private int Import_url_chk(
byte[] rel_url_prefix,
byte[] src,
int src_len,
int old_pos,
int find_bgn,
byte[] url_raw,
Bry_bfr bfr) {
if (find_bgn < Bry_import_len) return Bry_find_.Not_found;
if (!Bry_.Match(src, find_bgn - Bry_import_len, find_bgn, Bry_import))
return Bry_find_.Not_found;
byte[] css_url = url_raw;
int css_url_len = css_url.length;
if (css_url_len > 0
&& css_url[0]
== Byte_ascii
.Slash) { // css_url starts with "/"; EX: "/page" or "//site/page" DATE:2014-02-03
if (css_url_len > 1
&& css_url[1] != Byte_ascii.Slash) // skip if css_url starts with "//"; EX: "//site/page"
css_url = Bry_.Add(rel_url_prefix, css_url); // "/w/a.css" -> "//en.wikipedia.org/w/a.css"
}
css_url =
Bry_.Replace(
css_url,
Byte_ascii.Space,
Byte_ascii
.Underline); // NOTE: must replace spaces with underlines else download will fail;
// EX:https://it.wikivoyage.org/w/index.php?title=MediaWiki:Container e
// Infobox.css&action=raw&ctype=text/css; DATE:2015-03-08
byte[] css_src_bry = Import_url_build(stylesheet_prefix, rel_url_prefix, css_url);
String css_src_str = String_.new_u8(css_src_bry);
download_wkr.Download_xrg()
.Prog_fmt_hdr_(
usr_dlg.Log_many(
GRP_KEY, "logo.download", "downloading import for '~{0}'", css_src_str));
byte[] css_trg_bry = download_wkr.Download_xrg().Exec_as_bry(css_src_str);
if (css_trg_bry == null) {
usr_dlg.Warn_many("", "", "could not import css: url=~{0}", css_src_str);
return Bry_find_.Not_found; // css not found
}
bfr.Add_mid(src, old_pos, find_bgn - Bry_import_len).Add_byte_nl();
bfr.Add(Bry_comment_bgn).Add(css_url).Add(Bry_comment_end).Add_byte_nl();
if (Bry_find_.Find_fwd(css_url, Wikisource_dynimg_ttl) != -1)
css_trg_bry =
Bry_.Replace(
css_trg_bry,
Wikisource_dynimg_find,
Wikisource_dynimg_repl); // FreedImg hack;
// PAGE:en.s:Page:Notes_on_Osteology_of_Baptanodon._With_a_Description_of_a_New_Species.pdf/3 DATE:2014-09-06
bfr.Add(css_trg_bry).Add_byte_nl();
bfr.Add_byte_nl();
int semic_pos = Bry_find_.Find_fwd(src, Byte_ascii.Semic, find_bgn + url_raw.length, src_len);
return semic_pos + Int_.Const_dlm_len;
}
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;
}
}
@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("", "", "");
}