public class Xob_xfer_regy_log_tbl {
public static void Create_table(Db_conn p) {
gplx.dbs.engines.sqlite.Sqlite_engine_.Tbl_create_and_delete(p, Tbl_name, Tbl_sql);
}
public static Db_stmt Insert_stmt(Db_conn p) {
return Db_stmt_.new_insert_(
p, Tbl_name, Fld_lnki_id, Fld_xfer_status, Fld_xfer_bin_tid, Fld_xfer_bin_msg);
}
public static void Insert(Db_stmt stmt, byte status, int id, byte wkr_tid, String wkr_msg) {
stmt.Clear().Val_int(id).Val_byte(status).Val_byte(wkr_tid).Val_str(wkr_msg).Exec_insert();
}
private static final String Tbl_sql =
String_.Concat_lines_nl(
"CREATE TABLE IF NOT EXISTS xfer_regy_log",
"( lnki_id integer NOT NULL",
", xfer_status tinyint NOT NULL" // 0=todo; 1=fail; 2=pass; 3=done
,
", xfer_bin_tid tinyint NOT NULL",
", xfer_bin_msg varchar(255) NOT NULL",
");");
public static final String Tbl_name = "xfer_regy_log",
Fld_lnki_id = "lnki_id",
Fld_xfer_status = "xfer_status",
Fld_xfer_bin_tid = "xfer_bin_tid",
Fld_xfer_bin_msg = "xfer_bin_msg";
}
public void Select_tasks_steps(
Xobc_task_step_hash task_step_hash,
Xobc_step_map_tbl step_map_tbl,
byte[] wiki_abrv,
String wiki_date) {
task_step_hash.Clear();
Db_rdr rdr =
conn.Stmt_sql(
String_.Concat_lines_nl_skip_last(
"SELECT DISTINCT sm.task_id, sm.step_id",
"FROM " + tbl_name + " imps",
" JOIN " + step_map_tbl.Tbl_name() + " sm ON sm.step_id = imps.step_id",
"WHERE imps.wiki_abrv = ?",
"AND imps.wiki_date = ?"))
.Crt_bry_as_str(fld_wiki_abrv, wiki_abrv)
.Crt_str(fld_wiki_date, wiki_date)
.Exec_select__rls_auto();
try {
while (rdr.Move_next()) {
task_step_hash.Add(rdr.Read_int("task_id"), rdr.Read_int("step_id"));
}
} finally {
rdr.Rls();
}
}
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;
}
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();
}
}
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);
}
@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");
}