private static void Find_pages_in_ctg(
Ordered_hash rv,
Xowe_wiki wiki,
Xodb_load_mgr load_mgr,
Xowd_page_itm tmp_page,
Int_obj_ref tmp_id,
byte[] ctg_ttl) {
Xoctg_catpage_ctg ctg =
wiki.Ctg__catpage_mgr()
.Get_or_load_or_null(
Xoctg_catpage_url.New__blank(),
wiki.Ttl_parse(gplx.xowa.wikis.nss.Xow_ns_.Tid__category, ctg_ttl),
-1);
if (ctg == null) return;
// loop grps to get grp
for (byte ctg_tid = 0; ctg_tid < Xoa_ctg_mgr.Tid___max; ++ctg_tid) {
Xoctg_catpage_grp ctg_grp = ctg.Grp_by_tid(ctg_tid);
int itms_len = ctg_grp.Itms__len();
// loop itms in grp and add to hash
for (int i = 0; i < itms_len; ++i) {
Xoctg_catpage_itm ctg_itm = ctg_grp.Itms__get_at(i);
int itm_page_id = ctg_itm.Page_id();
if (rv.Has(tmp_id.Val_(itm_page_id))) continue;
rv.Add(Int_obj_ref.New(itm_page_id), ctg_itm);
// DELETE: recurse subcategories; PAGE:en.b:XML DATE:2016-09-18
// if (ctg_tid == Xoa_ctg_mgr.Tid__subc) {
// load_mgr.Load_by_id(tmp_page, itm_page_id);
// Find_pages_in_ctg(rv, wiki, load_mgr, tmp_page, tmp_id, tmp_page.Ttl_page_db());
// }
}
}
}
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);
}
private static void Del_old_pages_not_in_cur(
int i, Int_obj_ref tmp_id, Ordered_hash old_regy, Ordered_hash cur_regy, List_adp del_list) {
if (i == 0) return; // skip logic for first ctg (which doesn't have a predecessor)
int old_len = old_regy.Count();
for (int j = 0; j < old_len; j++) { // if cur is not in new, del it
Int_obj_ref old_id = (Int_obj_ref) old_regy.Get_at(j);
if (!cur_regy.Has(tmp_id.Val_(old_id.Val()))) // old_itm does not exist in cur_regy
del_list.Add(
old_id); // remove; EX: (A,B) in old; B only in cur; old should now be (A) only
}
int del_len = del_list.Count();
for (int j = 0; j < del_len; j++) {
Int_obj_ref old_itm = (Int_obj_ref) del_list.Get_at(j);
old_regy.Del(tmp_id.Val_(old_itm.Val()));
}
}
public static byte[] Write_wdata_links(List_adp slink_list, Xowe_wiki wiki, Xoa_ttl ttl, Wdata_external_lang_links_data external_links_mgr) {
try {
switch (wiki.Domain_tid()) {
case Xow_domain_type_.Tid_home: // home will never be in wikidata
case Xow_domain_type_.Tid_wikidata: // wikidata will never be in wikidata
return Qid_null;
}
Wdata_wiki_mgr wdata_mgr = wiki.Appe().Wiki_mgr().Wdata_mgr();
Wdata_doc doc = wdata_mgr.Pages_get(wiki, ttl); if (doc == null) return Qid_null; // no links
boolean external_links_mgr_enabled = external_links_mgr.Enabled();
Ordered_hash links = doc.Slink_list();
Bry_bfr tmp_bfr = wiki.Appe().Utl__bfr_mkr().Get_k004();
Xow_wiki_abrv wiki_abrv = new Xow_wiki_abrv();
int len = links.Count();
for (int i = 0; i < len; i++) {
Wdata_sitelink_itm slink = (Wdata_sitelink_itm)links.Get_at(i);
byte[] xwiki_key = slink.Site();
Xow_wiki_abrv_.parse_(wiki_abrv, xwiki_key, 0, xwiki_key.length);
if (wiki_abrv.Domain_tid() == Xow_wiki_abrv_.Tid_null) {
wiki.Appe().Usr_dlg().Warn_many("", "", "unknown wiki in wikidata: ttl=~{0} wiki=~{1}", ttl.Page_db_as_str(), String_.new_u8(xwiki_key));
continue;
}
if (wiki_abrv.Domain_tid() != wiki.Domain_tid()) continue; // ignore wikis in a different domain; EX: looking at enwiki:Earth, and wikidata has dewikiquote; ignore dewikiquote; DATE:2014-06-21
byte[] lang_key = wiki_abrv.Lang_itm().Key();
if (external_links_mgr_enabled && external_links_mgr.Langs_hide(lang_key, 0, lang_key.length)) continue;
tmp_bfr.Add(lang_key);
tmp_bfr.Add_byte(Byte_ascii.Colon);
tmp_bfr.Add(slink.Name());
Xoa_ttl slink_ttl = Xoa_ttl.parse_(wiki, tmp_bfr.Xto_bry_and_clear());
if (slink_ttl == null) continue; // invalid ttl
Xow_xwiki_itm xwiki_itm = slink_ttl.Wik_itm();
if ( xwiki_itm == null // not a known xwiki; EX: [[zzz:abc]]
|| Bry_.Eq(xwiki_itm.Domain_bry(), wiki.Domain_bry()) // skip if same as self; i.e.: do not include links to enwiki if already in enwiki
) continue;
slink.Page_ttl_(slink_ttl);
slink_list.Add(slink);
}
tmp_bfr.Mkr_rls();
if (external_links_mgr_enabled && external_links_mgr.Sort())
slink_list.Sort_by(Xoa_ttl_sorter._);
return doc.Qid();
} catch (Exception e) {Err_.Noop(e); return Qid_null;}
}
private static void Add_cur_pages_also_in_old(
int i,
Int_obj_ref tmp_id,
Ordered_hash old_regy,
Ordered_hash cur_regy,
Ordered_hash new_regy,
Ordered_hash exclude_pages,
int ns_filter) {
int found_len = cur_regy.Count();
for (int j = 0; j < found_len; j++) { // if new_page is in cur, add it
Xoctg_catpage_itm cur_itm = (Xoctg_catpage_itm) cur_regy.Get_at(j);
Xoa_ttl cur_ttl = cur_itm.Page_ttl();
if (cur_ttl == null) continue;
if (ns_filter != Dpl_itm.Ns_filter_null && ns_filter != cur_ttl.Ns().Id()) continue;
tmp_id.Val_(cur_itm.Page_id()); // set tmp_id, since it will be used at least once
if (exclude_pages.Has(tmp_id)) continue; // ignore excluded pages
if (i != 0) { // skip logic for first ctg (which doesn't have a predecessor)
if (!old_regy.Has(tmp_id)) continue; // cur_itm not in old_regy; ignore
}
new_regy.Add_as_key_and_val(Int_obj_ref.New(cur_itm.Page_id()));
}
}
public Wbase_claim_grp Get_at(int i) {
return (Wbase_claim_grp) hash.Get_at(i);
}
public int Len() {
return hash.Count();
}
public void Add(Wbase_claim_grp itm) {
hash.Add(itm.Id_ref(), itm);
}