wikiparser-node 0.4.0 → 0.5.0

package/index.js

@@ -75,6 +75,8 @@ const /** @type {Parser} */ Parser = {
 		'converter-rule-from': ['convert-rule-from', 'conversion-rule-from'],
 	},
 
+	promises: [Promise.resolve()],
+
 	warn(msg, ...args) {
 		if (this.warning) {
 			console.warn('\x1B[33m%s\x1B[0m', msg, ...args);
@@ -172,7 +174,7 @@ const /** @type {Parser} */ Parser = {
 
 	parse(wikitext, include, maxStage = Parser.MAX_STAGE, config = Parser.getConfig()) {
 		const Token = require('./src');
-		let token;
+		let /** @type {Token} */ token;
 		this.run(() => {
 			if (typeof wikitext === 'string') {
 				token = new Token(wikitext, config);
@@ -197,6 +199,27 @@ const /** @type {Parser} */ Parser = {
 				throw e;
 			}
 		});
+		if (this.debugging) {
+			let restored = String(token),
+				process = '解析';
+			if (restored === wikitext) {
+				const entities = {lt: '<', gt: '>', amp: '&'};
+				restored = token.print().replaceAll(
+					/<[^<]+?>|&([lg]t|amp);/gu,
+					/** @param {string} s */ (_, s) => s ? entities[s] : '',
+				);
+				process = '渲染HTML';
+			}
+			if (restored !== wikitext) {
+				const diff = require('./util/diff');
+				const {promises: {0: cur, length}} = this;
+				this.promises.unshift((async () => {
+					await cur;
+					this.error(`${process}过程中不可逆地修改了原始文本！`);
+					return diff(wikitext, restored, length);
+				})());
+			}
+		}
 		return token;
 	},
 
@@ -234,7 +257,7 @@ const /** @type {Parser} */ Parser = {
 
 const /** @type {PropertyDescriptorMap} */ def = {};
 for (const key in Parser) {
-	if (['aliases', 'MAX_STAGE', 'typeAliases'].includes(key)) {
+	if (['aliases', 'MAX_STAGE', 'typeAliases', 'promises'].includes(key)) {
 		def[key] = {enumerable: false, writable: false};
 	} else if (!['config', 'isInterwiki', 'normalizeTitle', 'parse', 'getTool'].includes(key)) {
 		def[key] = {enumerable: false};

package/lib/element.js

@@ -1,8 +1,9 @@
 'use strict';
 
 const fs = require('fs'),
+	path = require('path'),
 	{externalUse} = require('../util/debug'),
-	{toCase, noWrap} = require('../util/string'),
+	{toCase, noWrap, print} = require('../util/string'),
 	{nth} = require('./ranges'),
 	parseSelector = require('../parser/selector'),
 	Parser = require('..'),
@@ -138,6 +139,18 @@ class AstElement extends AstNode {
 		return previousSibling;
 	}
 
+	/** 内部高度 */
+	get clientHeight() {
+		const {innerText} = this;
+		return typeof innerText === 'string' ? innerText.split('\n').length : undefined;
+	}
+
+	/** 内部宽度 */
+	get clientWidth() {
+		const {innerText} = this;
+		return typeof innerText === 'string' ? innerText.split('\n').at(-1).length : undefined;
+	}
+
 	constructor() {
 		super();
 		this.seal('name');
@@ -228,11 +241,11 @@ class AstElement extends AstNode {
 		this.getAttribute('verifyChild')(i);
 		const /** @type {AstText} */ oldText = this.childNodes.at(i),
 			{type, data, constructor: {name}} = oldText;
-		if (type !== 'text') {
-			throw new RangeError(`第 ${i} 个子节点是 ${name}！`);
+		if (type === 'text') {
+			oldText.replaceData(str);
+			return data;
 		}
-		oldText.replaceData(str);
-		return data;
+		throw new RangeError(`第 ${i} 个子节点是 ${name}！`);
 	}
 
 	/** 是否受保护。保护条件来自Token，这里仅提前用于:required和:optional伪选择器。 */
@@ -298,7 +311,7 @@ class AstElement extends AstNode {
 							|| (type === 'file' || type === 'gallery-image' && link);
 					case ':local-link':
 						return (type === 'link' || type === 'file' || type === 'gallery-image')
-							&& link?.startsWith('#');
+							&& link?.[0] === '#';
 					case ':read-only':
 						return fixed;
 					case ':read-write':
@@ -358,7 +371,7 @@ class AstElement extends AstNode {
 			return Parser.run(() => {
 				const stack = parseSelector(selector),
 					/** @type {Set<string>} */
-					pseudos = new Set(stack.flat(2).filter(step => typeof step === 'string' && step.startsWith(':')));
+					pseudos = new Set(stack.flat(2).filter(step => typeof step === 'string' && step[0] === ':'));
 				if (pseudos.size > 0) {
 					Parser.warn('检测到伪选择器，请确认是否需要将":"转义成"\\:"。', pseudos);
 				}
@@ -521,204 +534,74 @@ class AstElement extends AstNode {
 		return String(this).split('\n', n + 1).at(-1);
 	}
 
-	/**
-	 * 将字符位置转换为行列号
-	 * @param {number} index 字符位置
-	 * @complexity `n`
-	 */
-	posFromIndex(index) {
-		if (typeof index !== 'number') {
-			this.typeError('posFromIndex', 'Number');
-		}
-		const text = String(this);
-		if (index >= -text.length && index < text.length && Number.isInteger(index)) {
-			const lines = text.slice(0, index).split('\n');
-			return {top: lines.length - 1, left: lines.at(-1).length};
-		}
-		return undefined;
-	}
-
-	/**
-	 * 将行列号转换为字符位置
-	 * @param {number} top 行号
-	 * @param {number} left 列号
-	 * @complexity `n`
-	 */
-	indexFromPos(top, left) {
-		if (typeof top !== 'number' || typeof left !== 'number') {
-			this.typeError('indexFromPos', 'Number');
-		}
-		const lines = String(this).split('\n');
-		return top >= 0 && left >= 0 && Number.isInteger(top) && Number.isInteger(left)
-			&& lines.length >= top + 1 && lines[top].length >= left
-			? lines.slice(0, top).reduce((acc, curLine) => acc + curLine.length + 1, 0) + left
-			: undefined;
-	}
-
-	/**
-	 * 获取行数和最后一行的列数
-	 * @complexity `n`
-	 */
-	#getDimension() {
-		const lines = String(this).split('\n');
-		return {height: lines.length, width: lines.at(-1).length};
-	}
+	static lintIgnoredHidden = new Set(['noinclude', 'double-underscore', 'hidden']);
+	static lintIgnoredSyntax = new Set(['magic-word-name', 'heading-trail', 'table-syntax']);
+	static lintIgnoredExt = new Set(['nowiki', 'pre', 'syntaxhighlight', 'source', 'math', 'timeline']);
 
 	/**
-	 * 获取当前节点的相对字符位置，或其第`j`个子节点的相对字符位置
-	 * @param {number|undefined} j 子节点序号
-	 * @complexity `n`
+	 * Linter
+	 * @param {number} start 起始位置
 	 */
-	getRelativeIndex(j) {
-		if (j !== undefined && typeof j !== 'number') {
-			this.typeError('getRelativeIndex', 'Number');
+	lint(start = 0) {
+		if (AstElement.lintIgnoredHidden.has(this.type) || AstElement.lintIgnoredSyntax.has(this.type)
+			|| this.type === 'ext-inner' && AstElement.lintIgnoredExt.has(this.name)
+		) {
+			return [];
 		}
-		let /** @type {this[]} */ childNodes;
-
-		/**
-		 * 获取子节点相对于父节点的字符位置，使用前需要先给`childNodes`赋值
-		 * @param {number} end 子节点序号
-		 * @param {this} parent 父节点
-		 * @returns {number}
-		 */
-		const getIndex = (end, parent) => childNodes.slice(0, end).reduce(
-			(acc, cur, i) => acc + String(cur).length + parent.getGaps(i),
-			0,
-		) + parent.getPadding();
-		if (j === undefined) {
-			const {parentNode} = this;
-			if (!parentNode) {
-				return 0;
-			}
-			({childNodes} = parentNode);
-			return getIndex(childNodes.indexOf(this), parentNode);
+		const /** @type {LintError[]} */ errors = [];
+		for (let i = 0, cur = start + this.getPadding(); i < this.childNodes.length; i++) {
+			const child = this.childNodes[i];
+			errors.push(...child.lint(cur));
+			cur += String(child).length + this.getGaps(i);
 		}
-		this.getAttribute('verifyChild')(j, 1);
-		({childNodes} = this);
-		return getIndex(j, this);
+		return errors;
 	}
 
 	/**
-	 * 获取当前节点的绝对位置
-	 * @returns {number}
-	 * @complexity `n`
-	 */
-	getAbsoluteIndex() {
-		const {parentNode} = this;
-		return parentNode ? parentNode.getAbsoluteIndex() + this.getRelativeIndex() : 0;
-	}
-
-	/**
-	 * 获取当前节点的相对位置，或其第`j`个子节点的相对位置
-	 * @param {number|undefined} j 子节点序号
-	 * @complexity `n`
+	 * 以HTML格式打印
+	 * @param {printOpt} opt 选项
+	 * @returns {string}
 	 */
-	#getPosition(j) {
-		return j === undefined
-			? this.parentNode?.posFromIndex(this.getRelativeIndex()) ?? {top: 0, left: 0}
-			: this.posFromIndex(this.getRelativeIndex(j));
-	}
-
-	/**
-	 * 获取当前节点的行列位置和大小
-	 * @complexity `n`
-	 */
-	getBoundingClientRect() {
-		const root = this.getRootNode();
-		return {...this.#getDimension(), ...root.posFromIndex(this.getAbsoluteIndex())};
-	}
-
-	/** 第一个子节点前的间距 */
-	getPadding() {
-		return 0;
-	}
-
-	/** 子节点间距 */
-	getGaps() {
-		return 0;
-	}
-
-	/**
-	 * 行数
-	 * @complexity `n`
-	 */
-	get offsetHeight() {
-		return this.#getDimension().height;
-	}
-
-	/**
-	 * 最后一行的列数
-	 * @complexity `n`
-	 */
-	get offsetWidth() {
-		return this.#getDimension().width;
-	}
-
-	/**
-	 * 行号
-	 * @complexity `n`
-	 */
-	get offsetTop() {
-		return this.#getPosition().top;
-	}
-
-	/**
-	 * 列号
-	 * @complexity `n`
-	 */
-	get offsetLeft() {
-		return this.#getPosition().left;
+	print(opt = {}) {
+		return this.childNodes.length === 0
+			? ''
+			: `<span class="wpb-${opt.class ?? this.type}">${print(this.childNodes, opt)}</span>`;
 	}
 
 	/**
-	 * 位置、大小和padding
-	 * @complexity `n`
+	 * 保存为JSON
+	 * @param {string} file 文件名
+	 * @returns {Record<string, any>}
 	 */
-	get style() {
-		return {...this.#getPosition(), ...this.#getDimension(), padding: this.getPadding()};
-	}
-
-	/** 内部高度 */
-	get clientHeight() {
-		const {innerText} = this;
-		return typeof innerText === 'string' ? innerText.split('\n').length : undefined;
-	}
-
-	/** 内部宽度 */
-	get clientWidth() {
-		const {innerText} = this;
-		return typeof innerText === 'string' ? innerText.split('\n').at(-1).length : undefined;
+	json(file) {
+		const {childNodes, ...prop} = this,
+			json = {
+				...prop,
+				childNodes: childNodes.map(child => child.type === 'text' ? String(child) : child.json()),
+			};
+		if (typeof file === 'string') {
+			fs.writeFileSync(
+				path.join(__dirname.slice(0, -4), 'printed', `${file}${file.endsWith('.json') ? '' : '.json'}`),
+				JSON.stringify(json, null, 2),
+			);
+		}
+		return json;
 	}
 
 	/**
 	 * 输出AST
-	 * @template {'markup'|'json'} T
-	 * @param {T} format 输出格式
-	 * @param {T extends 'markup' ? number : string} depth 输出深度
-	 * @returns {T extends 'markup' ? void : Record<string, any>}
+	 * @param {number} depth 当前深度
+	 * @returns {void}
 	 */
-	print(format = 'markup', depth = 0) {
-		if (format === 'json') {
-			const {childNodes, ...prop} = this,
-				json = {
-					...prop,
-					childNodes: childNodes.map(child => child.type === 'text' ? String(child) : child.print('json')),
-				};
-			if (typeof depth === 'string') {
-				fs.writeFileSync(
-					`${__dirname.slice(0, -3)}printed/${depth}${depth.endsWith('.json') ? '' : '.json'}`,
-					JSON.stringify(json, null, 2),
-				);
-			}
-			return json;
-		} else if (typeof depth !== 'number') {
+	echo(depth = 0) {
+		if (typeof depth !== 'number') {
 			this.typeError('print', 'Number');
 		}
 		const indent = '  '.repeat(depth),
 			str = String(this),
-			{childNodes, type, firstChild} = this,
+			{childNodes, type} = this,
 			{length} = childNodes;
-		if (!str || length === 0 || firstChild.type === 'text' && String(firstChild) === str) {
+		if (childNodes.every(child => child.type === 'text' || !String(child))) {
 			console.log(`${indent}\x1B[32m<%s>\x1B[0m${noWrap(str)}\x1B[32m</%s>\x1B[0m`, type, type);
 			return undefined;
 		}
@@ -736,11 +619,12 @@ class AstElement extends AstNode {
 			} else if (child.type === 'text') {
 				console.log(`${indent}  ${noWrap(String(child))}`);
 			} else {
-				child.print('markup', depth + 1);
+				child.echo(depth + 1);
 			}
-			i += childStr.length + gap;
+			i += childStr.length;
 			if (gap) {
-				console.log(`${indent}  ${noWrap(str.slice(i - gap, i))}`);
+				console.log(`${indent}  ${noWrap(str.slice(i, i + gap))}`);
+				i += gap;
 			}
 		}
 		if (i < str.length) {

package/lib/node.js

@@ -561,7 +561,9 @@ class AstNode {
 		const /** @type {AstText[]} */ childNodes = [...this.childNodes];
 		for (let i = childNodes.length - 1; i >= 0; i--) {
 			const {type, data} = childNodes[i];
-			if (data === '') {
+			if (this.getGaps(i - 1)) {
+				//
+			} else if (data === '') {
 				childNodes.splice(i, 1);
 			} else if (type === 'text' && childNodes[i - 1]?.type === 'text') {
 				childNodes[i - 1].setAttribute('data', childNodes[i - 1].data + data);
@@ -579,6 +581,162 @@ class AstNode {
 		}
 		return parentNode ?? this;
 	}
+
+	/**
+	 * 将字符位置转换为行列号
+	 * @param {number} index 字符位置
+	 * @complexity `n`
+	 */
+	posFromIndex(index) {
+		if (typeof index !== 'number') {
+			this.typeError('posFromIndex', 'Number');
+		}
+		const str = String(this);
+		if (index >= -str.length && index <= str.length && Number.isInteger(index)) {
+			const lines = str.slice(0, index).split('\n');
+			return {top: lines.length - 1, left: lines.at(-1).length};
+		}
+		return undefined;
+	}
+
+	/**
+	 * 将行列号转换为字符位置
+	 * @param {number} top 行号
+	 * @param {number} left 列号
+	 * @complexity `n`
+	 */
+	indexFromPos(top, left) {
+		if (typeof top !== 'number' || typeof left !== 'number') {
+			this.typeError('indexFromPos', 'Number');
+		}
+		const lines = String(this).split('\n');
+		return top >= 0 && left >= 0 && Number.isInteger(top) && Number.isInteger(left)
+			&& lines.length >= top + 1 && lines[top].length >= left
+			? lines.slice(0, top).reduce((acc, curLine) => acc + curLine.length + 1, 0) + left
+			: undefined;
+	}
+
+	/**
+	 * 获取行数和最后一行的列数
+	 * @complexity `n`
+	 */
+	#getDimension() {
+		const lines = String(this).split('\n');
+		return {height: lines.length, width: lines.at(-1).length};
+	}
+
+	/** 第一个子节点前的间距 */
+	getPadding() {
+		return 0;
+	}
+
+	/** 子节点间距 */
+	getGaps() {
+		return 0;
+	}
+
+	/**
+	 * 获取当前节点的相对字符位置，或其第`j`个子节点的相对字符位置
+	 * @param {number|undefined} j 子节点序号
+	 * @complexity `n`
+	 */
+	getRelativeIndex(j) {
+		if (j !== undefined && typeof j !== 'number') {
+			this.typeError('getRelativeIndex', 'Number');
+		}
+		let /** @type {this[]} */ childNodes;
+
+		/**
+		 * 获取子节点相对于父节点的字符位置，使用前需要先给`childNodes`赋值
+		 * @param {number} end 子节点序号
+		 * @param {this} parent 父节点
+		 * @returns {number}
+		 */
+		const getIndex = (end, parent) => childNodes.slice(0, end).reduce(
+			(acc, cur, i) => acc + String(cur).length + parent.getGaps(i),
+			0,
+		) + parent.getPadding();
+		if (j === undefined) {
+			const {parentNode} = this;
+			if (parentNode) {
+				({childNodes} = parentNode);
+				return getIndex(childNodes.indexOf(this), parentNode);
+			}
+			return 0;
+		}
+		this.getAttribute('verifyChild')(j, 1);
+		({childNodes} = this);
+		return getIndex(j, this);
+	}
+
+	/**
+	 * 获取当前节点的绝对位置
+	 * @returns {number}
+	 * @complexity `n`
+	 */
+	getAbsoluteIndex() {
+		const {parentNode} = this;
+		return parentNode ? parentNode.getAbsoluteIndex() + this.getRelativeIndex() : 0;
+	}
+
+	/**
+	 * 获取当前节点的相对位置，或其第`j`个子节点的相对位置
+	 * @param {number|undefined} j 子节点序号
+	 * @complexity `n`
+	 */
+	#getPosition(j) {
+		return j === undefined
+			? this.parentNode?.posFromIndex(this.getRelativeIndex()) ?? {top: 0, left: 0}
+			: this.posFromIndex(this.getRelativeIndex(j));
+	}
+
+	/**
+	 * 获取当前节点的行列位置和大小
+	 * @complexity `n`
+	 */
+	getBoundingClientRect() {
+		return {...this.#getDimension(), ...this.getRootNode().posFromIndex(this.getAbsoluteIndex())};
+	}
+
+	/**
+	 * 行数
+	 * @complexity `n`
+	 */
+	get offsetHeight() {
+		return this.#getDimension().height;
+	}
+
+	/**
+	 * 最后一行的列数
+	 * @complexity `n`
+	 */
+	get offsetWidth() {
+		return this.#getDimension().width;
+	}
+
+	/**
+	 * 相对于父容器的行号
+	 * @complexity `n`
+	 */
+	get offsetTop() {
+		return this.#getPosition().top;
+	}
+
+	/**
+	 * 相对于父容器的列号
+	 * @complexity `n`
+	 */
+	get offsetLeft() {
+		return this.#getPosition().left;
+	}
+
+	/**
+	 * 位置、大小和padding
+	 * @complexity `n`
+	 */
+	get style() {
+		return {...this.#getPosition(), ...this.#getDimension(), padding: this.getPadding()};
+	}
 }
 
 Parser.classes.AstNode = __filename;

package/lib/ranges.js

@@ -66,8 +66,7 @@ class Range {
 	 * @complexity `n`
 	 */
 	applyTo(arr) {
-		return new Array(typeof arr === 'number' ? arr : arr.length).fill().map((_, i) => i)
-			.slice(this.start, this.end)
+		return new Array(typeof arr === 'number' ? arr : arr.length).fill().map((_, i) => i).slice(this.start, this.end)
 			.filter((_, j) => j % this.step === 0);
 	}
 }

package/lib/text.js

@@ -33,6 +33,7 @@ class AstText extends AstNode {
 	 * @template {string} T
 	 * @param {T} key 属性键
 	 * @returns {TokenAttribute<T>}
+	 * @throws `Error` 文本节点没有子节点
 	 */
 	getAttribute(key) {
 		return key === 'verifyChild'
@@ -70,7 +71,6 @@ class AstText extends AstNode {
 	/**
 	 * 在后方添加字符串
 	 * @param {string} text 添加的字符串
-	 * @throws `Error` 禁止外部调用
 	 */
 	appendData(text) {
 		this.#setData(this.data + text);
@@ -80,8 +80,6 @@ class AstText extends AstNode {
 	 * 删减字符串
 	 * @param {number} offset 起始位置
 	 * @param {number} count 删减字符数
-	 * @throws `RangeError` 错误的删减位置
-	 * @throws `Error` 禁止外部调用
 	 */
 	deleteData(offset, count) {
 		this.#setData(this.data.slice(0, offset) + this.data.slice(offset + count));
@@ -91,8 +89,6 @@ class AstText extends AstNode {
 	 * 插入字符串
 	 * @param {number} offset 插入位置
 	 * @param {string} text 待插入的字符串
-	 * @throws `RangeError` 错误的插入位置
-	 * @throws `Error` 禁止外部调用
 	 */
 	insertData(offset, text) {
 		this.#setData(this.data.slice(0, offset) + text + this.data.slice(offset));
@@ -101,7 +97,6 @@ class AstText extends AstNode {
 	/**
 	 * 替换字符串
 	 * @param {string} text 替换的字符串
-	 * @throws `Error` 禁止外部调用
 	 */
 	replaceData(text = '') {
 		this.#setData(text);
@@ -140,6 +135,40 @@ class AstText extends AstNode {
 		parentNode.setAttribute('childNodes', childNodes);
 		return newText;
 	}
+
+	static errorSyntax = /[{}]+|\[{2,}|\[(?!(?:(?!https?\b)[^[])*\])|(?<=^|\])([^[]*?)\]+|<(?=\s*\/?\w+[\s/>])/giu;
+
+	/**
+	 * Linter
+	 * @param {number} start 起始位置
+	 * @returns {LintError[]}
+	 */
+	lint(start = 0) {
+		const {data} = this,
+			errors = [...data.matchAll(AstText.errorSyntax)];
+		if (errors.length > 0) {
+			const {top, left} = this.getRootNode().posFromIndex(start);
+			return errors.map(({0: error, 1: prefix, index}) => {
+				if (prefix) {
+					index += prefix.length;
+					error = error.slice(prefix.length);
+				}
+				const lines = data.slice(0, index).split('\n'),
+					startLine = lines.length + top - 1,
+					{length} = lines.at(-1),
+					startCol = lines.length > 1 ? length : left + length;
+				return {
+					message: `孤立的"${error[0]}"`,
+					severity: error[0] === '{' || error[0] === '}' ? 'error' : 'warning',
+					startLine,
+					endLine: startLine,
+					startCol,
+					endCol: startCol + error.length,
+				};
+			});
+		}
+		return [];
+	}
 }
 
 Parser.classes.AstText = __filename;

package/lib/title.js

@@ -55,7 +55,7 @@ class Title {
 		}
 		this.main = title && `${title[0].toUpperCase()}${title.slice(1)}`;
 		this.prefix = `${namespace}${namespace && ':'}`;
-		this.title = `${iw ? `${this.interwiki}:` : ''}${this.prefix}${this.main}`;
+		this.title = `${iw ? `${this.interwiki}:` : ''}${this.prefix}${this.main.replaceAll(' ', '_')}`;
 		this.valid = Boolean(this.main || this.fragment) && !/\0\d+[eh!+-]\x7F|[<>[\]{}|]/u.test(this.title);
 	}
 

package/mixin/fixedToken.js

@@ -28,11 +28,11 @@ const fixedToken = Constructor => class extends Constructor {
 	 * @throws `Error`
 	 */
 	insertAt(token, i = this.childNodes.length) {
-		if (!Parser.running) {
-			throw new Error(`${this.constructor.name} 不可插入元素！`);
+		if (Parser.running) {
+			super.insertAt(token, i);
+			return token;
 		}
-		super.insertAt(token, i);
-		return token;
+		throw new Error(`${this.constructor.name} 不可插入元素！`);
 	}
 };