static void putstr(int mode, STR tp)
{
if (mode == ESCOFF) {
fprintf(cur_fpo, "%s", tp);
} else if (*tp == '\0') {
fprintf(cur_fpo, "||"); // ヌル文字列を表す
} else {
if (num(tp)) { // read.h
fputc('\\', cur_fpo); // 数値ではなく数字の場合は見分けるために'\\'を接頭辞としてつける
}
do {
if (iskanji(*tp) && iskanji2(*(tp+1)))
{
fputc(*tp++, cur_fpo);
fputc(*tp++, cur_fpo);
}
else if (!isprkana(*tp))
{
fprintf(cur_fpo, "#\\%03d", *tp++);
}
else
{
if (isesc(*tp))
{
fputc('\\', cur_fpo);
}
fputc(*tp++, cur_fpo);
}
} while (*tp != '\0');
}
}
static void putstr(int mode, STR tp) {//N//
if(mode == ESCOFF) {
fprintf(cur_fpo, "%s", tp);
} else if(*tp == '\0') {
fprintf(cur_fpo, "||");
} else {
if(num(tp)) {
fputc('\\', cur_fpo);
}
do {
if(iskanji(*tp) && iskanji2(*(tp+1))) {
fputc(*tp++, cur_fpo);
fputc(*tp++, cur_fpo);
} else if(!isprkana(*tp)) {
fprintf(cur_fpo, "#\\%03d", *tp++);
} else {
if(isesc(*tp)) {
fputc('\\', cur_fpo);
}
fputc(*tp++, cur_fpo);
}
} while(*tp != '\0');
}
}
oo_token *lex(FILE *fp, size_t sz) {
size_t l;
char *source = malloc(sz);
oo_token *head = NULL, *curr = NULL;
if(!source) goto err1;
if(fseek(fp, 0L, SEEK_SET) != 0) goto err0;
l = fread(source, sizeof(char), sz, fp);
source[++l] = 0;
char *cp = source;
oo_floc empty = { .line = 0, .column = 0, .offset = 0};
curr = head = alloc_token(oot_START, empty, 0, NULL);
oo_floc loc = { .line = 1, .column = 1, .offset = -1 };
oo_states state = oos_start, prev = state;
uint32_t end_offset = 0;
oo_floc block;
do {
char a = *cp;
oo_tokens token = oot_UNKNOWN;
step(&state, &cp, &loc);
if(state == oos_err) goto syn;
if (state == oos_str || state == oos_ident || state == oos_num) {
block.line = loc.line;
block.column = loc.column;
block.offset = loc.offset;
}
int end_of_str = (state == oos_eo_str || state == oos_eo_id || state == oos_eo_num);
if (end_of_str) {
/* id, num & str */
end_offset = loc.offset;
int t = -1;
if (state == oos_eo_id) t = oot_IDENT;
if (state == oos_eo_str) t = oot_STR;
if (state == oos_eo_num) t = oot_NUM;
if (t == oot_IDENT || t == oot_STR || t == oot_NUM) {
curr = alloc_token(t, block, end_offset - block.offset, curr);
}
if(state == oos_eo_id || state == oos_eo_num) {
cp--;
loc.offset--;
}
end_offset = 0;
} else {
/* sym */
if (state == oos_sym) {
char c = *(source + (loc.offset));
token = oo_tokens_from_char(c);
}
if(token != oot_UNKNOWN) {
curr = alloc_token(token, loc, 1, curr);
}
}
syn:
printf("o: %d, s -> %d, s_s = '%s' (%d)", (int)loc.offset, (int)state, oo_state_to_string(state), (int)a);
if(isalpha(a) || isdigit(a) || issym(a)) {
printf(", '%c'", a);
}
printf("\n");
if(state == oos_err) {
if(prev == oos_in_str) {
printf("Syntax error, unterminated string constant, line %d column %d\n", block.line, block.column - 1);
}
/* TODO: Add oot_ERR token */
curr = alloc_token(oot_EOF, loc, 1, curr);
return head;
}
prev = state;
} while(state != oos_eof);
if(state == oos_eof) {
curr = alloc_token(oot_EOF, loc, 0, curr);
}
err0:
if(source) free(source), source = NULL;
err1:
return head;
}
static void step(oo_states *state, char **cp, oo_floc *loc) {
int c = **cp, k = 0;
int i = -1; /* unknown */
if(c != 0 && *(*cp+1) != 0) k = *(*cp +1);
i = isspace(c) ? 0 : i; /* ws */
i = c == '\n' ? 1 : i; /* new line */
i = issym(c) ? 2 : i; /* sym */
i = isdigit(c) ? 3 : i; /* numeric */
i = isid(c) ? 4 : i; /* id */
i = c == '"' ? 5 : i; /* str */
i = isesc(c, k) ? 6 : i; /* escape */
i = c == 0 ? 7 : i; /* eof */
if(i == -1) printf("---> '%c'\n", c);
oo_branch *b = &table[*state][i];
*state = b->state;
if(b->advance) {
if(b->state == oos_nl) {
loc->line++;
loc->column = 0;
}
loc->column++;
loc->offset++;
(*cp)++;
}
}
static oo_tokens oo_tokens_from_char(int t) {
if (t == '~') return oot_TILDE;
if (t == '`') return oot_GACCENT;
if (t == '!') return oot_BANG;
if (t == '@') return oot_AT;
if (t == '#') return oot_POUND;
if (t == '$') return oot_DOLLAR;
if (t == '%') return oot_PERCENT;
if (t == '^') return oot_EXP;
if (t == '&') return oot_AND;
if (t == '*') return oot_STAR;
if (t == '(') return oot_LPAREN;
if (t == ')') return oot_RPAREN;
if (t == '-') return oot_MINUS;
if (t == '=') return oot_EQUAL;
if (t == '_') return oot_UNDER;
if (t == '+') return oot_PLUS;
if (t == '[') return oot_LBRACKET;
if (t == ']') return oot_RBRACKET;
if (t == '\\') return oot_BSLASH;
if (t == '{') return oot_LBRACE;
if (t == '}') return oot_RBRACE;
if (t == '|') return oot_BAR;
if (t == ';') return oot_SEMI;
if (t == '\'') return oot_SQUOTE;
if (t == ':') return oot_COLON;
if (t == '"') return oot_DQUOTE;
if (t == ',') return oot_COMMA;
if (t == '.') return oot_DOT;
if (t == '/') return oot_FSLASH;
if (t == '<') return oot_LT;
if (t == '>') return oot_GT;
if (t == '?') return oot_WHAT;
return oot_UNKNOWN;
}
char *string_from_oo_tokens(oo_tokens t) {
if (t == oot_START) return "oot_START";
if (t == '~') return "oot_TILDE";
if (t == '`') return "oot_GACCENT";
if (t == '!') return "oot_BANG";
if (t == '@') return "oot_AT";
if (t == '#') return "oot_POUND";
if (t == '$') return "oot_DOLLAR";
if (t == '%') return "oot_PERCENT";
if (t == '^') return "oot_EXP";
if (t == '&') return "oot_AND";
if (t == '*') return "oot_STAR";
if (t == '(') return "oot_LPAREN";
if (t == ')') return "oot_RPAREN";
if (t == '-') return "oot_MINUS";
if (t == '=') return "oot_EQUAL";
if (t == '_') return "oot_UNDER";
if (t == '+') return "oot_PLUS";
if (t == '[') return "oot_LBRACKET";
if (t == ']') return "oot_RBRACKET";
if (t == '\\') return "oot_BSLASH";
if (t == '{') return "oot_LBRACE";
if (t == '}') return "oot_RBRACE";
if (t == '|') return "oot_BAR";
if (t == ';') return "oot_SEMI";
if (t == '\'') return "oot_SQUOTE";
if (t == ':') return "oot_COLON";
if (t == '"') return "oot_DQUOTE";
if (t == ',') return "oot_COMMA";
if (t == '.') return "oot_DOT";
if (t == '/') return "oot_FSLASH";
if (t == '<') return "oot_LT";
if (t == '>') return "oot_GT";
if (t == '?') return "oot_WHAT";
if (t == oot_IDENT) return "oot_IDENT";
if (t == oot_NUM) return "oot_NUM";
if (t == oot_STR) return "oot_STR";
if (t == oot_EOF) return "oot_EOF";
return "oot_UNKNOWN";
}
static char *oo_state_to_string(oo_states state) {
if (state == oos_start) return "oos_start";
if (state == oos_ws) return "oos_ws";
if (state == oos_nl) return "oos_nl";
if (state == oos_sym) return "oos_sym";
if (state == oos_num) return "oos_num";
if (state == oos_ident) return "oos_ident";
if (state == oos_str) return "oos_str";
if (state == oos_esc) return "oos_esc";
if (state == oos_in_str) return "oos_in_str";
if (state == oos_in_id) return "oos_in_id";
if (state == oos_in_num) return "oos_in_num";
if (state == oos_eo_str) return "oos_eo_str";
if (state == oos_eo_id) return "oos_eo_id";
if (state == oos_eo_num) return "oos_eo_num";
if (state == oos_err) return "oos_err";
if (state == oos_eof) return "oos_eof";
return "undefined";
}
static int issym(int c) {
return
c == '[' || c == ']' || c == '\\'
|| c == '{' || c == '}' || c == '|'
|| c == ';' || c == '\''|| c == ':' || c == '"'
|| c == ',' || c == '.' || c == '/'
|| c == '<' || c == '>' || c == '?'
|| c == '~' || c == '`' || c == '!' || c == '@' || c == '#' || c == '$'
|| c == '%' || c == '^' || c == '&' || c == '*' || c == '(' || c == ')'
|| c == '-' || c == '_' || c == '=' || c == '+'
;
}
static int isid(int c) {
return isalpha(c) || c == '_';
}
static int isesc(int c, int k) {
if(c == 0) return 0;
if(k == 0) return 0;
if(c == '\\') {
if (k == '"') return 1;
if (k == '\\') return 1;
if (k == 'r') return 1;
if (k == 'n') return 1;
}
return 0;
}