Shell Programming and Scripting

You seem to have at least heard of redirections, but >> is MEANT to append to files, so above described behaviour is no surprise. Don't use the single > to the input file either as it would destroy the target and thus the input before it can be read.
Try > TMP && mv TMP file (which might lose the original file's attributes).

If the command:

Code:

perl -pe 'BEGIN{$/="<name>"} if(/XXX<\/name/){s/Delay/Down/g}' file.xml

produces the data you want to appear in your file on the screen, and there are one or more hard links to file.xml, the following should do what you want safely:

Code:

perl -pe 'BEGIN{$/="<name>"} if(/XXX<\/name/){s/Delay/Down/g}' file.xml > file.xml$$ && cp file.xml$$ file.xml && rm file.xml$$

If you are absolutely positive that there will never be more than one hard link to file.xml and you don't care about preserving modes and ownership of the original file, you can change:

Code:

cp file.xml$$ file.xml && rm file.xml$$

in the above to:

Code:

mv file.xml$$ file.xml

to make it run a little bit faster, and avoid the period of time while the file is being copied where someone reading file.xml might find an incomplete file.

Just add the highlighted -i flag. Do not include the >> file.xml part.

Code:

perl -i -pe 'BEGIN{$/="<name>"} if(/XXX<\/name/){s/Delay/Down/g}' file.xml

Works fine.

Thank you so much

Viswa

@viswanath:

Please try to use single redirection symbol in your code like below.

Code:

perl -pe 'BEGIN{$/="<name>"} if(/XXX<\/name/){s/Delay/Down/g}' file.xml > file.xml

NO! Do not do this. The first thing that the above command does is reduce file.xml to length zero. The perl command then reads that zero length file and, finding no data to process, leaves you with all of your original data deleted.