ccl_xml_to_csv.js

const fs = require('fs');
const xml2js = require('xml2js');
const util = require('util');
const eyes = require('eyes');
const converter = require('json-2-csv');

const parser = new xml2js.Parser();

//const xml = fs.readFileSync('just_notes.xml', 'utf8');
const inspect = require('eyes').inspector({maxLength: false});


// parseString(xml, function (err, result) {
//     console.log(util.inspect(result, false, null));
// });





fs.readFile('just_notes.xml', function(err, data) {
    parser.parseString(data, function (err, result) {
        //console.log(util.inspect(result, false, null));
        const messyArr = result.rss['channel'][0].item;
        let bestarr = [];
        for (let i = 0; i < messyArr.length; i++) {
            let newObj = {};
            const element = messyArr[i];

            /* 
                process title 
               Make it an array of elements as spearated by space
               Pull values into appropriate meta fields
            */
            const messyTitle = element.title[0];
            const titleArr = messyTitle.split(" ");
            newObj.cTitle = titleArr[1]; 
            newObj.metaInitials = titleArr[0];
            newObj.metaCharsets = titleArr[2].replace(/[()]/g, '');;

            /* 
                process content: 
                get the raw cdata
                split it at strongs to get initial data headings in to array
                for each iten in the array, clean off the </strong>...
                then split results at colon to get arrays like [ 'Radical', '木' ]
                
            */
            const messyContent = element['content:encoded'][0];
            const messyContentArray = messyContent.split("<strong>");

            for ( let k = 0; k < messyContentArray.length; k++ ){
                mystr = messyContentArray[k];
                if ( mystr.length > 0 ){
                    subst = mystr.substring(0, mystr.lastIndexOf("</"));
                    messyContentArray[k] = subst;
                    messyContentArray[k] = subst.split(": ");
                }
            }

            //content is now an array of arrays
            //newObj.content = messyContentArray;
            //console.log(messyContentArray);
            for ( let z=0; z < messyContentArray.length; z++ ){

                let keyEl = messyContentArray[z][0];
                let valEl = messyContentArray[z][1];

                if (keyEl === "Definition"){
                    newObj.definition = '"' + String(valEl) +                                '"';
                }

                if (keyEl === "Stroke number"){
                    newObj.strokeNum = String(valEl);
                } 
                
                if (keyEl == "Radical" || keyEl == "Radical:"){
                    newObj.radical = String(valEl).trim();
                    if (newObj.radical.length > 2){
                        newObj.radical = 'no_radical_parsed';
                    }
                    newObj.radLen = newObj.radical.length;
                }

                if (keyEl === "Radical"){
                    newObj.radical = String(valEl).trim();
                    newObj.radLen = newObj.radical.length;
                    if (newObj.radical.length > 2){
                        newObj.radical = 'no_radical_parsed';
                    }
                }

                //now we need to make sure all objs have same schema
                if (newObj.hasOwnProperty('radical') !== true ){
                    newObj.radical = 'no_radical_found'
                }
                
                if (newObj.hasOwnProperty('cTitle') !== true ){
                    newObj.cTitle = 'no_cTitle_found';
                }

                if (newObj.hasOwnProperty('metaInitials') !== true ){
                    newObj.metaInitials = 'no_metaInitials_found'
                }

                if (newObj.hasOwnProperty('metaCharsets') !== true ){
                    newObj.metaCharsets = 'no_metaCharsets_found'
                }

                if (newObj.hasOwnProperty('definition') !== true ){
                    newObj.definition = 'no_definition_found'
                }

                if (newObj.hasOwnProperty('strokeNum') !== true ){
                    newObj.strokeNum = 'no_strokeNum_found'
                }

                if (newObj.hasOwnProperty('radLen') !== true ){
                    newObj.radLen = 'no_radLen_found'
                }

            }
            bestarr.push(newObj); 
        }
        const sha = JSON.stringify(bestarr);
        
        //convert object to json file
        fs.writeFile("sha.json", sha, 'utf8', function (err) {
            if (err) {
                return console.log(err);
            }
            console.log("The file was saved!");
            //console.log(sha);
        }); 

        //log it to console as a csv
        const json2csvCallback = function (err, csv) {
            if (err) throw err;
            console.log(csv);
        };
        
        converter.json2csv(bestarr, json2csvCallback);
    });
});