wikiparser-node 1.8.0-b → 1.8.0

package/dist/lib/element.js

@@ -0,0 +1,597 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.AstElement = void 0;
+const fs = require("fs");
+const path = require("path");
+const string_1 = require("../util/string");
+const debug_1 = require("../util/debug");
+const constants_1 = require("../util/constants");
+const ranges_1 = require("./ranges");
+const title_1 = require("./title");
+const node_1 = require("./node");
+/* NOT FOR BROWSER */
+/**
+ * optionally convert to lower cases
+ * @param val 属性值
+ * @param i 是否对大小写不敏感
+ */
+const toCase = (val, i) => i ? val.toLowerCase() : val;
+/**
+ * 检查某个下标是否符合表达式
+ * @param str 表达式
+ * @param i 待检查的下标
+ */
+const nth = (str, i) => new ranges_1.Ranges(str.split(',')).applyTo(i + 1).includes(i);
+/**
+ * 检测:lang()伪选择器
+ * @param node 节点
+ * @param node.attributes 节点属性
+ * @param regex 语言正则
+ */
+const matchesLang = ({ attributes }, regex) => {
+    const lang = attributes?.['lang'];
+    return typeof lang === 'string' && regex.test(lang);
+};
+const primitives = new Set(['string', 'number', 'boolean', 'undefined']);
+/* NOT FOR BROWSER END */
+/** 类似HTMLElement */
+class AstElement extends node_1.AstNode {
+    /** 子节点总数 */
+    get length() {
+        return this.childNodes.length;
+    }
+    /* NOT FOR BROWSER */
+    /** 全部非文本子节点 */
+    get children() {
+        return this.childNodes.filter((child) => child.type !== 'text');
+    }
+    /** 首位非文本子节点 */
+    get firstElementChild() {
+        return this.childNodes.find((child) => child.type !== 'text');
+    }
+    /** 末位非文本子节点 */
+    get lastElementChild() {
+        return this.children[this.childElementCount - 1];
+    }
+    /** 非文本子节点总数 */
+    get childElementCount() {
+        return this.children.length;
+    }
+    /** 父节点 */
+    get parentElement() {
+        return this.parentNode;
+    }
+    /** AstElement.prototype.text()的getter写法 */
+    get outerText() {
+        return this.text();
+    }
+    /** 不可见 */
+    get hidden() {
+        return this.text() === '';
+    }
+    /** 内部高度 */
+    get clientHeight() {
+        const { innerText } = this;
+        return typeof innerText === 'string' ? innerText.split('\n').length : undefined;
+    }
+    /** 内部宽度 */
+    get clientWidth() {
+        const { innerText } = this;
+        if (typeof innerText === 'string') {
+            const lines = innerText.split('\n');
+            return lines[lines.length - 1].length;
+        }
+        return undefined;
+    }
+    constructor() {
+        super();
+        this.seal('name');
+    }
+    /* NOT FOR BROWSER END */
+    /**
+     * 可见部分
+     * @param separator 子节点间的连接符
+     */
+    text(separator) {
+        return (0, string_1.text)(this.childNodes, separator);
+    }
+    /** 合并相邻的文本子节点 */
+    normalize() {
+        const childNodes = [...this.childNodes];
+        /**
+         * 移除子节点
+         * @param i 移除位置
+         */
+        const remove = (i) => {
+            childNodes[i].setAttribute('parentNode', undefined);
+            childNodes.splice(i, 1);
+        };
+        for (let i = childNodes.length - 1; i >= 0; i--) {
+            const { type, data } = childNodes[i];
+            if (type !== 'text' || this.getGaps(i - 1)) {
+                //
+            }
+            else if (data === '') {
+                remove(i);
+                /* NOT FOR BROWSER */
+            }
+            else {
+                const prev = childNodes[i - 1];
+                if (prev?.type === 'text') {
+                    prev.setAttribute('data', prev.data + data);
+                    remove(i);
+                }
+                /* NOT FOR BROWSER END */
+            }
+        }
+        this.setAttribute('childNodes', childNodes);
+    }
+    /**
+     * 移除子节点
+     * @param i 移除位置
+     */
+    removeAt(i) {
+        /* NOT FOR BROWSER */
+        this.verifyChild(i);
+        /* NOT FOR BROWSER END */
+        return (0, debug_1.setChildNodes)(this, i, 1)[0];
+    }
+    /**
+     * 插入子节点
+     * @param node 待插入的子节点
+     * @param i 插入位置
+     * @throws `RangeError` 不能插入祖先或子节点
+     */
+    insertAt(node, i = this.length) {
+        /* NOT FOR BROWSER */
+        if (node.contains(this)) {
+            throw new RangeError('不能插入祖先节点！');
+        }
+        if (this.childNodes.includes(node)) {
+            throw new RangeError('不能插入子节点！');
+        }
+        this.verifyChild(i, 1);
+        node.parentNode?.removeChild(node);
+        /* NOT FOR BROWSER END */
+        (0, debug_1.setChildNodes)(this, i, 0, [node]);
+        return node;
+    }
+    /**
+     * 将选择器转化为类型谓词
+     * @param selector 选择器
+     */
+    #getCondition(selector) {
+        /* NOT FOR BROWSER */
+        if (/[^a-z\-,\s]/u.test(selector)) {
+            const { parseSelector } = require('../parser/selector');
+            const stack = parseSelector(selector);
+            return (token => stack.some(copy => token.#matchesArray(copy)));
+        }
+        /* NOT FOR BROWSER END */
+        const types = new Set(selector.split(',').map(str => str.trim()));
+        return (({ type }) => types.has(type));
+    }
+    /**
+     * 最近的祖先节点
+     * @param selector 选择器
+     */
+    closest(selector) {
+        const condition = this.#getCondition(selector);
+        let { parentNode } = this;
+        while (parentNode) {
+            if (condition(parentNode)) {
+                return parentNode;
+            }
+            ({ parentNode } = parentNode);
+        }
+        return undefined;
+    }
+    /**
+     * 符合条件的第一个后代节点
+     * @param condition 条件
+     */
+    #getElementBy(condition) {
+        for (const child of this.childNodes) {
+            if (child.type === 'text') {
+                continue;
+            }
+            else if (condition(child)) {
+                return child;
+            }
+            const descendant = child.#getElementBy(condition);
+            if (descendant) {
+                return descendant;
+            }
+        }
+        return undefined;
+    }
+    /**
+     * 符合选择器的第一个后代节点
+     * @param selector 选择器
+     */
+    querySelector(selector) {
+        const condition = this.#getCondition(selector);
+        return this.#getElementBy(condition);
+    }
+    /**
+     * 符合条件的所有后代节点
+     * @param condition 条件
+     */
+    #getElementsBy(condition) {
+        const descendants = [];
+        for (const child of this.childNodes) {
+            if (child.type === 'text') {
+                continue;
+            }
+            else if (condition(child)) {
+                descendants.push(child);
+            }
+            descendants.push(...child.#getElementsBy(condition));
+        }
+        return descendants;
+    }
+    /**
+     * 符合选择器的所有后代节点
+     * @param selector 选择器
+     */
+    querySelectorAll(selector) {
+        const condition = this.#getCondition(selector);
+        return this.#getElementsBy(condition);
+    }
+    /**
+     * 在末尾批量插入子节点
+     * @param elements 插入节点
+     */
+    append(...elements) {
+        for (const element of elements) {
+            this.insertAt(element);
+        }
+    }
+    /**
+     * 批量替换子节点
+     * @param elements 新的子节点
+     */
+    replaceChildren(...elements) {
+        for (let i = this.length - 1; i >= 0; i--) {
+            this.removeAt(i);
+        }
+        this.append(...elements);
+    }
+    /**
+     * 修改文本子节点
+     * @param str 新文本
+     * @param i 子节点位置
+     * @throws `RangeError` 对应位置的子节点不是文本节点
+     */
+    setText(str, i = 0) {
+        i += i < 0 ? this.length : 0;
+        /* NOT FOR BROWSER */
+        this.verifyChild(i);
+        /* NOT FOR BROWSER END */
+        const oldText = this.childNodes[i];
+        if (oldText.type === 'text') { // eslint-disable-line @typescript-eslint/no-unnecessary-condition
+            const { data } = oldText;
+            oldText.replaceData(str);
+            return data;
+        }
+        /* NOT FOR BROWSER */
+        throw new RangeError(`第 ${i} 个子节点是 ${oldText.constructor.name}！`);
+    }
+    /** @private */
+    toString(separator = '') {
+        return this.childNodes.map(String).join(separator);
+    }
+    /**
+     * @override
+     * @param start
+     * @param re 用于`AstText`后代节点的正则表达式
+     */
+    lint(start = this.getAbsoluteIndex(), re) {
+        const errors = [];
+        for (let i = 0, cur = start + this.getAttribute('padding'); i < this.length; i++) {
+            const child = this.childNodes[i];
+            errors.push(...child.lint(cur, re));
+            cur += String(child).length + this.getGaps(i);
+        }
+        return errors;
+    }
+    /**
+     * @override
+     * @param opt 选项
+     */
+    print(opt = {}) {
+        return String(this) ? `<span class="wpb-${opt.class ?? this.type}">${(0, string_1.print)(this.childNodes, opt)}</span>` : '';
+    }
+    /**
+     * 保存为JSON
+     * @param file 文件名
+     * @param start
+     */
+    json(file, start = this.getAbsoluteIndex()) {
+        const json = {
+            ...this,
+            range: [start, start + String(this).length],
+            childNodes: [],
+        };
+        for (let i = 0, cur = start + this.getAttribute('padding'); i < this.length; i++) {
+            const child = this.childNodes[i], { length } = String(child);
+            if (child.type === 'text') {
+                json.childNodes.push({ data: child.data, range: [cur, cur + length] });
+            }
+            else {
+                json.childNodes.push(child.json(undefined, cur));
+            }
+            cur += length + this.getGaps(i);
+        }
+        /* NOT FOR BROWSER */
+        if (typeof file === 'string') {
+            fs.writeFileSync(path.join(__dirname.slice(0, -4), '..', 'printed', file + (file.endsWith('.json') ? '' : '.json')), JSON.stringify(json, null, 2));
+        }
+        /* NOT FOR BROWSER END */
+        return json;
+    }
+    /* NOT FOR BROWSER */
+    /** 是否受保护。保护条件来自Token，这里仅提前用于:required和:optional伪选择器。 */
+    #isProtected() {
+        const { parentNode } = this;
+        if (!parentNode) {
+            return undefined;
+        }
+        const { childNodes, fixed } = parentNode, protectedIndices = parentNode.getAttribute('protectedChildren').applyTo(childNodes);
+        return fixed || protectedIndices.includes(childNodes.indexOf(this));
+    }
+    /**
+     * 检查是否符合属性选择器
+     * @param key 属性键
+     * @param equal 比较符
+     * @param val 属性值
+     * @param i 是否对大小写不敏感
+     * @throws `RangeError` 复杂属性不能用于选择器
+     */
+    #matchesAttr(key, equal, val = '', i) {
+        const isAttr = typeof this.hasAttr === 'function' && typeof this.getAttr === 'function';
+        if (!(key in this) && (!isAttr || !this.hasAttr(key))) {
+            return equal === '!=';
+        }
+        const v = toCase(val, i);
+        let thisVal = this.getAttribute(key);
+        if (isAttr) {
+            const attr = this.getAttr(key);
+            if (attr !== undefined) {
+                thisVal = attr === true ? '' : attr;
+            }
+        }
+        if (!equal) {
+            return thisVal !== undefined && thisVal !== false;
+        }
+        else if (thisVal instanceof RegExp) {
+            thisVal = thisVal.source;
+        }
+        if (equal === '~=') {
+            const thisVals = typeof thisVal === 'string' ? thisVal.split(/\s/u) : thisVal;
+            return Boolean(thisVals?.[Symbol.iterator])
+                && [...thisVals].some(w => typeof w === 'string' && toCase(w, i) === v);
+        }
+        else if (!primitives.has(typeof thisVal) && !(thisVal instanceof title_1.Title)) {
+            throw new RangeError(`复杂属性 ${key} 不能用于选择器！`);
+        }
+        const stringVal = toCase(String(thisVal), i);
+        switch (equal) {
+            case '|=':
+                return stringVal === v || stringVal.startsWith(`${v}-`);
+            case '^=':
+                return stringVal.startsWith(v);
+            case '$=':
+                return stringVal.endsWith(v);
+            case '*=':
+                return stringVal.includes(v);
+            case '!=':
+                return stringVal !== v;
+            default: // `=`
+                return stringVal === v;
+        }
+    }
+    /**
+     * 检查是否符合解析后的选择器，不含节点关系
+     * @param step 解析后的选择器
+     * @throws `SyntaxError` 错误的正则伪选择器
+     * @throws `SyntaxError` 未定义的伪选择器
+     */
+    #matches(step) {
+        const { parentNode, type, name, childNodes, link } = this, children = parentNode?.children, childrenOfType = children?.filter(({ type: t }) => t === type), siblingsCount = children?.length ?? 1, siblingsCountOfType = childrenOfType?.length ?? 1, index = (children?.indexOf(this) ?? 0) + 1, indexOfType = (childrenOfType?.indexOf(this) ?? 0) + 1, lastIndex = siblingsCount - index + 1, lastIndexOfType = siblingsCountOfType - indexOfType + 1;
+        return step.every(selector => {
+            if (typeof selector === 'string') {
+                switch (selector) { // 情形1：简单伪选择器、type和name
+                    case '*':
+                        return true;
+                    case ':root':
+                        return !parentNode;
+                    case ':first-child':
+                        return index === 1;
+                    case ':first-of-type':
+                        return indexOfType === 1;
+                    case ':last-child':
+                        return lastIndex === 1;
+                    case ':last-of-type':
+                        return lastIndexOfType === 1;
+                    case ':only-child':
+                        return siblingsCount === 1;
+                    case ':only-of-type':
+                        return siblingsCountOfType === 1;
+                    case ':empty':
+                        return !childNodes.some(({ type: t, data }) => t !== 'text' || data);
+                    case ':parent':
+                        return childNodes.some(({ type: t, data }) => t !== 'text' || data);
+                    case ':header':
+                        return type === 'heading';
+                    case ':hidden':
+                        return this.text() === '';
+                    case ':visible':
+                        return this.text() !== '';
+                    case ':only-whitespace':
+                        return this.text().trim() === '';
+                    case ':any-link':
+                        return type === 'link'
+                            || type === 'free-ext-link'
+                            || type === 'ext-link'
+                            || (type === 'file' || type === 'gallery-image' && link);
+                    case ':local-link':
+                        return (type === 'link' || type === 'file' || type === 'gallery-image')
+                            && link instanceof title_1.Title
+                            && link.title === '';
+                    case ':invalid':
+                        return type === 'table-inter' || type === 'image-parameter' && name === 'invalid';
+                    case ':required':
+                        return this.#isProtected() === true;
+                    case ':optional':
+                        return this.#isProtected() === false;
+                    default: {
+                        const [t, n] = selector.split('#');
+                        return (!t || t === type) && (!n || n === name);
+                    }
+                }
+            }
+            else if (selector.length === 4) { // 情形2：属性选择器
+                return this.#matchesAttr(...selector);
+            }
+            const [s, pseudo] = selector; // 情形3：复杂伪选择器
+            switch (pseudo) {
+                case 'is':
+                    return this.matches(s);
+                case 'not':
+                    return !this.matches(s);
+                case 'nth-child':
+                    return nth(s, index);
+                case 'nth-of-type':
+                    return nth(s, indexOfType);
+                case 'nth-last-child':
+                    return nth(s, lastIndex);
+                case 'nth-last-of-type':
+                    return nth(s, lastIndexOfType);
+                case 'contains':
+                    return this.text().includes(s);
+                case 'has':
+                    return Boolean(this.querySelector(s));
+                case 'lang': {
+                    const regex = new RegExp(`^${s}(?:-|$)`, 'u');
+                    return matchesLang(this, regex)
+                        || this.getAncestors().some(ancestor => matchesLang(ancestor, regex));
+                }
+                case 'regex': {
+                    const mt = /^([^,]+),\s*\/(.+)\/([a-z]*)$/u.exec(s);
+                    if (!mt) {
+                        throw new SyntaxError('错误的伪选择器用法。请使用形如 ":regex(\'attr, /re/i\')" 的格式。');
+                    }
+                    try {
+                        return new RegExp(mt[2], mt[3]).test(String(this.getAttribute(mt[1].trim())));
+                    }
+                    catch {
+                        throw new SyntaxError(`错误的正则表达式：/${mt[2]}/${mt[3]}`);
+                    }
+                }
+                default:
+                    throw new SyntaxError(`未定义的伪选择器：${pseudo}`);
+            }
+        });
+    }
+    /**
+     * 检查是否符合解析后的选择器
+     * @param copy 解析后的选择器
+     */
+    #matchesArray(copy) {
+        const condition = [...copy], step = condition.pop();
+        if (this.#matches(step)) {
+            const { parentNode, previousElementSibling } = this;
+            switch (condition[condition.length - 1]?.relation) {
+                case undefined:
+                    return true;
+                case '>':
+                    return Boolean(parentNode && parentNode.#matchesArray(condition));
+                case '+':
+                    return Boolean(previousElementSibling && previousElementSibling.#matchesArray(condition));
+                case '~': {
+                    if (!parentNode) {
+                        return false;
+                    }
+                    const { children } = parentNode;
+                    return children.slice(0, children.indexOf(this))
+                        .some(child => child.#matchesArray(condition));
+                }
+                default: // ' '
+                    return this.getAncestors().some(ancestor => ancestor.#matchesArray(condition));
+            }
+        }
+        return false;
+    }
+    /**
+     * 检查是否符合选择器
+     * @param selector 选择器
+     */
+    matches(selector) {
+        return selector === undefined || this.#getCondition(selector)(this);
+    }
+    /**
+     * 类型选择器
+     * @param types
+     */
+    getElementByTypes(types) {
+        const typeSet = new Set(types.split(',').map(str => str.trim()));
+        return this.#getElementBy((({ type }) => typeSet.has(type)));
+    }
+    /**
+     * id选择器
+     * @param id id名
+     */
+    getElementById(id) {
+        return this.#getElementBy((token => 'id' in token && token.id === id));
+    }
+    /**
+     * 类选择器
+     * @param className 类名之一
+     */
+    getElementsByClassName(className) {
+        return this.#getElementsBy((token => 'classList' in token && token.classList.has(className)));
+    }
+    /**
+     * 标签名选择器
+     * @param tag 标签名
+     */
+    getElementsByTagName(tag) {
+        return this.#getElementsBy((({ type, name }) => name === tag && (type === 'html' || type === 'ext')));
+    }
+    /**
+     * 在开头批量插入子节点
+     * @param elements 插入节点
+     */
+    prepend(...elements) {
+        for (let i = 0; i < elements.length; i++) {
+            this.insertAt(elements[i], i);
+        }
+    }
+    /**
+     * 获取子节点的位置
+     * @param node 子节点
+     * @throws `RangeError` 找不到子节点
+     */
+    #getChildIndex(node) {
+        const i = this.childNodes.indexOf(node);
+        if (i === -1) {
+            throw new RangeError('Not a child node!');
+        }
+        return i;
+    }
+    /**
+     * 移除子节点
+     * @param node 子节点
+     */
+    removeChild(node) {
+        return this.removeAt(this.#getChildIndex(node));
+    }
+    insertBefore(child, reference) {
+        return reference === undefined
+            ? this.insertAt(child)
+            : this.insertAt(child, this.#getChildIndex(reference));
+    }
+}
+exports.AstElement = AstElement;
+constants_1.classes['AstElement'] = __filename;