wikiparser-node 0.9.2-b → 0.10.0-m

package/parser/quotes.js

@@ -0,0 +1,63 @@
+'use strict';
+
+const Parser = require('..'),
+	QuoteToken = require('../src/nowiki/quote');
+
+/**
+ * 解析单引号
+ * @param {string} text wikitext
+ * @param {accum} accum
+ */
+const parseQuotes = (text, config = Parser.getConfig(), accum = []) => {
+	const arr = text.split(/('{2,})/u),
+		{length} = arr;
+	if (length === 1) {
+		return text;
+	}
+	let nBold = 0,
+		nItalic = 0,
+		firstSingle, firstMulti, firstSpace;
+	for (let i = 1; i < length; i += 2) {
+		const {length: len} = arr[i];
+		switch (len) {
+			case 2:
+				nItalic++;
+				break;
+			case 4:
+				arr[i - 1] += `'`;
+				arr[i] = `'''`;
+				// fall through
+			case 3:
+				nBold++;
+				if (firstSingle) {
+					break;
+				} else if (arr[i - 1].endsWith(' ')) {
+					if (!firstMulti && !firstSpace) {
+						firstSpace = i;
+					}
+				} else if (arr[i - 1].at(-2) === ' ') {
+					firstSingle = i;
+				} else {
+					firstMulti ||= i;
+				}
+				break;
+			default:
+				arr[i - 1] += `'`.repeat(len - 5);
+				arr[i] = `'''''`;
+				nItalic++;
+				nBold++;
+		}
+	}
+	if (nItalic % 2 === 1 && nBold % 2 === 1) {
+		const i = firstSingle ?? firstMulti ?? firstSpace;
+		arr[i] = `''`;
+		arr[i - 1] += `'`;
+	}
+	for (let i = 1; i < length; i += 2) {
+		new QuoteToken(arr[i].length, config, accum);
+		arr[i] = `\0${accum.length - 1}q\x7F`;
+	}
+	return arr.join('');
+};
+
+module.exports = parseQuotes;

package/parser/table.js

@@ -0,0 +1,113 @@
+'use strict';
+
+const Parser = require('..'),
+	AstText = require('../lib/text'),
+	Token = require('../src'),
+	TableToken = require('../src/table'),
+	TrToken = require('../src/table/tr'),
+	TdToken = require('../src/table/td'),
+	DdToken = require('../src/nowiki/dd');
+
+/**
+ * 解析表格，注意`tr`和`td`包含开头的换行
+ * @param {Token & {firstChild: AstText}} root 根节点
+ * @param {accum} accum
+ */
+const parseTable = ({firstChild: {data}, type, name}, config = Parser.getConfig(), accum = []) => {
+	const /** @type {TrToken[]} */ stack = [],
+		lines = data.split('\n');
+	let out = type === 'root' || type === 'parameter-value' || type === 'ext-inner' && name === 'poem'
+		? ''
+		: `\n${lines.shift()}`;
+
+	/**
+	 * 向表格中插入纯文本
+	 * @param {string} str 待插入的文本
+	 * @param {TrToken} top 当前解析的表格或表格行
+	 */
+	const push = (str, top) => {
+		if (!top) {
+			out += str;
+			return;
+		}
+		const /** @type {Token}} */ {lastChild} = top;
+		if (lastChild.constructor === Token) {
+			lastChild.setText(String(lastChild) + str);
+		} else {
+			const token = new Token(str, config, true, accum);
+			token.type = 'table-inter';
+			top.insertAt(token.setAttribute('stage', 3));
+		}
+	};
+	for (const outLine of lines) {
+		let top = stack.pop();
+		const [spaces] = /^(?:\s|\0\d+c\x7F)*/u.exec(outLine),
+			line = outLine.slice(spaces.length),
+			matchesStart = /^(:*)((?:\s|\0\d+c\x7F)*)(\{\||\{(?:\0\d+c\x7F)*\0\d+!\x7F|\0\d+\{\x7F)(.*)$/u.exec(line);
+		if (matchesStart) {
+			while (top && top.type !== 'td') {
+				top = stack.pop();
+			}
+			const [, indent, moreSpaces, tableSyntax, attr] = matchesStart;
+			if (indent) {
+				new DdToken(indent, config, accum);
+			}
+			push(`\n${spaces}${indent && `\0${accum.length - 1}d\x7F`}${moreSpaces}\0${accum.length}b\x7F`, top);
+			const table = new TableToken(tableSyntax, attr, config, accum);
+			stack.push(...top ? [top] : [], table);
+			continue;
+		} else if (!top) {
+			out += `\n${outLine}`;
+			continue;
+		}
+		const matches = // eslint-disable-line operator-linebreak
+			/^(?:(\|\}|\0\d+!\x7F\}|\0\d+\}\x7F)|(\|-+|\0\d+!\x7F-+|\0\d+-\x7F-*)(?!-)|(!|(?:\||\0\d+!\x7F)\+?))(.*)$/u
+				.exec(line);
+		if (!matches) {
+			push(`\n${outLine}`, top);
+			stack.push(...top ? [top] : []);
+			continue;
+		}
+		const [, closing, row, cell, attr] = matches;
+		if (closing) {
+			while (!(top instanceof TableToken)) {
+				top = stack.pop();
+			}
+			top.close(`\n${spaces}${closing}`, true);
+			push(attr, stack.at(-1));
+		} else if (row) {
+			if (top.type === 'td') {
+				top = stack.pop();
+			}
+			if (top.type === 'tr') {
+				top = stack.pop();
+			}
+			const tr = new TrToken(`\n${spaces}${row}`, attr, config, accum);
+			stack.push(top, tr);
+			top.insertAt(tr);
+		} else {
+			if (top.type === 'td') {
+				top = stack.pop();
+			}
+			const regex = cell === '!'
+				? /!!|(?:\||\0\d+!\x7F){2}|\0\d+\+\x7F/gu
+				: /(?:\||\0\d+!\x7F){2}|\0\d+\+\x7F/gu;
+			let mt = regex.exec(attr),
+				lastIndex = 0,
+				lastSyntax = `\n${spaces}${cell}`;
+			while (mt) {
+				const td = new TdToken(lastSyntax, attr.slice(lastIndex, mt.index), config, accum);
+				top.insertAt(td);
+				({lastIndex} = regex);
+				[lastSyntax] = mt;
+				mt = regex.exec(attr);
+			}
+			const td = new TdToken(lastSyntax, attr.slice(lastIndex), config, accum);
+			stack.push(top, td);
+			top.insertAt(td);
+		}
+	}
+	return out.slice(1);
+};
+
+module.exports = parseTable;

package/src/arg.js

@@ -0,0 +1,93 @@
+'use strict';
+
+const {text} = require('../util/string'),
+	{generateForSelf, generateForChild} = require('../util/lint'),
+	Parser = require('..'),
+	Token = require('.');
+
+/**
+ * `{{{}}}`包裹的参数
+ * @classdesc `{childNodes: [AtomToken, ?Token, ...HiddenToken]}`
+ */
+class ArgToken extends Token {
+	type = 'arg';
+
+	/** default */
+	get default() {
+		return this.childNodes[1]?.text() ?? false;
+	}
+
+	/**
+	 * @param {string[]} parts 以'|'分隔的各部分
+	 * @param {accum} accum
+	 * @complexity `n`
+	 */
+	constructor(parts, config = Parser.getConfig(), accum = []) {
+		super(undefined, config, true, accum, {
+		});
+		for (let i = 0; i < parts.length; i++) {
+			if (i === 0 || i > 1) {
+				const AtomToken = i === 0 ? require('./atom') : require('./atom/hidden');
+				const token = new AtomToken(parts[i], i === 0 ? 'arg-name' : undefined, config, accum, {
+				});
+				super.insertAt(token);
+			} else {
+				const token = new Token(parts[i], config, true, accum);
+				token.type = 'arg-default';
+				super.insertAt(token.setAttribute('stage', 2));
+			}
+		}
+	}
+
+	/**
+	 * @override
+	 */
+	toString(selector) {
+		return `{{{${super.toString(selector, '|')}}}}`;
+	}
+
+	/**
+	 * @override
+	 * @returns {string}
+	 */
+	text() {
+		return `{{{${text(this.childNodes.slice(0, 2), '|')}}}}`;
+	}
+
+	/** @override */
+	getPadding() {
+		return 3;
+	}
+
+	/** @override */
+	getGaps() {
+		return 1;
+	}
+
+	/**
+	 * @override
+	 * @param {number} start 起始位置
+	 * @returns {LintError[]}
+	 */
+	lint(start = this.getAbsoluteIndex()) {
+		if (!this.getAttribute('include')) {
+			return [generateForSelf(this, {start}, 'unexpected template argument')];
+		}
+		const {childNodes: [argName, argDefault, ...rest]} = this,
+			errors = argName.lint(start + 3);
+		if (argDefault) {
+			errors.push(...argDefault.lint(start + 4 + String(argName).length));
+		}
+		if (rest.length > 0) {
+			const rect = {start, ...this.getRootNode().posFromIndex(start)};
+			errors.push(...rest.map(child => {
+				const error = generateForChild(child, rect, 'invisible content inside triple brackets'),
+					{startIndex, startCol, excerpt} = error;
+				return {...error, startIndex: startIndex - 1, startCol: startCol - 1, excerpt: `|${excerpt}`};
+			}));
+		}
+		return errors;
+	}
+}
+
+module.exports = ArgToken;

package/src/atom/hidden.js

@@ -0,0 +1,11 @@
+'use strict';
+
+const hidden = require('../../mixin/hidden'),
+	AtomToken = require('.');
+
+/** 不可见的节点 */
+class HiddenToken extends hidden(AtomToken) {
+	type = 'hidden';
+}
+
+module.exports = HiddenToken;

package/src/atom/index.js

@@ -0,0 +1,26 @@
+'use strict';
+
+const Parser = require('../..'),
+	Token = require('..');
+
+/**
+ * 不会被继续解析的plain Token
+ * @classdesc `{childNodes: ...AstText|Token}`
+ */
+class AtomToken extends Token {
+	type = 'plain';
+
+	/**
+	 * @param {string} wikitext wikitext
+	 * @param {string|undefined} type Token.type
+	 * @param {accum} accum
+	 */
+	constructor(wikitext, type, config = Parser.getConfig(), accum = [], acceptable = undefined) {
+		super(wikitext, config, true, accum, acceptable);
+		if (type) {
+			this.type = type;
+		}
+	}
+}
+
+module.exports = AtomToken;

package/src/attribute.js

@@ -0,0 +1,290 @@
+'use strict';
+
+const {generateForChild} = require('../util/lint'),
+	{removeComment} = require('../util/string'),
+	Parser = require('..'),
+	Token = require('.'),
+	AtomToken = require('./atom');
+
+const commonHtmlAttrs = new Set([
+		'id',
+		'class',
+		'style',
+		'lang',
+		'dir',
+		'title',
+		'tabindex',
+		'aria-describedby',
+		'aria-flowto',
+		'aria-hidden',
+		'aria-label',
+		'aria-labelledby',
+		'aria-owns',
+		'role',
+		'about',
+		'property',
+		'resource',
+		'datatype',
+		'typeof',
+		'itemid',
+		'itemprop',
+		'itemref',
+		'itemscope',
+		'itemtype',
+	]),
+	blockAttrs = new Set(['align']),
+	citeAttrs = new Set(['cite']),
+	citeAndAttrs = new Set(['cite', 'datetime']),
+	widthAttrs = new Set(['width']),
+	tdAttrs = new Set(
+		['align', 'valign', 'abbr', 'axis', 'headers', 'scope', 'rowspan', 'colspan', 'width', 'height', 'bgcolor'],
+	),
+	typeAttrs = new Set(['type']),
+	htmlAttrs = {
+		div: blockAttrs,
+		h1: blockAttrs,
+		h2: blockAttrs,
+		h3: blockAttrs,
+		h4: blockAttrs,
+		h5: blockAttrs,
+		h6: blockAttrs,
+		blockquote: citeAttrs,
+		q: citeAttrs,
+		p: blockAttrs,
+		br: new Set(['clear']),
+		pre: widthAttrs,
+		ins: citeAndAttrs,
+		del: citeAndAttrs,
+		ul: typeAttrs,
+		ol: new Set(['type', 'start', 'reversed']),
+		li: new Set(['type', 'value']),
+		table: new Set(
+			['summary', 'width', 'border', 'frame', 'rules', 'cellspacing', 'cellpadding', 'align', 'bgcolor'],
+		),
+		caption: blockAttrs,
+		tr: new Set(['bgcolor', 'align', 'valign']),
+		td: tdAttrs,
+		th: tdAttrs,
+		img: new Set(['alt', 'src', 'width', 'height', 'srcset']),
+		font: new Set(['size', 'color', 'face']),
+		hr: widthAttrs,
+		rt: new Set(['rbspan']),
+		data: new Set(['value']),
+		time: new Set(['datetime']),
+		meta: new Set(['itemprop', 'content']),
+		link: new Set(['itemprop', 'href', 'title']),
+		gallery: new Set(['mode', 'showfilename', 'caption', 'perrow', 'widths', 'heights', 'showthumbnails', 'type']),
+		poem: new Set(['compact', 'align']),
+		categorytree: new Set([
+			'align',
+			'hideroot',
+			'onlyroot',
+			'depth',
+			'mode',
+			'hideprefix',
+			'namespaces',
+			'showcount',
+			'notranslations',
+		]),
+		combooption: new Set(['name', 'for', 'inline', 'align']),
+	},
+	empty = new Set(),
+	extAttrs = {
+		nowiki: empty,
+		indicator: new Set(['name']),
+		langconvert: new Set(['from', 'to']),
+		ref: new Set(['group', 'name', 'extends', 'follow', 'dir']),
+		references: new Set(['group', 'responsive']),
+		charinsert: new Set(['label']),
+		choose: new Set(['uncached', 'before', 'after']),
+		option: new Set(['weight']),
+		imagemap: empty,
+		inputbox: empty,
+		templatestyles: new Set(['src', 'wrapper']),
+		dynamicpagelist: empty,
+		poll: new Set(['id', 'show-results-before-voting']),
+		sm2: typeAttrs,
+		flashmp3: typeAttrs,
+		tab: new Set([
+			'nested',
+			'name',
+			'index',
+			'class',
+			'block',
+			'inline',
+			'openname',
+			'closename',
+			'collapsed',
+			'dropdown',
+			'style',
+			'bgcolor',
+			'container',
+			'id',
+			'title',
+		]),
+		tabs: new Set(['plain', 'class', 'container', 'id', 'title', 'style']),
+		combobox: new Set(['placeholder', 'value', 'id', 'class', 'text', 'dropdown', 'style']),
+	},
+	insecureStyle = new RegExp(
+		`${
+			'expression'
+		}|${
+			'(?:filter|accelerator|-o-link(?:-source)?|-o-replace)\\s*:'
+		}|${
+			'(?:url|image(?:-set)?)\\s*\\('
+		}|${
+			'attr\\s*\\([^)]+[\\s,]url'
+		}`,
+		'u',
+	);
+
+/**
+ * 扩展和HTML标签属性
+ * @classdesc `{childNodes: [AtomToken, Token|AtomToken]}`
+ */
+class AttributeToken extends Token {
+	#equal;
+	#quotes;
+	#tag;
+
+	/** 引号是否匹配 */
+	get balanced() {
+		return !this.#equal || this.#quotes[0] === this.#quotes[1];
+	}
+
+	/** getValue()的getter */
+	get value() {
+		return this.getValue();
+	}
+
+	/** 标签名 */
+	get tag() {
+		return this.#tag;
+	}
+
+	/**
+	 * @param {'ext-attr'|'html-attr'|'table-attr'} type 标签类型
+	 * @param {string} tag 标签名
+	 * @param {string} key 属性名
+	 * @param {string} equal 等号
+	 * @param {string} value 属性值
+	 * @param {string[]} quotes 引号
+	 * @param {accum} accum
+	 */
+	constructor(type, tag, key, equal = '', value = '', quotes = [], config = Parser.getConfig(), accum = []) {
+		const keyToken = new AtomToken(key, 'attr-key', config, accum, {
+		});
+		let valueToken;
+		if (key === 'title') {
+			valueToken = new Token(value, config, true, accum, {
+			}).setAttribute('type', 'attr-value').setAttribute('stage', Parser.MAX_STAGE - 1);
+		} else if (tag === 'gallery' && key === 'caption') {
+			/** @type {ParserConfig} */
+			const newConfig = {...config, excludes: [...config.excludes, 'quote', 'extLink', 'magicLink', 'list']};
+			valueToken = new Token(value, newConfig, true, accum, {
+			}).setAttribute('type', 'attr-value').setAttribute('stage', 5);
+		} else if (tag === 'choose' && (key === 'before' || key === 'after')) {
+			/** @type {ParserConfig} */
+			const newConfig = {...config, excludes: [...config.excludes, 'heading', 'html', 'table', 'hr', 'list']};
+			valueToken = new Token(value, newConfig, true, accum, {
+			}).setAttribute('type', 'attr-value').setAttribute('stage', 1);
+		} else {
+			valueToken = new AtomToken(value, 'attr-value', config, accum, {
+			});
+		}
+		super(undefined, config, true, accum);
+		this.type = type;
+		this.append(keyToken, valueToken);
+		this.#equal = equal;
+		this.#quotes = quotes;
+		this.#tag = tag;
+		this.setAttribute('name', removeComment(key).trim().toLowerCase());
+	}
+
+	/** @override */
+	afterBuild() {
+		if (this.#equal.includes('\0')) {
+			this.#equal = this.getAttribute('buildFromStr')(this.#equal, 'string');
+		}
+		if (this.parentNode) {
+			this.#tag = this.parentNode.name;
+		}
+		this.setAttribute('name', this.firstChild.text().trim().toLowerCase());
+	}
+
+	/**
+	 * @override
+	 * @returns {string}
+	 */
+	toString(selector) {
+		const [quoteStart = '', quoteEnd = ''] = this.#quotes;
+		return this.#equal
+			? `${super.toString(selector, `${this.#equal}${quoteStart}`)}${quoteEnd}`
+			: this.firstChild.toString(selector);
+	}
+
+	/**
+	 * @override
+	 * @returns {string}
+	 */
+	text() {
+		return this.#equal ? `${super.text(`${this.#equal.trim()}"`)}"` : this.firstChild.text();
+	}
+
+	/** @override */
+	getGaps() {
+		return this.#equal ? this.#equal.length + (this.#quotes[0]?.length ?? 0) : 0;
+	}
+
+	/** @override */
+	print() {
+		const [quoteStart = '', quoteEnd = ''] = this.#quotes;
+		return this.#equal ? super.print({sep: `${this.#equal}${quoteStart}`, post: quoteEnd}) : super.print();
+	}
+
+	/**
+	 * @override
+	 * @param {number} start 起始位置
+	 */
+	lint(start = this.getAbsoluteIndex()) {
+		const errors = super.lint(start),
+			{balanced, firstChild, lastChild, type, name, parentNode, value} = this,
+			tagName = parentNode?.name;
+		let rect;
+		if (!balanced) {
+			const root = this.getRootNode();
+			rect = {start, ...root.posFromIndex(start)};
+			const e = generateForChild(lastChild, rect, 'unclosed quotes', 'warning'),
+				startIndex = e.startIndex - 1,
+				startCol = e.startCol - 1;
+			errors.push({...e, startIndex, startCol, excerpt: String(root).slice(startIndex, startIndex + 50)});
+		}
+		if (!/\{\{[^{]+\}\}/u.test(name) && (
+			type === 'ext-attr' && !(tagName in htmlAttrs) && extAttrs[tagName] && !extAttrs[tagName].has(name)
+			|| (type === 'html-attr' || type === 'table-attr' || tagName in htmlAttrs) && !htmlAttrs[tagName]?.has(name)
+			&& !/^(?:xmlns:[\w:.-]+|data-[^:]*)$/u.test(name)
+			&& (tagName === 'meta' || tagName === 'link' || !commonHtmlAttrs.has(name))
+		)) {
+			rect ||= {start, ...this.getRootNode().posFromIndex(start)};
+			errors.push(generateForChild(firstChild, rect, 'illegal attribute name'));
+		} else if (name === 'style' && typeof value === 'string' && insecureStyle.test(value)) {
+			rect ||= {start, ...this.getRootNode().posFromIndex(start)};
+			errors.push(generateForChild(lastChild, rect, 'insecure style'));
+		}
+		return errors;
+	}
+
+	/** 获取属性值 */
+	getValue() {
+		if (this.#equal) {
+			const value = this.lastChild.text();
+			if (this.#quotes[1]) {
+				return value;
+			}
+			return this.#quotes[0] ? value.trimEnd() : value.trim();
+		}
+		return true;
+	}
+}
+
+module.exports = AttributeToken;