wikiparser-node 0.0.2 → 0.2.0

package/src/converterFlags.js

@@ -0,0 +1,214 @@
+'use strict';
+
+const /** @type {Parser} */ Parser = require('..'),
+	Token = require('.'),
+	AtomToken = require('./atom');
+
+/**
+ * 转换flags
+ * @classdesc `{childNodes: ...AtomToken}`
+ */
+class ConverterFlagsToken extends Token {
+	type = 'converter-flags';
+	/** @type {string[]} */ #flags;
+
+	/**
+	 * @param {string[]} flags
+	 * @param {accum} accum
+	 */
+	constructor(flags, config = Parser.getConfig(), accum = []) {
+		super(undefined, config, false, accum, {AtomToken: ':'});
+		this.append(...flags.map(flag => new AtomToken(flag, 'converter-flag', config, accum)));
+	}
+
+	cloneNode() {
+		const cloned = this.cloneChildren(),
+			token = Parser.run(() => new ConverterFlagsToken([], this.getAttribute('config')));
+		token.append(...cloned);
+		token.afterBuild();
+		return token;
+	}
+
+	/** @complexity `n` */
+	afterBuild() {
+		this.#flags = this.children.map(child => child.text().trim());
+		const that = this,
+			/** @type {AstListener} */ converterFlagsListener = ({prevTarget}) => {
+				if (prevTarget) {
+					that.#flags[that.childNodes.indexOf(prevTarget)] = prevTarget.text().trim();
+				}
+			};
+		this.addEventListener(['remove', 'insert', 'text', 'replace'], converterFlagsListener);
+	}
+
+	/**
+	 * @template {string} T
+	 * @param {T} key
+	 * @returns {TokenAttribute<T>}
+	 */
+	getAttribute(key) {
+		if (key === 'flags') {
+			return Parser.debugging ? this.#flags : [...this.#flags];
+		}
+		return super.getAttribute(key);
+	}
+
+	/**
+	 * @param {number} i
+	 * @complexity `n`
+	 */
+	removeAt(i) {
+		const /** @type {AtomToken} */ token = super.removeAt(i);
+		this.#flags?.splice(i, 1);
+		return token;
+	}
+
+	/**
+	 * @param {AtomToken} token
+	 * @complexity `n`
+	 */
+	insertAt(token, i = this.childNodes.length) {
+		super.insertAt(token, i);
+		this.#flags?.splice(i, 0, token.text());
+		return token;
+	}
+
+	toString() {
+		return super.toString(';');
+	}
+
+	getGaps() {
+		return 1;
+	}
+
+	text() {
+		return super.text(';');
+	}
+
+	/** @returns {[number, string][]} */
+	plain() {
+		return [];
+	}
+
+	/**
+	 * @param {string} flag
+	 * @returns {AtomToken[]}
+	 * @complexity `n`
+	 */
+	getFlagToken(flag) {
+		return this.#flags.includes(flag) ? this.children.filter(child => child.text().trim() === flag) : [];
+	}
+
+	getAllFlags() {
+		return new Set(this.#flags);
+	}
+
+	/** @complexity `n` */
+	getEffectiveFlags() {
+		const {variants} = this.getAttribute('config'),
+			variantFlags = this.#flags.filter(flag => variants.includes(flag)),
+			unknownFlags = this.#flags.filter(flag => /{{.+}}/.test(flag));
+		if (variantFlags.length) {
+			return new Set([...variantFlags, ...unknownFlags]);
+		}
+		const validFlags = ['A', 'T', 'R', 'D', '-', 'H', 'N'],
+			flags = new Set([...this.#flags.filter(flag => validFlags.includes(flag)), ...unknownFlags]);
+		if (flags.size === 0) {
+			return new Set('S');
+		} else if (flags.has('R')) {
+			return new Set('R');
+		} else if (flags.has('N')) {
+			return new Set('N');
+		} else if (flags.has('-')) {
+			return new Set('-');
+		} else if (flags.has('H')) {
+			const hasT = flags.has('T'),
+				hasD = flags.has('D');
+			if (hasT && hasD) {
+				return new Set(['+', 'H', 'T', 'D']);
+			}
+			return new Set(['+', 'H', ...hasT ? ['T'] : [], ...hasD ? ['D'] : [], ...unknownFlags]);
+		}
+		if (flags.size === 1 && flags.has('T')) {
+			flags.add('H');
+		}
+		if (flags.has('A')) {
+			flags.add('+');
+			flags.add('S');
+		}
+		if (flags.has('D')) {
+			flags.delete('S');
+		}
+		return flags;
+	}
+
+	/** @complexity `n` */
+	getUnknownFlags() {
+		return [...this.getFlags()].filter(flag => /{{.+}}/.test(flag));
+	}
+
+	/** @param {string} flag */
+	hasFlag(flag) {
+		if (typeof flag !== 'string') {
+			this.typeError('hasFlag', 'String');
+		}
+		return this.#flags.includes(flag);
+	}
+
+	/**
+	 * @param {string} flag
+	 * @complexity `n`
+	 */
+	hasEffectiveFlag(flag) {
+		if (typeof flag !== 'string') {
+			this.typeError('hasFlag', 'String');
+		}
+		return this.getEffectiveFlags().has(flag);
+	}
+
+	/**
+	 * @param {string} flag
+	 * @complexity `n²`
+	 */
+	removeFlag(flag) {
+		for (const token of this.getFlagToken(flag)) {
+			token.remove();
+		}
+	}
+
+	/**
+	 * @param {string} flag
+	 * @complexity `n`
+	 */
+	#newFlag(flag) {
+		const token = Parser.run(() => new AtomToken(flag, 'converter-flag', this.getAttribute('config')));
+		this.appendChild(token);
+	}
+
+	/**
+	 * @param {string} flag
+	 * @complexity `n`
+	 */
+	setFlag(flag) {
+		if (!this.#flags.includes(flag)) {
+			this.#newFlag(flag);
+		} else if (!this.getEffectiveFlags().has(flag)) {
+			Parser.error('此 flag 不会生效', flag);
+		}
+	}
+
+	/**
+	 * @param {string} flag
+	 * @complexity `n²`
+	 */
+	toggleFlag(flag) {
+		if (!this.#flags.includes(flag)) {
+			this.#newFlag(flag);
+		} else {
+			this.removeFlag(flag);
+		}
+	}
+}
+
+Parser.classes.ConverterFlagsToken = __filename;
+module.exports = ConverterFlagsToken;

package/src/converterRule.js

@@ -0,0 +1,209 @@
+'use strict';
+
+const {undo} = require('../util/debug'),
+	{noWrap} = require('../util/string'),
+	/** @type {Parser} */ Parser = require('..'),
+	Token = require('.'),
+	AtomToken = require('./atom');
+
+/**
+ * 转换规则
+ * @classdesc `{childNodes: ...AtomToken)}`
+ */
+class ConverterRuleToken extends Token {
+	type = 'converter-rule';
+	variant = '';
+	unidirectional = false;
+	bidirectional = false;
+
+	/**
+	 * @param {string} rule
+	 * @param {accum} accum
+	 */
+	constructor(rule, hasColon = true, config = Parser.getConfig(), accum = []) {
+		super(undefined, config, true, accum, {AtomToken: ':'});
+		if (!hasColon) {
+			super.insertAt(new AtomToken(rule, 'converter-rule-noconvert', config, accum));
+		} else {
+			const i = rule.indexOf(':'),
+				j = rule.slice(0, i).indexOf('=>'),
+				v = (j === -1 ? rule.slice(0, i) : rule.slice(j + 2, i)).trim(),
+				{variants} = config;
+			if (!variants.includes(v)) {
+				super.insertAt(new AtomToken(rule, 'converter-rule-noconvert', config, accum));
+			} else {
+				super.insertAt(new AtomToken(v, 'converter-rule-variant', config, accum));
+				super.insertAt(new AtomToken(rule.slice(i + 1), 'converter-rule-to', config, accum));
+				if (j === -1) {
+					this.bidirectional = true;
+				} else {
+					super.insertAt(new AtomToken(rule.slice(0, j), 'converter-rule-from', config, accum), 0);
+					this.unidirectional = true;
+				}
+			}
+		}
+		this.seal(['variant', 'unidirectional', 'bidirectional']);
+	}
+
+	cloneNode() {
+		const cloned = this.cloneChildren(),
+			placeholders = ['', ':', '=>:'],
+			placeholder = placeholders[cloned.length - 1],
+			token = Parser.run(() => new ConverterRuleToken(placeholder, placeholder, this.getAttribute('config')));
+		for (let i = 0; i < cloned.length; i++) {
+			token.children[i].safeReplaceWith(cloned[i]);
+		}
+		token.afterBuild();
+		return token;
+	}
+
+	afterBuild() {
+		if (this.childNodes.length > 1) {
+			this.setAttribute('variant', this.children.at(-2).text().trim());
+		}
+		const that = this,
+			/** @type {AstListener} */ converterRuleListener = (e, data) => {
+				const {childNodes} = that,
+					{prevTarget} = e;
+				if (childNodes.length > 1 && childNodes.at(-2) === prevTarget) {
+					const v = prevTarget.text().trim(),
+						{variants} = that.getAttribute('config');
+					if (variants.includes(v)) {
+						that.setAttribute('variant', v);
+					} else {
+						undo(e, data);
+						throw new Error(`无效的语言变体：${v}`);
+					}
+				}
+			};
+		this.addEventListener(['remove', 'insert', 'text', 'replace'], converterRuleListener);
+	}
+
+	/**
+	 * @param {number} i
+	 * @returns {AtomToken}
+	 */
+	removeAt(i) {
+		if (i !== 0 && i !== -this.childNodes.length) {
+			throw new RangeError(`${this.constructor.name} 禁止移除第 ${i} 个子节点！`);
+		}
+		return super.removeAt(i);
+	}
+
+	insertAt() {
+		throw new Error('转换规则语法复杂，请勿尝试手动插入子节点！');
+	}
+
+	/** @returns {string} */
+	toString() {
+		if (this.childNodes.length === 3) {
+			const [from, variant, to] = this.children;
+			return `${from.toString()}=>${variant.toString()}:${to.toString()}`;
+		}
+		return super.toString(':');
+	}
+
+	/** @param {number} i */
+	getGaps(i = 0) {
+		const {length} = this.childNodes;
+		i = i < 0 ? i + length : i;
+		return i === 0 && length === 3 ? 2 : 1;
+	}
+
+	/** @returns {string} */
+	text() {
+		if (this.childNodes.length === 3) {
+			const [from, variant, to] = this.children;
+			return `${from.text()}=>${variant.text()}:${to.text()}`;
+		}
+		return super.text(':');
+	}
+
+	/** @returns {[number, string][]} */
+	plain() {
+		const {firstElementChild, lastElementChild, unidirectional} = this;
+		if (unidirectional) {
+			return [...firstElementChild.plain(), ...lastElementChild.plain()];
+		}
+		return lastElementChild.plain();
+	}
+
+	noConvert() {
+		for (let i = this.childNodes.length - 2; i >= 0; i--) {
+			super.removeAt(i);
+		}
+		this.setAttribute('unidirectional', false).setAttribute('bidirectional', false).setAttribute('variant', '');
+	}
+
+	/** @param {string} to */
+	setTo(to) {
+		to = String(to);
+		const config = this.getAttribute('config'),
+			root = Parser.parse(`-{|${config.variants[0]}:${to}}-`, this.getAttribute('include'), undefined, config),
+			{childNodes: {length}, firstElementChild} = root;
+		if (length !== 1 || firstElementChild.type !== 'converter' || firstElementChild.childNodes.length !== 2
+			|| firstElementChild.lastElementChild.childNodes.length !== 2
+		) {
+			throw new SyntaxError(`非法的转换目标：${noWrap(to)}`);
+		}
+		const {lastChild} = firstElementChild.lastElementChild;
+		firstElementChild.lastElementChild.removeAt(0);
+		this.lastElementChild.safeReplaceWith(lastChild);
+	}
+
+	/** @param {string} variant */
+	setVariant(variant) {
+		if (typeof variant !== 'string') {
+			this.typeError('setVariant', 'String');
+		}
+		const config = this.getAttribute('config'),
+			v = variant.trim();
+		if (!config.variants.includes(v)) {
+			throw new RangeError(`无效的语言变体：${v}`);
+		} else if (this.childNodes.length === 1) {
+			super.insertAt(Parser.run(() => new AtomToken(variant, 'converter-rule-variant', config)), 0);
+			this.setAttribute('bidirectional', true);
+		} else {
+			this.children.at(-2).setText(variant);
+		}
+		this.setAttribute('variant', v);
+	}
+
+	/** @param {string} from */
+	setFrom(from) {
+		const {variant} = this;
+		if (!variant) {
+			throw new Error('请先指定语言变体！');
+		}
+		from = String(from);
+		const config = this.getAttribute('config'),
+			root = Parser.parse(`-{|${from}=>${variant}:}-`, this.getAttribute('include'), undefined, config),
+			{childNodes: {length}, firstElementChild} = root;
+		if (length !== 1 || firstElementChild.type !== 'converter' || firstElementChild.childNodes.length !== 2
+			|| firstElementChild.lastElementChild.childNodes.length !== 3
+		) {
+			throw new SyntaxError(`非法的转换原文：${noWrap(from)}`);
+		}
+		if (this.unidirectional) {
+			this.firstElementChild.safeReplaceWith(firstElementChild.lastElementChild.firstChild);
+		} else {
+			super.insertAt(firstElementChild.lastElementChild.firstChild, 0);
+			this.setAttribute('unidirectional', true).setAttribute('bidirectional', false);
+		}
+	}
+
+	/** @param {string} from */
+	makeUnidirectional(from) {
+		this.setFrom(from);
+	}
+
+	makeBidirectional() {
+		if (this.unidirectional) {
+			super.removeAt(0);
+			this.setAttribute('unidirectional', false).setAttribute('bidirectional', true);
+		}
+	}
+}
+
+Parser.classes.ConverterRuleToken = __filename;
+module.exports = ConverterRuleToken;

package/src/extLink.js

@@ -1,6 +1,7 @@
 'use strict';
 
-const /** @type {Parser} */ Parser = require('..'),
+const {noWrap, normalizeSpace} = require('../util/string'),
+	/** @type {Parser} */ Parser = require('..'),
 	Token = require('.'),
 	MagicLinkToken = require('./magicLink');
 
@@ -32,9 +33,9 @@ class ExtLinkToken extends Token {
 		this.appendChild(new MagicLinkToken(url, true, config, accum));
 		this.#space = space;
 		if (text) {
-			const inner = new Token(text, config, true, accum);
+			const inner = new Token(text, config, true, accum, {'Stage-7': ':', ConverterToken: ':'});
 			inner.type = 'ext-link-text';
-			this.appendChild(inner.setAttribute('stage', 8));
+			this.appendChild(inner.setAttribute('stage', Parser.MAX_STAGE - 1));
 		}
 		this.protectChildren(0);
 	}
@@ -48,20 +49,32 @@ class ExtLinkToken extends Token {
 		}
 	}
 
+	#correct() {
+		if (!this.#space && this.childNodes.length > 1
+			// 都替换成`<`肯定不对，但无妨
+			&& /^[^[\]<>"{\x00-\x20\x7f\p{Zs}\ufffd]/u.test(this.lastElementChild.text().replace(/&[lg]t;/, '<'))
+		) {
+			this.#space = ' ';
+		}
+	}
+
 	toString() {
-		return `[${this.firstElementChild.toString()}${this.#space}${this.children[1]?.toString() ?? ''}]`;
+		this.#correct();
+		return `[${this.firstElementChild.toString()}${this.#space}${normalizeSpace(this.children[1])}]`;
 	}
 
 	getPadding() {
+		this.#correct();
 		return 1;
 	}
 
 	getGaps() {
+		this.#correct();
 		return this.#space.length;
 	}
 
 	text() {
-		return `[${super.text(' ')}]`;
+		return `[${super.text(' ').replaceAll('\n', ' ')}]`;
 	}
 
 	/**
@@ -69,7 +82,7 @@ class ExtLinkToken extends Token {
 	 * @complexity `n`
 	 */
 	plain() {
-		return this.childElementCount === 1 ? [] : this.lastElementChild.plain();
+		return this.childNodes.length === 1 ? [] : this.lastElementChild.plain();
 	}
 
 	/** @this {ExtLinkToken & {firstElementChild: MagicLinkToken}} */
@@ -82,7 +95,7 @@ class ExtLinkToken extends Token {
 		url = String(url);
 		const root = Parser.parse(`[${url}]`, this.getAttribute('include'), 8, this.getAttribute('config')),
 			{childNodes: {length}, firstElementChild} = root;
-		if (length !== 1 || firstElementChild?.type !== 'ext-link' || firstElementChild.childElementCount !== 1) {
+		if (length !== 1 || firstElementChild?.type !== 'ext-link' || firstElementChild.childNodes.length !== 1) {
 			throw new SyntaxError(`非法的外链目标：${url}`);
 		}
 		const {firstChild} = firstElementChild;
@@ -96,11 +109,11 @@ class ExtLinkToken extends Token {
 		text = String(text);
 		const root = Parser.parse(`[//url ${text}]`, this.getAttribute('include'), 8, this.getAttribute('config')),
 			{childNodes: {length}, firstElementChild} = root;
-		if (length !== 1 || firstElementChild?.type !== 'ext-link' || firstElementChild.childElementCount !== 2) {
-			throw new SyntaxError(`非法的外链文字：${text.replaceAll('\n', '\\n')}`);
+		if (length !== 1 || firstElementChild?.type !== 'ext-link' || firstElementChild.childNodes.length !== 2) {
+			throw new SyntaxError(`非法的外链文字：${noWrap(text)}`);
 		}
 		const {lastChild} = firstElementChild;
-		if (this.childElementCount === 1) {
+		if (this.childNodes.length === 1) {
 			this.appendChild(lastChild);
 		} else {
 			this.lastElementChild.replaceWith(lastChild);

package/src/heading.js

@@ -1,7 +1,7 @@
 'use strict';
 
-const {typeError} = require('../util/debug'),
-	fixedToken = require('../mixin/fixedToken'),
+const fixedToken = require('../mixin/fixedToken'),
+	sol = require('../mixin/sol'),
 	/** @type {Parser} */ Parser = require('..'),
 	Token = require('.');
 
@@ -9,7 +9,7 @@ const {typeError} = require('../util/debug'),
  * 章节标题
  * @classdesc `{childNodes: [Token, HiddenToken]}`
  */
-class HeadingToken extends fixedToken(Token) {
+class HeadingToken extends fixedToken(sol(Token)) {
 	type = 'heading';
 
 	/**
@@ -38,34 +38,29 @@ class HeadingToken extends fixedToken(Token) {
 		return token;
 	}
 
+	/** @this {HeadingToken & {prependNewLine(): ''|'\n', appendNewLine(): ''|'\n'}} */
 	toString() {
-		const equals = '='.repeat(Number(this.name)),
-			{previousVisibleSibling, nextVisibleSibling} = this;
-		return `${
-			typeof previousVisibleSibling === 'string' && !previousVisibleSibling.endsWith('\n')
-			|| previousVisibleSibling instanceof Token
-				? '\n'
-				: ''
-		}${equals}${super.toString(equals)}${
-			typeof nextVisibleSibling === 'string' && !nextVisibleSibling.startsWith('\n')
-			|| nextVisibleSibling instanceof Token
-				? '\n'
-				: ''
-		}`;
+		const equals = '='.repeat(Number(this.name));
+		return `${this.prependNewLine()}${equals}${
+			this.firstElementChild.toString()
+		}${equals}${this.lastElementChild.toString()}${this.appendNewLine()}`;
 	}
 
 	getPadding() {
-		return Number(this.name);
+		return super.getPadding() + Number(this.name);
 	}
 
 	getGaps() {
 		return Number(this.name);
 	}
 
-	/** @returns {string} */
+	/**
+	 * @this {HeadingToken & {prependNewLine(): ''|'\n', appendNewLine(): ''|'\n'}}
+	 * @returns {string}
+	 */
 	text() {
 		const equals = '='.repeat(Number(this.name));
-		return `${equals}${this.firstElementChild.text()}${equals}`;
+		return `${this.prependNewLine()}${equals}${this.firstElementChild.text()}${equals}${this.appendNewLine()}`;
 	}
 
 	/** @returns {[number, string][]} */
@@ -76,7 +71,7 @@ class HeadingToken extends fixedToken(Token) {
 	/** @param {number} n */
 	setLevel(n) {
 		if (typeof n !== 'number') {
-			typeError(this, 'setLevel', 'Number');
+			this.typeError('setLevel', 'Number');
 		}
 		n = Math.min(Math.max(n, 1), 6);
 		this.setAttribute('name', String(n)).firstElementChild.setAttribute('name', this.name);

package/src/html.js

@@ -1,6 +1,7 @@
 'use strict';
 
-const fixedToken = require('../mixin/fixedToken'),
+const {noWrap} = require('../util/string'),
+	fixedToken = require('../mixin/fixedToken'),
 	attributeParent = require('../mixin/attributeParent'),
 	/** @type {Parser} */ Parser = require('..'),
 	Token = require('.');
@@ -79,7 +80,7 @@ class HtmlToken extends attributeParent(fixedToken(Token)) {
 	findMatchingTag() {
 		const {html} = this.getAttribute('config'),
 			{name, parentElement, closing, selfClosing} = this,
-			string = this.toString().replaceAll('\n', '\\n');
+			string = noWrap(this.toString());
 		if (closing && selfClosing) {
 			throw new SyntaxError(`同时闭合和自封闭的标签：${string}`);
 		} else if (html[2].includes(name) || selfClosing && html[1].includes(name)) { // 自封闭标签
@@ -134,7 +135,7 @@ class HtmlToken extends attributeParent(fixedToken(Token)) {
 			this.selfClosing = false;
 			this.closing = true;
 		} else {
-			Parser.warn('无法修复无效自封闭标签', this.toString().replaceAll('\n', '\\n'));
+			Parser.warn('无法修复无效自封闭标签', noWrap(this.toString()));
 			throw new Error(`无法修复无效自封闭标签：前文共有 ${imbalance} 个未匹配的闭合标签`);
 		}
 	}

package/src/imageParameter.js

@@ -1,7 +1,6 @@
 'use strict';
 
-const {typeError} = require('../util/debug'),
-	{text, extUrlChar} = require('../util/string'),
+const {text, noWrap, extUrlChar} = require('../util/string'),
 	Title = require('../lib/title'),
 	/** @type {Parser} */ Parser = require('..'),
 	Token = require('.');
@@ -32,7 +31,7 @@ class ImageParameterToken extends Token {
 			if (!value) {
 				return this.#noLink;
 			}
-			const regex = new RegExp(`(?:${config.protocol}|//)${extUrlChar}`, 'ui');
+			const regex = new RegExp(`(?:${config.protocol}|//)${extUrlChar}(?=\x00\\d+t\x7f|$)`, 'ui');
 			if (regex.test(value)) {
 				return value;
 			}
@@ -200,13 +199,13 @@ class ImageParameterToken extends Token {
 	setValue(value) {
 		if (this.#isVoid()) {
 			if (typeof value !== 'boolean') {
-				typeError(this, 'setValue', 'Boolean');
+				this.typeError('setValue', 'Boolean');
 			} else if (value === false) {
 				this.remove();
 			}
 			return;
 		} else if (typeof value !== 'string') {
-			typeError(this, 'setValue', 'String');
+			this.typeError('setValue', 'String');
 		}
 		const root = Parser.parse(`[[File:F|${
 				this.#syntax ? this.#syntax.replace('$1', value) : value
@@ -214,9 +213,9 @@ class ImageParameterToken extends Token {
 			{childNodes: {length}, firstElementChild} = root,
 			param = firstElementChild?.lastElementChild;
 		if (length !== 1 || !firstElementChild?.matches('file#File:F')
-			|| firstElementChild.childElementCount !== 2 || param.name !== this.name
+			|| firstElementChild.childNodes.length !== 2 || param.name !== this.name
 		) {
-			throw new SyntaxError(`非法的 ${this.name} 参数：${value.replaceAll('\n', '\\n')}`);
+			throw new SyntaxError(`非法的 ${this.name} 参数：${noWrap(value)}`);
 		}
 		this.replaceChildren(...param.childNodes);
 	}