Node.js read/write stream skipping the first line on write

Question

I wrote a simple utility to convert a somewhat weird json file (multiple objects not in an array) to csv for some system testing purposes. The read and transformation themselves are fine, and the resulting string is logged to the console correctly, but sometimes the resulting csv file is missing the first data line (it shows header, 1 blank line, then rest of data). I'm using read and write streams, without any provisions for backpressure. I don't think the problem is backpressure, since only the 1st line gets skipped, but I could be wrong. Any ideas?

const fs = require('fs');
const readline = require('readline');
const JSONbig = require('json-bigint');

// Create read interface to stream each line
const readInterface = readline.createInterface({
    input: fs.createReadStream('./confirm.json'),
    // output: process.stdout,
    console: false
});

const writeHeader = fs.createWriteStream('./confirm.csv');
const header = "ACTION_TYPE,PROCESS_PICK,TYPE_FLAG,APP_ID,FACILITY_ID,CONTAINER_ID,USER_ID,CONFIRM_DATE_TS,PICK_QTY,REMAINING_QTY,PICK_STATUS,ASSIGNMENT_ID,LOCATION_ID,ITEM_ID,CLUSTER_ID,TOTAL_QTY,TOTAL_ITEMS,WAVE_NBR,QA_FLAG,WORK_DIRECTIVE_ID\n";
writeHeader.write(header);

// Create write interface to save each csv line
const writeDetail = fs.createWriteStream('./confirm.csv', {
    flags: 'a'
});

readInterface.on('line', function(line) {
    let task = JSONbig.parse(line);
    task.businessData.MESSAGE.RECORD[0].DETAIL.REG_DETAIL.forEach(element => {
        let csv = "I,PTB,0,VCO,PR9999999011,,cpicker1,2020121000000," + element.QUANTITYTOPICK.toString() + ",0,COMPLETED," +
            task.businessData.MESSAGE.RECORD[0].ASSIGNMENTNUMBER.toString() + "," + element.LOCATIONNUMBER.toString() + "," +
            element.ITEMNUMBER.toString() + ",,,," + 
            task.businessData.MESSAGE.RECORD[0].WAVE.toString() + ",N," + element.CARTONNUMBER.toString() + "\n";
        
        console.log(csv);
        try {
            writeDetail.write(csv);    
        } catch (err) {
            console.error(err);
        }
    });
});

Edit: Based on the feedback below, I consolidated the write streams into one (the missing line was still happening, but it's better coding anyway). I also added a try block around the JSON parse. Ran the code several times over different files, and no missing line. Maybe the write was happening before the parse was done? In any case, it seems my problem is resolved for the moment. I'll have to research how to properly handle backpressure later. Thanks for the help.

Answer 1

The code you show here is opening two separate writestreams on the same file and then writing to both of them without any timing coordination between them. That will clearly conflict.

You open one here:

const writeHeader = fs.createWriteStream('./confirm.csv');
const header = "ACTION_TYPE,PROCESS_PICK,TYPE_FLAG,APP_ID,FACILITY_ID,CONTAINER_ID,USER_ID,CONFIRM_DATE_TS,PICK_QTY,REMAINING_QTY,PICK_STATUS,ASSIGNMENT_ID,LOCATION_ID,ITEM_ID,CLUSTER_ID,TOTAL_QTY,TOTAL_ITEMS,WAVE_NBR,QA_FLAG,WORK_DIRECTIVE_ID\n";
writeHeader.write(header);

And, you open one here:

// Create write interface to save each csv line
const writeDetail = fs.createWriteStream('./confirm.csv', {
    flags: 'a'
});

And, then you write to the second one in your loop. Those clearly conflict. The write from the first is probably not complete when you open the second and it also may not be flushed to disk yet either. The second one opens for append, but doesn't accurately read the file position for appending because the first one hasn't yet succeeded.

This code doesn't show any reason for using separate write streams at all so the cleanest way to address this would be to just use one writestream that will accurately serialize the writes. Otherwise, you have to wait for the first writestream to finish and close before opening the second one.

---

And, your .forEach() loop needs to have backpressure support since you're repeatedly calling .write() and, at some data size, you can get backpressure. I agree that backpressure is not likely the cause of the issue you are asking about, but is something else you need to fix when rapdily writing in a loop.