sed matching within a line

[hailholyghost]

Hello all :guitar:

I want to extract numbers/coordinates from files named like

optppa18X270.49E179.970Z180.860.out

like grep would do, except within a single line.

For example, I would like to print after ppa and before the letter E, after E and before Z, and after Z but before .out.

I cannot find how to do this with sed or awk. If there is a way, how can I do it?

thanks for your expertise!:popcorn:
-Dave

[hailholyghost]

I actually solved this on my own, the first coordinate is shown as follows:


#!/bin/bash

#extract energies from a dihedral scan

for i in $(ls optppa*.out)
do
PPA="$(ls $i | sed -e 's/^o.*a//g' -e 's/X.*out//g')"
ENERGY="$(grep RwB97XD optppa18X260.49E179.970Z180.860.out | awk '/SCF/{print $5}')"
echo $PPA

done

[sisco311]

ls shows the representation of files. They are not file names. For simple names, they mostly happen to be equivalent. So, do NOT try to parse it. See: http://mywiki.wooledge.org/ParsingLs

In most cases you can use parameter expansion instead of echo "$var" | sed .... See: http://mywiki.wooledge.org/BashSheet#Parameter_Operations

There is no need to pipe grep in awk: awk '/FOO/ && /BAR/{whatever}' file

You should double quote every expansion and anything that could possibly contain a special character: echo "$ppa". See: http://mywiki.wooledge.org/Quotes

By convention, we capitalize environment variables (PAGER, EDITOR, SHELL, ...) and internal shell variables (BASH_VERSION, RANDOM, ...). All other variable names should contain at least one lowercase letter. Remember that variable names are case-sensitive; this convention avoids accidentally overriding environmental and internal variables.




for f in optppa*.out
do
f="${f#optppa}"
ppa="${f%%X*.out}"
energy="$(awk '/RwB97XD/&&/SCF/{print $5}' "$f")"
echo "$ppa"
done

[Vaphell]

if you want to extract multiple parts from string with single sed you can use subgroups


$ echo optppa18X270.49E179.970Z180.860.out | sed -r 's/[a-zA-Z]*(.*)X(.*)E(.*)Z(.*)[.]out/\1 : \2 : \3 : \4/'
18 : 270.49 : 179.970 : 180.860


each pair of parentheses gets an index that can be used to call it in the replacement part
we have 4 pairs here so \1 \2 \3 and \4

[sisco311]

if you want to extract multiple parts from string with single sed you can use subgroups


$ echo optppa18X270.49E179.970Z180.860.out | sed -r 's/[a-zA-Z]*(.*)X(.*)E(.*)Z(.*)[.]out/\1 : \2 : \3 : \4/'
18 : 270.49 : 179.970 : 180.860


each pair of parentheses gets an index that can be used to call it in the replacement part
we have 4 pairs here so \1 \2 \3 and \4

Nice. And the output can be stored in an array:

{ IFS=: read -a arr < <(echo optppa18X270.49E179.970Z180.860.out | sed -r 's/[a-zA-Z]*(.*)X(.*)E(.*)Z(.*)[.]out/\1 : \2 : \3 : \4/'); }
echo "${arr[0]}"
echo "${arr[1]}"
echo "${arr[2]}"
echo "${arr[3]}"

[hailholyghost]

Thank you so much sisco311 and Vaphell!

sisco: I took to hear the grep | awk advice you gave to me a while ago, and I implemented it, but I didn't know that they could be linked together with &&. I guess you learn new things every day!

Vaphell: thanks for pointing the subgroups out.

I will study the links and advice you've given. I appreciate your time!