# If you'd like to omit non-matching lines from the result; add ';d' to the end of the expression.
sed -E 's/^((?!(\?)|(\[))(.+(\n|$)))+//gm;t' <<< "this is paragraph line1 line 1
and line 2
?[question_2]{\"category\":[]}[who are you]=[] {1,2}
[] OPTION 1
[] OPTION 2
[] OPTION 3
this is paragraph line3
with a line break
?[question_3]{\"category\":[]}[picode]=_ [PLACEHOLDER_TEXT]
?[question_1]{\"category\":[]}[sex]=() {1}
() male
() femele
this is paragraph line3
in the end of file"
Please keep in mind that these code samples are automatically generated and are not guaranteed to work. If you find any syntax errors, feel free to submit a bug report. For a full regex reference for SED, please visit: https://www.gnu.org/software/sed/manual/html_node/The-_0022s_0022-Command.html