public Xowv_wiki(Xoav_app app, byte[] domain_bry, Io_url wiki_root_dir) {
this.app = app;
this.domain_bry = domain_bry;
this.domain_str = String_.new_u8(domain_bry);
this.domain_itm = Xow_domain_itm_.parse(domain_bry);
this.domain_tid = domain_itm.Domain_type_id();
this.domain_abrv = Xow_abrv_wm_.To_abrv(Xow_domain_itm_.parse(domain_bry));
this.ns_mgr = Xow_ns_mgr_.default_(app.Utl_case_mgr());
this.lang =
app.Lang_mgr()
.Get_by_or_en(
domain_itm
.Lang_actl_key()); // NOTE: must not be null, or causes null ref exception on
// redlinks in drd; DATE:2016-06-28
this.msg_mgr = new Xow_msg_mgr(this, lang);
this.html__hdump_mgr = new Xow_hdump_mgr(this);
this.special_mgr = new Xosp_special_mgr(this);
this.fsys_mgr =
new Xow_fsys_mgr(wiki_root_dir, app.Fsys_mgr().File_dir().GenSubDir(domain_str));
this.fsdb_mgr = new Xof_fsdb_mgr__sql();
this.url__parser = new Xow_url_parser(this);
this.xwiki_mgr = new Xow_xwiki_mgr(this);
this.stats = new Xow_site_stats_mgr(this);
this.lnki_bldr = new Xoh_lnki_bldr(app, href_wtr);
this.ctg_catpage_mgr = new Xoctg_catpage_mgr(this);
}
public void Init_by_wiki() {
if (!init_needed) return;
init_needed = false;
if (String_.Eq(domain_str, "xowa"))
return; // HACK: ignore "xowa" for now; WHEN:converting xowa to sqlitedb
data_mgr__core_mgr = new Xow_db_mgr(this, fsys_mgr.Root_dir());
data_mgr__core_mgr.Init_by_load(gplx.xowa.wikis.data.Xow_db_file__core_.Find_core_fil(this));
app.Html__css_installer()
.Install(this, Xowd_css_core_mgr.Key_mobile); // must init after data_mgr
this.db_core_mgr = Fsdb_db_mgr_.new_detect(this, fsys_mgr.Root_dir(), fsys_mgr.File_dir());
if (db_core_mgr
== null) // no fsdb; occurs during merge; also, will be null for xowa_db; DATE:2016-05-31
db_core_mgr = Fsdb_db_mgr__v2_bldr.Get_or_make(this, true);
else // fsdb exists; load it
fsdb_mgr.Mnt_mgr().Ctor_by_load(db_core_mgr);
file_mgr__repo_mgr.Add_repo(
app, fsys_mgr.File_dir(), Xow_domain_itm_.Bry__commons, Xow_domain_itm_.Bry__commons);
file_mgr__repo_mgr.Add_repo(app, fsys_mgr.File_dir(), domain_bry, domain_bry);
Xof_orig_tbl[] orig_tbls =
db_core_mgr == null ? new Xof_orig_tbl[0] : db_core_mgr.File__orig_tbl_ary();
orig_mgr.Init_by_wiki(this, file_mgr__fsdb_mode, orig_tbls, Xof_url_bldr.new_v2());
fsdb_mgr.Init_by_wiki(this);
data_mgr__core_mgr.Db__core().Tbl__ns().Select_all(ns_mgr);
data_mgr__core_mgr.Db__core().Tbl__site_stats().Select(stats);
html__hdump_mgr.Init_by_db(this);
app.Addon_mgr().Load_by_wiki(this);
ctg_pagebox_wtr.Init_by_wiki(this);
}
public static Fsdb_db_mgr new_src_bin_db_mgr(Xow_wiki wiki, String version) {
String domain_str = wiki.Domain_str();
Fsdb_db_mgr rv = null;
Io_url url = null;
if (String_.Eq(version, "v1")) {
url =
wiki.Fsys_mgr()
.File_dir()
.OwnerDir()
.GenSubDir(domain_str + "-prv"); // v1: EX: /xowa/file/en.wikipedia.org-prv/
rv = new Fsdb_db_mgr__v1(url);
} else if (String_.Eq(version, "v2")) {
url = wiki.Fsys_mgr().Root_dir().GenSubDir("prv"); // v2: EX: /xowa/wiki/en.wikipedia.org/prv/
rv = Fsdb_db_mgr_.new_detect(wiki, url, url); // note that v2 is prioritized over v1
} else throw Err_.new_wo_type("fsdb.make:unknown fsdb_type", "version", version);
if (rv == null)
throw Err_.new_wo_type(
"fsdb.make:source fsdb not found", "version", version, "url", url.Raw());
return rv;
}
private void Download_exec(Xodb_tbl_oimg_xfer_itm fsdb) {
Io_stream_rdr src_rdr = src_bin_mgr.Find_as_rdr(Xof_exec_tid.Tid_wiki_page, fsdb);
try {
if (src_rdr == Io_stream_rdr_.Noop) { // download failed
++exec_fail;
usr_dlg.Warn_many(
"",
"",
"failed: ttl=~{0}",
String_.Format("[[File:{0}|{1}px]]", fsdb.Orig_ttl(), fsdb.Html_w()));
Print_progress(fsdb);
} else { // download passed
long src_rdr_len = src_rdr.Len();
int lnki_tier_id = fsdb.Lnki_tier_id();
if (src_rdr_len > download_size_max && !Int_.In(lnki_tier_id, download_keep_tier_ids)) {
usr_dlg.Warn_many(
"",
"",
"skipped; ttl=~{0} w=~{1} size=~{2} tier=~{3}",
fsdb.Orig_ttl(),
fsdb.Lnki_w(),
src_rdr_len,
lnki_tier_id);
return;
}
if (trg_bin_fil == null) // no trg_bin_fil
Make_trg_bin_file(Bool_.Y, fsdb, src_rdr_len);
else if (trg_bin_fil.Bin_len() + src_rdr_len
> trg_bin_db_max) // or trg_bin_fil is out of space
Make_trg_bin_file(Bool_.N, fsdb, src_rdr_len);
else if (prv_lnki_tier_id != lnki_tier_id) { // or tier has changed
if (prv_lnki_tier_id != -1 && !bin_db_mgr.Schema_is_1()) // do not increment dbs for v1
Make_trg_bin_file(Bool_.Y, fsdb, src_rdr_len);
prv_lnki_tier_id = lnki_tier_id;
}
trg_bin_updater.Save_bin(trg_mnt_itm, trg_atr_fil, trg_bin_fil, fsdb, src_rdr, src_rdr_len);
}
} finally {
src_rdr.Rls();
}
}
@Override
public void Cmd_run() {
wiki.Init_by_wiki();
Xow_db_file core_db = wiki.Data__core_mgr().Db__core();
Db_conn core_db_conn = core_db.Conn();
Gfo_usr_dlg usr_dlg = Gfo_usr_dlg_.Instance;
usr_dlg.Plog_many("", "", "creating page_filter");
if (!core_db_conn.Meta_tbl_exists("page_filter")) {
core_db_conn.Meta_tbl_create(
Dbmeta_tbl_itm.New(
"page_filter",
new Dbmeta_fld_itm[] {
Dbmeta_fld_itm.new_int("page_id").Primary_y_(),
Dbmeta_fld_itm.new_int("page_text_db_id")
},
Dbmeta_idx_itm.new_normal_by_tbl(
"page_filter", "db_id__page", "page_text_db_id", "page_id"),
Dbmeta_idx_itm.new_normal_by_tbl("page_filter", "page_id", "page_id")));
}
core_db_conn.Exec_sql_plog_ntx(
"finding missing redirects",
String_.Concat_lines_nl_skip_last(
"INSERT INTO page_filter (page_id, page_text_db_id)",
"SELECT ptr.page_id, ptr.page_text_db_id",
"FROM page ptr",
" LEFT JOIN page orig ON ptr.page_redirect_id = orig.page_id",
"WHERE ptr.page_is_redirect = 1",
"AND orig.page_id IS NULL",
"UNION",
"SELECT ptr.page_id, ptr.page_text_db_id",
"FROM page ptr",
"WHERE ptr.page_is_redirect = 1",
"AND ptr.page_redirect_id = -1",
";"));
try {
Xow_db_file[] db_file_ary =
core_db.Tbl__db().Select_all(wiki.Data__core_mgr().Props(), wiki.Fsys_mgr().Root_dir());
int len = db_file_ary.length;
for (int i = 0; i < len; ++i) {
boolean db_file_is_text = Bool_.N, db_file_is_cat = Bool_.N, db_file_is_search = Bool_.N;
Xow_db_file db_file = db_file_ary[i];
switch (db_file.Tid()) {
case Xow_db_file_.Tid__core:
case Xow_db_file_.Tid__wiki_solo:
case Xow_db_file_.Tid__text_solo:
if (wiki.Data__core_mgr().Props().Layout_text().Tid_is_lot())
continue; // if mode is lot, then "core" db does not have text, cat, search; skip;
// DATE:2016-01-31
db_file_is_text = db_file_is_cat = db_file_is_search = Bool_.Y;
break;
case Xow_db_file_.Tid__text:
db_file_is_text = Bool_.Y;
break;
case Xow_db_file_.Tid__cat:
db_file_is_cat = Bool_.Y;
break;
case Xow_db_file_.Tid__search_core:
db_file_is_search = Bool_.Y;
break;
}
int db_id = db_file.Id();
if (db_file_is_text)
Run_sql(
core_db_conn,
db_file.Url(),
db_id,
"deleting text: " + db_id,
"DELETE FROM <data_db>text WHERE page_id IN (SELECT page_id FROM page_filter WHERE page_text_db_id = {0});");
if (db_file_is_cat)
Run_sql(
core_db_conn,
db_file.Url(),
db_id,
"deleting cat: " + db_id,
"DELETE FROM <data_db>cat_link WHERE cl_from IN (SELECT page_id FROM page_filter);");
if (db_file_is_search)
Run_sql(
core_db_conn,
db_file.Url(),
db_id,
"deleting search:" + db_id,
"DELETE FROM <data_db>search_link WHERE page_id IN (SELECT page_id FROM page_filter);");
if (db_file_is_text || db_file_is_cat || db_file_is_search) db_file.Conn().Env_vacuum();
}
} catch (Exception e) {
Gfo_usr_dlg_.Instance.Warn_many(
"", "", "fatal error during page deletion: err=~{0}", Err_.Message_gplx_log(e));
}
core_db_conn.Exec_sql_plog_ntx(
"deleting from table: page",
"DELETE FROM page WHERE page_id IN (SELECT page_id FROM page_filter);");
// core_db_conn.Meta_tbl_delete("page_filter");
core_db_conn.Env_vacuum();
usr_dlg.Plog_many("", "", "");
}
String Make_key(int ns_id, byte[] ttl) {
return Int_.To_str(ns_id) + "|" + String_.new_u8(ttl);
}
@Override
public void Cmd_bgn(Xob_bldr bldr) {
wiki.Init_assert();
this.poll_interval = poll_mgr.Poll_interval();
this.bin_db_mgr = new Xob_bin_db_mgr(ns_ids);
// src_bin_mgr
if (src_bin_mgr__fsdb_version != null) {
this.src_fsdb_wkr = Xof_bin_wkr__fsdb_sql.new_(wiki.File__mnt_mgr());
src_bin_mgr.Wkrs__add(src_fsdb_wkr);
src_fsdb_wkr.Mnt_mgr().Ctor_by_load(new_src_bin_db_mgr(wiki, src_bin_mgr__fsdb_version));
src_fsdb_wkr.Mnt_mgr()
.Mnts__get_main()
.Txn_bgn(); // NOTE: txn on atr speeds up from 50 -> 300; DATE:2015-03-21
if (src_bin_mgr__fsdb_skip_wkrs != null) {
src_fsdb_wkr.Skip_mgr_init(
src_fsdb_wkr.Mnt_mgr().Mnts__get_main().Cfg_mgr(), src_bin_mgr__fsdb_skip_wkrs);
}
if (src_bin_mgr__cache_enabled) {
usr_dlg.Prog_many("", "", "src_bin_mgr.cache.bgn");
src_fsdb_wkr.Mnt_mgr().Mnts__get_main().Atr_mgr().Db__core().Fil_cache_enabled_y_();
usr_dlg.Prog_many("", "", "src_bin_mgr.cache.end");
}
}
if (src_bin_mgr__wmf_enabled) {
Xof_bin_wkr__http_wmf wmf_wkr = Xof_bin_wkr__http_wmf.new_(wiki);
src_bin_mgr.Wkrs__add(wmf_wkr);
wmf_wkr.Fail_timeout_(
0); // 1000; NOTE: set Fail_timeout here; DATE:2014-06-21; NOTE: do not put in ctor, or
// else will be 1st wkr; DATE:2014-06-28
}
// trg_mnt_itm
this.trg_bin_db_max = app.Api_root().Bldr().Wiki().Import().File_db_max();
Io_url trg_file_dir_v1 =
String_.Eq(trg_bin_mgr__fsdb_version, "v1")
? wiki.Fsys_mgr().File_dir().GenNewNameOnly(wiki.Domain_str() + "-prv")
: wiki.Fsys_mgr().File_dir(); // NOTE: convoluted way of setting trg to -prv if
// trg_bin_mgr__fsdb_version_v1 is set; otherwise set to
// "en.wikipedia.org" which will noop; DATE:2015-12-02
Fsdb_db_mgr trg_db_mgr =
Fsdb_db_mgr_.new_detect(wiki, wiki.Fsys_mgr().Root_dir(), trg_file_dir_v1);
if (trg_db_mgr == null) trg_db_mgr = Fsdb_db_mgr__v2_bldr.Get_or_make(wiki, Bool_.Y);
Fsm_mnt_mgr trg_mnt_mgr = new Fsm_mnt_mgr();
trg_mnt_mgr.Ctor_by_load(trg_db_mgr);
trg_mnt_mgr.Mnts__get_insert_idx_(
Fsm_mnt_mgr
.Mnt_idx_main); // NOTE: do not delete; mnt_mgr default to Mnt_idx_user; DATE:2014-04-25
this.trg_mnt_itm = trg_mnt_mgr.Mnts__get_insert();
Fsm_mnt_mgr.Patch(
trg_mnt_itm.Cfg_mgr()
.Tbl()); // NOTE: always patch again; fsdb_make may be run separately without lnki_temp;
// DATE:2014-04-26
this.trg_atr_fil = trg_mnt_itm.Atr_mgr().Db__core();
this.trg_cfg_mgr = trg_mnt_itm.Cfg_mgr();
bin_db_mgr.Init_by_mnt_mgr(trg_mnt_mgr);
trg_atr_fil.Conn().Txn_bgn("bldr__fsdb_make__trg_atr_fil");
if (!trg_atr_fil.Conn()
.Eq(trg_cfg_mgr.Tbl().Conn())) // need to create txn for v1; DATE:2015-07-04
trg_cfg_mgr.Tbl().Conn().Txn_bgn("bldr__fsdb_make__trg_cfg_fil");
// bldr_db
Xob_db_file bldr_db = Xob_db_file.New__file_make(wiki.Fsys_mgr().Root_dir());
this.bldr_conn = bldr_db.Conn();
this.bldr_cfg_tbl = bldr_db.Tbl__cfg(); // NOTE: cfg and atr is in same db; use it
bldr_cfg_tbl.Conn().Txn_bgn("bldr__fsdb_make__bldr_cfg_tbl");
}
public void Test_load_page(int ns_id, int page_id, String expd) {
Xowe_wiki wiki = bldr_fxt.Wiki();
Xow_ns ns = wiki.Ns_mgr().Ids_get_or_null(ns_id);
wiki.Db_mgr_as_sql().Load_mgr().Load_page(actl.Id_(page_id), ns, false);
Tfds.Eq(expd, String_.new_a7(actl.Text()));
}
