Thread: Read 1000000 chars into array.

Hi,
I want to read 1MB chars from stdio into array. But I will never know how big is the input, it can be 1B and in next run it can be 432kB and so on. Allowed chars are:
- digits,
- letters,
- white cars.
scanf() does not work as it stops at white char. I know that I can use 'l' prefix and specify size e.g. 1000000 - however I'm not sure will scanf() stop after NULL if 'l=1000000'?

Do you know from what you've just read how large the input is? If so, you could use fread().

If you don't know, then there are two further options:
1. Read a smaller piece into a buffer (using fread), then using your own array scanner (e.g. looking fore "newline", "}" or whatever indicates the end of your data), to copy from this buffer to the actual data array.

2. Just loop around using fgetc(). This is the "simple, but slightly slower" method.

--
Mats

input is STDIO (keyboard)

Ok, so this brings the question who would type in 1MB of data all at once?

On the other hand, "stdio" is a file of the type FILE * just like any other file you would want to read from, so both of my solutions above would work, although fgets() is certainly the better choice here.. Knowing that it's stdin, and expected to be used by hand [assuming we're not redirecting from a file], the performance is not critical [it would be different if you were to read huge amounts of data across from a file - because a file from disk can supply several megabytes per second, whilst keyboard input is generally in the 10s of bytes per second, or less], so fgetc() is certainly looking like a good choice to me.

I'm suspecting you are asking something else, perhaps you are asking "How can I detect if the user stops typing, without hitting enter?"

--
Mats

Simply in Linux you can use ReadDataFromKB.out < file.txt
And I want to handle what ever is in the file.txt. Data must be process till NULL character.

Do you KNOW for sure that there is a NUL ('\0') character in your file? Otherwise, it makes more sense to process data until end of file.

However, either (or both) case(s) can be handled quite simply with a loop around fgetc(), so is there any particular reason you don't want to do that?

--
Mats

> However, either (or both) case(s) can be handled quite simply with a loop around fgetc(), so is there any particular reason you don't want to do that?
Can you not use fread() instead? ie read BUFSIZ chunks at a time. Or am I missing something?

Originally Posted by man libc

The value of BUFSIZ is chosen on each system so as to make stream I/O efficient.

No point driving a steam car if you can drive a Ferrari

Originally Posted by zacs7

Can you not use fread() instead? ie read BUFSIZ chunks at a time.

That's exactly the question I want to ask!

Originally Posted by matsp

Do you KNOW for sure that there is a NUL ('\0') character in your file? Otherwise, it makes more sense to process data until end of file.
Mats

Correct me if I'm wrong. When I do myfile.out < file.txt, file.txt is read in myfile.out as an keyboard, therefore I can not look for EOF. However I think that while(getc()) will work.

> Correct me if I'm wrong. When I do myfile.out < file.txt, file.txt is read in myfile.out as an keyboard, therefore I can not look for EOF. However I think that while(getc()) will work.

Wrong . EOF will be returned appon the end of the stream also. Matsp has already said, stdin is just a FILE *. Treat it the same. Don't think of 'STDIN' as the keyboard, think of it as the standard stream for input, file or keyboard it doesn't matter.

As matsp has said, fgetc() will work in a loop (same as getc()).

Originally Posted by zacs7

> Correct me if I'm wrong. When I do myfile.out < file.txt, file.txt is read in myfile.out as an keyboard, therefore I can not look for EOF. However I think that while(getc()) will work.

Wrong . EOF will be returned appon the end of the stream also. Matsp has already said, stdin is just a FILE *. Treat it the same.

As matsp has said, fgetc() will work in a loop (same as getc()).

And you will not get zero back from getc() unless the user presses CTRL-@ - or something equivalent to that.

--
Mats