Headers parsed with added whitespace at the beginning

• Operating System: Windows 10 x64
• Node Version: 11.0.0
• NPM Version: 6.4.1
• csv-parser Version: 2.10

Expected Behavior

Example

    let pokemonDescData = [];
    fs.createReadStream('./data/csv/pokemon_desc.csv')
        .pipe(csv(
            {
              separator: '`',
              mapHeaders: (object) => console.log(object),
            }))
        .on('headers', (headers) => {
          console.log(headers);
        })
        .on('data', (data) => pokemonDescData.push(
            {
              pokemon_id: data.pokemon_id,
              p_desc: data.p_desc,
            }
        ))

CSV file example

pokemon_id`p_desc
1`Bulbasaur can be seen nappin
2`There is a bud on this

Expected headers should be:

{ header: ‘pokemon_id’, index: 0 } { header: ‘p_desc’, index: 1 }

Actual Behavior

The data that is returned from the parsing will not contain ‘pokemon_id’, but contain ’ pokemon_id’.

If I use the mapHeaders function to trim() the headers then it works correctly.

Also, if I add an empty column with any dummy data, then the first column will have the whitespace, but the 2nd column, in this example ‘pokemon_id’ will be parsed correctly.

How Do We Reproduce?

Create a CSV file like:

pokemon_id`p_desc
1`Bulbasaur can be seen napping
2`There is a bud on this

Parse with csv({ separator: ‘`’ })

[shellscape]

This may be useful to you (and others) facing similar unknowns: https://github.com/sindresorhus/strip-bom-stream. Using that, a stream can be stripped of the BOM before piping to csv-parser.

[DryWaters]

@shellscape Figured out the problem, it was a data corruption error.
I could not see the character in Notepad++ or VSCode, but looking at the .csv file in a hex editor, I found a few scrambled characters. Weird.
Sorry!