diff --git a/src/HtmlAgilityPack.Shared/HtmlDocument.cs b/src/HtmlAgilityPack.Shared/HtmlDocument.cs
index 2871d40..37fa647 100644
--- a/src/HtmlAgilityPack.Shared/HtmlDocument.cs
+++ b/src/HtmlAgilityPack.Shared/HtmlDocument.cs
@@ -194,11 +194,18 @@ public static bool DisableBehaviorTagP
///
public bool OptionWriteEmptyNodes;
- #endregion
+ ///
+ /// The max number of nested child nodes.
+ /// Added to prevent stackoverflow problem when a page has tens of thousands of opening html tags with no closing tags
+ ///
+ public int OptionMaxNestedChildNodes = 0;
+
+
+ #endregion
- #region Static Members
+ #region Static Members
- internal static readonly string HtmlExceptionRefNotChild = "Reference node must be a child of this node";
+ internal static readonly string HtmlExceptionRefNotChild = "Reference node must be a child of this node";
internal static readonly string HtmlExceptionUseIdAttributeFalse = "You need to set UseIdAttribute property to true to enable this feature";
@@ -1093,7 +1100,7 @@ private string CurrentNodeName()
private void DecrementPosition()
{
_index--;
- if (_lineposition == 1)
+ if (_lineposition == 0)
{
_lineposition = _maxlineposition;
_line--;
@@ -1190,7 +1197,7 @@ private void IncrementPosition()
_maxlineposition = _lineposition;
if (_c == 10)
{
- _lineposition = 1;
+ _lineposition = 0;
_line++;
}
else
@@ -1228,7 +1235,7 @@ private bool NewCheck()
break;
case ParseState.BetweenAttributes:
- PushAttributeNameStart(_index - 1);
+ PushAttributeNameStart(_index - 1, _lineposition -1);
break;
case ParseState.WhichTag:
@@ -1255,7 +1262,7 @@ private bool NewCheck()
{
if (Text[_index] == '!')
{
- PushNodeStart(HtmlNodeType.Comment, _index - 1);
+ PushNodeStart(HtmlNodeType.Comment, _index - 1, _lineposition -1);
PushNodeNameStart(true, _index);
PushNodeNameEnd(_index + 1);
_state = ParseState.Comment;
@@ -1276,7 +1283,7 @@ private bool NewCheck()
}
}
- PushNodeStart(HtmlNodeType.Element, _index - 1);
+ PushNodeStart(HtmlNodeType.Element, _index - 1, _lineposition - 1);
return true;
}
@@ -1298,8 +1305,8 @@ private void Parse()
_fullcomment = false;
_parseerrors = new List();
_line = 1;
- _lineposition = 1;
- _maxlineposition = 1;
+ _lineposition = 0;
+ _maxlineposition = 0;
_state = ParseState.Text;
_oldstate = _state;
@@ -1312,7 +1319,7 @@ private void Parse()
_currentattribute = null;
_index = 0;
- PushNodeStart(HtmlNodeType.Text, 0);
+ PushNodeStart(HtmlNodeType.Text, 0, _lineposition);
while (_index < Text.Length)
{
_c = Text[_index];
@@ -1391,7 +1398,7 @@ private void Parse()
if (_state != ParseState.Tag)
continue;
_state = ParseState.Text;
- PushNodeStart(HtmlNodeType.Text, _index);
+ PushNodeStart(HtmlNodeType.Text, _index, _lineposition);
}
break;
@@ -1421,11 +1428,11 @@ private void Parse()
if (_state != ParseState.BetweenAttributes)
continue;
_state = ParseState.Text;
- PushNodeStart(HtmlNodeType.Text, _index);
+ PushNodeStart(HtmlNodeType.Text, _index, _lineposition);
continue;
}
- PushAttributeNameStart(_index - 1);
+ PushAttributeNameStart(_index - 1, _lineposition -1);
_state = ParseState.AttributeName;
break;
@@ -1445,7 +1452,7 @@ private void Parse()
if (_state != ParseState.EmptyTag)
continue;
_state = ParseState.Text;
- PushNodeStart(HtmlNodeType.Text, _index);
+ PushNodeStart(HtmlNodeType.Text, _index, _lineposition);
continue;
}
@@ -1496,7 +1503,7 @@ private void Parse()
if (_state != ParseState.AttributeName)
continue;
_state = ParseState.Text;
- PushNodeStart(HtmlNodeType.Text, _index);
+ PushNodeStart(HtmlNodeType.Text, _index, _lineposition);
continue;
}
@@ -1520,7 +1527,7 @@ private void Parse()
if (_state != ParseState.AttributeBeforeEquals)
continue;
_state = ParseState.Text;
- PushNodeStart(HtmlNodeType.Text, _index);
+ PushNodeStart(HtmlNodeType.Text, _index, _lineposition);
continue;
}
@@ -1562,7 +1569,7 @@ private void Parse()
if (_state != ParseState.AttributeAfterEquals)
continue;
_state = ParseState.Text;
- PushNodeStart(HtmlNodeType.Text, _index);
+ PushNodeStart(HtmlNodeType.Text, _index, _lineposition);
continue;
}
@@ -1594,7 +1601,7 @@ private void Parse()
if (_state != ParseState.AttributeValue)
continue;
_state = ParseState.Text;
- PushNodeStart(HtmlNodeType.Text, _index);
+ PushNodeStart(HtmlNodeType.Text, _index, _lineposition);
continue;
}
@@ -1645,7 +1652,7 @@ private void Parse()
}
_state = ParseState.Text;
- PushNodeStart(HtmlNodeType.Text, _index);
+ PushNodeStart(HtmlNodeType.Text, _index, _lineposition);
continue;
}
@@ -1706,11 +1713,11 @@ private void Parse()
script._outerlength = _index - 1 - script._outerstartindex;
script._streamposition = script._outerstartindex;
script._line = _currentnode.Line;
- script._lineposition = _currentnode.LinePosition + _currentnode._namelength + 2;
- _currentnode.AppendChild(script);
+ script._lineposition = _currentnode.LinePosition + _currentnode._namelength + 2;
+ _currentnode.AppendChild(script);
- PushNodeStart(HtmlNodeType.Element, _index - 1);
+ PushNodeStart(HtmlNodeType.Element, _index - 1, _lineposition -1);
PushNodeNameStart(false, _index - 1 + 2);
_state = ParseState.Tag;
IncrementPosition();
@@ -1743,12 +1750,12 @@ private void PushAttributeNameEnd(int index)
_currentnode.Attributes.Append(_currentattribute);
}
- private void PushAttributeNameStart(int index)
+ private void PushAttributeNameStart(int index, int lineposition)
{
_currentattribute = CreateAttribute();
_currentattribute._namestartindex = index;
_currentattribute.Line = _line;
- _currentattribute._lineposition = _lineposition;
+ _currentattribute._lineposition = lineposition;
_currentattribute._streamposition = index;
}
@@ -2056,16 +2063,11 @@ private void PushNodeNameStart(bool starttag, int index)
_currentnode._namestartindex = index;
}
- private void PushNodeStart(HtmlNodeType type, int index)
+ private void PushNodeStart(HtmlNodeType type, int index, int lineposition)
{
_currentnode = CreateNode(type, index);
_currentnode._line = _line;
- _currentnode._lineposition = _lineposition;
- if (type == HtmlNodeType.Element)
- {
- _currentnode._lineposition--;
- }
-
+ _currentnode._lineposition = lineposition;
_currentnode._streamposition = index;
}
diff --git a/src/HtmlAgilityPack.Shared/HtmlNode.cs b/src/HtmlAgilityPack.Shared/HtmlNode.cs
index cfceff1..5be7971 100644
--- a/src/HtmlAgilityPack.Shared/HtmlNode.cs
+++ b/src/HtmlAgilityPack.Shared/HtmlNode.cs
@@ -618,6 +618,12 @@ public string XPath
}
}
+
+ ///
+ /// The depth of the node relative to the opening root html element. This value is used to determine if a document has to many nested html nodes which can cause stack overflows
+ ///
+ public int Depth { get; set; }
+
#endregion
#region Public Methods
@@ -1859,6 +1865,23 @@ public string WriteTo()
}
}
+ ///
+ /// Sets the parent Html node and properly determines the current node's depth using the parent node's depth.
+ ///
+ public void SetParent(HtmlNode parent)
+ {
+ if (parent == null)
+ return;
+
+ ParentNode = parent;
+ if (OwnerDocument.OptionMaxNestedChildNodes > 0)
+ {
+ Depth = parent.Depth + 1;
+ if (Depth > OwnerDocument.OptionMaxNestedChildNodes)
+ throw new Exception(string.Format("Document has more than {0} nested tags. This is likely due to the page not closing tags properly.", OwnerDocument.OptionMaxNestedChildNodes));
+ }
+ }
+
#endregion
#region Internal Methods
diff --git a/src/HtmlAgilityPack.Shared/HtmlNodeCollection.cs b/src/HtmlAgilityPack.Shared/HtmlNodeCollection.cs
index ca194ed..43df9a2 100644
--- a/src/HtmlAgilityPack.Shared/HtmlNodeCollection.cs
+++ b/src/HtmlAgilityPack.Shared/HtmlNodeCollection.cs
@@ -220,9 +220,9 @@ public void Insert(int index, HtmlNode node)
if (next == node)
throw new InvalidProgramException("Unexpected error.");
- node._nextnode = next;
- node._parentnode = _parentnode;
- }
+ node._nextnode = next;
+ node.SetParent(_parentnode);
+ }
///
/// Remove node
@@ -315,8 +315,8 @@ public void Append(HtmlNode node)
_items.Add(node);
node._prevnode = last;
node._nextnode = null;
- node._parentnode = _parentnode;
- if (last == null) return;
+ node.SetParent(_parentnode);
+ if (last == null) return;
if (last == node)
throw new InvalidProgramException("Unexpected error.");
@@ -363,9 +363,9 @@ public void Prepend(HtmlNode node)
throw new InvalidProgramException("Unexpected error.");
node._nextnode = first;
node._prevnode = null;
- node._parentnode = _parentnode;
+ node.SetParent(_parentnode);
- if (first != null)
+ if (first != null)
first._prevnode = node;
}
@@ -415,9 +415,9 @@ public void Replace(int index, HtmlNode node)
throw new InvalidProgramException("Unexpected error.");
node._nextnode = next;
- node._parentnode = _parentnode;
+ node.SetParent(_parentnode);
- oldnode._prevnode = null;
+ oldnode._prevnode = null;
oldnode._nextnode = null;
oldnode._parentnode = null;
}