Lex and Yacc issue with comments

问题

I am trying to locate the root cause of an issue. I have the following line that needs to be parsed -

sample format "string";

Where sample and format need to be tokenized and whatever is in the inverted commas needs to be provided to the Parser file. There is a catch however, if I have a perl style comment # inside the string, then I get an error.

In the lexer.l, I have the following -

stringIdentifier [^"]+

<STRING_S>{stringIdentifier} {
    strncpy(yylval.str, yytext,1023);
    yylval.str[1023] = '\0';
  return IDENTIFIER;
}

<*>"//".*  {
}

<*>"#".* {
} 

<INITIAL>{s}{a}{m}{p}{l}{e} {
    BEGIN(SAMPLE_S);
    return SAMPLE;
}

<SAMPLE_S>{f}{o}{r}{m}{a}{t} {
    return FORMAT;
}
<SAMPLE_S>"\"" {
    BEGIN(STRING_S);
    return INVERTED_COMMA;
}

<STRING_S>"\"" {
    BEGIN(INITIAL);
    return INVERTED_COMMA;
}

In the Parser.y I have the following rule:

pass : SAMPLE FORMAT INVERTED_COMMA IDENTIFIER INVERTED_COMMA 
{

};

However, when I give sample format "abc;" it works, however, when I add a comment character # in the string it fails. Could you please help with this

回答1:

The answer lies in the way you have used the default start conditions. A quick read of the lex/flex manual explains their operation.

The <*> means that the following pattern is applied in every state. This includes inside a string, which is indicated by the S_STRING state. To stop the comment pattern operating inside the string you need to exclude the S_STRING state from <*>. You can do this by listing all the other applicable states, which enumerate, in your example, to <INITIAL,S_SAMPLE>. The comment rules then become:

<INITIAL,SAMPLE_S>"//".* {
}

<INITIAL,SAMPLE_S>"#".* {
}

And that's it. It now works! (I have tested it BTW)

来源：https://stackoverflow.com/questions/20532346/lex-and-yacc-issue-with-comments

标签

lex