Checking if two FILE *s are associated with the same stream.

[Edd]

Hello,

I have an array of strings containing filenames. I must open each in turn
and parse the data within. However, if a filename appears multiple times in
the list it must still only be read and parsed once.

Initially, to solve this problem I thought strcmp'ing the filenames would
suffice to check for duplicates in the list. However, it may be the case
that "somefile.in" identifies the same file as "~/inputs/somefile.in",
despite the strings being different. Thus simply using strcmp is inadequate.

Assuming:

FILE *fptr1 = fopen("somefile.in", "rb");
FILE *fptr2 = fopen("~/inputs/somefile.in", "rb");

both return non-NULL, is there a portable way of checking that fptr1 and
fptr2 are associated with the same stream?
As a really nasty hack, I could check equality of the appropriate fields in
the FILEs. But that probably wouldn't port outside of a given implementation
of the standard library, let alone across Operating Systems.

Can anybody suggest a way around my problem?

Thanks,
Edd

[Martin Dickopp]

"Edd" <ed***********@nunswithguns.net> writes:

I have an array of strings containing filenames. I must open each in turn
and parse the data within. However, if a filename appears multiple times in
the list it must still only be read and parsed once.

Initially, to solve this problem I thought strcmp'ing the filenames would
suffice to check for duplicates in the list. However, it may be the case
that "somefile.in" identifies the same file as "~/inputs/somefile.in",
despite the strings being different. Thus simply using strcmp is inadequate.

Assuming:

FILE *fptr1 = fopen("somefile.in", "rb");
FILE *fptr2 = fopen("~/inputs/somefile.in", "rb");

both return non-NULL, is there a portable way of checking that fptr1 and
fptr2 are associated with the same stream?
No, not in ISO C. But many operating systems provide an OS specific
way to check this, so you could ask in a group dedicated to your OS
(e.g. comp.unix.programmer if you're writing code for Unix).
As a really nasty hack, I could check equality of the appropriate fields in
the FILEs. But that probably wouldn't port outside of a given implementation
of the standard library, let alone across Operating Systems.

I'd recommend to adhere to some standard that provides a way to achieve
what you want (e.g. POSIX if you're writing code for Unix) instead of
relying on implementation internals.

Martin
--
,--. Martin Dickopp, Dresden, Germany ,= ,-_-. =.
/ ,- ) http://www.zero-based.org/ ((_/)o o(\_))
\ `-' `-'(. .)`-'
`-. Debian, a variant of the GNU operating system. \_/

[Stephen L.]

Martin Dickopp wrote:

"Edd" <ed***********@nunswithguns.net> writes:

I have an array of strings containing filenames. I must open each in turn
and parse the data within. However, if a filename appears multiple times in
the list it must still only be read and parsed once.

Initially, to solve this problem I thought strcmp'ing the filenames would
suffice to check for duplicates in the list. However, it may be the case
that "somefile.in" identifies the same file as "~/inputs/somefile.in",
despite the strings being different. Thus simply using strcmp is inadequate.

Assuming:

FILE *fptr1 = fopen("somefile.in", "rb");
FILE *fptr2 = fopen("~/inputs/somefile.in", "rb");

both return non-NULL, is there a portable way of checking that fptr1 and
fptr2 are associated with the same stream?

No, not in ISO C. But many operating systems provide an OS specific
way to check this, so you could ask in a group dedicated to your OS
(e.g. comp.unix.programmer if you're writing code for Unix).

Check out (in support of Martin's post) ->

http://david.tribble.com/text/c0xfstat.html
Stephen

[Edd]

Stephen L. wrote:

Martin Dickopp wrote:

"Edd" <ed***********@nunswithguns.net> writes:
[ 8< - - - snip ]

Assuming:

FILE *fptr1 = fopen("somefile.in", "rb");
FILE *fptr2 = fopen("~/inputs/somefile.in", "rb");

both return non-NULL, is there a portable way of checking that
fptr1 and fptr2 are associated with the same stream?

No, not in ISO C. But many operating systems provide an OS specific
way to check this, so you could ask in a group dedicated to your OS
(e.g. comp.unix.programmer if you're writing code for Unix).

Check out (in support of Martin's post) ->

http://david.tribble.com/text/c0xfstat.html

Ah, that's pretty handy! I'll probably be coding my project in a *nix
environment, but I'd like it to work on windows too. The code found at the
end of the link you posted should make things considerably easier -- thanks!

Edd

[Keith Thompson]

"Edd" <ed***********@nunswithguns.net> writes:
[...]

FILE *fptr1 = fopen("somefile.in", "rb");
FILE *fptr2 = fopen("~/inputs/somefile.in", "rb");

[...]

The C standard says very little about what file names are legal or how
they map to external files. In particular, the C standard doesn't say
anything about how the '~' and '/' characters in

"~/inputs/somefile.in"

might be interpreted (or the '.' character, for that matter).

If you're expecting '~' to have some particular special meaning, you
should be sure that that meaning applies to an fopen() call.

<OT>In Unix, it doesn't; if you have any questions about that, try
comp.unix.programmer.</OT>

--
Keith Thompson (The_Other_Keith) ks***@mib.org <http://www.ghoti.net/~kst>
San Diego Supercomputer Center <*> <http://users.sdsc.edu/~kst>
We must do something. This is something. Therefore, we must do this.