Perl One Liner : Capturing more than one matched regex per line and printing

uncle-c · 01-20-2012, 03:46 AM

Something similar to a previous thread of mine :

File = email.txt

Code:

$ cat email.txt
jjjlllll
jksdjklds <<user1@home.com>
<user2@home.com> <user3@home.com>
qwerty
<user4@home.com> 
$ perl -ne ' /<(.*?)>/ && print "$1\n" ' email.txt 
user1@home.com
user2@home.com
user4@home.com
$

What has happened is that a second regex, user3@home.com has not been matched. I have written a short perl script which puts each regex matched into an array and then prints them all but but how could I achieve this using/ modifiying the simple one-liner I have above ? From my limited knowledge I knew above was not going to work as I have used $\1. So how could this be tweaked to capture all the email addresses or will I have to use arrays ?

Thanks.

Cedrik · 01-20-2012, 05:02 AM

Code:

perl -ane 'print map {s/<|>//g;"$_\n"} grep {/<(.*)>/} @F' email.txt

-a : split line into @F array
grep field that contains <.*>
map results => remove < and >, add \n

uncle-c · 01-20-2012, 06:34 AM

Thanks Cedrik. One thing confuses me here is that why does map come before grep and why cannot you use grep twice ?

Cedrik · 01-20-2012, 06:43 AM

map is written before grep but is executed after...

map executes commands for each item in the list
(perldoc -f map)

while grep returns items from list for which the expression evaluated is true
(perldoc -f grep)

Here, the list comes from grep command which extracts fields from @F array that match /<(.*)>/
for each field, map executes:
- s/<|>//g; => remove all < and > from item
- "$_\n" => concatenate item with "\n" char
each item is in $_ variable, map could also be written as:

Code:

map {$_ =~ s/<|>//g; "$_\n"}

(edit)

I found that substitution could be done in grep expression, so no need to use map

Code:

perl -ane 'print grep {s/<+(.*)>/$1\n/} @F' email.txt

- I used ' s/<+(.*)>/$1\n/ ' to match 1 or more '<' like: "<<user1@home.com>"

This substitution could also work...

Code:

perl -ne 'print if s/.*?<+(.*?)>\s?\n?/$1\n/g' email.txt

uncle-c · 01-20-2012, 10:03 AM

Thanks Cedrik. I myself was trying to think up the one liner without having to use grep / arrays.

Code:

perl -ne 'print if s/.*?<+(.*?)>\s?\n?/$1\n/g' email.txt

was the the type of code I was thinking along the lines of. Could this be made even simpler ?

Thanks again.

Cedrik · 01-20-2012, 12:20 PM

It is even too simpler (fails with words after email for example)

Hey, look at this reg exp

:
http://www.ex-parrot.com/pdw/Mail-RFC822-Address.html

uncle-c · 01-22-2012, 04:16 AM

LOL. The regex to end all regexes !