URI.g4

// Simple URI parser with ANTLR4
// Omer Faruk Zorlu, Jun 2021
grammar URI;

prog                    : (segment)+  ;
segment                 : protocol? host path?                                                          ;
protocol                : VALID_CHARSET+ PROTOCOL_DELIMITER                                             ;
host                    : VALID_CHARSET+ ~'/'?                                                          ;
path                    : (VALID_CHARSET | PATH_DELIMITER)+ query?                                      ;
query                   : VALID_CHARSET? QUERY_DELIMITER query_name_value                               ;
query_name_value        : ((VALID_CHARSET (QUERY_NVP_DELIMITER VALID_CHARSET)?) QUERY_PRM_DELIMITER?)+  ;

QUERY_DELIMITER         : '?'                           ;
QUERY_PRM_DELIMITER     : '&'                           ;
QUERY_NVP_DELIMITER     : '='                           ;
PATH_DELIMITER          : '/'                           ;
PROTOCOL_DELIMITER      : '://' '/'?                    ;
VALID_CHARSET           : (ALPHA | DIGIT | '.' | '-')+  ;
EOL                     : EOF | '\r'? '\n'              ;

fragment ALPHA          : [a-zA-Z]                      ;
fragment DIGIT          : '0'| [1-9]                    ;