public Xow_search_tkn[] Scan(byte[] src) { this.src = src; this.src_len = src.length; tkns.Clear(); pos = 0; txt_bgn = -1; while (pos < src_len) { byte cur_b = src[pos]; Object cur_obj = trie.Match_bgn_w_byte(cur_b, src, pos, src_len); if (cur_obj == null) { // text character if (txt_bgn == -1) txt_bgn = pos; // 1st character not set; set it ++pos; } else { // AND, OR, (, ), -, \s, " int pos_end = trie.Match_pos(); byte cur_tid = ((Byte_obj_val) cur_obj).Val(); if (Cur_join_is_word(cur_tid, pos_end)) continue; // ignore words containing "and", "or"; EX: "random"; "for" if (txt_bgn != -1) { // pending word; create Tkns_add_word(Xow_search_tkn.Tid_word, txt_bgn, pos); txt_bgn = -1; } switch (cur_tid) { case Xow_search_tkn.Tid_space: // discard spaces pos = Bry_finder.Find_fwd_while(src, pos, src_len, Byte_ascii.Space); break; case Xow_search_tkn.Tid_quote: // find end quote and add as word int quote_bgn = pos + 1; int quote_end = Bry_finder.Find_fwd(src, Byte_ascii.Quote, quote_bgn, src_len); if (quote_end == Bry_.NotFound) throw Err_.new_fmt_("could not find end quote: {0}", String_.new_u8(src)); Tkns_add_word(Xow_search_tkn.Tid_word_quoted, quote_bgn, quote_end); pos = quote_end + 1; // +1 to place after quote break; case Xow_search_tkn.Tid_not: Tkns_add_word(Xow_search_tkn.Tid_not, pos, pos_end); pos = pos_end; break; case Xow_search_tkn.Tid_paren_bgn: case Xow_search_tkn.Tid_paren_end: case Xow_search_tkn.Tid_and: case Xow_search_tkn.Tid_or: tkns.Add(new_tkn(cur_tid, pos, pos_end)); pos = pos_end; break; default: throw Err_.unhandled(cur_tid); } } } if (txt_bgn != -1) { // pending word; create Tkns_add_word(Xow_search_tkn.Tid_word, txt_bgn, pos); txt_bgn = -1; } return (Xow_search_tkn[]) tkns.To_ary_and_clear(Xow_search_tkn.class); }
int[] Xto_int_ary(Xoctg_data_ctg ctg) { List_adp list = List_adp_.new_(); byte tid_max = Xoa_ctg_mgr.Tid__max; for (byte tid = 0; tid < tid_max; tid++) { Xoctg_idx_mgr grp = ctg.Grp_by_tid(tid); if (grp == null) continue; int len = grp.Itms_len(); for (int i = 0; i < len; i++) { Xoctg_idx_itm itm = grp.Itms_get_at(i); list.Add(itm.Id()); } } return (int[]) list.To_ary_and_clear(int.class); }