private void Make_trg_bin_file(boolean try_nth, Xodb_tbl_oimg_xfer_itm fsdb, long src_rdr_len) { boolean is_solo = trg_mnt_itm.Db_mgr().File__solo_file(); boolean make = true, use_txn = !is_solo; // solo file; should never open txn if (trg_bin_fil != null && use_txn) // pre-existing bin_file; trg_bin_fil.Conn().Txn_end(); // close txn before making new db int tier_id = fsdb.Lnki_tier_id(); Xob_bin_db_itm nth_bin_db = bin_db_mgr.Get_nth_by_tier(tier_id); if (try_nth) { // try_nth is true; occurs for new runs or changed tier if (nth_bin_db.Id() != -1 // nth exists; && nth_bin_db.Db_len() + src_rdr_len < trg_bin_db_max) { // if src_rdr_len exceeds make = false; // do not make; use existing } } if (make) { // no nth; make it; int ns_id = bin_db_mgr.Get_ns_id(tier_id); int pt_id = bin_db_mgr.Increment_pt_id(nth_bin_db); String new_bin_db_name = bin_db_mgr.Gen_name(wiki.Domain_str(), ns_id, pt_id); this.trg_bin_fil = trg_mnt_itm.Bin_mgr().Dbs__make(new_bin_db_name); if (!trg_mnt_itm.Db_mgr().File__solo_file()) { Fsdb_db_file trg_bin_db = trg_mnt_itm.Db_mgr() .File__bin_file__at(trg_mnt_itm.Id(), trg_bin_fil.Id(), new_bin_db_name); if (!bin_db_mgr.Schema_is_1()) Fsdb_db_mgr__v2_bldr.Make_cfg_data( wiki, trg_atr_fil.Url_rel(), trg_bin_db, Xow_db_file_.Tid__file_data, trg_bin_fil.Id() + List_adp_.Base1); } } else { // nth available; use it this.trg_bin_fil = trg_mnt_itm.Bin_mgr().Dbs__get_at(nth_bin_db.Id()); trg_bin_fil.Bin_len_(nth_bin_db.Db_len()); } if (use_txn) trg_bin_fil.Conn().Txn_bgn("bldr__fsdb_make__trg_bin_fil"); }
@Override public void Cmd_run() { Init_bldr_bmks(); this.time_bgn = System_.Ticks(); int total_pending = Xob_xfer_regy_tbl.Select_total_pending(bldr_conn); // if (total_pending > 250000 && src_bin_mgr__fsdb_version == null) usr_dlg.Note_many("", "", "total pending: ~{0}", total_pending); List_adp list = List_adp_.New(); boolean loop = true; while (loop) { byte rslt = Select_fsdb_itms(list); switch (rslt) { case Select_rv_stop: if (bin_db_mgr.Tier_id_is_last(tier_id_val)) loop = false; else { ++tier_id_val; page_id_val = 0; continue; } break; case Select_rv_next_page: ++page_id_val; lnki_id_val = 0; continue; case Select_rv_process: break; } if (!loop) break; // no more ttls found int len = list.Count(); usr_dlg.Prog_many("", "", "fetched pages: ~{0}", len); for (int i = 0; i < len; ++i) { Xodb_tbl_oimg_xfer_itm fsdb = (Xodb_tbl_oimg_xfer_itm) list.Get_at(i); Download_itm(fsdb); if (exit_now || exec_count >= exec_count_max || exec_fail >= exec_fail_max || page_id_val >= page_id_end) { this.Txn_sav(); return; } } } exec_done = true; }
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_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"); }