AG-37031 Fix link breakage due to deletion of amp string. #102

Squashed commit of the following: commit 00428fd Author: jellizaveta <[email protected]> Date: Thu Oct 24 13:39:42 2024 +0300 fix issue with selectors, remove redundant conditions commit a04db2a Author: jellizaveta <[email protected]> Date: Wed Oct 23 19:28:06 2024 +0300 clean added links too commit 7aad8cc Author: jellizaveta <[email protected]> Date: Wed Oct 23 18:14:01 2024 +0300 update changelog commit 84a2333 Author: jellizaveta <[email protected]> Date: Wed Oct 23 18:05:01 2024 +0300 fix vars, update config commit bda23bf Author: Slava Leleka <[email protected]> Date: Wed Oct 23 17:01:15 2024 +0300 CHANGELOG.md edited online with Bitbucket commit 56c2a9b Author: jellizaveta <[email protected]> Date: Wed Oct 23 16:23:49 2024 +0300 remove watch from webpack config commit a1a5d36 Author: jellizaveta <[email protected]> Date: Wed Oct 23 16:20:56 2024 +0300 remove redundant func commit 2c6e95d Author: jellizaveta <[email protected]> Date: Wed Oct 23 16:15:28 2024 +0300 remove replacement/deletion of 'amp' string in links, added watch mode for debugging, rewrite the code for redirects commit 395d5b0 Author: jellizaveta <[email protected]> Date: Tue Oct 22 18:27:04 2024 +0300 update meta commit 27450a7 Author: jellizaveta <[email protected]> Date: Tue Oct 22 18:22:34 2024 +0300 update changes commit 29f0a1b Author: jellizaveta <[email protected]> Date: Tue Oct 22 18:21:38 2024 +0300 limit fix commit 4ccec6d Author: jellizaveta <[email protected]> Date: Tue Oct 22 18:21:22 2024 +0300 limit fix commit a4e9314 Author: Slava Leleka <[email protected]> Date: Tue Oct 22 16:17:45 2024 +0300 CHANGELOG.md edited online with Bitbucket commit 70c3d83 Author: Slava Leleka <[email protected]> Date: Tue Oct 22 16:04:39 2024 +0300 src/google-amp.js edited online with Bitbucket commit 8d2c257 Author: jellizaveta <[email protected]> Date: Tue Oct 22 15:59:51 2024 +0300 update changelog commit 8b57d25 Author: jellizaveta <[email protected]> Date: Tue Oct 22 15:57:36 2024 +0300 AG-37031 handle cases where amp should be replaced by HTML file extension. #102
AdguardTeam · Oct 24, 2024 · 7d17631 · 7d17631
1 parent 59f05d4
commit 7d17631
Show file tree

Hide file tree

Showing 8 changed files with 1,569 additions and 2,301 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -7,6 +7,19 @@ The format is based on [Keep a Changelog], and this project adheres to [Semantic
 [Keep a Changelog]: https://keepachangelog.com/en/1.0.0/
 [Semantic Versioning]: https://semver.org/spec/v2.0.0.html
 
+## [1.0.51] - 2024-10-23
+
+### Added
+
+- `cleanAmpLink` function, that forces to open google news link in new tab, necessary for proper redirection. [#102].
+
+### Fixed
+
+- Link breakage due to deletion of amp string. [#102].
+
+[1.0.51]: https://github.com/AdguardTeam/Scriptlets/compare/v1.0.49...v1.0.51
+[#102]: https://github.com/AdguardTeam/DisableAMP/issues/102
+
 ## [1.0.49] - 2024-10-18
 
 ### Added

diff --git a/meta.template.js b/meta.template.js
@@ -22,7 +22,8 @@
 // @include      https://*/?amp*
 // @include      https://*/*&amp*
 // @include      https://*/*&amp=1*
-// @include      https://*?amp=1*
+// @include      https://*?amp=*
+// @include      https://*/*?amp=1*
 // @include      https://*/amp-*/*
 // @include      https://*/ampNews/*
 // @exclude

diff --git a/package.json b/package.json
@@ -7,6 +7,7 @@
         "dev": "NODE_ENV=DEV ./node_modules/.bin/webpack",
         "beta": "NODE_ENV=BETA ./node_modules/.bin/webpack",
         "release": "NODE_ENV=RELEASE ./node_modules/.bin/webpack",
+        "watch": "NODE_ENV=DEV ./node_modules/.bin/webpack --watch",
         "lint": "eslint ./src",
         "locales:download": "LOCALES=DOWNLOAD node locales.js",
         "locales:upload": "LOCALES=UPLOAD node locales.js",
@@ -38,7 +39,7 @@
         "html-webpack-plugin": "^3.2.0",
         "husky": "^4.3.0",
         "replace-in-file": "^4.1.1",
-        "webpack": "^4.35.3",
+        "webpack": "^5.95.0",
         "webpack-cli": "^3.3.5"
     }
 }
diff --git a/src/amp-project.js b/src/amp-project.js
diff --git a/src/google-amp.js b/src/google-amp.js
@@ -1,71 +1,56 @@
-import {
-    sanitizeUrl,
-    hideAmpIcon,
-    sanitizeLinkElement,
-    extractCanonicalFromJslog,
-} from './utils';
+import { hideAmpIcon, sanitizeLinkElement, extractCanonicalFromJslog } from './utils';
 
-const DISABLE_AMP_REDIRECTED = '__disable_amp_redirected';
 const URL_PATTERN_REGEX = /^https?:\/\/.+/i;
-const ARTICLE_VIEW_AMP = 'articleViewAmp';
-const expando = `__${Math.random()}`;
 const HTTPS = 'https://';
+const AMP_ATTRIBUTES_TO_REMOVE = [
+    'ping',
+    'data-ved',
+    'data-amp-cur',
+    'data-amp-title',
+    'data-amp',
+    'data-amp-vgi',
+];
 
 /**
- * Redirects amp version to normal
+ * Prevent amp links from open in google iframe
+ * e.g. google.com/amp/amp.website.com
  */
-export const ampRedirect = () => {
-    // timeout to prevent automatic redirects to amp
-    const REDIRECT_WAIT_TIME_OUT_MS = 30000;
-    const disableAmpRedirectedDate = Number(sessionStorage.getItem(DISABLE_AMP_REDIRECTED));
-    if (
-        // Prevent redirecting to another page if run in an iframe
-        window.self !== window.top
-        // Do not redirect if since last redirect past less than REDIRECT_WAIT_TIME_OUT_MS
-        || (disableAmpRedirectedDate
-            && Date.now() - disableAmpRedirectedDate < REDIRECT_WAIT_TIME_OUT_MS)
-    ) {
-        return;
-    }
-    const canonicalLink = document.querySelector('head > link[rel="canonical"]');
-    const ampProjectScript = document.querySelector('head > script[src^="https://cdn.ampproject.org"]');
-    if (ampProjectScript && canonicalLink && URL_PATTERN_REGEX.test(canonicalLink.href)) {
-        sessionStorage.setItem(DISABLE_AMP_REDIRECTED, Date.now());
-        window.top.location.href = canonicalLink.href;
-    }
+export const cleanAmpLink = () => {
+    const ampLinks = document.querySelectorAll('a[data-amp]');
+    ampLinks.forEach((link) => {
+        AMP_ATTRIBUTES_TO_REMOVE.forEach((attr) => {
+            link.removeAttribute(attr);
+        });
+    });
 };
 
 /**
- * Replaces amp links by data-amp-cur attribute value
+ * Redirects amp version to normal
  */
-const replaceByAmpCurAttribute = () => {
-    const elements = document.querySelectorAll('a[data-amp-cur]');
-    [...elements].forEach((el) => {
-        if (el[expando]) {
-            return;
-        }
+export const ampRedirect = () => {
+    const canonicalLink = document.querySelector('head > link[rel="canonical"]');
+    if (!canonicalLink) {
+        return;
+    }
 
-        // eslint-disable-next-line no-param-reassign
-        el[expando] = true;
+    if (!URL_PATTERN_REGEX.test(canonicalLink.href)) {
+        return;
+    }
 
-        const url = el.getAttribute('data-amp-cur') || el.getAttribute('data-amp');
-        if (url) {
-            sanitizeLinkElement(el, sanitizeUrl(url));
-        }
+    // iframe do not have this marker
+    const ampMarker = document.querySelector('script[src^="https://cdn.ampproject.org/"]');
+    if (!ampMarker) {
+        return;
+    }
 
-        // Some websites manage to break the data-amp-cur attribute and use articleViewAmp htmls instead,
-        // e.g http://www.mediatoday.co.kr/news/articleViewAmp.html?idxno=313292
-        const { href } = el;
-        if (href && href.includes(ARTICLE_VIEW_AMP)) {
-            sanitizeLinkElement(el, href.replace(ARTICLE_VIEW_AMP, 'articleView'));
-        }
-    });
+    // redirect to canonical link if current page is not iframe
+    document.location.href = canonicalLink.href;
 };
 
 /**
  * Replaces amp links provided by amp cdn
  */
-const replaceCdnAmp = () => {
+export const replaceCdnAmp = () => {
     const ampLinks = document.querySelectorAll('a[data-amp-cdn]');
     ampLinks.forEach((ampLink) => {
         let fixedUrl = ampLink.href;
@@ -93,11 +78,6 @@ const replaceCdnAmp = () => {
     });
 };
 
-export const preventAmp = () => {
-    replaceByAmpCurAttribute();
-    replaceCdnAmp();
-};
-
 export const preventNewsAmp = () => {
     if (window.self !== window.top) {
         return;

diff --git a/src/index.js b/src/index.js
@@ -1,25 +1,39 @@
 import { redirectTurboPages } from './yandex-turbo';
-import { redirectAmpProjectPages } from './amp-project';
-import { preventNewsAmp, preventAmp, ampRedirect } from './google-amp';
+import {
+    preventNewsAmp,
+    replaceCdnAmp,
+    ampRedirect,
+    cleanAmpLink,
+} from './google-amp';
 import { observeDomChanges } from './utils';
 
+const SEARCH_PATH = '/search';
+const GOOGLE_NEWS_PATH = 'news.google.';
+const GOOGLE_PATH = 'google.';
+const YANDEX_TURBO_PATH = 'https://yandex.ru/turbo';
+const TURBOPAGES_PATH = 'turbopages.org';
+
 (() => {
     const { href, origin } = document.location;
 
-    if (href.includes('https://yandex.ru/turbo') || href.includes('turbopages.org')) {
+    if (href.includes(YANDEX_TURBO_PATH) || href.includes(TURBOPAGES_PATH)) {
         redirectTurboPages();
         return;
     }
 
-    if (!origin.includes('.google.')) {
+    // clean links in google search result to prevent them from open in iframe
+    if (origin.includes(GOOGLE_PATH) && href.includes(SEARCH_PATH)) {
+        observeDomChanges(cleanAmpLink);
+    }
+
+    // redirect only amp links
+    if (!origin.includes(GOOGLE_PATH)) {
         ampRedirect();
-        redirectAmpProjectPages();
-        return;
     }
 
-    if (origin.includes('news.google.')) {
+    if (origin.includes(GOOGLE_NEWS_PATH)) {
         observeDomChanges(preventNewsAmp);
     } else {
-        observeDomChanges(preventAmp);
+        observeDomChanges(replaceCdnAmp);
     }
 })();
diff --git a/src/utils.js b/src/utils.js
@@ -35,14 +35,6 @@ export const sanitizeLinkElement = (el, url) => {
     hideAmpIcon(el);
 };
 
-/**
- * Removes amp marker from url
- *
- * @param {string} ampUrl amp url
- * @returns {string} url without amp marker
- */
-export const sanitizeUrl = (ampUrl) => ampUrl.replace(AMP_TOKEN_REGEXP, '');
-
 /**
  * Extracts canonical url from link tags which use jslog attribute to store
  * encoded data. Encoded data may store only canonical url or both canonical