I have a bunch of text files with different formats that somewhere in the file have email addresses. I would like to be able to parse through any number of these files for email addresses. Here are the types of input:

CFO: some_cfo@domain.com

misterman@domain.com

The Main Man mainman@domain.com

To take care of the situations I have the following seds:

#Removes line with an opening title
sed -e 's/^.*://'

#Removes opening and closing whitepsace
sed -e 's/^[ ^t]*//;s/[ ^t]*$//'

Those are both really simple, but for the life of me I can't figure out how to remove normal text from before the email address. I either end up clobbering the whole thing, or including it.

I just need to end up with something like keep what is directly attached to the '@' and delete anything after or before other whitespace

The Main Man mainman@domain.com
^not part of email. ^ and ^ are both parts of email.

Any clues anyone?

And grep saves the day. Next time I'll RTFM better.

grep -o "[[:alnum:][:graph:]]*@[[:alnum:][:graph:]]*"

I haven't tested for many bugs/quirks in the results yet, but a few quick checks seemed to work fine.

If anyone has any further ideas though they would still be greatly appreciated!

Here are a few I've used in the past (dunno if they'll work for you since SED isn't my strong point):

# get return address header
sed '/^Reply-To:/q; /^From:/h; /./d;g;q'

# get Subject header, but remove initial "Subject: " portion
sed '/^Subject: */!d; s///;q'

# get return address header
sed '/^Reply-To:/q; /^From:/h; /./d;g;q'

# parse out the address proper. Pulls out the e-mail address by itself
# from the 1-line return address header (see preceding script)
sed 's/ *(.*)//; s/>.*//; s/.*[:<] *//'

Those look awfully familiar. Did you grab those off of a "100 useful SED scripts" site?

I got them from my friend Josh who probably did pull them off that very site

hello,

I am working with:

cat filename.txt | grep @

and getting the email names reduced to something in the one line. I am thinking that this will help. What I wonder is if we can get grep to simply output the found expression instead of the whole dang line.

I am also wondering if AWK will do what you need.

Christian

Wow. Looks like a bunch of us working on it at the same time. Cool.

Christian

That is what I posted about:

grep -o "[[:alnum:][:graph:]]*@[[:alnum:][:graph:]]*"

grep -o returns the matched expression instead of the whole line matched

I realized that this can be cut down to:

grep -o "[[:graph:]]*@[[:graph:]]*"

Using SED...you could also find a pattern similar to the grep -o

sed -n 's/.*\(pattern\).*/\1/p' file


Is the * in your grep -o example = to any character? I've never used that in a grep command before...

* = any ammount of matches of the previous expression

For example:

[[:graph:]]* is really "Any printable and visible (non-space) character repeated any number of times"