How does this sed expression to remove non-alpha characters work?

Shell Programming and Scripting
1

Hello!

I know that this expression gets rid of non-alphanumeric characters:

sed 's/[^a-zA-z0-9]//g'

and I understand that it is replacing them with nothing - hence the '//'-, but I don't understand how it's doing it.
It seems it's finding strings that begin with alphanumeric and replacing them with nothing! Obviously that would not give the required output so I'm a little confused and clearly missing something...

Thanks for your help!

2

Could you please post your input code as well your expected output so that we can help you more on same.

Thanks,
R. Singh

3

The circumflex I marked in red in your code (and please use CODE tags) is not an anchor. When a circumflex is the first character in a matching expression (i.e., [expression specifying a set of characters to match] ), it specifies a non-matching expression where all of the characters except those specified by the expression following the circumflex will be matched. So, in this case everything that is not a lowercase letter, not an uppercase letter, and not a digit is removed (or, as you said, replaced by nothing).