This commit is contained in:
Neil 2022-02-15 19:48:50 -08:00
parent 2caa6f458b
commit 7c2c4785e4
3 changed files with 58 additions and 41 deletions

View File

@ -146,8 +146,8 @@ int main(int argc, char **argv) {
}
printf("Lexing finished: %s on %lu.\n",
lex_symbols[lex.symbol], lex.line);
if(lex.symbol != END) { errno = EILSEQ; goto catch; }
char_array_clear(&entry);
break; /* fixme */
}
int_array_clear(&days);

View File

@ -5,20 +5,29 @@ int lex_looks_like_month(const char *);
int lex_looks_like_day(const char *);
#define LEX_SYMBOL \
/* Results. */ X(END), X(SYNTAX), X(ILLEGAL), X(NOT_FOUND), \
/* Text. */ X(PARAGRAPH), X(TEXT), \
/* Edicts. */ X(SOURCE), X(DEFAULT), X(SIGNIFICANT), X(SCORE), X(MAP), \
/* Arguments. */ X(ARG_KEYWORD), X(ARG_DATE), X(ARG_NATURAL), \
X(ARG_FREEFORM)
\
/* Results. */ \
X(END), X(SYNTAX), X(ILLEGAL), X(NOT_FOUND), \
\
/* Text. */ \
X(PARAGRAPH), X(TEXT), X(CAPTION), \
\
/* Edicts. */ \
X(SOURCE), X(DEFAULT), X(SOURCE_RECALL), \
X(LOCATION), X(LOCATION_RECALL), \
X(SIGNIFICANT), X(SIGNIFICANT_RECALL), \
\
/* Arguments. */ \
X(ARG_KEYWORD), X(ARG_DATE), X(ARG_NATURAL), X(ARG_FREEFORM)
#define X(n) n
struct lex {
#define X(n) n
enum lex_symbol { LEX_SYMBOL } symbol;
#undef X
int ws_before, new_paragraph;
const char *s0, *s1;
size_t line;
};
#undef X
#define X(n) #n
static const char *const lex_symbols[] = { LEX_SYMBOL };
#undef X

View File

@ -72,7 +72,7 @@ int lex_looks_like_day(const char *const a) {
/* "[edict: expect; there; to; be; args]", in this case, expect would be a
stack of `size = 5` `EXPECT_KEYWORD`. This mirrors arguments in `LEX_SYMBOL`
and should also be an `edict_*` in <fn:lex_next> and <fn:expect_pop>. */
#define EXPECT_HEAD X(keyword, KEYWORD) X(date, DATE)
#define EXPECT_HEAD X(keyword, KEYWORD) X(date, DATE) X(natural, NATURAL)
#define EXPECT_CONS Y(freeform, FREEFORM)
#define EXPECT EXPECT_HEAD EXPECT_CONS
@ -125,6 +125,15 @@ int lex_next(struct lex *const x) {
re2c:define:YYGETCONDITION:naked = 1;
re2c:define:YYSETCONDITION = 'scan.condition = @@;';
re2c:define:YYSETCONDITION:naked = 1;
sentinel = "\x00";
illegal = [\x01-\x08\x0a-\x1f\x7f]; // unix-style control characters
newline = "\n";
ws = [ \t];
glyph = [^\x00-\x1f \x7f];
keyword = [a-zA-Z_][a-zA-Z0-9_\-]{0,63};
decimal = "-"? ([1-9][0-9]* | [0])? "." [0-9]+ | [1-9][0-9]* | [0];
natural = [1-9][0-9]*;
date = "-"? natural "-" [0-1][0-9] "-" [0-1][0-9];
*/
const char *s0, *s1;
/*!stags:re2c format = 'const char *@@;\n'; */
@ -133,39 +142,36 @@ int lex_next(struct lex *const x) {
x->s0 = x->s1 = 0;
scan:
/*!re2c
sentinel = "\x00";
illegal = [\x01-\x08\x0a-\x1f\x7f]; // unix-style control characters
newline = "\n";
ws = [ \t];
glyph = [^\x00-\x1f \x7f];
<*> illegal { return x->symbol = ILLEGAL, 0; }
<*> * { return x->symbol = SYNTAX, 0; }
<line> sentinel { return x->symbol = END, 0; }
<text, text, image, edict, edict_keyword, edict_date, edict_freeform, edict_end>
<text, expect_line, expect_caption, text, image, edict, edict_keyword, edict_date, edict_freeform, edict_end>
sentinel { return x->symbol = ILLEGAL, 0; }
<expect_line> newline => line { x->line = ++scan.line; goto scan; }
<line> newline { x->line = ++scan.line; return x->symbol = PARAGRAPH, 1; }
<line> "![" :=> image
<line> "--" :=> source
<line> "->" :=> location
<line> "[" :=> edict
<line> "" / glyph :=> text
<line> * { return x->symbol = SYNTAX, 1; }
<expect_caption> ws* @s0 ([^] \ (sentinel | illegal | newline | ws))
([^] \ (sentinel | illegal | newline))* @s1 ws* / newline => expect_line
{ x->s0 = s0, x->s1 = s1; return x->symbol = CAPTION, 1; }
<text> newline => line { x->line = ++scan.line; goto scan; }
<text> ws+ { goto scan; }
<text> @s0 glyph+ @s1 { x->s0 = s0, x->s1 = s1;
return x->symbol = TEXT, 1; }
<text> @s0 glyph+ @s1
{ x->s0 = s0, x->s1 = s1; return x->symbol = TEXT, 1; }
decimal = "-"? ([1-9][0-9]* | [0])? "." [0-9]+ | [1-9][0-9]* | [0];
<source> @s0 keyword @s1 => expect_line
{ x->s0 = s0, x->s1 = s1; return x->symbol = SOURCE_RECALL, 1; }
<image> ws* "osm" ws* "](geo:" @s0 decimal "," @s1 decimal ")" => text {
x->symbol = MAP, x->s0 = s0, x->s1 = s1;
printf("Got a map.\n");
return 1;
}
<image> * { printf("image(broken)\n");return 0; }
<location> "" / "![" :=> map
<location> @s0 keyword @s1 => expect_line
{ x->s0 = s0, x->s1 = s1; return x->symbol = LOCATION_RECALL, 1; }
//
natural = [1-9][0-9]*;
id = [a-zA-Z_][a-zA-Z_\-0-9]{0,63};
date = "-"? natural "-" [0-1][0-9] "-" [0-1][0-9];
<map> "![" ws* "osm" ws* "](geo:" @s0 decimal "," @s1 decimal ")" ws*
=> expect_caption
{ x->s0 = s0, x->s1 = s1; return x->symbol = LOCATION, 1; }
<edict> "source"
{ if(scan.is_ws_expected || scan.edict.size)
@ -181,12 +187,12 @@ scan:
scan.is_ws_expected = 1, scan.is_source = 0;
return x->symbol = DEFAULT, 1; }
// score
<edict> "significant"
{ if(scan.is_ws_expected || scan.edict.size)
return x->symbol = SYNTAX, 0;
scan.is_ws_expected = 1;
scan.edict.size = 2;
scan.edict.size = 3;
scan.edict.expect[2] = EXPECT_NATURAL;
scan.edict.expect[1] = EXPECT_FREEFORM;
scan.edict.expect[0] = EXPECT_DATE;
return x->symbol = SIGNIFICANT, 1; }
@ -195,28 +201,30 @@ scan:
return x->symbol = SYNTAX, 0;
scan.is_ws_expected = 1;
x->s0 = s0, x->s1 = s1;
return x->symbol = SCORE, 1; }
return x->symbol = SIGNIFICANT_RECALL, 1; }
<edict> ws+ { scan.is_ws_expected = 0; goto scan; }
<edict> ":"
{ if(!scan.edict.size) return x->symbol = SYNTAX, 0;
scan.is_ws_expected = 0, scan.is_source = 0;
expect_pop(); goto scan; }
<! edict_keyword, edict_date, edict_freeform> { expect_pop(); }
<edict_keyword> ws* @s0 id @s1 ws* ";"? / "]"?
{ x->s0 = s0, x->s1 = s1;
<edict_keyword> ws* @s0 keyword @s1 ws* ";"?
{ x->s0 = s0, x->s1 = s1; expect_pop();
return x->symbol = ARG_KEYWORD, 1; }
<edict_date> ws* @s0 date @s1 ws* ";"? / "]"?
{ x->s0 = s0, x->s1 = s1;
<edict_date> ws* @s0 date @s1 ws* ";"?
{ x->s0 = s0, x->s1 = s1; expect_pop();
return x->symbol = ARG_DATE, 1; }
<edict_natural> ws* @s0 natural @s1 ws* ";"?
{ x->s0 = s0, x->s1 = s1; expect_pop();
return x->symbol = ARG_NATURAL, 1; }
// fixme!
<edict_freeform> ws* @s0
[^ \t\n\r\v\f;[\]\x00][^\t\n\r\v\f;[\]\x00]*[^ \t\n\r\v\f;[\]\x00]*
@s1 ws* ";"? / "]"?
{ x->s0 = s0, x->s1 = s1;
{ x->s0 = s0, x->s1 = s1; expect_pop();
return x->symbol = ARG_FREEFORM, 1; }
<edict, edict_end> "]" => text
{ if(scan.edict.size) return 0;
goto scan; }
{ if(scan.edict.size) return 0; goto scan; }
<edict, edict_keyword, edict_date, edict_freeform, edict_end> *
{ return x->symbol = SYNTAX, 0; }
*/