Xtn.Math: Swap out math tags with UNIQ

400_xowa/src/gplx/xowa/Xoa_app_.java

@@ -32,7 +32,7 @@ public class Xoa_app_ {
 		}
 	}
 	public static final String		Name			= "xowa";
-	public static final String		Version			= "3.11.2.8";
+	public static final String		Version			= "3.11.2.9";
 	public static String	Build_date		= "2012-12-30 00:00:00";
 	public static String	Op_sys_str;
 	public static String	User_agent		= "";

400_xowa/src/gplx/xowa/Xop_fxt.java

@@ -417,6 +417,18 @@ public class Xop_fxt {
 		Xol_kwd_grp kwd_grp = kwd_mgr.Get_or_new(kwd_id);
 		kwd_grp.Srl_load(case_match, Bry_.Ary(kwds));
 	}
+	public void Init_lang_vnts(String... vnts) {
+		wiki.Lang().Vnt_mgr().Enabled_(true);
+		gplx.xowa.langs.vnts.Xol_vnt_regy vnt_regy = wiki.Lang().Vnt_mgr().Regy();
+		for (int i = 0; i < vnts.length; i++) {
+			byte[] vnt = Bry_.new_u8(vnts[i]);
+			vnt_regy.Add(vnt, vnt);
+			if (i == 0) {
+				wiki.Lang().Vnt_mgr().Cur_itm_(vnt);
+			}
+		}
+		wiki.Lang().Vnt_mgr().Init_end();
+	}
 	public void Init_xtn_pages() {
 		Io_mgr.Instance.InitEngine_mem();
 		wiki.Xtn_mgr().Xtn_proofread().Enabled_y_();
@@ -440,4 +452,17 @@ public class Xop_fxt {
             // Tfds.Dbg(tmp_bfr.To_str());
 		return tmp_bfr.To_str_and_clear();
 	}
+	public void Test__parse_to_html_mgr(String src_str, String expd) {
+		byte[] src_bry = Bry_.new_u8(src_str);
+		Xop_root_tkn root = Exec_parse_page_all_as_root(src_bry);
+		Xoae_page page = this.Page();
+		page.Root_(root);
+		byte[] actl = wiki.Html_mgr().Page_wtr_mgr().Gen(page, gplx.xowa.wikis.pages.Xopg_page_.Tid_read);
+		Tfds.Eq_str_lines(expd, String_.new_u8(actl));
+	}
+	public static Xop_fxt New_app_html() {
+		Xop_fxt fxt = new Xop_fxt();
+		fxt.Wiki().Html_mgr().Page_wtr_mgr().Page_read_fmtr().Fmt_("~{page_data}");
+		return fxt;
+	}
 }

400_xowa/src/gplx/xowa/htmls/Xoh_page_wtr_mgr_tst.java

@@ -30,4 +30,18 @@ public class Xoh_page_wtr_mgr_tst {
 		page_wtr_mgr.Gen(wiki.Parser_mgr().Ctx().Page(), Xopg_page_.Tid_read);
 		Tfds.Eq(String_.new_a7(portal_mgr.Div_logo_bry()), "/site/en.wikipedia.org/wiki/");
 	}
+	@Test   public void Skip__math__basic() {
+		Xop_fxt fxt = Xop_fxt.New_app_html();
+		fxt.Init_lang_vnts("zh-hans", "zh-hant");
+
+		fxt.Test__parse_to_html_mgr(String_.Concat_lines_nl_skip_last
+		( "<math>x_{1}-1</math>"
+		, "<math>x-{1+2}-1</math>"
+		, "-{zh-hans:A;zh-hant:B;}-"
+		), String_.Concat_lines_nl_skip_last
+		( "<span id='xowa_math_txt_0'>x_{1}-1</span>"		// not converted
+		, "<span id='xowa_math_txt_0'>x-{1+2}-1</span>"		// not converted
+		, "A" // converted
+		));
+	}
 }

400_xowa/src/gplx/xowa/htmls/Xoh_page_wtr_wkr.java

@@ -207,6 +207,9 @@ public class Xoh_page_wtr_wkr {
 		// translate if variants are enabled
 		Xol_vnt_mgr vnt_mgr = wiki.Lang().Vnt_mgr();
 		if (vnt_mgr.Enabled()) bfr.Add(vnt_mgr.Convert_lang().Parse_page(vnt_mgr.Cur_itm(), page.Db().Page().Id(), bfr.To_bry_and_clear()));
+
+		// handle uniqs
+		wiki.Parser_mgr().Uniq_mgr().Parse(bfr);
 	}
 	private void Write_body_pre(Bry_bfr bfr, Xoae_app app, Xowe_wiki wiki, Xoh_wtr_ctx hctx, byte[] data_raw, Bry_bfr tmp_bfr) {
 		Xoh_html_wtr_escaper.Escape(app.Parser_amp_mgr(), tmp_bfr, data_raw, 0, data_raw.length, false, false);

400_xowa/src/gplx/xowa/htmls/core/htmls/Xoh_html_wtr_tst.java

@@ -320,13 +320,6 @@ public class Xoh_html_wtr_tst {
 		fxt.Test_parse_page_wiki_str("<div id='a*'></div>", "<div id='a.2A'></div>");
 		fxt.Test_parse_page_wiki_str("<div id='a b'></div>", "<div id='a_b'></div>");
 	}
-	@Test  public void Math() {
-		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(false);
-		fxt.Test_parse_page_all_str("<math>x + y</math>", "<img id='xowa_math_img_0' src='' width='' height=''/><span id='xowa_math_txt_0'>x + y</span>");	// latex has img
-		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(true);
-		fxt.Test_parse_page_all_str("<math>x + y</math>", "<span id='xowa_math_txt_0'>x + y</span>");	// mathjax has no img
-		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(false);
-	}
 	@Test  public void Timeline() {// PURPOSE: embed timeline contents in pre; DATE:2014-05-22
 		fxt.Test_parse_page_wiki_str("<timeline>a</timeline>", "<pre class='xowa-timeline'>a</pre>");
 	}

400_xowa/src/gplx/xowa/htmls/modules/popups/Xow_popup_parser.java

@@ -128,6 +128,7 @@ public class Xow_popup_parser {
 		parser.Parse_to_src_end(wtxt_root, wtxt_ctx, tkn_mkr, wrdx_bry, wtxt_trie, Xop_parser_.Doc_bgn_bos, wrdx_bry.length);
 		wtxt_ctx.Parser__page_term(wtxt_root, wrdx_bry, wrdx_bry.length);
 		wiki.Html_mgr().Html_wtr().Write_doc(wrdx_bfr, wtxt_ctx, hctx, wrdx_bry, wtxt_root);
+		wiki.Parser_mgr().Uniq_mgr().Parse(wrdx_bfr);
 	}
 	private void Adjust_wrdx_end(Xow_popup_itm popup_itm, Bry_bfr wrdx_bfr) {
 		popup_itm.Words_found_(data.Words_found());

400_xowa/src/gplx/xowa/parsers/Xow_parser_mgr.java

@@ -20,6 +20,7 @@ import gplx.core.primitives.*; import gplx.core.brys.fmtrs.*;
 import gplx.xowa.wikis.*; import gplx.core.envs.*;
 import gplx.xowa.files.*;
 import gplx.xowa.xtns.scribunto.*; import gplx.xowa.xtns.wbases.hwtrs.*; import gplx.xowa.xtns.pfuncs.ifs.*; import gplx.xowa.xtns.pfuncs.times.*; import gplx.xowa.xtns.pfuncs.ttls.*;
+import gplx.xowa.parsers.uniqs.*;
 public class Xow_parser_mgr {
 	private final    Xowe_wiki wiki; private final    Xop_tkn_mkr tkn_mkr;
 	public Xow_parser_mgr(Xowe_wiki wiki) {
@@ -37,6 +38,7 @@ public class Xow_parser_mgr {
 	public Pft_func_formatdate_bldr Date_fmt_bldr()		{return date_fmt_bldr;} private final    Pft_func_formatdate_bldr date_fmt_bldr = new Pft_func_formatdate_bldr();
 	public Gfo_number_parser		Pp_num_parser()		{return pp_num_parser;} private final    Gfo_number_parser pp_num_parser = new Gfo_number_parser().Ignore_space_at_end_y_();
 	public int[]					Rel2abs_ary()		{return rel2abs_ary;} private final    int[] rel2abs_ary = new int[Pfunc_rel2abs.Ttl_max];
+	public Xop_uniq_mgr				Uniq_mgr()			{return uniq_mgr;} private final    Xop_uniq_mgr uniq_mgr = new Xop_uniq_mgr();
 	public boolean						Lst__recursing()	{return lst_recursing;} private boolean lst_recursing; public void	Lst__recursing_(boolean v) {lst_recursing = v;}
 	public Bry_bfr					Wbase__time__bfr()  {return wbase__time__bfr;} private final    Bry_bfr wbase__time__bfr = Bry_bfr_.New();
 	public Bry_fmtr					Wbase__time__fmtr() {return wbase__time__fmtr;} private final    Bry_fmtr wbase__time__fmtr = Bry_fmtr.new_();
@@ -76,6 +78,7 @@ public class Xow_parser_mgr {
 		if (!Env_.Mode_testing()) wiki.Init_assert();
 		tmpl_stack_ary = Bry_.Ary_empty;
 		tmpl_stack_ary_len = tmpl_stack_ary_max = 0;
+		uniq_mgr.Clear();
 
 		scrib.When_page_changed(page);	// notify scribunto about page changed
 		ctx.Page_(page);

400_xowa/src/gplx/xowa/parsers/uniqs/Xop_uniq_mgr.java

@@ -31,6 +31,11 @@ public class Xop_uniq_mgr {	// REF.MW:/parser/StripState.php
 		return key;
 	}
 	public byte[] Get(byte[] key) {return (byte[])general_trie.Match_exact(key, 0, key.length);}
+	public void Parse(Bry_bfr bfr) {
+		if (general_trie.Count() == 0) return;
+		byte[] rv = Parse(key_bfr, general_trie, bfr.To_bry_and_clear());
+		bfr.Add(rv);
+	}
 	public byte[] Parse(byte[] src) {return Parse(key_bfr, general_trie, src);}
 	private byte[] Parse(Bry_bfr bfr, Btrie_slim_mgr trie, byte[] src) {
 		int src_len = src.length;

400_xowa/src/gplx/xowa/parsers/vnts/Vnt_convert_lang.java

@@ -184,6 +184,5 @@ public class Vnt_convert_lang {
 	private static final    Btrie_fast_mgr trie = Btrie_fast_mgr.cs()
 	.Add_bry_byte(Bry__curly_bgn, Tid__curly_bgn)
 	.Add_bry_byte(Bry__curly_end, Tid__curly_end);
-	public static final    byte[] Bry__armor_bgn = Bry_.new_a7("-{R|"), Bry__armor_end = Bry_.new_a7("}-");
 	private static final int max_depth = 32; 
 }

400_xowa/src/gplx/xowa/parsers/xndes/Xop_xnde_wkr__basic_tst.java

@@ -18,7 +18,7 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 package gplx.xowa.parsers.xndes; import gplx.*; import gplx.xowa.*; import gplx.xowa.parsers.*;
 import org.junit.*; import gplx.xowa.wikis.nss.*;
 public class Xop_xnde_wkr__basic_tst {
-	private final Xop_fxt fxt = new Xop_fxt();
+	private final    Xop_fxt fxt = new Xop_fxt();
 	@After public void term() {fxt.Init_para_n_();}
 	@Test  public void Escape_lt() {	// PURPOSE: some templates have unknown tags; PAGE:en.w:PHP
 		fxt.Init_para_y_();
@@ -103,9 +103,9 @@ public class Xop_xnde_wkr__basic_tst {
 	}
 	@Test   public void CaseSensitivity_xtn_3() {// PURPOSE: xtn xnde must do case-insensitive match DATE:2013-12-02
 		fxt.Test_parse_page_all_str
-			(	"<matH>a</math> b <math>c</matH>"	// <matH> should match </math> not </matH>
+		( "<translatE>a</translate> b <translate>c</translatE>"	// <translatE> should match </translate> not </translatE>
-			,	"<span id='xowa_math_txt_0'>a</span> b <span id='xowa_math_txt_0'>c</span>"
+		, "a b c"
-			);
+		);
 	}
 	@Test  public void Whitelist() {
 		fxt.Test_parse_page_all_str("<span onload='alert()'></span>", "<span></span>");
@@ -122,11 +122,6 @@ public class Xop_xnde_wkr__basic_tst {
 	@Test   public void Script_in_syntaxhighlight() {
 		fxt.Test_parse_page_all_str("<syntaxhighlight><script>alert('fail');</script></syntaxhighlight>", "<div class=\"mw-highlight\"><pre style=\"overflow:auto\">&lt;script&gt;alert('fail');&lt;/script&gt;</pre></div>");
 	}
-	@Test   public void Script_in_math() {
-		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(false);
-		fxt.Test_parse_page_all_str("<math><script>alert('fail');</script></math>", "<img id='xowa_math_img_0' src='' width='' height=''/><span id='xowa_math_txt_0'>&lt;script>alert('fail');</script></span>");
-		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(true);
-	}
 	@Test  public void Html5_time() {// PURPOSE: HTML5; should output self (i.e.: must be whitelisted)
 		fxt.Test_parse_page_wiki_str("<time class=\"dtstart\" datetime=\"2010-10-10\">10 October 2010</time>", "<time class=\"dtstart\" datetime=\"2010-10-10\">10 October 2010</time>");
 	}

400_xowa/src/gplx/xowa/xtns/math/Xof_math_html_wtr.java

@@ -20,6 +20,7 @@ import gplx.core.brys.fmtrs.*;
 import gplx.xowa.htmls.*; import gplx.langs.htmls.entitys.*; import gplx.xowa.htmls.core.htmls.*;
 import gplx.xowa.parsers.*; import gplx.xowa.parsers.xndes.*; import gplx.xowa.parsers.vnts.*;
 public class Xof_math_html_wtr {
+	public static final    byte[] Bry__id_prefix = Bry_.new_a7("xowa_math_txt_");
 	private final    Bry_fmtr math_fmtr_latex		= Bry_fmtr.new_("<img id='xowa_math_img_~{math_idx}' src='' width='' height=''/><span id='xowa_math_txt_~{math_idx}'>~{math_text}</span>", "math_idx", "math_text");
 	private final    Bry_fmtr math_fmtr_mathjax		= Bry_fmtr.new_("<span id='xowa_math_txt_~{math_idx}'>~{math_text}</span>", "math_idx", "math_text");
 	public void Write(Xoh_html_wtr wtr, Xop_ctx ctx, Xoh_wtr_ctx opts, Bry_bfr bfr, byte[] src, Xop_xnde_tkn xnde) {
@@ -45,12 +46,16 @@ public class Xof_math_html_wtr {
 	private void Write_for_mathjax(Bry_bfr bfr, Xoae_page page, boolean enabled, boolean renderer_is_latex, byte[] math_bry, Bry_bfr tmp_bfr, Xof_math_itm math_itm) {
 		int id = page.File_math().Count();
 		Xof_math_itm new_math_itm = math_itm.Clone().Id_(id);
+
+//			boolean armor_math = page.Lang().Vnt_mgr().Enabled() && !renderer_is_latex;	// REF.MW:LangConverter.php|armourMath
+//			if (armor_math) bfr.Add(Vnt_convert_lang.Bry__armor_bgn);
+//			bfr.Add_bfr_and_clear(tmp_bfr);
+//			if (armor_math) bfr.Add(Vnt_convert_lang.Bry__armor_end);
+		byte[] unique_bry = page.Wikie().Parser_mgr().Uniq_mgr().Add(math_bry);
 		Bry_fmtr math_fmtr = renderer_is_latex ? math_fmtr_latex : math_fmtr_mathjax;
-		boolean armor_math = page.Lang().Vnt_mgr().Enabled() && !renderer_is_latex;	// REF.MW:LangConverter.php|armourMath
+		math_fmtr.Bld_bfr_many(tmp_bfr, id, unique_bry);
-		if (armor_math) bfr.Add(Vnt_convert_lang.Bry__armor_bgn);
-		math_fmtr.Bld_bfr_many(tmp_bfr, id, math_bry);
 		bfr.Add_bfr_and_clear(tmp_bfr);
-		if (armor_math) bfr.Add(Vnt_convert_lang.Bry__armor_end);
+
 		if (enabled && renderer_is_latex)	// NOTE: only generate images if math is enabled; otherwise "downloading" prints at bottom of screen, but no action (also a lot of file IO)
 			page.File_math().Add(new_math_itm);
 	}

400_xowa/src/gplx/xowa/xtns/math/Xof_math_mgr_html_tst.java

@@ -18,24 +18,33 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 package gplx.xowa.xtns.math; import gplx.*; import gplx.xowa.*; import gplx.xowa.xtns.*;
 import org.junit.*;
 public class Xof_math_mgr_html_tst {
-	@Before public void init() {} private final Xop_fxt fxt = new Xop_fxt();
+	@Before public void init() {} private final    Xop_fxt fxt = Xop_fxt.New_app_html();
-	@Test  public void Escape_lt_gt_mathjax() {	// PURPOSE: escape <>&"; EX:de.w:Vergleich_(Zahlen); DATE:2014-05-10; PAGE:s.w:Matrix_(mathematics) DATE:2014-07-19
+	@Test  public void Basic__latex() {
-		fxt.Test_html_full_str("<math>a<>b</math>", "<span id='xowa_math_txt_0'>a&lt;&gt;b</span>");
-	}
-	@Test  public void Escape_lt_gt_latex() {
 		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(false);
-		fxt.Test_html_full_str("<math>a<>b</math>", "<img id='xowa_math_img_0' src='' width='' height=''/><span id='xowa_math_txt_0'>a<>b</span>");
+		fxt.Test__parse_to_html_mgr("<math>x + y</math>", "<img id='xowa_math_img_0' src='' width='' height=''/><span id='xowa_math_txt_0'>x + y</span>");	// latex has img
+	}
+	@Test  public void Basic__mathjax() {
 		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(true);
+		fxt.Test__parse_to_html_mgr("<math>x + y</math>", "<span id='xowa_math_txt_0'>x + y</span>");	// mathjax has no img
+	}
+	@Test  public void Escape__mathjax() {	// PURPOSE: escape <>&"; EX:de.w:Vergleich_(Zahlen); DATE:2014-05-10; PAGE:s.w:Matrix_(mathematics) DATE:2014-07-19
+		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(true);
+		fxt.Test__parse_to_html_mgr("<math>a<>b</math>", "<span id='xowa_math_txt_0'>a&lt;&gt;b</span>");
+	}
+	@Test  public void Escape__latex() {
+		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(false);
+		fxt.Test__parse_to_html_mgr("<math>a<>b</math>", "<img id='xowa_math_img_0' src='' width='' height=''/><span id='xowa_math_txt_0'>a<>b</span>");
 	}
 	@Test  public void Amp() { // PURPOSE: assert that amp is not escaped; DATE:2014-07-20
-		fxt.Test_html_full_str("<math>a&b</math>", "<span id='xowa_math_txt_0'>a&b</span>");
+		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(true);
+		fxt.Test__parse_to_html_mgr("<math>a&b</math>", "<span id='xowa_math_txt_0'>a&b</span>");
 	}
 	@Test  public void Quote() { // PURPOSE: assert that quote is not escaped; DATE:2014-07-20
-		fxt.Test_html_full_str("<math>a\"b</math>", "<span id='xowa_math_txt_0'>a\"b</span>");
+		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(true);
+		fxt.Test__parse_to_html_mgr("<math>a\"b</math>", "<span id='xowa_math_txt_0'>a\"b</span>");
 	}
-	@Test  public void Vnt() { // PURPOSE: if vnt, armor math; PAGE:sr.w:Земља; DATE:2015-10-11
+	@Test   public void Script() {
-		fxt.Wiki().Lang().Vnt_mgr().Enabled_(Bool_.Y);
+		fxt.App().File_mgr().Math_mgr().Renderer_is_mathjax_(false);
-		fxt.Test_html_full_str("<math>abc</math>", "-{R|<span id='xowa_math_txt_0'>abc</span>}-");	// NOTE: this also armors the id which is good for the mathjax.js
+		fxt.Test__parse_to_html_mgr("<math><script>alert('fail');</script></math>", "<img id='xowa_math_img_0' src='' width='' height=''/><span id='xowa_math_txt_0'>&lt;script>alert('fail');</script></span>");
-		fxt.Wiki().Lang().Vnt_mgr().Enabled_(Bool_.N);
 	}
 }