行政院環境保護署 空氣品質監測資料 全年逐時資料 (http://taqm.epa.gov.tw/taqm/tw/YearlyDataDownload.aspx) 格式轉換

AirDataReformat.js

/**
 * 行政院環境保護署 空氣品質監測資料 全年逐時資料 (http://taqm.epa.gov.tw/taqm/tw/YearlyDataDownload.aspx) 格式轉換
 * keywords: 空污,空氣,品質,空氣品質,監測,逐年,逐時,資料,行政院,環保署,環境保護署
 *
 * 將原本以時間(小時)為欄，測項為列之資料轉換成
 * 以測項為欄，時間(小時)為列
 *
 * Usage: $ node AirDataReformat.js < /PATH/TO/oriAirData.xml
 */

var readline = require('readline');
var fs = require('fs');
var rl = readline.createInterface({
	input: process.stdin,
	output: process.stdout,
	terminal: false
});

var lineNum = 0;
/**
 * resultObj = {
 * 		'2016/01/01': {
 * 			'00': {
 * 				AMB_TEMP: 16
 * 			}
 * 		}
 * }
 */
var resultObj = {};
var resultCSV = '"date","station","hour",';
var area = '';
/**
 * head = {
 * 		AMB_TEMP: 1, 
 * 		CO: 1
 * }
 */
var head = {};

rl.on('line', function(line){
	lineNum ++;
	if (lineNum != 1) {
		storeData(line);
	}
});

rl.on('close', function(){
	reFormat();
});

function storeData(line) {
	var lineArr = line.split(',');
	var [date, station, item] = lineArr;
	area = station;
	head[item] = 1;

	for (let i = 3; i < lineArr.length; i++) {
		var value = `"${lineArr[i]}"`;
		var hour = `"${(i - 3)}"`;
		resultObj[date] = resultObj[date] || {};
		resultObj[date][hour] = resultObj[date][hour] || {};
		resultObj[date][hour][item] = value;
	}
}

function reFormat() {
	for (let item in head) {
		resultCSV += `${item},`;
	}
	resultCSV.slice(0, -1);
	resultCSV += '\n';

	for (let date in resultObj) {
		for (let hour in resultObj[date]) {
			resultCSV += `${date},${area},${hour},`;
			for (let item in head) {
				resultCSV += `${resultObj[date][hour][item] || ''},`;
			}
			resultCSV = resultCSV.slice(0, -1);
			resultCSV += '\n';
		}
	}

	fs.writeFile("./reFormatedAirData.csv", resultCSV, function(err) {
		if(err) {
			return console.log(err);
		} else {
			console.log('Done!');
		}
	}); 
}