wikiparser-node 0.3.0 → 0.3.1

package/.eslintrc.json

@@ -3,7 +3,16 @@
 		"node": true,
 		"es2021": true
 	},
-	"extends": "eslint:recommended",
+	"plugins": [
+		"promise",
+		"regexp"
+	],
+	"extends": [
+		"eslint:recommended",
+		"plugin:promise/recommended",
+		"plugin:n/recommended-script",
+		"plugin:regexp/recommended"
+	],
 	"parserOptions": {
 		"ecmaVersion": 13
 	},
@@ -135,7 +144,10 @@
 		"no-shadow": [
 			2,
 			{
-				"builtinGlobals": true
+				"builtinGlobals": true,
+				"allow": [
+					"root"
+				]
 			}
 		],
 		"no-undef-init": 2,
@@ -185,7 +197,7 @@
 		],
 		"require-await": 2,
 		"spaced-comment": 2,
-        "strict": 2,
+		"strict": 2,
 		"symbol-description": 2,
 		"yoda": 2,
 		"array-bracket-spacing": 2,
@@ -271,6 +283,37 @@
 		"wrap-iife": [
 			2,
 			"inside"
-		]
+		],
+		"promise/no-multiple-resolved": 1,
+		"n/exports-style": [
+			2,
+			"module.exports"
+		],
+		"n/no-path-concat": 2,
+		"n/no-missing-require": 2,
+		"regexp/no-contradiction-with-assertion": 2,
+		"regexp/no-missing-g-flag": 2,
+		"regexp/no-super-linear-backtracking": [
+			2,
+			{
+				"report": "potential"
+			}
+		],
+		"regexp/no-super-linear-move": [
+			2,
+			{
+				"report": "potential"
+			}
+		],
+		"regexp/no-extra-lookaround-assertions": 2,
+		"regexp/no-octal": 2,
+		"regexp/no-standalone-backslash": 2,
+		"regexp/prefer-escape-replacement-dollar-char": 2,
+		"regexp/prefer-quantifier": 2,
+		"regexp/prefer-regexp-test": 2,
+		"regexp/use-ignore-case": 2,
+		"regexp/hexadecimal-escape": 2,
+		"regexp/letter-case": 2,
+		"regexp/prefer-lookaround": 2
 	}
 }

package/config/default.json

@@ -124,34 +124,47 @@
 		"媒体": -2,
 		"媒体文件": -2,
 		"媒體": -2,
+		"メディア": -2,
 		"special": -1,
 		"特殊": -1,
+		"特別": -1,
 		"talk": 1,
 		"对话": 1,
 		"對話": 1,
 		"讨论": 1,
 		"討論": 1,
+		"トーク": 1,
+		"ノート": 1,
 		"user": 2,
 		"u": 2,
 		"使用者": 2,
 		"用戶": 2,
 		"用户": 2,
-		"user_talk": 3,
+		"利用者": 2,
+		"user talk": 3,
 		"使用者討論": 3,
 		"用戶對話": 3,
 		"用戶討論": 3,
 		"用户对话": 3,
 		"用户讨论": 3,
+		"利用者‐会話": 3,
+		"利用者・トーク": 3,
 		"llwiki": 4,
+		"mgp": 4,
+		"moegirlpedia": 4,
 		"萌娘百科": 4,
+		"萌えっ娘百科事典": 4,
 		"project": 4,
-		"llwiki_talk": 5,
+		"llwiki talk": 5,
 		"llwiki讨论": 5,
 		"llwiki討論": 5,
-		"萌娘百科_talk": 5,
+		"萌娘百科 talk": 5,
 		"萌娘百科讨论": 5,
 		"萌娘百科討論": 5,
-		"project_talk": 5,
+		"moegirlpedia talk": 5,
+		"萌えっ娘百科事典‐ノート": 5,
+		"萌えっ娘百科事典・トーク": 5,
+		"project talk": 5,
 		"file": 6,
 		"image": 6,
 		"图像": 6,
@@ -159,8 +172,10 @@
 		"文件": 6,
 		"档案": 6,
 		"檔案": 6,
-		"file_talk": 7,
-		"image_talk": 7,
+		"ファイル": 6,
+		"画像": 6,
+		"file talk": 7,
+		"image talk": 7,
 		"图像对话": 7,
 		"图像讨论": 7,
 		"圖像對話": 7,
@@ -173,16 +188,22 @@
 		"档案讨论": 7,
 		"檔案對話": 7,
 		"檔案討論": 7,
+		"ファイル‐ノート": 7,
+		"ファイル・トーク": 7,
+		"画像‐ノート": 7,
 		"mediawiki": 8,
-		"mediawiki_talk": 9,
+		"mediawiki talk": 9,
 		"mediawiki讨论": 9,
 		"mediawiki討論": 9,
+		"mediawiki‐ノート": 9,
+		"mediawiki・トーク": 9,
 		"template": 10,
 		"t": 10,
 		"样板": 10,
 		"模板": 10,
 		"樣板": 10,
-		"template_talk": 11,
+		"テンプレート": 10,
+		"template talk": 11,
 		"样板对话": 11,
 		"样板讨论": 11,
 		"模板对话": 11,
@@ -191,33 +212,44 @@
 		"模板討論": 11,
 		"樣板對話": 11,
 		"樣板討論": 11,
+		"template‐ノート": 11,
+		"テンプレート・トーク": 11,
+		"h": 12,
 		"help": 12,
 		"使用說明": 12,
 		"帮助": 12,
 		"幫助": 12,
-		"help_talk": 13,
+		"ヘルプ": 12,
+		"help talk": 13,
 		"使用說明討論": 13,
 		"帮助对话": 13,
 		"帮助讨论": 13,
 		"幫助對話": 13,
 		"幫助討論": 13,
+		"help‐ノート": 13,
+		"ヘルプ・トーク": 13,
 		"category": 14,
 		"cat": 14,
 		"分类": 14,
 		"分類": 14,
-		"category_talk": 15,
+		"カテゴリ": 14,
+		"category talk": 15,
 		"分类对话": 15,
 		"分类讨论": 15,
 		"分類對話": 15,
 		"分類討論": 15,
+		"category‐ノート": 15,
+		"カテゴリ・トーク": 15,
 		"widget": 274,
-		"widget_talk": 275,
+		"widget talk": 275,
 		"module": 828,
 		"模块": 828,
 		"模組": 828,
-		"module_talk": 829,
+		"モジュール": 828,
+		"module talk": 829,
 		"模块讨论": 829,
-		"模組討論": 829
+		"模組討論": 829,
+		"モジュール・トーク": 829
 	},
 	"parserFunction": [
 		[

package/config/llwiki.json

@@ -126,7 +126,7 @@
 		"使用者": 2,
 		"用戶": 2,
 		"用户": 2,
-		"user_talk": 3,
+		"user talk": 3,
 		"使用者討論": 3,
 		"用戶對話": 3,
 		"用戶討論": 3,
@@ -134,10 +134,10 @@
 		"用户讨论": 3,
 		"llwiki": 4,
 		"project": 4,
-		"llwiki_talk": 5,
+		"llwiki talk": 5,
 		"llwiki讨论": 5,
 		"llwiki討論": 5,
-		"project_talk": 5,
+		"project talk": 5,
 		"file": 6,
 		"image": 6,
 		"图像": 6,
@@ -145,8 +145,8 @@
 		"文件": 6,
 		"档案": 6,
 		"檔案": 6,
-		"file_talk": 7,
-		"image_talk": 7,
+		"file talk": 7,
+		"image talk": 7,
 		"图像对话": 7,
 		"图像讨论": 7,
 		"圖像對話": 7,
@@ -160,14 +160,14 @@
 		"檔案對話": 7,
 		"檔案討論": 7,
 		"mediawiki": 8,
-		"mediawiki_talk": 9,
+		"mediawiki talk": 9,
 		"mediawiki讨论": 9,
 		"mediawiki討論": 9,
 		"template": 10,
 		"样板": 10,
 		"模板": 10,
 		"樣板": 10,
-		"template_talk": 11,
+		"template talk": 11,
 		"样板对话": 11,
 		"样板讨论": 11,
 		"模板对话": 11,
@@ -180,7 +180,7 @@
 		"使用說明": 12,
 		"帮助": 12,
 		"幫助": 12,
-		"help_talk": 13,
+		"help talk": 13,
 		"使用說明討論": 13,
 		"帮助对话": 13,
 		"帮助讨论": 13,
@@ -189,17 +189,17 @@
 		"category": 14,
 		"分类": 14,
 		"分類": 14,
-		"category_talk": 15,
+		"category talk": 15,
 		"分类对话": 15,
 		"分类讨论": 15,
 		"分類對話": 15,
 		"分類討論": 15,
 		"widget": 274,
-		"widget_talk": 275,
+		"widget talk": 275,
 		"module": 828,
 		"模块": 828,
 		"模組": 828,
-		"module_talk": 829,
+		"module talk": 829,
 		"模块讨论": 829,
 		"模組討論": 829
 	},

package/config/moegirl.json

@@ -119,30 +119,43 @@
 		"媒体": -2,
 		"媒体文件": -2,
 		"媒體": -2,
+		"メディア": -2,
 		"special": -1,
 		"特殊": -1,
+		"特別": -1,
 		"talk": 1,
 		"对话": 1,
 		"對話": 1,
 		"讨论": 1,
 		"討論": 1,
+		"トーク": 1,
+		"ノート": 1,
 		"user": 2,
 		"u": 2,
 		"使用者": 2,
 		"用戶": 2,
 		"用户": 2,
-		"user_talk": 3,
+		"利用者": 2,
+		"user talk": 3,
 		"使用者討論": 3,
 		"用戶對話": 3,
 		"用戶討論": 3,
 		"用户对话": 3,
 		"用户讨论": 3,
+		"利用者‐会話": 3,
+		"利用者・トーク": 3,
+		"mgp": 4,
+		"moegirlpedia": 4,
 		"萌娘百科": 4,
+		"萌えっ娘百科事典": 4,
 		"project": 4,
-		"萌娘百科_talk": 5,
+		"萌娘百科 talk": 5,
 		"萌娘百科讨论": 5,
 		"萌娘百科討論": 5,
-		"project_talk": 5,
+		"moegirlpedia talk": 5,
+		"萌えっ娘百科事典‐ノート": 5,
+		"萌えっ娘百科事典・トーク": 5,
+		"project talk": 5,
 		"file": 6,
 		"image": 6,
 		"图像": 6,
@@ -150,8 +163,10 @@
 		"文件": 6,
 		"档案": 6,
 		"檔案": 6,
-		"file_talk": 7,
-		"image_talk": 7,
+		"ファイル": 6,
+		"画像": 6,
+		"file talk": 7,
+		"image talk": 7,
 		"图像对话": 7,
 		"图像讨论": 7,
 		"圖像對話": 7,
@@ -164,16 +179,22 @@
 		"档案讨论": 7,
 		"檔案對話": 7,
 		"檔案討論": 7,
+		"ファイル‐ノート": 7,
+		"ファイル・トーク": 7,
+		"画像‐ノート": 7,
 		"mediawiki": 8,
-		"mediawiki_talk": 9,
+		"mediawiki talk": 9,
 		"mediawiki讨论": 9,
 		"mediawiki討論": 9,
+		"mediawiki‐ノート": 9,
+		"mediawiki・トーク": 9,
 		"template": 10,
 		"t": 10,
 		"样板": 10,
 		"模板": 10,
 		"樣板": 10,
-		"template_talk": 11,
+		"テンプレート": 10,
+		"template talk": 11,
 		"样板对话": 11,
 		"样板讨论": 11,
 		"模板对话": 11,
@@ -182,33 +203,44 @@
 		"模板討論": 11,
 		"樣板對話": 11,
 		"樣板討論": 11,
+		"template‐ノート": 11,
+		"テンプレート・トーク": 11,
+		"h": 12,
 		"help": 12,
 		"使用說明": 12,
 		"帮助": 12,
 		"幫助": 12,
-		"help_talk": 13,
+		"ヘルプ": 12,
+		"help talk": 13,
 		"使用說明討論": 13,
 		"帮助对话": 13,
 		"帮助讨论": 13,
 		"幫助對話": 13,
 		"幫助討論": 13,
+		"help‐ノート": 13,
+		"ヘルプ・トーク": 13,
 		"category": 14,
 		"cat": 14,
 		"分类": 14,
 		"分類": 14,
-		"category_talk": 15,
+		"カテゴリ": 14,
+		"category talk": 15,
 		"分类对话": 15,
 		"分类讨论": 15,
 		"分類對話": 15,
 		"分類討論": 15,
+		"category‐ノート": 15,
+		"カテゴリ・トーク": 15,
 		"widget": 274,
-		"widget_talk": 275,
+		"widget talk": 275,
 		"module": 828,
 		"模块": 828,
 		"模組": 828,
-		"module_talk": 829,
+		"モジュール": 828,
+		"module talk": 829,
 		"模块讨论": 829,
-		"模組討論": 829
+		"模組討論": 829,
+		"モジュール・トーク": 829
 	},
 	"parserFunction": [
 		[

package/index.js

@@ -1,6 +1,7 @@
 'use strict';
 
 const fs = require('fs'),
+	path = require('path'),
 	{text} = require('./util/string');
 
 const /** @type {Parser} */ Parser = {
@@ -49,12 +50,12 @@ const /** @type {Parser} */ Parser = {
 			...Object.entries(this.mixins),
 			...Object.entries(this.parsers),
 		];
-		for (const [, path] of entries) {
-			delete require.cache[require.resolve(path)];
+		for (const [, filePath] of entries) {
+			delete require.cache[require.resolve(filePath)];
 		}
-		for (const [name, path] of entries) {
+		for (const [name, filePath] of entries) {
 			if (name in global) {
-				global[name] = require(path);
+				global[name] = require(filePath);
 			}
 		}
 	},
@@ -88,8 +89,7 @@ const /** @type {Parser} */ Parser = {
 
 	isInterwiki(title, {interwiki} = Parser.getConfig()) {
 		title = String(title);
-		return title.replaceAll('_', ' ').replace(/^\s*:?\s*/, '')
-			.match(new RegExp(`^(${interwiki.join('|')})\\s*:`, 'i'));
+		return RegExp(`^(${interwiki.join('|')})\\s*:`, 'i').exec(title.replaceAll('_', ' ').replace(/^\s*:?\s*/, ''));
 	},
 
 	normalizeTitle(title, defaultNs = 0, include = false, config = Parser.getConfig(), halfParsed = false) {
@@ -105,7 +105,7 @@ const /** @type {Parser} */ Parser = {
 		if (token) {
 			const build = /** @param {string[]} keys */ keys => {
 				for (const key of keys) {
-					if (titleObj[key].includes('\x00')) {
+					if (titleObj[key].includes('\0')) {
 						titleObj[key] = text(token.buildFromStr(titleObj[key]));
 					}
 				}
@@ -135,7 +135,7 @@ const /** @type {Parser} */ Parser = {
 				token.parse(maxStage, include);
 			} catch (e) {
 				if (e instanceof Error) {
-					const file = `${__dirname}/errors/${new Date().toISOString()}`,
+					const file = path.join(__dirname, 'errors', new Date().toISOString()),
 						stage = token.getAttribute('stage');
 					fs.writeFileSync(file, stage === this.MAX_STAGE ? wikitext : token.toString());
 					fs.writeFileSync(`${file}.err`, e.stack);
@@ -150,13 +150,13 @@ const /** @type {Parser} */ Parser = {
 	},
 
 	reparse(date) {
-		const path = `${__dirname}/errors/`,
-			main = fs.readdirSync(path).find(name => name.startsWith(date) && name.endsWith('Z'));
+		const main = fs.readdirSync(path.join(__dirname, 'errors'))
+			.find(name => name.startsWith(date) && name.endsWith('Z'));
 		if (!main) {
 			throw new RangeError(`找不到对应时间戳的错误记录：${date}`);
 		}
 		const Token = require('./src'),
-			file = `${path}${main}`,
+			file = path.join(__dirname, 'errors', main),
 			wikitext = fs.readFileSync(file, 'utf8'),
 			{stage, include, config} = require(`${file}.json`);
 		this.config = config;

package/lib/element.js

@@ -346,11 +346,11 @@ class AstElement extends AstNode {
 		visible: false,
 	};
 	/** @type {pseudo[]} */ static #pseudoKeys = Object.keys(AstElement.#pseudo);
-	static #pseudoRegex = new RegExp(
+	static #pseudoRegex = RegExp(
 		`:(${this.#pseudoKeys.join('|')})(?:\\(\\s*("[^"]*"|'[^']*'|[^()]*?)\\s*\\))?(?=:|\\s*(?:,|$))`,
 		'g',
 	);
-	static #simplePseudoRegex = new RegExp(`:(?:${this.#pseudoKeys.join('|')})(?:\\(.*?\\))?(?=:|\\s*(?:,|$))`, 'g');
+	static #simplePseudoRegex = RegExp(`:(?:${this.#pseudoKeys.join('|')})(?:\\(.*?\\))?(?=:|\\s*(?:,|$))`, 'g');
 
 	/**
 	 * @returns {boolean}
@@ -365,7 +365,7 @@ class AstElement extends AstNode {
 		simple &&= Parser.running;
 		const /** @type {Record<string, string>} */ escapedQuotes = {'"': '&quot;', "'": '&apos;'},
 			escapedSelector = selector.replace(/\\["']/g, m => escapedQuotes[m[1]]);
-		if (simple || !AstElement.#pseudoRegex.test(escapedSelector)) {
+		if (simple || escapedSelector.search(AstElement.#pseudoRegex) === -1) {
 			if (!simple && selector.includes(',')) {
 				return Parser.run(() => selector.split(',').some(str => this.matches(str, true)));
 			}
@@ -384,10 +384,10 @@ class AstElement extends AstNode {
 				plainSelector = selector.replace(
 					/&(comma|ratio);/g, /** @param {string} name */ (_, name) => entities[name],
 				).replace(
-					/\[\s*(\w+)\s*(?:([~|^$*!]?=)\s*("[^"]*"|'[^']*'|[^[\]]*?)\s*(\si)?\s*)?]/g,
+					/\[\s*(\w+)\s*(?:([~|^$*!]?=)\s*("[^"]*"|'[^']*'|[^\s[\]]+)(?:\s+(i))?\s*)?\]/g,
 					/** @type {function(...string): ''} */ (_, key, equal, val, i) => {
 						if (equal) {
-							const quotes = val.match(/^(["']).*\1$/)?.[1];
+							const quotes = /^(["']).*\1$/.exec(val)?.[1];
 							attributes.push([
 								key,
 								equal,
@@ -418,7 +418,7 @@ class AstElement extends AstNode {
 						calls[f].push('');
 						return m;
 					}
-					const quotes = arg.match(/^(["']).*\1$/)?.[1];
+					const quotes = /^(["']).*\1$/.exec(arg)?.[1];
 					calls[f].push(quotes ? arg.slice(1, -1).replaceAll(escapedQuotes[quotes], quotes) : arg);
 					return `:${f}(${calls[f].length - 1})`;
 				},

package/lib/ranges.js

@@ -32,7 +32,7 @@ class Range {
 				throw new RangeError(`步长 ${this.step} 应为整数！`);
 			}
 		} else {
-			const mt = str.match(/^([+-])?(\d+)?n(?:([+-])(\d+))?$/);
+			const mt = /^([+-])?(\d+)?n(?:([+-])(\d+))?$/.exec(str);
 			if (mt) {
 				const [, sgnA = '+', a = 1, sgnB = '+'] = mt,
 					b = Number(mt[4] ?? 0);

package/lib/title.js

@@ -53,7 +53,7 @@ class Title {
 		this.main = ucfirst(title);
 		this.prefix = `${namespace}${namespace && ':'}`;
 		this.title = `${iw ? `${this.interwiki}:` : ''}${this.prefix}${this.main}`;
-		this.valid = Boolean(this.main) && !/\x00\d+[eh!+-]\x7f|[<>[\]{}|]/.test(this.title);
+		this.valid = Boolean(this.main || this.fragment) && !/\0\d+[eh!+-]\x7f|[<>[\]{}|]/.test(this.title);
 	}
 }
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wikiparser-node",
-  "version": "0.3.0",
+  "version": "0.3.1",
   "description": "A Node.js parser for MediaWiki markup with AST",
   "keywords": [
     "mediawiki",
@@ -22,7 +22,10 @@
   },
   "devDependencies": {
     "@types/node": "^17.0.23",
-    "eslint": "^8.30.0"
+    "eslint": "^8.30.0",
+    "eslint-plugin-promise": "^6.1.1",
+    "eslint-plugin-n": "^15.6.0",
+    "eslint-plugin-regexp": "^1.11.0"
   },
   "engines": {
     "node": "^18.4.0"

package/parser/brackets.js

@@ -8,15 +8,20 @@ const {removeComment} = require('../util/string'),
  * @param {accum} accum
  */
 const parseBrackets = (text, config = Parser.getConfig(), accum = []) => {
-	const source = '(?<=^(?:\x00\\d+c\x7f)*)={1,6}|\\[\\[|{{2,}|-{(?!{)',
+	const source = '^(\0\\d+c\x7f)*={1,6}|\\[\\[|\\{{2,}|-\\{(?!\\{)',
 		/** @type {BracketExecArray[]} */ stack = [],
 		closes = {'=': '\n', '{': '}{2,}|\\|', '-': '}-', '[': ']]'},
 		/** @type {Record<string, string>} */ marks = {'!': '!', '!!': '+', '(!': '{', '!)': '}', '!-': '-', '=': '~'};
-	let regex = new RegExp(source, 'gm'),
+	let regex = RegExp(source, 'gm'),
 		/** @type {BracketExecArray} */ mt = regex.exec(text),
 		moreBraces = text.includes('}}'),
 		lastIndex;
 	while (mt || lastIndex <= text.length && stack.at(-1)?.[0]?.[0] === '=') {
+		if (mt?.[1]) {
+			const [, {length}] = mt;
+			mt[0] = mt[0].slice(length);
+			mt.index += length;
+		}
 		const {0: syntax, index: curIndex} = mt ?? {0: '\n', index: text.length},
 			/** @type {BracketExecArray} */ top = stack.pop() ?? {},
 			{0: open, index, parts} = top,
@@ -27,9 +32,9 @@ const parseBrackets = (text, config = Parser.getConfig(), accum = []) => {
 			lastIndex = curIndex + 1;
 			const {pos, findEqual} = stack.at(-1) ?? {};
 			if (!pos || findEqual || removeComment(text.slice(pos, index)) !== '') {
-				const rmt = text.slice(index, curIndex).match(/^(={1,6})(.+)\1((?:\s|\x00\d+c\x7f)*)$/);
+				const rmt = /^(={1,6})(.+)\1((?:\s|\0\d+c\x7f)*)$/.exec(text.slice(index, curIndex));
 				if (rmt) {
-					text = `${text.slice(0, index)}\x00${accum.length}h\x7f${text.slice(curIndex)}`;
+					text = `${text.slice(0, index)}\0${accum.length}h\x7f${text.slice(curIndex)}`;
 					lastIndex = index + 4 + String(accum.length).length;
 					const HeadingToken = require('../src/heading');
 					new HeadingToken(rmt[1].length, rmt.slice(2), config, accum);
@@ -71,7 +76,7 @@ const parseBrackets = (text, config = Parser.getConfig(), accum = []) => {
 			}
 			if (!skip) {
 				/* 标记{{!}}结束 */
-				text = `${text.slice(0, index + rest)}\x00${length}${ch}\x7f${text.slice(lastIndex)}`;
+				text = `${text.slice(0, index + rest)}\0${length}${ch}\x7f${text.slice(lastIndex)}`;
 				lastIndex = index + rest + 3 + String(length).length;
 				if (rest > 1) {
 					stack.push({0: open.slice(0, rest), index, pos: index + rest, parts: [[]]});
@@ -93,7 +98,7 @@ const parseBrackets = (text, config = Parser.getConfig(), accum = []) => {
 			stack.pop();
 			curTop = stack.at(-1);
 		}
-		regex = new RegExp(source + (curTop
+		regex = RegExp(source + (curTop
 			? `|${closes[curTop[0][0]]}${curTop.findEqual ? '|=' : ''}`
 			: ''
 		), 'gm');