Reading the Nth character in a file?

Ficheall

Suppose I have a file which contains just a string of 0s and 1s and nothing else.
I don't want to open the file, and I don't want to print anything or change anything within the file - I just want to know whether the Nth character is a 1 or a 0.
What would the most efficient way of doing this be?

Pygmalion

If you're writing a program to do it you can seek() (man fseek and/or lseek for the docs on the C standard library function).

For example to get the 7652th character you could do

//Open the file etc.
fseek(file, 7651, SEEK_SET);
//Read from the file etc.

No idea if there's a way to do this on the command line with the usual tools.

Worth noting that since data is generally read from the disk in blocks there might not actually be a real performance difference between reading 1 byte and reading 100 bytes, unless the bytes span more than one block on disk.

E39MSport

Technically, if you don't open the file you can't access it afaik.

Something like cut is simple.

echo "this is a test string" | cut -c6 will yield 'i' (edit, but that obviously line based. perl may better suit for an entire file)

Khannie

head -c <nthchar> <filename> | tail -c 1

Pygmalion

Khannie wrote: »

head -c <nthchar> <filename> | tail -c 1

Still involves reading in the first 'n' characters of the file, so can still be quite inefficient for very large files.

Ficheall

Cheers for the replies. Aye, yon cut -c seems to do exactly what I want but for a string. The problem is there are 8 files of about 400M, so I'd rather not have to read them if avoidable.

bnt

You could try this in Python, either scripted or in an interactive shell. Saying "open" opens a handle to the file, it doesn't mean you read the whole file in to memory. Something like this (via the docs):

f = open('/tmp/workfile', 'r+')
f.seek(7651)
f.read(1)

returns the 7652th byte of the file only.

Khannie

Pygmalion wrote: »

Still involves reading in the first 'n' characters of the file, so can still be quite inefficient for very large files.

Agreed, but only an issue if both n and the file are very large. Seeking is better of course, but it's quick and easy to type and very easy to remember (half the battle ). Even very large files should be done quickly though. I'd expect 80+MB/s.

backboiler

I'm late as usual, but what about the following?

dd bs=1 count=1 if=$inputfile skip=$position

PC here gave me the 300,000,000th character in an uncached 450 MiB file in 24 microseconds.

Khannie

Nice. dd.