How can I create a regex validation expression to allow this string? -

i have regex validation expression (thanks clasg , marvel308):

^((?p\begin{(?\w+)}\s*(?:-?\d+\s*(?:&\s*|\\\n))+\end{\g})|\$\s*(?p>matrix)\s*\$|\$\$\s*(?p>matrix)\s*\$\$|\[\s*(?p>matrix)\s*\]|\(\s*(?p>matrix)\s*\))$

at beginning of each row contain -?\d's allow (but not require) word "\left(". , @ end of each row contain -?\d's allow (but not require) word "\right)" how can edit validation expression satisfy this?

here's regex handles cases:

 (?:(?:\$\$?|\\[$[(]?)\s*|^)\\begin{(\w+)}\s*(?:-?\d+\s*(?:&\s*|\\\\\n))+\\end{\1}\s*(?:\$\$?|\\[]$)]?)?$ 

see here @ regex101.

edit

here's solution allows matching opening/closing *tags:

^((?p<matrix>\\begin{(?<token>\w+)}\s*(?:-?\d+\s*(?:&\s*|\\\\\n))+\\end{\g<token>})|\$\s*(?p>matrix)\s*\$|\$\$\s*(?p>matrix)\s*\$\$|\\\[\s*(?p>matrix)\s*\\\]|\\\(\s*(?p>matrix)\s*\\\))$ 

it uses subroutine - matrix - keep bit shorter.

here @ regex101.

here's explanation:

the gut of part of recognizes matrix:

\begin{name} 1 & 2 & 3 \\ 1 & -2 & 4 \\ 1 & 2 & 5 \\ \end{name} 

this done part

\\begin{(\w+)}\s*(?:-?\d+\s*(?:&\s*|\\\\\n))+\\end{\1} 

which matches backslash followed literal text begin. matches name of matrix inside curly brackets, capturing group one. thereafter matches number of positive or negative numbers (-?\d+) followed either ampersand (&) or 2 backslashes , newline (\\\\\n). these rows may repeat number of times. matches ending backslash , name (captured group 1) inside curly brackets \\end{\1}.

the first version allows (not mandatory) preceded , followed lines consisting of $, $$, \( , \[. no check made enveloping lines match each other.

the second version starts testing naked matrix using gut regex explained above, , doing stores regex subroutine named "matrix" - (?p<matrix>. difference between previous gut regex name of regex stored in named capture group - token. text end of matrix done \\end{\g<token>}) checking literal string inside curly brackets matches part captured group named token.

the test of naked regex followed alternations having allowed surrounding lines correct combinations of characters , using subroutine matrix between them. e.g. alternations

\$\s*(?p>matrix)\s*\$ 

checks (one) dollar sign, optional white space, actual matrix (by calling subroutine) followed matching terminating line, i.e. single dollar sign.