fix: put a white char before an opening tag in tree mode if it was in the source html

src/html/htmlparser.cpp

@@ -5,7 +5,7 @@
  */
 
 /* 
- * Copyright (c) 2008-2021, Tomasz Sowa
+ * Copyright (c) 2008-2022, Tomasz Sowa
  * All rights reserved.
  * 
  * Redistribution and use in source and binary forms, with or without
@@ -71,16 +71,17 @@ void HTMLParser::clear_input_flags()
 void HTMLParser::Item::Clear()
 {
 	name.clear();
-	type            = none;
-	is_commentary   = false;
-	is_cdata		= false;
-	porphans        = nullptr;
-	new_line_before = false;
-	new_line        = false;
+	type				= none;
+	is_commentary		= false;
+	is_cdata			= false;
+	porphans			= nullptr;
+	new_line_before		= false;
+	new_line_after		= false;
 	new_line_in_the_middle = false;
-	has_body_tag    = false;
-	tree_index      = 0;
-	space           = nullptr;
+	white_char_before	= false;
+	has_body_tag		= false;
+	tree_index			= 0;
+	space				= nullptr;
 }
 
 
@@ -1175,10 +1176,18 @@ bool HTMLParser::PutOpeningTag()
 		return false;
 	}
 
-	if( current_white_char_mode() == WHITE_MODE_TREE && LastItem().new_line_before )
+	if( current_white_char_mode() == WHITE_MODE_TREE )
 	{
-		Put(10);
-		PutTabs(LastItem().tree_index);
+		if( LastItem().new_line_before )
+		{
+			Put(10);
+			PutTabs(LastItem().tree_index);
+		}
+		else
+		if( LastItem().white_char_before )
+		{
+			Put(' ');
+		}
 	}
 
 	PutOpeningTagMark();
@@ -1414,6 +1423,8 @@ void HTMLParser::ReadTextUntilClosingTag(bool put_closing_tag_as_well)
 void HTMLParser::ReadText(bool is_cdata)
 {
 	new_item_has_new_line_before = false;
+	new_item_has_white_char_before = false;
+
 	bool was_white_char = false;
 	bool was_new_line = false;
 
@@ -1424,7 +1435,7 @@ void HTMLParser::ReadText(bool is_cdata)
 
 	if( current_white_char_mode() == WHITE_MODE_TREE )
 	{
-		if( LastItem().new_line || (wrap_line != 0 && LastItem().has_body_tag && line_len >= wrap_line) )
+		if( LastItem().new_line_after || (wrap_line != 0 && LastItem().has_body_tag && line_len >= wrap_line) )
 		{
 			allow_put_new_line = true;
 		}
@@ -1450,7 +1461,10 @@ void HTMLParser::ReadText(bool is_cdata)
 		was_closing_tag = PutNormalNonWhite(tmp_text, allow_put_new_line, allow_put_space, is_cdata);
 
 		if( lastc == -1 || was_closing_tag )
+		{
 			new_item_has_new_line_before = was_new_line;
+			new_item_has_white_char_before = was_white_char;
+		}
 
 		if( !tmp_text.empty() )
 		{
@@ -1486,7 +1500,7 @@ void HTMLParser::ReadText(bool is_cdata)
 					LastItem().new_line_in_the_middle = true;
 
 					if( !was_non_white_text )
-						LastItem().new_line = true;
+						LastItem().new_line_after = true;
 				}
 				else
 				{
@@ -1758,6 +1772,7 @@ bool HTMLParser::ReadItem()
 		return false;
 
 	LastItem().new_line_before = new_item_has_new_line_before; // new_item_has_new_line_before is set by ReadText() method
+	LastItem().white_char_before = new_item_has_white_char_before; // new_item_has_white_char_before is set by ReadText() method
 
 	if( stack_len > 1 && pstack[stack_len-2].new_line_in_the_middle )
 		LastItem().tree_index += 1;
@@ -1971,7 +1986,7 @@ void HTMLParser::CheckStackPrintRest()
 {
 	while( stack_len-- > 0 )
 	{
-		if( stack_len==0 || pstack[stack_len-1].new_line )
+		if( stack_len==0 || pstack[stack_len-1].new_line_after )
 		{
 			if( current_white_char_mode() == WHITE_MODE_TREE )
 			{
@@ -2030,7 +2045,7 @@ void HTMLParser::CheckClosingTags()
 
 		if( !skip_tags && IsTagSafe(LastItem().name) && !IsNameEqual(no_filter_tag, LastItem().name) )
 		{
-			if( pstack[z].new_line )
+			if( pstack[z].new_line_after )
 			{
 				if( current_white_char_mode() == WHITE_MODE_TREE )
 				{

src/html/htmlparser.h

@@ -5,7 +5,7 @@
  */
 
 /* 
- * Copyright (c) 2008-2021, Tomasz Sowa
+ * Copyright (c) 2008-2022, Tomasz Sowa
  * All rights reserved.
  * 
  * Redistribution and use in source and binary forms, with or without
@@ -134,14 +134,18 @@ public:
 
 		bool is_cdata;
 
+		// is a new line before this tag (or just a new line and some white characters)
 		bool new_line_before;
 
-		// is there a new line after this tag
-		bool new_line;
+		// is there a new line after this tag (or just some white characters and a new line)
+		bool new_line_after;
 
-		// is there a new
+		// is there a new line in the middle after this tag and before the next tag
 		bool new_line_in_the_middle;
 
+		// is there a white char (but not new line) before this tag
+		bool white_char_before;
+
 		// current orphans table
 		// (will be propagated)
 		Orphans * porphans;
@@ -448,6 +452,7 @@ protected:
 
 	//bool last_new_line;
 	bool new_item_has_new_line_before;
+	bool new_item_has_white_char_before;
 	int white_mode;
 	bool is_first_item;
 	size_t wrap_line;		// insert a new line character into long lines