☰

Text Frequency Analyzer with ASIN Page Support

Extract text from specific spans based on page URL, remove stop words, and do word frequency analysis.

  1. // ==UserScript==
  2. // @name         Text Frequency Analyzer with ASIN Page Support
  3. // @namespace    http://tampermonkey.net/
  4. // @version      1.10
  5. // @description  Extract text from specific spans based on page URL, remove stop words, and do word frequency analysis.
  6. // @author       Your Name
  7. // @match        https://www.amazon.com/gp/bestsellers*
  8. // @match        https://www.amazon.com/*/dp/*
  9. // @match        https://www.amazon.com/dp/*
  10. // @match        https://www.amazon.com/s?k=*
  11. // @match        https://www.amazon.com/s?*
  12.  
  13. // @license MIT
  14. // ==/UserScript==
  15.  
  16. (function () {
  17.     'use strict';
  18.  
  19.     const stopWords = new Set([
  20.         'with', 'of', 'for', 'and', 'at', 'if', 'to', 'on', 'by', 'from', 'as', 'than', 'too',
  21.         // ...省略其他停用词
  22.         'now'
  23.     ]);
  24.  
  25.     function extractText() {
  26.         let spans;
  27.         const url = window.location.href;
  28.  
  29.         if (url.includes('/gp/bestsellers')) {
  30.             spans = document.querySelectorAll('._cDEzb_p13n-sc-css-line-clamp-3_g3dy1');
  31.         } else if (url.includes('/s?k=')) {
  32.             spans = document.querySelectorAll('.a-size-base-plus.a-color-base.a-text-normal, .a-size-medium.a-color-base.a-text-normal');
  33.         } else if (url.includes('/dp/')) {
  34.             const titleElement = document.getElementById('productTitle');
  35.             const miniElements = document.querySelectorAll('.a-unordered-list.a-vertical.a-spacing-mini .a-spacing-mini');
  36.             let textContent = titleElement ? titleElement.innerText : '';
  37.             miniElements.forEach(el => {
  38.                 textContent += ' ' + el.innerText;
  39.             });
  40.             return textContent.trim();
  41.         } else {
  42.             alert('This script is not configured for this page.');
  43.             return '';
  44.         }
  45.  
  46.         let textContent = '';
  47.         spans.forEach(span => {
  48.             textContent += span.innerText + ' ';
  49.         });
  50.         return textContent.trim();
  51.     }
  52.  
  53.     function cleanText(text) {
  54.         return text.toLowerCase()
  55.             .replace(/[^a-z0-9\s\/"'.-]/g, '')  // 保留特定符号
  56.             .replace(/\s+/g, ' ')
  57.             .trim();
  58.     }
  59.  
  60.     function getWords(text, removeStopWords = true) {
  61.         const words = cleanText(text).split(/\s+/).filter(Boolean);
  62.         if (removeStopWords) {
  63.             return words.filter(word => !stopWords.has(word));
  64.         }
  65.         return words;
  66.     }
  67.  
  68.     function countFrequencies(words, n) {
  69.         const freqMap = new Map();
  70.         for (let i = 0; i <= words.length - n; i++) {
  71.             const phrase = words.slice(i, i + n).join(' ');
  72.             freqMap.set(phrase, (freqMap.get(phrase) || 0) + 1);
  73.         }
  74.         return Array.from(freqMap.entries()).sort((a, b) => b[1] - a[1]).slice(0, 10);
  75.     }
  76.  
  77.     function removePreviousHighlights() {
  78.         const highlightedElements = document.querySelectorAll('.highlight');
  79.         highlightedElements.forEach(el => {
  80.             el.outerHTML = el.innerText; // Replace the span with its text content
  81.         });
  82.     }
  83.  
  84.     function highlightText(phrase) {
  85.         removePreviousHighlights(); // Remove previous highlights
  86.  
  87.         const url = window.location.href;
  88.         const regex = new RegExp(`(${phrase.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')})`, 'gi');
  89.  
  90.         if (url.includes('/dp/')) {
  91.             const titleElement = document.getElementById('productTitle');
  92.             if (titleElement) {
  93.                 titleElement.innerHTML = titleElement.innerHTML.replace(regex, '<span class="highlight">$1</span>');
  94.             }
  95.             const miniElements = document.querySelectorAll('.a-unordered-list.a-vertical.a-spacing-mini .a-spacing-mini');
  96.             miniElements.forEach(el => {
  97.                 el.innerHTML = el.innerHTML.replace(regex, '<span class="highlight">$1</span>');
  98.             });
  99.         } else {
  100.             const classesToSearch = [
  101.                 '_cDEzb_p13n-sc-css-line-clamp-3_g3dy1',
  102.                 'a-size-base-plus.a-color-base.a-text-normal',
  103.                 'a-size-medium.a-color-base.a-text-normal'
  104.             ];
  105.  
  106.             classesToSearch.forEach(className => {
  107.                 document.querySelectorAll(`.${className}`).forEach(span => {
  108.                     span.innerHTML = span.innerHTML.replace(regex, '<span class="highlight">$1</span>');
  109.                 });
  110.             });
  111.         }
  112.     }
  113.  
  114.     function displayResults(results) {
  115.         const resultDiv = document.createElement('div');
  116.         resultDiv.style.position = 'fixed';
  117.         resultDiv.style.top = '10px';
  118.         resultDiv.style.right = '10px';
  119.         resultDiv.style.backgroundColor = 'white';
  120.         resultDiv.style.border = '1px solid black';
  121.         resultDiv.style.padding = '10px';
  122.         resultDiv.style.zIndex = '10000';
  123.         resultDiv.style.maxHeight = '90vh';
  124.         resultDiv.style.overflowY = 'auto';
  125.         resultDiv.innerHTML = '<h2>Word Frequency Analysis</h2>';
  126.  
  127.         results.forEach(([label, data]) => {
  128.             const title = document.createElement('h3');
  129.             title.textContent = label;
  130.             resultDiv.appendChild(title);
  131.             const list = document.createElement('ul');
  132.             data.forEach(([phrase, count]) => {
  133.                 const listItem = document.createElement('li');
  134.                 listItem.textContent = `${phrase}: ${count}`;
  135.                 listItem.addEventListener('click', () => highlightText(phrase));  // 绑定点击事件
  136.                 list.appendChild(listItem);
  137.             });
  138.             resultDiv.appendChild(list);
  139.         });
  140.  
  141.         document.body.appendChild(resultDiv);
  142.     }
  143.  
  144.     function analyzeText() {
  145.         const text = extractText();
  146.         if (!text) {
  147.             alert('No text found in the specified spans.');
  148.             return;
  149.         }
  150.  
  151.         const wordsForSingle = getWords(text);
  152.         const wordsForPhrases = getWords(text, false);
  153.  
  154.         const results = [
  155.             ['Top 10 Single Words', countFrequencies(wordsForSingle, 1)],
  156.             ['Top 10 Two-Word Phrases', countFrequencies(wordsForPhrases, 2)],
  157.             ['Top 10 Three-Word Phrases', countFrequencies(wordsForPhrases, 3)],
  158.             ['Top 10 Four-Word Phrases', countFrequencies(wordsForPhrases, 4)]
  159.         ];
  160.  
  161.         displayResults(results);
  162.     }
  163.  
  164.     const highlightStyle = document.createElement('style');
  165.     highlightStyle.innerHTML = `
  166.         .highlight {
  167.             background-color: yellow;
  168.             font-weight: bold;
  169.         }
  170.     `;
  171.     document.head.appendChild(highlightStyle);
  172.  
  173.     const analyzeButton = document.createElement('button');
  174.     analyzeButton.textContent = 'Analyze Text Frequency';
  175.     analyzeButton.style.position = 'fixed';
  176.     analyzeButton.style.bottom = '10px';
  177.     analyzeButton.style.right = '10px';
  178.     analyzeButton.style.zIndex = '10000';
  179.     analyzeButton.style.padding = '10px 20px';
  180.     analyzeButton.style.backgroundColor = '#007bff';
  181.     analyzeButton.style.color = 'white';
  182.     analyzeButton.style.border = 'none';
  183.     analyzeButton.style.borderRadius = '5px';
  184.     analyzeButton.style.cursor = 'pointer';
  185.  
  186.     analyzeButton.addEventListener('click', analyzeText);
  187.  
  188.     document.body.appendChild(analyzeButton);
  189. })();