# If you'd like to omit non-matching lines from the result; add ';d' to the end of the expression.
sed -E 's/^((\w[^\W]+)[\.\-]?){1,}\@(([0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3})|(([a-zA-Z\-0-9]+\.)+[a-zA-Z]{2,}))$//gm;t' <<< "JOAO!DA.SILVA@CNN,COM (INVALID)
joao@microsoft.com
mauricio.teste@site.ba.gov.br
santana.teste@sistema-ba.ba.gov.br
teste.teste.teste.igor.caza@google.com
teste.teste.teste.igor.caza@globo.com.br
ayslan-johnson@teste.com.br
joao@123.com
ayslanjohnson@google.com
teste.teste.teste.igor.caza@globo (INVALID)
SILVA.teste@SITE-ba.ba.gov.br. (INVALID)"
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