Get value from string based on RegEx

Question

I have a bash script that I am trying to learn from. I have a sql file that has in it:

create user myuser with password 'mypassword';

I want to use sed and store the value of the password in one bash variable and the pasword in another bash variable. The script that I am learning from has:

USERNAME := $(shell sed -n \
"s/^create user \(\S\+\) with password '\(\S\+\)';$$/\1/p" \
createdb.sql)
PASSWORD := $(shell sed -n \
"s/^create user \(\S\+\) with password '\(\S\+\)';$$/\2/p" \
createdb.sql)

No matter what I tried the regex would not match up. Part of my issue is I don't understand the basics of sed. How do I go about getting the value of parts of the string?

chepner · Accepted Answer · 2017-02-23 15:38:59Z

2

Don't use sed for this; use bash's built-in regular expression support.

str="create user myuser with password 'mypassword';"
regex='create user (.*) with password (.*);'
if [[ $str =~ $regex ]]; then
    username=${BASH_REMATCH[1]}
    password=${BASH_REMATCH[2]}
fi

As for how you get the appropriate line in the first place, use grep:

str=$(grep "create user" createddb.sql)

edited Feb 23, 2017 at 15:38

answered Feb 23, 2017 at 15:34

chepner

538k77 gold badges594 silver badges746 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

Dovid Bender Over a year ago

$str in this case would be the file?

clt60 Over a year ago

hm... using external grep and after bash builtin regex, instead of one external sed + read?

chepner Over a year ago

You could, but I find this a little simpler. It essentially uses a multi-character delimiter with password to separate the two values, rather than trying to come up with a safe single-character delimiter to use in IFS to let read resplit the output of sed.

chepner Over a year ago

As long as you could find a character that doesn't appear in the user (which in this case is probably trivial), something like IFS=% read user passwd < <(sed 's/create user (.*) with password (.*)/\1%\2/') would be fine. On the other extreme, you could even forgo the call to grep by using a while loop to read the file line by line, but I suspect grep would be faster unless you have a very small file or the line in question is found early. grep + =~ just happens to be (IMO) the simplest balance between two extremes.

Collectives™ on Stack Overflow

Get value from string based on RegEx

1 Answer 1

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related