Sed Question

[highst]

I might have the wrong forum here, but regardless i'm going to ask here first.

I want to replace a character with sed under the condition that a specific character doesn't follow.

Eg: i want to replace '+' but only when '+' is not followed by another '+' ('++')

The problem is i can't use s/\+[^\+]/ /g because it will replace the second character if it doesn't fail. Is there a method to either have it put back the second charactor or search for this case, but only replace the first character?

Another alternative would be to have it replace the first of the two '+'s but not the second. The problem is i need to use the (g)lobal option so that i replace cases after that.

An example string would be: This+is+a+string+with+a+following++
Result: This is a string with a following +

I'm sure this is hard to read, and for that i'm sorry but i'm really not sure how to best explain this.

[vsemaska]

How about this:

sed -e 's/++/##/g' -e 's/+/ /g' -e 's/##/ +/g'

It changes the '++' to '##' so the next substitution '+' to ' ' won't pick them up. The last substition replaces the '##' to ' +'.

Vic

[highst]

I had a similiar thought but since i don't control the input i could potentially end up replacing something the user had entered. I suppose though the longer and more unique the characters the less likly this is that you'll replace the user's input.

Thanks, i'll use that in the mean time, instead perhaps using $ since i can't think of an instance where I would see $$. Any other thoughts.

I looked at the man pages but couldn't decipher the section that looked interesting. It had something to do with &, \1 \2 and setting markers and such.

[tpl]

welcome to the forum

very clear description--an example always helps. A crude solution
employs a placeholder for the "++" thus:

sed 's/'\+\+'/GGGGGG/g;s/+/ /g;s/GGGGGG/ +/g' < test

where "test" is your example.