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[] Clean_img_url(byte[] raw, int raw_len) {
int pos_bgn = 0;
if (Bry_.Has_at_bgn(raw, Bry_fwd_slashes, 0, raw_len)) pos_bgn = Bry_fwd_slashes.length;
if (Bry_.Has_at_bgn(raw, Bry_http, 0, raw_len)) pos_bgn = Bry_http.length;
int pos_slash = Bry_find_.Find_fwd(raw, Byte_ascii.Slash, pos_bgn, raw_len);
if (pos_slash == Bry_find_.Not_found)
return null; // first segment is site_name; at least one slash must be present for image name;
// EX: site.org/img_name.jpg
if (pos_slash == raw_len - 1) return null; // "site.org/" is invalid
int pos_end = raw_len;
int pos_question = Bry_find_.Find_bwd(raw, Byte_ascii.Question);
if (pos_question != Bry_find_.Not_found)
pos_end = pos_question; // remove query params; EX: img_name?key=val
return Bry_.Mid(raw, pos_bgn, pos_end);
}
public void Special__gen(Xoa_app app, Xow_wiki wiki, Xoa_page page, Xoa_url url, Xoa_ttl ttl) {
int slash_pos =
Bry_find_.Find_fwd(ttl.Page_txt_wo_qargs(), Xoa_ttl.Subpage_spr); // check for slash
byte[] special_name =
slash_pos == Bry_find_.Not_found
? ttl
.Base_txt_wo_qarg() // no slash found; use base_txt; ignore qry args and just get
// page_names; EX: Search/Earth?fulltext=y;
// Allpages?from=Earth...
: Bry_.Mid(
ttl.Page_txt_wo_qargs(),
0,
slash_pos); // slash found; use root page; EX: Special:ItemByTitle/enwiki/Earth
special_name =
Xoa_ttl.Replace_spaces(special_name); // handle spaces; EX:Spezial:Zufällige_Seite
Xow_special_page special = (Xow_special_page) hash.Get_by_bry(special_name);
if (special != null) { // special found; generate it;
special = special.Special__clone();
page.Db().Page().Modified_on_(Datetime_now.Get());
try {
special.Special__gen(wiki, page, url, ttl);
} catch (Exception e) {
Gfo_log_.Instance.Warn(
"failed to generate special page",
"url",
url.To_str(),
"err",
Err_.Message_gplx_log(e));
}
}
}
private void Parse_href(Xoh_hdoc_ctx hctx, Gfh_tag anch_head) {
href_itm.Parse(rdr.Err_wkr(), hctx, src, anch_head);
this.href_bgn = href_itm.Ttl_bgn();
this.href_end = href_itm.Ttl_end();
switch (href_itm.Tid()) {
case Xoh_anch_href_data.Tid__wiki:
case Xoh_anch_href_data.Tid__site:
this.href_ns_id = href_itm.Ttl_ns_id();
this.href_src = href_itm.Ttl_full_txt();
this.href_bgn = 0;
this.href_end = href_src.length;
if (href_ns_id != Xow_ns_.Tid__main) { // not main; try to remove template name;
int colon_pos = Bry_find_.Find_fwd(href_src, Byte_ascii.Colon, href_bgn, href_end);
this.href_ns_name =
Xoa_ttl.Replace_unders(
Bry_.Mid(href_src, href_bgn, colon_pos + 1)); // EX: 11="Template talk:"
this.href_ns_name_len = href_ns_name.length;
}
break;
}
}
public void Get_by_url1(Xow_wiki wiki, Xoa_page page, Xoa_url url, Xoa_ttl ttl) {
int slash_pos =
Bry_find_.Find_fwd(ttl.Page_txt_wo_qargs(), Xoa_ttl.Subpage_spr); // check for slash
byte[] special_name =
slash_pos == Bry_find_.Not_found
? ttl.Base_txt_wo_qarg() // no slash found; use base_txt; ignore qry args and just get
// page_names; EX: Search/Earth?fulltext=y;
// Allpages?from=Earth...
: Bry_.Mid(
ttl.Page_txt_wo_qargs(),
0,
slash_pos); // slash found; use root page; EX: Special:ItemByTitle/enwiki/Earth
Object o = hash.Get_by_bry(special_name);
if (o == null) {
Xol_specials_itm special_itm = wiki.Lang().Specials_mgr().Get_by_alias(special_name);
if (special_itm != null) o = hash.Get_by_bry(special_itm.Special());
}
if (o != null) {
// Xow_special_page special = (Xow_special_page)o;
// page.Revision_data().Modified_on_(Datetime_now.Get());
// special.Special__gen(wiki, page, url, ttl);
}
}
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;
}
}
