Grep with regex in CLI returns nothing

Question

I have the following file.

//TESTCASES=3
//MARK=9

[runscript]
nc dec s10 s11
[/runscript]

[checks]

[/checks]

[testcase]
// List: 1, 2, 3, 5, 0xA, -1
.global LIST
.data
LIST: 
.word 1, 2, 3, 5, 0xA, -1
[/testcase]

I am trying to get the text between [runscript] and [/runscript] using grep and regex.

I have verified that the regex works on its own.

(?<=\[runscript\]\n)(.|\n)*(?=\[\/runscript\])

However, grep returns nothing. Is it an issue of the options? I have tried many of them alone and combined.

-P, -e, -E, -w, -o

What am I missing?

Answer 1

grep works on one line at a time, so the thing the regex is matched against will never contain a newline character.

Since you want to use perl/PCRE operators, you could use pcregrep which has a multi-line mode with -M and can be told to report part of a match:

pcregrep -xMo1 '(?s)\[runscript\]\n(.*?)\n\[/runscript]'

Or:

pcregrep -xMo1 '(?s)\[runscript\]\r?\n(.*?)\n\[/runscript]\r?'

To allow MS-DOS style CRLF line delimiters (here the CRs are included in the output if present in the input).

Answer 2

I wouldn't use grep but rather awk:

awk '
    $0=="[runscript]" {rs++; next}
    $0=="[/runscript]" {rs--}
    rs {print}
' file

Output

nc dec s10 s11

If you really want to use grep, this will work with PCREs and NUL-delimited data. But I would suggest it's harder for people to understand (and maintain) then the awk version, and less portable too:

grep -zoP '(?<=\[runscript\]\n)(.|\n)*?(?=\[\/runscript\])' file

Output (with an invisible trailing NUL)

nc dec s10 s11

I've tweaked your RE to cope with multiple matches should there be any.