static int _compilecode(const char **re_loc, ByteProg *prog, int sizecode)
{
const char *re = *re_loc;
char *code = sizecode ? NULL : prog->insts;
int start = PC;
int term = PC;
int alt_label = 0;
for (; *re && *re != ')'; re++) {
switch (*re) {
case '\\': {
re++;
if (!*re) goto syntax_error; // Trailing backslash
char c = *re | 0x20;
if (c == 'd' || c == 's' || c == 'w') {
term = PC;
EMIT(PC++, NamedClass);
EMIT(PC++, *re);
prog->len++;
break;
}
if ((c >= '0' && c <= '9') || (c >= 'a' && c <= 'z')) {
goto unsupported_escape;
}
}
default:
term = PC;
EMIT(PC++, Char);
EMIT(PC++, *re);
prog->len++;
break;
case '.':
term = PC;
EMIT(PC++, Any);
prog->len++;
break;
case '[': {
int cnt;
term = PC;
re++;
if (*re == '^') {
EMIT(PC++, ClassNot);
re++;
} else {
EMIT(PC++, Class);
}
PC++; // Skip "# of pairs" byte
prog->len++;
for (cnt = 0; *re != ']'; re++, cnt++) {
if (!*re) goto syntax_error;
if (*re == '\\') {
re++;
if (!*re) goto syntax_error;
if (*re != '\\' && *re != ']') goto unsupported_escape;
}
EMIT(PC++, *re);
if (re[1] == '-' && re[2] != ']') {
re += 2;
}
EMIT(PC++, *re);
}
EMIT(term + 1, cnt);
break;
}
case '(': {
term = PC;
int sub;
int capture = 1;
re++;
if (*re == '?') {
re++;
if (*re == ':') {
capture = 0;
re++;
} else {
*re_loc = re;
return RE1_5_UNSUPPORTED_SYNTAX;
}
}
if (capture) {
sub = ++prog->sub;
EMIT(PC++, Save);
EMIT(PC++, 2 * sub);
prog->len++;
}
int res = _compilecode(&re, prog, sizecode);
*re_loc = re;
if (res < 0) return res;
if (*re != ')') return RE1_5_SYNTAX_ERROR;
if (capture) {
EMIT(PC++, Save);
EMIT(PC++, 2 * sub + 1);
prog->len++;
}
break;
}
case '{':
*re_loc = re;
return RE1_5_UNSUPPORTED_SYNTAX;
case '?':
if (PC == term) goto syntax_error; // nothing to repeat
INSERT_CODE(term, 2, PC);
if (re[1] == '?') {
EMIT(term, RSplit);
re++;
} else {
EMIT(term, Split);
}
EMIT(term + 1, REL(term, PC));
prog->len++;
term = PC;
break;
case '*':
if (PC == term) goto syntax_error; // nothing to repeat
INSERT_CODE(term, 2, PC);
EMIT(PC, Jmp);
EMIT(PC + 1, REL(PC, term));
PC += 2;
if (re[1] == '?') {
EMIT(term, RSplit);
re++;
} else {
EMIT(term, Split);
}
EMIT(term + 1, REL(term, PC));
prog->len += 2;
term = PC;
break;
case '+':
if (PC == term) goto syntax_error; // nothing to repeat
if (re[1] == '?') {
EMIT(PC, Split);
re++;
} else {
EMIT(PC, RSplit);
}
EMIT(PC + 1, REL(PC, term));
PC += 2;
prog->len++;
term = PC;
break;
case '|':
if (alt_label) {
EMIT(alt_label, REL(alt_label, PC) + 1);
}
INSERT_CODE(start, 2, PC);
EMIT(PC++, Jmp);
alt_label = PC++;
EMIT(start, Split);
EMIT(start + 1, REL(start, PC));
prog->len += 2;
term = PC;
break;
case '^':
EMIT(PC++, Bol);
prog->len++;
term = PC;
break;
case '$':
EMIT(PC++, Eol);
prog->len++;
term = PC;
break;
}
}
if (alt_label) {
EMIT(alt_label, REL(alt_label, PC) + 1);
}
*re_loc = re;
return RE1_5_SUCCESS;
syntax_error:
*re_loc = re;
return RE1_5_SYNTAX_ERROR;
unsupported_escape:
*re_loc = re;
return RE1_5_UNSUPPORTED_ESCAPE;
}
static const char *_compilecode(const char *re, ByteProg *prog, int sizecode)
{
char *code = sizecode ? NULL : prog->insts;
int start = PC;
int term = PC;
int alt_label = 0;
for (; *re && *re != ')'; re++) {
switch (*re) {
case '\\':
re++;
if (!*re) return NULL; // Trailing backslash
if ((*re | 0x20) == 'd' || (*re | 0x20) == 's' || (*re | 0x20) == 'w') {
term = PC;
EMIT(PC++, NamedClass);
EMIT(PC++, *re);
prog->len++;
break;
}
default:
term = PC;
EMIT(PC++, Char);
EMIT(PC++, *re);
prog->len++;
break;
case '.':
term = PC;
EMIT(PC++, Any);
prog->len++;
break;
case '[': {
int cnt;
term = PC;
re++;
if (*re == '^') {
EMIT(PC++, ClassNot);
re++;
} else {
EMIT(PC++, Class);
}
PC++; // Skip # of pair byte
prog->len++;
for (cnt = 0; *re != ']'; re++, cnt++) {
if (!*re) return NULL;
EMIT(PC++, *re);
if (re[1] == '-') {
re += 2;
}
EMIT(PC++, *re);
}
EMIT(term + 1, cnt);
break;
}
case '(': {
term = PC;
int sub = 0;
int capture = re[1] != '?' || re[2] != ':';
if (capture) {
sub = ++prog->sub;
EMIT(PC++, Save);
EMIT(PC++, 2 * sub);
prog->len++;
} else {
re += 2;
}
re = _compilecode(re + 1, prog, sizecode);
if (re == NULL || *re != ')') return NULL; // error, or no matching paren
if (capture) {
EMIT(PC++, Save);
EMIT(PC++, 2 * sub + 1);
prog->len++;
}
break;
}
case '?':
if (PC == term) return NULL; // nothing to repeat
INSERT_CODE(term, 2, PC);
if (re[1] == '?') {
EMIT(term, RSplit);
re++;
} else {
EMIT(term, Split);
}
EMIT(term + 1, REL(term, PC));
prog->len++;
term = PC;
break;
case '*':
if (PC == term) return NULL; // nothing to repeat
INSERT_CODE(term, 2, PC);
EMIT(PC, Jmp);
EMIT(PC + 1, REL(PC, term));
PC += 2;
if (re[1] == '?') {
EMIT(term, RSplit);
re++;
} else {
EMIT(term, Split);
}
EMIT(term + 1, REL(term, PC));
prog->len += 2;
term = PC;
break;
case '+':
if (PC == term) return NULL; // nothing to repeat
if (re[1] == '?') {
EMIT(PC, Split);
re++;
} else {
EMIT(PC, RSplit);
}
EMIT(PC + 1, REL(PC, term));
PC += 2;
prog->len++;
term = PC;
break;
case '|':
if (alt_label) {
EMIT(alt_label, REL(alt_label, PC) + 1);
}
INSERT_CODE(start, 2, PC);
EMIT(PC++, Jmp);
alt_label = PC++;
EMIT(start, Split);
EMIT(start + 1, REL(start, PC));
prog->len += 2;
term = PC;
break;
case '^':
EMIT(PC++, Bol);
prog->len++;
term = PC;
break;
case '$':
EMIT(PC++, Eol);
prog->len++;
term = PC;
break;
}
}
if (alt_label) {
EMIT(alt_label, REL(alt_label, PC) + 1);
}
return re;
}
