wikiparser-node 1.12.5-b → 1.12.6

package/dist/util/debug.js

@@ -0,0 +1,95 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.undo = exports.mixin = exports.emptyArray = exports.setChildNodes = exports.isToken = exports.Shadow = void 0;
+exports.Shadow = {
+    running: false,
+    /** @private */
+    run(callback) {
+        const { running } = this;
+        this.running = true;
+        try {
+            const { Token: AnyToken } = require('../src/index');
+            const result = callback();
+            if (result instanceof AnyToken && !result.getAttribute('built')) {
+                result.afterBuild();
+            }
+            this.running = running;
+            return result;
+        }
+        catch (e) {
+            this.running = running;
+            throw e;
+        }
+    },
+    rev: 0,
+};
+/**
+ * 是否是某一特定类型的节点
+ * @param type 节点类型
+ */
+const isToken = (type) => (node) => node.type === type;
+exports.isToken = isToken;
+/**
+ * 更新chldNodes
+ * @param parent 父节点
+ * @param position 子节点位置
+ * @param deleteCount 移除的子节点数量
+ * @param inserted 插入的子节点
+ */
+const setChildNodes = (parent, position, deleteCount, inserted = []) => {
+    const childNodes = [...parent.childNodes], removed = childNodes.splice(position, deleteCount, ...inserted);
+    parent.setAttribute('childNodes', childNodes);
+    for (const node of inserted) {
+        node.setAttribute('parentNode', parent);
+    }
+    /* NOT FOR BROWSER */
+    for (const node of removed) {
+        node.setAttribute('parentNode', undefined);
+    }
+    /* NOT FOR BROWSER END */
+    return removed;
+};
+exports.setChildNodes = setChildNodes;
+/**
+ * 生成一个指定长度的空数组
+ * @param n 数组长度
+ * @param callback 回调函数
+ */
+const emptyArray = (n, callback) => new Array(n).fill(undefined).map((_, i) => callback(i));
+exports.emptyArray = emptyArray;
+/**
+ * 同步混入的类名
+ * @param target 混入的目标
+ * @param source 混入的源
+ */
+const mixin = (target, source) => {
+    Object.defineProperty(target, 'name', { value: source.name });
+};
+exports.mixin = mixin;
+/* 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':
+            (0, exports.setChildNodes)(target, data.position, 0, [data.removed]);
+            break;
+        case 'insert':
+            (0, exports.setChildNodes)(target, data.position, 1);
+            break;
+        case 'replace':
+            (0, exports.setChildNodes)(target.parentNode, data.position, 1, [data.oldToken]);
+            break;
+        case 'text':
+            target.replaceData(data.oldText);
+            break;
+        default:
+            throw new RangeError(`Unable to undo events with an unknown type: ${type}`);
+    }
+};
+exports.undo = undo;

package/dist/util/diff.js

@@ -0,0 +1,83 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.info = exports.error = exports.diff = exports.cmd = void 0;
+const fs = require("fs/promises");
+const child_process_1 = require("child_process");
+const chalk = require("chalk");
+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) => {
+    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,
+    ]);
+    console.log(stdout?.split('\n').slice(4).join('\n'));
+    await Promise.all([fs.unlink(oldFile), fs.unlink(newFile)]);
+};
+exports.diff = diff;
+/** @implements */
+const error = (msg, ...args) => {
+    console.error(chalk.red(msg), ...args);
+};
+exports.error = error;
+/** @implements */
+const info = (msg, ...args) => {
+    console.info(chalk.green(msg), ...args);
+};
+exports.info = info;

package/dist/util/html.js

@@ -0,0 +1,146 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.font = exports.html = exports.getCommon = void 0;
+/**
+ * get common prefix length
+ * @param prefix
+ * @param lastPrefix
+ */
+const getCommon = (prefix, lastPrefix) => prefix.startsWith(lastPrefix) ? lastPrefix.length : [...lastPrefix].findIndex((ch, i) => ch !== prefix[i]);
+exports.getCommon = getCommon;
+/**
+ * get next list item
+ * @param char list syntax
+ * @param state
+ * @param state.dt whether a <dt> tag is open
+ */
+const nextItem = (char, state) => {
+    if (char === '*' || char === '#') {
+        return '</li>\n<li>';
+    }
+    const close = state.dt[0] ? '</dt>\n' : '</dd>\n';
+    if (char === ';') {
+        state.dt[0] = true;
+        return `${close}<dt>`;
+    }
+    state.dt[0] = false;
+    return `${close}<dd>`;
+};
+/**
+ * close list item
+ * @param chars list syntax
+ * @param state
+ * @param state.dt whether a <dt> tag is open
+ */
+const closeList = (chars, state) => {
+    let result = '';
+    for (let i = chars.length - 1; i >= 0; i--) {
+        const char = chars[i];
+        switch (char) {
+            case '*':
+            case '#':
+                result += `</li></${char === '*' ? 'ul' : 'ol'}>`;
+                break;
+            case ':':
+                result += `</${state.dt.shift() ? 'dt' : 'dd'}></dl>`;
+                break;
+            default:
+            //
+        }
+    }
+    return result;
+};
+/**
+ * open list item
+ * @param chars list syntax
+ * @param state
+ * @param state.dt whether a <dt> tag is open
+ */
+const openList = (chars, state) => {
+    let result = '';
+    for (const char of chars) {
+        switch (char) {
+            case '*':
+            case '#':
+                result += `<${char === '*' ? 'ul' : 'ol'}><li>`;
+                break;
+            case ':':
+                state.dt.unshift(false);
+                result += '<dl><dd>';
+                break;
+            default:
+                state.dt.unshift(true);
+                result += '<dl><dt>';
+        }
+    }
+    return result;
+};
+/**
+ * convert to HTML
+ * @param childNodes a Token's contents
+ * @param separator delimiter between nodes
+ * @param nowrap whether to normalize newlines
+ * @param nocc whether to perform language conversion
+ */
+const html = (childNodes, separator = '', nowrap, nocc) => {
+    let lastPrefix = '';
+    const results = [], state = { dt: [] };
+    for (let j = 0; j < childNodes.length; j++) {
+        const child = childNodes[j];
+        let result = child.toHtmlInternal(nowrap, nocc);
+        if (child.is('list-range')) {
+            result = result.trim();
+            const { previousSibling: { firstChild: { data } } } = child, prefix = data.trim(), prefix2 = prefix.replaceAll(';', ':'), commonPrefixLength = (0, exports.getCommon)(prefix2, lastPrefix);
+            let pre = closeList(lastPrefix.slice(commonPrefixLength), state);
+            if (prefix.length === commonPrefixLength) {
+                pre += nextItem(prefix.slice(-1), state);
+            }
+            else {
+                if (state.dt[0] && prefix[commonPrefixLength - 1] === ':') {
+                    pre += nextItem(':', state);
+                }
+                if (lastPrefix) {
+                    pre += '\n';
+                }
+                pre += openList(prefix.slice(commonPrefixLength), state);
+            }
+            result = pre + result;
+            let { nextSibling } = child;
+            while (nextSibling?.type === 'dd') {
+                const next = nextSibling.nextSibling;
+                for (let i = 0; i < nextSibling.indent; i++) {
+                    result += nextItem(':', state);
+                }
+                result += next.toHtmlInternal(nowrap).trim();
+                ({ nextSibling } = next);
+                j += 2;
+            }
+            if (nextSibling?.type === 'text'
+                && nextSibling.data === '\n'
+                && nextSibling.nextVisibleSibling?.type === 'list') {
+                j += 2;
+                lastPrefix = prefix2;
+            }
+            else {
+                lastPrefix = '';
+                result += closeList(prefix2, state);
+            }
+        }
+        results.push(result);
+    }
+    return results.join(separator);
+};
+exports.html = html;
+/**
+ * wrap text with <b> and <i> tags
+ * @param node
+ * @param strOrOmit
+ */
+const font = (node, strOrOmit) => {
+    const { font: { bold, italic } } = node, str = strOrOmit ?? '', i = str.indexOf('\n');
+    const wrap = /** @ignore */ (s) => s === ''
+        ? ''
+        : (italic ? '<i>' : '') + (bold ? '<b>' : '') + (s ?? '') + (bold ? '</b>' : '') + (italic ? '</i>' : '');
+    return i === -1 ? wrap(strOrOmit) : wrap(str.slice(0, i)) + str.slice(i);
+};
+exports.font = font;

package/dist/util/lint.js

@@ -0,0 +1,32 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.generateForSelf = exports.generateForChild = void 0;
+const rect_1 = require("../lib/rect");
+const index_1 = require("../index");
+/**
+ * 生成lint函数
+ * @param func lint函数
+ */
+const factory = (func) => (token, rect, rule, msg, severity = 'error') => {
+    const { start } = rect, { top, left } = rect instanceof rect_1.BoundingRect ? rect : new rect_1.BoundingRect(token, start), { offsetHeight, offsetWidth } = token, { startIndex, startLine, startCol } = func(token, start, top, left);
+    return {
+        rule,
+        message: index_1.default.msg(msg),
+        severity,
+        startIndex,
+        endIndex: startIndex + token.toString().length,
+        startLine,
+        endLine: startLine + offsetHeight - 1,
+        startCol,
+        endCol: offsetHeight === 1 ? startCol + offsetWidth : offsetWidth,
+    };
+};
+exports.generateForChild = factory((child, start, line, col) => {
+    const index = child.getRelativeIndex(), { top, left } = child.parentNode.posFromIndex(index);
+    return {
+        startIndex: start + index,
+        startLine: line + top,
+        startCol: top ? left : col + left,
+    };
+});
+exports.generateForSelf = factory((_, startIndex, startLine, startCol) => ({ startIndex, startLine, startCol }));

package/dist/util/string.js

@@ -0,0 +1,107 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.removeCommentLine = exports.newline = exports.sanitizeAlt = exports.sanitizeId = exports.sanitizeAttr = exports.sanitize = exports.normalizeSpace = exports.encode = exports.print = exports.escape = exports.noWrap = exports.decodeNumber = exports.decodeHtml = exports.text = exports.rawurldecode = exports.escapeRegExp = exports.removeComment = exports.tidy = exports.extUrlChar = exports.extUrlCharFirst = void 0;
+const commonExtUrlChar = String.raw `[^[\]<>"\0-\x1F\x7F\p{Zs}\uFFFD]`;
+exports.extUrlCharFirst = String.raw `(?:\[[\da-f:.]+\]|${commonExtUrlChar})`;
+exports.extUrlChar = String.raw `(?:${commonExtUrlChar}|\0\d+[cn!~]\x7F)*`;
+/**
+ * 生成正则替换函数
+ * @param regex 正则表达式
+ * @param replace 替换字符串或函数
+ */
+const factory = (regex, replace) => (str) => str.replace(regex, replace);
+/** 清理解析专用的不可见字符 */
+exports.tidy = factory(/[\0\x7F]|\r$/gmu, '');
+/** remove half-parsed comment-like tokens */
+exports.removeComment = factory(/\0\d+[cn]\x7F/gu, '');
+/** escape special chars for RegExp constructor */
+exports.escapeRegExp = factory(/[\\{}()|.?*+^$[\]]/gu, String.raw `\$&`);
+/**
+ * PHP的`rawurldecode`函数的JavaScript实现
+ * @param str 要解码的字符串
+ */
+const rawurldecode = (str) => decodeURIComponent(str.replace(/%(?![\da-f]{2})/giu, '%25'));
+exports.rawurldecode = rawurldecode;
+/**
+ * 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;
+const names = { lt: '<', gt: '>', lbrack: '[', rbrack: ']', lbrace: '{', rbrace: '}', nbsp: ' ', amp: '&', quot: '"' };
+/** decode HTML entities */
+exports.decodeHtml = factory(/&(?:#(\d+|[Xx][\da-fA-F]+)|([lLgG][tT]|[lr]brac[ke]|nbsp|amp|AMP|quot|QUOT));/gu, (_, code, name) => code
+    ? String.fromCodePoint(Number((/^x/iu.test(code) ? '0' : '') + code))
+    : names[name.toLowerCase()]);
+/** decode numbered HTML entities */
+exports.decodeNumber = factory(/&#(\d+|x[\da-f]+);/giu, (_, code) => String.fromCodePoint(Number((/^x/iu.test(code) ? '0' : '') + code)));
+/** escape newlines */
+exports.noWrap = factory(/\n/gu, String.raw `\n`);
+const entities = { '&': 'amp', '<': 'lt', '>': 'gt', '"': 'quot', '\n': '#10' };
+/**
+ * replace by HTML entities
+ * @param re regex
+ */
+const replaceEntities = (re) => factory(re, p => `&${entities[p]};`);
+/** escape HTML entities */
+exports.escape = replaceEntities(/[&<>]/gu);
+/**
+ * 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 */
+/** encode URI */
+exports.encode = factory(/[<>[\]#|=]+/gu, encodeURIComponent);
+/**
+ * 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;
+/** escape HTML entities */
+exports.sanitize = replaceEntities(/[<>]|&(?=amp(?!;))/giu);
+/** escape HTML entities in attributes */
+exports.sanitizeAttr = replaceEntities(/[<>"]/gu);
+/** escape HTML entities in heading id */
+exports.sanitizeId = replaceEntities(/["&]/gu);
+/**
+ * sanitize selected HTML attributes
+ * @param str attribute value
+ */
+const sanitizeAlt = (str) => str?.replace(/<\/?[a-z].*?>/gu, '').trim().replace(/\s+/gu, ' ').replaceAll('"', '&quot;');
+exports.sanitizeAlt = sanitizeAlt;
+/** escape newline */
+exports.newline = factory(/\n/gu, '&#10;');
+/**
+ * remove lines that only contain comments
+ * @param str
+ * @param standalone whether for a standalone document
+ */
+const removeCommentLine = (str, standalone) => {
+    const lines = str.split('\n'), { length } = lines;
+    if (!standalone && length < 3) {
+        return (0, exports.removeComment)(str);
+    }
+    const offset = standalone ? 0 : 1, end = length - offset;
+    return (0, exports.removeComment)([
+        ...lines.slice(0, offset),
+        ...lines.slice(offset, end).filter(line => !/^(?!\s*$)(?:\s|\0\d+c\x7F)*$/u.test(line)),
+        ...lines.slice(end),
+    ].join('\n'));
+};
+exports.removeCommentLine = removeCommentLine;

package/errors/README

@@ -0,0 +1,3 @@
+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。
+
+When parsing fails, the half-parsed wikitext and the error message are recorded here.

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "wikiparser-node",
-	"version": "1.12.5-b",
+	"version": "1.12.6",
 	"description": "A Node.js parser for MediaWiki markup with AST",
 	"keywords": [
 		"mediawiki",
@@ -14,50 +14,43 @@
 	"license": "GPL-3.0",
 	"author": "Bhsd",
 	"files": [
-		"/bundle/bundle.min.js",
-		"/extensions/dist/",
-		"/extensions/*.css",
-		"!/extensions/dist/gh-page.js",
-		"!/extensions/gh-page.css",
-		"!/extensions/dist/parserTests.js",
-		"!/extensions/test-page.css",
+		"/errors/README",
+		"/printed/README",
 		"/config/",
 		"/i18n/",
-		"!/config/.schema.json"
+		"/dist/",
+		"!/dist/[bt]*/"
 	],
-	"browser": "/bundle/bundle.min.js",
-	"main": "./bundle/bundle.min.js",
-	"types": "./extensions/typings.d.ts",
+	"main": "./dist/index.js",
+	"types": "./dist/index.d.ts",
 	"repository": {
 		"type": "git",
 		"url": "git+https://github.com/bhsd-harry/wikiparser-node.git"
 	},
 	"scripts": {
-		"prepublishOnly": "npm run build",
-		"build:core": "esbuild index.ts --bundle --minify --target=es2018 --sourcemap --outfile=bundle/bundle.min.js && eslint --no-eslintrc -c .eslintrc.browser.json bundle/bundle.min.js",
-		"build:ext": "bash build-ext.sh",
-		"build:test": "npm run build:core && tsc --project tsconfig.dist.json",
-		"build": "npm run build:core && npm run build:ext",
+		"toc": "node ./dist/bin/toc.js",
+		"declaration": "grep -rl --include='*.d.ts' '@private' dist/ | xargs bash sed.sh -i -E '/^\\s+\\/\\*\\* @private/,+1d'; grep -rl --include='*.d.ts' '/util/' dist/ | xargs bash sed.sh -i -E '/^import .+\\/util\\//d'; bash sed.sh -i -E 's/abstract (lint|print|text)\\b/\\1/' dist/lib/node.d.ts; node ./dist/bin/declaration.js",
+		"prepublishOnly": "npm run build:core",
+		"build:core": "bash build.sh",
+		"build": "npm run build:core && node ./dist/bin/parserTests.js",
 		"diff": "bash diff.sh",
 		"diff:stat": "f() { git diff --stat --ignore-all-space --color=always $1 $2 -- . ':!extensions/' ':!bin/' | grep '\\.ts'; }; f",
-		"lint:ts": "tsc --noEmit && tsc --noEmit --project extensions/tsconfig.json && tsc --noEmit --project extensions/tsconfig.codejar.json && eslint --cache .",
+		"lint:ts": "tsc --noEmit && eslint --cache .",
 		"lint:json": "v8r -s config/.schema.json config/*.json",
-		"lint:css": "stylelint extensions/*.css",
-		"lint": "npm run lint:ts && npm run lint:json && npm run lint:css",
-		"server": "npm run test:end; http-server .. -c-1 --cors &",
-		"test": " node --enable-source-maps dist/test/test.js",
-		"test:parser": "node --enable-source-maps dist/test/parserTests.js",
-		"test:server": "npm run build && bash sed.sh -i 's|importScripts.*|importScripts(\"http://127.0.0.1:8080/wikiparser-node/bundle/bundle.min.js\");|' extensions/dist/base.js; npm run server",
+		"lint": "npm run lint:ts && npm run lint:json",
+		"prof": "node dist/test/prof.js",
+		"test": "node dist/test/test.js",
+		"test:parser": "node dist/test/parserTests.js",
 		"test:end": "pkill -x http-server",
-		"test:real": "node --enable-source-maps dist/test/real.js"
+		"test:real": "node dist/test/real.js"
+	},
+	"dependencies": {
+		"chalk": "^4.1.2"
 	},
 	"devDependencies": {
-		"@bhsd/codemirror-mediawiki": "^2.14.0",
 		"@bhsd/common": "^0.0.0",
 		"@codemirror/lint": "^6.8.0",
 		"@types/node": "^20.11.6",
-		"@types/prismjs": "^1.26.3",
-		"chalk": "^4.1.2",
 		"codejar-async": "^4.2.0",
 		"monaco-editor": "^0.50.0",
 		"v8r": "^3.0.0"

package/printed/README

@@ -0,0 +1,3 @@
+这里存放以 JSON 格式打印的 AST。
+
+AST as JSON files are saved here.