Function AR_SPLIT

src/arithmetic/string_funcs/string_funcs.c:665–723

returns a list of strings resulting from the splitting of the original string around matches of the given delimiter

// returns a list of strings resulting from the splitting of the original string around matches of the given delimiter
SIValue AR_SPLIT(SIValue *argv, int argc, void *private_data) {
	if(SIValue_IsNull(argv[0]) || SIValue_IsNull(argv[1])) {
		return SI_NullVal();
	}

	char       *str           = argv[0].stringval;
	const char *delimiter     = argv[1].stringval;
	size_t      str_len       = strlen(str);
	size_t      delimiter_len = strlen(delimiter);
	SIValue     tokens        = SIArray_New(1);

	if(delimiter_len == 0) {
		if(strlen(str) == 0) {
			SIArray_Append(&tokens, SI_ConstStringVal(""));
		} else {
			utf8proc_int32_t c;
			utf8proc_uint8_t token[5];
			const utf8proc_uint8_t *str_i = (const utf8proc_uint8_t *)str;
			while(str_i[0] != 0) {
				str_i += utf8proc_iterate(str_i, -1, &c);
				int i  = utf8proc_encode_char(utf8proc_tolower(c), token);
				token[i] = '\0';
				SIArray_Append(&tokens, SI_ConstStringVal((const char *)token));
			}
		}
	} else if(str_len == 0) {
		SIArray_Append(&tokens, SI_ConstStringVal(""));
	} else {
		size_t rest_len   = str_len;
		const char *start = str;
		bool delimiter_found = false;
		while(rest_len >= delimiter_len) {
			// find bytes length from start to delimiter
			int len = 0;
			delimiter_found = false;
			while(len <= rest_len - delimiter_len) {
				if(strncmp(start + len, delimiter, delimiter_len) == 0) {
					delimiter_found = true;
					break;
				}
				len++;
			}
			if(!delimiter_found) {
				break;
			}
			SIValue si_token = SI_TransferStringVal(rm_strndup(start, len));
			SIArray_Append(&tokens, si_token);
			SIValue_Free(si_token);
			start += len + delimiter_len;
			rest_len -= len + delimiter_len;
		}
		if(rest_len > 0 || delimiter_found) {
			SIValue si_token = SI_ConstStringVal(start);
			SIArray_Append(&tokens, si_token);
		}
	}

	return tokens;