wikiparser-node 0.4.0 → 0.6.1

package/src/index.js

@@ -3,7 +3,7 @@
 /*
  * PHP解析器的步骤：
  * -1. 替换签名和`{{subst:}}`，参见Parser::preSaveTransform；这在revision中不可能保留，可以跳过
- * 0. 移除特定字符`\0`和`\x7f`，参见Parser::parse
+ * 0. 移除特定字符`\0`和`\x7F`，参见Parser::parse
  * 1. 注释/扩展标签（'<'相关），参见Preprocessor_Hash::buildDomTreeArrayFromText和Sanitizer::decodeTagAttributes
  * 2. 模板/模板变量/标题，注意rightmost法则，以及`-{`和`[[`可以破坏`{{`或`{{{`语法，
  *    参见Preprocessor_Hash::buildDomTreeArrayFromText
@@ -19,7 +19,7 @@
  */
 
 /*
- * \0\d+.\x7f标记Token：
+ * \0\d+.\x7F标记Token：
  * e: ExtToken
  * c: CommentToken、NoIncludeToken和IncludeToken
  * !: `{{!}}`专用
@@ -28,6 +28,7 @@
  * -: `{{!-}}`专用
  * +: `{{!!}}`专用
  * ~: `{{=}}`专用
+ * m: `{{fullurl:}}`、`{{canonicalurl:}}`或`{{filepath:}}`
  * t: ArgToken或TranscludeToken
  * h: HeadingToken
  * x: HtmlToken
@@ -58,20 +59,12 @@ class Token extends AstElement {
 	type = 'root';
 	#stage = 0; // 解析阶段，参见顶部注释。只对plain Token有意义。
 	#config;
-	// 这个数组起两个作用：1. 数组中的Token会在build时替换`/\0\d+.\x7f/`标记；2. 数组中的Token会依次执行parseOnce和build方法。
+	// 这个数组起两个作用：1. 数组中的Token会在build时替换`/\0\d+.\x7F/`标记；2. 数组中的Token会依次执行parseOnce和build方法。
 	#accum;
 	/** @type {Record<string, Ranges>} */ #acceptable;
 	#protectedChildren = new Ranges();
 	/** @type {boolean} */ #include;
 
-	/**
-	 * 保护部分子节点不被移除
-	 * @param {...string|number|Range} args 子节点范围
-	 */
-	#protectChildren = (...args) => {
-		this.#protectedChildren.push(...new Ranges(args));
-	};
-
 	/**
 	 * 将维基语法替换为占位符
 	 * @param {number} n 解析阶段
@@ -144,9 +137,36 @@ class Token extends AstElement {
 		throw new Error(`解析错误！未正确标记的 Token：${s}`);
 	});
 
+	/**
+	 * 将占位符替换为子Token
+	 * @complexity `n`
+	 */
+	#build = () => {
+		this.#stage = MAX_STAGE;
+		const {length, firstChild} = this,
+			str = String(firstChild);
+		if (length === 1 && firstChild.type === 'text' && str.includes('\0')) {
+			this.replaceChildren(...this.#buildFromStr(str));
+			this.normalize();
+			if (this.type === 'root') {
+				for (const token of this.#accum) {
+					token.getAttribute('build')();
+				}
+			}
+		}
+	};
+
+	/**
+	 * 保护部分子节点不被移除
+	 * @param {...string|number|Range} args 子节点范围
+	 */
+	#protectChildren = (...args) => {
+		this.#protectedChildren.push(...new Ranges(args));
+	};
+
 	/** 所有图片，包括图库 */
 	get images() {
-		return this.querySelectorAll('file, gallery-image');
+		return this.querySelectorAll('file, gallery-image, imagemap-image');
 	}
 
 	/** 所有内链、外链和自由外链 */
@@ -167,7 +187,7 @@ class Token extends AstElement {
 	constructor(wikitext, config = Parser.getConfig(), halfParsed = false, accum = [], acceptable = null) {
 		super();
 		if (typeof wikitext === 'string') {
-			this.appendChild(halfParsed ? wikitext : wikitext.replaceAll('\0', '').replaceAll('\x7F', ''));
+			this.insertAt(halfParsed ? wikitext : wikitext.replaceAll(/[\0\x7F]/gu, ''));
 		}
 		this.#config = config;
 		this.#accum = accum;
@@ -175,34 +195,6 @@ class Token extends AstElement {
 		accum.push(this);
 	}
 
-	/**
-	 * 深拷贝所有子节点
-	 * @complexity `n`
-	 * @returns {(AstText|Token)[]}
-	 */
-	cloneChildNodes() {
-		return this.childNodes.map(child => child.cloneNode());
-	}
-
-	/**
-	 * 深拷贝节点
-	 * @complexity `n`
-	 * @throws `Error` 未定义复制方法
-	 */
-	cloneNode() {
-		if (!this.isPlain()) {
-			throw new Error(`未定义 ${this.constructor.name} 的复制方法！`);
-		}
-		const cloned = this.cloneChildNodes();
-		return Parser.run(() => {
-			const token = new Token(undefined, this.#config, false, [], this.#acceptable);
-			token.type = this.type;
-			token.append(...cloned);
-			token.getAttribute('protectChildren')(...this.#protectedChildren);
-			return token;
-		});
-	}
-
 	/**
 	 * @override
 	 * @template {string} T
@@ -211,22 +203,24 @@ class Token extends AstElement {
 	 */
 	getAttribute(key) {
 		switch (key) {
-			case 'stage':
-				return this.#stage;
 			case 'config':
 				return structuredClone(this.#config);
 			case 'accum':
 				return this.#accum;
+			case 'parseOnce':
+				return this.#parseOnce;
+			case 'buildFromStr':
+				return this.#buildFromStr;
+			case 'build':
+				return this.#build;
+			case 'stage':
+				return this.#stage;
 			case 'acceptable':
 				return this.#acceptable ? {...this.#acceptable} : null;
 			case 'protectChildren':
 				return this.#protectChildren;
 			case 'protectedChildren':
 				return new Ranges(this.#protectedChildren);
-			case 'parseOnce':
-				return this.#parseOnce;
-			case 'buildFromStr':
-				return this.#buildFromStr;
 			case 'include': {
 				if (this.#include !== undefined) {
 					return this.#include;
@@ -252,16 +246,8 @@ class Token extends AstElement {
 	 * @template {string} T
 	 * @param {T} key 属性键
 	 * @param {TokenAttribute<T>} value 属性值
-	 * @throws `RangeError` 禁止手动指定私有属性
 	 */
 	setAttribute(key, value) {
-		if (key === 'include' || !Parser.running && (key === 'config' || key === 'accum')) {
-			throw new RangeError(`禁止手动指定私有的 #${key} 属性！`);
-		} else if (!Parser.debugging && (key === 'stage' || key === 'acceptable' || key === 'protectedChildren')
-			&& externalUse('setAttribute')
-		) {
-			throw new RangeError(`使用 ${this.constructor.name}.setAttribute 方法设置私有属性 #${key} 仅用于代码调试！`);
-		}
 		switch (key) {
 			case 'stage':
 				if (this.#stage === 0 && this.type === 'root') {
@@ -269,15 +255,6 @@ class Token extends AstElement {
 				}
 				this.#stage = value;
 				return this;
-			case 'config':
-				this.#config = value;
-				return this;
-			case 'accum':
-				this.#accum = value;
-				return this;
-			case 'protectedChildren':
-				this.#protectedChildren = value;
-				return this;
 			case 'acceptable': {
 				const /** @type {acceptable} */ acceptable = {};
 				if (value) {
@@ -308,36 +285,6 @@ class Token extends AstElement {
 		return this.constructor === Token;
 	}
 
-	/**
-	 * @override
-	 * @param {number} i 移除位置
-	 * @returns {Token}
-	 * @complexity `n`
-	 * @throws `Error` 不可移除的子节点
-	 */
-	removeAt(i) {
-		if (typeof i !== 'number') {
-			this.typeError('removeAt', 'Number');
-		}
-		const iPos = i < 0 ? i + this.childNodes.length : i;
-		if (!Parser.running) {
-			const protectedIndices = this.#protectedChildren.applyTo(this.childNodes);
-			if (protectedIndices.includes(iPos)) {
-				throw new Error(`${this.constructor.name} 的第 ${i} 个子节点不可移除！`);
-			} else if (this.#acceptable) {
-				const acceptableIndices = Object.fromEntries(
-						Object.entries(this.#acceptable)
-							.map(([str, ranges]) => [str, ranges.applyTo(this.childNodes.length - 1)]),
-					),
-					nodesAfter = i === -1 ? [] : this.childNodes.slice(i + 1);
-				if (nodesAfter.some(({constructor: {name}}, j) => !acceptableIndices[name].includes(i + j))) {
-					throw new Error(`移除 ${this.constructor.name} 的第 ${i} 个子节点会破坏规定的顺序！`);
-				}
-			}
-		}
-		return super.removeAt(i);
-	}
-
 	/**
 	 * @override
 	 * @template {string|Token} T
@@ -372,6 +319,45 @@ class Token extends AstElement {
 		return token;
 	}
 
+	/**
+	 * 规范化页面标题
+	 * @param {string} title 标题（含或不含命名空间前缀）
+	 * @param {number} defaultNs 命名空间
+	 */
+	normalizeTitle(title, defaultNs = 0, halfParsed = false) {
+		return Parser.normalizeTitle(title, defaultNs, this.#include, this.#config, halfParsed);
+	}
+
+	/**
+	 * @override
+	 * @param {number} i 移除位置
+	 * @returns {Token}
+	 * @complexity `n`
+	 * @throws `Error` 不可移除的子节点
+	 */
+	removeAt(i) {
+		if (!Number.isInteger(i)) {
+			this.typeError('removeAt', 'Number');
+		}
+		const iPos = i < 0 ? i + this.childNodes.length : i;
+		if (!Parser.running) {
+			const protectedIndices = this.#protectedChildren.applyTo(this.childNodes);
+			if (protectedIndices.includes(iPos)) {
+				throw new Error(`${this.constructor.name} 的第 ${i} 个子节点不可移除！`);
+			} else if (this.#acceptable) {
+				const acceptableIndices = Object.fromEntries(
+						Object.entries(this.#acceptable)
+							.map(([str, ranges]) => [str, ranges.applyTo(this.childNodes.length - 1)]),
+					),
+					nodesAfter = i === -1 ? [] : this.childNodes.slice(i + 1);
+				if (nodesAfter.some(({constructor: {name}}, j) => !acceptableIndices[name].includes(i + j))) {
+					throw new Error(`移除 ${this.constructor.name} 的第 ${i} 个子节点会破坏规定的顺序！`);
+				}
+			}
+		}
+		return super.removeAt(i);
+	}
+
 	/**
 	 * 替换为同类节点
 	 * @param {Token} token 待替换的节点
@@ -444,6 +430,120 @@ class Token extends AstElement {
 		throw new RangeError(`非法的标签名！${tagName}`);
 	}
 
+	/**
+	 * 创建纯文本节点
+	 * @param {string} data 文本内容
+	 */
+	createTextNode(data = '') {
+		return typeof data === 'string' ? new AstText(data) : this.typeError('createComment', 'String');
+	}
+
+	/**
+	 * 找到给定位置所在的节点
+	 * @param {number} index 位置
+	 */
+	caretPositionFromIndex(index) {
+		if (index === undefined) {
+			return undefined;
+		} else if (!Number.isInteger(index)) {
+			this.typeError('caretPositionFromIndex', 'Number');
+		}
+		const {length} = String(this);
+		if (index > length || index < -length) {
+			return undefined;
+		} else if (index < 0) {
+			index += length;
+		}
+		let child = this, // eslint-disable-line unicorn/no-this-assignment
+			acc = 0,
+			start = 0;
+		while (child.type !== 'text') {
+			const {childNodes} = child;
+			acc += child.getPadding();
+			for (let i = 0; acc <= index && i < childNodes.length; i++) {
+				const cur = childNodes[i],
+					{length: l} = String(cur);
+				acc += l;
+				if (acc >= index) {
+					child = cur;
+					acc -= l;
+					start = acc;
+					break;
+				}
+				acc += child.getGaps(i);
+			}
+			if (child.childNodes === childNodes) {
+				return {offsetNode: child, offset: index - start};
+			}
+		}
+		return {offsetNode: child, offset: index - start};
+	}
+
+	/**
+	 * 找到给定位置所在的节点
+	 * @param {number} x 列数
+	 * @param {number} y 行数
+	 */
+	caretPositionFromPoint(x, y) {
+		return this.caretPositionFromIndex(this.indexFromPos(y, x));
+	}
+
+	/**
+	 * 找到给定位置所在的最外层节点
+	 * @param {number} index 位置
+	 * @throws `Error` 不是根节点
+	 */
+	elementFromIndex(index) {
+		if (index === undefined) {
+			return undefined;
+		} else if (!Number.isInteger(index)) {
+			this.typeError('elementFromIndex', 'Number');
+		} else if (this.type !== 'root') {
+			throw new Error('elementFromIndex方法只可用于根节点！');
+		}
+		const {length} = String(this);
+		if (index > length || index < -length) {
+			return undefined;
+		} else if (index < 0) {
+			index += length;
+		}
+		const {childNodes} = this;
+		let acc = 0,
+			i = 0;
+		for (; acc < index && i < childNodes.length; i++) {
+			const {length: l} = String(childNodes[i]);
+			acc += l;
+		}
+		return childNodes[i && i - 1];
+	}
+
+	/**
+	 * 找到给定位置所在的最外层节点
+	 * @param {number} x 列数
+	 * @param {number} y 行数
+	 */
+	elementFromPoint(x, y) {
+		return this.elementFromIndex(this.indexFromPos(y, x));
+	}
+
+	/**
+	 * 找到给定位置所在的所有节点
+	 * @param {number} index 位置
+	 */
+	elementsFromIndex(index) {
+		const offsetNode = this.caretPositionFromIndex(index)?.offsetNode;
+		return offsetNode && [...offsetNode.getAncestors().reverse(), offsetNode];
+	}
+
+	/**
+	 * 找到给定位置所在的所有节点
+	 * @param {number} x 列数
+	 * @param {number} y 行数
+	 */
+	elementsFromPoint(x, y) {
+		return this.elementsFromIndex(this.indexFromPos(y, x));
+	}
+
 	/**
 	 * 判断标题是否是跨维基链接
 	 * @param {string} title 标题
@@ -453,12 +553,31 @@ class Token extends AstElement {
 	}
 
 	/**
-	 * 规范化页面标题
-	 * @param {string} title 标题（含或不含命名空间前缀）
-	 * @param {number} defaultNs 命名空间
+	 * 深拷贝所有子节点
+	 * @complexity `n`
+	 * @returns {(AstText|Token)[]}
 	 */
-	normalizeTitle(title, defaultNs = 0, halfParsed = false) {
-		return Parser.normalizeTitle(title, defaultNs, this.#include, this.#config, halfParsed);
+	cloneChildNodes() {
+		return this.childNodes.map(child => child.cloneNode());
+	}
+
+	/**
+	 * 深拷贝节点
+	 * @complexity `n`
+	 * @throws `Error` 未定义复制方法
+	 */
+	cloneNode() {
+		if (this.constructor !== Token) {
+			throw new Error(`未定义 ${this.constructor.name} 的复制方法！`);
+		}
+		const cloned = this.cloneChildNodes();
+		return Parser.run(() => {
+			const token = new Token(undefined, this.#config, false, [], this.#acceptable);
+			token.type = this.type;
+			token.append(...cloned);
+			token.getAttribute('protectChildren')(...this.#protectedChildren);
+			return token;
+		});
 	}
 
 	/**
@@ -499,7 +618,7 @@ class Token extends AstElement {
 	 * @complexity `n`
 	 */
 	section(n) {
-		return typeof n === 'number' ? this.sections()?.[n] : this.typeError('section', 'Number');
+		return Number.isInteger(n) ? this.sections()?.[n] : this.typeError('section', 'Number');
 	}
 
 	/**
@@ -521,27 +640,28 @@ class Token extends AstElement {
 		if (!parentNode) {
 			return undefined;
 		}
-		const {children} = parentNode,
-			index = children.indexOf(this);
+		const {childNodes} = parentNode,
+			index = childNodes.indexOf(this);
 		let i;
 		for (i = index - 1; i >= 0; i--) {
-			if (children[i].matches(`html${tag && '#'}${tag ?? ''}[selfClosing=false][closing=false]`)) {
+			const {type, name, selfClosing, closing} = childNodes[i];
+			if (type === 'html' && (!tag || name === tag) && selfClosing === false && closing === false) {
 				break;
 			}
 		}
 		if (i === -1) {
 			return parentNode.findEnclosingHtml(tag);
 		}
-		const opening = children[i],
-			{name} = opening;
-		for (i = index + 1; i < children.length; i++) {
-			if (children[i].matches(`html#${name}[selfClosing=false][closing=true]`)) {
+		const opening = childNodes[i];
+		for (i = index + 1; i < childNodes.length; i++) {
+			const {type, name, selfClosing, closing} = childNodes[i];
+			if (type === 'html' && name === opening.name && selfClosing === false && closing === true) {
 				break;
 			}
 		}
-		return i === children.length
+		return i === childNodes.length
 			? parentNode.findEnclosingHtml(tag)
-			: [opening, children[i]];
+			: [opening, childNodes[i]];
 	}
 
 	/**
@@ -589,29 +709,6 @@ class Token extends AstElement {
 		this.normalize();
 	}
 
-	/**
-	 * 将占位符替换为子Token
-	 * @complexity `n`
-	 */
-	build() {
-		if (!Parser.debugging && externalUse('build')) {
-			this.debugOnly('build');
-		}
-		this.#stage = MAX_STAGE;
-		const {childNodes: {length}, firstChild} = this,
-			str = String(firstChild);
-		if (length === 1 && firstChild.type === 'text' && str.includes('\0')) {
-			this.replaceChildren(...this.#buildFromStr(str));
-			this.normalize();
-			if (this.type === 'root') {
-				for (const token of this.#accum) {
-					token.build();
-				}
-			}
-		}
-		return this;
-	}
-
 	/** 生成部分Token的`name`属性 */
 	afterBuild() {
 		if (!Parser.debugging && externalUse('afterBuild')) {
@@ -630,16 +727,18 @@ class Token extends AstElement {
 	 * @param {boolean} include 是否嵌入
 	 */
 	parse(n = MAX_STAGE, include = false) {
-		if (typeof n !== 'number') {
+		if (!Number.isInteger(n)) {
 			this.typeError('parse', 'Number');
-		} else if (n < MAX_STAGE && !Parser.debugging && Parser.warning && externalUse('parse')) {
-			Parser.warn('指定解析层级的方法仅供熟练用户使用！');
 		}
 		this.#include = Boolean(include);
 		while (this.#stage < n) {
 			this.#parseOnce(this.#stage, include);
 		}
-		return n ? this.build().afterBuild() : this;
+		if (n) {
+			this.#build();
+			this.afterBuild();
+		}
+		return this;
 	}
 
 	/**
@@ -654,7 +753,9 @@ class Token extends AstElement {
 	/** 解析花括号 */
 	#parseBrackets() {
 		const parseBrackets = require('../parser/brackets');
-		this.setText(parseBrackets(String(this), this.#config, this.#accum));
+		const str = this.type === 'root' ? String(this) : `\0${String(this)}`,
+			parsed = parseBrackets(str, this.#config, this.#accum);
+		this.setText(this.type === 'root' ? parsed : parsed.slice(1));
 	}
 
 	/** 解析HTML标签 */
@@ -718,9 +819,13 @@ class Token extends AstElement {
 
 	/** 解析列表 */
 	#parseList() {
+		if (this.type === 'image-parameter') {
+			return;
+		}
 		const parseList = require('../parser/list');
 		const lines = String(this).split('\n');
-		for (let i = this.type === 'root' ? 0 : 1; i < lines.length; i++) {
+		let i = this.type === 'root' || this.type === 'ext-inner' && this.type === 'poem' ? 0 : 1;
+		for (; i < lines.length; i++) {
 			lines[i] = parseList(lines[i], this.#config, this.#accum);
 		}
 		this.setText(lines.join('\n'));

package/src/link/category.js

@@ -2,8 +2,7 @@
 
 const Title = require('../../lib/title'),
 	Parser = require('../..'),
-	LinkToken = require('.'),
-	Token = require('..');
+	LinkToken = require('.');
 
 /**
  * 分类
@@ -12,17 +11,13 @@ const Title = require('../../lib/title'),
 class CategoryToken extends LinkToken {
 	type = 'category';
 
-	setLangLink = undefined;
-	setFragment = undefined;
-	asSelfLink = undefined;
-	pipeTrick = undefined;
-
 	/** 分类排序关键字 */
 	get sortkey() {
-		return this.children[1]?.text()
-			?.replaceAll(/&#(\d+);/gu, /** @param {string} p */ (_, p) => String.fromCodePoint(Number(p)))
-			?.replaceAll(/&#x([\da-f]+);/giu, /** @param {string} p */ (_, p) => String.fromCodePoint(parseInt(p, 16)))
-			?.replaceAll('\n', '') ?? '';
+		return this.childNodes[1]?.text()?.replaceAll(
+			/&#(\d+|x[\da-f]+);|\n/gu,
+			/** @param {string} p */
+			(_, p) => p ? String.fromCodePoint(p[0] === 'x' ? parseInt(p.slice(1), 16) : Number(p)) : '',
+		);
 	}
 
 	set sortkey(text) {
@@ -34,10 +29,11 @@ class CategoryToken extends LinkToken {
 	 * @param {string|undefined} text 排序关键字
 	 * @param {Title} title 分类页面标题对象
 	 * @param {accum} accum
+	 * @param {string} delimiter `|`
 	 */
-	constructor(link, text, title, config = Parser.getConfig(), accum = []) {
-		super(link, text, title, config, accum);
-		this.seal(['setFragment', 'asSelfLink', 'pipeTrick'], true);
+	constructor(link, text, title, config = Parser.getConfig(), accum = [], delimiter = '|') {
+		super(link, text, title, config, accum, delimiter);
+		this.seal(['selfLink', 'interwiki', 'setLangLink', 'setFragment', 'asSelfLink', 'pipeTrick'], true);
 	}
 
 	/**