wikiparser-node 0.11.0-m → 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

@@ -66,10 +66,19 @@ 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();
-		this.data = text;
+		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 */
@@ -155,7 +164,12 @@ class AstText extends AstNode {
 	 */
 	#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});
+		}
 	}
 
 	/**
@@ -165,6 +179,87 @@ class AstText extends AstNode {
 	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

@@ -9,6 +9,10 @@ class Title {
 	ns = 0;
 	fragment;
 	encoded = false;
+	title = '';
+	main = '';
+	prefix = '';
+	interwiki = '';
 
 	/**
 	 * @param {string} title 标题（含或不含命名空间前缀）
@@ -32,6 +36,11 @@ class Title {
 			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()]];
@@ -45,12 +54,30 @@ class Title {
 		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.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

@@ -1,5 +1,7 @@
 'use strict';
 
+const Parser = require('..');
+
 /**
  * 解析后不可见的类
  * @template T
@@ -15,4 +17,5 @@ const hidden = Constructor => class extends Constructor {
 	}
 };
 
+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;

package/mixin/sol.js

@@ -0,0 +1,54 @@
+'use strict';
+
+const Parser = require('..'),
+	Token = require('../src');
+
+/**
+ * 只能位于行首的类
+ * @template T
+ * @param {T} Constructor 基类
+ * @returns {T}
+ */
+const sol = Constructor => class SolToken extends Constructor {
+	/**
+	 * 是否可以视为root节点
+	 * @this {Token}
+	 */
+	#isRoot() {
+		const {parentNode, type} = this;
+		return parentNode?.type === 'root'
+			|| type !== 'heading' && parentNode?.type === 'ext-inner' && parentNode.name === 'poem';
+	}
+
+	/**
+	 * 在前方插入newline
+	 * @this {SolToken & Token}
+	 */
+	prependNewLine() {
+		return (this.previousVisibleSibling || !this.#isRoot()) && !String(this.previousVisibleSibling).endsWith('\n')
+			? '\n'
+			: '';
+	}
+
+	/**
+	 * 还原为wikitext
+	 * @this {SolToken & Token}
+	 * @param {string} selector
+	 */
+	toString(selector) {
+		return selector && this.matches(selector) ? '' : `${this.prependNewLine()}${super.toString(selector)}`;
+	}
+
+	/** 获取padding */
+	getPadding() {
+		return this.prependNewLine().length;
+	}
+
+	/** 可见部分 */
+	text() {
+		return `${this.prependNewLine()}${super.text()}`;
+	}
+};
+
+Parser.mixins.sol = __filename;
+module.exports = sol;

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "wikiparser-node",
-	"version": "0.11.0-m",
+	"version": "0.11.0",
 	"description": "A Node.js parser for MediaWiki markup with AST",
 	"keywords": [
 		"mediawiki",
@@ -21,6 +21,7 @@
 		"/mixin/",
 		"/parser/",
 		"/src/",
+		"/tool/",
 		"/util/",
 		"/typings/",
 		"/dist/"
@@ -32,9 +33,11 @@
 	},
 	"scripts": {
 		"build": "rm -rf dist/ && tsc && grep -rl '/typings' dist/ | xargs gsed -i 's|/typings|/../typings|g' && node ./bin/tsc.js",
+		"doc": "node ./bin/doc.js",
+		"toc": "node ./bin/toc.js",
 		"test": "eslint . && ajv -s config/.schema.json -d 'config/*.json' --strict=true --strict-required=false",
 		"real": "node test/real.js",
-		"single": "node test/single.js"
+		"single": "node --prof test/single.js && node --prof-process isolate-0x*-v8.log > test/processed.txt && rm isolate-0x*-v8.log"
 	},
 	"devDependencies": {
 		"@types/node": "^17.0.23",

package/parser/brackets.js

@@ -124,4 +124,5 @@ const parseBrackets = (text, config = Parser.getConfig(), accum = []) => {
 	return text;
 };
 
+Parser.parsers.parseBrackets = __filename;
 module.exports = parseBrackets;

package/parser/commentAndExt.js

@@ -58,4 +58,5 @@ const parseCommentAndExt = (text, config = Parser.getConfig(), accum = [], inclu
 	);
 };
 
+Parser.parsers.parseCommentAndExt = __filename;
 module.exports = parseCommentAndExt;

package/parser/converter.js

@@ -42,4 +42,5 @@ const parseConverter = (wikitext, config = Parser.getConfig(), accum = []) => {
 	return wikitext;
 };
 
+Parser.parsers.parseConverter = __filename;
 module.exports = parseConverter;

package/parser/externalLinks.js

@@ -29,4 +29,5 @@ const parseExternalLinks = (wikitext, config = Parser.getConfig(), accum = []) =
 	});
 };
 
+Parser.parsers.parseExternalLinks = __filename;
 module.exports = parseExternalLinks;

package/parser/hrAndDoubleUnderscore.js

@@ -45,4 +45,5 @@ const parseHrAndDoubleUnderscore = ({firstChild: {data}, type, name}, config = P
 	return type === 'root' || type === 'ext-inner' && name === 'poem' ? data : data.slice(1);
 };
 
+Parser.parsers.parseHrAndDoubleUnderscore = __filename;
 module.exports = parseHrAndDoubleUnderscore;

package/parser/html.js

@@ -38,4 +38,5 @@ const parseHtml = (wikitext, config = Parser.getConfig(), accum = []) => {
 	return text;
 };
 
+Parser.parsers.parseHtml = __filename;
 module.exports = parseHtml;

package/parser/links.js

@@ -44,12 +44,12 @@ const parseLinks = (wikitext, config = Parser.getConfig(), accum = []) => {
 			continue;
 		}
 		const title = Parser.normalizeTitle(link, 0, false, config, true, true, true),
-			{ns, valid} = title;
+			{ns, interwiki, valid} = title;
 		if (!valid) {
 			s += `[[${x}`;
 			continue;
 		} else if (mightBeImg) {
-			if (ns !== 6) {
+			if (interwiki || ns !== 6) {
 				s += `[[${x}`;
 				continue;
 			}
@@ -79,9 +79,9 @@ const parseLinks = (wikitext, config = Parser.getConfig(), accum = []) => {
 		s += `\0${accum.length}l\x7F${after}`;
 		let SomeLinkToken = LinkToken;
 		if (!force) {
-			if (ns === 6) {
+			if (!interwiki && ns === 6) {
 				SomeLinkToken = FileToken;
-			} else if (ns === 14) {
+			} else if (!interwiki && ns === 14) {
 				SomeLinkToken = CategoryToken;
 			}
 		}
@@ -90,4 +90,5 @@ const parseLinks = (wikitext, config = Parser.getConfig(), accum = []) => {
 	return s;
 };
 
+Parser.parsers.parseLinks = __filename;
 module.exports = parseLinks;

package/parser/list.js

@@ -55,4 +55,5 @@ const parseList = (text, config = Parser.getConfig(), accum = []) => {
 	return text;
 };
 
+Parser.parsers.parseList = __filename;
 module.exports = parseList;

package/parser/magicLinks.js

@@ -37,4 +37,5 @@ const parseMagicLinks = (wikitext, config = Parser.getConfig(), accum = []) => {
 	});
 };
 
+Parser.parsers.parseMagicLinks = __filename;
 module.exports = parseMagicLinks;

package/parser/quotes.js

@@ -60,4 +60,5 @@ const parseQuotes = (text, config = Parser.getConfig(), accum = []) => {
 	return arr.join('');
 };
 
+Parser.parsers.parseQuotes = __filename;
 module.exports = parseQuotes;