Is it possible to set priorities for rules to avoid the “longest-earliest” matching pattern?

Question

Another simple question : is there any way to tell flex to prefer a rule that matches a short thing over a rule that matches a longer thing ? I can\'t find any good documentatio

Answer 1

From the Dragon Book 2nd edition, Section 3.5.3 "Conflict Resolution in Lex":

We have alluded to the two rules that Lex uses to decide on the proper lexeme
to select, when several prefixes of the input match one or more patterns:
    1. Always prefer a longer prefix to a shorter prefix.
    2. If the longest possible prefix matches two or more patterns, prefer the
       pattern listed first in the Lex program.

The rule above also applies to Flex. Here is what the Flex manual says (Chapter 7: How the input is matched.)

When the generated scanner is run, it analyzes its input looking for strings 
which match any of its patterns. If it finds more than one match, it takes the 
one matching the most text (for trailing context rules, this includes the length 
of the trailing part, even though it will then be returned to the input). If it 
finds two or more matches of the same length, the rule listed first in the flex 
input file is chosen.

If I understood correctly, your lexer treats keywords like Endif as an identifier, so it will be considered as part of an expression afterwards. If this is your problem, simply put the rules of keywords on top of your specification, such as the following: (suppose each word in uppercase is a predefined enum corresponding to a token)

"If"                      { return IF;         }
"Then"                    { return THEN;       }
"Endif"                   { return ENDIF;      }
"While"                   { return WHILE;      }
"Do"                      { return DO;         }
"EndWhile"                { return ENDWHILE;   }
\"(\\.|[^\\"])*\"         { return STRING;     }
[a-zA-Z_][a-zA-Z0-9_]*    { return IDENTIFIER; }

Then the keywords will always matched before the identifier due to Rule No. 2.

EDIT:

Thank you for your comment, kol. I forgot to add the rule for string. But I don't think my solution is wrong. for example, if an identifier called If_this_is_an_identifier, rule 1 will apply, thus the identifier rule will take effect (Since it matches the longest string). I wrote a simple test case and saw no problem in my solution. Here is my lex.l file:

%{
  #include <iostream>
  using namespace std;
%}

ID       [a-zA-Z_][a-zA-Z0-9_]*

%option noyywrap
%%

"If"                      { cout << "IF: " << yytext << endl;         }
"Then"                    { cout << "THEN: " << yytext << endl;       }
"Endif"                   { cout << "ENDIF: " << yytext << endl;      }
"While"                   { cout << "WHILE: " << yytext << endl;      }
"Do"                      { cout << "DO: " << yytext << endl;         }
"EndWhile"                { cout << "ENDWHILE: " << yytext << endl;   }
\"(\\.|[^\\"])*\"         { cout << "STRING: " << yytext << endl;     }
{ID}                      { cout << "IDENTIFIER: " << yytext << endl; }
.                         { cout << "Ignore token: " << yytext << endl; }

%%

int main(int argc, char* argv[]) {
  ++argv, --argc;  /* skip over program name */
  if ( argc > 0 )
    yyin = fopen( argv[0], "r" );
  else
    yyin = stdin;

  yylex();
}

I tested my solution with the following test case:

If If_this_is_an_identifier > 0 Then read(b); Endif
    c := "If I were...";
While While_this_is_also_an_identifier > 5 Do d := d + 1 Endwhile

and it gives me the following output (other output not relevant to the problem you mentioned is ignored.)

IF: If
IDENTIFIER: If_this_is_an_identifier
......
STRING: "If I were..."
......
WHILE: While
IDENTIFIER: While_this_is_also_an_identifier

The lex.l program is modified base on an example from the flex manual: (which use the same method to match keyword out of identifiers)

Also have a look at the ANSI C grammar, Lex specification.

I also used this approach in my personal project, and so far I didn't find any problem.