Shell Programming and Scripting

I have done a script using sed which replaces a string in file.
I face 2 problems, while using sed command.

1)last line of the file is missing.
2)if i am are using wild card character in the command, Its not taking the next matching pattern ie if I am giving the pattern abc*def to be replaced with abcXdef : and the input file contains:-

abctttttttttttdefmmmmmmmmmabcttttttttttdef

the o/p i am expecting is abcXdefmmmmmmmmmabcXdef

but the o/p I am getting is abcXdef.


the first problem is getting solved when i put an enter in the last line in the file. But i am having pratical difficulty that Its ahecking almost 600files.


My script code is giving below

script.sh
.................................

Code:

#!/bin/sh

find /opt/users/adm5tca1/simulator/OSDI/OSDISimulator/jakarta-tomcat-5.0.28/webapps/OSDI_SIM/WEB-INF/classes/workarea/ -name 'ResponseBody.txt' 2>/dev/null > pathy.txt
FILE_NAME=pathy.txt

while read a_line ; do

echo $a_line
/usr/bin/sed -e 's!<ns6\:originatingOrderLines><ns6\:item>.*<\/ns6\:item><\/ns6\:originatingOrderLines>!<ns6\:originatingOrderLines><ns6\:item>DEFAULT<\/ns6\:item><\/ns6\:originatingOrderLines>!g' $a_line > bb.txt
mv bb.txt $a_line


done < $FILE_NAME

.................................................

Can anyone plz help me in solving out.

Your 2nd problem:
Regexp matching is greedy by default, so it matches the most characters possible i.e. /abc.*def/ matches everything between the first "abc" and the last "def". You need a non-greedy match.
In Perl you get a non-greedy match by appending "?" to the quantifier, e.g.

Code:

echo zzzabcZZZdefYYYabcAAAdefSSS | perl -pn -e 's/(abc).*?(def)/$1X$2/g'

gives you "zzzabcXdefYYYabcXdefSSS", which is what you want, right?

I can't get the same thing to work with sed -- maybe someone else knows sed better. But if you don't mind using Perl instead, you could write

Code:

perl -pin -e 's!(<ns6:originatingOrderLines><ns6:item>).*(</ns6:item></ns6:originatingOrderLines>)!$1DEFAULT$2!g'  $a_line

N.B:These features exist in sed too - check 'em out in the man page.

Looks like you forgot to use .*? instead of .* in the final example, though. Also, Perl's -p and -n options are basically complementary; if you have -p, you don't need -n (and in fact -pin means -p and -i with an "n" argument, meaning create a backup file with an "n" suffix added to the end of the file name).