Parser: Add namespace to Template title [#784]

400_xowa/src/gplx/xowa/langs/cases/Xol_case_cvt.java

@@ -0,0 +1,113 @@
+/*
+XOWA: the XOWA Offline Wiki Application
+Copyright (C) 2020 https://github.com/desb42
+
+XOWA is licensed under the terms of the General Public License (GPL) Version 3,
+or alternatively under the terms of the Apache License Version 2.0.
+
+You may use XOWA according to either of these licenses as is most appropriate
+for your project on a case-by-case basis.
+
+The terms of each license can be found in the source code repository:
+
+GPLv3 License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-GPLv3.txt
+Apache License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-APACHE2.txt
+*/
+package gplx.xowa.langs.cases;
+
+import gplx.Bool_;
+import gplx.Bry_;
+import gplx.Bry_bfr;
+import gplx.Bry_bfr_;
+
+public class Xol_case_cvt {
+	public static byte[] Upper_1st(byte[] src, int pos, int src_len, boolean reuse) {return Up_low_1st(src, pos, src_len, Bool_.Y, reuse);}
+	public static byte[] Upper_1st(byte[] src, int pos, int src_len) {return Up_low_1st(src, pos, src_len, Bool_.Y);}
+	public static byte[] Lower_1st(byte[] src, int pos, int src_len) {return Up_low_1st(src, pos, src_len, Bool_.N);}
+	public static byte[] Up_low_1st(byte[] src, int pos, int src_len, boolean upper) {
+		return Up_low_1st(src, pos, src_len, upper, Bool_.Y);
+	}
+	public static byte[] Up_low_1st(byte[] src, int pos, int src_len, boolean upper, boolean reuse) {
+		if (src_len == 0) return Bry_.Empty;
+		byte b = src[pos];
+		int b_len = gplx.core.intls.Utf8_.Len_of_char_by_1st_byte(b);
+		if (b_len > src_len) return Bry_.Empty; // bad unicode
+		if (reuse) {
+			return Upper_Lower_1st(src, pos, src_len, b_len, upper);
+		}
+		else {
+			int copyLen = src_len - pos;
+			byte[] copy = new byte[copyLen];
+			for (int i = 0; i < copyLen; i++) {
+				copy[i] = src[pos + i];
+			}
+			return Upper_Lower_1st(copy, 0, copyLen, b_len, upper);
+		}
+	}
+	public static byte[] Upper_1st(byte[] src, int pos, int src_len, int b_len) {return Upper_Lower_1st(src, pos, src_len, b_len, Bool_.Y);}
+	public static byte[] Lower_1st(byte[] src, int pos, int src_len, int b_len) {return Upper_Lower_1st(src, pos, src_len, b_len, Bool_.N);}
+	private static byte[] Upper_Lower_1st(byte[] src, int pos, int src_len, int b_len, boolean upper) {
+		byte[] ucase;
+		if (upper)
+			ucase = Xol_case_cvt_.Upper(src, pos, b_len);
+		else
+			ucase = Xol_case_cvt_.Lower(src, pos, b_len);
+		if (ucase == Xol_case_cvt_.byte_NOCHANGE)
+			return src;
+		if (ucase.length == b_len) {
+			for (int i = 0; i < b_len; i++)
+				src[pos+i] = ucase[i];
+			return src;
+		} else {
+			// need to rebuild the byte string
+			Bry_bfr tmp_bfr = Bry_bfr_.New();
+			tmp_bfr.Add_mid(src, 0, pos);
+			tmp_bfr.Add(ucase);
+			tmp_bfr.Add_mid(src, pos + b_len, src_len);
+			return tmp_bfr.To_bry_and_clear();
+		}
+	}
+
+	public static byte[] Uppercase(byte[] src, int src_len) {return Case_cvt(src, src_len, Bool_.Y);}
+	public static byte[] Lowercase(byte[] src, int src_len) {return Case_cvt(src, src_len, Bool_.N);}
+	public static byte[] Case_cvt(byte[] src, int src_len, boolean upper) {
+		Bry_bfr tmp_bfr = null;
+		int pos = 0;
+		while (pos < src_len) {
+			byte b = src[pos];
+			int b_len = gplx.core.intls.Utf8_.Len_of_char_by_1st_byte(b);
+			byte[] ucase;
+			if (upper)
+				ucase = Xol_case_cvt_.Upper(src, pos, b_len);
+			else
+				ucase = Xol_case_cvt_.Lower(src, pos, b_len);
+			if (tmp_bfr == null) {
+				if (ucase != Xol_case_cvt_.byte_NOCHANGE) {
+					if (ucase.length == b_len) {
+						for (int i = 0; i < b_len; i++)
+							src[pos+i] = ucase[i];
+					} else {
+						// need to rebuild the byte string
+						tmp_bfr = Bry_bfr_.New();
+						tmp_bfr.Add_mid(src, 0, pos);
+						tmp_bfr.Add(ucase);
+					}
+				}
+			} else {
+				if (ucase != Xol_case_cvt_.byte_NOCHANGE) {
+					tmp_bfr.Add(ucase);
+				} else {
+					if (b_len == 1)
+						tmp_bfr.Add_byte(b);
+					else
+						tmp_bfr.Add_mid(src, pos, pos + b_len);
+				}
+			}
+			pos += b_len;
+		}
+		if (tmp_bfr == null)
+			return src;
+		else
+			return tmp_bfr.To_bry_and_clear();
+	}
+}

400_xowa/src/gplx/xowa/parsers/tmpls/Xot_defn_tmpl_.java

@@ -1,6 +1,6 @@
 /*
 XOWA: the XOWA Offline Wiki Application
-Copyright (C) 2012-2017 gnosygnu@gmail.com
+Copyright (C) 2012-2020 gnosygnu@gmail.com
 
 XOWA is licensed under the terms of the General Public License (GPL) Version 3,
 or alternatively under the terms of the Apache License Version 2.0.
@@ -13,59 +13,83 @@ The terms of each license can be found in the source code repository:
 GPLv3 License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-GPLv3.txt
 Apache License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-APACHE2.txt
 */
-package gplx.xowa.parsers.tmpls; import gplx.*; import gplx.xowa.*; import gplx.xowa.parsers.*;
+package gplx.xowa.parsers.tmpls;
-import gplx.xowa.wikis.nss.*;
+
-public class Xot_defn_tmpl_ {
+import gplx.Bool_;
-	public static Xot_invk CopyNew(Xop_ctx ctx, Xot_defn_tmpl orig_defn, Xot_invk orig, Xot_invk caller, byte[] src, int frame_ns, byte[] frame_ttl) {	// SEE:NOTE_1
+import gplx.Bry_;
-		Xop_tkn_mkr tkn_mkr = ctx.Tkn_mkr();
+import gplx.Bry_bfr;
-		byte[] orig_src = orig_defn.Data_raw();
+import gplx.Bry_bfr_;
-		Xowe_wiki wiki = ctx.Wiki();
+import gplx.xowa.Xoa_ttl;
-		Xot_invk_temp rv = Xot_invk_temp.New(orig.Defn_tid(), ctx.Page().Ttl().Page_txt(), orig.Name_tkn(), orig_src, caller.Src_bgn(), caller.Src_end());
+import gplx.xowa.Xowe_wiki;
-		frame_ttl = wiki.Lang().Case_mgr().Case_reuse_1st_upper(frame_ttl);	// NOTE: always uppercase 1st; EX:{{navbox -> "Template:Navbox"; PAGE:en.w:Achilles DATE:2014-06-21
+import gplx.xowa.langs.cases.Xol_case_cvt;
-		frame_ttl = Xoa_ttl.Replace_unders(frame_ttl);
+import gplx.xowa.langs.cases.Xol_case_mgr;
-		if (frame_ns == Xow_ns_.Tid__template)
+import gplx.xowa.parsers.Xop_ctx;
-			frame_ttl = Bry_.Add(wiki.Ns_mgr().Ns_template().Name_db_w_colon(), Xoa_ttl.Replace_unders(frame_ttl)); // NOTE: always prepend "Template:" to frame_ttl; DATE:2014-06-13; always use spaces; DATE:2014-08-14; must be local language; Russian "Шаблон" not English "Template"; PAGE:ru.w:Королевство_Нидерландов DATE:2016-11-23
+import gplx.xowa.parsers.Xop_tkn_mkr;
-		rv.Frame_ttl_(frame_ttl);
+import gplx.xowa.wikis.nss.Xow_ns_;
-		int orig_args_len = orig.Args_len();
+
-		boolean tmpl_args_parsing_orig = ctx.Tmpl_args_parsing();
+public class Xot_defn_tmpl_ {
-		ctx.Tmpl_args_parsing_(true);
+	public static Xot_invk CopyNew(Xop_ctx ctx, Xot_defn_tmpl orig_defn, Xot_invk orig, Xot_invk caller, byte[] src, int frame_ns, byte[] frame_ttl) {    // SEE:NOTE_1
-		for (int i = 0; i < orig_args_len; i++) {
+		Xop_tkn_mkr tkn_mkr = ctx.Tkn_mkr();
-			Arg_nde_tkn orig_arg = orig.Args_get_by_idx(i);
+		byte[] orig_src = orig_defn.Data_raw();
-			Arg_nde_tkn copy_arg = tkn_mkr.ArgNde(-1, 0);
+		Xowe_wiki wiki = ctx.Wiki();
-			if (orig_arg.KeyTkn_exists()) {
+		Xot_invk_temp rv = Xot_invk_temp.New(orig.Defn_tid(), ctx.Page().Ttl().Page_txt(), orig.Name_tkn(), orig_src, caller.Src_bgn(), caller.Src_end());
-				Arg_itm_tkn key_tkn = orig_arg.Key_tkn();
+
-				copy_arg.Key_tkn_(Make_itm(false, ctx, tkn_mkr, src, key_tkn, caller, orig_arg));
+		// DATE:2014-06-21: always uppercase 1st; EX:{{navbox -> "Template:Navbox"; PAGE:en.w:Achilles
-				rv.Args_add_by_key(copy_arg.Key_tkn().Dat_ary(), copy_arg);	// NOTE: was originally Bry_.Mid(caller.Src(), key_tkn.Dat_bgn(), key_tkn.Dat_end()) which was wrong; caused {{{increment}}} instead of "increment"
+		// DATE:2020-08-09: ISSUE#:784; uppercase non-ascii chars; NOTE: do not reuse byte array, else will cause Xot_defn_trace tests to fail
-			}
+		frame_ttl = Xol_case_cvt.Upper_1st(frame_ttl, 0, frame_ttl.length, Bool_.N);
-			else
+
-				rv.Args_add_by_idx(copy_arg);	// NOTE: not a key, so add to idx_hash; DATE:2014-07-23
+		// DATE:2014-08-14: always use spaces
-			copy_arg.Val_tkn_(Make_itm(true, ctx, tkn_mkr, src, orig_arg.Val_tkn(), caller, orig_arg));
+		frame_ttl = Xoa_ttl.Replace_unders(frame_ttl);
-			rv.Args_add(copy_arg);
+
-		}
+		// DATE:2014-06-13: always prepend "Template:" to frame_ttl
-		ctx.Tmpl_args_parsing_(tmpl_args_parsing_orig);
+		// DATE:2016-11-23: must be local language; Russian "Шаблон" not English "Template"; PAGE:ru.w:Королевство_Нидерландов
-		return rv;
+		// DATE:2020-08-09: ISSUE#:784; apply to all non-main namespaces; PAGE:en.w:Wikipedia:Wikipedia_Signpost/2015-07-15/Op-ed
-	}
+		if (frame_ns != Xow_ns_.Tid__main) {
-	private static Arg_itm_tkn Make_itm(boolean val_tkn, Xop_ctx ctx, Xop_tkn_mkr tkn_mkr, byte[] src, Arg_itm_tkn orig, Xot_invk caller, Arg_nde_tkn orig_arg) {
+			byte[] nsBry = wiki.Ns_mgr().Ids_get_or_null(frame_ns).Name_db_w_colon();
-		int subs_len = orig.Subs_len();
+			frame_ttl = Bry_.Add(nsBry, frame_ttl);
-		Bry_bfr arg_bfr = Bry_bfr_.New();
+		}
-		for (int i = 0; i < subs_len; i++)
+
-			orig.Subs_get(i).Tmpl_evaluate(ctx, src, caller, arg_bfr);
+		rv.Frame_ttl_(frame_ttl);
-		Arg_itm_tkn rv = tkn_mkr.ArgItm(-1, -1);	// NOTE: was -1, 0; DATE:2013-04-10
+		int orig_args_len = orig.Args_len();
-		byte[] rv_ary = orig_arg.KeyTkn_exists() && val_tkn ? arg_bfr.To_bry_and_clear_and_trim() : arg_bfr.To_bry_and_clear();	// // NOTE: must trim if key_exists; DUPE:TRIM_IF_KEY; PAGE:en.w:Coord in Chernobyl disaster, Sahara
+		boolean tmpl_args_parsing_orig = ctx.Tmpl_args_parsing();
-		rv.Dat_ary_(rv_ary);
+		ctx.Tmpl_args_parsing_(true);
-		return rv;
+		for (int i = 0; i < orig_args_len; i++) {
-	}
+			Arg_nde_tkn orig_arg = orig.Args_get_by_idx(i);
-}
+			Arg_nde_tkn copy_arg = tkn_mkr.ArgNde(-1, 0);
-/*
+			if (orig_arg.KeyTkn_exists()) {
-NOTE_1: Creates an invk_temp from an invk
+				Arg_itm_tkn key_tkn = orig_arg.Key_tkn();
-
+				copy_arg.Key_tkn_(Make_itm(false, ctx, tkn_mkr, src, key_tkn, caller, orig_arg));
-page	{{test_1|a}}
+				rv.Args_add_by_key(copy_arg.Key_tkn().Dat_ary(), copy_arg);	// NOTE: was originally Bry_.Mid(caller.Src(), key_tkn.Dat_bgn(), key_tkn.Dat_end()) which was wrong; caused {{{increment}}} instead of "increment"
-test_1	{{test_2|{{{1|nil_1}}}}}
+			}
-test_2	{{{1|nil_2}}}
+			else
-
+				rv.Args_add_by_idx(copy_arg);	// NOTE: not a key, so add to idx_hash; DATE:2014-07-23
-page   : invk_temp gets created for {{test1|a}} where name=test1 and arg1=a
+			copy_arg.Val_tkn_(Make_itm(true, ctx, tkn_mkr, src, orig_arg.Val_tkn(), caller, orig_arg));
-test_1 : invk_temp gets created for {{test_2|{{{1|nil_1}}}}}
+			rv.Args_add(copy_arg);
-1)	create the invk_tmp tkn, with name=test2
+		}
-2)	copy the args and resolve; in this case -> {{test2|a}}
+		ctx.Tmpl_args_parsing_(tmpl_args_parsing_orig);
-now we can use the invk_temp to call test_2 (and so on if needed)
+		return rv;
-*/
+	}
+	private static Arg_itm_tkn Make_itm(boolean val_tkn, Xop_ctx ctx, Xop_tkn_mkr tkn_mkr, byte[] src, Arg_itm_tkn orig, Xot_invk caller, Arg_nde_tkn orig_arg) {
+		int subs_len = orig.Subs_len();
+		Bry_bfr arg_bfr = Bry_bfr_.New();
+		for (int i = 0; i < subs_len; i++)
+			orig.Subs_get(i).Tmpl_evaluate(ctx, src, caller, arg_bfr);
+		Arg_itm_tkn rv = tkn_mkr.ArgItm(-1, -1);	// NOTE: was -1, 0; DATE:2013-04-10
+		byte[] rv_ary = orig_arg.KeyTkn_exists() && val_tkn ? arg_bfr.To_bry_and_clear_and_trim() : arg_bfr.To_bry_and_clear();	// // NOTE: must trim if key_exists; DUPE:TRIM_IF_KEY; PAGE:en.w:Coord in Chernobyl disaster, Sahara
+		rv.Dat_ary_(rv_ary);
+		return rv;
+	}
+}
+/*
+NOTE_1: Creates an invk_temp from an invk
+
+page	{{test_1|a}}
+test_1	{{test_2|{{{1|nil_1}}}}}
+test_2	{{{1|nil_2}}}
+
+page   : invk_temp gets created for {{test1|a}} where name=test1 and arg1=a
+test_1 : invk_temp gets created for {{test_2|{{{1|nil_1}}}}}
+1)	create the invk_tmp tkn, with name=test2
+2)	copy the args and resolve; in this case -> {{test2|a}}
+now we can use the invk_temp to call test_2 (and so on if needed)
+*/

400_xowa/tst/gplx/xowa/parsers/tmpls/Xot_defn_tmpl_Test.java

@@ -0,0 +1,69 @@
+/*
+XOWA: the XOWA Offline Wiki Application
+Copyright (C) 2012-2020 gnosygnu@gmail.com
+
+XOWA is licensed under the terms of the General Public License (GPL) Version 3,
+or alternatively under the terms of the Apache License Version 2.0.
+
+You may use XOWA according to either of these licenses as is most appropriate
+for your project on a case-by-case basis.
+
+The terms of each license can be found in the source code repository:
+
+GPLv3 License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-GPLv3.txt
+Apache License: https://github.com/gnosygnu/xowa/blob/master/LICENSE-APACHE2.txt
+*/
+
+package gplx.xowa.parsers.tmpls;
+
+import gplx.Bry_;
+import gplx.core.tests.Gftest;
+import gplx.xowa.Xop_fxt;
+import gplx.xowa.Xowe_wiki;
+import gplx.xowa.parsers.Xop_ctx;
+import gplx.xowa.wikis.nss.Xow_ns_;
+import org.junit.Test;
+
+public class Xot_defn_tmpl_Test {
+	private final Xot_defn_tmpl__fxt fxt = new Xot_defn_tmpl__fxt();
+
+	@Test
+    public void CopyNewMain() {
+    	fxt.Test_CopyNew_FrameTtl(Xow_ns_.Tid__main, "a", "A");
+	}
+
+	@Test
+    public void CopyNewUppercaseNonAscii() {
+    	fxt.Test_CopyNew_FrameTtl(Xow_ns_.Tid__main, "à", "À");
+	}
+
+	@Test
+    public void CopyNewSpaces() {
+    	fxt.Test_CopyNew_FrameTtl(Xow_ns_.Tid__help_talk, "a_b", "Help_talk:A b");
+	}
+
+	@Test
+    public void CopyNewTemplate() {
+    	fxt.Test_CopyNew_FrameTtl(Xow_ns_.Tid__template, "a", "Template:A");
+	}
+
+	@Test
+    public void CopyNewNonMain() {
+    	fxt.Test_CopyNew_FrameTtl(Xow_ns_.Tid__project, "a", "Wikipedia:A");
+	}
+}
+class Xot_defn_tmpl__fxt {
+	private final Xop_fxt fxt = new Xop_fxt();
+	public void Test_CopyNew_FrameTtl(int frameNs, String frameTtlStr, String expdFrameTtl) {
+		Xowe_wiki wiki = fxt.Wiki();
+		Xop_ctx ctx = Xop_ctx.New__top(wiki);
+		byte[] frameTtlBry = Bry_.new_u8(frameTtlStr);
+		Xot_defn_tmpl orig_defn = new Xot_defn_tmpl();
+		Xot_invk orig = Xot_invk_temp.New_root(Bry_.new_u8("orig"));
+		Xot_invk caller = Xot_invk_temp.New_root(Bry_.new_u8("caller"));
+		byte[] src = Bry_.Empty;
+
+		Xot_invk tmpl =  Xot_defn_tmpl_.CopyNew(ctx, orig_defn, orig, caller, src, frameNs, frameTtlBry);
+		Gftest.Eq__bry(Bry_.new_u8(expdFrameTtl), tmpl.Frame_ttl());
+	}
+}