Is it possible to use PCRE2 on variables in bash-CodePudding

I'm trying to use PCRE2 on stdin in a bash script but I'm not sure it's possible based on reading the docs. I know I can give it the filename but I'm doing a test and it specifically says on stdin.

For example I have this text in input.txt

foo blah is a bar
this shouldn't work

I then call this command: cat input.txt | sh program.sh "foo %{0} is a %{1S3}"

Which fires this shell script:

#!/bin/bash

# Give access to Elixir module
chmod  x ./lib/my_mod.ex

# Get arg passed to program.sh
token=$1

# Create array for cat and insert into it
array=()
while IFS= read line; do
array =("${line}")
        done

# Get regex based on pattern sequence
regex=$(mix run -e "my_mod.main('${token}')")

pcre "${regex}" ./input.txt

# Call pcre2grep on each item and match with regex
# for value in "${array[@]}"
# do
#      echo "${value}"
#      pcre2grep "${regex}" "${value}"
# done

This works: pcre "${regex}" ./input.txt

This doesn't

# Call pcre2grep on each item and match with regex
    for value in "${array[@]}"
    do
         echo "${value}"
         pcre2grep "${regex}" "${value}"
    done

I'm supposed to use PCRE (Or supposedly PCRE2) but I feel like I'm doing something stupid.

Any ideas on if this is possible using PCRE?

Thanks Nick

Update

If I hard code the regex returned from $(mix run -e "my_mod.main('${token}')") it doesn't work but if I hardcode the regex, it does work:

pcre2grep "foo [a-zA-Z ]{0,} is a \W*(\w (\W |$)){1,4}$"

Calling Mix is affecting cat input.txt | sh program.sh "foo %{0} is a %{1S3}"

CodePudding user response：

Just run it, just like grep:

regex=$(mix run -e "my_mod.main('${token}')" <&-)
pcre2grep "${regex}"

It happens that mix is reading from standard input, leaving nothing for pcre2grep to read. Close standard input for the process with <&-.