1 files changed, 77 insertions, 3 deletions

diff --git a/unicode/biditest2.C b/unicode/biditest2.C
index 8e9d7da..a14b3ea 100644
--- a/unicode/biditest2.C
+++ b/unicode/biditest2.C
@@ -204,14 +204,14 @@ void character_test()
 			? unicode::bidi_calc(s, direction)
 			: unicode::bidi_calc(s);
 
-		if (std::get<1>(ret) != paragraph_embedding_level)
+		if (std::get<1>(ret).direction != paragraph_embedding_level)
 		{
 			std::cerr << "Regression, line "
 				  << linenum
 				  << ": expected "
 				  << paragraph_embedding_level
 				  << " paragraph embedding level, got "
-				  << (int)std::get<1>(ret)
+				  << (int)std::get<1>(ret).direction
 				  << std::endl;
 			exit(1);
 		}
@@ -274,6 +274,9 @@ void character_test()
 				 std::reverse(b+index, b+index+n);
 			 });
 
+		size_t cleaned_size=unicode_bidi_cleaned_size(s.c_str(),
+							      s.size(), 0);
+
 		n=0;
 		unicode::bidi_cleanup
 			(s, levels,
@@ -285,6 +288,17 @@ void character_test()
 				 ++n;
 			 });
 
+		if (cleaned_size != s.size())
+		{
+			std::cerr << "Regression, line "
+				  << linenum
+				  << ": default cleaned size"
+				  << std::endl
+				  << "   Expected size: " << cleaned_size
+				  << ", actual size: " << s.size()
+				  << std::endl;
+			exit(1);
+		}
 		if (render_order != actual_render_order)
 		{
 			std::cerr << "Regression, line "
@@ -408,6 +422,12 @@ void character_test()
 			}
 
 			unicode::bidi_reorder(new_string, std::get<0>(ret));
+
+			cleaned_size=unicode_bidi_cleaned_size
+				(new_string.c_str(),
+				 new_string.size(),
+				 UNICODE_BIDI_CLEANUP_CANONICAL);
+
 			unicode::bidi_cleanup(new_string,
 					      std::get<0>(ret),
 					      []
@@ -416,6 +436,20 @@ void character_test()
 					      },
 					      UNICODE_BIDI_CLEANUP_CANONICAL);
 
+			if (cleaned_size != new_string.size())
+			{
+				std::cerr << "Regression, line "
+					  << linenum
+					  << ": canonoical cleaned size"
+					  << std::endl
+					  << "   Expected size: "
+					  << cleaned_size
+					  << ", actual size: "
+					  << new_string.size()
+					  << std::endl;
+				exit(1);
+			}
+
 			/* New string is now back in logical order */
 
 			if (new_string == s && std::get<0>(ret) == levels)
@@ -548,34 +582,62 @@ void null_character_test()
 			      },
 			      UNICODE_BIDI_CLEANUP_EXTRA,
 			      0, 3);
+
+	s=U"";
+	res=unicode::bidi_calc(s, UNICODE_BIDI_RL);
+
+	if (std::get<1>(res).direction != UNICODE_BIDI_RL)
+	{
+		std::cerr << "Paragraph embedding level not honored"
+			  << std::endl;
+		exit(1);
+	}
 }
 
 void direction_test()
 {
 	static const struct {
-		const char32_t *str;
+		std::u32string str;
 		unicode_bidi_level_t direction;
 		int is_explicit;
+		bool needs_embed;
 	} tests[]={
 		{
 			U"Hello",
 			UNICODE_BIDI_LR,
 			1,
+			true,
 		},
 		{
 			U" ",
 			UNICODE_BIDI_LR,
 			0,
+			true,
 		},
 		{
 			U"",
 			UNICODE_BIDI_LR,
 			0,
+			true,
 		},
 		{
 			U"שלום",
 			UNICODE_BIDI_RL,
 			1,
+			true,
+		},
+		{
+			U"Helloש",
+			UNICODE_BIDI_LR,
+			1,
+			true,
+		},
+		{
+			U"Hello" + std::u32string{unicode::literals::LRO}
+			+ U"ש",
+			UNICODE_BIDI_LR,
+			1,
+			false,
 		},
 	};
 
@@ -589,6 +651,18 @@ void direction_test()
 			std::cerr << "direction_test failed\n";
 			exit(1);
 		}
+
+		std::u32string s=t.str;
+		auto levels=std::get<0>(unicode::bidi_calc(s, t.direction));
+		unicode::bidi_reorder(s, levels);
+		unicode::bidi_cleanup(s, levels);
+
+		if (unicode::bidi_needs_embed(s, levels, &t.direction)
+		    != t.needs_embed)
+		{
+			std::cerr << "needs embed failed\n";
+			exit(1);
+		}
 	}
 }