%{
#include "mp4.tab.h"
  void print_unknown();
  void print_string();
  void print_error(char* type);

  unsigned int line_number=1;
  unsigned int column_number=1;

  unsigned int string_line_number;
  unsigned int string_column_number;

  char current_string[5000];
%}

digit [0-9]
letter [A-Za-z]
eol \n

%x multiline_comment
%x string_type

%%

char { column_number += yyleng; return CHAR; }  
class { column_number += yyleng; return CLASS; } 
else { column_number += yyleng; return ELSE; } 
if { column_number += yyleng; return IF; } 
int { column_number += yyleng; return INT; } 
private { column_number += yyleng; return PRIVATE; } 
public { column_number += yyleng; return PUBLIC; } 
return { column_number += yyleng; return RETURN; } 
void { column_number += yyleng; return VOID; } 
while { column_number += yyleng; return WHILE; } 
cin { column_number += yyleng; return CIN; } 
cout { column_number += yyleng; return COUT; } 

({letter}|_)({letter}|{digit}|_)* { column_number += yyleng; return ident; }

{digit}+ { column_number += yyleng; return num; }

"<<" { column_number += yyleng; return doublelessthan; } 
"->" { column_number += yyleng; return arrow; } 

"{" | 
"}" | 
"[" | 
"]" | 
"(" | 
")" | 
";" | 
":" | 
"." | 
"+" | 
"*" | 
"/" | 
"%" |
"&" | 
"|" | 
"!" | 
"=" | 
"<" | 
">" | 
"-" |
"," { column_number += yyleng; return yytext[0]; } 

\'[^\\]\' |
\'\\[\'\"\?\\fnrt]\' |
\'\\[0-7]{1,3}\' { column_number += yyleng; return charconst; }

\'\\. { print_error("Invalid escape sequence"); column_number += yyleng; }
\'. { print_error("Invalid character"); column_number += yyleng; }

\" BEGIN(string_type); { string_column_number = column_number; string_line_number = line_number; column_number++; current_string[0] = '\0'; }

<string_type>([^\\\"\n]|\\[\'\"\?\\fnrt]|\\[0-7]{1,3}) { strcat(current_string, yytext); column_number += yyleng; }
<string_type>\\{eol} { column_number = 1; line_number++; }
<string_type>\" BEGIN(INITIAL); { column_number++; return string; }

<string_type>\\. { print_error("Invalid escape sequence"); column_number += yyleng; }
<string_type>{eol} { print_error("Invalid newline within string"); column_number += yyleng; }

{eol} { column_number = 1; line_number++; }

"/*" BEGIN(multiline_comment); { column_number += 2; }

<multiline_comment>{eol} { column_number = 1; line_number++; }
<multiline_comment>"*/" BEGIN(INITIAL); { column_number += 2; }
<multiline_comment>. { column_number++; }

"//".*{eol} { column_number = 1; line_number++; }

[ \t\f] { column_number++; }

. { print_unknown(); column_number++; }

%%

void print_unknown()
{
  printf("%-11s %4d   %4d   \\%o\n", 
	 "unknown",
	 line_number,
	 column_number,
	 yytext[0]);
}

void print_string()
{
  printf("%-11s %4d   %4d   %s\n", 
	 "string",
	 string_line_number,
	 string_column_number,
	 current_string);
}

void print_error(char* type)
{
  printf("%-11s %4d   %4d   %s\n", 
	 "ERROR",
	 line_number,
	 column_number,
	 type);
}