136 lines
3.4 KiB
C
136 lines
3.4 KiB
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 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 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 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*/
|
|
enum symbol { END, TEXT, BANG, BRACKET, WHITE, MAP };
|
|
|
|
/** scanner reads a file and extracts semantic information. Valid to access
|
|
only while underlying pointers do not change. */
|
|
struct scanner {
|
|
/* `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, *s0, *s1;
|
|
enum condition condition;
|
|
enum symbol symbol;
|
|
size_t line;
|
|
int ws_before;
|
|
};
|
|
|
|
/*!re2c
|
|
re2c:flags:tags = 1;
|
|
re2c:define:YYCURSOR = s->cursor;
|
|
re2c:define:YYMARKER = s->marker;
|
|
re2c:define:YYCTXMARKER = s->ctx_marker;
|
|
re2c:define:YYCONDTYPE = 'condition';
|
|
re2c:define:YYGETCONDITION = 's->condition';
|
|
re2c:define:YYGETCONDITION:naked = 1;
|
|
re2c:define:YYSETCONDITION = 's->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(struct scanner *const s) {
|
|
const char *s0, *s1;
|
|
/*!stags:re2c format = 'const char *@@;\n'; */
|
|
s->ws_before = 0;
|
|
scan:
|
|
/*!re2c
|
|
<text> end { return s->symbol = END, 1; }
|
|
// fixme: paragraphs.
|
|
<text> newline { s->line++; s->ws_before = 1; goto scan; }
|
|
<text> ws+ { s->ws_before = 1; goto scan; }
|
|
<text> @s0 glyph+ @s1 { s->s0 = s0, s->s1 = s1;
|
|
return s->symbol = TEXT, 1; }
|
|
<text> @s0 "!" @s1 { s->s0 = s0, s->s1 = s1;
|
|
return s->symbol = BANG, 1; }
|
|
<text> "\\" @s0 "[" @s1 { s->s0 = s0, s->s1 = s1;
|
|
return s->symbol = BRACKET, 1; }
|
|
<text> "![" :=> image
|
|
<text> "[" :=> command
|
|
<image> * { return 0; }
|
|
<image> ws* "osm" ws* "](geo:" @s0 number "," @s1 number ")" {
|
|
s->condition = yyctext;
|
|
s->s0 = s0, s->s1 = s1;
|
|
printf("Got a map.\n");
|
|
return 1;
|
|
}
|
|
<command> * { return 0; }
|
|
*/
|
|
}
|