More complex; read args in commands.

This commit is contained in:
Neil 2022-02-14 22:02:32 -08:00
parent 794dc2206f
commit a75385f71a
3 changed files with 79 additions and 19 deletions

View File

@ -135,10 +135,12 @@ int main(int argc, char **argv) {
while(lex_next(&lex)) { while(lex_next(&lex)) {
printf("%lu: %s", printf("%lu: %s",
(unsigned long)lex.line, lex_symbols[lex.symbol]); (unsigned long)lex.line, lex_symbols[lex.symbol]);
if(lex.symbol == TEXT) { if(lex.symbol == TEXT || lex.symbol == ARG_KEYWORD
|| lex.symbol == ARG_DATE
|| lex.symbol == ARG_FREEFORM) {
if(lex.s0 + INT_MAX < lex.s1) if(lex.s0 + INT_MAX < lex.s1)
{ errno = EILSEQ; goto catch; } { errno = EILSEQ; goto catch; }
printf(" %.*s", (int)(lex.s1 - lex.s0), lex.s0); printf(" <<%.*s>>", (int)(lex.s1 - lex.s0), lex.s0);
} }
printf(".\n"); printf(".\n");
} }

View File

@ -6,7 +6,8 @@ int lex_looks_like_day(const char *);
#define LEX_SYMBOL X(END), X(ERROR), X(TEXT), X(BANG), X(BRACKET), X(WHITE), \ #define LEX_SYMBOL X(END), X(ERROR), X(TEXT), X(BANG), X(BRACKET), X(WHITE), \
X(MAP), \ X(MAP), \
/* Commands */ X(SIGNIFICANT), X(DEFAULT), X(SOURCE) /* Commands */ X(SOURCE), X(DEFAULT), X(SIGNIFICANT), X(SCORE), \
/* Arguments */ X(ARG_KEYWORD), X(ARG_DATE), X(ARG_NATURAL), X(ARG_FREEFORM)
#define X(n) n #define X(n) n
struct lex { struct lex {

View File

@ -66,6 +66,8 @@ int lex_looks_like_day(const char *const a) {
/* This defines `enum condition`. */ /* This defines `enum condition`. */
/*!types:re2c*/ /*!types:re2c*/
#define EXPECT X(KEYWORD), X(DATE), X(FREEFORM)
/** scanner reads a file and extracts semantic information. Valid to access /** scanner reads a file and extracts semantic information. Valid to access
only while underlying pointers do not change. */ only while underlying pointers do not change. */
static struct scan { static struct scan {
@ -75,7 +77,12 @@ static struct scan {
const char *label, *buffer; const char *label, *buffer;
enum condition condition; enum condition condition;
size_t line; size_t line;
int is_ws_expected; int is_ws_expected, is_source;
#define X(n) EXPECT_ ## n
/* "[something: expect; there; to; be; args]", in this case, `size = 5` and
expect would be 5 `EXPECT_KEYWORD`. */
struct { unsigned size; enum { EXPECT } expect[20]; } command;
#undef X
} scan; /* Terrible, gah. */ } scan; /* Terrible, gah. */
void lex_reset(const char *const buffer) { void lex_reset(const char *const buffer) {
@ -108,8 +115,19 @@ glyphs = glyph+;
natural = [1-9][0-9]*; natural = [1-9][0-9]*;
decimal = "-"? ([1-9][0-9]* | [0])? "." [0-9]+ | [1-9][0-9]* | [0]; decimal = "-"? ([1-9][0-9]* | [0])? "." [0-9]+ | [1-9][0-9]* | [0];
id = [a-zA-Z_][a-zA-Z_\-0-9]{0,63}; id = [a-zA-Z_][a-zA-Z_\-0-9]{0,63};
date = "-"? natural "-" [0-1][0-9] "-" [0-1][0-9];
*/ */
static void expect_pop(void) {
printf("<expect_pop>");
if(!scan.command.size) { printf("nostack\n"); scan.condition = yyccommand_end; return; }
switch(scan.command.expect[--scan.command.size]) {
case EXPECT_KEYWORD: printf("keyword\n");scan.condition = yyccommand_keyword; break;
case EXPECT_DATE: printf("date\n");scan.condition = yyccommand_date; break;
case EXPECT_FREEFORM: printf("freeform\n");scan.condition = yyccommand_freeform; break;
}
}
int lex_next(struct lex *const x) { int lex_next(struct lex *const x) {
const char *s0, *s1; const char *s0, *s1;
const size_t prev_line = scan.line; const size_t prev_line = scan.line;
@ -123,7 +141,8 @@ int lex_next(struct lex *const x) {
scan: scan:
/*!re2c /*!re2c
<text> end { printf("end\n");return x->symbol = END, 0; } <text> end { printf("end\n");return x->symbol = END, 0; }
<image, command, command_args> end { return x->symbol = ERROR, 0; } <image, command, command_keyword, command_date, command_end> end
{ return x->symbol = ERROR, 0; }
<text> newline <text> newline
{ x->line = ++scan.line; { x->line = ++scan.line;
@ -147,20 +166,58 @@ scan:
} }
<image> * { printf("image(broken)\n");return 0; } <image> * { printf("image(broken)\n");return 0; }
<command> ws+ { scan.is_ws_expected = 0; goto scan; } // source
<command> @s0 natural @s1
{ if(scan.is_ws_expected) return x->symbol = ERROR, 0;
return x->symbol = SIGNIFICANT, scan.is_ws_expected = 1,
x->s0 = s0, x->s1 = s1, 1; }
<command> "source" <command> "source"
{ if(scan.is_ws_expected) return x->symbol = ERROR, 0; { if(scan.is_ws_expected || scan.command.size)
return scan.is_ws_expected = 1, x->symbol = SOURCE, 1; } return x->symbol = ERROR, 0;
scan.is_ws_expected = 1, scan.is_source = 1;
scan.command.size = 2;
scan.command.expect[1] = EXPECT_KEYWORD;
scan.command.expect[0] = EXPECT_FREEFORM;
return x->symbol = SOURCE, 1; }
<command> "default" <command> "default"
{ if(scan.is_ws_expected) return x->symbol = ERROR, 0; { if(scan.is_ws_expected || !scan.is_source)
return scan.is_ws_expected = 1, x->symbol = DEFAULT, 1; } return x->symbol = ERROR, 0;
<command> ":" => command_args { scan.is_ws_expected = 0; goto scan; } scan.is_ws_expected = 1, scan.is_source = 0;
<command_args> [^\n\r\]\x00]+ { printf("Command args fixme\n"); goto scan; } return x->symbol = DEFAULT, 1; }
<command, command_args> "]" :=> text
<command, command_args> * { return x->symbol = ERROR, 0; } // score
*/ <command> "significant"
{ if(scan.is_ws_expected || scan.command.size)
return x->symbol = ERROR, 0;
scan.is_ws_expected = 1;
scan.command.size = 2;
scan.command.expect[1] = EXPECT_FREEFORM;
scan.command.expect[0] = EXPECT_DATE;
return x->symbol = SIGNIFICANT, 1; }
<command> @s0 natural @s1
{ if(scan.is_ws_expected || scan.command.size)
return x->symbol = ERROR, 0;
scan.is_ws_expected = 1;
x->s0 = s0, x->s1 = s1;
return x->symbol = SCORE, 1; }
// general command stuff
<command> ws+ { scan.is_ws_expected = 0; goto scan; }
<command> ":"
{ if(!scan.command.size) return x->symbol = ERROR, 0;
scan.is_ws_expected = 0, scan.is_source = 0;
expect_pop(); goto scan; }
<command_keyword> ws* @s0 id @s1 ws* ";"? / "]"?
{ x->s0 = s0, x->s1 = s1; expect_pop();
return x->symbol = ARG_KEYWORD, 1; }
<command_date> ws* @s0 date @s1 ws* ";"? / "]"?
{ x->s0 = s0, x->s1 = s1; expect_pop();
return x->symbol = ARG_DATE, 1; }
<command_freeform> ws* @s0 [^\t\n\r\v\f;[\]\x00]* @s1 ws* ";"? / "]"?
{ x->s0 = s0, x->s1 = s1; expect_pop();
return x->symbol = ARG_FREEFORM, 1; }
<command, command_end> "]" => text
{ printf("]\n"); if(scan.command.size) { printf("huh?\n"); return 0; }
goto scan;
}
<command, command_keyword, command_date, command_freeform, command_end> *
{ return x->symbol = ERROR, 0; }
*/
assert(0);
} }