please help with bash script loop and sed, thx :)

[spicemint]

Hi ppl, I need some help with some bash scripting, this is what i want to accomplish:
in a directory there are the files a.log b.log c.log and many more, in all files the word "INTERNET" occurs somewhere, i want something that replaces the word "INTERNET" with "a" in file a.log and saves it again as a.log, the word "INTERNET" with "b" in file b.log and saves it as b.log and so on for every file in the directory, I hope i made it clear what I want to accomplish, I have tried different approaches e.g.:
<code>
#!/bin/bash
VAR=`ls ../`
for g in $( ls )
do
sed s/INTERNET/"$VAR"/ $g
done
</code>
or
<code>
#!/bin/bash
for f in $( ls ); do
for g in $( ls ); do
sed s/INTERNET/"$f"/ $g
done
done
</code>
but it is not complete, but i hope close, please help me, i have no clue how to continue, kinda new to this, thx in advance
jody


addition: got now what i wanted, but want to save the output in the files again how to achieve that? script so far
#!/bin/sh
for i in `ls *.log`
do
bn=`basename $i .log`
sed "s/INTERNET/$bn/g" $i
done

[i92guboj]

Hello,

#!/bin/bash
VAR=`ls ../`
for g in $( ls )

No need to, just do:

for file in ../*

The sed command should be something in the lines of:

sed -i 's/test/'"${file%.*}"'/g' $file

-i allow editing in-place, otherwise, you will get an empty file. The final /g on the sed command guarantees global edition (so, ALL the instances will be changed, not only the first one).

About ${file%.*}, google for "bash string mangling". It's used to remove the extension of the file name, which is what you intended, I think.

EDIT: I changed your "g" variable by "file", to avoid confusion with the sed "g" command.

[scm]

The sed command should be something in the lines of:

sed -i 's/test/'"${file%.*}"'/g' $file

No need for the fancy quoting, in this example the following will suffice:

sed -i "s/test/${file%.*}/g" $file