Regular expression matches any word except for some words in flex

Question

I would like to write a regular expression that matches any word. I use [a-zA-Z]* except for some words, for example WORD1 and WORD2.

So somethingsomething matches, but the words WORD1 and WORD2 won't. Is it possible in flex?

I have tried:

[a-zA-Z]*|[^"WORD1""WORD2] and [a-zA-Z]*{-}["WORD1""WORD2"] but neither works.

(Now I know why they don't work but I still don't know the solution.)

[Lookaheads](http://www.regular-expressions.info/lookaround.html) are not available in flex regex are they? — bobble bubble, Feb 29 '16 at 10:16
@bobblebubble: [It doesn't look good.](http://stackoverflow.com/q/22326399/20670) — Tim Pietzcker, Feb 29 '16 at 10:41
I really don't understand what you're asking here. `[a-zA-Z]*` won't match `WORD1`, although it will match `WORD`. What did you want to happen when WORD1 is encountered? An error? A different token type? Two tokens? Only if we know what you want the result of scanning `WORD1` to be can we provide a suggestion for how to implement it. — rici, Mar 01 '16 at 00:28
What do you mean by "any word"? As in, any English word, in which case you need to allow for apostrophes (or you *won't* match words like "don't")? — nnnnnn, Mar 02 '16 at 09:46

score 1 · Answer 1 · answered Mar 02 '16 at 09:40

The usual approach in lex/flex is to use a combination of patterns and rules to select the desired behavior:

You could make a lexer which matches (and ignores) those words and then falls through to the expression for other identifiers, or
Simply match all identifiers and filter out the results with a lookup table.

score 1 · Answer 2 · answered Mar 03 '16 at 11:55

It is possible to write a regular expression for the situation you presented.

In order to match all words except word you can write:

w|wo|wor|word[a-z]+|([^w]|w[^o]|wo[^r]|wor[^d])[a-z]*

But as @Thomas and @rici pointed out, you have much better solutions (specially when you define a concrete problem).

Example: "count all words except the word word" is in fact very simple (using Thomas proposal):

%%
  int i;

word        {    }
[a-z]+      {i++;}

.|\n        {    }
<<EOF>>     { printf("%d\n",i); return 0; }
%%

(untested)

2 Answers2