wikiparser-node 0.3.1 → 0.5.0

package/src/link/index.js

@@ -1,9 +1,10 @@
 'use strict';
 
-const Title = require('../../lib/title'), // eslint-disable-line no-unused-vars
-	{text, noWrap} = require('../../util/string'),
+const Title = require('../../lib/title'),
+	{noWrap} = require('../../util/string'),
 	{undo} = require('../../util/debug'),
-	/** @type {Parser} */ Parser = require('../..'),
+	Parser = require('../..'),
+	AstText = require('../../lib/text'),
 	Token = require('..');
 
 /**
@@ -12,14 +13,67 @@ const Title = require('../../lib/title'), // eslint-disable-line no-unused-vars
  */
 class LinkToken extends Token {
 	type = 'link';
-	selfLink;
-	fragment;
-	interwiki;
+
+	/** 完整链接，和FileToken保持一致 */
+	get link() {
+		return String(this.#getTitle());
+	}
+
+	set link(link) {
+		this.setTarget(link);
+	}
+
+	/** 是否链接到自身 */
+	get selfLink() {
+		return !this.#getTitle().title;
+	}
+
+	set selfLink(selfLink) {
+		if (selfLink === true) {
+			this.asSelfLink();
+		}
+	}
+
+	/** fragment */
+	get fragment() {
+		return this.#getTitle().fragment;
+	}
+
+	set fragment(fragment) {
+		this.setFragment(fragment);
+	}
+
+	/** interwiki */
+	get interwiki() {
+		return this.#getTitle().interwiki;
+	}
+
+	set interwiki(interwiki) {
+		if (typeof interwiki !== 'string') {
+			this.typeError('set interwiki', 'String');
+		}
+		const {prefix, main, fragment} = this.#getTitle(),
+			link = `${interwiki}:${prefix}${main}${fragment && '#'}${fragment}`;
+		if (interwiki && !this.isInterwiki(link)) {
+			throw new RangeError(`${interwiki} 不是合法的跨维基前缀!`);
+		}
+		this.setTarget(link);
+	}
+
+	/** 链接显示文字 */
+	get innerText() {
+		if (this.type === 'link') {
+			return this.childNodes.length > 1
+				? this.lastChild.text()
+				: this.firstChild.text().replace(/^\s*:/u, '');
+		}
+		return undefined;
+	}
 
 	/**
-	 * @param {string} link
-	 * @param {string|undefined} linkText
-	 * @param {Title} title
+	 * @param {string} link 链接标题
+	 * @param {string|undefined} linkText 链接显示文字
+	 * @param {Title} title 链接标题对象
 	 * @param {accum} accum
 	 */
 	constructor(link, linkText, title, config = Parser.getConfig(), accum = []) {
@@ -33,149 +87,175 @@ class LinkToken extends Token {
 			inner.type = 'link-text';
 			this.appendChild(inner.setAttribute('stage', Parser.MAX_STAGE - 1));
 		}
-		this.selfLink = !title.title;
-		this.fragment = title.fragment;
-		this.interwiki = title.interwiki;
-		this.setAttribute('name', title.title).seal(['selfLink', 'fragment', 'interwiki']).protectChildren(0);
+		this.setAttribute('name', title.title).getAttribute('protectChildren')(0);
 	}
 
+	/** 生成Title对象 */
+	#getTitle() {
+		return this.normalizeTitle(this.firstChild.text());
+	}
+
+	/** @override */
 	cloneNode() {
-		const [link, ...linkText] = this.cloneChildren();
+		const [link, ...linkText] = this.cloneChildNodes();
 		return Parser.run(() => {
-			const /** @type {typeof LinkToken} */ Constructor = this.constructor,
-				token = new Constructor('', undefined, {
-					title: this.name, interwiki: this.interwiki, fragment: this.fragment,
-				}, this.getAttribute('config'));
-			token.firstElementChild.safeReplaceWith(link);
+			/** @type {this & {constructor: typeof LinkToken}} */
+			const {constructor} = this,
+				token = new constructor('', undefined, this.#getTitle(), this.getAttribute('config'));
+			token.firstChild.safeReplaceWith(link);
 			token.append(...linkText);
 			return token.afterBuild();
 		});
 	}
 
+	/**
+	 * @override
+	 * @throws `Error` 非法的内链目标
+	 * @throws `Error` 不可更改命名空间
+	 */
 	afterBuild() {
-		if (this.name.includes('\0')) {
-			this.setAttribute('name', text(this.buildFromStr(this.name)));
-		}
-		if (this.fragment.includes('\0')) {
-			this.setAttribute('fragment', text(this.buildFromStr(this.fragment)));
-		}
-		const that = this;
 		const /** @type {AstListener} */ linkListener = (e, data) => {
 			const {prevTarget} = e;
 			if (prevTarget?.type === 'link-target') {
 				const name = prevTarget.text(),
-					{title, interwiki, fragment, ns, valid} = that.normalizeTitle(name);
+					{title, interwiki, ns, valid} = this.normalizeTitle(name);
 				if (!valid) {
 					undo(e, data);
 					throw new Error(`非法的内链目标：${name}`);
-				} else if (that.type === 'category' && (interwiki || ns !== 14)
-					|| that.type === 'file' && (interwiki || ns !== 6)
+				} else if (this.type === 'category' && (interwiki || ns !== 14)
+					|| this.type === 'file' && (interwiki || ns !== 6)
 				) {
 					undo(e, data);
-					throw new Error(`${that.type === 'file' ? '文件' : '分类'}链接不可更改命名空间：${name}`);
-				} else if (that.type === 'link' && !interwiki && [6, 14].includes(ns) && !name.trim().startsWith(':')) {
-					const {firstChild} = prevTarget;
-					if (typeof firstChild === 'string') {
-						prevTarget.setText(`:${firstChild}`);
+					throw new Error(`${this.type === 'file' ? '文件' : '分类'}链接不可更改命名空间：${name}`);
+				} else if (this.type === 'link' && !interwiki && (ns === 6 || ns === 14) && name.trim()[0] !== ':') {
+					const /** @type {{firstChild: AstText}} */ {firstChild} = prevTarget;
+					if (firstChild.type === 'text') {
+						firstChild.insertData(0, ':');
 					} else {
 						prevTarget.prepend(':');
 					}
 				}
-				that.setAttribute('selfLink', !title).setAttribute('interwiki', interwiki)
-					.setAttribute('name', title).setAttribute('fragment', fragment);
+				this.setAttribute('name', title);
 			}
 		};
 		this.addEventListener(['remove', 'insert', 'replace', 'text'], linkListener);
 		return this;
 	}
 
-	toString() {
-		const str = super.toString('|');
-		return this.type === 'gallery-image' ? str : `[[${str}]]`;
+	/**
+	 * @override
+	 * @param {string} selector
+	 */
+	toString(selector) {
+		const str = super.toString(selector, '|');
+		return this.type === 'gallery-image' || selector && this.matches(selector) ? str : `[[${str}]]`;
 	}
 
+	/** @override */
 	getPadding() {
 		return 2;
 	}
 
+	/** @override */
 	getGaps() {
 		return 1;
 	}
 
+	/** @override */
+	print() {
+		return super.print(this.type === 'gallery-image' ? {sep: '|'} : {pre: '[[', post: ']]', sep: '|'});
+	}
+
+	/** @override */
 	text() {
 		const str = super.text('|');
 		return this.type === 'gallery-image' ? str : `[[${str}]]`;
 	}
 
-	/** @param {string} link */
+	/**
+	 * 设置链接目标
+	 * @param {string} link 链接目标
+	 * @throws `SyntaxError` 非法的链接目标
+	 */
 	setTarget(link) {
 		link = String(link);
-		if (link.type === 'link' && !/^\s*[:#]/.test(link)) {
+		if (this.type === 'link' && !/^\s*[:#]/u.test(link)) {
 			link = `:${link}`;
 		}
 		const root = Parser.parse(`[[${link}]]`, this.getAttribute('include'), 6, this.getAttribute('config')),
-			{childNodes: {length}, firstElementChild} = root;
-		if (length !== 1 || firstElementChild?.type !== this.type || firstElementChild.childNodes.length !== 1) {
+			{childNodes: {length}, firstChild: wikiLink} = root,
+			{type, firstChild, childNodes: {length: linkLength}} = wikiLink;
+		if (length !== 1 || type !== this.type || linkLength !== 1) {
 			const msgs = {link: '内链', file: '文件链接', category: '分类', 'gallery-image': '文件链接'};
 			throw new SyntaxError(`非法的${msgs[this.type]}目标：${link}`);
 		}
-		const {firstChild} = firstElementChild;
-		root.destroy();
-		firstElementChild.destroy();
-		this.firstElementChild.safeReplaceWith(firstChild);
+		wikiLink.destroy(true);
+		this.firstChild.safeReplaceWith(firstChild);
 	}
 
 	/**
-	 * @param {string} lang
-	 * @param {string} link
+	 * 设置跨语言链接
+	 * @param {string} lang 语言前缀
+	 * @param {string} link 页面标题
+	 * @throws `SyntaxError` 非法的跨语言链接
 	 */
 	setLangLink(lang, link) {
 		if (typeof lang !== 'string') {
 			this.typeError('setLangLink', 'String');
 		}
 		link = String(link).trim();
-		if (link.startsWith('#')) {
+		const [char] = link;
+		if (char === '#') {
 			throw new SyntaxError(`跨语言链接不能仅为fragment！`);
-		} else if (link.startsWith(':')) {
+		} else if (char === ':') {
 			link = link.slice(1);
 		}
 		const root = Parser.parse(`[[${lang}:${link}]]`, this.getAttribute('include'), 6, this.getAttribute('config')),
-			/** @type {Token & {firstElementChild: LinkToken}} */ {childNodes: {length}, firstElementChild} = root;
-		if (length !== 1 || firstElementChild?.type !== 'link' || firstElementChild.childNodes.length !== 1
-			|| firstElementChild.interwiki !== lang.toLowerCase()
-		) {
+			/** @type {Token & {firstChild: LinkToken}} */ {childNodes: {length}, firstChild: wikiLink} = root,
+			{type, childNodes: {length: linkLength}, interwiki, firstChild} = wikiLink;
+		if (length !== 1 || type !== 'link' || linkLength !== 1 || interwiki !== lang.toLowerCase()) {
 			throw new SyntaxError(`非法的跨语言链接目标：${lang}:${link}`);
 		}
-		const {firstChild} = firstElementChild;
-		root.destroy();
-		firstElementChild.destroy();
-		this.firstElementChild.safeReplaceWith(firstChild);
+		wikiLink.destroy(true);
+		this.firstChild.safeReplaceWith(firstChild);
 	}
 
-	/** @param {string} fragment */
+	/**
+	 * 设置fragment
+	 * @param {string} fragment fragment
+	 * @param {boolean} page 是否是其他页面
+	 * @throws `SyntaxError` 非法的fragment
+	 */
 	#setFragment(fragment, page = true) {
-		fragment = String(fragment).replace(/[<>[]#|=!\]/g, p => encodeURIComponent(p));
+		fragment = String(fragment).replaceAll(/[<>[\]#|=]/gu, p => encodeURIComponent(p));
 		const include = this.getAttribute('include'),
 			config = this.getAttribute('config'),
 			root = Parser.parse(`[[${page ? `:${this.name}` : ''}#${fragment}]]`, include, 6, config),
-			{childNodes: {length}, firstElementChild} = root;
-		if (length !== 1 || firstElementChild?.type !== 'link' || firstElementChild.childNodes.length !== 1) {
+			{childNodes: {length}, firstChild: wikiLink} = root,
+			{type, childNodes: {length: linkLength}, firstChild} = wikiLink;
+		if (length !== 1 || type !== 'link' || linkLength !== 1) {
 			throw new SyntaxError(`非法的 fragment：${fragment}`);
 		}
 		if (page) {
 			Parser.warn(`${this.constructor.name}.setFragment 方法会同时规范化页面名！`);
 		}
-		const {firstChild} = firstElementChild;
-		root.destroy();
-		firstElementChild.destroy();
-		this.firstElementChild.safeReplaceWith(firstChild);
+		wikiLink.destroy(true);
+		this.firstChild.safeReplaceWith(firstChild);
 	}
 
-	/** @param {string} fragment */
+	/**
+	 * 设置fragment
+	 * @param {string} fragment fragment
+	 */
 	setFragment(fragment) {
 		this.#setFragment(fragment);
 	}
 
+	/**
+	 * 修改为到自身的链接
+	 * @param {string} fragment fragment
+	 * @throws `RangeError` 空fragment
+	 */
 	asSelfLink(fragment = this.fragment) {
 		fragment = String(fragment);
 		if (!fragment.trim()) {
@@ -184,46 +264,55 @@ class LinkToken extends Token {
 		this.#setFragment(fragment, false);
 	}
 
+	/**
+	 * 设置链接显示文字
+	 * @param {string} linkText 链接显示文字
+	 * @throws `SyntaxError` 非法的链接显示文字
+	 */
 	setLinkText(linkText = '') {
 		linkText = String(linkText);
-		let lastElementChild;
+		let lastChild;
 		const config = this.getAttribute('config');
 		if (linkText) {
 			const root = Parser.parse(`[[${
 					this.type === 'category' ? 'Category:' : ''
 				}L|${linkText}]]`, this.getAttribute('include'), 6, config),
-				{childNodes: {length}, firstElementChild} = root;
-			if (length !== 1 || firstElementChild?.type !== this.type || firstElementChild.childNodes.length !== 2) {
+				{childNodes: {length}, firstChild: wikiLink} = root;
+			if (length !== 1 || wikiLink.type !== this.type || wikiLink.childNodes.length !== 2) {
 				throw new SyntaxError(`非法的${this.type === 'link' ? '内链文字' : '分类关键字'}：${noWrap(linkText)}`);
 			}
-			({lastElementChild} = firstElementChild);
+			({lastChild} = wikiLink);
 		} else {
-			lastElementChild = Parser.run(() => new Token('', config));
-			lastElementChild.setAttribute('stage', 7).type = 'link-text';
+			lastChild = Parser.run(() => new Token('', config));
+			lastChild.setAttribute('stage', 7).type = 'link-text';
 		}
 		if (this.childNodes.length === 1) {
-			this.appendChild(lastElementChild);
+			this.appendChild(lastChild);
 		} else {
-			this.lastElementChild.safeReplaceWith(lastElementChild);
+			this.lastChild.safeReplaceWith(lastChild);
 		}
 	}
 
+	/**
+	 * 自动生成管道符后的链接文字
+	 * @throws `Error` 带有"#"或"%"时不可用
+	 */
 	pipeTrick() {
-		const linkText = this.firstElementChild.text();
-		if (/[#%]/.test(linkText)) {
+		const linkText = this.firstChild.text();
+		if (linkText.includes('#') || linkText.includes('%')) {
 			throw new Error('Pipe trick 不能用于带有"#"或"%"的场合！');
 		}
-		const m1 = /^:?(?:[ \w\x80-\xff-]+:)?([^(]+)\(.+\)$/.exec(linkText);
+		const m1 = /^:?(?:[ \w\x80-\xFF-]+:)?([^(]+)\(.+\)$/u.exec(linkText);
 		if (m1) {
 			this.setLinkText(m1[1].trim());
 			return;
 		}
-		const m2 = /^:?(?:[ \w\x80-\xff-]+:)?([^（]+)（.+）$/.exec(linkText);
+		const m2 = /^:?(?:[ \w\x80-\xFF-]+:)?([^（]+)（.+）$/u.exec(linkText);
 		if (m2) {
 			this.setLinkText(m2[1].trim());
 			return;
 		}
-		const m3 = /^:?(?:[ \w\x80-\xff-]+:)?(.+?)(?:(?<!\()\(.+\))?(?:, |，|، )./.exec(linkText);
+		const m3 = /^:?(?:[ \w\x80-\xFF-]+:)?(.+?)(?:(?<!\()\(.+\))?(?:, |，|، )./u.exec(linkText);
 		if (m3) {
 			this.setLinkText(m3[1].trim());
 			return;

package/src/magicLink.js

@@ -1,52 +1,94 @@
 'use strict';
 
-const /** @type {Parser} */ Parser = require('..'),
+const {generateForChild} = require('../util/lint'),
+	Parser = require('..'),
+	AstText = require('../lib/text'),
 	Token = require('.');
 
 /**
  * 自由外链
- * @classdesc `{childNodes: [...string|CommentToken|IncludeToken|NoincludeToken]}`
+ * @classdesc `{childNodes: [...AstText|CommentToken|IncludeToken|NoincludeToken]}`
  */
 class MagicLinkToken extends Token {
 	type = 'free-ext-link';
 	#protocolRegex;
 
+	/** 协议 */
 	get protocol() {
 		return this.#protocolRegex.exec(this.text())?.[0];
 	}
+
 	set protocol(value) {
 		if (typeof value !== 'string') {
 			this.typeError('protocol', 'String');
 		}
-		if (!RegExp(`${this.#protocolRegex.source}$`, 'i').test(value)) {
+		if (!new RegExp(`${this.#protocolRegex.source}$`, 'iu').test(value)) {
 			throw new RangeError(`非法的外链协议：${value}`);
 		}
 		this.replaceChildren(this.text().replace(this.#protocolRegex, value));
 	}
 
+	/** 和内链保持一致 */
+	get link() {
+		return this.text();
+	}
+
+	set link(url) {
+		this.setTarget(url);
+	}
+
 	/**
-	 * @param {string} url
+	 * @param {string} url 网址
+	 * @param {boolean} doubleSlash 是否接受"//"作为协议
 	 * @param {accum} accum
 	 */
-	constructor(url, doubleSlash = false, config = Parser.getConfig(), accum = []) {
+	constructor(url, doubleSlash, config = Parser.getConfig(), accum = []) {
 		super(url, config, true, accum, {'Stage-1': ':', '!ExtToken': ''});
 		if (doubleSlash) {
 			this.type = 'ext-link-url';
 		}
-		this.#protocolRegex = RegExp(`^(?:${config.protocol}${doubleSlash ? '|//' : ''})`, 'i');
+		this.#protocolRegex = new RegExp(`^(?:${config.protocol}${doubleSlash ? '|//' : ''})`, 'iu');
 	}
 
+	/**
+	 * @override
+	 * @param {number} start 起始位置
+	 */
+	lint(start = 0) {
+		const errors = super.lint(start);
+		let /** @type {{top: number, left: number}} */ rect;
+		for (const child of this.childNodes) {
+			const str = String(child);
+			if (child.type !== 'text' || !/[，；。：！？（）【】]/u.test(str)) {
+				continue;
+			}
+			rect ||= this.getRootNode().posFromIndex(start);
+			const refError = generateForChild(child, rect, 'URL中的全角标点', 'warning');
+			errors.push(...[...str.matchAll(/[，；。：！？（）【】]/gu)].map(({index}) => {
+				const lines = str.slice(0, index).split('\n'),
+					{length: top} = lines,
+					{length: left} = lines.at(-1),
+					startLine = refError.startLine + top - 1,
+					startCol = top > 1 ? left : refError.startCol + left;
+				return {...refError, startLine, endLine: startLine, startCol, endCol: startCol + 1};
+			}));
+		}
+		return errors;
+	}
+
+	/** @override */
 	afterBuild() {
-		const ParameterToken = require('./parameter'),
-			/** @type {ParameterToken} */ parameter = this.closest('parameter');
+		const ParameterToken = require('./parameter');
+		const /** @type {ParameterToken} */ parameter = this.closest('parameter');
 		if (parameter?.getValue() === this.text()) {
-			this.replaceWith(this.toString());
+			this.replaceWith(...this.childNodes);
 		}
 		return this;
 	}
 
+	/** @override */
 	cloneNode() {
-		const cloned = this.cloneChildren(),
+		const cloned = this.cloneChildNodes(),
 			token = Parser.run(() => new MagicLinkToken(
 				undefined, this.type === 'ext-link-url', this.getAttribute('config'),
 			));
@@ -55,6 +97,10 @@ class MagicLinkToken extends Token {
 		return token;
 	}
 
+	/**
+	 * 获取网址
+	 * @throws `Error` 非标准协议
+	 */
 	getUrl() {
 		let url = this.text();
 		if (url.startsWith('//')) {
@@ -70,15 +116,19 @@ class MagicLinkToken extends Token {
 		}
 	}
 
-	/** @param {string|URL} url */
+	/**
+	 * 设置外链目标
+	 * @param {string|URL} url 含协议的网址
+	 * @throws `SyntaxError` 非法的自由外链目标
+	 */
 	setTarget(url) {
 		url = String(url);
 		const root = Parser.parse(url, this.getAttribute('include'), 9, this.getAttribute('config')),
-			{childNodes: {length}, firstElementChild} = root;
-		if (length !== 1 || firstElementChild?.type !== 'free-ext-link') {
+			{childNodes: {length}, firstChild: freeExtLink} = root;
+		if (length !== 1 || freeExtLink.type !== 'free-ext-link') {
 			throw new SyntaxError(`非法的自由外链目标：${url}`);
 		}
-		this.replaceChildren(...firstElementChild.childNodes);
+		this.replaceChildren(...freeExtLink.childNodes);
 	}
 }
 

package/src/nowiki/comment.js

@@ -1,44 +1,70 @@
 'use strict';
 
 const hidden = require('../../mixin/hidden'),
-	/** @type {Parser} */ Parser = require('../..'),
+	{generateForSelf} = require('../../util/lint'),
+	Parser = require('../..'),
 	NowikiToken = require('.');
 
 /**
  * HTML注释，不可见
- * @classdesc `{childNodes: [string]}`
+ * @classdesc `{childNodes: [AstText]}`
  */
 class CommentToken extends hidden(NowikiToken) {
 	type = 'comment';
 	closed;
 
+	/** 内部wikitext */
+	get innerText() {
+		return String(this.firstChild);
+	}
+
 	/**
-	 * @param {string} wikitext
+	 * @param {string} wikitext wikitext
+	 * @param {boolean} closed 是否闭合
 	 * @param {accum} accum
 	 */
 	constructor(wikitext, closed = true, config = Parser.getConfig(), accum = []) {
 		super(wikitext, config, accum);
 		this.closed = closed;
+		Object.defineProperty(this, 'closed', {enumerable: false});
 	}
 
-	/** @this {CommentToken & {firstChild: string}} */
+	/** @override */
 	cloneNode() {
-		return Parser.run(() => new CommentToken(this.firstChild, this.closed, this.getAttribute('config')));
+		return Parser.run(() => new CommentToken(String(this.firstChild), this.closed, this.getAttribute('config')));
 	}
 
-	/** @this {CommentToken & {firstChild: string}} */
-	toString() {
-		const {firstChild, closed, nextSibling} = this;
-		if (!closed && nextSibling) {
+	/**
+	 * @override
+	 * @param {string} selector
+	 */
+	toString(selector) {
+		if (!this.closed && this.nextSibling) {
 			Parser.error('自动闭合HTML注释', this);
 			this.closed = true;
 		}
-		return `<!--${firstChild}${this.closed ? '-->' : ''}`;
+		return selector && this.matches(selector)
+			? ''
+			: `<!--${String(this.firstChild)}${this.closed ? '-->' : ''}`;
 	}
 
+	/** @override */
 	getPadding() {
 		return 4;
 	}
+
+	/** @override */
+	print() {
+		return super.print({pre: '&lt;!--', post: this.closed ? '--&gt;' : ''});
+	}
+
+	/**
+	 * @override
+	 * @param {number} start 起始位置
+	 */
+	lint(start = 0) {
+		return this.closed ? [] : [generateForSelf(this, this.getRootNode().posFromIndex(start), '未闭合的HTML注释')];
+	}
 }
 
 Parser.classes.CommentToken = __filename;