From b2835347ba65165d740937faa999ccdcce2511c5 Mon Sep 17 00:00:00 2001
From: Samuel Macbeth <smacbeth@duckduckgo.com>
Date: Wed, 28 Dec 2022 10:49:27 -0500
Subject: [PATCH] Add element collector

---
 collectors/ElementCollector.js | 97 ++++++++++++++++++++++++++++++++++
 main.js                        |  2 +
 2 files changed, 99 insertions(+)
 create mode 100644 collectors/ElementCollector.js

diff --git a/collectors/ElementCollector.js b/collectors/ElementCollector.js
new file mode 100644
index 00000000..fbc8efca
--- /dev/null
+++ b/collectors/ElementCollector.js
@@ -0,0 +1,97 @@
+const fs = require('fs').promises;
+const BaseCollector = require('./BaseCollector');
+
+class ElementCollector extends BaseCollector {
+
+    id() {
+        return 'elements';
+    }
+
+    /**
+     * @param {import('./BaseCollector').CollectorInitOptions} options 
+     */
+    init(options) {
+        this.context = options.context;
+        this.log = options.log;
+        /**
+         * @type {import('puppeteer').Frame[]}
+         */
+        this.frames = [];
+        this.cookieSelectors = this._loadCookieMonsterRules();
+    }
+
+    async _loadCookieMonsterRules() {
+        const contents = String(await fs.readFile('./fanboy-cookiemonster.txt', {encoding: 'utf-8'}));
+        return contents.split('\n').filter(line => line.startsWith('##')).map(line => line.slice(2));
+    }
+
+    /**
+     * @param {{cdpClient: import('puppeteer').CDPSession, url: string, type: import('./TargetCollector').TargetType}} targetInfo 
+     */
+    // eslint-disable-next-line no-unused-vars
+    addTarget(targetInfo) {
+        if (targetInfo.type === 'page') {
+            this.context.pages().then(pages => {
+                this.page = pages[0];
+                this.check = new Promise(resolve => {
+                    this.page.on('load', async () => {
+                        // check CSS rule matches
+                        const selectors = await this.cookieSelectors;
+                        const isMatched = await Promise.all(selectors.map(async selector => (await this.page.$(selector)) !== null));
+                        const present = selectors.filter((_, i) => isMatched[i]);
+                        const visible = await this.page.evaluate(testSelectors => {
+                            /**
+                             * @param {HTMLElement} elem
+                             */
+                            function isElementVisible(elem) {
+                                if (!elem) {
+                                    return false;
+                                }
+                                if (elem.offsetParent !== null) {
+                                    return true;
+                                }
+                                // eslint-disable-next-line no-undef
+                                const css = window.getComputedStyle(elem);
+                                if (css.position === 'fixed' && css.display !== "none") { // fixed elements may be visible even if the parent is not
+                                    return true;
+                                }
+                                return false;
+                            }
+                            return testSelectors.filter((/** @type {string} */ s) => {
+                                // eslint-disable-next-line no-undef
+                                const elem = document.querySelectorAll(s);
+                                const results = new Array(elem.length);
+                                elem.forEach((e, i) => {
+                                    // check for display: none
+                                    // @ts-ignore
+                                    results[i] = isElementVisible(e);
+                                    if (results[i]) {
+                                        e.setAttribute('style', 'border: 4px dashed red;');
+                                    }
+                                });
+                                return results.some(r => r);
+                            });
+                        }, present);
+                        resolve({
+                            present,
+                            visible,
+                        });
+                    });
+                });
+            }, () => this.log('Unable to get pages'));
+        }
+    }
+
+    /**
+     * Called after the crawl to retrieve the data. Can be async, can throw errors.
+     * 
+     * @param {{finalUrl: string, urlFilter?: function(string):boolean}} options
+     * @returns {Promise<Object>|Object}
+     */
+    // eslint-disable-next-line no-unused-vars
+    getData(options) {
+        return this.check;
+    }
+}
+
+module.exports = ElementCollector;
\ No newline at end of file
diff --git a/main.js b/main.js
index 1095c0ab..43256d12 100644
--- a/main.js
+++ b/main.js
@@ -8,6 +8,7 @@ const TargetCollector = require('./collectors/TargetCollector');
 const TraceCollector = require('./collectors/TraceCollector');
 const ScreenshotCollector = require('./collectors/ScreenshotCollector');
 const CMPCollector = require('./collectors/CMPCollector');
+const ElementCollector = require('./collectors/ElementCollector');
 
 // reexport main pieces of code so that they can be easily imported when this project is used as a dependency
 // e.g. `const {crawlerConductor} = require('3p-crawler');`
@@ -22,4 +23,5 @@ module.exports = {
     TraceCollector,
     ScreenshotCollector,
     CMPCollector,
+    ElementCollector,
 };
\ No newline at end of file