I have to compare the text file outputs (outputs produced in two
different platforms). Right now, I sed the two files and then compare
them. See the script below.

#! /bin/csh
set sunfile=$1.res
set vaxfile=$1.out
sed -f fixvaxf.sed $vaxfile >$vaxfile.1
sed -f fixsunf.sed $sunfile >$sunfile.1
diff -bwi $vaxfile.1 $sunfile.1 >$1.diff
rm $vaxfile.1 $sunfile.1

This script produces a diff output such as

130c130
< 1 0.0 0.10000E+01 0.67763E-20 0.10000E
+01 -0.23962E-05 0.16255E+01 0.16255E+01
---
> 1 0.0 0.10000E+01 0.00000E+00 0.10000E+01 -0.23962E-05 0.16255E+01 0.16255E+01

I would prefer an output of

130c130
<
0.67763E-20
---
> 0.00000E+00

In other words, I want to refine the current diff output to show only
the different characters in a diff line output. Is there any way to
produce a diff output like the above? I am a newbie to this forum and
would welcome any better way of doing the above. Thanks.

Looks like I lost some information while posting (due to automatic
reformatting). Hopefully this message will display correctly.

Original output:
274c274
< 1 0.0 0.10000E+01 0.00000E+00 0.10000E+01
---
> 1 0.0 0.10000E+01 -0.34694E-17 0.10000E+01

Preferred output:
274c274
< 0.00000E+00
---
> -0.34694E-17

Ragu wrote:
> Looks like I lost some information while posting (due to automatic
> reformatting). Hopefully this message will display correctly.
>
> Original output:
> 274c274
> < 1 0.0 0.10000E+01 0.00000E+00 0.10000E+01
> ---
>> 1 0.0 0.10000E+01 -0.34694E-17 0.10000E+01
>
> Preferred output:
> 274c274
> < 0.00000E+00
> ---
>> -0.34694E-17


The problen is that the real difference would look like this (no "0." and
no "E"):

274c274
< 00000 +00
---
> - 34694 -17

--
Best regards | Monica Lewinsky's X-Boyfriend's
Cyrus | Wife for President

> < 00000 +00
> ---
> > - 34694 -17

It is OK. Such an output will me to find the differences quickly.

Is there a way to show the outputs like you had mentioned? Thanks.

On Mar 24, 10:48am, Cyrus Kriticos <cyrus.kriti...@googlemail.com>
wrote:
> Ragu wrote:
> > Looks like I lost some information while posting (due to automatic
> > reformatting). Hopefully this message will display correctly.
>
> > Original output:
> > 274c274
> > < 1 0.0 0.10000E+01 0.00000E+00 0.10000E+01
> > ---
> >> 1 0.0 0.10000E+01 -0.34694E-17 0.10000E+01
>
> > Preferred output:
> > 274c274
> > < 0.00000E+00
> > ---
> >> -0.34694E-17
>
> The problen is that the real difference would look like this (no "0." and
> no "E"):
>
> 274c274
> < 00000 +00
> ---
> > - 34694 -17
>
> --
> Best regards | Monica Lewinsky's X-Boyfriend's
> Cyrus | Wife for President

On Mon, 24 Mar 2008 07:16:53 -0700, Ragu wrote:

> I have to compare the text file outputs (outputs produced in two
> different platforms). Right now, I sed the two files and then compare
> them. See the script below.
>
> #! /bin/csh
> set sunfile=$1.res
> set vaxfile=$1.out
> sed -f fixvaxf.sed $vaxfile >$vaxfile.1 sed -f fixsunf.sed $sunfile
> >$sunfile.1 diff -bwi $vaxfile.1 $sunfile.1 >$1.diff rm $vaxfile.1
> $sunfile.1
>
> This script produces a diff output such as
>
> 130c130
> < 1 0.0 0.10000E+01 0.67763E-20 0.10000E +01
> -0.23962E-05 0.16255E+01 0.16255E+01 ---
>> 1 0.0 0.10000E+01 0.00000E+00 0.10000E+01
>> -0.23962E-05 0.16255E+01 0.16255E+01
>
> I would prefer an output of
>
> 130c130
> <
> 0.67763E-20
> ---
>> 0.00000E+00
>
> In other words, I want to refine the current diff output to show only
> the different characters in a diff line output. Is there any way to
> produce a diff output like the above? I am a newbie to this forum and
> would welcome any better way of doing the above. Thanks.

See if you have "wdiff" installed.

Ragu wrote:
>> < 00000 +00
>> ---
>> > - 34694 -17
>
> It is OK. Such an output will me to find the differences quickly.
>
> Is there a way to show the outputs like you had mentioned? Thanks.

I don't know. But you can write a script to do what you asked first.

e.g. with GNU's bash:

--- cut here ---
#!/bin/bash

FILE1="/somewhere/filename1"
FILE2="/somewhere/filename2"


while read NO1 LINE1; do
read -u 3 NO2 LINE2
if [ "$LINE1" != "$LINE2" ]; then
echo line: $NO1
LA1=($LINE1)
LA2=($LINE2)
for ((I=0; I<=${#LA1[@]}; I++)); do
if [ "${LA1[$I]}" = "${LA2[$I]}" ]; then
OUT1="${OUT1}\t"
OUT2="${OUT2}\t"
else
OUT1="${OUT1}${LA1[$I]}"
OUT2="${OUT2}${LA2[$I]}"
fi
done
echo -e "< $OUT1\n> $OUT2\n---"
unset OUT1 OUT2
fi
done < <(cat -n "$FILE1") 3< <(cat -n "$FILE2")
--- cut here ---

Try to avoid commas (,) in both of your files.

--
Best regards | Monica Lewinsky's X-Boyfriend's
Cyrus | Wife for President