QCaml/Nuclei/Xyz_parser.mly

/* Parses nuclear coordinates in xyz format */

%{
exception InputError of string
%}

%token EOL
%token <string> SPACE
%token <string> WORD
%token <int> INTEGER
%token <float> FLOAT
%token EOF

%start input
%type <(Element.t * Coordinate.t) array> input

%% /* Grammar rules and actions follow */

input:
  | integer title atoms_xyz { 
     let len = List.length $3 in
     if len <> $1 then
       let error_msg = Printf.sprintf "%d atoms entered, expected %d" len $1 in
       raise (InputError error_msg)
     else
        Array.of_list $3
     }

integer:
  | INTEGER EOL { $1 }
  | INTEGER SPACE EOL { $1 }
  | SPACE INTEGER EOL { $2 }
  | SPACE INTEGER SPACE EOL { $2 }

title:
  | title_list EOL { $1 }

text:
  | WORD    { $1 }
  | SPACE   { $1 }
  | FLOAT   { (string_of_float $1)}
  | INTEGER { (string_of_int $1)}

title_list:
  | { "" }
  | title_list text { ($1 ^ $2) }

atoms_xyz:
  | atoms_list EOL { List.rev $1 }
  | atoms_list EOF { List.rev $1 }


atoms_list:
  | { [] }
  | atoms_list WORD    SPACE FLOAT SPACE FLOAT SPACE FLOAT       EOL { (Element.of_string $2, Coordinate.of_3_floats $4 $6 $8 `Angstrom ) :: $1 }
  | atoms_list WORD    SPACE FLOAT SPACE FLOAT SPACE FLOAT SPACE EOL { (Element.of_string $2, Coordinate.of_3_floats $4 $6 $8 `Angstrom ) :: $1 }
  | atoms_list INTEGER SPACE FLOAT SPACE FLOAT SPACE FLOAT       EOL { (Element.of_int $2, Coordinate.of_3_floats $4 $6 $8 `Angstrom ) :: $1 }
  | atoms_list INTEGER SPACE FLOAT SPACE FLOAT SPACE FLOAT SPACE EOL { (Element.of_int $2, Coordinate.of_3_floats $4 $6 $8 `Angstrom ) :: $1 }
  | atoms_list SPACE WORD    SPACE FLOAT SPACE FLOAT SPACE FLOAT       EOL { (Element.of_string $3, Coordinate.of_3_floats $5 $7 $9 `Angstrom ) :: $1 }
  | atoms_list SPACE WORD    SPACE FLOAT SPACE FLOAT SPACE FLOAT SPACE EOL { (Element.of_string $3, Coordinate.of_3_floats $5 $7 $9 `Angstrom ) :: $1 }
  | atoms_list SPACE INTEGER SPACE FLOAT SPACE FLOAT SPACE FLOAT       EOL { (Element.of_int $3, Coordinate.of_3_floats $5 $7 $9 `Angstrom ) :: $1 }
  | atoms_list SPACE INTEGER SPACE FLOAT SPACE FLOAT SPACE FLOAT SPACE EOL { (Element.of_int $3, Coordinate.of_3_floats $5 $7 $9 `Angstrom ) :: $1 }