wikiparser-node 0.8.0 → 0.8.1-m

package/config/minimum.json

@@ -0,0 +1,142 @@
+{
+	"ext": [],
+	"html": [
+		[
+			"b",
+			"bdi",
+			"del",
+			"i",
+			"ins",
+			"u",
+			"font",
+			"big",
+			"small",
+			"sub",
+			"sup",
+			"h1",
+			"h2",
+			"h3",
+			"h4",
+			"h5",
+			"h6",
+			"cite",
+			"code",
+			"em",
+			"s",
+			"strike",
+			"strong",
+			"tt",
+			"var",
+			"div",
+			"center",
+			"blockquote",
+			"ol",
+			"ul",
+			"dl",
+			"table",
+			"caption",
+			"pre",
+			"ruby",
+			"rb",
+			"rp",
+			"rt",
+			"rtc",
+			"p",
+			"span",
+			"abbr",
+			"dfn",
+			"kbd",
+			"samp",
+			"data",
+			"time",
+			"mark",
+			"tr",
+			"td",
+			"th",
+			"q",
+			"bdo"
+		],
+		[
+			"li",
+			"dt",
+			"dd"
+		],
+		[
+			"br",
+			"wbr",
+			"hr",
+			"meta",
+			"link",
+			"img"
+		]
+	],
+	"namespaces": {},
+	"nsid": {},
+	"parserFunction": [
+		{
+			"#language": "language",
+			"#special": "special",
+			"#speciale": "speciale",
+			"#tag": "tag",
+			"#formatdate": "formatdate",
+			"#invoke": "invoke",
+			"#while": "while",
+			"#dowhile": "dowhile",
+			"#loop": "loop",
+			"#forargs": "forargs",
+			"#fornumargs": "fornumargs",
+			"#if": "if",
+			"#ifeq": "ifeq",
+			"#switch": "switch",
+			"#ifexist": "ifexist",
+			"#ifexpr": "ifexpr",
+			"#iferror": "iferror",
+			"#time": "time",
+			"#timel": "timel",
+			"#expr": "expr",
+			"#rel2abs": "rel2abs",
+			"#titleparts": "titleparts",
+			"#categorytree": "categorytree",
+			"#urldecode": "urldecode",
+			"#choose": "choose",
+			"#var": "var",
+			"#varexists": "varexists",
+			"#var_final": "var_final",
+			"#vardefine": "vardefine",
+			"#vardefineecho": "vardefineecho",
+			"#widget": "widget",
+			"#related": "related",
+			"#cscore": "cscore"
+		},
+		[],
+		[
+			"msg",
+			"原始",
+			"raw"
+		],
+		[
+			"替代",
+			"subst",
+			"安全替代",
+			"safesubst"
+		]
+	],
+	"doubleUnderscore": [
+		[],
+		[]
+	],
+	"protocol": "bitcoin:|ftp://|ftps://|geo:|git://|gopher://|http://|https://|irc://|ircs://|magnet:|mailto:|mms://|news:|nntp://|redis://|sftp://|sip:|sips:|sms:|ssh://|svn://|tel:|telnet://|urn:|worldwind://|xmpp:",
+	"interwiki": [],
+	"img": {},
+	"variants": [
+		"zh",
+		"zh-hans",
+		"zh-hant",
+		"zh-cn",
+		"zh-tw",
+		"zh-hk",
+		"zh-sg",
+		"zh-my",
+		"zh-mo"
+	]
+}

package/index.js

@@ -1,128 +1,16 @@
 'use strict';
 
-const fs = require('fs'),
-	path = require('path');
-
 const /** @type {Parser} */ Parser = {
-	config: './config/default',
+	config: undefined,
+	minConfig: require('./config/minimum'),
 
 	MAX_STAGE: 11,
 
-	warning: true,
-	debugging: false,
-	running: false,
-
-	classes: {},
-	mixins: {},
-	parsers: {},
-	tool: {},
-
-	aliases: [
-		['AstText'],
-		['CommentToken', 'ExtToken', 'IncludeToken', 'NoincludeToken'],
-		['ArgToken', 'TranscludeToken', 'HeadingToken'],
-		['HtmlToken'],
-		['TableToken'],
-		['HrToken', 'DoubleUnderscoreToken'],
-		['LinkToken', 'FileToken', 'CategoryToken'],
-		['QuoteToken'],
-		['ExtLinkToken'],
-		['MagicLinkToken'],
-		['ListToken', 'DdToken'],
-		['ConverterToken'],
-	],
-	typeAliases: {
-		text: ['string', 'str'],
-		plain: ['regular', 'normal'],
-		// comment and extension
-		onlyinclude: ['only-include'],
-		noinclude: ['no-include'],
-		include: ['includeonly', 'include-only'],
-		comment: undefined,
-		ext: ['extension'],
-		'ext-attrs': ['extension-attrs', 'ext-attributes', 'extension-attributes'],
-		'ext-attr-dirty': ['extension-attr-dirty', 'ext-attribute-dirty', 'extension-attribute-dirty'],
-		'ext-attr': ['extension-attr', 'ext-attribute', 'extension-attribute'],
-		'attr-key': ['attribute-key'],
-		'attr-value': ['attribute-value', 'attr-val', 'attribute-val'],
-		'ext-inner': ['extension-inner'],
-		// triple brackets
-		arg: ['argument'],
-		'arg-name': ['argument-name'],
-		'arg-default': ['argument-default'],
-		hidden: ['arg-redundant'],
-		// double brackets
-		'magic-word': ['parser-function', 'parser-func'],
-		'magic-word-name': ['parser-function-name', 'parser-func-name'],
-		'invoke-function': ['invoke-func', 'lua-function', 'lua-func', 'module-function', 'module-func'],
-		'invoke-module': ['lua-module'],
-		template: undefined,
-		'template-name': undefined,
-		parameter: ['param'],
-		'parameter-key': ['param-key'],
-		'parameter-value': ['parameter-val', 'param-value', 'param-val'],
-		// heading
-		heading: ['header'],
-		'heading-title': ['header-title'],
-		'heading-trail': ['header-trail'],
-		// html
-		html: undefined,
-		'html-attrs': ['html-attributes'],
-		'html-attr-dirty': ['html-attribute-dirty'],
-		'html-attr': ['html-attribute'],
-		// table
-		table: undefined,
-		tr: ['table-row'],
-		td: ['table-cell', 'table-data'],
-		'table-syntax': undefined,
-		'table-attrs': ['tr-attrs', 'td-attrs', 'table-attributes', 'tr-attributes', 'td-attributes'],
-		'table-attr-dirty':
-			['tr-attr-dirty', 'td-attr-dirty', 'table-attribute-dirty', 'tr-attribute-dirty', 'td-attribute-dirty'],
-		'table-attr': ['tr-attr', 'td-attr', 'table-attribute', 'tr-attribute', 'td-attribute'],
-		'table-inter': undefined,
-		'td-inner': ['table-cell-inner', 'table-data-inner'],
-		// hr and double-underscore
-		hr: ['horizontal'],
-		'double-underscore': ['underscore', 'behavior-switch', 'behaviour-switch'],
-		// link
-		link: ['wikilink'],
-		'link-target': ['wikilink-target'],
-		'link-text': ['wikilink-text'],
-		category: ['category-link', 'cat', 'cat-link'],
-		file: ['file-link', 'image', 'image-link', 'img', 'img-link'],
-		'gallery-image': ['gallery-file', 'gallery-img'],
-		'imagemap-image': ['imagemap-file', 'imagemap-img', 'image-map-image', 'image-map-file', 'image-map-img'],
-		'image-parameter': ['img-parameter', 'image-param', 'img-param'],
-		// quotes
-		quote: ['quotes', 'quot', 'apostrophe', 'apostrophes', 'apos'],
-		// external link
-		'ext-link': ['external-link'],
-		'ext-link-text': ['external-link-text'],
-		'ext-link-url': ['external-link-url'],
-		// magic link
-		'free-ext-link': ['free-external-link', 'magic-link'],
-		// list
-		list: ['ol', 'ordered-list', 'ul', 'unordered-list', 'dl', 'description-list'],
-		dd: ['indent', 'indentation'],
-		// converter
-		converter: ['convert', 'conversion'],
-		'converter-flags': ['convert-flags', 'conversion-flags'],
-		'converter-flag': ['convert-flag', 'conversion-flag'],
-		'converter-rule': ['convert-rule', 'conversion-rule'],
-		'converter-rule-noconvert': ['convert-rule-noconvert', 'conversion-rule-noconvert'],
-		'converter-rule-variant': ['convert-rule-variant', 'conversion-rule-variant'],
-		'converter-rule-to': ['convert-rule-to', 'conversion-rule-to'],
-		'converter-rule-from': ['convert-rule-from', 'conversion-rule-from'],
-		// specific extensions
-		'param-line': ['parameter-line'],
-		'charinsert-line': undefined,
-		'imagemap-link': ['image-map-link'],
-	},
-
-	promises: [Promise.resolve()],
-
-	getConfig() {
-		return {...require(this.config), excludes: []};
+	getConfig(path) {
+		if (path) {
+			this.config = require(path);
+		}
+		return {...this.minConfig, ...this.config, excludes: []};
 	},
 
 	normalizeTitle(
@@ -147,22 +35,6 @@ const /** @type {Parser} */ Parser = {
 		}
 		const Title = require('./lib/title');
 		const titleObj = new Title(String(title), defaultNs, config, decode, selfLink);
-		if (token) {
-			/**
-			 * 重建部分属性值
-			 * @param {string[]} keys 属性键
-			 */
-			const build = keys => {
-				for (const key of keys) {
-					if (titleObj[key].includes('\0')) {
-						titleObj[key] = token.getAttribute('buildFromStr')(titleObj[key], 'text');
-					}
-				}
-			};
-			this.run(() => {
-				build(['title', 'main', 'fragment']);
-			});
-		}
 		return titleObj;
 	},
 
@@ -176,96 +48,13 @@ const /** @type {Parser} */ Parser = {
 			token = new Token(wikitext, config);
 			try {
 				token.parse(maxStage, include);
-			} catch (e) {
-				if (e instanceof Error) {
-					const file = path.join(__dirname, 'errors', new Date().toISOString()),
-						stage = token.getAttribute('stage');
-					fs.writeFileSync(file, stage === this.MAX_STAGE ? wikitext : String(token));
-					fs.writeFileSync(`${file}.err`, e.stack);
-					fs.writeFileSync(`${file}.json`, JSON.stringify({
-						stage, include: token.getAttribute('include'), config: this.config,
-					}, null, '\t'));
-				}
-				throw e;
-			}
+			} catch {}
 		});
-		if (this.debugging) {
-			let restored = String(token),
-				process = '解析';
-			if (restored === wikitext) {
-				const entities = {lt: '<', gt: '>', amp: '&'};
-				restored = token.print().replace(
-					/<[^<]+?>|&([lg]t|amp);/gu,
-					/** @param {string} s */ (_, s) => s ? entities[s] : '',
-				);
-				process = '渲染HTML';
-			}
-			if (restored !== wikitext) {
-				const diff = require('./util/diff');
-				const {promises: {0: cur, length}} = this;
-				this.promises.unshift((async () => {
-					await cur;
-					this.error(`${process}过程中不可逆地修改了原始文本！`);
-					return diff(wikitext, restored, length);
-				})());
-			}
-		}
 		return token;
 	},
 
 	run(callback) {
-		const {running} = this;
-		this.running = true;
-		try {
-			const result = callback();
-			this.running = running;
-			return result;
-		} catch (e) {
-			this.running = running;
-			throw e;
-		}
-	},
-
-	warn(msg, ...args) {
-		if (this.warning) {
-			console.warn('\x1B[33m%s\x1B[0m', msg, ...args);
-		}
-	},
-	debug(msg, ...args) {
-		if (this.debugging) {
-			console.debug('\x1B[34m%s\x1B[0m', msg, ...args);
-		}
-	},
-	error(msg, ...args) {
-		console.error('\x1B[31m%s\x1B[0m', msg, ...args);
-	},
-	info(msg, ...args) {
-		console.info('\x1B[32m%s\x1B[0m', msg, ...args);
-	},
-
-	log(f) {
-		if (typeof f === 'function') {
-			console.log(String(f));
-		}
-	},
-
-	clearCache() {
-		const entries = [
-			...Object.entries(this.classes),
-			...Object.entries(this.mixins),
-			...Object.entries(this.parsers),
-			...Object.entries(this.tool),
-		];
-		for (const [, filePath] of entries) {
-			try {
-				delete require.cache[require.resolve(filePath)];
-			} catch {}
-		}
-		for (const [name, filePath] of entries) {
-			if (name in global) {
-				global[name] = require(filePath);
-			}
-		}
+		return callback();
 	},
 
 	isInterwiki(title, {interwiki} = Parser.getConfig()) {
@@ -273,42 +62,11 @@ const /** @type {Parser} */ Parser = {
 		return new RegExp(`^(${interwiki.join('|')})\\s*:`, 'iu')
 			.exec(title.replaceAll('_', ' ').replace(/^\s*:?\s*/u, ''));
 	},
-
-	reparse(date) {
-		const main = fs.readdirSync(path.join(__dirname, 'errors'))
-			.find(name => name.startsWith(date) && name.at(-1) === 'Z');
-		if (!main) {
-			throw new RangeError(`找不到对应时间戳的错误记录：${date}`);
-		}
-		const file = path.join(__dirname, 'errors', main),
-			wikitext = fs.readFileSync(file, 'utf8');
-		const {stage, include, config} = require(`${file}.json`),
-			Token = require('./src');
-		this.config = config;
-		return this.run(() => {
-			const halfParsed = stage < this.MAX_STAGE,
-				token = new Token(wikitext, this.getConfig(), halfParsed);
-			if (halfParsed) {
-				token.setAttribute('stage', stage).getAttribute('parseOnce')(stage, include);
-			} else {
-				token.parse(undefined, include);
-			}
-			fs.unlinkSync(file);
-			fs.unlinkSync(`${file}.err`);
-			fs.unlinkSync(`${file}.json`);
-			return token;
-		});
-	},
-
-	getTool() {
-		delete require.cache[require.resolve('./tool')];
-		return require('./tool');
-	},
 };
 
 const /** @type {PropertyDescriptorMap} */ def = {},
-	immutable = new Set(['MAX_STAGE', 'aliases', 'typeAliases', 'promises']),
-	enumerable = new Set(['config', 'normalizeTitle', 'parse', 'isInterwiki', 'getTool']);
+	immutable = new Set(['MAX_STAGE', 'minConfig']),
+	enumerable = new Set(['config', 'normalizeTitle', 'parse', 'isInterwiki']);
 for (const key in Parser) {
 	if (immutable.has(key)) {
 		def[key] = {enumerable: false, writable: false};