docs.sheetjs.com/docz/static/dom/SheetJSCheerio.js

24 lines
1.3 KiB
JavaScript
Raw Permalink Normal View History

2023-05-18 22:41:23 +00:00
const XLSX = require("xlsx");
const { readFileSync } = require("fs");
const cheerio = require("cheerio");
/* obtain HTML string. This example reads from test.html */
const html_str = readFileSync("SheetJSTable.html", "utf8");
/* get first TABLE element */
const $ = cheerio.load(html_str);
const doc = $("TABLE").first()[0];
/* FIX THE CHEERIO LIBRARY */
Object.defineProperty(doc.__proto__, "tagName", { get: function() { return Object.entries(this).find(r => r[0] == "tagName" || r[0] == "name")[1].toUpperCase(); }});
Object.defineProperty(doc.__proto__, "rows", { get: function() { return $(this).children("tbody").children("tr"); }});
Object.defineProperty(doc.__proto__, "cells", { get: function() { return $(this).children("td, th"); }});
Object.defineProperty(doc.__proto__, "ownerDocument", { get: function() { return {}; }});
doc.__proto__.hasAttribute = function(name) { return Object.hasOwnProperty.call(this.attribs, name); }
doc.__proto__.getAttribute = function(name) { return this.attribs[name]; }
Object.defineProperty(doc.__proto__, "innerHTML", { get: function() { return $(this).prop('innerHTML'); }});
doc.__proto__.getElementsByTagName = function(name) { return ($(this).children(name))}
/* generate workbook */
const workbook = XLSX.utils.table_to_book(doc);
XLSX.writeFile(workbook, "SheetJSCheerio.xlsx");