Esempio n. 1
0
  public static void Find_pages(List_adp rv, Xowe_wiki wiki, Dpl_itm itm) {
    rv.Clear();
    List_adp includes = itm.Ctg_includes();
    if (includes == null) return;
    int includes_len = includes.Count();
    Ordered_hash old_regy = Ordered_hash_.New(),
        new_regy = Ordered_hash_.New(),
        cur_regy = Ordered_hash_.New();
    Xodb_load_mgr load_mgr = wiki.Db_mgr().Load_mgr();
    Xowd_page_itm tmp_page = new Xowd_page_itm();
    Int_obj_ref tmp_id = Int_obj_ref.New_zero();
    List_adp del_list = List_adp_.New();
    int ns_filter = itm.Ns_filter();
    Ordered_hash exclude_pages = Ordered_hash_.New();
    Find_excludes(exclude_pages, wiki, load_mgr, tmp_page, tmp_id, itm.Ctg_excludes());

    for (int i = 0; i < includes_len; i++) { // loop over includes
      byte[] include = (byte[]) includes.Get_at(i);
      cur_regy.Clear();
      del_list.Clear();
      Find_pages_in_ctg(cur_regy, wiki, load_mgr, tmp_page, tmp_id, include);
      Del_old_pages_not_in_cur(i, tmp_id, old_regy, cur_regy, del_list);
      Add_cur_pages_also_in_old(i, tmp_id, old_regy, cur_regy, new_regy, exclude_pages, ns_filter);
      old_regy = new_regy;
      new_regy = Ordered_hash_.New();
    }
    int pages_len = old_regy.Count();
    for (int i = 0; i < pages_len; i++) { // loop over old and create pages
      Int_obj_ref old_id = (Int_obj_ref) old_regy.Get_at(i);
      rv.Add(new Xowd_page_itm().Id_(old_id.Val()));
    }
    wiki.Db_mgr().Load_mgr().Load_by_ids(Cancelable_.Never, rv, 0, pages_len);
    rv.Sort_by(Xowd_page_itm_sorter.IdAsc);
  }
Esempio n. 2
0
  public static Keyval[] new_() {
    Ordered_hash translated = Ordered_hash_.New_bry();
    List_adp untranslated = List_adp_.New();
    Add_itm_many(
        translated,
        Xol_lang_stub_.Id_en,
        Xol_lang_stub_.Id_de,
        Xol_lang_stub_.Id_pl,
        Xol_lang_stub_.Id_zh_hans,
        Xol_lang_stub_
            .Id_zh_hant); // add langs with translations first, so they alphabetize to top of list
    int len = Xol_lang_stub_.Id__max;
    for (int i = 0; i < len; i++) { // add rest of langs, but sort by code
      Xol_lang_stub itm = Xol_lang_stub_.Get_by_id(i);
      if (translated.Has(itm.Key())) continue;
      untranslated.Add(itm);
    }
    untranslated.Sort_by(Xol_lang_stub_.Comparer_key);

    Keyval[] rv = new Keyval[len];
    int translated_max = translated.Count();
    for (int i = 0; i < translated_max; i++) rv[i] = new_itm((Xol_lang_stub) translated.Get_at(i));

    for (int i = translated_max; i < len; i++)
      rv[i] = new_itm((Xol_lang_stub) untranslated.Get_at(i - translated_max));
    return rv;
  }
public class Wbase_claim_grp_list {
  private Ordered_hash hash = Ordered_hash_.New();

  public void Add(Wbase_claim_grp itm) {
    hash.Add(itm.Id_ref(), itm);
  }

  public int Len() {
    return hash.Count();
  }

  public Wbase_claim_grp Get_at(int i) {
    return (Wbase_claim_grp) hash.Get_at(i);
  }
}
Esempio n. 4
0
class Xow_search_scanner {
  private final List_adp tkns = List_adp_.new_();
  private byte[] src;
  private int src_len, pos, txt_bgn;
  private final Ordered_hash tmp_list = Ordered_hash_.new_();
  private final Bry_bfr tmp_bfr = Bry_bfr.new_();

  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;
  }

  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));
  }

  private Xow_search_tkn new_tkn(byte tid, int val_bgn, int val_end) {
    return Xow_search_tkn.new_pos(tid, val_bgn, val_end);
  }

  private static final byte[] Bry_and = Bry_.new_a7("AND");
  private static final Btrie_slim_mgr trie =
      Btrie_slim_mgr.ci_ascii_() // NOTE:ci.ascii:OR / AND only
          .Add_str_byte(" ", Xow_search_tkn.Tid_space)
          .Add_str_byte("\"", Xow_search_tkn.Tid_quote)
          .Add_str_byte("-", Xow_search_tkn.Tid_not)
          .Add_str_byte("(", Xow_search_tkn.Tid_paren_bgn)
          .Add_str_byte(")", Xow_search_tkn.Tid_paren_end)
          .Add_str_byte("or", Xow_search_tkn.Tid_or)
          .Add_str_byte("and", Xow_search_tkn.Tid_and);
}
Esempio n. 5
0
 private static Ordered_hash Make_wkrs(Merge2_wkr... wkrs) {
   Ordered_hash rv = Ordered_hash_.New();
   for (Merge2_wkr wkr : wkrs) rv.Add(wkr.Tbl().Tbl_name(), wkr);
   return rv;
 }