@Test
public void Bidi() {
fxt.Init_ttl("A" + String_.new_u8(Bry_.ints_(226, 128, 142)) + "B").Expd_page_txt("AB").Test();
fxt.Init_ttl("A" + String_.new_u8(Bry_.ints_(226, 128, 97)) + "B")
.Expd_page_txt("A" + String_.new_u8(Bry_.ints_(226, 128, 97)) + "B")
.Test();
}
public boolean Chk(Xowe_wiki wiki) {
if (!wiki.App().App_type().Uid_is_gui())
return true; // NOTE: ignore during Server / Console modes; DATE:2015-04-01
if (in_progress_hash.Has(wiki.Domain_bry()))
return true; // NOTE: ignore if currently building; different bldr commands call
// wiki.Init_assert() which may lead to fals checks;
Io_url url = url_(wiki);
if (!Io_mgr.I.ExistsFil(url)) return true;
Xoae_app app = wiki.Appe();
app.Usr_dlg().Log_many("", "", "import.marker: marker found: url=~{0}", url.Raw());
byte[] incompete_msg_bry =
app.Usere()
.Msg_mgr()
.Val_by_key_args(Bry_.new_a7("api-xowa.import.core.incomplete"), wiki.Domain_str());
int rslt = app.Gui_mgr().Kit().Ask_yes_no_cancel("", "", String_.new_u8(incompete_msg_bry));
switch (rslt) {
case Gfui_dlg_msg_.Btn_yes:
Xob_cleanup_cmd.Delete_wiki_sql(wiki);
End(wiki);
return false; // delete wiki
case Gfui_dlg_msg_.Btn_no:
End(wiki);
return true; // delete marker
case Gfui_dlg_msg_.Btn_cancel:
return true; // noop
default:
throw Err_.unhandled(rslt);
}
}
public Xowv_wiki(Xoav_app app, byte[] domain_bry, Io_url wiki_root_dir) {
this.app = app;
this.domain_bry = domain_bry;
this.domain_str = String_.new_u8(domain_bry);
this.domain_itm = Xow_domain_itm_.parse(domain_bry);
this.domain_tid = domain_itm.Domain_type_id();
this.domain_abrv = Xow_abrv_wm_.To_abrv(Xow_domain_itm_.parse(domain_bry));
this.ns_mgr = Xow_ns_mgr_.default_(app.Utl_case_mgr());
this.lang =
app.Lang_mgr()
.Get_by_or_en(
domain_itm
.Lang_actl_key()); // NOTE: must not be null, or causes null ref exception on
// redlinks in drd; DATE:2016-06-28
this.msg_mgr = new Xow_msg_mgr(this, lang);
this.html__hdump_mgr = new Xow_hdump_mgr(this);
this.special_mgr = new Xosp_special_mgr(this);
this.fsys_mgr =
new Xow_fsys_mgr(wiki_root_dir, app.Fsys_mgr().File_dir().GenSubDir(domain_str));
this.fsdb_mgr = new Xof_fsdb_mgr__sql();
this.url__parser = new Xow_url_parser(this);
this.xwiki_mgr = new Xow_xwiki_mgr(this);
this.stats = new Xow_site_stats_mgr(this);
this.lnki_bldr = new Xoh_lnki_bldr(app, href_wtr);
this.ctg_catpage_mgr = new Xoctg_catpage_mgr(this);
}
public byte[] Bld(Xowe_wiki cur_wiki, Xoae_page page, Xow_popup_itm popup_itm, Bry_bfr wrdx_bfr) {
if (output_js_clean) cur_wiki.Html_mgr().Js_cleaner().Clean_bfr(wiki, page.Ttl(), wrdx_bfr, 0);
if (output_tidy)
cur_wiki.Html_mgr().Tidy_mgr().Exec_tidy(wrdx_bfr, Bool_.Y, page.Url_bry_safe());
byte[] hdom_bry = wrdx_bfr.To_bry_and_clear();
String page_url =
wrdx_bfr.Add(page.Wiki().Domain_bry())
.Add(gplx.xowa.htmls.hrefs.Xoh_href_.Bry__wiki)
.Add(
gplx.langs.htmls.encoders.Gfo_url_encoder_.Href.Encode(
page.Ttl()
.Full_db())) // NOTE: was page.Url().Raw(), but that doesn't work for
// Special:Search; PAGE:en.w:Earth and "Quotations";
// DATE:2014-06-29
.To_str_and_clear();
fmtr_popup.Bld_bfr_many(
wrdx_bfr,
hdom_bry,
wiki.Lang().Dir_ltr_bry(),
page_url,
String_.new_u8(page.Ttl().Full_txt_w_ttl_case()),
popup_itm.Popup_id(),
Xow_popup_html_bldr_.Bld_fmtr_wiki(
fmtr_wiki,
wrdx_bfr,
cur_wiki.Domain_bry(),
page.Wiki().Domain_bry()) // NOTE: use cur_wiki, not page_wiki; DATE:2014-06-28
,
gplx.core.ios.Io_size_.To_str(page.Db().Text().Text_bry().length),
page.Db().Page().Modified_on().XtoStr_fmt_yyyy_MM_dd_HH_mm_ss(),
Xow_popup_html_bldr_.Bld_fmtr_viewed(fmtr_viewed, app, wiki, wrdx_bfr, page.Ttl()),
app.Fsys_mgr().Root_dir().To_http_file_bry());
return wrdx_bfr.To_bry_and_clear();
}
Xof_repo_itm Ini_repo_add(Xof_file_mgr file_mgr, byte[] key, String root, String wiki, boolean trg) {
Xof_repo_itm repo = file_mgr.Repo_mgr().Set(String_.new_u8(key), root, wiki).Ext_rules_(Xof_rule_grp.Grp_app_default).Dir_depth_(2);
if (trg) {
byte[][] ary = repo.Mode_names();
ary[0] = Bry_.new_a7("raw");
ary[1] = Bry_.new_a7("fit");
}
return repo;
}
public static Xopg_tag_itm New_js_file(Io_url src_url) {
byte[] src = src_url.To_http_file_bry();
return new Xopg_tag_itm(
Tid__js_file,
Gfh_tag_.Bry__script,
src,
null,
Tag__type_js,
Keyval_.new_("src", String_.new_u8(src)));
}
public Xow_search_tkn[] Scan(byte[] src) {
this.src = src;
this.src_len = src.length;
tkns.Clear();
pos = 0;
txt_bgn = -1;
while (pos < src_len) {
byte cur_b = src[pos];
Object cur_obj = trie.Match_bgn_w_byte(cur_b, src, pos, src_len);
if (cur_obj == null) { // text character
if (txt_bgn == -1) txt_bgn = pos; // 1st character not set; set it
++pos;
} else { // AND, OR, (, ), -, \s, "
int pos_end = trie.Match_pos();
byte cur_tid = ((Byte_obj_val) cur_obj).Val();
if (Cur_join_is_word(cur_tid, pos_end))
continue; // ignore words containing "and", "or"; EX: "random"; "for"
if (txt_bgn != -1) { // pending word; create
Tkns_add_word(Xow_search_tkn.Tid_word, txt_bgn, pos);
txt_bgn = -1;
}
switch (cur_tid) {
case Xow_search_tkn.Tid_space: // discard spaces
pos = Bry_finder.Find_fwd_while(src, pos, src_len, Byte_ascii.Space);
break;
case Xow_search_tkn.Tid_quote: // find end quote and add as word
int quote_bgn = pos + 1;
int quote_end = Bry_finder.Find_fwd(src, Byte_ascii.Quote, quote_bgn, src_len);
if (quote_end == Bry_.NotFound)
throw Err_.new_fmt_("could not find end quote: {0}", String_.new_u8(src));
Tkns_add_word(Xow_search_tkn.Tid_word_quoted, quote_bgn, quote_end);
pos = quote_end + 1; // +1 to place after quote
break;
case Xow_search_tkn.Tid_not:
Tkns_add_word(Xow_search_tkn.Tid_not, pos, pos_end);
pos = pos_end;
break;
case Xow_search_tkn.Tid_paren_bgn:
case Xow_search_tkn.Tid_paren_end:
case Xow_search_tkn.Tid_and:
case Xow_search_tkn.Tid_or:
tkns.Add(new_tkn(cur_tid, pos, pos_end));
pos = pos_end;
break;
default:
throw Err_.unhandled(cur_tid);
}
}
}
if (txt_bgn != -1) { // pending word; create
Tkns_add_word(Xow_search_tkn.Tid_word, txt_bgn, pos);
txt_bgn = -1;
}
return (Xow_search_tkn[]) tkns.To_ary_and_clear(Xow_search_tkn.class);
}
private static boolean Navigate(
Gfo_usr_dlg usr_dlg,
Xoae_app app,
Wdata_wiki_mgr wdata_mgr,
Xoae_page page,
byte[] site_bry,
byte[] page_bry) {
page_bry =
Xoa_app_.Utl__encoder_mgr()
.Http_url()
.Decode(page_bry); // NOTE: space is converted to + on postback to url; decode
byte[] wiki_domain = Xow_wiki_alias.Parse_wmf_key(site_bry);
if (wiki_domain == null) {
usr_dlg.Warn_many("", "", "site_bry parse failed; site_bry:~{0}", String_.new_u8(site_bry));
return false;
}
Xowe_wiki wiki = app.Wiki_mgr().Get_by_key_or_make(wiki_domain);
if (wiki == null) {
usr_dlg.Warn_many(
"", "", "wiki_domain does not exist; wiki_domain:~{0}", String_.new_u8(wiki_domain));
return false;
}
Xoa_ttl wdata_ttl = Xoa_ttl.parse_(wiki, page_bry);
if (wdata_ttl == null) {
usr_dlg.Warn_many("", "", "ttl is invalid; ttl:~{0}", String_.new_u8(page_bry));
return false;
}
Wdata_doc doc = wdata_mgr.Pages_get(wiki, wdata_ttl);
if (doc == null) {
usr_dlg.Warn_many(
"", "", "ttl cannot be found in wikidata; ttl:~{0}", String_.new_u8(wdata_ttl.Raw()));
return false;
}
byte[] qid_bry = doc.Qid();
Xoae_page qid_page = wdata_mgr.Wdata_wiki().Data_mgr().Redirect(page, qid_bry);
if (qid_page.Missing()) {
usr_dlg.Warn_many(
"", "", "qid cannot be found in wikidata; qid:~{0}", String_.new_u8(qid_bry));
return false;
}
return true;
}
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 static Xopg_tag_itm New_css_file(Io_url href_url) {
byte[] href = href_url.To_http_file_bry();
return new Xopg_tag_itm(
Tid__css_file,
Gfh_tag_.Bry__link,
href,
null,
Tag__type_css,
Tag__rel_stylesheet,
Keyval_.new_("href", String_.new_u8(href)));
}
public byte[] Cartouche_img(
Xoh_wtr_ctx hctx,
boolean bgn,
byte[] glyph) { // render open / close cartouche; note that MW has two branches, but they are
// both the same
int height = (int) ((Hiero_html_mgr.Max_height * Hiero_html_mgr.scale) / 100);
Hiero_phoneme_itm phoneme_itm = phoneme_mgr.Get_by_key(glyph);
if (phoneme_itm == null) throw Err_.new_fmt_("missing phoneme: {0}", String_.new_u8(glyph));
byte[] code = phoneme_itm.Gardiner_code();
byte[] title = bgn ? Html_entity_.Lt_bry : Html_entity_.Gt_bry;
return cartouche_img_fmtr.Bld_bry_many(temp_bfr, hiero_img_dir, code, height, title);
}
public Object Invk(GfsCtx ctx, int ikey, String k, GfoMsg m) {
if (ctx.Match(k, Invk_clear)) this.Clear();
else if (ctx.Match(k, Invk_separators)) return separators_mgr;
else if (ctx.Match(k, Invk_digits)) {
digits_translate = true;
return digits_mgr;
} // NOTE: only langes with a digit_transform_table will call digits; DATE:2014-05-28
else if (ctx.Match(k, Invk_digit_grouping_pattern))
return String_.new_u8(num_grp_fmtr.Digit_grouping_pattern());
else if (ctx.Match(k, Invk_digit_grouping_pattern_))
num_grp_fmtr.Digit_grouping_pattern_(m.ReadBry("v"));
else return Gfo_invk_.Rv_unhandled;
return this;
}
public void Save_as() {
if (this.Active_tab_is_null()) return;
Xog_tab_itm tab = win.Tab_mgr().Active_tab();
String file_name =
Xoa_app_.Utl__encoder_mgr()
.Fsys_safe()
.Encode_str(String_.new_u8(tab.Page().Ttl().Full_url()))
+ ".html";
String file_url =
app.Gui_mgr()
.Kit()
.New_dlg_file(Gfui_kit_.File_dlg_type_save, "Select file to save to:")
.Init_file_(file_name)
.Ask();
if (String_.Len_eq_0(file_url)) return;
Io_mgr.I.SaveFilStr(file_url, tab.Html_box().Text());
app.Usr_dlg().Prog_many("", "", "saved page: file=~{0}", file_url);
}
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;
}
}
String Replace_invalid_chars_str(String raw_str) {
return String_.new_u8(Replace_invalid_chars(Bry_.new_u8(raw_str)));
}
public static byte[] Write_wdata_links(List_adp slink_list, Xowe_wiki wiki, Xoa_ttl ttl, Wdata_external_lang_links_data external_links_mgr) {
try {
switch (wiki.Domain_tid()) {
case Xow_domain_type_.Tid_home: // home will never be in wikidata
case Xow_domain_type_.Tid_wikidata: // wikidata will never be in wikidata
return Qid_null;
}
Wdata_wiki_mgr wdata_mgr = wiki.Appe().Wiki_mgr().Wdata_mgr();
Wdata_doc doc = wdata_mgr.Pages_get(wiki, ttl); if (doc == null) return Qid_null; // no links
boolean external_links_mgr_enabled = external_links_mgr.Enabled();
Ordered_hash links = doc.Slink_list();
Bry_bfr tmp_bfr = wiki.Appe().Utl__bfr_mkr().Get_k004();
Xow_wiki_abrv wiki_abrv = new Xow_wiki_abrv();
int len = links.Count();
for (int i = 0; i < len; i++) {
Wdata_sitelink_itm slink = (Wdata_sitelink_itm)links.Get_at(i);
byte[] xwiki_key = slink.Site();
Xow_wiki_abrv_.parse_(wiki_abrv, xwiki_key, 0, xwiki_key.length);
if (wiki_abrv.Domain_tid() == Xow_wiki_abrv_.Tid_null) {
wiki.Appe().Usr_dlg().Warn_many("", "", "unknown wiki in wikidata: ttl=~{0} wiki=~{1}", ttl.Page_db_as_str(), String_.new_u8(xwiki_key));
continue;
}
if (wiki_abrv.Domain_tid() != wiki.Domain_tid()) continue; // ignore wikis in a different domain; EX: looking at enwiki:Earth, and wikidata has dewikiquote; ignore dewikiquote; DATE:2014-06-21
byte[] lang_key = wiki_abrv.Lang_itm().Key();
if (external_links_mgr_enabled && external_links_mgr.Langs_hide(lang_key, 0, lang_key.length)) continue;
tmp_bfr.Add(lang_key);
tmp_bfr.Add_byte(Byte_ascii.Colon);
tmp_bfr.Add(slink.Name());
Xoa_ttl slink_ttl = Xoa_ttl.parse_(wiki, tmp_bfr.Xto_bry_and_clear());
if (slink_ttl == null) continue; // invalid ttl
Xow_xwiki_itm xwiki_itm = slink_ttl.Wik_itm();
if ( xwiki_itm == null // not a known xwiki; EX: [[zzz:abc]]
|| Bry_.Eq(xwiki_itm.Domain_bry(), wiki.Domain_bry()) // skip if same as self; i.e.: do not include links to enwiki if already in enwiki
) continue;
slink.Page_ttl_(slink_ttl);
slink_list.Add(slink);
}
tmp_bfr.Mkr_rls();
if (external_links_mgr_enabled && external_links_mgr.Sort())
slink_list.Sort_by(Xoa_ttl_sorter._);
return doc.Qid();
} catch (Exception e) {Err_.Noop(e); return Qid_null;}
}
String Make_key(int ns_id, byte[] ttl) {
return Int_.To_str(ns_id) + "|" + String_.new_u8(ttl);
}
private static Keyval new_itm(Xol_lang_stub itm) {
String key_str = String_.new_u8(itm.Key());
String name_str = String_.new_u8(itm.Canonical_name());
return Keyval_.new_(key_str, name_str + " [" + key_str + "]");
}