☰

extractsentence

提取选中词所在句子

ეს სკრიპტი არ უნდა იყოს პირდაპირ დაინსტალირებული. ეს ბიბლიოთეკაა, სხვა სკრიპტებისთვის უნდა ჩართეთ მეტა-დირექტივაში // @require https://update.greasyfork.org/scripts/534393/1580082/extractsentence.js. 

// altered from https://github.com/ninja33/ODH/blob/master/src/fg/js/text.js
;const {
    getSentence, calSentence, cutSentence
} = (() => {
    const HtmlTagsToReplace = {
        '&': '&',
        '<': '&lt;',
        '>': '&gt;'
    };

    function replaceHtmlTag(tag) {
        return HtmlTagsToReplace[tag] || tag;
    }

    function escapeHtmlTag(string) {
        return string.replace(/[&<>]/g, replaceHtmlTag);
    }

    function escapeRegExp(string) {
        return string.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); // $& means the whole matched string
    }

// peculiarly, it's will change unsafeWindow's String replaceAll if not rename
    String.prototype.replaceAllX = function (search, replacement, flag = 'g') {
        let target = this;
        search = escapeRegExp(search);
        return target.replace(new RegExp(search, flag), replacement);
    };

    String.prototype.searchAll = function (search) {
        let target = this;
        search = escapeRegExp(search);
        let regex = new RegExp(search, 'gi');
        let result = 0;
        let indices = [];
        while ((result = regex.exec(target)) && result) {
            indices.push(result.index);
        }
        return indices;
    };

    function isPDFJSPage() {
        return (document.querySelectorAll('div#viewer.pdfViewer,pdf-viewer#viewer').length > 0);
    }

    function isEmpty(word) {
        return (!word);
    }

    function isShortandNum(word) {
        let numReg = /\d/;
        return (word.length < 3 || numReg.test(word));
    }

    function isChinese(word) {
        let cnReg = /[\u4e00-\u9fa5]+/gi;
        return (cnReg.test(word));
    }

    function isInvalid(word) {
        if (isChinese(word)) return false;
        return (isChinese(word) && isEmpty(word) || isShortandNum(word));
    }

    function cutSentence(word, offset, sentence, sentenceNum, wordFormat, sentenceFormat) {
        if (!word) {
            return '';
        }
        if (!wordFormat) {
            wordFormat = '<b>{$bold}</b>';
        }
        if (!sentenceFormat) {
            sentenceFormat = '{$sentence}';
        }
        wordFormat = wordFormat.split('{$bold}').join('\$&');
        if (sentenceNum > 0) {
            let arr = sentence.match(/((?![.!?;:。！？]['"’”]?\s).|\n)*[.!?;:。！？]['"’”]?(\s|.*$)/g);
            if (arr && arr.length > 1) {
                arr = arr.reduceRight((accumulation, current) => {
                    if (current.search(/\.\w{0,3}\.\s$/g) !== -1) {
                        accumulation[0] = current + accumulation[0];
                    } else {
                        accumulation.unshift(current);
                    }
                    return accumulation;
                }, ['']);
                arr = arr.filter(x => x.length);
            } else {
                arr = [sentence];
            }

            let index = arr.findIndex(ele => { //try to exactly match to word based on offset.
                if (ele.indexOf(word) !== -1 && ele.searchAll(word).indexOf(offset) !== -1)
                    return true;
                else
                    offset -= ele.length;
            });

            if (index === -1) // fallback if can not exactly find word.
                index = arr.findIndex(ele => ele.indexOf(word) !== -1);

            let left = Math.ceil((sentenceNum - 1) / 2);
            let start = index - left;
            let end = index + ((sentenceNum - 1) - left);

            if (start < 0) {
                start = 0;
                end = sentenceNum - 1;
            } else if (end > (arr.length - 1)) {
                end = arr.length - 1;

                if ((end - (sentenceNum - 1)) < 0) {
                    start = 0;
                } else {
                    start = end - (sentenceNum - 1);
                }
            }

            sentence = arr.slice(start, end + 1).join('').replaceAllX(word, word.replace(/\S+/g, wordFormat), 'gi');
        } else {
            sentence = sentence.replace(word, word.replace(/\S+/g, wordFormat));
        }
        return sentenceFormat.replaceAll('{$sentence}', sentence);
    }

    function getSelectionOffset(node) {
        const range = window.getSelection().getRangeAt(0);
        const clone = range.cloneRange();
        clone.selectNodeContents(node);
        clone.setEnd(range.startContainer, range.startOffset);
        let start = clone.toString().length;
        clone.setEnd(range.endContainer, range.endOffset);
        let end = clone.toString().length;
        return {start, end};

    }

    function getPDFNode(node) {

        let backwardindex = 0;
        do {
            node = node.parentNode;
        } while (node.name && node.nodeName.toUpperCase() !== 'SPAN' && node.nodeName.toUpperCase() !== 'DIV');
        let currentspan = node;

        let sentenceNodes = [currentspan];
        let previous = null;
        while ((previous = node.previousSibling)) {
            sentenceNodes.unshift(previous);
            backwardindex += 1;
            if (previous.textContent.search(/[.!?;:。！？]['"’”]?(\s|.*$)/g) !== -1)
                break;
            else
                node = previous;
        }

        node = currentspan;
        let next = null;
        while ((next = node.nextSibling)) {
            sentenceNodes.push(next);
            if (node.nextSibling.textContent.search(/[.!?;:。！？]['"’”]?(\s|.*$)/g) !== -1)
                break;
            else
                node = next;
        }

        let sentence = '';
        let offset = 0;
        sentenceNodes = sentenceNodes.filter(x => x.textContent !== '' || x.textContent !== '-');
        for (const node of sentenceNodes) {
            if (backwardindex === 0)
                offset = sentence.length + window.getSelection().getRangeAt(0).startOffset;
            backwardindex -= 1;
            let nodetext = node.textContent;
            if (nodetext === '-')
                sentence = sentence.slice(0, sentence.length - 1);
            else
                sentence += (nodetext[nodetext.length - 1] === '-') ? nodetext.slice(0, nodetext.length - 1) : nodetext + ' ';
        }

        return {sentence, offset};
    }

    function calSentence() {
        let sentence = '';
        let offset = 0;
        const upNum = 4;

        const selection = window.getSelection();
        let word = (selection.toString() || '').trim();
        const res = {sentence, offset, word}
        if (selection.rangeCount < 1)
            return res;

        let node = selection.getRangeAt(0).commonAncestorContainer;

        if (['INPUT', 'TEXTAREA'].indexOf(node.tagName) !== -1) {
            return res;
        }

        if (isPDFJSPage()) {
            let pdfcontext = getPDFNode(node);
            sentence = escapeHtmlTag(pdfcontext.sentence);
            offset = pdfcontext.offset;
        } else {
            node = getWebNode(node, upNum);

            if (node !== document) {
                sentence = escapeHtmlTag(node.textContent);
                offset = getSelectionOffset(node).start;
            }
        }
        return {sentence, offset, word}
    }

    function getSentence(sentenceNum, wordFormat, sentenceFormat) {
        const {word, offset, sentence} = calSentence()
        if (word === '') {
            return '';
        }
        return cutSentence(word, offset, sentence, sentenceNum, wordFormat, sentenceFormat);
    }

    function getWebNode(node, deep) {
        const blockTags = ['LI', 'P', 'DIV', 'BODY', 'PRE', 'CODE'];
        const nodeName = node.nodeName.toUpperCase();
        if (blockTags.includes(nodeName) || deep === 0) {
            return node;
        } else {
            return getWebNode(node.parentElement, deep - 1);
        }
    }

    function selectedText() {
        const selection = window.getSelection();
        return (selection.toString() || '').trim();
    }

    function isValidElement() {
        // if (document.activeElement.getAttribute('contenteditable'))
        //     return false;

        const invalidTags = ['INPUT', 'TEXTAREA'];
        const nodeName = document.activeElement.nodeName.toUpperCase();
        return !invalidTags.includes(nodeName);
    }

    return {
        getSentence, calSentence, cutSentence
    }
})();