wikiparser-node 0.0.3 → 0.1.0

package/index.js

@@ -76,6 +76,7 @@ const /** @type {Parser} */ Parser = {
 		['QuoteToken'],
 		['ExtLinkToken'],
 		['MagicLinkToken'],
+		['ListToken', 'DdToken'],
 	],
 
 	config: './config/default',

package/mixin/sol.js

@@ -0,0 +1,38 @@
+'use strict';
+
+const /** @type {Parser} */ Parser = require('..'),
+	Token = require('../src'); // eslint-disable-line no-unused-vars
+
+/**
+ * @template T
+ * @param {T} constructor
+ * @returns {T}
+ */
+const sol = constructor => class extends constructor {
+	/** @this {Token} */
+	prependNewLine() {
+		const {previousVisibleSibling} = this;
+		return previousVisibleSibling && !String(previousVisibleSibling).endsWith('\n') ? '\n' : '';
+	}
+
+	/** @this {Token} */
+	appendNewLine() {
+		const {nextVisibleSibling} = this;
+		return nextVisibleSibling && !String(nextVisibleSibling).startsWith('\n') ? '\n' : '';
+	}
+
+	toString(ownLine = false) {
+		return `${this.prependNewLine()}${super.toString()}${ownLine ? this.appendNewLine() : ''}`;
+	}
+
+	getPadding() {
+		return this.prependNewLine().length;
+	}
+
+	text(ownLine = false) {
+		return `${this.prependNewLine()}${super.text()}${ownLine ? this.appendNewLine() : ''}`;
+	}
+};
+
+Parser.mixins.sol = __filename;
+module.exports = sol;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wikiparser-node",
-  "version": "0.0.3",
+  "version": "0.1.0",
   "description": "A Node.js parser for MediaWiki markup with AST",
   "keywords": [
     "mediawiki",

package/parser/list.js

@@ -0,0 +1,58 @@
+'use strict';
+
+const /** @type {Parser} */ Parser = require('..');
+
+/**
+ * @param {string} text
+ * @param {accum} accum
+ */
+const parseList = (text, config = Parser.getConfig(), accum = []) => {
+	const mt = text.match(/^(?:[;:*#]|\x00\d+c\x7f)*[;:*#]/);
+	if (!mt) {
+		return text;
+	}
+	const ListToken = require('../src/nowiki/list'),
+		[prefix] = mt;
+	text = `\x00${accum.length}d\x7f${text.slice(prefix.length)}`;
+	new ListToken(prefix, config, accum);
+	let dt = prefix.split(';').length - 1;
+	if (!dt) {
+		return text;
+	}
+	const DdToken = require('../src/nowiki/dd');
+	let regex = /:+|-{/g,
+		ex = regex.exec(text),
+		lc = 0;
+	while (ex && dt) {
+		const {0: syntax, index} = ex;
+		if (syntax[0] === ':') {
+			if (syntax.length >= dt) {
+				new DdToken(':'.repeat(dt), config, accum);
+				return `${text.slice(0, index)}\x00${accum.length - 1}d\x7f${text.slice(index + dt)}`;
+			}
+			text = `${text.slice(0, index)}\x00${accum.length}d\x7f${text.slice(regex.lastIndex)}`;
+			dt -= syntax.length;
+			regex.lastIndex = index + 4 + String(accum.length).length;
+			new DdToken(syntax, config, accum);
+		} else if (syntax === '-{') {
+			if (!lc) {
+				const {lastIndex} = regex;
+				regex = /-{|}-/g;
+				regex.lastIndex = lastIndex;
+			}
+			lc++;
+		} else {
+			lc--;
+			if (!lc) {
+				const {lastIndex} = regex;
+				regex = /:+|-{/g;
+				regex.lastIndex = lastIndex;
+			}
+		}
+		ex = regex.exec(text);
+	}
+	return text;
+};
+
+Parser.parsers.parseList = __filename;
+module.exports = parseList;

package/src/attribute.js

@@ -28,14 +28,13 @@ class AttributeToken extends Token {
 		) {
 			equal = '{{=}}';
 		}
-		const str = [...this.#attr].map(([k, v]) => {
+		return [...this.#attr].map(([k, v]) => {
 			if (v === true) {
 				return k;
 			}
 			const quote = v.includes('"') ? "'" : '"';
 			return `${k}${equal}${quote}${v}${quote}`;
 		}).join(' ');
-		return str && ` ${str}`;
 	}
 
 	/** @complexity `n` */
@@ -56,17 +55,17 @@ class AttributeToken extends Token {
 	 */
 	#parseAttr() {
 		this.#attr.clear();
-		const config = this.getAttribute('config'),
-			include = this.getAttribute('include'),
-			/** @type {Token & {firstChild: string}} */ token = this.type !== 'ext-attr' && !Parser.running
-				? Parser.run(() => new Token(string, config).parseOnce(0, include).parseOnce())
-				: undefined,
-			string = removeComment(token?.firstChild ?? this.toString()).replace(/\x00\d+~\x7f/g, '=');
-		const build = /** @param {string|boolean} str */ str => {
-			return typeof str === 'boolean' || !(token instanceof Token)
-				? str
-				: token.buildFromStr(str).map(String).join('');
-		};
+		let string = this.toString(),
+			/** @type {Token & {firstChild: string}} */ token;
+		if (this.type !== 'ext-attr' && !Parser.running) {
+			const config = this.getAttribute('config'),
+				include = this.getAttribute('include');
+			token = Parser.run(() => new Token(string, config).parseOnce(0, include).parseOnce());
+			string = token.firstChild;
+		}
+		string = removeComment(string).replace(/\x00\d+~\x7f/g, '=');
+		const build = /** @param {string|boolean} str */ str =>
+			typeof str === 'boolean' || !token ? str : token.buildFromStr(str).map(String).join('');
 		for (const [, key,, quoted, unquoted] of string
 			.matchAll(/([^\s/][^\s/=]*)(?:\s*=\s*(?:(["'])(.*?)(?:\2|$)|(\S*)))?/sg)
 		) {
@@ -241,13 +240,24 @@ class AttributeToken extends Token {
 		this.setAttr(key, force === true || force === undefined && value === false);
 	}
 
+	#leadingSpace(str = super.toString()) {
+		return str && !/^\s/.test(str) ? ' ' : '';
+	}
+
 	toString() {
-		const str = super.toString();
+		let str = super.toString();
+		str = `${this.#leadingSpace(str)}${str}`;
 		return this.type === 'table-attr' ? str.replaceAll('\n', ' ') : str;
 	}
 
+	getPadding() {
+		return this.#leadingSpace().length;
+	}
+
 	text() {
-		return this.#updateFromAttr();
+		let str = this.#updateFromAttr();
+		str = `${this.#leadingSpace(str)}${str}`;
+		return this.type === 'table-attr' ? str.replaceAll('\n', ' ') : str;
 	}
 
 	/** @returns {[number, string][]} */

package/src/extLink.js

@@ -49,15 +49,27 @@ class ExtLinkToken extends Token {
 		}
 	}
 
+	#correct() {
+		if (!this.#space
+			// 都替换成`<`肯定不对，但无妨
+			&& /^[^[\]<>"\x00-\x20\x7f\p{Zs}\ufffd]/u.test(this.children[1]?.text().replace(/&[lg]t;/, '<'))
+		) {
+			this.#space = ' ';
+		}
+	}
+
 	toString() {
+		this.#correct();
 		return `[${this.firstElementChild.toString()}${this.#space}${this.children[1]?.toString() ?? ''}]`;
 	}
 
 	getPadding() {
+		this.#correct();
 		return 1;
 	}
 
 	getGaps() {
+		this.#correct();
 		return this.#space.length;
 	}
 

package/src/heading.js

@@ -1,6 +1,7 @@
 'use strict';
 
 const fixedToken = require('../mixin/fixedToken'),
+	sol = require('../mixin/sol'),
 	/** @type {Parser} */ Parser = require('..'),
 	Token = require('.');
 
@@ -8,7 +9,7 @@ const fixedToken = require('../mixin/fixedToken'),
  * 章节标题
  * @classdesc `{childNodes: [Token, HiddenToken]}`
  */
-class HeadingToken extends fixedToken(Token) {
+class HeadingToken extends fixedToken(sol(Token)) {
 	type = 'heading';
 
 	/**
@@ -37,34 +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][]} */

package/src/index.js

@@ -37,6 +37,7 @@
  * l: LinkToken
  * q: QuoteToken
  * w: ExtLinkToken
+ * d: ListToken
  */
 
 const {externalUse} = require('../util/debug'),
@@ -436,11 +437,7 @@ class Token extends AstElement {
 				this.#parseLinks();
 				break;
 			case 6: {
-				const lines = this.firstChild.split('\n');
-				for (let i = 0; i < lines.length; i++) {
-					lines[i] = this.#parseQuotes(lines[i]);
-				}
-				this.setText(lines.join('\n'));
+				this.#parseQuotes();
 				break;
 			}
 			case 7:
@@ -450,6 +447,7 @@ class Token extends AstElement {
 				this.#parseMagicLinks();
 				break;
 			case 9:
+				this.#parseList();
 				break;
 			case 10:
 				// no default
@@ -579,10 +577,14 @@ class Token extends AstElement {
 		this.setText(parseLinks(this.firstChild, this.#config, this.#accum));
 	}
 
-	/** @param {string} text */
-	#parseQuotes(text) {
-		const parseQuotes = require('../parser/quotes');
-		return parseQuotes(text, this.#config, this.#accum);
+	/** @this {Token & {firstChild: string}} */
+	#parseQuotes() {
+		const parseQuotes = require('../parser/quotes'),
+			lines = this.firstChild.split('\n');
+		for (let i = 0; i < lines.length; i++) {
+			lines[i] = parseQuotes(lines[i], this.#config, this.#accum);
+		}
+		this.setText(lines.join('\n'));
 	}
 
 	/** @this {Token & {firstChild: string}} */
@@ -596,6 +598,15 @@ class Token extends AstElement {
 		const parseMagicLinks = require('../parser/magicLinks');
 		this.setText(parseMagicLinks(this.firstChild, this.#config, this.#accum));
 	}
+
+	#parseList() {
+		const parseList = require('../parser/list'),
+			lines = this.firstChild.split('\n');
+		for (let i = 0; i < lines.length; i++) {
+			lines[i] = parseList(lines[i], this.#config, this.#accum);
+		}
+		this.setText(lines.join('\n'));
+	}
 }
 
 Parser.classes.Token = __filename;

package/src/nowiki/dd.js

@@ -0,0 +1,49 @@
+'use strict';
+
+const /** @type {Parser} */ Parser = require('../..'),
+	NowikiToken = require('.');
+
+/**
+ * :
+ * @classdesc `{childNodes: [string]}`
+ */
+class DdToken extends NowikiToken {
+	type = 'dd';
+	dt;
+	ul;
+	ol;
+	indent;
+
+	/** @param {string} str */
+	#update(str) {
+		this.setAttribute('ul', str.includes('*')).setAttribute('ol', str.includes('#'))
+			.setAttribute('dt', str.includes(';')).setAttribute('indent', str.split(':').length - 1);
+	}
+
+	/**
+	 * @param {string} str
+	 * @param {accum} accum
+	 */
+	constructor(str, config = Parser.getConfig(), accum = []) {
+		super(str, config, accum);
+		this.seal(['dt', 'ul', 'ol', 'indent']).#update(str);
+	}
+
+	/** @returns {[number, string][]} */
+	plain() {
+		return [];
+	}
+
+	/** @param {string} str */
+	setText(str) {
+		const src = this.type === 'dd' ? ':' : ';:*#';
+		if (new RegExp(`[^${src}]`).test(str)) {
+			throw new RangeError(`${this.constructor.name} 仅能包含${src.split('').map(c => `"${c}"`).join('、')}！`);
+		}
+		this.#update(str);
+		return super.setText(str);
+	}
+}
+
+Parser.classes.DdToken = __filename;
+module.exports = DdToken;

package/src/nowiki/hr.js

@@ -1,13 +1,14 @@
 'use strict';
 
-const /** @type {Parser} */ Parser = require('../..'),
+const sol = require('../../mixin/sol'),
+	/** @type {Parser} */ Parser = require('../..'),
 	NowikiToken = require('.');
 
 /**
  * `<hr>`
  * @classdesc `{childNodes: [string]}`
  */
-class HrToken extends NowikiToken {
+class HrToken extends sol(NowikiToken) {
 	type = 'hr';
 
 	/**

package/src/nowiki/list.js

@@ -0,0 +1,16 @@
+'use strict';
+
+const sol = require('../../mixin/sol'),
+	/** @type {Parser} */ Parser = require('../..'),
+	DdToken = require('./dd');
+
+/**
+ * ;:*#
+ * @classdesc `{childNodes: [string]}`
+ */
+class ListToken extends sol(DdToken) {
+	type = 'list';
+}
+
+Parser.classes.ListToken = __filename;
+module.exports = ListToken;

package/src/table/index.js

@@ -70,15 +70,15 @@ class TableToken extends TrToken {
 	}
 
 	/**
-	 * @template {string|Token} T
+	 * @template {TrToken|SyntaxToken} T
 	 * @param {T} token
 	 * @returns {T}
 	 * @complexity `n`
 	 */
 	insertAt(token, i = this.childNodes.length) {
-		const previous = this.childNodes.at(i - 1),
+		const previous = this.children.at(i - 1),
 			{closingPattern} = TableToken;
-		if (token instanceof TrToken && token.type === 'td' && previous instanceof TrToken && previous.type === 'tr') {
+		if (token.type === 'td' && previous.type === 'tr') {
 			Parser.warn('改为将单元格插入当前行。');
 			return previous.appendChild(token);
 		} else if (!Parser.running && i === this.childNodes.length && token instanceof SyntaxToken
@@ -417,14 +417,14 @@ class TableToken extends TrToken {
 	/** @complexity `n` */
 	#prependTableRow() {
 		const row = Parser.run(() => new TrToken('\n|-', undefined, this.getAttribute('config'))),
-			{childNodes} = this,
-			[,, plain] = childNodes,
-			start = typeof plain === 'string' || plain.isPlain() ? 3 : 2,
-			/** @type {TdToken[]} */ children = childNodes.slice(start),
-			index = children.findIndex(({type}) => type !== 'td');
+			{children} = this,
+			[,, plain] = children,
+			start = plain?.isPlain() ? 3 : 2,
+			/** @type {TdToken[]} */ tdChildren = children.slice(start),
+			index = tdChildren.findIndex(({type}) => type !== 'td');
 		this.insertAt(row, index === -1 ? -1 : index + start);
 		Parser.run(() => {
-			for (const cell of children.slice(0, index === -1 ? undefined : index)) {
+			for (const cell of tdChildren.slice(0, index === -1 ? undefined : index)) {
 				if (cell.subtype !== 'caption') {
 					row.appendChild(cell);
 				}

package/src/table/tr.js

@@ -48,15 +48,13 @@ class TrToken extends attributeParent(Token, 1) {
 	}
 
 	#correct() {
-		const [,, child] = this.childNodes;
-		if (typeof child === 'string' && !child.startsWith('\n')) {
-			this.setText(`\n${child}`, 2);
-		} else if (typeof child !== 'string' && child?.isPlain()) {
+		const [,, child] = this.children;
+		if (child?.isPlain()) {
 			const {firstChild} = child;
 			if (typeof firstChild !== 'string') {
 				child.prepend('\n');
 			} else if (!firstChild.startsWith('\n')) {
-				child.setText(`\n${firstChild}`, 0);
+				child.setText(`\n${firstChild}`);
 			}
 		}
 	}

package/typings/node.d.ts

@@ -7,7 +7,7 @@ declare global {
 		T extends 'childNodes' ? (string|Token)[] :
 		T extends 'parentNode' ? Token|undefined :
 		T extends 'optional'|'tags' ? string[] :
-		T extends 'stage' ? number :
+		T extends 'stage'|'indent' ? number :
 		T extends 'config' ? ParserConfig :
 		T extends 'accum' ? accum :
 		T extends 'acceptable' ? Record<string, Ranges> :
@@ -15,7 +15,7 @@ declare global {
 		T extends 'keys' ? Set<string> :
 		T extends 'args' ? Record<string, Set<ParameterToken>> :
 		T extends 'attr' ? Map<string, string|true> :
-		T extends 'include'|'selfLink' ? boolean :
+		T extends 'include'|'selfLink'|'ul'|'ol'|'dt' ? boolean :
 		T extends 'pattern' ? RegExp :
 		string;
 }

package/src/listToken.js

@@ -1,47 +0,0 @@
-'use strict';
-const Token = require('.'),
-	AtomToken = require('./atom'),
-	{fixToken} = require('./util');
-
-class ListToken extends fixToken(Token) {
-	type = 'list';
-
-	/**
-	 * @param {string} syntax
-	 * @param {?string|number|Token|(string|Token)[]} content
-	 * @param {Object<string, any>} config
-	 * @param {Token} parent
-	 * @param {Token[]} accum
-	 */
-	constructor(syntax, content, config = require(Token.config), parent = null, accum = [], isTable = false) {
-		if (/[^:;#*]/.test(syntax)) {
-			throw new RangeError('List语法只接受":"、";"、"#"或"*"！');
-		}
-		super(new AtomToken(syntax, 'list-syntax'), config, true, parent, accum, ['AtomToken', 'Token']);
-		const inner = new Token(content, config, true, this, accum);
-		inner.type = 'list-inner';
-		inner.set('stage', isTable ? 4 : 10);
-		this.lists = new Set(syntax.split(''));
-		this.seal();
-	}
-
-	isDt() {
-		return this.$children[0].contains(';');
-	}
-
-	idDd() {
-		return this.$children[0].contains(':');
-	}
-
-	isOl() {
-		return this.$children[0].contains('#');
-	}
-
-	isUl() {
-		return this.$children[0].contains('*');
-	}
-}
-
-Token.classes.ListToken = ListToken;
-
-module.exports = ListToken;