[SOLVED] Ignore case only for part of regular expression

luvshines · 02-15-2010, 04:13 PM

What i want to achieve is, if I have following in my file:

x y
X z

I should be able to find the pair, x Y and replace it with x a.

So, basically I want to search for case sensitive 'x' but search and replace case insensitive 'y'.

How should I do it ??
Can this be achieved with 'sed' command ??

AlucardZero · 02-15-2010, 05:40 PM

Depending on the flavor of regex - (?i) to turn INsensitivity on and (?-i) to turn it sensitive again.

luvshines · 02-16-2010, 12:25 AM

Does this works for Shell script ??

echo "fooFoOBAR" | egrep "(?i)foo(?-i)"

returned nothing

pixellany · 02-16-2010, 01:18 AM

I've never seen the construct that Alucar* shows--perhaps he can give us a reference.

The common way to specify a character which can be either case is the "character class"---eg "[aA]" matches "a" or "A".

Various utilities have flags for ignoring case--eg "grep -i"

luvshines · 02-16-2010, 04:55 AM

Finally got it,

Single sed command and it's done.

sed -i /regex/s/regex/replacement text/i <file>

Example:

sed -i '/^[ \t]*x [ \t]*/s/[ \t]* y[ \t]*$/ abc/gi' <file>

This is working for me.

Suggestions/comments are most welcome

pixellany · 02-16-2010, 08:23 AM

Quote:

Originally Posted by luvshines

Finally got it,

Single sed command and it's done.

sed -i /regex/s/regex/replacement text/i <file>

Example:

sed -i '/^[ \t]*x [ \t]*/s/[ \t]* y[ \t]*$/ abc/gi' <file>

This is working for me.

Suggestions/comments are most welcome

This may not be doing what you think....

First, it appears to be looking for any number of "tab" or space characters in several places. Is that relevant to your problem?

Second the "I" or "i" at the end of the s command makes it case-INsensistive, but just for the s command. This is a Gnu extension, which I found here:
http://www.gnu.org/software/sed/manual/sed.html

Regardless of the spaces and tabs, it's finding lower case "x", and then doing the substitution on "y" or "Y".

luvshines · 02-25-2010, 12:39 PM

Quote:

Originally Posted by pixellany

This may not be doing what you think....

First, it appears to be looking for any number of "tab" or space characters in several places. Is that relevant to your problem?

Second the "I" or "i" at the end of the s command makes it case-INsensistive, but just for the s command. This is a Gnu extension, which I found here:
http://www.gnu.org/software/sed/manual/sed.html

Regardless of the spaces and tabs, it's finding lower case "x", and then doing the substitution on "y" or "Y".

The "tabs" and "spaces" are very much relevant and I think it's working exactly as expected.

The problem I had was regarding, yp.conf

domain <domainname> server <servername>

entries

The domain in NIS is case sensitive but the servername is case insensitive.
I had to write a script to find and modify server name for a particular domain
So, I had to look for case-sensitive <domainname> and then replace all occurrences of <servername> with the new server.
The x in my sed command denotes the domainname and y/Y is for server

AlucardZero · 02-25-2010, 12:53 PM

Quote:

Originally Posted by pixellany

I've never seen the construct that Alucar* shows--perhaps he can give us a reference.

Maybe it's specific to Unreal IRCd?

mattseanbachman · 02-25-2010, 10:39 PM

I think i know how to do this with the -P switch with grep (P standing for Perl syntax). For that you should be able to specify things like [a-z]\x20[a-zA-z] THe first thing in braces will just match lowercase a-z, the middle is a space, and the next matches a character that's either upper or lower case.

I hope this helps you out. This is all assuming you're using grep I guess, I'm uncertain what tool you have so I can't give more help unless I know that.