public void Chk(byte[] wiki_domain, Io_url css_fil) {
this.wiki_domain = wiki_domain;
List_adp img_list = List_adp_.New();
byte[] old_bry = Io_mgr.Instance.LoadFilBry(css_fil);
byte[] rel_url_prefix = Bry_.Add(Bry_fwd_slashes, wiki_domain);
byte[] new_bry = Convert_to_local_urls(rel_url_prefix, old_bry, img_list);
Io_url img_dir = css_fil.OwnerDir();
Download_fils(img_dir, img_list.To_str_ary());
Io_mgr.Instance.SaveFilBry(css_fil, new_bry);
}
@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 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;
}
}
