wikiparser-node 1.2.0-b → 1.2.0

package/dist/util/constants.js

@@ -0,0 +1,108 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.promises = exports.typeAliases = exports.aliases = exports.parsers = exports.mixins = exports.classes = exports.MAX_STAGE = void 0;
+exports.MAX_STAGE = 11;
+/* NOT FOR BROWSER */
+exports.classes = {};
+exports.mixins = {};
+exports.parsers = {};
+exports.aliases = [
+    ['AstText'],
+    ['CommentToken', 'ExtToken', 'IncludeToken', 'NoincludeToken'],
+    ['ArgToken', 'TranscludeToken', 'HeadingToken'],
+    ['HtmlToken'],
+    ['TableToken'],
+    ['HrToken', 'DoubleUnderscoreToken'],
+    ['LinkToken', 'FileToken', 'CategoryToken'],
+    ['QuoteToken'],
+    ['ExtLinkToken'],
+    ['MagicLinkToken'],
+    ['ListToken', 'DdToken'],
+    ['ConverterToken'],
+];
+exports.typeAliases = {
+    text: ['string', 'str'],
+    plain: ['regular', 'normal'],
+    root: undefined,
+    // 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 braces
+    arg: ['argument'],
+    'arg-name': ['argument-name'],
+    'arg-default': ['argument-default'],
+    hidden: ['arg-redundant'],
+    // double braces
+    '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-variant': ['convert-rule-variant', 'conversion-rule-variant'],
+    'converter-rule-to': ['convert-rule-to', 'conversion-rule-to', 'converter-rule-noconvert'],
+    'converter-rule-from': ['convert-rule-from', 'conversion-rule-from'],
+    // specific extensions
+    'param-line': ['parameter-line'],
+    'imagemap-link': ['image-map-link'],
+};
+exports.promises = [Promise.resolve()];

package/dist/util/debug.js

@@ -0,0 +1,64 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.undo = exports.Shadow = void 0;
+exports.Shadow = {
+    /* NOT FOR BROWSER */
+    running: false,
+    /* NOT FOR BROWSER END */
+    /** @private */
+    run(callback) {
+        const { running } = this;
+        this.running = true;
+        try {
+            const result = callback();
+            this.running = running;
+            return result;
+        }
+        catch (e) {
+            this.running = running;
+            throw e;
+        }
+    },
+};
+/* NOT FOR BROWSER */
+/**
+ * 撤销最近一次Mutation
+ * @param e 事件
+ * @param data 事件数据
+ * @throws `RangeError` 无法撤销的事件类型
+ */
+const undo = (e, data) => {
+    const { target, type } = e;
+    switch (data.type) {
+        case 'remove': {
+            const childNodes = [...target.childNodes];
+            childNodes.splice(data.position, 0, data.removed);
+            data.removed.setAttribute('parentNode', target);
+            target.setAttribute('childNodes', childNodes);
+            break;
+        }
+        case 'insert': {
+            const childNodes = [...target.childNodes];
+            childNodes.splice(data.position, 1);
+            data.inserted.setAttribute('parentNode', undefined);
+            target.setAttribute('childNodes', childNodes);
+            break;
+        }
+        case 'replace': {
+            const { parentNode } = target, childNodes = [...parentNode.childNodes];
+            childNodes.splice(data.position, 1, data.oldToken);
+            data.oldToken.setAttribute('parentNode', parentNode);
+            data.newToken.setAttribute('parentNode', undefined);
+            parentNode.setAttribute('childNodes', childNodes);
+            break;
+        }
+        case 'text':
+            if (target.type === 'text') {
+                target.replaceData(data.oldText);
+            }
+            break;
+        default:
+            throw new RangeError(`无法撤销未知类型的事件：${String(type)}`);
+    }
+};
+exports.undo = undo;

package/dist/util/diff.js

@@ -0,0 +1,72 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.diff = exports.cmd = void 0;
+const fs = require("fs/promises");
+const child_process_1 = require("child_process");
+process.on('unhandledRejection', e => {
+    console.error(e);
+});
+/**
+ * 将shell命令转化为Promise对象
+ * @param command shell指令
+ * @param args shell输入参数
+ */
+const cmd = (command, args) => new Promise(resolve => {
+    let timer, shell;
+    /**
+     * 清除进程并返回
+     * @param val 返回值
+     */
+    const r = (val) => {
+        clearTimeout(timer);
+        shell?.kill('SIGINT');
+        resolve(val);
+    };
+    try {
+        shell = (0, child_process_1.spawn)(command, args);
+        timer = setTimeout(() => {
+            shell.kill('SIGINT');
+        }, 60 * 1000);
+        let buf = '';
+        shell.stdout.on('data', data => {
+            buf += String(data);
+        });
+        shell.stdout.on('end', () => {
+            r(buf);
+        });
+        shell.on('exit', () => {
+            r(shell.killed ? undefined : '');
+        });
+        shell.on('error', () => {
+            r(undefined);
+        });
+    }
+    catch {
+        r(undefined);
+    }
+});
+exports.cmd = cmd;
+/**
+ * 比较两个文件
+ * @param oldStr 旧文本
+ * @param newStr 新文本
+ * @param uid 唯一标识
+ */
+const diff = async (oldStr, newStr, uid = -1) => {
+    if (oldStr === newStr) {
+        return;
+    }
+    const oldFile = `diffOld${uid}`, newFile = `diffNew${uid}`;
+    await Promise.all([fs.writeFile(oldFile, oldStr), fs.writeFile(newFile, newStr)]);
+    const stdout = await (0, exports.cmd)('git', [
+        'diff',
+        '--color-words=[\xC0-\xFF][\x80-\xBF]+|<?/?\\w+/?>?|[^[:space:]]',
+        '-U0',
+        '--no-index',
+        oldFile,
+        newFile,
+    ]);
+    await Promise.all([fs.unlink(oldFile), fs.unlink(newFile)]);
+    console.log(stdout?.split('\n').slice(4).join('\n'));
+};
+exports.diff = diff;

package/dist/util/lint.js

@@ -0,0 +1,48 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.generateForSelf = exports.generateForChild = void 0;
+const Parser = require("../index");
+/**
+ * 生成对于子节点的LintError对象
+ * @param child 子节点
+ * @param boundingRect 父节点的绝对定位
+ * @param msg 错误信息
+ * @param severity 严重程度
+ */
+const generateForChild = (child, boundingRect, msg, severity = 'error') => {
+    const index = child.getRelativeIndex(), { offsetHeight, offsetWidth, parentNode } = child, { top: offsetTop, left: offsetLeft } = parentNode.posFromIndex(index), { start } = boundingRect, { top, left } = 'top' in boundingRect ? boundingRect : child.getRootNode().posFromIndex(start), str = String(child), startIndex = start + index, startLine = top + offsetTop, startCol = offsetTop ? offsetLeft : left + offsetLeft;
+    return {
+        message: Parser.msg(msg),
+        severity,
+        startIndex,
+        endIndex: startIndex + str.length,
+        startLine,
+        endLine: startLine + offsetHeight - 1,
+        startCol,
+        endCol: offsetHeight === 1 ? startCol + offsetWidth : offsetWidth,
+        excerpt: str.slice(0, 50),
+    };
+};
+exports.generateForChild = generateForChild;
+/**
+ * 生成对于自己的LintError对象
+ * @param token 节点
+ * @param boundingRect 绝对定位
+ * @param msg 错误信息
+ * @param severity 严重程度
+ */
+const generateForSelf = (token, boundingRect, msg, severity = 'error') => {
+    const { start } = boundingRect, { offsetHeight, offsetWidth } = token, str = String(token), { top, left } = 'top' in boundingRect ? boundingRect : token.getRootNode().posFromIndex(start);
+    return {
+        message: Parser.msg(msg),
+        severity,
+        startIndex: start,
+        endIndex: start + str.length,
+        startLine: top,
+        endLine: top + offsetHeight - 1,
+        startCol: left,
+        endCol: offsetHeight === 1 ? left + offsetWidth : offsetWidth,
+        excerpt: str.slice(0, 50),
+    };
+};
+exports.generateForSelf = generateForSelf;

package/dist/util/string.js

@@ -0,0 +1,61 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.normalizeSpace = exports.print = exports.noWrap = exports.decodeHtml = exports.text = exports.escapeRegExp = exports.removeComment = exports.extUrlChar = exports.extUrlCharFirst = void 0;
+exports.extUrlCharFirst = '(?:\\[[\\da-f:.]+\\]|[^[\\]<>"\\0-\\x1F\\x7F\\p{Zs}\\uFFFD])';
+exports.extUrlChar = '(?:[^[\\]<>"\\0-\\x1F\\x7F\\p{Zs}\\uFFFD]|\\0\\d+[c!~]\\x7F)*';
+/**
+ * remove half-parsed comment-like tokens
+ * @param str 原字符串
+ */
+const removeComment = (str) => str.replace(/\0\d+c\x7F/gu, '');
+exports.removeComment = removeComment;
+/**
+ * escape special chars for RegExp constructor
+ * @param str RegExp source
+ */
+const escapeRegExp = (str) => str.replace(/[\\{}()|.?*+^$[\]]/gu, '\\$&');
+exports.escapeRegExp = escapeRegExp;
+/**
+ * extract effective wikitext
+ * @param childNodes a Token's contents
+ * @param separator delimiter between nodes
+ */
+const text = (childNodes, separator = '') => childNodes.map(child => typeof child === 'string' ? child : child.text()).join(separator);
+exports.text = text;
+/**
+ * decode HTML entities
+ * @param str 原字符串
+ */
+const decodeHtml = (str) => str.replace(/&#(\d+|x[\da-f]+);/giu, (_, code) => String.fromCodePoint(Number(`${code.toLowerCase().startsWith('x') ? '0' : ''}${code}`)));
+exports.decodeHtml = decodeHtml;
+/**
+ * escape newlines
+ * @param str 原字符串
+ */
+const noWrap = (str) => str.replaceAll('\n', '\\n');
+exports.noWrap = noWrap;
+/**
+ * 以HTML格式打印
+ * @param childNodes 子节点
+ * @param opt 选项
+ */
+const print = (childNodes, opt = {}) => {
+    const { pre = '', post = '', sep = '' } = opt;
+    return `${pre}${childNodes.map(child => child.print()).join(sep)}${post}`;
+};
+exports.print = print;
+/* NOT FOR BROWSER */
+/**
+ * convert newline in text nodes to single whitespace
+ * @param token 父节点
+ */
+const normalizeSpace = (token) => {
+    if (token) {
+        for (const child of token.childNodes) {
+            if (child.type === 'text') {
+                child.replaceData(child.data.replace(/\n+/gu, ' '));
+            }
+        }
+    }
+};
+exports.normalizeSpace = normalizeSpace;

package/errors/README

@@ -0,0 +1 @@
+这里记录解析失败时处于半解析状态的维基文本以及错误信息。

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "wikiparser-node",
-	"version": "1.2.0-b",
+	"version": "1.2.0",
 	"description": "A Node.js parser for MediaWiki markup with AST",
 	"keywords": [
 		"mediawiki",
@@ -14,54 +14,37 @@
 	"license": "GPL-3.0",
 	"author": "Bhsd",
 	"files": [
-		"/bundle/bundle.min.js",
-		"/extensions/dist/",
-		"/extensions/*.css",
-		"!/extensions/dist/gh-page.js",
-		"!/extensions/gh-page.css",
+		"/errors/README",
+		"/printed/README",
 		"/config/",
 		"/i18n/",
-		"!/config/.schema.json"
+		"/dist/"
 	],
-	"browser": "/bundle/bundle.min.js",
-	"main": "./bundle/bundle.js",
-	"types": "./extensions/typings.d.ts",
+	"main": "./dist/index.js",
 	"repository": {
 		"type": "git",
 		"url": "git+https://github.com/bhsd-harry/wikiparser-node.git"
 	},
 	"scripts": {
-		"prepublishOnly": "npm run build:prod && npm run build:ext",
-		"build:dev": "esbuild index.ts --bundle --target=es2018 --outfile=bundle/bundle.js",
-		"build:prod": "esbuild index.ts --bundle --minify --target=es2018 --outfile=bundle/bundle.min.js",
-		"build:core": "rm bundle/*.js; npm run build:dev; npm run build:prod; eslint --no-eslintrc -c .eslintrc.browser.json bundle/",
-		"build:ext": "bash build-ext.sh",
-		"build:test": "rm -rf dist/; tsc; gsed -i 's|../config/minimum|../../config/minimum|' dist/util/constants.js",
-		"build": "npm run build:core && npm run build:ext",
+		"toc": "node ./dist/bin/toc.js",
+		"prepublishOnly": "npm run build; rm dist/internal.js dist/base.js dist/[bmpu]*/*.d.ts; rm -r dist/test",
+		"build": "rm -rf dist/; tsc; grep -rl --include='*.d.ts' '@private' dist/ | xargs gsed -i '/@private/,+1d'",
 		"diff": "git diff --ignore-all-space --color-moved",
-		"lint:ts": "tsc --noEmit && tsc --noEmit --project extensions/tsconfig.json && eslint --cache .",
+		"lint:ts": "tsc --noEmit && eslint --cache .",
 		"lint:json": "ajv -s config/.schema.json -d 'config/*.json' --strict=true --strict-required=false",
-		"lint:css": "stylelint extensions/*.css",
-		"lint": "npm run lint:ts && npm run lint:json && npm run lint:css",
-		"server": "http-server .. -c-1 --cors &",
-		"test": "npm run server; open http://127.0.0.1:8080/wikiparser-node/index.html",
-		"test:end": "pkill -x http-server",
+		"lint": "npm run lint:ts && npm run lint:json",
+		"test": "node dist/test/test.js",
 		"test:real": "node dist/test/real.js",
-		"test:single": "node dist/test/single.js"
+		"test:single": "node dist/test/single.js; node --prof dist/test/single.js && node --prof-process isolate-0x*-v8.log > test/processed.txt && rm isolate-0x*-v8.log"
 	},
 	"devDependencies": {
-		"@bhsd/codemirror-mediawiki": "^2.0.13",
-		"@codemirror/lint": "^6.4.2",
-		"@codemirror/view": "^6.22.2",
 		"@cypress/request": "^3.0.1",
 		"@types/node": "^20.9.0",
 		"@types/request": "^2.48.12",
 		"@typescript-eslint/eslint-plugin": "^6.12.0",
 		"@typescript-eslint/parser": "^6.12.0",
 		"ajv-cli": "^5.0.0",
-		"esbuild": "^0.19.7",
 		"eslint": "^8.56.0",
-		"eslint-plugin-es-x": "^7.3.0",
 		"eslint-plugin-eslint-comments": "^3.2.0",
 		"eslint-plugin-jsdoc": "^47.0.2",
 		"eslint-plugin-json-es": "^1.5.7",
@@ -69,9 +52,6 @@
 		"eslint-plugin-promise": "^6.1.1",
 		"eslint-plugin-regexp": "^2.1.1",
 		"eslint-plugin-unicorn": "^50.0.1",
-		"http-server": "^14.1.0",
-		"stylelint": "^15.11.0",
-		"stylelint-config-recommended": "^13.0.0",
 		"typescript": "^5.1.6"
 	},
 	"engines": {

package/printed/README

@@ -0,0 +1 @@
+这里存放以 JSON 格式打印的 AST。