[Feature request]Custom column list support for csv dump

[Anutrix]

We don't always need all columns when generating the csv/json dump. Requesting a way to supply column list for csv dump.

[omerbenamram]

use https://github.com/BurntSushi/xsv

[Anutrix]

I thought less columns during load and dump meant little more speed boost.

I am trying to reduce the time for mft to csv conversion so xsv won't help.

[omerbenamram]

from my intuition i doubt that would be meaningful. simply transmuting fields to csv is a trivial amount of compute. you're welcome to try and modify the code and check the results if you'd like, or you can integrate the library to whatever downstream task you're doing for better performance.

[Anutrix]

Reducing columns didn't help much. So I tried a different approach.

I tried simply skipping the parse process if it's deleted in from_entry function:

let is_deleted: bool = !entry.header.flags.contains(EntryFlags::ALLOCATED);

if is_deleted {
    return None;
};

And skipped it from serialization/writing it in print_csv_entry function:

if flat_entry.is_some() {
    writer.serialize(flat_entry)?;
}

Got about 60% faster conversion for my data.
I know it might just be my case to skip deleted files but it worked. Thx for the awesome library.

[omerbenamram]

sounds very reasonable, i'm glad you found something that works for you :)

[Anutrix]

sounds very reasonable, i'm glad you found something that works for you :)

Should I make MR out of my changes so that it works with a flag like --exclude-deleted or is it not needed?