node-html-parser
Advanced tools
node-html-parser - npm Package Compare versions
Comparing version 2.2.1 to 3.0.0
@@ -1,241 +0,101 @@ | ||
| /** | ||
| * Cache to store generated match functions | ||
| * @type {Object} | ||
| */ | ||
| let pMatchFunctionCache = {}; | ||
| function compare_tagname(tag1, tag2) { | ||
| if (!tag1) { | ||
| return !tag2; | ||
| } | ||
| if (!tag2) { | ||
| return !tag1; | ||
| } | ||
| return tag1.toLowerCase() === tag2.toLowerCase(); | ||
| import NodeType from './nodes/type'; | ||
| function isTag(node) { | ||
| return node.nodeType === NodeType.ELEMENT_NODE; | ||
| } | ||
| /** | ||
| * Function cache | ||
| */ | ||
| const functionCache = { | ||
| f145(el, tagName, classes) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| if (el.id !== tagName.substr(1)) { | ||
| return false; | ||
| } | ||
| for (let cls = classes, i = 0; i < cls.length; i++) { | ||
| if (el.classNames.indexOf(cls[i]) === -1) { | ||
| return false; | ||
| function getAttributeValue(elem, name) { | ||
| return elem.getAttribute(name); | ||
| } | ||
| function getName(elem) { | ||
| return (elem.rawTagName || '').toLowerCase(); | ||
| } | ||
| function getChildren(node) { | ||
| return node.childNodes; | ||
| } | ||
| function getParent(node) { | ||
| return node.parentNode; | ||
| } | ||
| function getText(node) { | ||
| return node.text; | ||
| } | ||
| function removeSubsets(nodes) { | ||
| let idx = nodes.length; | ||
| let node; | ||
| let ancestor; | ||
| let replace; | ||
| // Check if each node (or one of its ancestors) is already contained in the | ||
| // array. | ||
| while (--idx > -1) { | ||
| node = ancestor = nodes[idx]; | ||
| // Temporarily remove the node under consideration | ||
| nodes[idx] = null; | ||
| replace = true; | ||
| while (ancestor) { | ||
| if (nodes.indexOf(ancestor) > -1) { | ||
| replace = false; | ||
| nodes.splice(idx, 1); | ||
| break; | ||
| } | ||
| ancestor = getParent(ancestor); | ||
| } | ||
| return true; | ||
| }, | ||
| f45(el, tagName, classes) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| for (let cls = classes, i = 0; i < cls.length; i++) { | ||
| if (el.classNames.indexOf(cls[i]) === -1) { | ||
| return false; | ||
| } | ||
| // If the node has been found to be unique, re-insert it. | ||
| if (replace) { | ||
| nodes[idx] = node; | ||
| } | ||
| return true; | ||
| }, | ||
| f15(el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| if (el.id !== tagName.substr(1)) { | ||
| return false; | ||
| } | ||
| return nodes; | ||
| } | ||
| function existsOne(test, elems) { | ||
| return elems.some((elem) => { | ||
| return isTag(elem) ? test(elem) || existsOne(test, getChildren(elem)) : false; | ||
| }); | ||
| } | ||
| function getSiblings(node) { | ||
| const parent = getParent(node); | ||
| return parent && getChildren(parent); | ||
| } | ||
| function hasAttrib(elem, name) { | ||
| return getAttributeValue(elem, name) !== undefined; | ||
| } | ||
| function findOne(test, elems) { | ||
| let elem = null; | ||
| for (let i = 0, l = elems.length; i < l && !elem; i++) { | ||
| const el = elems[i]; | ||
| if (test(el)) { | ||
| elem = el; | ||
| } | ||
| return true; | ||
| }, | ||
| f1(el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| if (el.id !== tagName.substr(1)) { | ||
| return false; | ||
| } | ||
| }, | ||
| f5() { | ||
| 'use strict'; | ||
| return true; | ||
| }, | ||
| f55(el, tagName, classes, attr_key) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = attr_key || ''; | ||
| const attrs = el.attributes; | ||
| return attrs.hasOwnProperty(attr_key); | ||
| }, | ||
| f245(el, tagName, classes, attr_key, value) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = (attr_key || '').toLowerCase(); | ||
| value = value || ''; | ||
| const attrs = el.attributes; | ||
| return Object.keys(attrs).some((key) => { | ||
| const val = attrs[key]; | ||
| return key.toLowerCase() === attr_key && val === value; | ||
| }); | ||
| // for (let cls = classes, i = 0; i < cls.length; i++) {if (el.classNames.indexOf(cls[i]) === -1){ return false;}} | ||
| // return true; | ||
| }, | ||
| f25(el, tagName, classes, attr_key, value) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = (attr_key || '').toLowerCase(); | ||
| value = value || ''; | ||
| const attrs = el.attributes; | ||
| return Object.keys(attrs).some((key) => { | ||
| const val = attrs[key]; | ||
| return key.toLowerCase() === attr_key && val === value; | ||
| }); | ||
| // return true; | ||
| }, | ||
| f2(el, tagName, classes, attr_key, value) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = (attr_key || '').toLowerCase(); | ||
| value = value || ''; | ||
| const attrs = el.attributes; | ||
| return Object.keys(attrs).some((key) => { | ||
| const val = attrs[key]; | ||
| return key.toLowerCase() === attr_key && val === value; | ||
| }); | ||
| }, | ||
| f345(el, tagName, classes) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| if (!compare_tagname(el.tagName, tagName)) { | ||
| return false; | ||
| } | ||
| for (let cls = classes, i = 0; i < cls.length; i++) { | ||
| if (el.classNames.indexOf(cls[i]) === -1) { | ||
| return false; | ||
| else { | ||
| const childs = getChildren(el); | ||
| if (childs && childs.length > 0) { | ||
| elem = findOne(test, childs); | ||
| } | ||
| } | ||
| return true; | ||
| }, | ||
| f35(el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| return compare_tagname(el.tagName, tagName); | ||
| }, | ||
| f3(el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| // if (el.tagName !== tagName) { | ||
| // return false; | ||
| // } | ||
| return compare_tagname(el.tagName, tagName); | ||
| } | ||
| }; | ||
| /** | ||
| * Matcher class to make CSS match | ||
| * | ||
| * @class Matcher | ||
| */ | ||
| export default class Matcher { | ||
| /** | ||
| * Creates an instance of Matcher. | ||
| * @param {string} selector | ||
| * | ||
| * @memberof Matcher | ||
| */ | ||
| constructor(selector) { | ||
| this.nextMatch = 0; | ||
| this.matchers = selector.split(' ').map((matcher) => { | ||
| if (pMatchFunctionCache[matcher]) { | ||
| return pMatchFunctionCache[matcher]; | ||
| } | ||
| const parts = matcher.split('.'); | ||
| const tagName = parts[0]; | ||
| const classes = parts.slice(1).sort(); | ||
| // let source = '"use strict";'; | ||
| let function_name = 'f'; | ||
| let attr_key = ''; | ||
| let value = ''; | ||
| if (tagName && tagName !== '*') { | ||
| if (tagName.startsWith('#')) { | ||
| // source += 'if (el.id != ' + JSON.stringify(tagName.substr(1)) + ') return false;';// 1 | ||
| function_name += '1'; | ||
| } | ||
| else { | ||
| // https://github.com/taoqf/node-html-parser/issues/86 | ||
| // const reg = /\[\s*([\w-]+)(\s*=\s*(((?<quote>'|")\s*(.*)(\k<quote>))|(\S*)))?\s*\]/.exec(tagName); | ||
| // `[a-b]`,`[ a-b ]`,`[a-b=c]`, `[a-b=c'd]`,`[a-b='c\' d"e ']`,`[ a-b = 'c\' d"e ' ]`,`[a-b="c' d\"e " ]`,`[ a-b = "c' d\"e " ]` | ||
| const reg = /\[\s*([\w-]+)(\s*=\s*(('\s*(.*)'|"\s*(.*)")|(\S*)))?\s*\]/.exec(tagName); | ||
| if (reg) { | ||
| attr_key = reg[1]; | ||
| value = reg[5] || reg[6] || reg[7]; | ||
| // source += `let attrs = el.attributes;for (let key in attrs){const val = attrs[key]; if (key == "${attr_key}" && val == "${value}"){return true;}} return false;`;// 2 | ||
| function_name += '2'; | ||
| } | ||
| else { | ||
| // source += 'if (el.tagName != ' + JSON.stringify(tagName) + ') return false;';// 3 | ||
| function_name += '3'; | ||
| } | ||
| } | ||
| } | ||
| if (classes.length > 0) { | ||
| // source += 'for (let cls = ' + JSON.stringify(classes) + ', i = 0; i < cls.length; i++) if (el.classNames.indexOf(cls[i]) === -1) return false;';// 4 | ||
| function_name += '4'; | ||
| } | ||
| // source += 'return true;';// 5 | ||
| function_name += '5'; | ||
| const obj = { | ||
| func: functionCache[function_name], | ||
| tagName: tagName || '', | ||
| classes: classes || '', | ||
| attr_key: attr_key || '', | ||
| value: value || '' | ||
| }; | ||
| // source = source || ''; | ||
| return (pMatchFunctionCache[matcher] = obj); | ||
| }); | ||
| return elem; | ||
| } | ||
| function findAll(test, nodes) { | ||
| let result = []; | ||
| for (let i = 0, j = nodes.length; i < j; i++) { | ||
| if (!isTag(nodes[i])) | ||
| continue; | ||
| if (test(nodes[i])) | ||
| result.push(nodes[i]); | ||
| const childs = getChildren(nodes[i]); | ||
| if (childs) | ||
| result = result.concat(findAll(test, childs)); | ||
| } | ||
| /** | ||
| * Trying to advance match pointer | ||
| * @param {HTMLElement} el element to make the match | ||
| * @return {bool} true when pointer advanced. | ||
| */ | ||
| advance(el) { | ||
| if (this.nextMatch < this.matchers.length && | ||
| this.matchers[this.nextMatch].func(el, this.matchers[this.nextMatch].tagName, this.matchers[this.nextMatch].classes, this.matchers[this.nextMatch].attr_key, this.matchers[this.nextMatch].value)) { | ||
| this.nextMatch++; | ||
| return true; | ||
| } | ||
| return false; | ||
| } | ||
| /** | ||
| * Rewind the match pointer | ||
| */ | ||
| rewind() { | ||
| this.nextMatch--; | ||
| } | ||
| /** | ||
| * Trying to determine if match made. | ||
| * @return {bool} true when the match is made | ||
| */ | ||
| get matched() { | ||
| return this.nextMatch === this.matchers.length; | ||
| } | ||
| /** | ||
| * Rest match pointer. | ||
| * @return {[type]} [description] | ||
| */ | ||
| reset() { | ||
| this.nextMatch = 0; | ||
| } | ||
| /** | ||
| * flush cache to free memory | ||
| */ | ||
| flushCache() { | ||
| pMatchFunctionCache = {}; | ||
| } | ||
| return result; | ||
| } | ||
| export default { | ||
| isTag, | ||
| getAttributeValue, | ||
| getName, | ||
| getChildren, | ||
| getParent, | ||
| getText, | ||
| removeSubsets, | ||
| existsOne, | ||
| getSiblings, | ||
| hasAttrib, | ||
| findOne, | ||
| findAll | ||
| }; |
| import Node from './node'; | ||
| import NodeType from './type'; | ||
| export default class CommentNode extends Node { | ||
| constructor(rawText) { | ||
| super(); | ||
| constructor(rawText, parentNode) { | ||
| super(parentNode); | ||
| this.rawText = rawText; | ||
@@ -7,0 +7,0 @@ /** |
| import he from 'he'; | ||
| import { selectAll, selectOne } from 'css-select'; | ||
| import Node from './node'; | ||
@@ -44,4 +45,4 @@ import NodeType from './type'; | ||
| */ | ||
| constructor(tagName, keyAttrs, rawAttrs = '', parentNode = null) { | ||
| super(); | ||
| constructor(tagName, keyAttrs, rawAttrs = '', parentNode) { | ||
| super(parentNode); | ||
| this.rawAttrs = rawAttrs; | ||
@@ -126,3 +127,3 @@ this.parentNode = parentNode; | ||
| set textContent(val) { | ||
| const content = [new TextNode(val)]; | ||
| const content = [new TextNode(val, this)]; | ||
| this.childNodes = content; | ||
@@ -204,3 +205,3 @@ } | ||
| const r = parse(content, options); | ||
| content = r.childNodes.length ? r.childNodes : [new TextNode(content)]; | ||
| content = r.childNodes.length ? r.childNodes : [new TextNode(content, this)]; | ||
| } | ||
@@ -288,60 +289,66 @@ this.childNodes = content; | ||
| * @param {string} selector Simplified CSS selector | ||
| * @param {Matcher} selector A Matcher instance | ||
| * @return {HTMLElement[]} matching elements | ||
| */ | ||
| querySelectorAll(selector) { | ||
| let matcher; | ||
| if (selector instanceof Matcher) { | ||
| matcher = selector; | ||
| matcher.reset(); | ||
| } | ||
| else { | ||
| if (selector.includes(',')) { | ||
| const selectors = selector.split(','); | ||
| return Array.from(selectors.reduce((pre, cur) => { | ||
| const result = this.querySelectorAll(cur.trim()); | ||
| return result.reduce((p, c) => { | ||
| return p.add(c); | ||
| }, pre); | ||
| }, new Set())); | ||
| } | ||
| matcher = new Matcher(selector); | ||
| } | ||
| const stack = []; | ||
| return this.childNodes.reduce((res, cur) => { | ||
| stack.push([cur, 0, false]); | ||
| while (stack.length) { | ||
| const state = arr_back(stack); // get last element | ||
| const el = state[0]; | ||
| if (state[1] === 0) { | ||
| // Seen for first time. | ||
| if (el.nodeType !== NodeType.ELEMENT_NODE) { | ||
| stack.pop(); | ||
| continue; | ||
| } | ||
| const html_el = el; | ||
| state[2] = matcher.advance(html_el); | ||
| if (state[2]) { | ||
| if (matcher.matched) { | ||
| res.push(html_el); | ||
| res.push(...(html_el.querySelectorAll(selector))); | ||
| // no need to go further. | ||
| matcher.rewind(); | ||
| stack.pop(); | ||
| continue; | ||
| } | ||
| } | ||
| } | ||
| if (state[1] < el.childNodes.length) { | ||
| stack.push([el.childNodes[state[1]++], 0, false]); | ||
| } | ||
| else { | ||
| if (state[2]) { | ||
| matcher.rewind(); | ||
| } | ||
| stack.pop(); | ||
| } | ||
| } | ||
| return res; | ||
| }, []); | ||
| return selectAll(selector, this, { | ||
| xmlMode: true, | ||
| adapter: Matcher | ||
| }); | ||
| // let matcher: Matcher; | ||
| // if (selector instanceof Matcher) { | ||
| // matcher = selector; | ||
| // matcher.reset(); | ||
| // } else { | ||
| // if (selector.includes(',')) { | ||
| // const selectors = selector.split(','); | ||
| // return Array.from(selectors.reduce((pre, cur) => { | ||
| // const result = this.querySelectorAll(cur.trim()); | ||
| // return result.reduce((p, c) => { | ||
| // return p.add(c); | ||
| // }, pre); | ||
| // }, new Set<HTMLElement>())); | ||
| // } | ||
| // matcher = new Matcher(selector); | ||
| // } | ||
| // interface IStack { | ||
| // 0: Node; // node | ||
| // 1: number; // children | ||
| // 2: boolean; // found flag | ||
| // } | ||
| // const stack = [] as IStack[]; | ||
| // return this.childNodes.reduce((res, cur) => { | ||
| // stack.push([cur, 0, false]); | ||
| // while (stack.length) { | ||
| // const state = arr_back(stack); // get last element | ||
| // const el = state[0]; | ||
| // if (state[1] === 0) { | ||
| // // Seen for first time. | ||
| // if (el.nodeType !== NodeType.ELEMENT_NODE) { | ||
| // stack.pop(); | ||
| // continue; | ||
| // } | ||
| // const html_el = el as HTMLElement; | ||
| // state[2] = matcher.advance(html_el); | ||
| // if (state[2]) { | ||
| // if (matcher.matched) { | ||
| // res.push(html_el); | ||
| // res.push(...(html_el.querySelectorAll(selector))); | ||
| // // no need to go further. | ||
| // matcher.rewind(); | ||
| // stack.pop(); | ||
| // continue; | ||
| // } | ||
| // } | ||
| // } | ||
| // if (state[1] < el.childNodes.length) { | ||
| // stack.push([el.childNodes[state[1]++], 0, false]); | ||
| // } else { | ||
| // if (state[2]) { | ||
| // matcher.rewind(); | ||
| // } | ||
| // stack.pop(); | ||
| // } | ||
| // } | ||
| // return res; | ||
| // }, [] as HTMLElement[]); | ||
| } | ||
@@ -351,45 +358,46 @@ /** | ||
| * @param {string} selector Simplified CSS selector | ||
| * @param {Matcher} selector A Matcher instance | ||
| * @return {HTMLElement} matching node | ||
| */ | ||
| querySelector(selector) { | ||
| let matcher; | ||
| if (selector instanceof Matcher) { | ||
| matcher = selector; | ||
| matcher.reset(); | ||
| } | ||
| else { | ||
| matcher = new Matcher(selector); | ||
| } | ||
| const stack = []; | ||
| for (const node of this.childNodes) { | ||
| stack.push([node, 0, false]); | ||
| while (stack.length) { | ||
| const state = arr_back(stack); | ||
| const el = state[0]; | ||
| if (state[1] === 0) { | ||
| // Seen for first time. | ||
| if (el.nodeType !== NodeType.ELEMENT_NODE) { | ||
| stack.pop(); | ||
| continue; | ||
| } | ||
| state[2] = matcher.advance(el); | ||
| if (state[2]) { | ||
| if (matcher.matched) { | ||
| return el; | ||
| } | ||
| } | ||
| } | ||
| if (state[1] < el.childNodes.length) { | ||
| stack.push([el.childNodes[state[1]++], 0, false]); | ||
| } | ||
| else { | ||
| if (state[2]) { | ||
| matcher.rewind(); | ||
| } | ||
| stack.pop(); | ||
| } | ||
| } | ||
| } | ||
| return null; | ||
| return selectOne(selector, this, { | ||
| xmlMode: true, | ||
| adapter: Matcher | ||
| }); | ||
| // let matcher: Matcher; | ||
| // if (selector instanceof Matcher) { | ||
| // matcher = selector; | ||
| // matcher.reset(); | ||
| // } else { | ||
| // matcher = new Matcher(selector); | ||
| // } | ||
| // const stack = [] as { 0: Node; 1: 0 | 1; 2: boolean }[]; | ||
| // for (const node of this.childNodes) { | ||
| // stack.push([node, 0, false]); | ||
| // while (stack.length) { | ||
| // const state = arr_back(stack); | ||
| // const el = state[0]; | ||
| // if (state[1] === 0) { | ||
| // // Seen for first time. | ||
| // if (el.nodeType !== NodeType.ELEMENT_NODE) { | ||
| // stack.pop(); | ||
| // continue; | ||
| // } | ||
| // state[2] = matcher.advance(el as HTMLElement); | ||
| // if (state[2]) { | ||
| // if (matcher.matched) { | ||
| // return el as HTMLElement; | ||
| // } | ||
| // } | ||
| // } | ||
| // if (state[1] < el.childNodes.length) { | ||
| // stack.push([el.childNodes[state[1]++], 0, false]); | ||
| // } else { | ||
| // if (state[2]) { | ||
| // matcher.rewind(); | ||
| // } | ||
| // stack.pop(); | ||
| // } | ||
| // } | ||
| // } | ||
| // return null; | ||
| } | ||
@@ -404,5 +412,3 @@ /** | ||
| this.childNodes.push(node); | ||
| if (node instanceof HTMLElement) { | ||
| node.parentNode = this; | ||
| } | ||
| node.parentNode = this; | ||
| return node; | ||
@@ -733,3 +739,3 @@ } | ||
| } | ||
| const root = new HTMLElement(null, {}); | ||
| const root = new HTMLElement(null, {}, '', null); | ||
| let currentParent = root; | ||
@@ -746,3 +752,3 @@ const stack = [root]; | ||
| const text = data.substring(lastTextPos, kMarkupPattern.lastIndex - match[0].length); | ||
| currentParent.appendChild(new TextNode(text)); | ||
| currentParent.appendChild(new TextNode(text, currentParent)); | ||
| } | ||
@@ -759,3 +765,3 @@ } | ||
| const text = data.substring(lastTextPos - 3, lastTextPos - match[0].length + 4); | ||
| currentParent.appendChild(new CommentNode(text)); | ||
| currentParent.appendChild(new CommentNode(text, currentParent)); | ||
| } | ||
@@ -782,3 +788,3 @@ continue; | ||
| // https://github.com/taoqf/node-html-parser/issues/38 | ||
| currentParent = currentParent.appendChild(new HTMLElement(match[2], attrs, match[3])); | ||
| currentParent = currentParent.appendChild(new HTMLElement(match[2], attrs, match[3], null)); | ||
| stack.push(currentParent); | ||
@@ -804,3 +810,3 @@ if (is_block_text_element(match[2])) { | ||
| if (text.length > 0) { | ||
| currentParent.appendChild(new TextNode(text)); | ||
| currentParent.appendChild(new TextNode(text, currentParent)); | ||
| } | ||
@@ -807,0 +813,0 @@ } |
@@ -5,3 +5,4 @@ /** | ||
| export default class Node { | ||
| constructor() { | ||
| constructor(parentNode = null) { | ||
| this.parentNode = parentNode; | ||
| this.childNodes = []; | ||
@@ -8,0 +9,0 @@ } |
@@ -8,4 +8,4 @@ import NodeType from './type'; | ||
| export default class TextNode extends Node { | ||
| constructor(rawText) { | ||
| super(); | ||
| constructor(rawText, parentNode) { | ||
| super(parentNode); | ||
| this.rawText = rawText; | ||
@@ -12,0 +12,0 @@ /** |
588
dist/main.js
@@ -50,3 +50,5 @@ var __extends = (this && this.__extends) || (function () { | ||
| var Node = /** @class */ (function () { | ||
| function Node() { | ||
| function Node(parentNode) { | ||
| if (parentNode === void 0) { parentNode = null; } | ||
| this.parentNode = parentNode; | ||
| this.childNodes = []; | ||
@@ -82,4 +84,4 @@ } | ||
| __extends(CommentNode, _super); | ||
| function CommentNode(rawText) { | ||
| var _this = _super.call(this) || this; | ||
| function CommentNode(rawText, parentNode) { | ||
| var _this = _super.call(this, parentNode) || this; | ||
| _this.rawText = rawText; | ||
@@ -122,4 +124,4 @@ /** | ||
| __extends(TextNode, _super); | ||
| function TextNode(rawText) { | ||
| var _this = _super.call(this) || this; | ||
| function TextNode(rawText, parentNode) { | ||
| var _this = _super.call(this, parentNode) || this; | ||
| _this.rawText = rawText; | ||
@@ -162,252 +164,106 @@ /** | ||
| }); | ||
| define("matcher", ["require", "exports"], function (require, exports) { | ||
| define("matcher", ["require", "exports", "nodes/type"], function (require, exports, type_3) { | ||
| "use strict"; | ||
| Object.defineProperty(exports, "__esModule", { value: true }); | ||
| /** | ||
| * Cache to store generated match functions | ||
| * @type {Object} | ||
| */ | ||
| var pMatchFunctionCache = {}; | ||
| function compare_tagname(tag1, tag2) { | ||
| if (!tag1) { | ||
| return !tag2; | ||
| } | ||
| if (!tag2) { | ||
| return !tag1; | ||
| } | ||
| return tag1.toLowerCase() === tag2.toLowerCase(); | ||
| type_3 = __importDefault(type_3); | ||
| function isTag(node) { | ||
| return node.nodeType === type_3.default.ELEMENT_NODE; | ||
| } | ||
| /** | ||
| * Function cache | ||
| */ | ||
| var functionCache = { | ||
| f145: function (el, tagName, classes) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| if (el.id !== tagName.substr(1)) { | ||
| return false; | ||
| } | ||
| for (var cls = classes, i = 0; i < cls.length; i++) { | ||
| if (el.classNames.indexOf(cls[i]) === -1) { | ||
| return false; | ||
| function getAttributeValue(elem, name) { | ||
| return elem.getAttribute(name); | ||
| } | ||
| function getName(elem) { | ||
| return (elem.rawTagName || '').toLowerCase(); | ||
| } | ||
| function getChildren(node) { | ||
| return node.childNodes; | ||
| } | ||
| function getParent(node) { | ||
| return node.parentNode; | ||
| } | ||
| function getText(node) { | ||
| return node.text; | ||
| } | ||
| function removeSubsets(nodes) { | ||
| var idx = nodes.length; | ||
| var node; | ||
| var ancestor; | ||
| var replace; | ||
| // Check if each node (or one of its ancestors) is already contained in the | ||
| // array. | ||
| while (--idx > -1) { | ||
| node = ancestor = nodes[idx]; | ||
| // Temporarily remove the node under consideration | ||
| nodes[idx] = null; | ||
| replace = true; | ||
| while (ancestor) { | ||
| if (nodes.indexOf(ancestor) > -1) { | ||
| replace = false; | ||
| nodes.splice(idx, 1); | ||
| break; | ||
| } | ||
| ancestor = getParent(ancestor); | ||
| } | ||
| return true; | ||
| }, | ||
| f45: function (el, tagName, classes) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| for (var cls = classes, i = 0; i < cls.length; i++) { | ||
| if (el.classNames.indexOf(cls[i]) === -1) { | ||
| return false; | ||
| } | ||
| // If the node has been found to be unique, re-insert it. | ||
| if (replace) { | ||
| nodes[idx] = node; | ||
| } | ||
| return true; | ||
| }, | ||
| f15: function (el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| if (el.id !== tagName.substr(1)) { | ||
| return false; | ||
| } | ||
| return nodes; | ||
| } | ||
| function existsOne(test, elems) { | ||
| return elems.some(function (elem) { | ||
| return isTag(elem) ? test(elem) || existsOne(test, getChildren(elem)) : false; | ||
| }); | ||
| } | ||
| function getSiblings(node) { | ||
| var parent = getParent(node); | ||
| return parent && getChildren(parent); | ||
| } | ||
| function hasAttrib(elem, name) { | ||
| return getAttributeValue(elem, name) !== undefined; | ||
| } | ||
| function findOne(test, elems) { | ||
| var elem = null; | ||
| for (var i = 0, l = elems.length; i < l && !elem; i++) { | ||
| var el = elems[i]; | ||
| if (test(el)) { | ||
| elem = el; | ||
| } | ||
| return true; | ||
| }, | ||
| f1: function (el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| if (el.id !== tagName.substr(1)) { | ||
| return false; | ||
| } | ||
| }, | ||
| f5: function () { | ||
| 'use strict'; | ||
| return true; | ||
| }, | ||
| f55: function (el, tagName, classes, attr_key) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = attr_key || ''; | ||
| var attrs = el.attributes; | ||
| return attrs.hasOwnProperty(attr_key); | ||
| }, | ||
| f245: function (el, tagName, classes, attr_key, value) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = (attr_key || '').toLowerCase(); | ||
| value = value || ''; | ||
| var attrs = el.attributes; | ||
| return Object.keys(attrs).some(function (key) { | ||
| var val = attrs[key]; | ||
| return key.toLowerCase() === attr_key && val === value; | ||
| }); | ||
| // for (let cls = classes, i = 0; i < cls.length; i++) {if (el.classNames.indexOf(cls[i]) === -1){ return false;}} | ||
| // return true; | ||
| }, | ||
| f25: function (el, tagName, classes, attr_key, value) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = (attr_key || '').toLowerCase(); | ||
| value = value || ''; | ||
| var attrs = el.attributes; | ||
| return Object.keys(attrs).some(function (key) { | ||
| var val = attrs[key]; | ||
| return key.toLowerCase() === attr_key && val === value; | ||
| }); | ||
| // return true; | ||
| }, | ||
| f2: function (el, tagName, classes, attr_key, value) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = (attr_key || '').toLowerCase(); | ||
| value = value || ''; | ||
| var attrs = el.attributes; | ||
| return Object.keys(attrs).some(function (key) { | ||
| var val = attrs[key]; | ||
| return key.toLowerCase() === attr_key && val === value; | ||
| }); | ||
| }, | ||
| f345: function (el, tagName, classes) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| if (!compare_tagname(el.tagName, tagName)) { | ||
| return false; | ||
| } | ||
| for (var cls = classes, i = 0; i < cls.length; i++) { | ||
| if (el.classNames.indexOf(cls[i]) === -1) { | ||
| return false; | ||
| else { | ||
| var childs = getChildren(el); | ||
| if (childs && childs.length > 0) { | ||
| elem = findOne(test, childs); | ||
| } | ||
| } | ||
| return true; | ||
| }, | ||
| f35: function (el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| return compare_tagname(el.tagName, tagName); | ||
| }, | ||
| f3: function (el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| // if (el.tagName !== tagName) { | ||
| // return false; | ||
| // } | ||
| return compare_tagname(el.tagName, tagName); | ||
| } | ||
| return elem; | ||
| } | ||
| function findAll(test, nodes) { | ||
| var result = []; | ||
| for (var i = 0, j = nodes.length; i < j; i++) { | ||
| if (!isTag(nodes[i])) | ||
| continue; | ||
| if (test(nodes[i])) | ||
| result.push(nodes[i]); | ||
| var childs = getChildren(nodes[i]); | ||
| if (childs) | ||
| result = result.concat(findAll(test, childs)); | ||
| } | ||
| return result; | ||
| } | ||
| exports.default = { | ||
| isTag: isTag, | ||
| getAttributeValue: getAttributeValue, | ||
| getName: getName, | ||
| getChildren: getChildren, | ||
| getParent: getParent, | ||
| getText: getText, | ||
| removeSubsets: removeSubsets, | ||
| existsOne: existsOne, | ||
| getSiblings: getSiblings, | ||
| hasAttrib: hasAttrib, | ||
| findOne: findOne, | ||
| findAll: findAll | ||
| }; | ||
| /** | ||
| * Matcher class to make CSS match | ||
| * | ||
| * @class Matcher | ||
| */ | ||
| var Matcher = /** @class */ (function () { | ||
| /** | ||
| * Creates an instance of Matcher. | ||
| * @param {string} selector | ||
| * | ||
| * @memberof Matcher | ||
| */ | ||
| function Matcher(selector) { | ||
| this.nextMatch = 0; | ||
| this.matchers = selector.split(' ').map(function (matcher) { | ||
| if (pMatchFunctionCache[matcher]) { | ||
| return pMatchFunctionCache[matcher]; | ||
| } | ||
| var parts = matcher.split('.'); | ||
| var tagName = parts[0]; | ||
| var classes = parts.slice(1).sort(); | ||
| // let source = '"use strict";'; | ||
| var function_name = 'f'; | ||
| var attr_key = ''; | ||
| var value = ''; | ||
| if (tagName && tagName !== '*') { | ||
| if (tagName.startsWith('#')) { | ||
| // source += 'if (el.id != ' + JSON.stringify(tagName.substr(1)) + ') return false;';// 1 | ||
| function_name += '1'; | ||
| } | ||
| else { | ||
| // https://github.com/taoqf/node-html-parser/issues/86 | ||
| // const reg = /\[\s*([\w-]+)(\s*=\s*(((?<quote>'|")\s*(.*)(\k<quote>))|(\S*)))?\s*\]/.exec(tagName); | ||
| // `[a-b]`,`[ a-b ]`,`[a-b=c]`, `[a-b=c'd]`,`[a-b='c\' d"e ']`,`[ a-b = 'c\' d"e ' ]`,`[a-b="c' d\"e " ]`,`[ a-b = "c' d\"e " ]` | ||
| var reg = /\[\s*([\w-]+)(\s*=\s*(('\s*(.*)'|"\s*(.*)")|(\S*)))?\s*\]/.exec(tagName); | ||
| if (reg) { | ||
| attr_key = reg[1]; | ||
| value = reg[5] || reg[6] || reg[7]; | ||
| // source += `let attrs = el.attributes;for (let key in attrs){const val = attrs[key]; if (key == "${attr_key}" && val == "${value}"){return true;}} return false;`;// 2 | ||
| function_name += '2'; | ||
| } | ||
| else { | ||
| // source += 'if (el.tagName != ' + JSON.stringify(tagName) + ') return false;';// 3 | ||
| function_name += '3'; | ||
| } | ||
| } | ||
| } | ||
| if (classes.length > 0) { | ||
| // source += 'for (let cls = ' + JSON.stringify(classes) + ', i = 0; i < cls.length; i++) if (el.classNames.indexOf(cls[i]) === -1) return false;';// 4 | ||
| function_name += '4'; | ||
| } | ||
| // source += 'return true;';// 5 | ||
| function_name += '5'; | ||
| var obj = { | ||
| func: functionCache[function_name], | ||
| tagName: tagName || '', | ||
| classes: classes || '', | ||
| attr_key: attr_key || '', | ||
| value: value || '' | ||
| }; | ||
| // source = source || ''; | ||
| return (pMatchFunctionCache[matcher] = obj); | ||
| }); | ||
| } | ||
| /** | ||
| * Trying to advance match pointer | ||
| * @param {HTMLElement} el element to make the match | ||
| * @return {bool} true when pointer advanced. | ||
| */ | ||
| Matcher.prototype.advance = function (el) { | ||
| if (this.nextMatch < this.matchers.length && | ||
| this.matchers[this.nextMatch].func(el, this.matchers[this.nextMatch].tagName, this.matchers[this.nextMatch].classes, this.matchers[this.nextMatch].attr_key, this.matchers[this.nextMatch].value)) { | ||
| this.nextMatch++; | ||
| return true; | ||
| } | ||
| return false; | ||
| }; | ||
| /** | ||
| * Rewind the match pointer | ||
| */ | ||
| Matcher.prototype.rewind = function () { | ||
| this.nextMatch--; | ||
| }; | ||
| Object.defineProperty(Matcher.prototype, "matched", { | ||
| /** | ||
| * Trying to determine if match made. | ||
| * @return {bool} true when the match is made | ||
| */ | ||
| get: function () { | ||
| return this.nextMatch === this.matchers.length; | ||
| }, | ||
| enumerable: false, | ||
| configurable: true | ||
| }); | ||
| /** | ||
| * Rest match pointer. | ||
| * @return {[type]} [description] | ||
| */ | ||
| Matcher.prototype.reset = function () { | ||
| this.nextMatch = 0; | ||
| }; | ||
| /** | ||
| * flush cache to free memory | ||
| */ | ||
| Matcher.prototype.flushCache = function () { | ||
| pMatchFunctionCache = {}; | ||
| }; | ||
| return Matcher; | ||
| }()); | ||
| exports.default = Matcher; | ||
| }); | ||
@@ -463,3 +319,3 @@ define("parse", ["require", "exports", "back", "nodes/html"], function (require, exports, back_1, html_1) { | ||
| }); | ||
| define("nodes/html", ["require", "exports", "he", "nodes/node", "nodes/type", "nodes/text", "matcher", "back", "nodes/comment", "parse"], function (require, exports, he_1, node_3, type_3, text_1, matcher_1, back_2, comment_1, parse_1) { | ||
| define("nodes/html", ["require", "exports", "he", "css-select", "nodes/node", "nodes/type", "nodes/text", "matcher", "back", "nodes/comment", "parse"], function (require, exports, he_1, css_select_1, node_3, type_4, text_1, matcher_1, back_2, comment_1, parse_1) { | ||
| "use strict"; | ||
@@ -470,3 +326,3 @@ Object.defineProperty(exports, "__esModule", { value: true }); | ||
| node_3 = __importDefault(node_3); | ||
| type_3 = __importDefault(type_3); | ||
| type_4 = __importDefault(type_4); | ||
| text_1 = __importDefault(text_1); | ||
@@ -515,4 +371,3 @@ matcher_1 = __importDefault(matcher_1); | ||
| if (rawAttrs === void 0) { rawAttrs = ''; } | ||
| if (parentNode === void 0) { parentNode = null; } | ||
| var _this = _super.call(this) || this; | ||
| var _this = _super.call(this, parentNode) || this; | ||
| _this.rawAttrs = rawAttrs; | ||
@@ -524,3 +379,3 @@ _this.parentNode = parentNode; | ||
| */ | ||
| _this.nodeType = type_3.default.ELEMENT_NODE; | ||
| _this.nodeType = type_4.default.ELEMENT_NODE; | ||
| _this.rawTagName = tagName; | ||
@@ -609,3 +464,3 @@ _this.rawAttrs = rawAttrs || ''; | ||
| set: function (val) { | ||
| var content = [new text_1.default(val)]; | ||
| var content = [new text_1.default(val, this)]; | ||
| this.childNodes = content; | ||
@@ -636,3 +491,3 @@ }, | ||
| function dfs(node) { | ||
| if (node.nodeType === type_3.default.ELEMENT_NODE) { | ||
| if (node.nodeType === type_4.default.ELEMENT_NODE) { | ||
| if (kBlockElements.get(node.rawTagName)) { | ||
@@ -651,3 +506,3 @@ if (currentBlock.length > 0) { | ||
| } | ||
| else if (node.nodeType === type_3.default.TEXT_NODE) { | ||
| else if (node.nodeType === type_4.default.TEXT_NODE) { | ||
| if (node.isWhitespace) { | ||
@@ -705,3 +560,3 @@ // Whitespace node, postponed output | ||
| var r = parse_1.default(content, options); | ||
| content = r.childNodes.length ? r.childNodes : [new text_1.default(content)]; | ||
| content = r.childNodes.length ? r.childNodes : [new text_1.default(content, this)]; | ||
| } | ||
@@ -725,3 +580,3 @@ this.childNodes = content; | ||
| var childNode = this.childNodes[i]; | ||
| if (childNode.nodeType === type_3.default.ELEMENT_NODE) { | ||
| if (childNode.nodeType === type_4.default.ELEMENT_NODE) { | ||
| childNode.trimRight(pattern); | ||
@@ -757,6 +612,6 @@ } | ||
| node.childNodes.forEach(function (childNode) { | ||
| if (childNode.nodeType === type_3.default.ELEMENT_NODE) { | ||
| if (childNode.nodeType === type_4.default.ELEMENT_NODE) { | ||
| dfs(childNode); | ||
| } | ||
| else if (childNode.nodeType === type_3.default.TEXT_NODE) { | ||
| else if (childNode.nodeType === type_4.default.TEXT_NODE) { | ||
| if (!childNode.isWhitespace) { | ||
@@ -783,3 +638,3 @@ write('#text'); | ||
| this.childNodes.forEach(function (node) { | ||
| if (node.nodeType === type_3.default.TEXT_NODE) { | ||
| if (node.nodeType === type_4.default.TEXT_NODE) { | ||
| if (node.isWhitespace) { | ||
@@ -790,3 +645,3 @@ return; | ||
| } | ||
| else if (node.nodeType === type_3.default.ELEMENT_NODE) { | ||
| else if (node.nodeType === type_4.default.ELEMENT_NODE) { | ||
| node.removeWhitespace(); | ||
@@ -802,61 +657,66 @@ } | ||
| * @param {string} selector Simplified CSS selector | ||
| * @param {Matcher} selector A Matcher instance | ||
| * @return {HTMLElement[]} matching elements | ||
| */ | ||
| HTMLElement.prototype.querySelectorAll = function (selector) { | ||
| var _this = this; | ||
| var matcher; | ||
| if (selector instanceof matcher_1.default) { | ||
| matcher = selector; | ||
| matcher.reset(); | ||
| } | ||
| else { | ||
| if (selector.includes(',')) { | ||
| var selectors = selector.split(','); | ||
| return Array.from(selectors.reduce(function (pre, cur) { | ||
| var result = _this.querySelectorAll(cur.trim()); | ||
| return result.reduce(function (p, c) { | ||
| return p.add(c); | ||
| }, pre); | ||
| }, new Set())); | ||
| } | ||
| matcher = new matcher_1.default(selector); | ||
| } | ||
| var stack = []; | ||
| return this.childNodes.reduce(function (res, cur) { | ||
| stack.push([cur, 0, false]); | ||
| while (stack.length) { | ||
| var state = back_2.default(stack); // get last element | ||
| var el = state[0]; | ||
| if (state[1] === 0) { | ||
| // Seen for first time. | ||
| if (el.nodeType !== type_3.default.ELEMENT_NODE) { | ||
| stack.pop(); | ||
| continue; | ||
| } | ||
| var html_el = el; | ||
| state[2] = matcher.advance(html_el); | ||
| if (state[2]) { | ||
| if (matcher.matched) { | ||
| res.push(html_el); | ||
| res.push.apply(res, (html_el.querySelectorAll(selector))); | ||
| // no need to go further. | ||
| matcher.rewind(); | ||
| stack.pop(); | ||
| continue; | ||
| } | ||
| } | ||
| } | ||
| if (state[1] < el.childNodes.length) { | ||
| stack.push([el.childNodes[state[1]++], 0, false]); | ||
| } | ||
| else { | ||
| if (state[2]) { | ||
| matcher.rewind(); | ||
| } | ||
| stack.pop(); | ||
| } | ||
| } | ||
| return res; | ||
| }, []); | ||
| return css_select_1.selectAll(selector, this, { | ||
| xmlMode: true, | ||
| adapter: matcher_1.default | ||
| }); | ||
| // let matcher: Matcher; | ||
| // if (selector instanceof Matcher) { | ||
| // matcher = selector; | ||
| // matcher.reset(); | ||
| // } else { | ||
| // if (selector.includes(',')) { | ||
| // const selectors = selector.split(','); | ||
| // return Array.from(selectors.reduce((pre, cur) => { | ||
| // const result = this.querySelectorAll(cur.trim()); | ||
| // return result.reduce((p, c) => { | ||
| // return p.add(c); | ||
| // }, pre); | ||
| // }, new Set<HTMLElement>())); | ||
| // } | ||
| // matcher = new Matcher(selector); | ||
| // } | ||
| // interface IStack { | ||
| // 0: Node; // node | ||
| // 1: number; // children | ||
| // 2: boolean; // found flag | ||
| // } | ||
| // const stack = [] as IStack[]; | ||
| // return this.childNodes.reduce((res, cur) => { | ||
| // stack.push([cur, 0, false]); | ||
| // while (stack.length) { | ||
| // const state = arr_back(stack); // get last element | ||
| // const el = state[0]; | ||
| // if (state[1] === 0) { | ||
| // // Seen for first time. | ||
| // if (el.nodeType !== NodeType.ELEMENT_NODE) { | ||
| // stack.pop(); | ||
| // continue; | ||
| // } | ||
| // const html_el = el as HTMLElement; | ||
| // state[2] = matcher.advance(html_el); | ||
| // if (state[2]) { | ||
| // if (matcher.matched) { | ||
| // res.push(html_el); | ||
| // res.push(...(html_el.querySelectorAll(selector))); | ||
| // // no need to go further. | ||
| // matcher.rewind(); | ||
| // stack.pop(); | ||
| // continue; | ||
| // } | ||
| // } | ||
| // } | ||
| // if (state[1] < el.childNodes.length) { | ||
| // stack.push([el.childNodes[state[1]++], 0, false]); | ||
| // } else { | ||
| // if (state[2]) { | ||
| // matcher.rewind(); | ||
| // } | ||
| // stack.pop(); | ||
| // } | ||
| // } | ||
| // return res; | ||
| // }, [] as HTMLElement[]); | ||
| }; | ||
@@ -866,46 +726,46 @@ /** | ||
| * @param {string} selector Simplified CSS selector | ||
| * @param {Matcher} selector A Matcher instance | ||
| * @return {HTMLElement} matching node | ||
| */ | ||
| HTMLElement.prototype.querySelector = function (selector) { | ||
| var matcher; | ||
| if (selector instanceof matcher_1.default) { | ||
| matcher = selector; | ||
| matcher.reset(); | ||
| } | ||
| else { | ||
| matcher = new matcher_1.default(selector); | ||
| } | ||
| var stack = []; | ||
| for (var _i = 0, _a = this.childNodes; _i < _a.length; _i++) { | ||
| var node = _a[_i]; | ||
| stack.push([node, 0, false]); | ||
| while (stack.length) { | ||
| var state = back_2.default(stack); | ||
| var el = state[0]; | ||
| if (state[1] === 0) { | ||
| // Seen for first time. | ||
| if (el.nodeType !== type_3.default.ELEMENT_NODE) { | ||
| stack.pop(); | ||
| continue; | ||
| } | ||
| state[2] = matcher.advance(el); | ||
| if (state[2]) { | ||
| if (matcher.matched) { | ||
| return el; | ||
| } | ||
| } | ||
| } | ||
| if (state[1] < el.childNodes.length) { | ||
| stack.push([el.childNodes[state[1]++], 0, false]); | ||
| } | ||
| else { | ||
| if (state[2]) { | ||
| matcher.rewind(); | ||
| } | ||
| stack.pop(); | ||
| } | ||
| } | ||
| } | ||
| return null; | ||
| return css_select_1.selectOne(selector, this, { | ||
| xmlMode: true, | ||
| adapter: matcher_1.default | ||
| }); | ||
| // let matcher: Matcher; | ||
| // if (selector instanceof Matcher) { | ||
| // matcher = selector; | ||
| // matcher.reset(); | ||
| // } else { | ||
| // matcher = new Matcher(selector); | ||
| // } | ||
| // const stack = [] as { 0: Node; 1: 0 | 1; 2: boolean }[]; | ||
| // for (const node of this.childNodes) { | ||
| // stack.push([node, 0, false]); | ||
| // while (stack.length) { | ||
| // const state = arr_back(stack); | ||
| // const el = state[0]; | ||
| // if (state[1] === 0) { | ||
| // // Seen for first time. | ||
| // if (el.nodeType !== NodeType.ELEMENT_NODE) { | ||
| // stack.pop(); | ||
| // continue; | ||
| // } | ||
| // state[2] = matcher.advance(el as HTMLElement); | ||
| // if (state[2]) { | ||
| // if (matcher.matched) { | ||
| // return el as HTMLElement; | ||
| // } | ||
| // } | ||
| // } | ||
| // if (state[1] < el.childNodes.length) { | ||
| // stack.push([el.childNodes[state[1]++], 0, false]); | ||
| // } else { | ||
| // if (state[2]) { | ||
| // matcher.rewind(); | ||
| // } | ||
| // stack.pop(); | ||
| // } | ||
| // } | ||
| // } | ||
| // return null; | ||
| }; | ||
@@ -920,5 +780,3 @@ /** | ||
| this.childNodes.push(node); | ||
| if (node instanceof HTMLElement) { | ||
| node.parentNode = this; | ||
| } | ||
| node.parentNode = this; | ||
| return node; | ||
@@ -1282,3 +1140,3 @@ }; | ||
| } | ||
| var root = new HTMLElement(null, {}); | ||
| var root = new HTMLElement(null, {}, '', null); | ||
| var currentParent = root; | ||
@@ -1295,3 +1153,3 @@ var stack = [root]; | ||
| var text = data.substring(lastTextPos, kMarkupPattern.lastIndex - match[0].length); | ||
| currentParent.appendChild(new text_1.default(text)); | ||
| currentParent.appendChild(new text_1.default(text, currentParent)); | ||
| } | ||
@@ -1308,3 +1166,3 @@ } | ||
| var text = data.substring(lastTextPos - 3, lastTextPos - match[0].length + 4); | ||
| currentParent.appendChild(new comment_1.default(text)); | ||
| currentParent.appendChild(new comment_1.default(text, currentParent)); | ||
| } | ||
@@ -1331,3 +1189,3 @@ return "continue"; | ||
| // https://github.com/taoqf/node-html-parser/issues/38 | ||
| currentParent = currentParent.appendChild(new HTMLElement(match[2], attrs, match[3])); | ||
| currentParent = currentParent.appendChild(new HTMLElement(match[2], attrs, match[3], null)); | ||
| stack.push(currentParent); | ||
@@ -1353,3 +1211,3 @@ if (is_block_text_element(match[2])) { | ||
| if (text.length > 0) { | ||
| currentParent.appendChild(new text_1.default(text)); | ||
| currentParent.appendChild(new text_1.default(text, currentParent)); | ||
| } | ||
@@ -1411,3 +1269,3 @@ } | ||
| }); | ||
| define("index", ["require", "exports", "nodes/comment", "nodes/html", "parse", "valid", "nodes/node", "nodes/text", "nodes/type"], function (require, exports, comment_2, html_3, parse_2, valid_1, node_4, text_2, type_4) { | ||
| define("index", ["require", "exports", "nodes/comment", "nodes/html", "parse", "valid", "nodes/node", "nodes/text", "nodes/type"], function (require, exports, comment_2, html_3, parse_2, valid_1, node_4, text_2, type_5) { | ||
| "use strict"; | ||
@@ -1423,3 +1281,3 @@ Object.defineProperty(exports, "__esModule", { value: true }); | ||
| Object.defineProperty(exports, "TextNode", { enumerable: true, get: function () { return __importDefault(text_2).default; } }); | ||
| Object.defineProperty(exports, "NodeType", { enumerable: true, get: function () { return __importDefault(type_4).default; } }); | ||
| Object.defineProperty(exports, "NodeType", { enumerable: true, get: function () { return __importDefault(type_5).default; } }); | ||
| }); |
@@ -0,41 +1,6 @@ | ||
| import { Adapter } from 'css-select/lib/types'; | ||
| import HTMLElement from './nodes/html'; | ||
| /** | ||
| * Matcher class to make CSS match | ||
| * | ||
| * @class Matcher | ||
| */ | ||
| export default class Matcher { | ||
| private matchers; | ||
| private nextMatch; | ||
| /** | ||
| * Creates an instance of Matcher. | ||
| * @param {string} selector | ||
| * | ||
| * @memberof Matcher | ||
| */ | ||
| constructor(selector: string); | ||
| /** | ||
| * Trying to advance match pointer | ||
| * @param {HTMLElement} el element to make the match | ||
| * @return {bool} true when pointer advanced. | ||
| */ | ||
| advance(el: HTMLElement): boolean; | ||
| /** | ||
| * Rewind the match pointer | ||
| */ | ||
| rewind(): void; | ||
| /** | ||
| * Trying to determine if match made. | ||
| * @return {bool} true when the match is made | ||
| */ | ||
| get matched(): boolean; | ||
| /** | ||
| * Rest match pointer. | ||
| * @return {[type]} [description] | ||
| */ | ||
| reset(): void; | ||
| /** | ||
| * flush cache to free memory | ||
| */ | ||
| flushCache(): void; | ||
| } | ||
| import Node from './nodes/node'; | ||
| export declare type Predicate = (node: Node) => node is HTMLElement; | ||
| declare const _default: Adapter<Node, HTMLElement>; | ||
| export default _default; |
| "use strict"; | ||
| var __importDefault = (this && this.__importDefault) || function (mod) { | ||
| return (mod && mod.__esModule) ? mod : { "default": mod }; | ||
| }; | ||
| Object.defineProperty(exports, "__esModule", { value: true }); | ||
| /** | ||
| * Cache to store generated match functions | ||
| * @type {Object} | ||
| */ | ||
| var pMatchFunctionCache = {}; | ||
| function compare_tagname(tag1, tag2) { | ||
| if (!tag1) { | ||
| return !tag2; | ||
| } | ||
| if (!tag2) { | ||
| return !tag1; | ||
| } | ||
| return tag1.toLowerCase() === tag2.toLowerCase(); | ||
| var type_1 = __importDefault(require("./nodes/type")); | ||
| function isTag(node) { | ||
| return node.nodeType === type_1.default.ELEMENT_NODE; | ||
| } | ||
| /** | ||
| * Function cache | ||
| */ | ||
| var functionCache = { | ||
| f145: function (el, tagName, classes) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| if (el.id !== tagName.substr(1)) { | ||
| return false; | ||
| } | ||
| for (var cls = classes, i = 0; i < cls.length; i++) { | ||
| if (el.classNames.indexOf(cls[i]) === -1) { | ||
| return false; | ||
| function getAttributeValue(elem, name) { | ||
| return elem.getAttribute(name); | ||
| } | ||
| function getName(elem) { | ||
| return (elem.rawTagName || '').toLowerCase(); | ||
| } | ||
| function getChildren(node) { | ||
| return node.childNodes; | ||
| } | ||
| function getParent(node) { | ||
| return node.parentNode; | ||
| } | ||
| function getText(node) { | ||
| return node.text; | ||
| } | ||
| function removeSubsets(nodes) { | ||
| var idx = nodes.length; | ||
| var node; | ||
| var ancestor; | ||
| var replace; | ||
| // Check if each node (or one of its ancestors) is already contained in the | ||
| // array. | ||
| while (--idx > -1) { | ||
| node = ancestor = nodes[idx]; | ||
| // Temporarily remove the node under consideration | ||
| nodes[idx] = null; | ||
| replace = true; | ||
| while (ancestor) { | ||
| if (nodes.indexOf(ancestor) > -1) { | ||
| replace = false; | ||
| nodes.splice(idx, 1); | ||
| break; | ||
| } | ||
| ancestor = getParent(ancestor); | ||
| } | ||
| return true; | ||
| }, | ||
| f45: function (el, tagName, classes) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| for (var cls = classes, i = 0; i < cls.length; i++) { | ||
| if (el.classNames.indexOf(cls[i]) === -1) { | ||
| return false; | ||
| } | ||
| // If the node has been found to be unique, re-insert it. | ||
| if (replace) { | ||
| nodes[idx] = node; | ||
| } | ||
| return true; | ||
| }, | ||
| f15: function (el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| if (el.id !== tagName.substr(1)) { | ||
| return false; | ||
| } | ||
| return nodes; | ||
| } | ||
| function existsOne(test, elems) { | ||
| return elems.some(function (elem) { | ||
| return isTag(elem) ? test(elem) || existsOne(test, getChildren(elem)) : false; | ||
| }); | ||
| } | ||
| function getSiblings(node) { | ||
| var parent = getParent(node); | ||
| return parent && getChildren(parent); | ||
| } | ||
| function hasAttrib(elem, name) { | ||
| return getAttributeValue(elem, name) !== undefined; | ||
| } | ||
| function findOne(test, elems) { | ||
| var elem = null; | ||
| for (var i = 0, l = elems.length; i < l && !elem; i++) { | ||
| var el = elems[i]; | ||
| if (test(el)) { | ||
| elem = el; | ||
| } | ||
| return true; | ||
| }, | ||
| f1: function (el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| if (el.id !== tagName.substr(1)) { | ||
| return false; | ||
| } | ||
| }, | ||
| f5: function () { | ||
| 'use strict'; | ||
| return true; | ||
| }, | ||
| f55: function (el, tagName, classes, attr_key) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = attr_key || ''; | ||
| var attrs = el.attributes; | ||
| return attrs.hasOwnProperty(attr_key); | ||
| }, | ||
| f245: function (el, tagName, classes, attr_key, value) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = (attr_key || '').toLowerCase(); | ||
| value = value || ''; | ||
| var attrs = el.attributes; | ||
| return Object.keys(attrs).some(function (key) { | ||
| var val = attrs[key]; | ||
| return key.toLowerCase() === attr_key && val === value; | ||
| }); | ||
| // for (let cls = classes, i = 0; i < cls.length; i++) {if (el.classNames.indexOf(cls[i]) === -1){ return false;}} | ||
| // return true; | ||
| }, | ||
| f25: function (el, tagName, classes, attr_key, value) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = (attr_key || '').toLowerCase(); | ||
| value = value || ''; | ||
| var attrs = el.attributes; | ||
| return Object.keys(attrs).some(function (key) { | ||
| var val = attrs[key]; | ||
| return key.toLowerCase() === attr_key && val === value; | ||
| }); | ||
| // return true; | ||
| }, | ||
| f2: function (el, tagName, classes, attr_key, value) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| attr_key = (attr_key || '').toLowerCase(); | ||
| value = value || ''; | ||
| var attrs = el.attributes; | ||
| return Object.keys(attrs).some(function (key) { | ||
| var val = attrs[key]; | ||
| return key.toLowerCase() === attr_key && val === value; | ||
| }); | ||
| }, | ||
| f345: function (el, tagName, classes) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| classes = classes || []; | ||
| if (!compare_tagname(el.tagName, tagName)) { | ||
| return false; | ||
| } | ||
| for (var cls = classes, i = 0; i < cls.length; i++) { | ||
| if (el.classNames.indexOf(cls[i]) === -1) { | ||
| return false; | ||
| else { | ||
| var childs = getChildren(el); | ||
| if (childs && childs.length > 0) { | ||
| elem = findOne(test, childs); | ||
| } | ||
| } | ||
| return true; | ||
| }, | ||
| f35: function (el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| return compare_tagname(el.tagName, tagName); | ||
| }, | ||
| f3: function (el, tagName) { | ||
| 'use strict'; | ||
| tagName = tagName || ''; | ||
| // if (el.tagName !== tagName) { | ||
| // return false; | ||
| // } | ||
| return compare_tagname(el.tagName, tagName); | ||
| } | ||
| return elem; | ||
| } | ||
| function findAll(test, nodes) { | ||
| var result = []; | ||
| for (var i = 0, j = nodes.length; i < j; i++) { | ||
| if (!isTag(nodes[i])) | ||
| continue; | ||
| if (test(nodes[i])) | ||
| result.push(nodes[i]); | ||
| var childs = getChildren(nodes[i]); | ||
| if (childs) | ||
| result = result.concat(findAll(test, childs)); | ||
| } | ||
| return result; | ||
| } | ||
| exports.default = { | ||
| isTag: isTag, | ||
| getAttributeValue: getAttributeValue, | ||
| getName: getName, | ||
| getChildren: getChildren, | ||
| getParent: getParent, | ||
| getText: getText, | ||
| removeSubsets: removeSubsets, | ||
| existsOne: existsOne, | ||
| getSiblings: getSiblings, | ||
| hasAttrib: hasAttrib, | ||
| findOne: findOne, | ||
| findAll: findAll | ||
| }; | ||
| /** | ||
| * Matcher class to make CSS match | ||
| * | ||
| * @class Matcher | ||
| */ | ||
| var Matcher = /** @class */ (function () { | ||
| /** | ||
| * Creates an instance of Matcher. | ||
| * @param {string} selector | ||
| * | ||
| * @memberof Matcher | ||
| */ | ||
| function Matcher(selector) { | ||
| this.nextMatch = 0; | ||
| this.matchers = selector.split(' ').map(function (matcher) { | ||
| if (pMatchFunctionCache[matcher]) { | ||
| return pMatchFunctionCache[matcher]; | ||
| } | ||
| var parts = matcher.split('.'); | ||
| var tagName = parts[0]; | ||
| var classes = parts.slice(1).sort(); | ||
| // let source = '"use strict";'; | ||
| var function_name = 'f'; | ||
| var attr_key = ''; | ||
| var value = ''; | ||
| if (tagName && tagName !== '*') { | ||
| if (tagName.startsWith('#')) { | ||
| // source += 'if (el.id != ' + JSON.stringify(tagName.substr(1)) + ') return false;';// 1 | ||
| function_name += '1'; | ||
| } | ||
| else { | ||
| // https://github.com/taoqf/node-html-parser/issues/86 | ||
| // const reg = /\[\s*([\w-]+)(\s*=\s*(((?<quote>'|")\s*(.*)(\k<quote>))|(\S*)))?\s*\]/.exec(tagName); | ||
| // `[a-b]`,`[ a-b ]`,`[a-b=c]`, `[a-b=c'd]`,`[a-b='c\' d"e ']`,`[ a-b = 'c\' d"e ' ]`,`[a-b="c' d\"e " ]`,`[ a-b = "c' d\"e " ]` | ||
| var reg = /\[\s*([\w-]+)(\s*=\s*(('\s*(.*)'|"\s*(.*)")|(\S*)))?\s*\]/.exec(tagName); | ||
| if (reg) { | ||
| attr_key = reg[1]; | ||
| value = reg[5] || reg[6] || reg[7]; | ||
| // source += `let attrs = el.attributes;for (let key in attrs){const val = attrs[key]; if (key == "${attr_key}" && val == "${value}"){return true;}} return false;`;// 2 | ||
| function_name += '2'; | ||
| } | ||
| else { | ||
| // source += 'if (el.tagName != ' + JSON.stringify(tagName) + ') return false;';// 3 | ||
| function_name += '3'; | ||
| } | ||
| } | ||
| } | ||
| if (classes.length > 0) { | ||
| // source += 'for (let cls = ' + JSON.stringify(classes) + ', i = 0; i < cls.length; i++) if (el.classNames.indexOf(cls[i]) === -1) return false;';// 4 | ||
| function_name += '4'; | ||
| } | ||
| // source += 'return true;';// 5 | ||
| function_name += '5'; | ||
| var obj = { | ||
| func: functionCache[function_name], | ||
| tagName: tagName || '', | ||
| classes: classes || '', | ||
| attr_key: attr_key || '', | ||
| value: value || '' | ||
| }; | ||
| // source = source || ''; | ||
| return (pMatchFunctionCache[matcher] = obj); | ||
| }); | ||
| } | ||
| /** | ||
| * Trying to advance match pointer | ||
| * @param {HTMLElement} el element to make the match | ||
| * @return {bool} true when pointer advanced. | ||
| */ | ||
| Matcher.prototype.advance = function (el) { | ||
| if (this.nextMatch < this.matchers.length && | ||
| this.matchers[this.nextMatch].func(el, this.matchers[this.nextMatch].tagName, this.matchers[this.nextMatch].classes, this.matchers[this.nextMatch].attr_key, this.matchers[this.nextMatch].value)) { | ||
| this.nextMatch++; | ||
| return true; | ||
| } | ||
| return false; | ||
| }; | ||
| /** | ||
| * Rewind the match pointer | ||
| */ | ||
| Matcher.prototype.rewind = function () { | ||
| this.nextMatch--; | ||
| }; | ||
| Object.defineProperty(Matcher.prototype, "matched", { | ||
| /** | ||
| * Trying to determine if match made. | ||
| * @return {bool} true when the match is made | ||
| */ | ||
| get: function () { | ||
| return this.nextMatch === this.matchers.length; | ||
| }, | ||
| enumerable: false, | ||
| configurable: true | ||
| }); | ||
| /** | ||
| * Rest match pointer. | ||
| * @return {[type]} [description] | ||
| */ | ||
| Matcher.prototype.reset = function () { | ||
| this.nextMatch = 0; | ||
| }; | ||
| /** | ||
| * flush cache to free memory | ||
| */ | ||
| Matcher.prototype.flushCache = function () { | ||
| pMatchFunctionCache = {}; | ||
| }; | ||
| return Matcher; | ||
| }()); | ||
| exports.default = Matcher; |
@@ -5,3 +5,3 @@ import Node from './node'; | ||
| rawText: string; | ||
| constructor(rawText: string); | ||
| constructor(rawText: string, parentNode: Node); | ||
| /** | ||
@@ -8,0 +8,0 @@ * Node Type declaration. |
@@ -25,4 +25,4 @@ "use strict"; | ||
| __extends(CommentNode, _super); | ||
| function CommentNode(rawText) { | ||
| var _this = _super.call(this) || this; | ||
| function CommentNode(rawText, parentNode) { | ||
| var _this = _super.call(this, parentNode) || this; | ||
| _this.rawText = rawText; | ||
@@ -29,0 +29,0 @@ /** |
| import Node from './node'; | ||
| import NodeType from './type'; | ||
| import Matcher from '../matcher'; | ||
| export interface KeyAttributes { | ||
@@ -26,3 +25,3 @@ id?: string; | ||
| private rawAttrs; | ||
| parentNode: HTMLElement; | ||
| parentNode: HTMLElement | null; | ||
| private _attrs; | ||
@@ -44,3 +43,3 @@ private _rawAttrs; | ||
| */ | ||
| constructor(tagName: string, keyAttrs: KeyAttributes, rawAttrs?: string, parentNode?: HTMLElement); | ||
| constructor(tagName: string, keyAttrs: KeyAttributes, rawAttrs: string, parentNode: HTMLElement | null); | ||
| /** | ||
@@ -102,13 +101,11 @@ * Remove current element | ||
| * @param {string} selector Simplified CSS selector | ||
| * @param {Matcher} selector A Matcher instance | ||
| * @return {HTMLElement[]} matching elements | ||
| */ | ||
| querySelectorAll(selector: string | Matcher): HTMLElement[]; | ||
| querySelectorAll(selector: string): HTMLElement[]; | ||
| /** | ||
| * Query CSS Selector to find matching node. | ||
| * @param {string} selector Simplified CSS selector | ||
| * @param {Matcher} selector A Matcher instance | ||
| * @return {HTMLElement} matching node | ||
| */ | ||
| querySelector(selector: string | Matcher): HTMLElement; | ||
| querySelector(selector: string): HTMLElement; | ||
| /** | ||
@@ -115,0 +112,0 @@ * Append a child node to childNodes |
@@ -28,2 +28,3 @@ "use strict"; | ||
| var he_1 = __importDefault(require("he")); | ||
| var css_select_1 = require("css-select"); | ||
| var node_1 = __importDefault(require("./node")); | ||
@@ -74,4 +75,3 @@ var type_1 = __importDefault(require("./type")); | ||
| if (rawAttrs === void 0) { rawAttrs = ''; } | ||
| if (parentNode === void 0) { parentNode = null; } | ||
| var _this = _super.call(this) || this; | ||
| var _this = _super.call(this, parentNode) || this; | ||
| _this.rawAttrs = rawAttrs; | ||
@@ -167,3 +167,3 @@ _this.parentNode = parentNode; | ||
| set: function (val) { | ||
| var content = [new text_1.default(val)]; | ||
| var content = [new text_1.default(val, this)]; | ||
| this.childNodes = content; | ||
@@ -261,3 +261,3 @@ }, | ||
| var r = parse_1.default(content, options); | ||
| content = r.childNodes.length ? r.childNodes : [new text_1.default(content)]; | ||
| content = r.childNodes.length ? r.childNodes : [new text_1.default(content, this)]; | ||
| } | ||
@@ -354,61 +354,66 @@ this.childNodes = content; | ||
| * @param {string} selector Simplified CSS selector | ||
| * @param {Matcher} selector A Matcher instance | ||
| * @return {HTMLElement[]} matching elements | ||
| */ | ||
| HTMLElement.prototype.querySelectorAll = function (selector) { | ||
| var _this = this; | ||
| var matcher; | ||
| if (selector instanceof matcher_1.default) { | ||
| matcher = selector; | ||
| matcher.reset(); | ||
| } | ||
| else { | ||
| if (selector.includes(',')) { | ||
| var selectors = selector.split(','); | ||
| return Array.from(selectors.reduce(function (pre, cur) { | ||
| var result = _this.querySelectorAll(cur.trim()); | ||
| return result.reduce(function (p, c) { | ||
| return p.add(c); | ||
| }, pre); | ||
| }, new Set())); | ||
| } | ||
| matcher = new matcher_1.default(selector); | ||
| } | ||
| var stack = []; | ||
| return this.childNodes.reduce(function (res, cur) { | ||
| stack.push([cur, 0, false]); | ||
| while (stack.length) { | ||
| var state = back_1.default(stack); // get last element | ||
| var el = state[0]; | ||
| if (state[1] === 0) { | ||
| // Seen for first time. | ||
| if (el.nodeType !== type_1.default.ELEMENT_NODE) { | ||
| stack.pop(); | ||
| continue; | ||
| } | ||
| var html_el = el; | ||
| state[2] = matcher.advance(html_el); | ||
| if (state[2]) { | ||
| if (matcher.matched) { | ||
| res.push(html_el); | ||
| res.push.apply(res, (html_el.querySelectorAll(selector))); | ||
| // no need to go further. | ||
| matcher.rewind(); | ||
| stack.pop(); | ||
| continue; | ||
| } | ||
| } | ||
| } | ||
| if (state[1] < el.childNodes.length) { | ||
| stack.push([el.childNodes[state[1]++], 0, false]); | ||
| } | ||
| else { | ||
| if (state[2]) { | ||
| matcher.rewind(); | ||
| } | ||
| stack.pop(); | ||
| } | ||
| } | ||
| return res; | ||
| }, []); | ||
| return css_select_1.selectAll(selector, this, { | ||
| xmlMode: true, | ||
| adapter: matcher_1.default | ||
| }); | ||
| // let matcher: Matcher; | ||
| // if (selector instanceof Matcher) { | ||
| // matcher = selector; | ||
| // matcher.reset(); | ||
| // } else { | ||
| // if (selector.includes(',')) { | ||
| // const selectors = selector.split(','); | ||
| // return Array.from(selectors.reduce((pre, cur) => { | ||
| // const result = this.querySelectorAll(cur.trim()); | ||
| // return result.reduce((p, c) => { | ||
| // return p.add(c); | ||
| // }, pre); | ||
| // }, new Set<HTMLElement>())); | ||
| // } | ||
| // matcher = new Matcher(selector); | ||
| // } | ||
| // interface IStack { | ||
| // 0: Node; // node | ||
| // 1: number; // children | ||
| // 2: boolean; // found flag | ||
| // } | ||
| // const stack = [] as IStack[]; | ||
| // return this.childNodes.reduce((res, cur) => { | ||
| // stack.push([cur, 0, false]); | ||
| // while (stack.length) { | ||
| // const state = arr_back(stack); // get last element | ||
| // const el = state[0]; | ||
| // if (state[1] === 0) { | ||
| // // Seen for first time. | ||
| // if (el.nodeType !== NodeType.ELEMENT_NODE) { | ||
| // stack.pop(); | ||
| // continue; | ||
| // } | ||
| // const html_el = el as HTMLElement; | ||
| // state[2] = matcher.advance(html_el); | ||
| // if (state[2]) { | ||
| // if (matcher.matched) { | ||
| // res.push(html_el); | ||
| // res.push(...(html_el.querySelectorAll(selector))); | ||
| // // no need to go further. | ||
| // matcher.rewind(); | ||
| // stack.pop(); | ||
| // continue; | ||
| // } | ||
| // } | ||
| // } | ||
| // if (state[1] < el.childNodes.length) { | ||
| // stack.push([el.childNodes[state[1]++], 0, false]); | ||
| // } else { | ||
| // if (state[2]) { | ||
| // matcher.rewind(); | ||
| // } | ||
| // stack.pop(); | ||
| // } | ||
| // } | ||
| // return res; | ||
| // }, [] as HTMLElement[]); | ||
| }; | ||
@@ -418,46 +423,46 @@ /** | ||
| * @param {string} selector Simplified CSS selector | ||
| * @param {Matcher} selector A Matcher instance | ||
| * @return {HTMLElement} matching node | ||
| */ | ||
| HTMLElement.prototype.querySelector = function (selector) { | ||
| var matcher; | ||
| if (selector instanceof matcher_1.default) { | ||
| matcher = selector; | ||
| matcher.reset(); | ||
| } | ||
| else { | ||
| matcher = new matcher_1.default(selector); | ||
| } | ||
| var stack = []; | ||
| for (var _i = 0, _a = this.childNodes; _i < _a.length; _i++) { | ||
| var node = _a[_i]; | ||
| stack.push([node, 0, false]); | ||
| while (stack.length) { | ||
| var state = back_1.default(stack); | ||
| var el = state[0]; | ||
| if (state[1] === 0) { | ||
| // Seen for first time. | ||
| if (el.nodeType !== type_1.default.ELEMENT_NODE) { | ||
| stack.pop(); | ||
| continue; | ||
| } | ||
| state[2] = matcher.advance(el); | ||
| if (state[2]) { | ||
| if (matcher.matched) { | ||
| return el; | ||
| } | ||
| } | ||
| } | ||
| if (state[1] < el.childNodes.length) { | ||
| stack.push([el.childNodes[state[1]++], 0, false]); | ||
| } | ||
| else { | ||
| if (state[2]) { | ||
| matcher.rewind(); | ||
| } | ||
| stack.pop(); | ||
| } | ||
| } | ||
| } | ||
| return null; | ||
| return css_select_1.selectOne(selector, this, { | ||
| xmlMode: true, | ||
| adapter: matcher_1.default | ||
| }); | ||
| // let matcher: Matcher; | ||
| // if (selector instanceof Matcher) { | ||
| // matcher = selector; | ||
| // matcher.reset(); | ||
| // } else { | ||
| // matcher = new Matcher(selector); | ||
| // } | ||
| // const stack = [] as { 0: Node; 1: 0 | 1; 2: boolean }[]; | ||
| // for (const node of this.childNodes) { | ||
| // stack.push([node, 0, false]); | ||
| // while (stack.length) { | ||
| // const state = arr_back(stack); | ||
| // const el = state[0]; | ||
| // if (state[1] === 0) { | ||
| // // Seen for first time. | ||
| // if (el.nodeType !== NodeType.ELEMENT_NODE) { | ||
| // stack.pop(); | ||
| // continue; | ||
| // } | ||
| // state[2] = matcher.advance(el as HTMLElement); | ||
| // if (state[2]) { | ||
| // if (matcher.matched) { | ||
| // return el as HTMLElement; | ||
| // } | ||
| // } | ||
| // } | ||
| // if (state[1] < el.childNodes.length) { | ||
| // stack.push([el.childNodes[state[1]++], 0, false]); | ||
| // } else { | ||
| // if (state[2]) { | ||
| // matcher.rewind(); | ||
| // } | ||
| // stack.pop(); | ||
| // } | ||
| // } | ||
| // } | ||
| // return null; | ||
| }; | ||
@@ -472,5 +477,3 @@ /** | ||
| this.childNodes.push(node); | ||
| if (node instanceof HTMLElement) { | ||
| node.parentNode = this; | ||
| } | ||
| node.parentNode = this; | ||
| return node; | ||
@@ -834,3 +837,3 @@ }; | ||
| } | ||
| var root = new HTMLElement(null, {}); | ||
| var root = new HTMLElement(null, {}, '', null); | ||
| var currentParent = root; | ||
@@ -847,3 +850,3 @@ var stack = [root]; | ||
| var text = data.substring(lastTextPos, kMarkupPattern.lastIndex - match[0].length); | ||
| currentParent.appendChild(new text_1.default(text)); | ||
| currentParent.appendChild(new text_1.default(text, currentParent)); | ||
| } | ||
@@ -860,3 +863,3 @@ } | ||
| var text = data.substring(lastTextPos - 3, lastTextPos - match[0].length + 4); | ||
| currentParent.appendChild(new comment_1.default(text)); | ||
| currentParent.appendChild(new comment_1.default(text, currentParent)); | ||
| } | ||
@@ -883,3 +886,3 @@ return "continue"; | ||
| // https://github.com/taoqf/node-html-parser/issues/38 | ||
| currentParent = currentParent.appendChild(new HTMLElement(match[2], attrs, match[3])); | ||
| currentParent = currentParent.appendChild(new HTMLElement(match[2], attrs, match[3], null)); | ||
| stack.push(currentParent); | ||
@@ -905,3 +908,3 @@ if (is_block_text_element(match[2])) { | ||
| if (text.length > 0) { | ||
| currentParent.appendChild(new text_1.default(text)); | ||
| currentParent.appendChild(new text_1.default(text, currentParent)); | ||
| } | ||
@@ -908,0 +911,0 @@ } |
@@ -6,2 +6,3 @@ import NodeType from './type'; | ||
| export default abstract class Node { | ||
| parentNode: Node; | ||
| abstract nodeType: NodeType; | ||
@@ -12,2 +13,3 @@ childNodes: Node[]; | ||
| abstract toString(): string; | ||
| constructor(parentNode?: Node); | ||
| get innerText(): string; | ||
@@ -14,0 +16,0 @@ get textContent(): string; |
@@ -7,3 +7,5 @@ "use strict"; | ||
| var Node = /** @class */ (function () { | ||
| function Node() { | ||
| function Node(parentNode) { | ||
| if (parentNode === void 0) { parentNode = null; } | ||
| this.parentNode = parentNode; | ||
| this.childNodes = []; | ||
@@ -10,0 +12,0 @@ } |
@@ -9,3 +9,3 @@ import NodeType from './type'; | ||
| rawText: string; | ||
| constructor(rawText: string); | ||
| constructor(rawText: string, parentNode: Node); | ||
| /** | ||
@@ -12,0 +12,0 @@ * Node Type declaration. |
@@ -29,4 +29,4 @@ "use strict"; | ||
| __extends(TextNode, _super); | ||
| function TextNode(rawText) { | ||
| var _this = _super.call(this) || this; | ||
| function TextNode(rawText, parentNode) { | ||
| var _this = _super.call(this, parentNode) || this; | ||
| _this.rawText = rawText; | ||
@@ -33,0 +33,0 @@ /** |
| { | ||
| "name": "node-html-parser", | ||
| "version": "2.2.1", | ||
| "version": "3.0.0", | ||
| "description": "A very fast HTML parser, generating a simplified DOM, with basic element query support.", | ||
@@ -31,2 +31,3 @@ "main": "dist/index.js", | ||
| "dependencies": { | ||
| "css-select": "^3.1.2", | ||
| "he": "1.2.0" | ||
@@ -33,0 +34,0 @@ }, |
@@ -129,2 +129,6 @@ # Fast HTML Parser [](http://badge.fury.io/js/node-html-parser) [](https://travis-ci.org/taoqf/node-html-parser) | ||
| ### HTMLElement#setAttributes(attrs: Record<string, string>) | ||
| Set attributes of the element. | ||
| ### HTMLElement#removeAttribute(key: string) | ||
@@ -209,1 +213,5 @@ | ||
| Get or Set textContent of current element, more efficient than [set_content](#htmlelementset_contentcontent-string--node--node). | ||
| ### HTMLElement#attributes | ||
| Get all attributes of current element. **Notice: do not try to change the returned value.** |
New alerts
License Policy Violation
LicenseThis package is not allowed per your license policy. Review the package's license to ensure compliance.
Found 1 instance in 1 package
Fixed alerts
License Policy Violation
LicenseThis package is not allowed per your license policy. Review the package's license to ensure compliance.
Found 1 instance in 1 package
Improved metrics
- Number of lines in readme file
- increased by3.85%
216
Worsened metrics
- Total package byte prevSize
- decreased by-10.12%
151142
- Dependency count
- increased by100%
2
- Lines of code
- decreased by-9.83%
4117
Dependency changes
+ Addedcss-select@^3.1.2
+ Addedboolbase@1.0.0(transitive)
+ Addedcss-select@3.1.2(transitive)
+ Addedcss-what@4.0.0(transitive)
+ Addeddom-serializer@1.4.1(transitive)
+ Addeddomelementtype@2.3.0(transitive)
+ Addeddomhandler@4.3.1(transitive)
+ Addeddomutils@2.8.0(transitive)
+ Addedentities@2.2.0(transitive)
+ Addednth-check@2.1.1(transitive)