[SOLVED] conditional substitution

sharky · 08-17-2011, 06:50 PM

In a CDL formatted netlist we have some lines that look like this;

Quote:

Xh2l<26>.lvl_shft_h2l lvl_shft_h2l $PINS out=out_lv<26> in=in_hv<26> gnd=gnd

and other lines that look like this;

Quote:

Xlevel_shift_l2h level_shift_l2h_2 $PINS in_lv<85>=l2h_in<85>

In the former I want to replace the first occurrence only of "<", ">", or "." with "_". Any occurrence of these symbols after a space need to be left as is.

The second example shows that sometimes there is no occurrence of the symbols prior to a space. In those cases the line is left completely unchanged.

If I were using awk with the default delimiter I would only do a substitutions on the first field. Sounds simple but I don't know how to substitute in awk, I always use sed. However, in sed I have no idea how to substitute characters only when they occur before a space.

Hope this is clear.

David the H. · 08-17-2011, 07:04 PM

Your wording is a little vague. Do you mean you want to replace all instances of all three [<>.] characters, but only in the first field? Or you only want to replace the first instance of each character in the line, as long as it's in the first field? Or only the first instance of any of the three? Or what?

I'm going to assume the first, in which case simply use awk's gsub function.

Code:

awk 'gsub( /[<>.]/ , "_"  ,$1 ) { print }' file

http://www.gnu.org/software/gawk/man...Functions.html

Also, please use [code][/code] tags around your code, to preserve formatting and to improve readability. Do not use quote tags, as they don't preserve whitespace.

sharky · 08-18-2011, 12:28 AM

Quote:

Originally Posted by David the H.

Your wording is a little vague. Do you mean you want to replace all instances of all three [<>.] characters, but only in the first field? Or you only want to replace the first instance of each character in the line, as long as it's in the first field? Or only the first instance of any of the three? Or what?

I'm going to assume the first, in which case simply use awk's gsub function.

Code:

awk 'gsub( /[<>.]/ , "_"  ,$1 ) { print }' file

http://www.gnu.org/software/gawk/man...Functions.html

Also, please use [code][/code] tags around your code, to preserve formatting and to improve readability. Do not use quote tags, as they don't preserve whitespace.

Your assumption was correct. I'll try this in the morning when I get to work.

Thx...again.

sharky · 08-18-2011, 10:54 AM

Quote:

Originally Posted by David the H.

Your wording is a little vague. Do you mean you want to replace all instances of all three [<>.] characters, but only in the first field? Or you only want to replace the first instance of each character in the line, as long as it's in the first field? Or only the first instance of any of the three? Or what?

I'm going to assume the first, in which case simply use awk's gsub function.

Code:

awk 'gsub( /[<>.]/ , "_"  ,$1 ) { print }' file

http://www.gnu.org/software/gawk/man...Functions.html

Also, please use [code][/code] tags around your code, to preserve formatting and to improve readability. Do not use quote tags, as they don't preserve whitespace.

This was almost perfect. The only problem was that it does exactly what I said I wanted. Unfortunately there were a few cases where I did not want a "." changed. CDL has keywords that alway have a leading ".", e.g. .ENDS, .SUBCKT, and .INCLUDE but it's easy enough to pipe the awk command to sed and restore the keywords.

Thanks again.