private void Parse_title(Gfh_atr title_atr) { // Tfds.Dbg(Bry_.Mid(href_src, href_bgn, href_end), Bry_.Mid(src, capt_bgn, capt_end), // Bry_.Mid(src, title_bgn, title_end)); title_bgn = title_atr.Val_bgn(); title_end = title_atr.Val_end(); if (href_ns_name != null) { // ns_name exists int title_bgn_wo_ns = title_bgn + href_ns_name_len; if (Bry_.Match(src, title_bgn, title_bgn_wo_ns, href_ns_name)) // title matches href_ns; title_bgn = title_bgn_wo_ns; // skip ns; "Help:" else title_missing_ns = true; } if (title_end == -1) title_tid = Title__missing; else { if (Bry_.Match(src, title_bgn, title_end, href_src, href_bgn, href_end) && !title_missing_ns) // NOTE: do not mark title=href if href omitted title; // PAGE:en.b:Wikibooks:WikiProject; DATE:2016-01-20 title_tid = Title__href; else if (Bry_.Match(src, title_bgn, title_end, src, capt_bgn, capt_end)) title_tid = Title__capt; else { title_tid = Title__diff; if (href_ns_name != null) title_bgn = title_atr.Val_bgn(); // since title is different, add back ns_name; EX: "<a // href='/wiki/Help:A_b#c' title='Help:A b'>a</a>"; title should be // "Help:A b", not "A b" } } }
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; }
private void Parse_capt(Gfh_tag_rdr tag_rdr, Gfh_tag anch_head) { this.capt_bgn = anch_head.Src_end(); // capt starts after <a> Gfh_tag anch_tail = tag_rdr.Tag__move_fwd_tail(Gfh_tag_.Id__a); // </a> this.capt_end = anch_tail.Src_bgn(); // get capt between "<a>" and "</a> this.src_end = anch_tail.Src_end(); // skip ns in href / capt if (href_ns_id != Xow_ns_.Tid__main) { // not main; try to remove template name; int capt_bgn_wo_ns = capt_bgn + href_ns_name_len; href_bgn += href_ns_name_len; // skip ns_name for href; EX: "Help:A" -> "A"; "Help" will be saved as // encoded number if (Bry_.Match( capt_src, capt_bgn, capt_bgn_wo_ns, href_ns_name)) { // capt matches ns_name; EX: <a href='/wiki/Help:A'>Help:A</a> -> // "Help:A" matches "Help:" capt_bgn = capt_bgn_wo_ns; // skip ns; "Help:" capt_has_ns = true; } } // get text splits this.text_tid = href_itm.Tid() == Xoh_anch_href_data.Tid__anch ? Xoh_anch_capt_itm.Tid__diff : capt_itm.Parse( rdr, capt_has_ns, href_src, href_bgn, href_end, src, capt_bgn, capt_end); int split_pos = capt_itm.Split_pos(); this.text_0_src = href_src; this.text_0_bgn = href_bgn; this.text_0_end = href_end; this.text_1_src = capt_src; this.text_1_bgn = capt_bgn; this.text_1_end = capt_end; switch (text_tid) { case Xoh_anch_capt_itm.Tid__same: // case Xoh_anch_capt_itm.Tid__href_pipe: case Xoh_anch_capt_itm.Tid__diff: // nothing to do; href / capt already set above break; case Xoh_anch_capt_itm.Tid__more: this.text_1_bgn = split_pos; break; case Xoh_anch_capt_itm.Tid__less: this.text_0_end = split_pos; this.text_1_src = href_src; this.text_1_bgn = split_pos; this.text_1_end = href_end; break; } }
public boolean Init_by_parse( Xoh_hdoc_wkr hdoc_wkr, Xoh_hdoc_ctx hctx, Gfh_tag_rdr tag_rdr, byte[] src, Gfh_tag anch_head, Gfh_tag unused) { Gfh_tag img_tag = anch_head; Bry_err_wkr err_wkr = tag_rdr.Err_wkr(); this.img_wo_anch = anch_head.Name_id() == Gfh_tag_.Id__img; if (img_wo_anch) { Gfh_atr xowa_title = anch_head.Atrs__get_by_or_empty( Xoh_img_xoimg_data.Bry__data_xowa_title); // data-xowa-title='A.png' anch_xo_ttl.Val_(xowa_title.Val()); } else { if (anch_head.Name_id() == Gfh_tag_.Id__div) { // video / audio if (!anch_head.Atrs__cls_eq( gplx.xowa.htmls.core.wkrs.thms.Xoh_thm_data.Atr__id__xowa_media_div)) return false; // handle fake-thumbs created through en.w:Template:Image_label_begin; // PAGE:en.w:Blackburnshire; DATE:2016-01-04 img_is_vid = true; tag_rdr.Tag__move_fwd_head(); // next <div> anch_head = tag_rdr.Tag__move_fwd_head(); // next <div> } this.src_bgn = anch_head.Src_bgn(); // <a if (!anch_href.Parse(err_wkr, hctx, src, anch_head)) return false; // href='/wiki/File:A.png' if (!anch_cls.Parse(err_wkr, src, anch_head)) return false; // class='image' Gfh_atr anch_title = anch_head.Atrs__get_by_or_empty(Gfh_atr_.Bry__title); // title='abc' anch_title_bgn = anch_title.Val_bgn(); anch_title_end = anch_title.Val_end(); Gfh_atr xowa_title = anch_head.Atrs__get_by_or_empty(Bry__atr__xowa_title); // xowa_title='A.png' if (xowa_title.Val_dat_exists()) anch_xo_ttl.Val_(xowa_title.Val()); img_tag = tag_rdr.Tag__move_fwd_head(); } img_tag.Chk_name_or_fail(Gfh_tag_.Id__img); // <img if (img_tag.Atrs__cls_has( gplx.xowa.xtns.pagebanners.Pgbnr_xtn_mgr .Bry__cls__wpb_banner_image)) { // handle pagebanner; EX: <img class="wpb-banner-image"> img_pgbnr.Init_by_parse(img_tag); } img_xoimg.Parse(err_wkr, src, img_tag); // data-xoimg='...' this.img_w = img_tag.Atrs__get_as_int_or(Gfh_atr_.Bry__width, Xof_img_size.Size__neg1); // width='220' this.img_h = img_tag.Atrs__get_as_int_or(Gfh_atr_.Bry__height, Xof_img_size.Size__neg1); // height='110' Gfh_atr img_alt = img_tag.Atrs__get_by_or_empty(Gfh_atr_.Bry__alt); // alt='File:A.png' img_alt_bgn = img_alt.Val_bgn(); img_alt_end = img_alt.Val_end(); img_cls.Init_by_parse(err_wkr, src, img_tag); // class='thumbborder' img_alt__diff_anch_title = !Bry_.Match(src, img_alt_bgn, img_alt_end, src, anch_title_bgn, anch_title_end); if (!img_src.Parse(err_wkr, hctx, hctx.Wiki__domain_bry(), img_tag)) return false; // src='...' if (anch_xo_ttl.Val_is_empty()) { anch_xo_ttl.Val_(img_src.File_ttl_bry()); if (anch_xo_ttl.Val_is_empty()) anch_xo_ttl.Val_(anch_href.Ttl_page_db()); } this.img_imap_idx = Get_imap_idx(tag_rdr.Err_wkr(), img_tag); if (img_wo_anch) { src_end = img_tag.Src_end(); return true; } Gfh_tag anch_tail = tag_rdr.Tag__move_fwd_tail(Gfh_tag_.Id__a); // </a> this.src_end = anch_tail.Src_end(); if (anch_href.Site_exists()) { Xow_domain_itm itm = Xow_domain_itm_.parse(Bry_.Mid(src, anch_href.Site_bgn(), anch_href.Site_end())); anch_rel_is_nofollow = itm.Domain_type_id() == Xow_domain_tid_.Tid__other; } if (img_is_vid) { tag_rdr.Tag__move_fwd_tail(Gfh_tag_.Id__div); anch_head = tag_rdr.Tag__move_fwd_tail(Gfh_tag_.Id__div); src_end = anch_head.Src_end(); } return true; }