薇星幻海
292 lines
16 KiB
JavaScript
292 lines
16 KiB
JavaScript
/* QuestionBankCore: 从已保存或当前练习页面读取题目，始终以“参考答案”为准。 */
|
||
(function (root) {
|
||
'use strict';
|
||
|
||
const SELECTORS = {
|
||
question: '.pc_examtest_content',
|
||
stem: '.pc_examtest_content_question',
|
||
option: '.pc_examtest_content_list_des',
|
||
label: '.examtest_content_list_des_optionsort',
|
||
reference: '.examtest_content_answer_left',
|
||
explanation: '.pc_examtest_content_resolve',
|
||
card: '.examtestcard_count_defult'
|
||
};
|
||
// 考试页的分数和难度属于界面标记；只按 DOM class 去除，保留题干中的括号和数字。
|
||
const CONTROL = '.pc_examtest_content_question_showanswer,.pc_examtest_content_question_value,.pc_examtest_content_question_difficulty,script,style,button,input,select,textarea,iframe,object,embed,link';
|
||
const TYPE_RE = /^\s*[\[【](单选题?|多选题?|判断题?|填空题?|简答题?|问答题?)[\]】]\s*/;
|
||
const NUMBER_RE = /^\s*(\d+)\s*[.．、]\s*/;
|
||
|
||
function normalize(value) {
|
||
return String(value == null ? '' : value).replace(/\u00a0/g, ' ').replace(/[\t\n\r ]+/g, ' ').trim();
|
||
}
|
||
|
||
function cleanClone(node) {
|
||
if (!node) return null;
|
||
const copy = node.cloneNode(true);
|
||
copy.querySelectorAll(CONTROL).forEach(function (item) { item.remove(); });
|
||
return copy;
|
||
}
|
||
|
||
function textOf(node) {
|
||
const copy = cleanClone(node);
|
||
return copy ? normalize(copy.textContent) : '';
|
||
}
|
||
|
||
function savedSourceURL(doc) {
|
||
const walker = doc.createTreeWalker(doc, 128);
|
||
let comment;
|
||
while ((comment = walker.nextNode())) {
|
||
const match = comment.nodeValue.match(/saved from url=\(\d+\)([\s\S]+?)\s*$/i);
|
||
if (match) return match[1].trim();
|
||
}
|
||
return '';
|
||
}
|
||
|
||
function usableURL(value) {
|
||
return /^(https?:|file:|blob:)/i.test(value || '') ? value : '';
|
||
}
|
||
|
||
function resolveURL(raw, base) {
|
||
if (!raw) return '';
|
||
raw = raw.trim();
|
||
if (/^(javascript|vbscript):/i.test(raw) || /^data:(?!image\/)/i.test(raw)) return '';
|
||
try { return new URL(raw, base || undefined).href; } catch (_) { return raw; }
|
||
}
|
||
|
||
function originalAttribute(node, name) {
|
||
return node.getAttribute('data-qb-original-' + name) || node.getAttribute(name) || '';
|
||
}
|
||
|
||
function baseURL(doc, sourceUrl) {
|
||
const base = doc.querySelector('base');
|
||
return base ? resolveURL(originalAttribute(base, 'href'), sourceUrl) || sourceUrl : sourceUrl;
|
||
}
|
||
|
||
function imageResources(node, base) {
|
||
if (!node) return [];
|
||
return Array.from(node.querySelectorAll('img')).map(function (img) {
|
||
const originalSrc = originalAttribute(img, 'src') || img.getAttribute('data-src') || img.getAttribute('data-original') || '';
|
||
return {
|
||
src: originalSrc,
|
||
url: resolveURL(originalSrc, base),
|
||
alt: normalize(img.getAttribute('alt'))
|
||
};
|
||
}).filter(function (item) { return item.url; });
|
||
}
|
||
|
||
function uniqueImages(resources) {
|
||
return Array.from(new Set(resources.map(function (item) { return item.url; })));
|
||
}
|
||
|
||
// HTML 仅作为附加数据导出。预览界面只使用 textContent，不插入这些 HTML。
|
||
function staticHTML(node, base) {
|
||
const copy = cleanClone(node);
|
||
if (!copy) return '';
|
||
const linkURLs = Array.from(copy.querySelectorAll('a')).map(function (link) { return resolveURL(originalAttribute(link, 'href'), base); });
|
||
copy.querySelectorAll('*').forEach(function (element) {
|
||
Array.from(element.attributes).forEach(function (attr) {
|
||
if (/^on|^_ng|^ng-|^style$|^data-qb-original-/i.test(attr.name)) element.removeAttribute(attr.name);
|
||
});
|
||
});
|
||
const sourceImages = node.querySelectorAll('img');
|
||
copy.querySelectorAll('img').forEach(function (img, index) {
|
||
const source = sourceImages[index];
|
||
const raw = source && (originalAttribute(source, 'src') || source.getAttribute('data-src') || source.getAttribute('data-original'));
|
||
const url = resolveURL(raw, base);
|
||
if (url) img.setAttribute('src', url); else img.removeAttribute('src');
|
||
img.removeAttribute('srcset');
|
||
});
|
||
copy.querySelectorAll('a').forEach(function (link, index) {
|
||
const href = linkURLs[index];
|
||
if (href) link.setAttribute('href', href); else link.removeAttribute('href');
|
||
});
|
||
return copy.innerHTML.trim();
|
||
}
|
||
|
||
function inferType(stem, options) {
|
||
const match = stem.match(TYPE_RE);
|
||
if (match) return match[1].replace(/题$/, '');
|
||
if (options.some(function (option) { return option.node.querySelector('.examtest_content_list_des_multiplesort'); })) return '多选';
|
||
const texts = options.map(function (option) { return option.text; });
|
||
if (texts.length === 2 && texts.every(function (value) { return /^(正确|错误|对|错|是|否)$/.test(value); })) return '判断';
|
||
return '未知';
|
||
}
|
||
|
||
function readAnswer(box) {
|
||
const reference = box.querySelector(SELECTORS.reference);
|
||
if (!reference) return '';
|
||
const pieces = Array.from(reference.querySelectorAll('.answer')).map(textOf).filter(Boolean);
|
||
if (pieces.length) {
|
||
// 多选可保存为一个 ABC span，也可分别保存为 A、B、C 多个 span。
|
||
if (pieces.every(function (piece) { return /^[A-Z\s,，、;；]+$/i.test(piece); })) return pieces.join('').replace(/[\s,，、;；]+/g, '').toUpperCase();
|
||
return pieces.join('；');
|
||
}
|
||
return textOf(reference).replace(/^\s*参考答案\s*[:：]\s*/, '').trim();
|
||
}
|
||
|
||
function answerOptions(answer, options) {
|
||
if (!answer) return [];
|
||
const normalizedAnswer = answer.replace(/[\s,，、;；]+/g, '').toUpperCase();
|
||
const letterLabels = options.every(function (option) { return /^[A-Z]$/i.test(option.label); });
|
||
if (letterLabels && /^[A-Z]+$/.test(normalizedAnswer)) {
|
||
return options.filter(function (option) { return normalizedAnswer.includes(option.label.toUpperCase()); });
|
||
}
|
||
const pieces = answer.split(/[\s,，、;；]+/).filter(Boolean);
|
||
return options.filter(function (option) { return pieces.includes(option.label) || pieces.includes(option.text); });
|
||
}
|
||
|
||
function extract(doc, sourceUrl) {
|
||
if (!doc || typeof doc.querySelectorAll !== 'function') throw new TypeError('extract 需要 Document 对象。');
|
||
const source = usableURL(sourceUrl) || usableURL(doc.__questionBankSourceURL) || savedSourceURL(doc) || usableURL(doc.URL) || '';
|
||
const resourceBase = baseURL(doc, source);
|
||
const title = textOf(doc.querySelector('.pc_examtest_title')) || normalize(doc.title) || '练习题库';
|
||
const questions = Array.from(doc.querySelectorAll(SELECTORS.question)).map(function (box, index) {
|
||
const stemNode = box.querySelector(SELECTORS.stem);
|
||
const rawStem = textOf(stemNode);
|
||
const optionItems = Array.from(box.querySelectorAll(SELECTORS.option)).map(function (node, optionIndex) {
|
||
const label = textOf(node.querySelector(SELECTORS.label)) || String.fromCharCode(65 + optionIndex);
|
||
const copy = cleanClone(node);
|
||
copy.querySelectorAll(SELECTORS.label).forEach(function (item) { item.remove(); });
|
||
// 保存网页把选项图片放在与文字相邻的 _list_img 内。
|
||
const container = node.parentElement;
|
||
const resources = imageResources(container || node, resourceBase);
|
||
return { node: node, label: label, text: normalize(copy.textContent), images: uniqueImages(resources), imageResources: resources, html: staticHTML(copy, resourceBase) };
|
||
});
|
||
const type = inferType(rawStem, optionItems);
|
||
const withoutType = rawStem.replace(TYPE_RE, '');
|
||
const numberMatch = withoutType.match(NUMBER_RE);
|
||
const number = numberMatch ? Number(numberMatch[1]) : index + 1;
|
||
const question = withoutType.replace(NUMBER_RE, '').trim();
|
||
const options = optionItems.map(function (option) {
|
||
return { label: option.label, text: option.text, images: option.images, imageResources: option.imageResources, html: option.html };
|
||
});
|
||
const answer = readAnswer(box);
|
||
const matchedOptions = answerOptions(answer, options);
|
||
const resolveNode = box.querySelector(SELECTORS.explanation);
|
||
const explanation = textOf(resolveNode).replace(/^\s*(?:题目解析|答案解析|解析)\s*[:：]\s*/, '').trim();
|
||
const questionResources = imageResources(stemNode, resourceBase);
|
||
const explanationResources = imageResources(resolveNode, resourceBase);
|
||
const allResources = questionResources.concat(options.flatMap(function (option) { return option.imageResources; }), explanationResources);
|
||
return {
|
||
number: number,
|
||
type: type,
|
||
question: question,
|
||
options: options,
|
||
answer: answer,
|
||
answerText: matchedOptions.map(function (option) { return option.label + '. ' + (option.text || '[图片选项]'); }).join('；'),
|
||
explanation: explanation,
|
||
images: uniqueImages(allResources),
|
||
questionImages: uniqueImages(questionResources),
|
||
explanationImages: uniqueImages(explanationResources),
|
||
imageResources: allResources,
|
||
questionHTML: staticHTML(stemNode, resourceBase),
|
||
explanationHTML: staticHTML(resolveNode, resourceBase)
|
||
};
|
||
});
|
||
const cardCount = doc.querySelectorAll(SELECTORS.card).length;
|
||
const expectedTotal = cardCount || null;
|
||
const missingAnswers = questions.filter(function (question) { return !question.answer; }).map(function (question) { return question.number; });
|
||
return {
|
||
title: title,
|
||
sourceUrl: source,
|
||
resourceBaseUrl: resourceBase,
|
||
exportedAt: new Date().toISOString(),
|
||
total: questions.length,
|
||
answered: questions.length - missingAnswers.length,
|
||
missingAnswers: missingAnswers,
|
||
expectedTotal: expectedTotal,
|
||
complete: questions.length > 0 && missingAnswers.length === 0 && (expectedTotal === null || expectedTotal === questions.length),
|
||
questions: questions
|
||
};
|
||
}
|
||
|
||
// 中和可能加载外部资源的属性后才交给 DOMParser；原值用于生成资源链接。
|
||
// 此函数不会把解析结果插入当前文档，也不执行保存网页中的任何脚本。
|
||
function parseHTML(html, sourceUrl) {
|
||
if (typeof DOMParser === 'undefined') throw new Error('此功能需要浏览器 DOMParser。');
|
||
let inert = String(html).replace(/<script\b[^>]*>[\s\S]*?<\/script\s*>/gi, '').replace(/<style\b[^>]*>[\s\S]*?<\/style\s*>/gi, '');
|
||
inert = inert.replace(/<[a-z][^>]*>/gi, function (tag) {
|
||
return tag.replace(/\s(src|srcset|href|poster|data|style|background|action|formaction)\s*=\s*("[^"]*"|'[^']*'|[^\s>]+)/gi, function (_, name, value) {
|
||
return ' data-qb-original-' + name.toLowerCase() + '=' + value;
|
||
}).replace(/\s(on[a-z]+)\s*=\s*("[^"]*"|'[^']*'|[^\s>]+)/gi, '');
|
||
});
|
||
const doc = new DOMParser().parseFromString(inert, 'text/html');
|
||
doc.__questionBankSourceURL = sourceUrl || '';
|
||
return doc;
|
||
}
|
||
|
||
function csvCell(value) {
|
||
let text = String(value == null ? '' : value);
|
||
// Excel 会在引号包裹的 CSV 单元格中依旧识别公式；前置单引号使其按文本打开。
|
||
if (/^[\s\uFEFF]*[=+\-@]/.test(text)) text = "'" + text;
|
||
return '"' + text.replace(/"/g, '""') + '"';
|
||
}
|
||
|
||
function toCSV(bank) {
|
||
const labels = [];
|
||
bank.questions.forEach(function (question) {
|
||
question.options.forEach(function (option) { if (!labels.includes(option.label)) labels.push(option.label); });
|
||
});
|
||
const header = ['题号', '题型', '题目'].concat(labels.map(function (label) { return '选项' + label; }), ['全部选项', '参考答案', '答案对应选项', '题目解析', '图片链接', '来源网址', '题库名称', '完整性', '网页题数', '缺失答案题号']);
|
||
const rows = bank.questions.map(function (question) {
|
||
return [question.number, question.type, question.question].concat(labels.map(function (label) {
|
||
const option = question.options.find(function (item) { return item.label === label; });
|
||
return option ? [option.text].concat(option.images).filter(Boolean).join('\n') : '';
|
||
}), [question.options.map(function (option) { return option.label + '. ' + option.text + (option.images.length ? '\n' + option.images.join('\n') : ''); }).join('\n'), question.answer, question.answerText, question.explanation, question.images.join('\n'), bank.sourceUrl, bank.title, bank.complete ? (bank.expectedTotal === null ? '当前页面完整（网页题数未知）' : '完整') : '不完整', bank.expectedTotal === null ? '未知' : bank.expectedTotal, bank.missingAnswers.join('、')]);
|
||
});
|
||
return '\uFEFF' + [header].concat(rows).map(function (row) { return row.map(csvCell).join(','); }).join('\r\n') + '\r\n';
|
||
}
|
||
|
||
function toJSON(bank) { return JSON.stringify(bank, null, 2); }
|
||
|
||
function toText(bank) {
|
||
const completeness = bank.complete ? (bank.expectedTotal === null ? '当前页面完整（网页题数未知）' : '完整') : '不完整';
|
||
const lines = [bank.title, '来源：' + (bank.sourceUrl || '本地网页'), '完整性：' + completeness, '题目：' + bank.total + '；有参考答案：' + bank.answered, '网页答题卡题数：' + (bank.expectedTotal === null ? '未知，只能核对当前页面' : bank.expectedTotal), '缺失答案题号：' + (bank.missingAnswers.length ? bank.missingAnswers.join('、') : '无')];
|
||
if (bank.expectedTotal !== null && bank.total < bank.expectedTotal) lines.push('缺失题目数：' + (bank.expectedTotal - bank.total));
|
||
lines.push('');
|
||
bank.questions.forEach(function (question) {
|
||
lines.push(question.number + '. [' + question.type + '] ' + question.question);
|
||
question.questionImages.forEach(function (url) { lines.push('题目图片：' + url); });
|
||
question.options.forEach(function (option) {
|
||
lines.push(option.label + '. ' + option.text);
|
||
option.images.forEach(function (url) { lines.push('选项图片：' + url); });
|
||
});
|
||
lines.push('参考答案：' + (question.answer || '网页未提供'));
|
||
if (question.answerText) lines.push('对应选项：' + question.answerText);
|
||
lines.push('题目解析：' + (question.explanation || '网页未提供'));
|
||
question.explanationImages.forEach(function (url) { lines.push('解析图片：' + url); });
|
||
lines.push('');
|
||
});
|
||
return lines.join('\r\n');
|
||
}
|
||
|
||
function typeCounts(bank) {
|
||
return bank.questions.reduce(function (counts, question) { counts[question.type] = (counts[question.type] || 0) + 1; return counts; }, {});
|
||
}
|
||
|
||
function filenameFor(bank, format) {
|
||
const name = (bank.title || '练习题库').replace(/[<>:"/\\|?*\x00-\x1F]/g, '_').replace(/[. ]+$/, '').slice(0, 100) || '练习题库';
|
||
return name + '_题库_' + bank.total + '题' + (bank.complete ? '' : '_不完整') + '.' + format;
|
||
}
|
||
|
||
function download(bank, format) {
|
||
format = String(format || 'csv').toLowerCase();
|
||
const formats = { csv: { encode: toCSV, mime: 'text/csv;charset=utf-8' }, json: { encode: toJSON, mime: 'application/json;charset=utf-8' }, txt: { encode: toText, mime: 'text/plain;charset=utf-8' } };
|
||
if (!formats[format]) throw new Error('导出格式仅支持 csv、json、txt。');
|
||
const filename = filenameFor(bank, format);
|
||
const blob = new Blob([formats[format].encode(bank)], { type: formats[format].mime });
|
||
const url = URL.createObjectURL(blob);
|
||
const link = root.document.createElement('a');
|
||
link.href = url;
|
||
link.download = filename;
|
||
root.document.body.appendChild(link);
|
||
link.click();
|
||
link.remove();
|
||
// 留出时间让浏览器接管下载后再回收 URL。
|
||
root.setTimeout(function () { URL.revokeObjectURL(url); }, 30000);
|
||
return filename;
|
||
}
|
||
|
||
root.QuestionBankCore = { version: '1.0.0', selectors: SELECTORS, extract: extract, parseHTML: parseHTML, savedSourceURL: savedSourceURL, resolveURL: resolveURL, toCSV: toCSV, toJSON: toJSON, toText: toText, typeCounts: typeCounts, filenameFor: filenameFor, download: download };
|
||
})(typeof window !== 'undefined' ? window : globalThis);
|