wikiparser-node 1.3.8 → 1.3.9

package/dist/base.d.ts

@@ -33,6 +33,11 @@ export interface AstNode {
     /** 以HTML格式打印 */
     print(): string;
 }
+/** 类似HTMLElement */
+interface AstElement extends AstNode {
+    /** 保存为JSON */
+    json(): object;
+}
 export interface Parser {
     config: string | Config;
     i18n: string | Record<string, string> | undefined;
@@ -41,5 +46,6 @@ export interface Parser {
      * @param include 是否嵌入
      * @param maxStage 最大解析层级
      */
-    parse(wikitext: string, include?: boolean, maxStage?: number, config?: Config): AstNode;
+    parse(wikitext: string, include?: boolean, maxStage?: number, config?: Config): AstElement;
 }
+export {};

package/dist/index.js

@@ -162,8 +162,10 @@ const Parser = {
     },
     /** @implements */
     isInterwiki(title, { interwiki } = Parser.getConfig()) {
-        return new RegExp(`^(${interwiki.join('|')})\\s*:`, 'diu')
-            .exec(title.replace(/_/gu, ' ').replace(/^\s*:?\s*/u, ''));
+        return interwiki.length > 0
+            ? new RegExp(`^(${interwiki.join('|')})\\s*:`, 'diu')
+                .exec(title.replace(/_/gu, ' ').replace(/^\s*:?\s*/u, ''))
+            : null;
     },
     /** @implements */
     reparse(date) {

package/dist/lib/title.js

@@ -86,7 +86,7 @@ class Title {
         const m = title.split(':');
         if (m.length > 1) {
             const id = nsid[m[0].trim().toLowerCase()];
-            if (id !== undefined) {
+            if (id) {
                 ns = id;
                 title = m.slice(1).join(':').trim();
             }
@@ -112,7 +112,7 @@ class Title {
         }
         this.valid = Boolean(title
             || this.interwiki
-            || selfLink && this.fragment !== undefined) && !/\0\d+[eh!+-]\x7F|[<>[\]{}|]|%[\da-f]{2}/iu.test(title);
+            || selfLink && this.fragment !== undefined) && !/^:|\0\d+[eh!+-]\x7F|[<>[\]{}|]|%[\da-f]{2}/iu.test(title);
         this.main = title;
         Object.defineProperties(this, {
             valid: { writable: false },

package/dist/parser/braces.js

@@ -85,7 +85,7 @@ const parseBraces = (wikitext, config = Parser.getConfig(), accum = []) => {
                     }
                 }
                 catch (e) {
-                    if (e instanceof SyntaxError && e.message.startsWith('非法的模板名称：')) {
+                    if (e instanceof SyntaxError && e.message === '非法的模板名称') {
                         skip = true;
                     }
                     else {

package/dist/src/attribute.js

@@ -35,7 +35,7 @@ const commonHtmlAttrs = new Set([
     'itemref',
     'itemscope',
     'itemtype',
-]), blockAttrs = new Set(['align']), citeAttrs = new Set(['cite']), citeAndAttrs = new Set(['cite', 'datetime']), widthAttrs = new Set(['width']), tdAttrs = new Set(['align', 'valign', 'abbr', 'axis', 'headers', 'scope', 'rowspan', 'colspan', 'width', 'height', 'bgcolor']), typeAttrs = new Set(['type']), htmlAttrs = {
+]), blockAttrs = new Set(['align']), citeAttrs = new Set(['cite']), citeAndAttrs = new Set(['cite', 'datetime']), widthAttrs = new Set(['width']), obsoleteTdAttrs = new Set(['axis', 'align', 'bgcolor', 'height', 'width', 'valign']), tdAttrs = new Set([...obsoleteTdAttrs, 'abbr', 'headers', 'scope', 'rowspan', 'colspan']), typeAttrs = new Set(['type']), obsoleteTableAttrs = new Set(['summary', 'align', 'bgcolor', 'cellpadding', 'cellspacing', 'frame', 'rules', 'width']), brAttrs = new Set(['clear']), trAttrs = new Set(['bgcolor', 'align', 'valign']), htmlAttrs = {
     div: blockAttrs,
     h1: blockAttrs,
     h2: blockAttrs,
@@ -46,16 +46,16 @@ const commonHtmlAttrs = new Set([
     blockquote: citeAttrs,
     q: citeAttrs,
     p: blockAttrs,
-    br: new Set(['clear']),
+    br: brAttrs,
     pre: widthAttrs,
     ins: citeAndAttrs,
     del: citeAndAttrs,
     ul: typeAttrs,
     ol: new Set(['type', 'start', 'reversed']),
     li: new Set(['type', 'value']),
-    table: new Set(['summary', 'width', 'border', 'frame', 'rules', 'cellspacing', 'cellpadding', 'align', 'bgcolor']),
+    table: new Set([...obsoleteTableAttrs, 'border']),
     caption: blockAttrs,
-    tr: new Set(['bgcolor', 'align', 'valign']),
+    tr: trAttrs,
     td: tdAttrs,
     th: tdAttrs,
     img: new Set(['alt', 'src', 'width', 'height', 'srcset']),
@@ -152,7 +152,26 @@ const commonHtmlAttrs = new Set([
     + '|'
     + '(?:url|image(?:-set)?)\\s*\\('
     + '|'
-    + 'attr\\s*\\([^)]+[\\s,]url', 'u');
+    + 'attr\\s*\\([^)]+[\\s,]url', 'u'), obsoleteAttrs = {
+    table: obsoleteTableAttrs,
+    td: new Set([...obsoleteTdAttrs, 'scope']),
+    th: obsoleteTdAttrs,
+    br: brAttrs,
+    caption: blockAttrs,
+    div: blockAttrs,
+    hr: widthAttrs,
+    h1: blockAttrs,
+    h2: blockAttrs,
+    h3: blockAttrs,
+    h4: blockAttrs,
+    h5: blockAttrs,
+    h6: blockAttrs,
+    li: typeAttrs,
+    p: blockAttrs,
+    pre: widthAttrs,
+    tr: trAttrs,
+    ul: typeAttrs,
+};
 /**
  * 扩展和HTML标签属性
  * @classdesc `{childNodes: [AtomToken, Token|AtomToken]}`
@@ -293,6 +312,10 @@ class AttributeToken extends (0, fixed_1.fixed)(index_1.Token) {
             rect ??= { start, ...this.getRootNode().posFromIndex(start) };
             errors.push((0, lint_1.generateForChild)(firstChild, rect, 'illegal attribute name'));
         }
+        else if (obsoleteAttrs[tag]?.has(name)) {
+            rect ??= { start, ...this.getRootNode().posFromIndex(start) };
+            errors.push((0, lint_1.generateForChild)(firstChild, rect, 'obsolete attribute', 'warning'));
+        }
         else if (name === 'style' && typeof value === 'string' && insecureStyle.test(value)) {
             rect ??= { start, ...this.getRootNode().posFromIndex(start) };
             errors.push((0, lint_1.generateForChild)(lastChild, rect, 'insecure style'));

package/dist/src/heading.d.ts

@@ -5,7 +5,7 @@ import type { LintError } from '../base';
 declare const HeadingToken_base: (abstract new (...args: any[]) => {
     prependNewLine(): string;
     toString(omit?: Set<string> | undefined): string;
-    getAttribute<S extends string>(key: S): TokenAttributeGetter<S>;
+    getAttribute<S extends string>(key: S): TokenAttributeGetter<S>; /** 标题格式的等号 */
     text(): string;
     readonly length: number;
     lint(start?: number | undefined): Parser.LintError[];

package/dist/src/heading.js

@@ -80,18 +80,27 @@ class HeadingToken extends (0, sol_1.sol)((0, fixed_1.fixed)(index_1.Token)) {
     }
     /** @override */
     lint(start = this.getAbsoluteIndex()) {
-        const errors = super.lint(start), { firstChild } = this, innerStr = String(firstChild);
-        let refError;
+        const errors = super.lint(start), { firstChild, level } = this, innerStr = String(firstChild), quotes = firstChild.childNodes.filter((node) => node.type === 'quote'), boldQuotes = quotes.filter(({ bold }) => bold), italicQuotes = quotes.filter(({ italic }) => italic);
+        let rect;
         if (this.level === 1) {
-            refError = (0, lint_1.generateForChild)(firstChild, { start }, '<h1>');
-            errors.push(refError);
+            rect = { start, ...this.getRootNode().posFromIndex(start) };
+            errors.push((0, lint_1.generateForChild)(firstChild, rect, '<h1>'));
         }
         if (innerStr.startsWith('=') || innerStr.endsWith('=')) {
-            refError ??= (0, lint_1.generateForChild)(firstChild, { start }, '');
-            errors.push({ ...refError, message: Parser.msg('unbalanced "=" in a section header') });
+            rect ??= { start, ...this.getRootNode().posFromIndex(start) };
+            errors.push((0, lint_1.generateForChild)(firstChild, rect, Parser.msg('unbalanced $1 in a section header', '"="')));
         }
         if (this.closest('html-attrs, table-attrs')) {
-            errors.push({ ...(0, lint_1.generateForSelf)(this, { start }, ''), message: Parser.msg('section header in a HTML tag') });
+            rect ??= { start, ...this.getRootNode().posFromIndex(start) };
+            errors.push((0, lint_1.generateForSelf)(this, rect, 'section header in a HTML tag'));
+        }
+        if (boldQuotes.length % 2) {
+            rect ??= { start, ...this.getRootNode().posFromIndex(start) };
+            errors.push((0, lint_1.generateForChild)(boldQuotes[boldQuotes.length - 1], { ...rect, start: start + level, left: rect.left + level }, Parser.msg('unbalanced $1 in a section header', 'bold apostrophes')));
+        }
+        if (italicQuotes.length % 2) {
+            rect ??= { start, ...this.getRootNode().posFromIndex(start) };
+            errors.push((0, lint_1.generateForChild)(italicQuotes[italicQuotes.length - 1], { start: start + level }, Parser.msg('unbalanced $1 in a section header', 'italic apostrophes')));
         }
         return errors;
     }

package/dist/src/html.js

@@ -9,7 +9,33 @@ const fixed_1 = require("../mixin/fixed");
 const attributesParent_1 = require("../mixin/attributesParent");
 const Parser = require("../index");
 const index_1 = require("./index");
-const magicWords = new Set(['if', 'ifeq', 'ifexpr', 'ifexist', 'iferror', 'switch']);
+const magicWords = new Set(['if', 'ifeq', 'ifexpr', 'ifexist', 'iferror', 'switch']), formattingTags = new Set([
+    'b',
+    'big',
+    'center',
+    'cite',
+    'code',
+    'del',
+    'dfn',
+    'em',
+    'font',
+    'i',
+    'ins',
+    'kbd',
+    'mark',
+    'pre',
+    'q',
+    's',
+    'samp',
+    'small',
+    'strike',
+    'strong',
+    'sub',
+    'sup',
+    'tt',
+    'u',
+    'var',
+]), obsoleteTags = new Set(['strike', 'big', 'center', 'font', 'tt']);
 /**
  * HTML标签
  * @classdesc `{childNodes: [AttributesToken]}`
@@ -119,7 +145,7 @@ class HtmlToken extends (0, attributesParent_1.attributesParent)((0, fixed_1.fix
                 const { message: errorMsg } = e;
                 refError ??= (0, lint_1.generateForSelf)(this, { start }, '');
                 const [msg] = errorMsg.split(':'), error = { ...refError, message: Parser.msg(msg) };
-                if (msg === 'unclosed tag') {
+                if (msg === 'unclosed tag' && !formattingTags.has(this.name)) {
                     error.severity = 'warning';
                 }
                 else if (msg === 'unmatched closing tag') {
@@ -131,6 +157,14 @@ class HtmlToken extends (0, attributesParent_1.attributesParent)((0, fixed_1.fix
                 errors.push(error);
             }
         }
+        if (obsoleteTags.has(this.name)) {
+            refError ??= (0, lint_1.generateForSelf)(this, { start }, '');
+            errors.push({
+                ...refError,
+                message: Parser.msg('obsolete HTML tag'),
+                severity: 'warning',
+            });
+        }
         return errors;
     }
     /**

package/dist/src/link/base.js

@@ -150,6 +150,10 @@ class LinkBaseToken extends index_1.Token {
             rect ??= { start, ...this.getRootNode().posFromIndex(start) };
             errors.push((0, lint_1.generateForChild)(target, rect, 'useless fragment'));
         }
+        if (linkType === 'link' && this.closest('ext-link-text')) {
+            rect ??= { start, ...this.getRootNode().posFromIndex(start) };
+            errors.push((0, lint_1.generateForSelf)(this, rect, 'internal link in an external link'));
+        }
         return errors;
     }
     /** @private */

package/dist/src/nowiki/quote.d.ts

@@ -18,7 +18,13 @@ declare const QuoteToken_base: (abstract new (...args: any[]) => {
 /** `''`和`'''` */
 export declare class QuoteToken extends QuoteToken_base {
     readonly type = "quote";
+    /** 是否粗体 */
+    get bold(): boolean;
+    /** 是否斜体 */
+    get italic(): boolean;
     /** @override */
     lint(start?: number): LintError[];
+    /** @override */
+    json(): object;
 }
 export {};

package/dist/src/nowiki/quote.js

@@ -10,6 +10,14 @@ const base_1 = require("./base");
 // @ts-expect-error not implementing all abstract methods
 class QuoteToken extends (0, syntax_1.syntax)(base_1.NowikiBaseToken, /^(?:'{5}|'''?)$/u) {
     type = 'quote';
+    /** 是否粗体 */
+    get bold() {
+        return this.innerText.length !== 2;
+    }
+    /** 是否斜体 */
+    get italic() {
+        return this.innerText.length !== 3;
+    }
     /** @override */
     lint(start = this.getAbsoluteIndex()) {
         const { previousSibling, nextSibling } = this, message = Parser.msg('lonely "$1"', `'`), errors = [];
@@ -40,6 +48,14 @@ class QuoteToken extends (0, syntax_1.syntax)(base_1.NowikiBaseToken, /^(?:'{5}|
         }
         return errors;
     }
+    /** @override */
+    json() {
+        return {
+            ...super.json(),
+            bold: this.bold,
+            italic: this.italic,
+        };
+    }
 }
 exports.QuoteToken = QuoteToken;
 constants_1.classes['QuoteToken'] = __filename;

package/dist/src/transclude.js

@@ -90,9 +90,10 @@ class TranscludeToken extends index_1.Token {
         }
         if (this.type === 'template') {
             const name = (0, string_1.removeComment)((0, string_1.decodeHtml)(title)).split('#')[0].trim();
-            if (!name || /\0\d+[eh!+-]\x7F|[<>[\]{}\n]|%[\da-f]{2}/u.test(name)) {
+            if (!name || /^:[\s_]*:|\0\d+[eh!+-]\x7F|[<>[\]{}\n]|%[\da-f]{2}/iu.test(name)) {
                 accum.pop();
-                throw new SyntaxError(`非法的模板名称：${(0, string_1.noWrap)(name)}`);
+                Parser.debug(`非法的模板名称：${(0, string_1.noWrap)(name)}`);
+                throw new SyntaxError('非法的模板名称');
             }
             const token = new atom_1.AtomToken(title, 'template-name', config, accum, {
                 'Stage-2': ':', '!HeadingToken': '',

package/i18n/zh-hans.json

@@ -4,6 +4,7 @@
 	"additional \"|\" in a table cell": "表格单元格中多余的\"|\"",
 	"additional \"|\" in the link text": "链接文本中多余的\"|\"",
 	"attributes of a closing tag": "位于闭合标签的属性",
+	"bold": "粗体单引号",
 	"conflicting image $1 parameter": "冲突的图片$1参数",
 	"containing invalid attribute": "包含无效属性",
 	"content to be moved out from the table": "将被移出表格的内容",
@@ -19,6 +20,7 @@
 	"illegal attribute name": "非法的属性名",
 	"illegal module name": "非法的模块名称",
 	"insecure style": "不安全的样式",
+	"internal link in an external link": "外链中的内链",
 	"invalid content in <$1>": "<$1>内的无效内容",
 	"invalid conversion flag": "无效的转换标记",
 	"invalid gallery image": "无效的图库图片",
@@ -27,17 +29,20 @@
 	"invalid parameter of $1": "$1的无效参数",
 	"invalid self-closing tag": "无效自封闭标签",
 	"invisible content inside triple brackets": "三重括号内的不可见部分",
+	"italic": "斜体单引号",
 	"lonely \"$1\"": "孤立的\"$1\"",
 	"missing module function": "缺少模块函数",
 	"missing module name": "缺少模块名称",
 	"nonzero tabindex": "不为0的tabindex",
 	"nothing should be in <$1>": "<$1>标签内不应有任何内容",
+	"obsolete attribute": "过时的属性",
+	"obsolete HTML tag": "过时的HTML标签",
 	"quotes": "引号",
 	"section header in a HTML tag": "HTML标签属性中的段落标题",
 	"table": "表格",
 	"tag that is both closing and self-closing": "同时闭合和自封闭的标签",
 	"template in an internal link target": "内链目标包含模板",
-	"unbalanced \"=\" in a section header": "段落标题中不平衡的\"=\"",
+	"unbalanced $1 in a section header": "段落标题中不平衡的$1",
 	"unclosed $1": "未闭合的$1",
 	"unclosed tag": "未闭合的标签",
 	"unescaped query string in an anonymous parameter": "匿名参数中未转义的查询参数",

package/i18n/zh-hant.json

@@ -4,6 +4,7 @@
 	"additional \"|\" in a table cell": "表哥單元格中多餘的\"|\"",
 	"additional \"|\" in the link text": "連結文本中多餘的\"|\"",
 	"attributes of a closing tag": "位於閉合標籤的屬性",
+	"bold apostrophes": "粗體撇號",
 	"conflicting image $1 parameter": "衝突的圖片$1參數",
 	"containing invalid attribute": "包含無效屬性",
 	"content to be moved out from the table": "將被移出表格的內容",
@@ -19,6 +20,7 @@
 	"illegal attribute name": "非法的屬性名",
 	"illegal module name": "非法的模組名稱",
 	"insecure style": "不安全的樣式",
+	"internal link in an external link": "外部連結中的內部連結",
 	"invalid content in <$1>": "<$1>內的無效內容",
 	"invalid conversion flag": "無效的轉換標記",
 	"invalid gallery image": "無效的圖庫圖片",
@@ -27,17 +29,20 @@
 	"invalid parameter of $1": "$1的無效參數",
 	"invalid self-closing tag": "無效自封閉標籤",
 	"invisible content inside triple brackets": "三重括號內的不可見部分",
+	"italic apostrophes": "斜體撇號",
 	"lonely \"$1\"": "孤立的\"$1\"",
 	"missing module function": "缺少模組函式",
 	"missing module name": "缺少模組名稱",
 	"nonzero tabindex": "不為0的tabindex",
 	"nothing should be in <$1>": "<$1>標籤內不應有任何內容",
+	"obsolete attribute": "過時的屬性",
+	"obsolete HTML tag": "過時的HTML標籤",
 	"quotes": "引號",
 	"section header in a HTML tag": "HTML標籤屬性中的段落標題",
 	"table": "表格",
 	"tag that is both closing and self-closing": "同時閉合和自封閉的標籤",
 	"template in an internal link target": "內部連結目標包含模板",
-	"unbalanced \"=\" in a section header": "段落標題中不平衡的\"=\"",
+	"unbalanced $1 in a section header": "段落標題中不平衡的$1",
 	"unclosed $1": "未閉合的$1",
 	"unclosed tag": "未閉合的標籤",
 	"unescaped query string in an anonymous parameter": "匿名參數中未轉義的查詢參數",

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "wikiparser-node",
-	"version": "1.3.8",
+	"version": "1.3.9",
 	"description": "A Node.js parser for MediaWiki markup with AST",
 	"keywords": [
 		"mediawiki",
@@ -35,6 +35,7 @@
 		"lint:json": "ajv -s config/.schema.json -d 'config/*.json' --strict=true --strict-required=false",
 		"lint": "npm run lint:ts && npm run lint:json",
 		"test": "node dist/test/test.js",
+		"test:end": "pkill -x http-server",
 		"test:real": "node dist/test/real.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"
 	},