interpret/src/lex.re_c.c

/** @license 2022 Neil Edelman, distributed under the terms of the
 [MIT License](https://opensource.org/licenses/MIT).

 Lexer for journal entries.

 @std C89/90 */

#include "../src/lex.h"
#include <stdlib.h>
#include <stdio.h>
#include <assert.h>
#include <limits.h>
#include <errno.h>

/*!re2c
re2c:yyfill:enable = 0;
re2c:define:YYCTYPE = char;
*/

int lex_looks_like_year(const char *const a, int *const year) {
	const char *YYCURSOR = a, *YYMARKER = a, *s0;
	/*!stags:re2c format = 'const char *@@;\n'; */
	assert(a && year);
	/*!re2c
	@s0 ("-"? [1-9][0-9]* | "0") "\x00" {
		int sign = 1, mag;
		if(*s0 == '-') { sign = -1; s0++; }
		for(mag = 0; *s0 != '\0'; s0++) {
			int d = *s0 - '0';
			if((INT_MAX - d) / 10 < mag) return 0;
			mag = mag * 10 + d;
		}
		*year = sign * mag;
		return 1;
	}
	* { return 0; }
	*/
}

int lex_looks_like_month(const char *const a) {
	const char *YYCURSOR = a, *YYMARKER = a, *s0;
	/*!stags:re2c format = 'const char *@@;\n'; */
	assert(a);
	/*!re2c
	@s0 [0-1][0-9] "\x00" {
		int val = 10 * (s0[0] - '0') + (s0[1] - '0');
		return val < 1 || val > 12 ? 0 : val;
	}
	* { return 0; }
	*/
}

int lex_looks_like_day(const char *const a) {
	const char *YYCURSOR = a, *YYMARKER = a, *s0;
	/*!stags:re2c format = 'const char *@@;\n'; */
	assert(a);
	/*!re2c
	@s0 [0-3][0-9] ".txt\x00" {
		int val = 10 * (s0[0] - '0') + (s0[1] - '0');
		return val < 1 || val > 31 ? 0 : val;
	}
	* { return 0; }
	*/
}

/* This defines `enum condition`. */
/*!types:re2c*/

/** scanner reads a file and extracts semantic information. Valid to access
 only while underlying pointers do not change. */
static struct scan {
	/* `re2c` variables; these point directly into `buffer`. */
	const char *marker, *ctx_marker, *from, *cursor;
	/* Weird `c2re` stuff: these fields have to come after when >5? */
	const char *label, *buffer;
	enum condition condition;
	size_t line;
	int is_ws_expected;
} scan; /* Terrible, gah. */

void lex_reset(const char *const buffer) {
	scan.marker = scan.ctx_marker = scan.from = scan.cursor = scan.label
		= scan.buffer = buffer;
	scan.condition = 0;
	scan.line = 1;
}

/*!re2c
re2c:flags:tags = 1;
re2c:define:YYCURSOR = scan.cursor;
re2c:define:YYMARKER = scan.marker;
re2c:define:YYCTXMARKER = scan.ctx_marker;
re2c:define:YYCONDTYPE = 'condition';
re2c:define:YYGETCONDITION = 'scan.condition';
re2c:define:YYGETCONDITION:naked = 1;
re2c:define:YYSETCONDITION = 'scan.condition = @@;';
re2c:define:YYSETCONDITION:naked = 1;

// Eof is marked by null when preparing files for lexing.
// Mutually exclusive; only !, [, are not covered.
end = "\x00";
newline = "\n" | "\r" "\n"?;
ws = [ \t\v\f];
glyph = [^ \t\n\r\v\f![\x00];
glyphs = glyph+;

// inside the block
decimal = [1-9][0-9]*;
number = ([1-9][0-9]* | [0])? "." [0-9]+ | [1-9][0-9]* | [0];
id = [a-zA-Z_][a-zA-Z_\-0-9]{0,63};
*/

int lex_next(struct lex *const x) {
	const char *s0, *s1;
	const size_t prev_line = scan.line;
	/*!stags:re2c format = 'const char *@@;\n'; */
	assert(x);
	if(!scan.buffer) return 0;
	x->s0 = x->s1 = 0;
	x->line = prev_line;
	x->ws_before = 0;
	x->new_paragraph = 0;
scan:
	/*!re2c
	<text> end { printf("end\n");return x->symbol = END, 0; }
	<image, command, command_args> end { return x->symbol = ERROR, 0; }

	<text> newline
		{ x->line = ++scan.line;
		x->new_paragraph = prev_line + 2 >= scan.line;
		x->ws_before = 1;
		goto scan; }
	<text> ws+ { x->ws_before = 1; goto scan; }
	<text> @s0 glyph+ @s1 { x->s0 = s0, x->s1 = s1;
		return x->symbol = TEXT, 1; }
	<text> @s0 "!" @s1 { x->s0 = s0, x->s1 = s1;
		return x->symbol = BANG, 1; }
	 <text> "\\" @s0 "[" @s1 { x->s0 = s0, x->s1 = s1;
		return x->symbol = BRACKET, 1; }
	<text> "![" :=> image
	<text> "[" :=> command

	<image> * { printf("image(broken)\n");return 0; }
	<image> ws* "osm" ws* "](geo:" @s0 number "," @s1 number ")" {
		scan.condition = yyctext;
		x->s0 = s0, x->s1 = s1;
		printf("Got a map.\n");
		return 1;
	}

	<command> ws+ { scan.is_ws_expected = 0; goto scan; }
	<command> "source"
		{ if(scan.is_ws_expected) return x->symbol = ERROR, 0;
		return scan.is_ws_expected = 1, x->symbol = SOURCE, 1; }
	<command> "default"
		{ if(scan.is_ws_expected) return x->symbol = ERROR, 0;
		return scan.is_ws_expected = 1, x->symbol = DEFAULT, 1; }
	<command> ":" => command_args { scan.is_ws_expected = 0; goto scan; }
	<command_args> [^\n\r\]\x00]+ { printf("Command args fixme\n"); goto scan; }
	<command_args> "]" :=> text
	<command, command_args> * { return x->symbol = ERROR, 0; }
	 */
}