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 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 List_adp Load(
Xomp_mgr_db mgr_db, String machine_name, List_adp list, int list_idx, int list_len) {
List_adp rv = List_adp_.New();
// add remaining pages from old pool to new_pool;
for (int i = list_idx; i < list_len; ++i) {
rv.Add((Xomp_page_itm) list.Get_at(i));
}
// load pages into new pool
Xomp_lock_mgr lock_mgr = mgr_db.Lock_mgr();
int uid_db = lock_mgr.Uid_prv__get(machine_name);
if (uid_db == Xomp_lock_mgr__fsys.Uid__eos)
return rv; // assert that uids must be incrementally larger; handle one machine reaching end,
// and putting -1 in queue;
int uid_new = 0;
try {
uid_new = this.Load_from_db(rv, uid_db);
} finally {
lock_mgr.Uid_prv__rls(machine_name, uid_new);
}
if (show_msg__fetched_pool)
Gfo_usr_dlg_.Instance.Note_many(
"", "", "fetched new pool: old=~{0} new=~{1}", uid_db, uid_new);
return rv;
}
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);
}
public Xoud_site_row[] Select_all() {
List_adp rv = List_adp_.new_();
Db_rdr rdr = conn.Stmt_select(tbl_name, flds).Exec_select__rls_auto();
try {
while (rdr.Move_next()) rv.Add(new_row(rdr));
return (Xoud_site_row[]) rv.To_ary_and_clear(Xoud_site_row.class);
} finally {
rdr.Rls();
}
}
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);
}
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()));
}
}
private int Load_from_db(List_adp list, int uid_prv) {
// prepare for page_tbl
String sql =
String_.Format(
String_.Concat_lines_nl_skip_last // ANSI.Y
(
"SELECT mp.xomp_uid",
", pp.page_id",
", pp.page_namespace",
", pp.page_title",
", pp.page_text_db_id",
"FROM xomp_page mp",
" JOIN <page_db>page pp ON mp.page_id = pp.page_id",
"WHERE mp.xomp_uid > {0}",
"AND mp.page_status = 0",
"LIMIT {1}"),
uid_prv,
num_pages_per_load);
this.attach_mgr.Conn_links_(
new Db_attach_itm("page_db", wiki.Data__core_mgr().Db__core().Conn()));
sql = attach_mgr.Resolve_sql(sql);
// run page_tbl
int rv = -1;
Xomp_text_db_loader text_db_loader = new Xomp_text_db_loader(wiki);
attach_mgr.Attach();
Db_rdr rdr = make_conn.Stmt_sql(sql).Exec_select__rls_auto();
try {
while (rdr.Move_next()) {
rv = rdr.Read_int("xomp_uid");
int text_db_id = rdr.Read_int("page_text_db_id");
Xomp_page_itm ppg = new Xomp_page_itm(rdr.Read_int("page_id"));
ppg.Init_by_page(
rdr.Read_int("page_namespace"), rdr.Read_bry_by_str("page_title"), text_db_id);
list.Add(ppg);
text_db_loader.Add(text_db_id, ppg);
}
} finally {
rdr.Rls();
}
attach_mgr.Detach();
text_db_loader.Load();
return rv;
}
private static void New_subs(Io_url owner_dir, List_adp list, IoItmList subs, byte[] dir_cmd) {
subs.Sort();
int len = subs.Len();
int list_total = list.Len();
byte[] owner_dir_bry = owner_dir.RawBry();
for (int i = 0; i < len; ++i) {
IoItm_base src = (IoItm_base) subs.Get_at(i);
Xow_import_doc trg = null;
if (src.Type_dir()) {
byte[] trg_url = src.Url().RawBry();
trg =
new Xow_import_doc(
Bool_.Y,
Bool_.N,
list_total + i,
owner_dir_bry,
trg_url,
Bry_.new_u8(src.Url().NameAndExt_noDirSpr()),
Bry_.Empty,
Bry_.Empty,
dir_cmd,
Ary_empty);
} else {
IoItmFil src_as_fil = (IoItmFil) src;
String size_str = Io_size_.To_str(src_as_fil.Size(), "#,###");
boolean is_xowa_core =
gplx.xowa.wikis.data.Xow_db_file__core_.Is_core_fil_name(
owner_dir.NameOnly(), src.Url().NameAndExt());
trg =
new Xow_import_doc(
Bool_.N,
is_xowa_core,
list_total + i,
owner_dir_bry,
src.Url().RawBry(),
Bry_.new_u8(src.Name()),
Bry_.new_u8(src_as_fil.ModifiedTime().XtoStr_fmt("yyyy-MM-dd")),
Bry_.new_u8(size_str),
dir_cmd,
Ary_empty);
}
list.Add(trg);
}
}
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;}
}
public Xobc_import_step_itm[] Select_by_task_id(int task_id) {
List_adp list = List_adp_.New();
Db_rdr rdr =
conn.Stmt_sql(
Db_sql_.Make_by_fmt(
String_.Ary(
"SELECT s.*",
"FROM import_step s",
" JOIN step_map sm ON s.step_id = sm.step_id",
"WHERE sm.task_id = {0}"),
task_id))
.Exec_select__rls_auto();
try {
while (rdr.Move_next()) {
list.Add(New_itm(rdr));
}
} finally {
rdr.Rls();
}
return (Xobc_import_step_itm[]) list.To_ary_and_clear(Xobc_import_step_itm.class);
}
public Xol_interval_itm[] Get_duration_intervals(long seconds, Xol_duration_itm[] intervals) {
if (intervals == null) intervals = Xol_duration_itm_.Ary_default;
Array_.Sort(intervals, Xol_duration_itm_sorter.Instance);
int intervals_len = intervals.length;
long val = seconds;
List_adp rv = List_adp_.New();
for (int i = 0; i < intervals_len; i++) {
Xol_duration_itm itm = intervals[i];
long itm_seconds = itm.Seconds();
val = seconds / itm_seconds;
if (val > 0
|| (i == intervals_len - 1
&& rv.Count()
== 0) // always add one seg; EX: 40 seconds, but minutes requested -> 0 minutes;
// DATE:2014-05-10
) {
seconds -= val * itm_seconds;
rv.Add(new Xol_interval_itm(itm, val));
}
}
return (Xol_interval_itm[]) rv.To_ary(Xol_interval_itm.class);
}
public static int[] Ary_get(Xow_domain_itm domain_itm, boolean wmf) {
if (wmf && domain_itm != null && domain_itm.Domain_type().Tid() != Xow_domain_tid_.Tid__home)
return Ary_wmf;
if (Ary_nonwmf == null) {
List_adp list = List_adp_.New();
int len = Ary_wmf.length;
for (int i = 0; i < len; ++i) {
list.Add(Ary_wmf[i]);
}
list.Add_many(
Xol_kwd_grp_.Id_strx_len,
Xol_kwd_grp_.Id_strx_pos,
Xol_kwd_grp_.Id_strx_rpos,
Xol_kwd_grp_.Id_strx_sub,
Xol_kwd_grp_.Id_strx_count,
Xol_kwd_grp_.Id_strx_replace,
Xol_kwd_grp_.Id_strx_explode,
Xol_kwd_grp_.Id_strx_urldecode,
Xol_kwd_grp_.Id_new_window_link);
Ary_nonwmf = (int[]) list.To_ary_and_clear(int.class);
}
return Ary_nonwmf;
}
private byte Select_fsdb_itms(List_adp list) {
list.Clear();
boolean pages_found = false, links_found = false;
DataRdr rdr =
Xob_xfer_regy_tbl.Select_by_tier_page(bldr_conn, tier_id_val, page_id_val, select_interval);
try {
while (rdr.MoveNextPeer()) {
pages_found = true; // at least one page found; set true
Xodb_tbl_oimg_xfer_itm itm = Xodb_tbl_oimg_xfer_itm.new_rdr_(rdr);
if (itm.Lnki_page_id() == page_id_val // same page_id
&& itm.Lnki_id() <= lnki_id_val // ... but lnki_id < last
) continue; // ... ignore; note that select is by page_id, not page_id + link_id; needed
// else restarts would not resume exactly at same point;
links_found = true;
list.Add(itm);
}
} finally {
rdr.Rls();
}
if (pages_found && !links_found)
return Select_rv_next_page; // pages found, but all links processed
else if (!pages_found) return Select_rv_stop; // no more pages found
else return Select_rv_process; // pages and links found
}
public byte[] Convert_to_local_urls(byte[] rel_url_prefix, byte[] src, List_adp list) {
try {
int src_len = src.length;
int prv_pos = 0;
Bry_bfr bfr = Bry_bfr_.New_w_size(src_len);
Hash_adp img_hash = Hash_adp_bry.cs();
while (true) {
int url_pos = Bry_find_.Find_fwd(src, Bry_url, prv_pos);
if (url_pos == Bry_find_.Not_found) {
bfr.Add_mid(src, prv_pos, src_len);
break;
} // no more "url("; exit;
int bgn_pos = url_pos + Bry_url_len; // set bgn_pos after "url("
byte bgn_byte = src[bgn_pos];
byte end_byte = Byte_ascii.Null;
boolean quoted = true;
switch (bgn_byte) { // find end_byte
case Byte_ascii.Quote:
case Byte_ascii.Apos: // quoted; end_byte is ' or "
end_byte = bgn_byte;
++bgn_pos;
break;
default: // not quoted; end byte is ")"
end_byte = Byte_ascii.Paren_end;
quoted = false;
break;
}
int end_pos = Bry_find_.Find_fwd(src, end_byte, bgn_pos, src_len);
if (end_pos
== Bry_find_.Not_found) { // unclosed "url("; exit since nothing else will be found
usr_dlg.Warn_many(
GRP_KEY,
"parse.invalid_url.end_missing",
"could not find end_sequence for 'url(': bgn='~{0}' end='~{1}'",
prv_pos,
String_.new_u8__by_len(src, prv_pos, prv_pos + 25));
bfr.Add_mid(src, prv_pos, src_len);
break;
}
if (end_pos - bgn_pos == 0) { // empty; "url()"; ignore
usr_dlg.Warn_many(
GRP_KEY,
"parse.invalid_url.empty",
"'url(' is empty: bgn='~{0}' end='~{1}'",
prv_pos,
String_.new_u8__by_len(src, prv_pos, prv_pos + 25));
bfr.Add_mid(src, prv_pos, bgn_pos);
prv_pos = bgn_pos;
continue;
}
byte[] img_raw = Bry_.Mid(src, bgn_pos, end_pos);
int img_raw_len = img_raw.length;
if (Bry_.Has_at_bgn(img_raw, Bry_data_image, 0, img_raw_len)) { // base64
bfr.Add_mid(src, prv_pos, end_pos); // nothing to download; just add entire String
prv_pos = end_pos;
continue;
}
int import_url_end =
Import_url_chk(
rel_url_prefix,
src,
src_len,
prv_pos,
url_pos,
img_raw,
bfr); // check for embedded stylesheets via @import tag
if (import_url_end != Bry_find_.Not_found) {
prv_pos = import_url_end;
continue;
}
byte[] img_cleaned = Xob_url_fixer.Fix(wiki_domain, img_raw, img_raw_len);
if (img_cleaned == null) { // could not clean img
usr_dlg.Warn_many(
GRP_KEY,
"parse.invalid_url.clean_failed",
"could not extract valid http src: bgn='~{0}' end='~{1}'",
prv_pos,
String_.new_u8(img_raw));
bfr.Add_mid(src, prv_pos, bgn_pos);
prv_pos = bgn_pos;
continue;
}
if (!img_hash.Has(img_cleaned)) { // only add unique items for download;
img_hash.Add_as_key_and_val(img_cleaned);
list.Add(String_.new_u8(img_cleaned));
}
img_cleaned =
Replace_invalid_chars(
Bry_.Copy(
img_cleaned)); // NOTE: must call ByteAry.Copy else img_cleaned will change
// *inside* hash
bfr.Add_mid(src, prv_pos, bgn_pos);
if (!quoted) bfr.Add_byte(Byte_ascii.Quote);
bfr.Add(img_cleaned);
if (!quoted) bfr.Add_byte(Byte_ascii.Quote);
prv_pos = end_pos;
}
return bfr.To_bry_and_clear();
} catch (Exception e) {
usr_dlg.Warn_many(
"",
"",
"failed to convert local_urls: ~{0} ~{1}",
String_.new_u8(rel_url_prefix),
Err_.Message_gplx_full(e));
return src;
}
}
public void Add(Xog_tab_close_lnr lnr) {
list.Add(lnr);
}