wikiparser-node 0.11.0-b → 0.11.0

package/lib/ranges.js

@@ -0,0 +1,130 @@
+'use strict';
+
+const Parser = require('..');
+
+/** 模拟Python的Range对象。除`step`至少为`1`外，允许负数、小数或`end < start`的情形。 */
+class Range {
+	start;
+	end;
+	step;
+
+	/**
+	 * @param {string|Range} str 表达式
+	 * @throws `RangeError` 起点、终点和步长均应为整数
+	 * @throws `RangeError` n的系数不能为0
+	 * @throws `RangeError` 应使用CSS选择器或Python切片的格式
+	 */
+	constructor(str) {
+		if (str instanceof Range) {
+			Object.assign(this, str);
+			return;
+		}
+		str = str.trim();
+		if (str === 'odd') {
+			Object.assign(this, {start: 1, end: Infinity, step: 2});
+		} else if (str === 'even') {
+			Object.assign(this, {start: 0, end: Infinity, step: 2});
+		} else if (str.includes(':')) {
+			const [start, end, step = '1'] = str.split(':', 3);
+			this.start = Number(start);
+			this.end = Number(end || Infinity);
+			this.step = Math.max(Number(step), 1);
+			if (!Number.isInteger(this.start)) {
+				throw new RangeError(`起点 ${this.start} 应为整数！`);
+			} else if (this.end !== Infinity && !Number.isInteger(this.end)) {
+				throw new RangeError(`终点 ${this.end} 应为整数！`);
+			} else if (!Number.isInteger(this.step)) {
+				throw new RangeError(`步长 ${this.step} 应为整数！`);
+			}
+		} else {
+			const mt = /^([+-])?(\d+)?n(?:([+-])(\d+))?$/u.exec(str);
+			if (mt) {
+				const [, sgnA = '+', a = 1, sgnB = '+'] = mt,
+					b = Number(mt[4] ?? 0);
+				this.step = Number(a);
+				if (this.step === 0) {
+					throw new RangeError(`参数 ${str} 中 "n" 的系数不允许为 0！`);
+				} else if (sgnA === '+') {
+					this.start = sgnB === '+' || b === 0 ? b : this.step - 1 - (b - 1) % this.step;
+					this.end = Infinity;
+				} else if (sgnB === '-') {
+					this.start = 0;
+					this.end = b > 0 ? 0 : this.step;
+				} else {
+					this.start = b % this.step;
+					this.end = this.step + b;
+				}
+			} else {
+				throw new RangeError(`参数 ${str} 应写作CSS选择器的 "an+b" 形式或Python切片！`);
+			}
+		}
+	}
+
+	/**
+	 * 将Range转换为针对特定数组的下标集
+	 * @param {number|*[]} arr 参考数组
+	 * @complexity `n`
+	 */
+	applyTo(arr) {
+		return new Array(typeof arr === 'number' ? arr : arr.length).fill().map((_, i) => i).slice(this.start, this.end)
+			.filter((_, j) => j % this.step === 0);
+	}
+}
+
+/** @extends {Array<number|Range>} */
+class Ranges extends Array {
+	/** @param {number|string|Range|(number|string|Range)[]} arr 表达式数组 */
+	constructor(arr) {
+		super();
+		if (arr === undefined) {
+			return;
+		}
+		arr = Array.isArray(arr) ? arr : [arr];
+		for (const ele of arr) {
+			if (ele instanceof Range) {
+				this.push(new Range(ele));
+				continue;
+			}
+			const number = Number(ele);
+			if (Number.isInteger(number)) {
+				this.push(number);
+			} else if (typeof ele === 'string' && Number.isNaN(number)) {
+				try {
+					const range = new Range(ele);
+					this.push(range);
+				} catch {}
+			}
+		}
+	}
+
+	/**
+	 * 将Ranges转换为针对特定Array的下标集
+	 * @param {number|*[]} arr 参考数组
+	 * @complexity `n`
+	 */
+	applyTo(arr) {
+		const length = typeof arr === 'number' ? arr : arr.length;
+		return [
+			...new Set(
+				[...this].flatMap(ele => {
+					if (typeof ele === 'number') {
+						return ele < 0 ? ele + length : ele;
+					}
+					return ele.applyTo(length);
+				}),
+			),
+		].filter(i => i >= 0 && i < length).sort();
+	}
+
+	/**
+	 * 检查某个下标是否符合表达式
+	 * @param {string} str 表达式
+	 * @param {number} i 待检查的下标
+	 */
+	static nth(str, i) {
+		return new Ranges(str.split(',')).applyTo(i + 1).includes(i);
+	}
+}
+
+Parser.classes.Ranges = __filename;
+module.exports = Ranges;

package/lib/text.js

@@ -0,0 +1,265 @@
+'use strict';
+
+/** @typedef {import('../typings/token').LintError} LintError */
+
+const Parser = require('..'),
+	AstNode = require('./node'),
+	AstElement = require('./element');
+
+const errorSyntax = /https?:\/\/|\{+|\}+|\[{2,}|\[(?![^[]*\])|(?<=^|\])([^[]*?)\]+|\]{2,}|<\s*\/?([a-z]\w*)/giu,
+	errorSyntaxUrl = /\{+|\}+|\[{2,}|\[(?![^[]*\])|(?<=^|\])([^[]*?)\]+|\]{2,}|<\s*\/?([a-z]\w*)/giu,
+	disallowedTags = [
+		'html',
+		'base',
+		'head',
+		'style',
+		'title',
+		'body',
+		'menu',
+		'a',
+		'area',
+		'audio',
+		'img',
+		'map',
+		'track',
+		'video',
+		'embed',
+		'iframe',
+		'object',
+		'picture',
+		'source',
+		'canvas',
+		'script',
+		'col',
+		'colgroup',
+		'tbody',
+		'tfoot',
+		'thead',
+		'button',
+		'datalist',
+		'fieldset',
+		'form',
+		'input',
+		'label',
+		'legend',
+		'meter',
+		'optgroup',
+		'option',
+		'output',
+		'progress',
+		'select',
+		'textarea',
+		'details',
+		'dialog',
+		'slot',
+		'template',
+		'dir',
+		'frame',
+		'frameset',
+		'marquee',
+		'param',
+		'xmp',
+	];
+
+/** 文本节点 */
+class AstText extends AstNode {
+	type = 'text';
+	/** @type {string} */ data;
+
+	/** 文本长度 */
+	get length() {
+		return this.data.length;
+	}
+
+	/** @param {string} text 包含文本 */
+	constructor(text = '') {
+		super();
+		Object.defineProperties(this, {
+			data: {value: text, writable: false},
+			childNodes: {enumerable: false, configurable: false},
+			type: {enumerable: false, writable: false, configurable: false},
+		});
+	}
+
+	/** 输出字符串 */
+	toString() {
+		return this.data;
+	}
+
+	/** @override */
+	text() {
+		return this.data;
+	}
+
+	/**
+	 * Linter
+	 * @this {AstText & {parentNode: AstElement}}
+	 * @param {number} start 起始位置
+	 * @returns {LintError[]}
+	 */
+	lint(start = this.getAbsoluteIndex()) {
+		const {data, parentNode, nextSibling, previousSibling} = this,
+			type = parentNode?.type,
+			name = parentNode?.name,
+			nextType = nextSibling?.type,
+			previousType = previousSibling?.type,
+			errorRegex
+			= type === 'free-ext-link' || type === 'ext-link-url' || type === 'image-parameter' && name === 'link'
+				? errorSyntaxUrl
+				: errorSyntax,
+			errors = [...data.matchAll(errorRegex)],
+			{ext, html} = this.getRootNode().getAttribute('config');
+		if (errors.length > 0) {
+			const root = this.getRootNode(),
+				{top, left} = root.posFromIndex(start),
+				tags = new Set([ext, html, disallowedTags].flat(2));
+			return errors.map(/** @returns {LintError} */ ({0: error, 1: prefix, 2: tag, index}) => {
+				if (prefix) {
+					index += prefix.length;
+					error = error.slice(prefix.length);
+				}
+				const startIndex = start + index,
+					lines = data.slice(0, index).split('\n'),
+					startLine = lines.length + top - 1,
+					line = lines.at(-1),
+					startCol = lines.length > 1 ? line.length : left + line.length,
+					{0: char, length} = error,
+					endIndex = startIndex + length,
+					end = char === '}' || char === ']' ? endIndex + 1 : startIndex + 49,
+					rootStr = String(root),
+					nextChar = rootStr[endIndex],
+					previousChar = rootStr[startIndex - 1],
+					severity = length > 1 && (char !== '<' || /[\s/>]/u.test(nextChar))
+						|| char === '{' && (nextChar === char || previousChar === '-')
+						|| char === '}' && (previousChar === char || nextChar === '-')
+						|| char === '[' && (
+							nextChar === char || type === 'ext-link-text'
+							|| !data.slice(index + 1).trim() && nextType === 'free-ext-link'
+						)
+						|| char === ']' && (
+							previousChar === char
+							|| !data.slice(0, index).trim() && previousType === 'free-ext-link'
+						)
+						? 'error'
+						: 'warning';
+				return (char !== 'h' || index > 0) && (char !== '<' || tags.has(tag.toLowerCase())) && {
+					message: Parser.msg('lonely "$1"', char === 'h' ? error : char),
+					severity,
+					startIndex,
+					endIndex,
+					startLine,
+					endLine: startLine,
+					startCol,
+					endCol: startCol + length,
+					excerpt: rootStr.slice(Math.max(0, end - 50), end),
+				};
+			}).filter(Boolean);
+		}
+		return [];
+	}
+
+	/**
+	 * 修改内容
+	 * @param {string} text 新内容
+	 */
+	#setData(text) {
+		text = String(text);
+		const {data} = this,
+			e = new Event('text', {bubbles: true});
+		this.setAttribute('data', text);
+		if (data !== text) {
+			this.dispatchEvent(e, {oldText: data, newText: text});
+		}
+	}
+
+	/**
+	 * 替换字符串
+	 * @param {string} text 替换的字符串
+	 */
+	replaceData(text = '') {
+		this.#setData(text);
+	}
+
+	/** 复制 */
+	cloneNode() {
+		return new AstText(this.data);
+	}
+
+	/**
+	 * @override
+	 * @template {string} T
+	 * @param {T} key 属性键
+	 * @returns {import('../typings/node').TokenAttribute<T>}
+	 * @throws `Error` 文本节点没有子节点
+	 */
+	getAttribute(key) {
+		return key === 'verifyChild'
+			? () => {
+				throw new Error('文本节点没有子节点！');
+			}
+			: super.getAttribute(key);
+	}
+
+	/**
+	 * 在后方添加字符串
+	 * @param {string} text 添加的字符串
+	 */
+	appendData(text) {
+		this.#setData(this.data + text);
+	}
+
+	/**
+	 * 删减字符串
+	 * @param {number} offset 起始位置
+	 * @param {number} count 删减字符数
+	 */
+	deleteData(offset, count) {
+		this.#setData(this.data.slice(0, offset) + this.data.slice(offset + count));
+	}
+
+	/**
+	 * 插入字符串
+	 * @param {number} offset 插入位置
+	 * @param {string} text 待插入的字符串
+	 */
+	insertData(offset, text) {
+		this.#setData(this.data.slice(0, offset) + text + this.data.slice(offset));
+	}
+
+	/**
+	 * 提取子串
+	 * @param {number} offset 起始位置
+	 * @param {number} count 字符数
+	 */
+	substringData(offset, count) {
+		return this.data.slice(offset, offset + count);
+	}
+
+	/**
+	 * 将文本子节点分裂为两部分
+	 * @param {number} offset 分裂位置
+	 * @throws `RangeError` 错误的断开位置
+	 * @throws `Error` 没有父节点
+	 */
+	splitText(offset) {
+		if (!Number.isInteger(offset)) {
+			this.typeError('splitText', 'Number');
+		} else if (offset > this.length || offset < -this.length) {
+			throw new RangeError(`错误的断开位置！${offset}`);
+		}
+		const {parentNode, data} = this;
+		if (!parentNode) {
+			throw new Error('待分裂的文本节点没有父节点！');
+		}
+		const newText = new AstText(data.slice(offset)),
+			childNodes = [...parentNode.childNodes];
+		this.setAttribute('data', data.slice(0, offset));
+		childNodes.splice(childNodes.indexOf(this) + 1, 0, newText);
+		newText.setAttribute('parentNode', parentNode);
+		parentNode.setAttribute('childNodes', childNodes);
+		return newText;
+	}
+}
+
+Parser.classes.AstText = __filename;
+module.exports = AstText;

package/lib/title.js

@@ -0,0 +1,83 @@
+'use strict';
+
+const {decodeHtml} = require('../util/string'),
+	Parser = require('..');
+
+/** MediaWiki页面标题对象 */
+class Title {
+	valid = true;
+	ns = 0;
+	fragment;
+	encoded = false;
+	title = '';
+	main = '';
+	prefix = '';
+	interwiki = '';
+
+	/**
+	 * @param {string} title 标题（含或不含命名空间前缀）
+	 * @param {number} defaultNs 命名空间
+	 * @param {boolean} decode 是否需要解码
+	 * @param {boolean} selfLink 是否允许selfLink
+	 */
+	constructor(title, defaultNs = 0, config = Parser.getConfig(), decode = false, selfLink = false) {
+		const {namespaces, nsid} = config;
+		let namespace = namespaces[defaultNs];
+		title = decodeHtml(title);
+		if (decode && title.includes('%')) {
+			try {
+				const encoded = /%(?!21|3[ce]|5[bd]|7[b-d])[\da-f]{2}/iu.test(title);
+				title = decodeURIComponent(title);
+				this.encoded = encoded;
+			} catch {}
+		}
+		title = title.replaceAll('_', ' ').trim();
+		if (title[0] === ':') {
+			namespace = '';
+			title = title.slice(1).trim();
+		}
+		const iw = defaultNs ? undefined : Parser.isInterwiki(title, config);
+		if (iw) {
+			this.interwiki = iw[1].toLowerCase();
+			title = title.slice(iw[0].length);
+		}
+		const m = title.split(':');
+		if (m.length > 1) {
+			const id = namespaces[nsid[m[0].trim().toLowerCase()]];
+			if (id !== undefined) {
+				namespace = id;
+				title = m.slice(1).join(':').trim();
+			}
+		}
+		this.ns = nsid[namespace.toLowerCase()];
+		const i = title.indexOf('#');
+		let fragment;
+		if (i !== -1) {
+			fragment = title.slice(i + 1).trimEnd();
+			if (fragment.includes('%')) {
+				try {
+					fragment = decodeURIComponent(fragment);
+				} catch {}
+			} else if (fragment.includes('.')) {
+				try {
+					fragment = decodeURIComponent(fragment.replaceAll('.', '%'));
+				} catch {}
+			}
+			title = title.slice(0, i).trim();
+		}
+		this.valid = Boolean(title || selfLink && fragment !== undefined || this.interwiki)
+			&& !/\0\d+[eh!+-]\x7F|[<>[\]{}|]|%[\da-f]{2}/iu.test(title);
+		this.fragment = fragment;
+		this.main = title && `${title[0].toUpperCase()}${title.slice(1)}`;
+		this.prefix = `${namespace}${namespace && ':'}`;
+		this.title = `${iw ? `${this.interwiki}:` : ''}${this.prefix}${this.main.replaceAll(' ', '_')}`;
+	}
+
+	/** @override */
+	toString() {
+		return `${this.title}${this.fragment === undefined ? '' : `#${this.fragment}`}`;
+	}
+}
+
+Parser.classes.Title = __filename;
+module.exports = Title;

package/mixin/attributeParent.js

@@ -0,0 +1,117 @@
+'use strict';
+
+const Parser = require('..'),
+	AttributesToken = require('../src/attributes');
+
+/**
+ * 子节点含有AttributesToken的类
+ * @template T
+ * @param {T} Constructor 基类
+ * @param {number} i AttributesToken子节点的位置
+ * @returns {T}
+ */
+const attributeParent = (Constructor, i = 0) => class extends Constructor {
+	/** getAttrs()方法的getter写法 */
+	get attributes() {
+		return this.getAttrs();
+	}
+
+	/** 以字符串表示的class属性 */
+	get className() {
+		const attr = this.getAttr('class');
+		return typeof attr === 'string' ? attr : '';
+	}
+
+	set className(className) {
+		this.setAttr('class', className);
+	}
+
+	/** 以Set表示的class属性 */
+	get classList() {
+		return new Set(this.className.split(/\s/u));
+	}
+
+	/** id属性 */
+	get id() {
+		const attr = this.getAttr('id');
+		return typeof attr === 'string' ? attr : '';
+	}
+
+	set id(id) {
+		this.setAttr('id', id);
+	}
+
+	/**
+	 * AttributesToken子节点是否具有某属性
+	 * @this {{childNodes: AttributesToken[]}}
+	 * @param {string} key 属性键
+	 */
+	hasAttr(key) {
+		return this.childNodes.at(i).hasAttr(key);
+	}
+
+	/**
+	 * 获取AttributesToken子节点的属性
+	 * @this {{childNodes: AttributesToken[]}}
+	 * @param {string} key 属性键
+	 */
+	getAttr(key) {
+		return this.childNodes.at(i).getAttr(key);
+	}
+
+	/**
+	 * 列举AttributesToken子节点的属性键
+	 * @this {{childNodes: AttributesToken[]}}
+	 */
+	getAttrNames() {
+		return this.childNodes.at(i).getAttrNames();
+	}
+
+	/**
+	 * AttributesToken子节点是否具有任意属性
+	 * @this {{childNodes: AttributesToken[]}}
+	 */
+	hasAttrs() {
+		return this.childNodes.at(i).hasAttrs();
+	}
+
+	/**
+	 * 获取AttributesToken子节点的全部标签属性
+	 * @this {{childNodes: AttributesToken[]}}
+	 */
+	getAttrs() {
+		return this.childNodes.at(i).getAttrs();
+	}
+
+	/**
+	 * 对AttributesToken子节点设置属性
+	 * @this {{childNodes: AttributesToken[]}}
+	 * @param {string} key 属性键
+	 * @param {string|boolean} value 属性值
+	 */
+	setAttr(key, value) {
+		return this.childNodes.at(i).setAttr(key, value);
+	}
+
+	/**
+	 * 移除AttributesToken子节点的某属性
+	 * @this {{childNodes: AttributesToken[]}}
+	 * @param {string} key 属性键
+	 */
+	removeAttr(key) {
+		this.childNodes.at(i).removeAttr(key);
+	}
+
+	/**
+	 * 开关AttributesToken子节点的某属性
+	 * @this {{childNodes: AttributesToken[]}}
+	 * @param {string} key 属性键
+	 * @param {boolean|undefined} force 强制开启或关闭
+	 */
+	toggleAttr(key, force) {
+		this.childNodes.at(i).toggleAttr(key, force);
+	}
+};
+
+Parser.mixins.attributeParent = __filename;
+module.exports = attributeParent;

package/mixin/fixedToken.js

@@ -0,0 +1,40 @@
+'use strict';
+
+const Parser = require('..'),
+	Token = require('../src');
+
+/**
+ * 不可增删子节点的类
+ * @template T
+ * @param {T} Constructor 基类
+ * @returns {T}
+ */
+const fixedToken = Constructor => class extends Constructor {
+	static fixed = true;
+
+	/**
+	 * 移除子节点
+	 * @throws `Error`
+	 */
+	removeAt() {
+		throw new Error(`${this.constructor.name} 不可删除元素！`);
+	}
+
+	/**
+	 * 插入子节点
+	 * @template {Token} T
+	 * @param {T} token 待插入的子节点
+	 * @param {number} i 插入位置
+	 * @throws `Error`
+	 */
+	insertAt(token, i = this.length) {
+		if (Parser.running) {
+			super.insertAt(token, i);
+			return token;
+		}
+		throw new Error(`${this.constructor.name} 不可插入元素！`);
+	}
+};
+
+Parser.mixins.fixedToken = __filename;
+module.exports = fixedToken;

package/mixin/hidden.js

@@ -0,0 +1,21 @@
+'use strict';
+
+const Parser = require('..');
+
+/**
+ * 解析后不可见的类
+ * @template T
+ * @param {T} Constructor 基类
+ * @returns {T}
+ */
+const hidden = Constructor => class extends Constructor {
+	static hidden = true;
+
+	/** 没有可见部分 */
+	text() { // eslint-disable-line class-methods-use-this
+		return '';
+	}
+};
+
+Parser.mixins.hidden = __filename;
+module.exports = hidden;

package/mixin/singleLine.js

@@ -0,0 +1,31 @@
+'use strict';
+
+const Parser = require('..');
+
+/**
+ * 不可包含换行符的类
+ * @template {Function} T
+ * @param {T} Constructor 基类
+ * @returns {T}
+ */
+const singleLine = Constructor => {
+	const SingleLineConstructor = class extends Constructor {
+		/**
+		 * @override
+		 * @param {string} selector
+		 */
+		toString(selector) {
+			return super.toString(selector).replaceAll('\n', ' ');
+		}
+
+		/** @override */
+		text() {
+			return super.text().replaceAll('\n', ' ');
+		}
+	};
+	Object.defineProperty(SingleLineConstructor, 'name', {value: `SingleLine${Constructor.name}`});
+	return SingleLineConstructor;
+};
+
+Parser.mixins.singleLine = __filename;
+module.exports = singleLine;