wikiparser-node 1.41.0 → 1.43.0

package/dist/mixin/elementLike.js

@@ -7,30 +7,41 @@ const selector_1 = require("../util/selector");
 const constants_1 = require("../util/constants");
 /** @ignore */
 const elementLike = (constructor) => {
-    LINT: {
-        class ElementLike extends constructor {
-            /* NOT FOR BROWSER */
-            get children() {
-                return this.childNodes.filter((child) => child.type !== 'text');
-            }
-            get firstElementChild() {
-                return this.childNodes.find((child) => child.type !== 'text');
-            }
-            get lastElementChild() {
-                return this.childNodes.findLast((child) => child.type !== 'text');
-            }
-            get childElementCount() {
-                return this.children.length;
-            }
-            /* NOT FOR BROWSER END */
-            #getCondition(selector) {
-                return (0, selector_1.getCondition)(selector, 
-                // eslint-disable-next-line unicorn/no-negated-condition
-                !('type' in this) ?
-                    undefined :
-                    this);
+    class ElementLike extends constructor {
+        /* NOT FOR BROWSER */
+        get children() {
+            return this.childNodes.filter((child) => child.type !== 'text');
+        }
+        get firstElementChild() {
+            return this.childNodes.find((child) => child.type !== 'text');
+        }
+        get lastElementChild() {
+            return this.childNodes.findLast((child) => child.type !== 'text');
+        }
+        get childElementCount() {
+            return this.children.length;
+        }
+        /* NOT FOR BROWSER END */
+        #getCondition(selector) {
+            return (0, selector_1.getCondition)(selector, 
+            // eslint-disable-next-line unicorn/no-negated-condition
+            !('type' in this) ?
+                undefined :
+                this);
+        }
+        closest(selector) {
+            const condition = this.#getCondition(selector);
+            let { parentNode } = this;
+            while (parentNode) {
+                if (condition(parentNode)) {
+                    return parentNode;
+                }
+                ({ parentNode } = parentNode);
             }
-            getElementBy(condition) {
+            return undefined;
+        }
+        getElementBy(condition) {
+            LINT: {
                 const stack = [...this.childNodes].reverse();
                 while (stack.length > 0) {
                     const child = stack.pop(), { type, childNodes } = child;
@@ -46,10 +57,12 @@ const elementLike = (constructor) => {
                 }
                 return undefined;
             }
-            querySelector(selector) {
-                return this.getElementBy(this.#getCondition(selector));
-            }
-            getElementsBy(condition) {
+        }
+        querySelector(selector) {
+            LINT: return this.getElementBy(this.#getCondition(selector));
+        }
+        getElementsBy(condition) {
+            LINT: {
                 const stack = [...this.childNodes].reverse(), descendants = [];
                 while (stack.length > 0) {
                     const child = stack.pop(), { type, childNodes } = child;
@@ -65,36 +78,36 @@ const elementLike = (constructor) => {
                 }
                 return descendants;
             }
-            querySelectorAll(selector) {
-                return this.getElementsBy(this.#getCondition(selector));
-            }
-            escape() {
-                LSP: {
-                    for (const child of this.childNodes) {
-                        child.escape();
-                    }
-                    /* NOT FOR BROWSER */
-                    this.detach?.();
+        }
+        querySelectorAll(selector) {
+            LINT: return this.getElementsBy(this.#getCondition(selector));
+        }
+        escape() {
+            LSP: {
+                for (const child of this.childNodes) {
+                    child.escape();
                 }
+                /* NOT FOR BROWSER */
+                this.detach?.();
             }
-            /* NOT FOR BROWSER */
-            getElementByTypes(types) {
-                const typeSet = new Set(types.split(',').map(str => str.trim()));
-                return this.getElementBy((({ type }) => typeSet.has(type)));
-            }
-            getElementById(id) {
-                return this.getElementBy((token => 'id' in token && token.id === id));
-            }
-            getElementsByClassName(className) {
-                return this.getElementsBy((token => 'classList' in token && token.classList.has(className)));
-            }
-            getElementsByTagName(tag) {
-                return this.getElementsBy((({ type, name }) => name === tag && (type === 'html' || type === 'ext')));
-            }
         }
-        (0, debug_1.mixin)(ElementLike, constructor);
-        return ElementLike;
+        /* NOT FOR BROWSER */
+        getElementByTypes(types) {
+            const typeSet = new Set(types.split(',').map(str => str.trim()));
+            return this.getElementBy((({ type }) => typeSet.has(type)));
+        }
+        getElementById(id) {
+            return this.getElementBy((token => 'id' in token && token.id === id));
+        }
+        getElementsByClassName(className) {
+            return this.getElementsBy((token => 'classList' in token && token.classList.has(className)));
+        }
+        getElementsByTagName(tag) {
+            return this.getElementsBy((({ type, name }) => name === tag && (type === 'html' || type === 'ext')));
+        }
     }
+    (0, debug_1.mixin)(ElementLike, constructor);
+    return ElementLike;
 };
 exports.elementLike = elementLike;
 constants_1.mixins['elementLike'] = __filename;

package/dist/parser/selector.js

@@ -163,8 +163,8 @@ const matches = (token, step, scope, has) => {
             }
         }
         else if (selector.length === 4) { // 情形2：属性选择器
-            const [key, equal, val = '', i] = selector, isAttr = typeof token.hasAttr === 'function' && typeof token.getAttr === 'function';
-            if (!(key in token || isAttr && token.hasAttr(key))) {
+            const [key, equal, val = '', i] = selector;
+            if (!(key in token || typeof token.hasAttr === 'function' && token.hasAttr(key))) {
                 return equal === '!=';
             }
             const v = toCase(val, i), thisVal = getAttr(token, key);
@@ -234,7 +234,7 @@ const matches = (token, step, scope, has) => {
                 for (; node; node = node.parentNode) {
                     const lang = node.attributes?.['lang'];
                     if (lang !== undefined) {
-                        return typeof lang === 'string' && regex.test(lang);
+                        return regex.test(lang);
                     }
                 }
                 return false;

package/dist/render/expand.js

@@ -186,7 +186,7 @@ const expand = (wikitext, page, callPage, config, include, context, now = index_
                 clean(accum, args);
                 return implicitNewLine(constants_1.functionHooks.get(name)(target, context || undefined), prev);
             }
-            else if (magicWords_1.expandedMagicWords.has(name)) {
+            try {
                 const result = (0, magicWords_1.expandMagicWord)(name, args.map(({ anon, name: key, value }) => anon ? value : `${key}=${value}`), callPage, config, now, accum);
                 if (result === false) {
                     return m;
@@ -194,7 +194,9 @@ const expand = (wikitext, page, callPage, config, include, context, now = index_
                 clean(accum, args);
                 return implicitNewLine(result, prev);
             }
-            return m;
+            catch {
+                return m;
+            }
         });
         plain.setText(expanded);
         if (plain.type === 'parameter-key') {

package/dist/render/extension.js

@@ -3,6 +3,14 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.renderExt = void 0;
 const constants_1 = require("../util/constants");
 const string_1 = require("../util/string");
+const sharable_1 = require("../util/sharable");
+/** @ignore */
+const getCiteNoteId = (i, refName) => `cite_note${refName ? `-${(0, string_1.sanitizeAttr)(refName, true)}` : ''}-${i}`, getCiteRefId = (i, count, refName) => `cite_ref-${refName ? `${(0, string_1.sanitizeAttr)(refName, true)}_${i}-${count - 1}` : i}`, updateRef = (ref, content, dir) => {
+    if (!ref.content) {
+        ref.content = content;
+        ref.dir = dir;
+    }
+};
 /**
  * 将扩展标签渲染为HTML
  * @param token 扩展标签节点
@@ -33,14 +41,16 @@ const renderExt = (token, opt) => {
                 .trim()}${padding}</div>`;
         }
         case 'gallery': {
-            const caption = firstChild.getAttrToken('caption'), perrow = parseInt(String(firstChild.getAttr('perrow'))), mode = firstChild.getAttr('mode'), { classList } = firstChild, nolines = typeof mode === 'string' && mode.toLowerCase() === 'nolines', padding = nolines ? 9 : 43;
+            const caption = firstChild.getAttrToken('caption'), perrow = parseInt(firstChild.getAttr('perrow') || ''), { classList } = firstChild, mode = firstChild.getAttr('mode')?.toLowerCase(), nolines = mode === 'nolines', padding = nolines ? 9 : 43;
             classList.add('gallery');
             if (nolines) {
                 classList.add('mw-gallery-nolines');
             }
+            else if (!mode || mode === 'traditional') {
+                classList.add('mw-gallery-traditional');
+            }
             if (perrow > 0) {
-                const style = firstChild.getAttr('style');
-                firstChild.setAttr('style', `max-width: ${(lastChild.widths + padding) * perrow}px;${typeof style === 'string' ? style : ''}`);
+                firstChild.setAttr('style', `max-width: ${(lastChild.widths + padding) * perrow}px;${firstChild.getAttr('style') || ''}`);
             }
             return `<ul${firstChild.toHtmlInternal()}>\n${caption
                 ? `\t<li class="gallerycaption">${caption.lastChild.toHtmlInternal({ nowrap: true })}</li>\n`
@@ -51,7 +61,7 @@ const renderExt = (token, opt) => {
             let html = lastChild.toHtmlInternal().trimEnd().replace(/^\n+/u, ''), lexer = firstChild.getAttr('lang');
             const dir = firstChild.getAttr('dir') === 'rtl' ? ' rtl' : 'ltr', isInline = firstChild.getAttr('enclose') === 'none' || firstChild.hasAttr('inline'), showLines = firstChild.hasAttr('line'), { classList } = firstChild;
             classList.add('mw-highlight');
-            if (lexer && lexer !== true) {
+            if (lexer) {
                 const { Prism, loadLanguage } = require('./syntaxhighlight');
                 lexer = lexer.toLowerCase();
                 if (Prism) {
@@ -64,7 +74,7 @@ const renderExt = (token, opt) => {
                     }
                     catch { }
                 }
-                const highlight = firstChild.getAttr('highlight'), lines = typeof highlight === 'string' && new Set(highlight.split(',').flatMap((str) => {
+                const highlight = firstChild.getAttr('highlight'), lines = highlight && new Set(highlight.split(',').flatMap((str) => {
                     const num = Number(str);
                     if (Number.isInteger(num) && num > 0) {
                         return num;
@@ -83,14 +93,14 @@ const renderExt = (token, opt) => {
                     if (linenos) {
                         const linelinks = firstChild.getAttr('linelinks'), startAttr = firstChild.getAttr('start');
                         lineReplace = '<span class="linenos" data-line="$1"></span>';
-                        if (startAttr && startAttr !== true) {
+                        if (startAttr) {
                             start = Number(startAttr);
                             if (!Number.isInteger(start) || start < 0) {
                                 start = 1;
                             }
                         }
-                        if (linelinks && linelinks !== true) {
-                            lineReplace = `<a href="#${linelinks}-$1">${lineReplace}</a>`;
+                        if (linelinks) {
+                            lineReplace = `<a href="#${(0, string_1.sanitizeId)(linelinks)}-$1">${lineReplace}</a>`;
                             begin = `${linelinks}-`;
                             end = '</span>';
                         }
@@ -145,6 +155,127 @@ const renderExt = (token, opt) => {
                 ? `<code${firstChild.toHtmlInternal()}>${html.trim().replaceAll('\n', ' ')}</code>`
                 : `<div${firstChild.toHtmlInternal()}>${html && `<pre>${(0, string_1.newline)(html)}</pre>`}</div>`;
         }
+        case 'ref': {
+            const refs = constants_1.states.get(token.getRootNode())?.refs;
+            if (!refs) {
+                return '';
+            }
+            const follow = firstChild.getAttr('follow') || '';
+            if (/^\d+$/u.test(follow)) {
+                return '';
+            }
+            let refName = firstChild.getAttr('name') || '';
+            if (!/\D/u.test(refName)) {
+                refName = '';
+            }
+            else if (refName && follow) {
+                return '';
+            }
+            let dir = firstChild.getAttr('dir')?.toLowerCase();
+            if (dir !== 'ltr' && dir !== 'rtl') {
+                dir = undefined;
+            }
+            const text = token.innerText?.trim(), references = token.closest('ext#references');
+            if (references) {
+                const { referencesGroup } = refs.get(references.getAttr('group') || '');
+                if (refName && text) {
+                    const ref = referencesGroup.find(({ name: n }) => n === refName);
+                    if (ref) {
+                        updateRef(ref, lastChild, dir);
+                    }
+                }
+                return '';
+            }
+            else if (!refName && !text || text && /<references\b[^>]*>/iu.test(text)) {
+                return '';
+            }
+            else if (text && /<ref\b[^>]*>/iu.test(text)) {
+                const inner = lastChild.cloneNode();
+                for (const ref of inner.querySelectorAll('ext#ref')) {
+                    ref.remove();
+                }
+                if (/<ref\b[^>]*>/iu.test(inner.toString())) {
+                    return '';
+                }
+            }
+            const group = firstChild.getAttr('group') || '';
+            if (!refs.has(group)) {
+                refs.set(group, { referencesGroup: [], follows: [] });
+            }
+            const { referencesGroup, follows } = refs.get(group);
+            if (follow) {
+                const ref = referencesGroup.find(({ name: n }) => n === follow);
+                if (ref) {
+                    if (ref.content) {
+                        ref.content.safeAppend([' ', ...lastChild.childNodes]);
+                    }
+                    else {
+                        ref.content = lastChild;
+                    }
+                }
+                else {
+                    refs.id++;
+                    follows.push({ content: lastChild });
+                }
+                return '';
+            }
+            let i = refName ? referencesGroup.findIndex(({ name: n }) => n === refName) : -1, count = 1, ref;
+            if (i === -1) {
+                i = referencesGroup.length;
+                ref = {
+                    ...refName && { name: refName },
+                    ...text && { content: lastChild },
+                    dir,
+                    count,
+                    id: ++refs.id,
+                };
+                referencesGroup.push(ref);
+            }
+            else {
+                ref = referencesGroup[i];
+                ref.count++;
+                ({ count } = ref);
+                if (text) {
+                    updateRef(ref, lastChild, dir);
+                }
+            }
+            return `<sup id="${getCiteRefId(ref.id, count, refName)}" class="reference"><a href="#${getCiteNoteId(ref.id, refName)}"><span class="cite-bracket">[</span>${group}${group && ' '}${i + 1}<span class="cite-bracket">]</span></a></sup>`;
+        }
+        case 'references': {
+            const refs = constants_1.states.get(token.getRootNode())?.refs;
+            if (!refs
+                || firstChild.childNodes.filter(node => node.is('ext-attr'))
+                    .some(({ name: key }) => !sharable_1.extAttrs['references'].has(key))) {
+                return '';
+            }
+            const group = firstChild.getAttr('group') || '';
+            if (!refs.has(group)) {
+                return '';
+            }
+            const html = lastChild.toHtmlInternal();
+            if (!refs.has(group)) { // 嵌套的`<ref>`
+                return html;
+            }
+            const { referencesGroup, follows } = refs.get(group);
+            if (referencesGroup.length === 0 && follows.length === 0) {
+                return '';
+            }
+            let ol = `<ol class="references"${group && ` data-mw-group="${group}"`}>`;
+            for (const { content } of follows) {
+                ol += `\n<p><span class="reference-text">${content.toHtmlInternal()}</span>\n</p>`;
+            }
+            for (let i = 0; i < referencesGroup.length; i++) {
+                const { content, count, dir, name: refName, id } = referencesGroup[i];
+                ol += `\n<li id="${getCiteNoteId(id, refName)}"${dir ? ` class="mw-cite-dir-${dir}"` : ''}><span class="mw-cite-backlink">${count === 1
+                    ? `<a href="#${getCiteRefId(id, 1, refName)}">↑</a>`
+                    : `↑${Array.from({ length: count }, (_, j) => ` <sup><a href="#${getCiteRefId(id, j + 1, refName)}">${i + 1}.${j}</a></sup>`).join('')}`}</span> <span class="reference-text">${content?.toHtmlInternal() ?? ''}</span>\n</li>`;
+            }
+            ol += '\n</ol>';
+            refs.delete(group);
+            return firstChild.getAttr('responsive') === '0'
+                ? ol
+                : `<div class="mw-references-wrap${referencesGroup.length > 10 ? ' mw-references-columns' : ''}">${ol}</div>`;
+        }
         default:
             return '';
     }

package/dist/render/html.js

@@ -2,7 +2,13 @@
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.toHtml = void 0;
 const constants_1 = require("../util/constants");
-const blockElems = 'table|h[1-6]|pre|p|[uod]l', antiBlockElems = 't[dh]';
+const blockElems = 'table|h[1-6]|pre|p|[uod]l', antiBlockElems = 't[dh]', allowed = new Set(['sup', 'sub', 'bdi', 'i', 'b', 's', 'strike', 'q']);
+const tocContainer = '<div id="toc" class="toc" role="navigation" aria-labelledby="mw-toc-heading">'
+    + '<input type="checkbox" role="button" id="toctogglecheckbox" class="toctogglecheckbox" style="display:none" />'
+    + '<div class="toctitle">'
+    + '<h2 id="mw-toc-heading">Contents</h2>'
+    + '<span class="toctogglespan"><label class="toctogglelabel" for="toctogglecheckbox"></label></span>'
+    + '</div>';
 // eslint-disable-next-line @typescript-eslint/no-unused-expressions
 /<(?:table|\/t[dh]|\/?(?:tr|caption|d[td]|li))\b/iu;
 const openRegex = new RegExp(String.raw `<(?:${blockElems}|\/${antiBlockElems}|\/?(?:tr|caption|d[td]|li))\b`, 'iu');
@@ -14,8 +20,70 @@ const closeRegex = new RegExp(String.raw `<(?:\/(?:${blockElems})|${antiBlockEle
  * @param token 展开后的节点
  */
 const toHtml = (token) => {
-    constants_1.states.set(token, { headings: new Set(), categories: new Set() });
+    constants_1.states.set(token, {
+        headings: new Set(),
+        categories: new Set(),
+        refs: Object.assign(new Map(), { id: 0 }),
+    });
+    // 前处理引用
+    const hasCite = token.getAttribute('config').ext.includes('references');
+    if (hasCite) {
+        token.append('\n', token.createElement('references', { selfClosing: true }));
+    }
+    // 处理目录
+    const tocSwitch = token.querySelector('double-underscore#toc'), forcetoc = tocSwitch || token.querySelector('double-underscore#forcetoc'), tocData = [];
+    if (forcetoc || !token.querySelector('double-underscore#notoc')) {
+        const headings = token.querySelectorAll('heading,html#h1,html#h2,html#h3,html#h4,html#h5,html#h6');
+        let firstHeading;
+        for (const heading of headings) {
+            if (heading.type === 'heading') {
+                tocData.push([heading.level, heading.getRenderedId(), heading.firstChild.toHtmlInternal().trim()]);
+                firstHeading ??= heading;
+            }
+            else {
+                const tocLine = heading.getTocLine();
+                if (tocLine) {
+                    tocData.push([Number(heading.name.slice(1)), ...tocLine]);
+                    firstHeading ??= heading;
+                }
+            }
+        }
+        const { length } = tocData;
+        if (forcetoc && length || length > 3) {
+            const levels = [], tocNumbers = [];
+            let toc = tocContainer, i = 0;
+            for (const [level, id, text] of tocData) {
+                const n = levels.length;
+                let j = levels.findIndex(l => l >= level), prefix;
+                if (j === -1) {
+                    j = n;
+                    prefix = '\n<ul>';
+                    levels.push(level);
+                    tocNumbers.push(1);
+                }
+                else {
+                    prefix = `${'</li>\n</ul>\n'.repeat(n - j - 1)}</li>`;
+                    levels.splice(j, Infinity, level);
+                    tocNumbers.length = j + 1;
+                    tocNumbers[j]++;
+                }
+                toc += `${prefix}\n<li class="toclevel-${j + 1} tocsection-${++i}"><a href="#${id}"><span class="tocnumber">${tocNumbers.join('.')}</span> <span class="toctext">${text.replaceAll(/<(\/?)([a-z]\w*)\b.*?>/gu, (_, slash, tag) => allowed.has(tag) ? `<${slash}${tag}>` : '').trim()}</span></a>`;
+            }
+            toc = i === 0 ? '' : `${toc}${'</li>\n</ul>\n'.repeat(levels.length)}</div>\n`;
+            if (tocSwitch) {
+                tocSwitch.tocData = toc;
+            }
+            else {
+                firstHeading.tocData = toc;
+            }
+        }
+    }
+    // 后处理引用
     const lines = token.toHtmlInternal().split('\n');
+    if (hasCite && lines.at(-1) === '') {
+        lines.pop();
+    }
+    // 处理正文
     let output = '', inBlockElem = false, pendingPTag = false, inBlockquote = false, lastParagraph = '';
     const /** @ignore */ closeParagraph = () => {
         if (lastParagraph) {
@@ -78,9 +146,10 @@ const toHtml = (token) => {
         }
     }
     output += closeParagraph();
+    let html = output.trimEnd();
+    // 处理分类
     const { categories } = constants_1.states.get(token);
     constants_1.states.delete(token);
-    let html = output.trimEnd();
     if (categories.size > 0) {
         html += `
 <div id="catlinks" class="catlinks"><div><a href="${token.normalizeTitle('Special:Categories', -1, { temporary: true }).getUrl()}" title="Special:Categories">Categories</a>: <ul>${[...categories].map(catlink => `<li>${catlink}</li>`).join('')}</div></div>`;