Hi there - I am new to regular expressions, but i was wondering how i would do the following...

Say I have a file with this in it:

Now say I wanted to write a regular expression to say match and delete (using sed) all of the #2 entry, how would I do that?

It is the regular expressoin part I am confused about.

My file would end up looking like this:

Thanks in advance.

PS - I have been googling, but my results (when I test what I read) are always all messed up.

I'd say the easiest way would be to get the line number of "start 2" and "end 2" and then simply use sed to delete that range of line numbers.

Yeah I figured as much, but I was looking for the reg ex example, it seems like it should be something simple, but for some reason my tests have come back with undesirable results.

Thanks!

The 'd' command in sed accepts an address range, so you want to address the range from 'start 2' to 'end 2'.

The regexp's would be:

and

I added the * after the space following # because I can't tell for sure if you have a space there or not. The * allows for there to be zero or more space characters there.

The ^ character specifies that the match happen beginnin at the first character in the line.

So the entire sed command would look something like:

Hmmm I just gave that a go. But for some reason the file is not changed at all.

I am writing a conceptual script to add and delete apache vhost entries. Just more of a "practice" for myself. So maybe if i post actual code it might make more sense as to why it is not working...

I have a "fake" vhost file with this in it:

And my actual script is simply this thus far (hard coded - the dustin.ca would be replaced with $1 later):

Now I am sure I need to be careful with the quotes in the SED command, especially when I change those entries to the $1.

With sed, you need to redirect it to a new file or it's just going to print the changes on the console or terminal.

Yeah, that's an important point.

If you are looking to write the resulting data back to the same file it may be easier to use ed (as opposed to sed) which uses similar syntax but operates on the full file in a buffer rather than line-by-line.

or

which may be slightly more readable by making the two-line command input to ed clear.

Or there may be a much better way still that I don't know.

Worked great!

Now onto figuring out how to remove an empty line if there are 2 or more (thus just leaving a single blank line).

Thanks!

Glad that worked. There is one caveat, though, which may or may not pose a problem for your application. That ed command is only going to delete the first section that matches.

If you had another section further down bounded by the same text it would be left intact. There may be a way to get ed to perform that action on all sections that match, but I'm not sure how.

sed would do it as it will process the whole file line-by-line but you would then have to do the redirection as noted above to achieve writing the new file.

However, this will probably be slightly more complicated than just:

because the '> filename' will open filename for writing and truncate it to zero, meaning all the data in the file will be lost before sed does anything.

Instead, you'll have to use a temporary file for the redirection and then mv or cp that file onto the original filename.

There may be a way to tell the shell to use a temporary file descriptor or something and avoid creating a temp file but that's beyond the scope of my knowledge.