foo.ts

import { readFileSync, writeFileSync } from 'fs'
import { parse, HTMLElement, TextNode } from 'node-html-parser'
// import { CssSelectorParser } from 'css-selector-parser'

const zipWith = <A, B, T>(a: A[], b: B[], func: (a: A, b: B) => T): T[] =>
	Array.from({length: a.length > b.length ? b.length : a.length}, (_, i) => func(a[i], b[i]))

// const parser = new CssSelectorParser()
// 	.registerSelectorPseudos('not', 'has', 'host', 'host-context', 'is', 'where')
// 	.registerAttrEqualityMods('^', '*', '~', '$')
// 	.registerNestingOperators('+', '~', '>')
// 	.disableSubstitutes()

// console.log(JSON.stringify(parser.parse(':a(a[href])')))

// http://www.kobis.or.kr/kobis/business/mast/thea/findTheaterInfoListXls.do
const html = readFileSync('./kobis.html', 'utf8')

const el = parse(html, {
	script: false,
	style: false,
	lowerCaseTagName: false,
	pre: false
})

const getText = (v: HTMLElement | TextNode): string => {
	if (v instanceof TextNode) {
		return v.rawText
	}

	return v.childNodes
		.filter(z => z instanceof TextNode)
		.map(z => z.rawText).join(' ')
};

if ('querySelector' in el) {
	const table = el.querySelector('table.tbl_exc')

	const ths = table.querySelector('thead')
		.querySelectorAll('th')
		.map(getText)

	const trs = table.querySelector('tbody')
		.querySelectorAll('tr')
		.map(v => v.querySelectorAll('td').map(getText))

	const data = trs.map(row =>
		Object.assign({}, ...zipWith(ths, row, (k, v) => ({[k]: v})))
	).filter(row => row.영업상태 === '영업')
	.map(row => ({
		code: row.영화상영관코드,
		name: row.영화상영관명,
		city: row.광역단체,
		area: row.기초단체,
	}))

	writeFileSync('bar.json', JSON.stringify(data))
}