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);
}
private boolean Cur_join_is_word(
byte cur_tid,
int
pos_end) { // extra logic to handle and / or occuring in unquoted strings; EX: "random";
// "for"
switch (cur_tid) {
default:
return false; // only look at AND, OR, -
case Xow_search_tkn.Tid_and:
case Xow_search_tkn.Tid_or:
case Xow_search_tkn.Tid_not:
break;
}
boolean join_is_word = true;
if (txt_bgn == -1) { // no pending word;
if (cur_tid == Xow_search_tkn.Tid_not)
return false; // NOT is only operator if no pending tkn; EX: -abc -> NOT abc; a-b -> a-b
byte nxt_b = pos_end < src_len ? src[pos_end] : Byte_ascii.Nil;
Object nxt_obj = trie.Match_bgn_w_byte(nxt_b, src, pos_end, src_len);
if (nxt_obj == null) // next tkn is text; join must be word
join_is_word = true;
else { // next tkn is tkn
byte nxt_tid = ((Byte_obj_val) nxt_obj).Val();
switch (nxt_tid) {
case Xow_search_tkn.Tid_space:
case Xow_search_tkn.Tid_quote:
case Xow_search_tkn.Tid_paren_bgn:
case Xow_search_tkn.Tid_paren_end:
join_is_word =
false; // next tkn is sym; and/or is not word; EX: a AND ; a AND"b"; a AND(b)
break;
case Xow_search_tkn.Tid_not:
case Xow_search_tkn.Tid_and:
case Xow_search_tkn.Tid_or:
join_is_word = true; // next tkn is and or not; and/or is word; EX: andor; oror; or-abc;
break;
default:
throw Err_.unhandled(cur_tid);
}
}
} else { // pending word; cur join must be word; EX: "grand": "and" invoked and "gr" pending
join_is_word = true;
}
if (join_is_word) {
if (txt_bgn == -1) txt_bgn = pos; // 1st character not set; set it
pos = pos_end;
return true;
}
if (txt_bgn != -1) {
Tkns_add_word(Xow_search_tkn.Tid_word, txt_bgn, pos); // create word
txt_bgn = -1;
}
return false;
}