UNIX for Dummies Questions & Answers

Hi,

I have a file with a URL text written in it within double quotes e.g.

"http://abcd.xyz.com/mno/somefile.dtd"

I want the above text to get replaced by a single space character.

I tried

cat File1.txt | sed -e 's/("http)*(dtd")/ /g' > File2.txt

But it didnt work out. Can someone suggest an sed command which replaces this url text, including double quotes, with a space.

Thanks

Try this:

Code:

sed 's!^"http://abcd.*somefile.dtd"$! !' < File1.txt > File2.txt

Regards

Thanks....but the command didnt work.......it did not delete/replace

Regards

It works fine for me. Try this with the line you provide. It's substitute by an X instead of a space with the sed command:

Code:

echo '"http://abcd.xyz.com/mno/somefile.dtd"'|sed 's!^"http://abcd.*somefile.dtd"$!X!'

Regards

Hi Franklin,

The echo | sed works fine, but the moment i try to the same thing in the file it doesnt work.

SED works and substitutes text if I specify any other text/pattern (which doesnt have any wild character or forward slash)

cat file1.txt | sed -e 's/DOCTYPE/ /' > file2.txt

But somehow when I use wild characters in sed its not working. I can specify the url since it may change with time.

I have to remove the url from following stmt

<!DOCTYPE tag SYSTEM "http://someurl/fileabc.dtd">

Thanks

To rephrase your problem: you search for a string of the form "http://...." enclosed in double quotes and ending in a white space, yes?

Further, we do not have to consider the end-of-line situation, as the URL will always be part of some HTML-tag and therefore cannot be at the end of the line. (Otherwise we would have to consider the two cases '"http://..." ' and '"http://...."$'.)

Then the solution is:

Code:

sed 's/"http:\/\/[^ ]*dtd"/ /' file1 > file2

The reason why you code didn't work was:

Code:

cat File1.txt | sed -e 's/("http)*(dtd")/ /g' > File2.txt

First, the bracket ("()") are just simple characters if you do not escape them: "\(...\)". As you have no brackets in your search string this must fail.

Second, even if that were grouping characters the asterisk ("*") after "http" would then make the whole group optional. "*" is not "any string" like in DOS, but "the last expression zero or more times".

So, what you have been searching for was: the literal string '("http' followed by an optional ')', followed by '(dtd")'.

I hope this helps.

bakunin