Transparent handling of #line directives in input files

[skvadrik]

I was playing around with a project where I pass the re2c input file through a preprocessor (m4) first. For best results in this situation, it’s best if re2c can handle #file directives in its input file.

The attached patch to 0.13.3 accomplishes this. While it takes reasonable precautions against false matches for #line directives, it does not strictly check for the directives to start at the beginning of a line (as far as I know, re2c does not yet support the ^pattern syntax, does it?)

While I tried to match the general style of re2c’s source code, I’m open to reworking the patch to meet any further standards I may have missed.

Original comment by: neeri

[skvadrik]

Original comment by: neeri

[skvadrik]

Logged In: YES
user_id=271023
Originator: NO

I think it is a good idea and I am going to add it. However I need to change your patch a bit. And I will need a few tests for it. Also a patch for documentation might be good. Dunno how right now, tough.

Original comment by: helly

[skvadrik]

- priority: 5 —> 3
- assigned_to: nobody —> helly
- status: open —> open-accepted

Original comment by: helly

[skvadrik]

Logged In: YES
user_id=271023
Originator: NO

I’ve added this feature along with an initial test. Turned out to be much more complex than you thought, when implementing it completely.

Original comment by: helly

[skvadrik]

- status: open-accepted —> closed-fixed

Original comment by: helly

[pskocik]

This doesn't appear to be working. I tried to use lemon output (with #line directives) as re2c input and large parts of the generated source got deleted. If I try feeding the output of running re2c on

typedef enum num_t { ERR, BIN, OCT, DEC, HEX } num_t;
static num_t lex(const char *YYCURSOR)
{
	const char *YYMARKER;
	/*!re2c
		re2c:define:YYCTYPE = char;
		re2c:yyfill:enable = 0;

		end = "\x00";
		bin = '0b' [01]+;
		oct = "0" [0-7]*;
		dec = [1-9][0-9]*;
		hex = '0x' [0-9a-fA-F]+;

		*       { return ERR; }
		bin end { return BIN; }
		oct end { return OCT; }
		dec end { return DEC; }
		hex end { return HEX; }
	*/
}

(from the docs) back to re2c with

re2c example.c.re |re2c -

I get

/* Generated by re2c 1.0.3 on Mon Jun 25 20:48:07 2018 */
#line 1 ""

}

(I'm on an older browser and I'm having some difficulties with the commenting tool, so apologies for the formatting.)

[skvadrik]

Pushed a fix: 96aea70, thanks for reporting! The correct behaviour was broken between 0.16 and 1.0.