Re: How do I handle multiple line quoted strings in fLex?

Tim Perry <tperry@blinksoft.com> wrote:
>I wan't to be able to handle multiple line strings in flex, where I return
>one token to bison of that string minus the quotes. I have a pattern such
>as this that currently works, with a few snags...
>
>\"[^\"]*\" { yyval.string = strdup(yytext[yyleng]); RETURN STRING }
>
>This doen't get rid of the " nor does it supply a method of counting the
>current line numbers since the \n rule that I have will not get executed.
>
>I have thought about makeing a STRING state that when it encounters the
>first " I initialize a state that accepts everything besides a \n, I
>discard the \n since I don't want them in my string anyway, increase my
>linecounter and then strcat the yytext to a temp buffer till I get the
>closeing ".
>[I'd do what you suggest, use the STRING state to recognize a line at
>a time and buffer up the string myself, since it's unwise to expect
>lex to handle enormous single tokens. -John]




I was working a some parser which has similar string (String starts
with " and ending with ". It also takes any escape character.). I
used lex for lexical analyzer. I implemented the follwing way to
increment line number if '\n' occurs in the middle of the
string. Implementation is as follows.


\" {
                                                        yyleng = 0; c1=0; c2=0;
                                                        end_str = FALSE;
                                                        c1 = input();
                                                        c2 = c1;
                                                        while (!end_str) {
                                                                if (c1 == '"' && c2 != '\\') {
                                                                        end_str = TRUE;
                                                                }
                                                                yytext[yyleng++] = c1;
                                                                if (c1 == '\n') {
                                                                        lineno++;
                                                                }
                                                                c2 = c1;
                                                                c1 = input();
                                                        }
/*
* To remove the last quotation
*/
                                                        yytext[yyleng-1] = '\0';
                                                        yylval.sliteral = (char *)strdup(yytext);
                                                        unput(c1);
                                                        RETURN_VAL(LITERAL);
                                                }


It is giving correct line numbers. I hope I understand your problem correctly.


Good luck
Haren Myneni
[This won't work in flex nor any other version of lex where yytext isn't an
array. -John]




--

Return to the comp.compilers page.
Search the comp.compilers archives again.