Preprocessor fails on escaped double quote within quotes

[tored]

From the IUP header file iupkey.h line 15

https://sourceforge.net/p/iup/iup/HEAD/tree/trunk/iup/include/iupkey.h

$parser = new  \PHPCParser\PreProcessor\Parser();
$parser->parse('iupkey.h', <<<HEADER
#define K_quotedbl    '\"'  /* 34 */
HEADER);`

Fatal error: Uncaught RuntimeException: Unterminated " in PreProcessor\Parser.php:112

Removing the block comment generates a different error

$parser = new  \PHPCParser\PreProcessor\Parser();
$parser->parse('iupkey.h', <<<HEADER
#define K_quotedbl    '\"'
HEADER);`

Fatal error: Uncaught LogicException: Unknown character literal escape sequence: '\\"' in PreProcessor\Tokenizer.php:101

[ircmaxell]

Is that even legal syntax?

Anyway, you can add a branch to this switch to check for the double quote: https://github.com/ircmaxell/php-c-parser/blob/master/lib/PreProcessor/Tokenizer.php#L93-L102

If that works (for the second error at least).

As far as the first error, it's because it needs to parse comments as part of comment stripping. https://github.com/ircmaxell/php-c-parser/blob/master/lib/PreProcessor/Parser.php#L100-L130

Basically, an if block would need to be added to intercept ' characters, and skip those as well...

[tored]

I'm far from an expert on C & the preprocessor, but if I'm reading the C standard correctly it should be legal to escape double quote like that in the preprocessor

http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1256.pdf

Define statement is

# define identifier replacement-list new-line

and replacement-list expands thru pp-tokens, preprocessing-token, character-constant, c-char-sequence, c-char, escape-sequence and finally to simple-escape-sequence and that is one of

\' \" \? \\ \a \b \f \n \r \t \v

Caveat, I normally don't read the C standard.

Given this test program

#include <stdio.h>
#define K_quotedbl    '\"'  /* 34 */

int main()
{
    printf("%c", K_quotedbl);
    return 0;
}

gcc -E -std=c89 -pedantic -Wall -Wextra macro.c

expands macro to

printf("%c", '\"');

That program will print double quote, Removing the escape of the double quote in the macro, the resulting program will expand to the following & but print the same result.

printf("%d", '"');

I see if I maybe have sometime next week to make a pull request to at least fix the double quote.

[ircmaxell]

Well, in that case, perhaps the first block should just eat any escape sequence. So the switch would also need cases for a, b, f, n, r, t, v. Something like;

                    switch ($buffer[1]) {
                        case '0':
                            $value = chr(0);
                            break;
                        case 'a':
                            $value = chr(0x07);
                            break;
                        case 'b':
                            $value = chr(0x08);
                            break;
                        case 'f':
                            $value = chr(0x0C);
                            break;
                        case 'n':
                            $value = chr(0x0A);
                            break;
                        case 'r':
                            $value = chr(0x0D);
                            break;
                        case 't':
                            $value = chr(0x09);
                            break;
                        case 'v':
                            $value = chr(0x0B);
                            break;
                        case 'x':
                            $value = chr(intval(substr($buffer, 2), 16));
                            break;
                        case 'n':
                            $value = chr(intval(substr($buffer, 2), 8));
                            break;
                        default: 
                            // default to the literal
                            $value = $buffer[1];
                    }

Thanks for digging into this!