How can I delete all null characters from a file?

Question

I had a directory with a lot of PGN chess files, from which I wanted to remove the move times (written as [%emt {a_number}]. I wrote this script:

regex = /\[.emt[^\]]+\]/
directory = "path/to/files"
extension = ".pgn"

Dir.chdir(directory)
Dir.foreach(directory) do |file_name|
    file_object = File.open(file_name, "r+")
    contents = file_object.read
    new_contents = contents.gsub(regex, "")
    File.truncate(directory + "/" + file_name, 0)
    file_object.puts(new_contents)
    file_object.close
end

This removed all the move times, but curiously it appended a large number of null characters to the beginning of the files (I suspect this number is equal to the number of bytes in the file). So I replaced the line new_contents = contents.gsub(regex, "") with contents.delete("\0"), but this only made it worse, appending even more null characters to the beginning of the files. How can I remove them?

Answer 1

It should work OK if you replace:

File.truncate(directory + "/" + file_name, 0)

with:

file_object.rewind

or

file_object.seek(0)

File.truncate should not be applied to open files (as here), and file_object.truncate should not be followed by any file operation other than file_object.close.

If you already have a file with nulls that you want to remove, read the file into a string str, close the file, execute

str.delete!("\000")

and then write str back to file.

Answer 2

Instead of truncating the file, better just re-open it for writing instead as it would automatically be truncated with it. And I believe you missed specifying the proper paths:

file_path = File.join(directory, file_name)
contents = File.read(file_path)  ## Previously just file_name.
new_contents = contents.gsub(regex, "")
File.open(file_path, 'w') do |file_object|
    file_object.puts(new_contents)
end

Also perhaps you didn't want to use puts as somehow it's different when writing binary data compared to writing ascii data:

File.write(file_path)