Hello all,

I'm trying to use fscanf to read a certain string, using this formatting :

" someid=%u%[;]c"

The latter part, "%[;]c" is a trick to detect if the whole string is read :
Only if the ";" is encountered the result of fscanf equals 2 (can't use
"%n", as that doesn't up the result count).

The thing is that I actually took a look at what got stored, and found that
instead of a single byte, 0x3B, *two* where stored, 0x003B. When I remove
the "[;]" it goes back to storing a just single byte.

Can somebody explain to my why the extra 0x00 happens (what the reason for
it is) ?

Second question : Is there some way I can tell fscanf that it should match
upto the terminating zero (so I can do away with the "%[;]c" cludge
altogether) ?

Regards,
Rudy Wieser

Hmmm ...

> The latter part, "%[;]c" is a trick to detect if the whole string is read

Hmmm ... It seems I misunderstood what the "[","]" brackets are for, and
from it I see that in the above the "c" isn't regarded as a formatting
specifier. It also explains the 'extra' "0x00" in the result.

And as a result the above needs to be changed to "%1[;]".

The second part of my question still stands though, as I would gladly do
away with the the whole trick (of sorts).

... remarkable how answers always seem to pop up /after/ the question about
them is submitted. :-)

Regards,
Rudy Wieser

On Thursday, June 16, 2022 at 3:54:41 PM UTC+1, R.Wieser wrote:
> Hello all,
>
> I'm trying to use fscanf to read a certain string, using this formatting :
>
> " someid=%u%[;]c"
>
> The latter part, "%[;]c" is a trick to detect if the whole string is read :
> Only if the ";" is encountered the result of fscanf equals 2 (can't use
> "%n", as that doesn't up the result count).

(snip)

> Second question : Is there some way I can tell fscanf that it should match
> upto the terminating zero (so I can do away with the "%[;]c" cludge
> altogether) ?

If you're using fscanf there presumably isn't a terminating zero, you will be looking at '\n' or EOF or the like. I think it is recommended not to use fscanf, rather to read in a line and process it using sscanf, which gives more control. Of course whether this is worthwhile will depend on the details, eg whether you want to simply ignore and/or complain about badly formed lines, or whether you want to abort the whole reading.

Paul,

> If you're using fscanf there presumably isn't a terminating zero,

The terminating zero was ment in regard to the formatting string.

The problem is that (f|s)scanf will simply ignore any part it cannot match
to the provided format string. And in my case that just doesn't work.

> Of course whether this is worthwhile will depend on the details, eg
> whether you
> want to simply ignore and/or complain about badly formed lines, or whether
> you
> want to abort the whole reading.

I have not considered trying to find the end of a badly-formatted (part of
a) line so I can continue reading the next string. So, aborting - or
rather terminating the scanning for more strings - on a "badly-formatted"
line it is.

I'm already considering a (better) way to check if I have an unrecognised
formatting string, or if the target is actually badly formatted.

It doesn't change anything to the question though - can I tell (f|s)scanf to
force a full matching of the formatting string (upto its terminating zero) ?
Other than the kludges I've mentioned I mean.

Regards,
Rudy Wieser