wikiparser-node 0.4.0 → 0.5.0

package/mixin/sol.js

@@ -9,25 +9,35 @@ const Parser = require('..'),
  * @param {T} Constructor 基类
  * @returns {T}
  */
-const sol = Constructor => class extends Constructor {
+const sol = Constructor => class SolToken extends Constructor {
 	/**
-	 * 在前方插入newline
+	 * 是否可以视为root节点
 	 * @this {Token}
+	 * @param {boolean} includeHeading 是否包括HeadingToken
+	 */
+	#isRoot(includeHeading) {
+		const {parentNode, type} = this;
+		return parentNode?.type === 'root'
+			|| parentNode?.type === 'ext-inner' && parentNode?.name === 'poem'
+				&& (includeHeading || type !== 'heading');
+	}
+
+	/**
+	 * 在前方插入newline
+	 * @this {SolToken & Token}
 	 */
 	prependNewLine() {
-		const {previousVisibleSibling = '', parentNode} = this;
-		return (previousVisibleSibling || parentNode?.type !== 'root') && String(previousVisibleSibling).at(-1) !== '\n'
+		return (this.previousVisibleSibling || !this.#isRoot()) && String(this.previousVisibleSibling).at(-1) !== '\n'
 			? '\n'
 			: '';
 	}
 
 	/**
 	 * 在后方插入newline
-	 * @this {Token}
+	 * @this {SolToken & Token}
 	 */
 	appendNewLine() {
-		const {nextVisibleSibling = '', parentNode} = this;
-		return (nextVisibleSibling || parentNode?.type !== 'root') && String(nextVisibleSibling ?? '')[0] !== '\n'
+		return (this.nextVisibleSibling || !this.#isRoot(true)) && String(this.nextVisibleSibling ?? '')[0] !== '\n'
 			? '\n'
 			: '';
 	}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wikiparser-node",
-  "version": "0.4.0",
+  "version": "0.5.0",
   "description": "A Node.js parser for MediaWiki markup with AST",
   "keywords": [
     "mediawiki",
@@ -13,6 +13,16 @@
   },
   "license": "GPL-3.0",
   "author": "Bhsd",
+  "files": [
+    "/index.js",
+    "/parser/",
+    "/util/",
+    "/lib/",
+    "/src/",
+    "/config/",
+    "/mixin/",
+    "/tool/"
+  ],
   "repository": {
     "type": "git",
     "url": "git+https://github.com/bhsd-harry/wikiparser-node.git"

package/parser/commentAndExt.js

@@ -35,7 +35,7 @@ const parseCommentAndExt = (text, config = Parser.getConfig(), accum = [], inclu
 			+ `<(${noincludeRegex})(\\s[^>]*?)?(?:/>|>(.*?)(?:</(\\5\\s*)>|$))`, // <noinclude>
 			'gisu',
 		);
-	return text.replace(
+	return text.replaceAll(
 		regex,
 		/** @type {function(...string): string} */
 		(substr, name, attr, inner, closing, include, includeAttr, includeInner, includeClosing) => {

package/parser/converter.js

@@ -22,7 +22,7 @@ const parseConverter = (wikitext, config = Parser.getConfig(), accum = []) => {
 				str = wikitext.slice(top.index + 2, index),
 				i = str.indexOf('|'),
 				[flags, text] = i === -1 ? [[], str] : [str.slice(0, i).split(';'), str.slice(i + 1)],
-				temp = text.replace(/(&[#a-z\d]+);/iu, '$1\x01'),
+				temp = text.replaceAll(/(&[#a-z\d]+);/giu, '$1\x01'),
 				variants = `(?:${config.variants.join('|')})`,
 				rules = temp.split(new RegExp(`;(?=\\s*(?:${variants}|[^;]*?=>\\s*${variants})\\s*:)`, 'u'))
 					.map(rule => rule.replaceAll('\x01', ';'));

package/parser/externalLinks.js

@@ -14,7 +14,7 @@ const parseExternalLinks = (wikitext, config = Parser.getConfig(), accum = []) =
 		`\\[((?:${config.protocol}|//)${extUrlChar})(\\p{Zs}*)([^\\]\x01-\x08\x0A-\x1F\uFFFD]*)\\]`,
 		'giu',
 	);
-	return wikitext.replace(regex, /** @type {function(...string): string} */ (_, url, space, text) => {
+	return wikitext.replaceAll(regex, /** @type {function(...string): string} */ (_, url, space, text) => {
 		const {length} = accum,
 			mt = /&[lg]t;/u.exec(url);
 		if (mt) {

package/parser/hrAndDoubleUnderscore.js

@@ -1,18 +1,19 @@
 'use strict';
 
 const Parser = require('..'),
-	AstText = require('../lib/text');
+	AstText = require('../lib/text'),
+	Token = require('../src');
 
 /**
  * 解析\<hr\>和状态开关
- * @param {{firstChild: AstText, type: string}} root 根节点
+ * @param {Token & {firstChild: AstText}} root 根节点
  * @param {accum} accum
  */
-const parseHrAndDoubleUnderscore = ({firstChild: {data}, type}, config = Parser.getConfig(), accum = []) => {
+const parseHrAndDoubleUnderscore = ({firstChild: {data}, type, name}, config = Parser.getConfig(), accum = []) => {
 	const HrToken = require('../src/nowiki/hr'),
 		DoubleUnderscoreToken = require('../src/nowiki/doubleUnderscore');
 	const {doubleUnderscore} = config;
-	if (type !== 'root') {
+	if (type !== 'root' && (type !== 'ext-inner' || name !== 'poem')) {
 		data = `\0${data}`;
 	}
 	data = data.replaceAll(/^((?:\0\d+c\x7F)*)(-{4,})/gmu, (_, lead, m) => {
@@ -28,7 +29,7 @@ const parseHrAndDoubleUnderscore = ({firstChild: {data}, type}, config = Parser.
 			return m;
 		},
 	);
-	return type === 'root' ? data : data.slice(1);
+	return type === 'root' || type === 'ext-inner' && name === 'poem' ? data : data.slice(1);
 };
 
 Parser.parsers.parseHrAndDoubleUnderscore = __filename;

package/parser/links.js

@@ -1,7 +1,6 @@
 'use strict';
 
-const Parser = require('..'),
-	Token = require('../src');
+const Parser = require('..');
 
 /**
  * 解析内部链接

package/parser/magicLinks.js

@@ -11,7 +11,7 @@ const {extUrlChar} = require('../util/string'),
 const parseMagicLinks = (wikitext, config = Parser.getConfig(), accum = []) => {
 	const MagicLinkToken = require('../src/magicLink');
 	const regex = new RegExp(`\\b(?:${config.protocol})(${extUrlChar})`, 'giu');
-	return wikitext.replace(regex, /** @param {string} p1 */ (m, p1) => {
+	return wikitext.replaceAll(regex, /** @param {string} p1 */ (m, p1) => {
 		let trail = '',
 			url = m;
 		const m2 = /&(?:lt|gt|nbsp|#x0*(?:3[ce]|a0)|#0*(?:6[02]|160));/iu.exec(url);

package/parser/selector.js

@@ -46,7 +46,7 @@ const /** @type {pseudo[]} */ simplePseudos = [
 		['&', '&'],
 	],
 	pseudoRegex = new RegExp(`:(${complexPseudos.join('|')})$`, 'u'),
-	regularRegex = /[[(,>+~]|\s+/u, // eslint-disable-line regexp/no-super-linear-move
+	regularRegex = /[[(,>+~]|\s+/u,
 	attributeRegex = /^\s*(\w+)\s*(?:([~|^$*!]?=)\s*("[^"]*"|'[^']*'|[^\s[\]]+)(?:\s+(i))?\s*)?\]/u,
 	functionRegex = /^(\s*"[^"]*"\s*|\s*'[^']*'\s*|[^()]*)\)/u;
 
@@ -164,11 +164,11 @@ const parseSelector = selector => {
 		}
 		mt = regex.exec(sanitized);
 	}
-	if (regex !== regularRegex) {
-		throw new SyntaxError(`非法的选择器！\n${selector}\n检测到未闭合的'${regex === attributeRegex ? '[' : '('}'`);
+	if (regex === regularRegex) {
+		pushSimple(step, sanitized);
+		return stack;
 	}
-	pushSimple(step, sanitized);
-	return stack;
+	throw new SyntaxError(`非法的选择器！\n${selector}\n检测到未闭合的'${regex === attributeRegex ? '[' : '('}'`);
 };
 
 Parser.parsers.parseSelector = __filename;

package/parser/table.js

@@ -1,22 +1,22 @@
 'use strict';
 
 const Parser = require('..'),
-	AstText = require('../lib/text');
+	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 {{firstChild: AstText, type: string}} root 根节点
+ * @param {Token & {firstChild: AstText}} root 根节点
  * @param {accum} accum
  */
-const parseTable = ({firstChild: {data}, type}, config = Parser.getConfig(), accum = []) => {
-	const Token = require('../src'),
-		TableToken = require('../src/table'),
-		TrToken = require('../src/table/tr'),
-		TdToken = require('../src/table/td'),
-		DdToken = require('../src/nowiki/dd');
+const parseTable = ({firstChild: {data}, type, name}, config = Parser.getConfig(), accum = []) => {
 	const /** @type {TrToken[]} */ stack = [],
 		lines = data.split('\n');
-	let out = type === 'root' ? '' : `\n${lines.shift()}`;
+	let out = type === 'root' || type === 'ext-inner' && name === 'poem' ? '' : `\n${lines.shift()}`;
 
 	/**
 	 * 向表格中插入纯文本
@@ -28,9 +28,9 @@ const parseTable = ({firstChild: {data}, type}, config = Parser.getConfig(), acc
 			out += str;
 			return;
 		}
-		const {lastElementChild} = top;
-		if (lastElementChild.isPlain()) {
-			lastElementChild.setText(String(lastElementChild) + str);
+		const /** @type {Token & {firstChild: AstText}} */ {lastChild} = top;
+		if (lastChild.isPlain()) {
+			lastChild.firstChild.appendData(str);
 		} else {
 			const token = new Token(str, config, true, accum);
 			token.type = 'table-inter';

package/src/arg.js

@@ -1,6 +1,7 @@
 'use strict';
 
 const {text, noWrap} = require('../util/string'),
+	{generateForChild} = require('../util/lint'),
 	Parser = require('..'),
 	Token = require('.');
 
@@ -20,10 +21,10 @@ class ArgToken extends Token {
 		super(undefined, config, true, accum, {AtomToken: 0, Token: 1, HiddenToken: '2:'});
 		for (let i = 0; i < parts.length; i++) {
 			if (i === 0 || i > 1) {
-				const AtomToken = i === 0 ? require('./atom') : require('./atom/hidden'),
-					token = new AtomToken(parts[i], i === 0 ? 'arg-name' : undefined, config, accum, {
-						'Stage-2': ':', '!HeadingToken': '',
-					});
+				const AtomToken = i === 0 ? require('./atom') : require('./atom/hidden');
+				const token = new AtomToken(parts[i], i === 0 ? 'arg-name' : undefined, config, accum, {
+					'Stage-2': ':', '!HeadingToken': '',
+				});
 				this.appendChild(token);
 			} else {
 				const token = new Token(parts[i], config, true, accum);
@@ -39,7 +40,7 @@ class ArgToken extends Token {
 		const [name, ...cloned] = this.cloneChildNodes();
 		return Parser.run(() => {
 			const token = new ArgToken([''], this.getAttribute('config'));
-			token.firstElementChild.safeReplaceWith(name);
+			token.firstChild.safeReplaceWith(name);
 			token.append(...cloned);
 			return token.afterBuild();
 		});
@@ -47,9 +48,9 @@ class ArgToken extends Token {
 
 	/** @override */
 	afterBuild() {
-		this.setAttribute('name', this.firstElementChild.text().trim());
+		this.setAttribute('name', this.firstChild.text().trim());
 		const /** @type {AstListener} */ argListener = ({prevTarget}) => {
-			if (prevTarget === this.firstElementChild) {
+			if (prevTarget === this.firstChild) {
 				this.setAttribute('name', prevTarget.text().trim());
 			}
 		};
@@ -75,12 +76,17 @@ class ArgToken extends Token {
 		return 1;
 	}
 
+	/** @override */
+	print() {
+		return super.print({pre: '{{{', post: '}}}', sep: '|'});
+	}
+
 	/**
 	 * @override
 	 * @returns {string}
 	 */
 	text() {
-		return `{{{${text(this.children.slice(0, 2), '|')}}}}`;
+		return `{{{${text(this.childNodes.slice(0, 2), '|')}}}}`;
 	}
 
 	/**
@@ -95,6 +101,24 @@ class ArgToken extends Token {
 		});
 	}
 
+	/**
+	 * @override
+	 * @param {number} start 起始位置
+	 * @returns {LintError[]}
+	 */
+	lint(start = 0) {
+		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 = this.getRootNode().posFromIndex(start);
+			errors.push(...rest.map(child => generateForChild(child, rect, '三重括号内的不可见部分')));
+		}
+		return errors;
+	}
+
 	/**
 	 * 移除子节点，且在移除`arg-default`子节点时自动移除全部多余子节点
 	 * @param {number} i 移除位置
@@ -133,13 +157,13 @@ class ArgToken extends Token {
 	setName(name) {
 		name = String(name);
 		const root = Parser.parse(`{{{${name}}}}`, this.getAttribute('include'), 2, this.getAttribute('config')),
-			{childNodes: {length}, firstElementChild} = root;
-		if (length !== 1 || firstElementChild?.type !== 'arg' || firstElementChild.childNodes.length !== 1) {
+			{childNodes: {length}, firstChild: arg} = root;
+		if (length !== 1 || arg.type !== 'arg' || arg.childNodes.length !== 1) {
 			throw new SyntaxError(`非法的参数名称：${noWrap(name)}`);
 		}
-		const {firstElementChild: newName} = firstElementChild;
-		firstElementChild.destroy(true);
-		this.firstElementChild.safeReplaceWith(newName);
+		const {firstChild} = arg;
+		arg.destroy(true);
+		this.firstChild.safeReplaceWith(firstChild);
 	}
 
 	/**
@@ -150,17 +174,17 @@ class ArgToken extends Token {
 	setDefault(value) {
 		value = String(value);
 		const root = Parser.parse(`{{{|${value}}}}`, this.getAttribute('include'), 2, this.getAttribute('config')),
-			{childNodes: {length}, firstElementChild} = root;
-		if (length !== 1 || firstElementChild?.type !== 'arg' || firstElementChild.childNodes.length !== 2) {
+			{childNodes: {length}, firstChild: arg} = root;
+		if (length !== 1 || arg.type !== 'arg' || arg.childNodes.length !== 2) {
 			throw new SyntaxError(`非法的参数预设值：${noWrap(value)}`);
 		}
-		const {children: [, oldDefault]} = this,
-			{lastElementChild} = firstElementChild;
-		firstElementChild.destroy(true);
+		const {childNodes: [, oldDefault]} = this,
+			{lastChild} = arg;
+		arg.destroy(true);
 		if (oldDefault) {
-			oldDefault.safeReplaceWith(lastElementChild);
+			oldDefault.safeReplaceWith(lastChild);
 		} else {
-			this.appendChild(lastElementChild);
+			this.appendChild(lastChild);
 		}
 	}
 }

package/src/attribute.js

@@ -1,6 +1,7 @@
 'use strict';
 
 const {externalUse} = require('../util/debug'),
+	{generateForSelf} = require('../util/lint'),
 	{toCase, removeComment, normalizeSpace} = require('../util/string'),
 	Parser = require('..'),
 	Token = require('.');
@@ -14,6 +15,7 @@ const stages = {'ext-attr': 0, 'html-attr': 2, 'table-attr': 3};
 class AttributeToken extends Token {
 	/** @type {Map<string, string|true>} */ #attr = new Map();
 	#sanitized = true;
+	#quoteBalance = true;
 
 	/**
 	 * @override
@@ -88,11 +90,10 @@ class AttributeToken extends Token {
 	 */
 	#updateFromAttr() {
 		let equal = '=';
-		const ParameterToken = require('./parameter');
-		const parent = this.closest('ext, parameter');
-		if (parent instanceof ParameterToken && parent.anon
-			&& parent.parentNode?.matches('template, magic-word#invoke')
-		) {
+		const ParameterToken = require('./parameter'),
+			TranscludeToken = require('./transclude');
+		const /** @type {ParameterToken & {parentNode: TranscludeToken}} */ parent = this.closest('ext, parameter');
+		if (parent instanceof ParameterToken && parent.anon && parent.parentNode?.isTemplate()) {
 			equal = '{{=}}';
 		}
 		return [...this.#attr].map(([k, v]) => {
@@ -117,6 +118,7 @@ class AttributeToken extends Token {
 			this.replaceChildren(...token.childNodes, true);
 		});
 		this.#sanitized = true;
+		this.#quoteBalance = true;
 	}
 
 	/**
@@ -146,11 +148,13 @@ class AttributeToken extends Token {
 		 */
 		const build = str =>
 			typeof str === 'boolean' || !token ? str : token.getAttribute('buildFromStr')(str).map(String).join('');
-		for (const [, key,, quoted, unquoted] of string
-			.matchAll(/([^\s/][^\s/=]*)(?:\s*=\s*(?:(["'])(.*?)(?:\2|$)|(\S*)))?/gsu)
+		for (const [, key, quoteStart, quoted, quoteEnd, unquoted] of string
+			.matchAll(/([^\s/][^\s/=]*)(?:\s*=\s*(?:(["'])(.*?)(\2|$)|(\S*)))?/gsu)
 		) {
 			if (!this.setAttr(build(key), build(quoted ?? unquoted ?? true), true)) {
 				this.#sanitized = false;
+			} else if (quoteStart !== quoteEnd) {
+				this.#quoteBalance = false;
 			}
 		}
 	}
@@ -355,6 +359,29 @@ class AttributeToken extends Token {
 		return this.#leadingSpace().length;
 	}
 
+	/**
+	 * @override
+	 * @this {AttributeToken & {parentNode: HtmlToken}}
+	 * @param {number} start 起始位置
+	 */
+	lint(start = 0) {
+		const HtmlToken = require('./html');
+		const errors = super.lint(start);
+		let /** @type {{top: number, left: number}} */ rect;
+		if (this.type === 'html-attr' && this.parentNode.closing && this.text().trim()) {
+			rect = this.getRootNode().posFromIndex(start);
+			errors.push(generateForSelf(this, rect, '位于闭合标签的属性'));
+		}
+		if (!this.#sanitized) {
+			rect ||= this.getRootNode().posFromIndex(start);
+			errors.push(generateForSelf(this, rect, '包含无效属性'));
+		} else if (!this.#quoteBalance) {
+			rect ||= this.getRootNode().posFromIndex(start);
+			errors.push(generateForSelf(this, rect, '未闭合的引号', 'warning'));
+		}
+		return errors;
+	}
+
 	/** @override */
 	text() {
 		if (this.type === 'table-attr') {

package/src/converter.js

@@ -1,6 +1,6 @@
 'use strict';
 
-const {text} = require('../util/string'),
+const {text, print} = require('../util/string'),
 	Parser = require('..'),
 	Token = require('.'),
 	ConverterFlagsToken = require('./converterFlags'),
@@ -47,7 +47,7 @@ class ConverterToken extends Token {
 	cloneNode() {
 		const [flags, ...rules] = this.cloneChildNodes(),
 			token = Parser.run(() => new ConverterToken([], [], this.getAttribute('config')));
-		token.firstElementChild.safeReplaceWith(flags);
+		token.firstChild.safeReplaceWith(flags);
 		token.append(...rules);
 		return token;
 	}
@@ -57,7 +57,7 @@ class ConverterToken extends Token {
 	 * @param {string} selector
 	 */
 	toString(selector) {
-		const {children: [flags, ...rules]} = this;
+		const {childNodes: [flags, ...rules]} = this;
 		return selector && this.matches(selector)
 			? ''
 			: `-{${flags.toString(selector)}${flags.childNodes.length > 0 ? '|' : ''}${rules.map(String).join(';')}}-`;
@@ -74,12 +74,20 @@ class ConverterToken extends Token {
 	 */
 	getGaps(i = 0) {
 		i = i < 0 ? i + this.childNodes.length : i;
-		return i || this.firstElementChild.childNodes.length > 0 ? 1 : 0;
+		return i || this.firstChild.childNodes.length > 0 ? 1 : 0;
 	}
 
 	/** @override */
-	text() {
+	print() {
 		const {children: [flags, ...rules]} = this;
+		return `<span class="wpb-converter">-{${flags.print()}${
+			flags.childNodes.length > 0 ? '|' : ''
+		}${print(rules, {sep: ';'})}}-</span>`;
+	}
+
+	/** @override */
+	text() {
+		const {childNodes: [flags, ...rules]} = this;
 		return `-{${flags.text()}|${text(rules, ';')}}-`;
 	}
 

package/src/converterFlags.js

@@ -1,6 +1,7 @@
 'use strict';
 
-const Parser = require('..'),
+const {generateForChild} = require('../util/lint'),
+	Parser = require('..'),
 	Token = require('.'),
 	AtomToken = require('./atom');
 
@@ -35,7 +36,7 @@ class ConverterFlagsToken extends Token {
 	 * @complexity `n`
 	 */
 	afterBuild() {
-		this.#flags = this.children.map(child => child.text().trim());
+		this.#flags = this.childNodes.map(child => child.text().trim());
 		const /** @type {AstListener} */ converterFlagsListener = ({prevTarget}) => {
 			if (prevTarget) {
 				this.#flags[this.childNodes.indexOf(prevTarget)] = prevTarget.text().trim();
@@ -94,6 +95,37 @@ class ConverterFlagsToken extends Token {
 		return 1;
 	}
 
+	/** @override */
+	print() {
+		return super.print({sep: ';'});
+	}
+
+	/**
+	 * @override
+	 * @param {number} start 起始位置
+	 */
+	lint(start = 0) {
+		const variantFlags = this.getVariantFlags(),
+			unknownFlags = this.getUnknownFlags(),
+			emptyFlags = this.#flags.filter(flag => !flag),
+			validFlags = this.#flags.filter(flag => ['A', 'T', 'R', 'D', '-', 'H', 'N'].includes(flag)),
+			knownFlagCount = this.#flags.length - unknownFlags.length - emptyFlags,
+			errors = super.lint(start);
+		if (variantFlags.length === knownFlagCount || validFlags.length === knownFlagCount) {
+			return errors;
+		}
+		const rect = this.getRootNode().posFromIndex(start);
+		for (const child of this.childNodes) {
+			const flag = child.text().trim();
+			if (flag && !variantFlags.includes(flag) && !unknownFlags.includes(flag)
+				&& (variantFlags.length > 0 || !validFlags.includes(flag))
+			) {
+				errors.push(generateForChild(child, rect, '无效的转换标记'));
+			}
+		}
+		return errors;
+	}
+
 	/** @override */
 	text() {
 		return super.text(';');
@@ -106,7 +138,7 @@ class ConverterFlagsToken extends Token {
 	 * @complexity `n`
 	 */
 	getFlagToken(flag) {
-		return this.#flags.includes(flag) ? this.children.filter(child => child.text().trim() === flag) : [];
+		return this.#flags.includes(flag) ? this.childNodes.filter(child => child.text().trim() === flag) : [];
 	}
 
 	/** 获取所有转换类型标记 */
@@ -122,13 +154,18 @@ class ConverterFlagsToken extends Token {
 		return this.#flags.filter(flag => /\{\{[^{}]+\}\}/u.test(flag));
 	}
 
+	/** 获取指定语言变体的转换标记 */
+	getVariantFlags() {
+		const {variants} = this.getAttribute('config');
+		return this.#flags.filter(flag => variants.includes(flag));
+	}
+
 	/**
 	 * 获取有效转换类型标记
 	 * @complexity `n`
 	 */
 	getEffectiveFlags() {
-		const {variants} = this.getAttribute('config'),
-			variantFlags = this.#flags.filter(flag => variants.includes(flag)),
+		const variantFlags = this.getVariantFlags(),
 			unknownFlags = this.getUnknownFlags();
 		if (variantFlags.length > 0) {
 			return new Set([...variantFlags, ...unknownFlags]);