Question

Is it possible to write a lex or flex script containing a set of rules where the maximal-munch matching step will always match a length-zero string?

I ask this because the mathematical definition of regular expressions permits regular expressions that only match zero-length strings. For example, the simple regular expression ε only matches the empty string. However, the flex pattern documentation doesn’t seem to allow you to specify regular expressions of this form.

Answer 1

I don’t know about lex, but flex can match the empty string by using a trailing context:

""/. { /* handle empty string */ }

This will match an empty string before any character, but it can be almost any pattern (see flex documentation about limitations).

<INITIAL>""/[-]{2} { BEGIN(sqlcomment); }
<sqlcomment>[^\n]* { /* Handle sql comment, including -- */ BEGIN(INITIAL); }

If you want to match the end of the input, you have to use the special symbol <<EOF>> (see flex documentation about eof).

EDIT: Now that i think a little more about it, the trailing context might not be mandatory. Just “” may work. If it does, i still highly recommand to use a trailing context in this case when possible. It prevents many side effects with malformed inputs and helps debugging.