☰

extractsentence

提取选中词所在句子

此脚本不应直接安装。它是供其他脚本使用的外部库，要使用该库请加入元指令 // @require https://update.gf.qytechs.cn/scripts/534393/1605572/extractsentence.js 

  1. // altered from https://github.com/ninja33/ODH/blob/master/src/fg/js/text.js
  2. ;const {
  3.     getSentence, calSentence, cutSentence, escapeRegExp
  4. } = (() => {
  5.     const HtmlTagsToReplace = {
  6.         '&': '&',
  7.         '<': '&lt;',
  8.         '>': '&gt;'
  9.     };
  10.  
  11.     function replaceHtmlTag(tag) {
  12.         return HtmlTagsToReplace[tag] || tag;
  13.     }
  14.  
  15.     function escapeHtmlTag(string) {
  16.         return string.replace(/[&<>]/g, replaceHtmlTag);
  17.     }
  18.  
  19.     function escapeRegExp(string) {
  20.         return string.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); // $& means the whole matched string
  21.     }
  22.  
  23. // peculiarly, it's will change unsafeWindow's String replaceAll if not rename
  24.     String.prototype.replaceAllX = function (search, replacement, flag = 'g') {
  25.         let target = this;
  26.         search = escapeRegExp(search);
  27.         return target.replace(new RegExp(search, flag), replacement);
  28.     };
  29.  
  30.     String.prototype.searchAll = function (search) {
  31.         let target = this;
  32.         search = escapeRegExp(search);
  33.         let regex = new RegExp(search, 'gi');
  34.         let result = 0;
  35.         let indices = [];
  36.         while ((result = regex.exec(target)) && result) {
  37.             indices.push(result.index);
  38.         }
  39.         return indices;
  40.     };
  41.  
  42.     function isPDFJSPage() {
  43.         return (document.querySelectorAll('div#viewer.pdfViewer,pdf-viewer#viewer').length > 0);
  44.     }
  45.  
  46.     function isEmpty(word) {
  47.         return (!word);
  48.     }
  49.  
  50.     function isShortandNum(word) {
  51.         let numReg = /\d/;
  52.         return (word.length < 3 || numReg.test(word));
  53.     }
  54.  
  55.     function isChinese(word) {
  56.         let cnReg = /[\u4e00-\u9fa5]+/gi;
  57.         return (cnReg.test(word));
  58.     }
  59.  
  60.     function isInvalid(word) {
  61.         if (isChinese(word)) return false;
  62.         return (isChinese(word) && isEmpty(word) || isShortandNum(word));
  63.     }
  64.  
  65.     function cutSentence(word, offset, sentence, sentenceNum, wordFormat, sentenceFormat) {
  66.         if (!word) {
  67.             return '';
  68.         }
  69.         if (!wordFormat) {
  70.             wordFormat = '<b>{$bold}</b>';
  71.         }
  72.         if (!sentenceFormat) {
  73.             sentenceFormat = '{$sentence}';
  74.         }
  75.         wordFormat = wordFormat.split('{$bold}').join('\$&');
  76.         if (sentenceNum > 0) {
  77.             let arr = sentence.match(/((?![.!?;:。！？]['"’”]?\s).|\n)*[.!?;:。！？]['"’”]?(\s|.*$)/g);
  78.             if (arr && arr.length > 1) {
  79.                 arr = arr.reduceRight((accumulation, current) => {
  80.                     if (current.search(/\.\w{0,3}\.\s$/g) !== -1) {
  81.                         accumulation[0] = current + accumulation[0];
  82.                     } else {
  83.                         accumulation.unshift(current);
  84.                     }
  85.                     return accumulation;
  86.                 }, ['']);
  87.                 arr = arr.filter(x => x.length);
  88.             } else {
  89.                 arr = [sentence];
  90.             }
  91.  
  92.             let index = arr.findIndex(ele => { //try to exactly match to word based on offset.
  93.                 if (ele.indexOf(word) !== -1 && ele.searchAll(word).indexOf(offset) !== -1)
  94.                     return true;
  95.                 else
  96.                     offset -= ele.length;
  97.             });
  98.  
  99.             if (index === -1) // fallback if can not exactly find word.
  100.                 index = arr.findIndex(ele => ele.indexOf(word) !== -1);
  101.  
  102.             let left = Math.ceil((sentenceNum - 1) / 2);
  103.             let start = index - left;
  104.             let end = index + ((sentenceNum - 1) - left);
  105.  
  106.             if (start < 0) {
  107.                 start = 0;
  108.                 end = sentenceNum - 1;
  109.             } else if (end > (arr.length - 1)) {
  110.                 end = arr.length - 1;
  111.  
  112.                 if ((end - (sentenceNum - 1)) < 0) {
  113.                     start = 0;
  114.                 } else {
  115.                     start = end - (sentenceNum - 1);
  116.                 }
  117.             }
  118.  
  119.             sentence = arr.slice(start, end + 1).join('').replaceAllX(word, word.replace(/\S+/g, wordFormat), 'gi');
  120.         } else {
  121.             sentence = sentence.replace(word, word.replace(/\S+/g, wordFormat));
  122.         }
  123.         return sentenceFormat.replaceAll('{$sentence}', sentence);
  124.     }
  125.  
  126.     function getSelectionOffset(node) {
  127.         const range = window.getSelection().getRangeAt(0);
  128.         const clone = range.cloneRange();
  129.         clone.selectNodeContents(node);
  130.         clone.setEnd(range.startContainer, range.startOffset);
  131.         let start = clone.toString().length;
  132.         clone.setEnd(range.endContainer, range.endOffset);
  133.         let end = clone.toString().length;
  134.         return {start, end};
  135.  
  136.     }
  137.  
  138.     function getPDFNode(node) {
  139.  
  140.         let backwardindex = 0;
  141.         do {
  142.             node = node.parentNode;
  143.         } while (node.name && node.nodeName.toUpperCase() !== 'SPAN' && node.nodeName.toUpperCase() !== 'DIV');
  144.         let currentspan = node;
  145.  
  146.         let sentenceNodes = [currentspan];
  147.         let previous = null;
  148.         while ((previous = node.previousSibling)) {
  149.             sentenceNodes.unshift(previous);
  150.             backwardindex += 1;
  151.             if (previous.textContent.search(/[.!?;:。！？]['"’”]?(\s|.*$)/g) !== -1)
  152.                 break;
  153.             else
  154.                 node = previous;
  155.         }
  156.  
  157.         node = currentspan;
  158.         let next = null;
  159.         while ((next = node.nextSibling)) {
  160.             sentenceNodes.push(next);
  161.             if (node.nextSibling.textContent.search(/[.!?;:。！？]['"’”]?(\s|.*$)/g) !== -1)
  162.                 break;
  163.             else
  164.                 node = next;
  165.         }
  166.  
  167.         let sentence = '';
  168.         let offset = 0;
  169.         sentenceNodes = sentenceNodes.filter(x => x.textContent !== '' || x.textContent !== '-');
  170.         for (const node of sentenceNodes) {
  171.             if (backwardindex === 0)
  172.                 offset = sentence.length + window.getSelection().getRangeAt(0).startOffset;
  173.             backwardindex -= 1;
  174.             let nodetext = node.textContent;
  175.             if (nodetext === '-')
  176.                 sentence = sentence.slice(0, sentence.length - 1);
  177.             else
  178.                 sentence += (nodetext[nodetext.length - 1] === '-') ? nodetext.slice(0, nodetext.length - 1) : nodetext + ' ';
  179.         }
  180.  
  181.         return {sentence, offset};
  182.     }
  183.  
  184.     function calSentence() {
  185.         let sentence = '';
  186.         let offset = 0;
  187.         const upNum = 4;
  188.  
  189.         const selection = window.getSelection();
  190.         let word = (selection.toString() || '').trim();
  191.         const res = {sentence, offset, word}
  192.         if (selection.rangeCount < 1)
  193.             return res;
  194.  
  195.         let node = selection.getRangeAt(0).commonAncestorContainer;
  196.  
  197.         if (['INPUT', 'TEXTAREA'].indexOf(node.tagName) !== -1) {
  198.             return res;
  199.         }
  200.  
  201.         if (isPDFJSPage()) {
  202.             let pdfcontext = getPDFNode(node);
  203.             sentence = escapeHtmlTag(pdfcontext.sentence);
  204.             offset = pdfcontext.offset;
  205.         } else {
  206.             node = getWebNode(node, upNum);
  207.  
  208.             if (node !== document) {
  209.                 sentence = escapeHtmlTag(node.textContent);
  210.                 offset = getSelectionOffset(node).start;
  211.             }
  212.         }
  213.         return {sentence, offset, word}
  214.     }
  215.  
  216.     function getSentence(sentenceNum, wordFormat, sentenceFormat) {
  217.         const {word, offset, sentence} = calSentence()
  218.         if (word === '') {
  219.             return '';
  220.         }
  221.         return cutSentence(word, offset, sentence, sentenceNum, wordFormat, sentenceFormat);
  222.     }
  223.  
  224.     function getWebNode(node, deep) {
  225.         const blockTags = ['LI', 'P', 'DIV', 'BODY', 'PRE', 'CODE'];
  226.         const nodeName = node.nodeName.toUpperCase();
  227.         if (blockTags.includes(nodeName) || deep === 0) {
  228.             return node;
  229.         } else {
  230.             return getWebNode(node.parentElement, deep - 1);
  231.         }
  232.     }
  233.  
  234.     function selectedText() {
  235.         const selection = window.getSelection();
  236.         return (selection.toString() || '').trim();
  237.     }
  238.  
  239.     function isValidElement() {
  240.         // if (document.activeElement.getAttribute('contenteditable'))
  241.         //     return false;
  242.  
  243.         const invalidTags = ['INPUT', 'TEXTAREA'];
  244.         const nodeName = document.activeElement.nodeName.toUpperCase();
  245.         return !invalidTags.includes(nodeName);
  246.     }
  247.  
  248.     return {
  249.         getSentence, calSentence, cutSentence, escapeRegExp
  250.     }
  251. })();

欢迎访问Greasy Fork镜像

土豆服务器，请按需使用

镜像地址随时可能被墙，建议加群获取最新地址

加入Q群

QingJ © 2025

镜像随时可能失效，请加Q群300939539或关注我们的公众号极客氢云获取最新地址