[plt-scheme] lex problem

On Tue, Sep 29, 2009 at 2:22 PM, Jon Rafkind <rafkind at cs.utah.edu> wrote:
> Stephen Bloch wrote:
>>
>> I'm trying to teach myself to use the scanner and parser tools in time to
>> provide my students a parser for a C-like language so they can write an
>> interpreter for it.
>>
>> I'm calling
>> (lexer (some-ws (get-token input-port))
>>         ((eof) 'eof)
>>         ...
>>         (#\+ (token-plus-operator))
>>         ...
>>         ((concatenation #\" (repetition 0 +inf.0 (complement #\")) #\")
>>          (token-str-lit-token (chop-fl lexeme)))
>>         )
>>
>> So this should recognize anything that starts with a quotation mark, ends
>> with a quotation mark, and doesn't contain any other quotation marks, as a
>> "str-lit-token".
>>
>> I call the lexer on the input "\"hello\" + \"there\""
>> and it gives me back a single str-lit-token with value "hello\" + \"there"
>> rather than (as I expected) a str-lit-token, a plus-operator, and another
>> str-lit-token.
>>
>> What gives?
>>
>
> Sort of a wild guess but does this work instead?
>
> replace (complement #\") with
>
> (intersection (complement #\") any-char)

Or replace complement with char-complement.

--Carl