private void Tkns_add_word(byte tid, int src_bgn, int src_end) {
if (tkns.Count() > 0) { // at least 1 tkn; check for "auto-and"
Xow_search_tkn last_tkn = (Xow_search_tkn) tkns.Get_at_last();
if (last_tkn.Tid()
== Xow_search_tkn.Tid_word) // previous tkn is word; auto "AND" words; EX: A B -> A AND B
tkns.Add(Xow_search_tkn.new_bry(Xow_search_tkn.Tid_and, Bry_and));
}
if (tid
== Xow_search_tkn
.Tid_word) { // if word has symbol, convert to quoted; EX: a-b should become "a-b";
// otherwise searcher would search for a single word a-b
byte[] cur_word = Bry_.Mid(src, src_bgn, src_end);
byte[][] words =
gplx.xowa.bldrs.cmds.texts.Xob_search_base.Split_ttl_into_words(
null, tmp_list, tmp_bfr, cur_word);
int words_len = words.length;
if (words_len == 1 // only one word
&& !Bry_.Eq(words[0], cur_word) // split word not same as raw
&& Bry_finder.Find_fwd(cur_word, Byte_ascii.Star) == -1 // no asterisk
) {
tkns.Add(Xow_search_tkn.new_bry(tid, words[0]));
return;
}
if (words.length > 1) // multiple words; add as quoted-term; EX: "a-b"
tid = Xow_search_tkn.Tid_word_quoted;
}
tkns.Add(new_tkn(tid, src_bgn, src_end));
}
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);
}