', details.url) && (enabledSites.includes('#options_disable_gb_wsj') || (!details.url.match(/((\w)+[%\-]+){3,}/) && details.type === 'main_frame' && mobile)));
  var bingbotEnabled = matchUrlDomain(use_bing_bot, details.url);
  var facebookbotEnabled = matchUrlDomain(use_facebook_bot, details.url);

--- a/changelog.txt
+++ b/changelog.txt
@ -5,6 +5,7 @@ Updates (install signed xpi-file): https://gitlab.com/magnolia1234/bypass-paywal
 Post-release
 Remove DN.no (fix obsolete)
 Fix Mediahuis Noord (json)
+Fix WSJ (amp 404 & restore option to disable Googlebot)

 * v3.3.0.0 (2023-08-20)
 Add Cieletespace.fr
--- a/contentScript.js
+++ b/contentScript.js
@ -2045,6 +2045,12 @@ else if (matchDomain('ilmanifesto.it')) {

 else if (matchDomain(['iltirreno.it', 'lanuovasardegna.it']) || matchDomain(['gazzettadimodena.it', 'gazzettadireggio.it', 'lanuovaferrara.it'])) {
  if (window.location.pathname.includes('/news/')) {
+    let paywall = document.querySelector('span > img[alt*="Paywall"]');
+    if (paywall) {
+      let header = paywall.parentNode.parentNode;
+      header_nofix(header);
+      removeDOMElement(paywall.parentNode);
+    }
    window.setTimeout(function () {
      let banners = document.querySelectorAll('div.MuiSnackbar-root, div.css-16cchgy');
      removeDOMElement(...banners);
@ -5196,17 +5202,24 @@ else if (matchDomain('winnipegfreepress.com')) {
 }

 else if (matchDomain('wsj.com')) {
-  if (window.location.pathname.includes('/amp/')) {
-    amp_unhide_subscr_section();
-    let masthead_link = document.querySelector('div.masthead > a[href*="/articles/"]');
-    if (masthead_link)
-      masthead_link.href = 'https://www.wsj.com';
-  } else {
-    let snippet = false;//document.querySelector('.snippet-promotion, div#cx-snippet-overlay');
-    let wsj_pro = document.querySelector('meta[name="page.site"][content="wsjpro"]');
-    if (snippet || wsj_pro) {
-      removeDOMElement(snippet, wsj_pro);
-      window.location.href = window.location.href.replace('wsj.com', 'wsj.com/amp');
+  let url_article = window.location.pathname.includes('/articles/');
+  let path_article = window.location.pathname.match(/((\w)+(\-)+){3,}\w+/);
+  if (url_article || path_article) {
+    if (window.location.pathname.startsWith('/amp/')) {
+      amp_unhide_subscr_section();
+      let masthead_link = document.querySelector('div.masthead > a[href*="-"]');
+      if (masthead_link)
+        masthead_link.href = 'https://www.wsj.com';
+    } else {
+      let snippet = document.querySelector('.snippet-promotion, div#cx-snippet-overlay');
+      let wsj_pro = document.querySelector('meta[name="page.site"][content="wsjpro"]');
+      if (snippet || wsj_pro) {
+        removeDOMElement(snippet, wsj_pro);
+        if (url_article)
+          window.location.href = window.location.href.replace('wsj.com', 'wsj.com/amp');
+        else
+          window.location.href = '/amp/articles/' + path_article[0];
+      }
    }
  }
  let ads = document.querySelectorAll('div.wsj-ad, div.adWrapper, div.uds-ad-container');
--- a/custom/manifest.json
+++ b/custom/manifest.json
@ -51,5 +51,5 @@
    "webRequestBlocking",
    "*://*/*"
  ],
-  "version": "3.3.0.2"
+  "version": "3.3.0.3"
 }
--- a/manifest.json
+++ b/manifest.json
@ -789,5 +789,5 @@
    "*://*.wyleex.com/*",
    "*://webcache.googleusercontent.com/*"
  ],
-  "version": "3.3.0.2"
+  "version": "3.3.0.3"
 }
--- a/sites.js
+++ b/sites.js
@ -2293,7 +2293,7 @@ var defaultSites = {
    allow_cookies: 1,
    cs_dompurify: 1
  },
-  "The Wall Street Journal": {
+  "The Wall Street Journal (when blocked disable Googlebot in BPC-settings)": {
    domain: "wsj.com",
    allow_cookies: 1,
    block_regex: /(cdn\.cxense\.com\/|cdn\.ampproject\.org\/v\d\/amp-(access|subscriptions)-.+\.js)/,
@ -2601,6 +2601,9 @@ var defaultSites = {
  "Barron's - no Googlebot (http error 500)": {
    domain: "#options_disable_gb_barrons"
  },
+  "The Wall Street Journal - no Googlebot (http error 500)": {
+    domain: "#options_disable_gb_wsj"
+  },
 }

 if (typeof browser !== 'object') {
--- a/sites_updated.json
+++ b/sites_updated.json
@ -1,22 +1,4 @@
 {
-  "America's Test Kitchen": {
-    "domain": "americastestkitchen.com",
-    "allow_cookies": 1,
-    "block_regex": "\\.tinypass\\.com",
-    "upd_version": "3.2.7.9"
-  },
-  "Autocar.co.uk": {
-    "domain": "autocar.co.uk",
-    "allow_cookies": 1,
-    "ld_google_webcache": "div.ms-block,div.register-block|div.content-wrapper",
-    "upd_version": "3.2.7.5"
-  },
-  "Bar and Bench": {
-    "domain": "barandbench.com",
-    "allow_cookies": 1,
-    "ld_json": "div#paywall-banner|div.arr--story-page-card-wrapper",
-    "upd_version": "3.2.7.7"
-  },
  "Bloomberg": {
    "domain": "bloomberg.com",
    "allow_cookies": 1,
@ -35,12 +17,6 @@
    "block_regex": "\\.crusoe\\.com\\.br\\/assets\\/js\\/swg-wallcontent-crusoe\\.js",
    "upd_version": "3.2.9.8"
  },
-  "EUobserver": {
-    "domain": "euobserver.com",
-    "allow_cookies": 1,
-    "cs_code": "[{\"cond\":\"div.membership-upsell.show\", \"rm_class\":\"show\"}]",
-    "upd_version": "3.2.7.4"
-  },
  "GenomeWeb": {
    "domain": "###_usa_genomeweb",
    "group": [
@ -93,13 +69,6 @@
    "block_regex": "\\.precisionmedicineonline\\.com\\/.+\\/js\\/js_.+\\.js",
    "upd_version": "3.2.9.5"
  },
-  "Swarajyamag": {
-    "domain": "swarajyamag.com",
-    "allow_cookies": 1,
-    "remove_cookies": 1,
-    "cs_code": "[{\"cond\":\"div#story-notification\", \"rm_elem\": 1, \"elems\":[{\"cond\":\"div#non-subscriber-text\", \"rm_attrib\": \"id\"}, {\"cond\":\"div.subscriber-text\", \"rm_attrib\": \"class\"}]}]",
-    "upd_version": "3.2.7.8"
-  },
  "The Athletic": {
    "domain": "###_usa_theathletic",
    "group": [
@ -183,11 +152,5 @@
    "domain": "###_usa_townnews",
    "block_regex_general": "\\/shared-content\\/art\\/tncms\\/api\\/access\\.js",
    "upd_version": "3.2.8.3"
-  },
-  "Washington Examiner": {
-    "domain": "washingtonexaminer.com",
-    "allow_cookies": 1,
-    "block_regex": "\\.zephr\\.com\\/zephr-browser",
-    "upd_version": "3.2.7.8"
  }
 }
--- a/updates.json
+++ b/updates.json
@ -2,8 +2,8 @@
  "addons": {
    "magnolia@12.34": {
      "updates": [
-        { "version": "3.2.9.0",
-            "update_link": "https://gitlab.com/magnolia1234/bpc-uploads/-/raw/master/bypass_paywalls_clean-3.2.9.0.xpi" }
+        { "version": "3.3.0.0",
+            "update_link": "https://gitlab.com/magnolia1234/bpc-uploads/-/raw/master/bypass_paywalls_clean-3.3.0.0.xpi" }
        ]
    }
  }