Gawk regexp to select sequence

Question

sorry for the nth simple question on regexp but I'm not able to get what I need without a what seems to me a too complicated solution. I'm parsing a file containing sequence of only 3 letters A,E,D as in

AADDEEDDA

EEEEEEEE

AEEEDEEA

AEEEDDAAA

and I'd like to identify only those that start with E and ends in D with only one change in the sequence as for example in

EDDDDDDDD

EEEDDDDDD

EEEEEEEED

I'm fighting with the proper regexp to do that. Here my last attempt

echo "1,AAEDDEED,1\n2,EEEEDDDD,2\n3,EDEDEDED" | gawk -F, '{if($2 ~ /^E[(ED){1,1}]*D$/ && $2 !~ /^E[(ED){2,}]*D$/) print $0}'

which does not work. Any help?

Thanks in advance.

Show your attempts. Doing that helps because it allows people who answer to explain where you went wrong and explain how their answers differ from your attempts (what the material confusion/etc. was). — Etan Reisner, Nov 11 '15 at 20:42
edit your question to include your attempts to solve your problem. Don't extend a conversation in comments about basic details of your problem. OK? :-) Good luck. — shellter, Nov 11 '15 at 20:59

Giuseppe Ricupero · Answer 1 · 2015-11-11T21:34:54.000

5

If i understand correctly your request a simple

awk '/^E+D+$/' file.input

will do the trick.

UPDATE: if the line format contains pre/post numbers (with post optional) as showed later in the example, this can be a possible pure regex adaptation (alternative to the use of field switch-F,):

awk '/^[0-9]+,E+D+(,[0-9]+)?$/' input.test

edited Nov 11 '15 at 21:34

answered Nov 11 '15 at 21:09

Giuseppe Ricupero

6,134
3
23
32

How about `$2 ~ /^E+[^ED]*D+$/`? – shadowtalker Nov 11 '15 at 21:19
Thanks, it works perfectly and I definitely need to find some time to study regexp. – G. Tartifola Nov 11 '15 at 21:24
@G.Tartifola check out [RexEgg](http://www.rexegg.com) for some tutorials. Also see my answer below for a more detailed explanation of why the correct AWK program is a lot simpler than the one you wrote. – shadowtalker Nov 11 '15 at 21:30

score 2 · Answer 2 · answered Nov 11 '15 at 21:29

2

First of all, you need the regular expression:

^E+[^ED]*D+$

This matches one or more Es at the beginning, zero or more characters that are neither E nor D in the middle, and one or more Ds at the end.

Then your AWK program will look like

$2 ~ /^E+[^ED]*D+$/

$2 refers to the 2nd field of the current record, ~ is the regex matching operator, and /s delimit a regular expression. Together, these components form what is known in AWK jargon as a "pattern", which amounts to a boolean filter for input records. Note that there is no "action" (a series of statements in {s) specified here. That's because when no action is specified, AWK assumes that the action should be { print $0 }, which prints the entire line.

answered Nov 11 '15 at 21:29

shadowtalker

12,529
3
53
96

Thanks for the explanation. – G. Tartifola Nov 11 '15 at 21:36
This solution does not work as expected, the ``[^ED]`` section of the regex will match unwanted string such as EEEAADDDD ([^ED]* will match 'AA'). View this example [online](https://regex101.com/r/xQ9vA4/1) – Giuseppe Ricupero Nov 11 '15 at 21:42
I see. Thanks for the link, the page seems very useful. – G. Tartifola Nov 11 '15 at 21:55
@GsusRecovery maybe I misread the OP, but I thought that's what they wanted – shadowtalker Nov 11 '15 at 21:56

score 1 · Answer 3 · answered Nov 11 '15 at 21:11

1

If I understand you correct you want to match patterns that starts with at least one E and then continues with at least one D until the end.

echo "1,AAEDDEED,1\n2,EEEEDDDD,2\n3,EDEDEDED" | gawk -F, '{if($2 ~ /^E+D+$) print $0}'

answered Nov 11 '15 at 21:11

Roger Lindsjö

11,330
1
42
53

Gawk regexp to select sequence

3 Answers3