From f02dd1093a1bac9b4a99465e256ce2e1504ff5ff Mon Sep 17 00:00:00 2001
From: Tomasz Sowa <t.sowa@ttmath.org>
Date: Tue, 16 Apr 2024 09:35:47 +0200
Subject: [PATCH] fix(HtmlParser): correctly remove an item from the space
 struct when requested from a callback

while here:
- implement the removing algorithm for the compact_mode
---
 src/html/htmlparser.cpp | 54 +++++++++++++++++++++++++++++++----------
 src/html/htmlparser.h   |  2 +-
 src/space/space.cpp     | 17 ++++++++++++-
 src/space/space.h       |  3 ++-
 4 files changed, 60 insertions(+), 16 deletions(-)

diff --git a/src/html/htmlparser.cpp b/src/html/htmlparser.cpp
index 46b2ef8..d08f09b 100644
--- a/src/html/htmlparser.cpp
+++ b/src/html/htmlparser.cpp
@@ -5,7 +5,7 @@
  */
 
 /* 
- * Copyright (c) 2008-2023, Tomasz Sowa
+ * Copyright (c) 2008-2024, Tomasz Sowa
  * All rights reserved.
  * 
  * Redistribution and use in source and binary forms, with or without
@@ -2119,14 +2119,13 @@ void HTMLParser::CheckClosingTags()
 		return;
 	}
 
-	// CHECK ME
-	if( RemoveIfNeeded(stack_len - 2) )
-	{
-		RemoveLastSpace(i);
-	}
-
 	for(int z=(int)stack_len-2 ; z >= i ; --z)
 	{
+		if( RemoveIfNeeded(z) )
+		{
+			RemoveLastSpace(z);
+		}
+
 		CheckWhiteCharsExceptions(pstack[z]);
 
 		if( !skip_tags && IsTagSafe(LastItem().name) && !IsNameEqual(no_filter_tag, LastItem().name) )
@@ -2238,7 +2237,6 @@ void HTMLParser::AddItemToSpace()
 }
 
 
-
 void HTMLParser::RemoveLastSpace(size_t index)
 {
 	if( out_space )
@@ -2252,22 +2250,52 @@ void HTMLParser::RemoveLastSpace(size_t index)
 
 		if( xml_compact_mode )
 		{
-			// IMPLEMENT ME
+			if( parent->is_object() )
+			{
+				Space::ObjectType::iterator i = parent->value.value_object.find(pstack[index].name);
+
+				if( i != parent->value.value_object.end() )
+				{
+					if( i->second->is_table() )
+					{
+						size_t len = i->second->table_size();
+
+						if( len > 0 && i->second->value.value_table[len-1] == pstack[index].space )
+						{
+							i->second->remove(len - 1);
+							pstack[index].space = nullptr;
+						}
+					}
+					else
+					{
+						if( i->second == pstack[index].space )
+						{
+							parent->remove(i);
+							pstack[index].space = nullptr;
+						}
+					}
+				}
+			}
 		}
 		else
 		{
 			Space * childs_tab = parent->get_space(L"childs");
-			size_t len = childs_tab->table_size();
 
-			if( childs_tab && childs_tab->is_table() && len > 0 && childs_tab->value.value_table[len-1] == pstack[stack_len-2].space )
+			if( childs_tab && childs_tab->is_table() )
 			{
-				childs_tab->remove(len - 1);
-				pstack[stack_len-2].space = nullptr;
+				size_t len = childs_tab->table_size();
+
+				if( len > 0 && childs_tab->value.value_table[len-1] == pstack[index].space )
+				{
+					childs_tab->remove(len - 1);
+					pstack[index].space = nullptr;
+				}
 			}
 		}
 	}
 }
 
+
 void HTMLParser::AddTextSpaceToSpaceTree(const Space & space)
 {
 	const std::wstring * text = space.get_wstr(L"text");
diff --git a/src/html/htmlparser.h b/src/html/htmlparser.h
index e5d5157..2f80e84 100644
--- a/src/html/htmlparser.h
+++ b/src/html/htmlparser.h
@@ -5,7 +5,7 @@
  */
 
 /* 
- * Copyright (c) 2008-2023, Tomasz Sowa
+ * Copyright (c) 2008-2024, Tomasz Sowa
  * All rights reserved.
  * 
  * Redistribution and use in source and binary forms, with or without
diff --git a/src/space/space.cpp b/src/space/space.cpp
index a9dddb8..8d7146d 100644
--- a/src/space/space.cpp
+++ b/src/space/space.cpp
@@ -5,7 +5,7 @@
  */
 
 /* 
- * Copyright (c) 2008-2023, Tomasz Sowa
+ * Copyright (c) 2008-2024, Tomasz Sowa
  * All rights reserved.
  * 
  * Redistribution and use in source and binary forms, with or without
@@ -2223,6 +2223,21 @@ void Space::remove(const std::wstring & field)
 }
 
 
+void Space::remove(ObjectType::iterator & iterator)
+{
+	if( type == type_object )
+	{
+		if( iterator != value.value_object.end() )
+		{
+			delete iterator->second;
+			iterator->second = nullptr;
+
+			value.value_object.erase(iterator);
+		}
+	}
+}
+
+
 void Space::remove(size_t table_index)
 {
 	if( type == type_table && table_index < value.value_table.size() )
diff --git a/src/space/space.h b/src/space/space.h
index de659a7..3964a09 100644
--- a/src/space/space.h
+++ b/src/space/space.h
@@ -5,7 +5,7 @@
  */
 
 /* 
- * Copyright (c) 2010-2023, Tomasz Sowa
+ * Copyright (c) 2010-2024, Tomasz Sowa
  * All rights reserved.
  * 
  * Redistribution and use in source and binary forms, with or without
@@ -538,6 +538,7 @@ public:
 	// remove a field from an object
 	void remove(const wchar_t * field);
 	void remove(const std::wstring & field);
+	void remove(ObjectType::iterator & iterator);
 
 	// remove a table item
 	void remove(size_t table_index);