private void Download_itm(Xodb_tbl_oimg_xfer_itm fsdb) {
try {
tier_id_val = fsdb.Lnki_tier_id();
page_id_val = fsdb.Lnki_page_id();
lnki_id_val = fsdb.Lnki_id();
fsdb.Orig_repo_name_(
fsdb.Orig_repo_id() == Xof_repo_tid_.Tid__local
? wiki.Domain_bry()
: Xow_domain_itm_.Bry__commons);
Download_exec(fsdb);
++exec_count;
if (exec_count % progress_interval == 0) Print_progress(fsdb);
if (exec_count % poll_interval == 0) poll_mgr.Poll();
if (exec_count % commit_interval == 0) Txn_sav();
if (exec_count % delete_interval == 0) Delete_files();
} catch (Exception exc) {
++exec_fail;
usr_dlg.Warn_many(
"",
"",
"download error; ttl=~{0} w=~{1} err=~{2}",
fsdb.Orig_ttl(),
fsdb.Lnki_w(),
Err_.Message_gplx_full(exc));
}
}
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;
}
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;
}
}