Unix Shell - Setting IFS to the null character

Author

Setting IFS to the null character

David Wake

2007-05-21, 7:26 pm

The "find" utilityQ offers the "-print0" option which can be used in conjunction
with "xargs". For example:

find -name "*.jar" -print0 | xargs -0 unzip -l

This uses the null character, instead of whitespace, as the field
separator and so works correctly even if some files contain whitespace
in their paths.

I'm trying to find a way to do something similar using a "for" loop.
My aim is to make the "for" loop split on null characters. The manual
says that the way to do this is to set the IFS variable, so I'm trying
to make IFS be the null character but I'm not having much luck:

> touch 1 2 3 4 5

> export IFS=$(echo -en '\0'); for file in $(find -type f -print0); do echo $file; echo "next"; done
../2./4./5./3./1
next

> export IFS=; for file in $(find -type f -print0); do echo $file; echo "next"; done
../2./4./5./3./1
next

I'd appreciate any help!

Thanks,

David

Chris F.A. Johnson

2007-05-21, 7:26 pm

On 2007-05-21, David Wake wrote:[vbcol=seagreen]
>
> The "find" utilityQ offers the "-print0" option which can be used in conjunction
> with "xargs". For example:
>
> find -name "*.jar" -print0 | xargs -0 unzip -l
>
> This uses the null character, instead of whitespace, as the field
> separator and so works correctly even if some files contain whitespace
> in their paths.
>
> I'm trying to find a way to do something similar using a "for" loop.
> My aim is to make the "for" loop split on null characters. The manual
> says that the way to do this is to set the IFS variable, so I'm trying
> to make IFS be the null character but I'm not having much luck:
>
>

So long as the filenames do not contain newlines:

find -type f |
while IFS= read -r
do
printf "%s\n" "$file"
echo "next"
done

--
Chris F.A. Johnson, author <http://cfaj.freeshell.org/shell/>
Shell Scripting Recipes: A Problem-Solution Approach (2005, Apress)
===== My code in this post, if any, assumes the POSIX locale
===== and is released under the GNU General Public Licence

Bill Marcum

2007-05-21, 7:26 pm

On 21 May 2007 15:59:48 -0700, David Wake
<dwake@stanfordalumni.org> wrote:
>
>
>
> The "find" utilityQ offers the "-print0" option which can be used in
> conjunction with "xargs". For example:
>
> find -name "*.jar" -print0 | xargs -0 unzip -l
>
> This uses the null character, instead of whitespace, as the field
> separator and so works correctly even if some files contain whitespace
> in their paths.
>
> I'm trying to find a way to do something similar using a "for" loop.
> My aim is to make the "for" loop split on null characters. The manual
> says that the way to do this is to set the IFS variable, so I'm trying
> to make IFS be the null character but I'm not having much luck:
>
Good luck, but I don't think it's possible.

--
Nothing is but what is not.

Chris F.A. Johnson

2007-05-22, 1:18 am

On 2007-05-22, Chris F.A. Johnson wrote:
> On 2007-05-21, David Wake wrote:
>
> So long as the filenames do not contain newlines:
>
> find -type f |
> while IFS= read -r

Missing var name; fixed below:

while IFS= read -r file

> do
> printf "%s\n" "$file"
> echo "next"
> done

--
Chris F.A. Johnson, author <http://cfaj.freeshell.org/shell/>
Shell Scripting Recipes: A Problem-Solution Approach (2005, Apress)
===== My code in this post, if any, assumes the POSIX locale
===== and is released under the GNU General Public Licence

Martin Krischik

2007-05-22, 7:18 am

David Wake schrieb:
> The "find" utilityQ offers the "-print0" option which can be used in conjunction
> with "xargs". For example:
>
> find -name "*.jar" -print0 | xargs -0 unzip -l
>
> This uses the null character, instead of whitespace, as the field
> separator and so works correctly even if some files contain whitespace
> in their paths.
>
> I'm trying to find a way to do something similar using a "for" loop.
> My aim is to make the "for" loop split on null characters. The manual
> says that the way to do this is to set the IFS variable, so I'm trying
> to make IFS be the null character but I'm not having much luck:
>
>
> ./2./4./5./3./1
> next
>
> ./2./4./5./3./1
> next
>
> I'd appreciate any help!

Am I glad that I discovered Z-Shell:

--------------------------------------------------
>touch 1 2 3 4 5 'with space' 'with
quote>linefeed'
>setopt Extended_Glob;
>for i in *(.); do
for> echo "${i}";
for> echo "next";
for>done
1
next
2
next
3
next
4
next
5
next
with
linefeed
next
with space
next
--------------------------------------------------

No find, no pipe, no IFS twisting, no endless trying out different
variations, just a for loop which works first time.

And if you need recursive search then use "./**/*(.)" as pattern.

Martin
--
Martin Krischik

Stephane CHAZELAS

2007-05-22, 7:18 am

2007-05-21, 15:59(-07), David Wake:
>
> The "find" utilityQ offers the "-print0" option which can be used in conjunction
> with "xargs". For example:
>
> find -name "*.jar" -print0 | xargs -0 unzip -l
>
> This uses the null character, instead of whitespace, as the field
> separator and so works correctly even if some files contain whitespace
> in their paths.
>
> I'm trying to find a way to do something similar using a "for" loop.
> My aim is to make the "for" loop split on null characters. The manual
> says that the way to do this is to set the IFS variable, so I'm trying
> to make IFS be the null character but I'm not having much luck:
>
>
> ./2./4./5./3./1
> next
>
> ./2./4./5./3./1
> next
>
> I'd appreciate any help!
[...]

Only zsh can have a NUL byte in its variables. Environment
variables can't have NUL bytes in them as they are NUL
terminated strings.

But:

IFS=$'\0'
for file in $(find . -type f -print0); do
...
done

can actually be written in zsh as

for file in ./**/*(ND.oN); do
...
done

Has zsh has find features as part of its globbing.

Anyway, the best way to perform actions for every file found by
find is to use find's -exec predicate.

--
Stéphane