public override IHtmlElement Create(HtmlAttributesCollection attributes, string html, IHtmlParserManager htmlParserManager)
{
HtmlDocTypeElement document = new HtmlDocTypeElement();
document.Attributes = attributes;
return(document);
}
public override IHtmlElement Create(HtmlAttributesCollection attributes, string html, int startContentIndex, int endContentIndex, IHtmlParserManager htmlParserManager)
{
HtmlNodeElement element = (HtmlNodeElement)CreateInstance();
// add children and texts
if (endContentIndex - startContentIndex > 0)
{
element.Children = htmlParserManager.ParseElements(startContentIndex, endContentIndex);
Dictionary <int, string> texts = htmlParserManager.ParseText(startContentIndex, endContentIndex);
foreach (var text in texts)
{
string decodedText = HttpUtility.HtmlDecode(text.Value);
string value = text.Value;
if (Regex.IsMatch(decodedText, HtmlRegexParserManager.commentRegex))
{
value = decodedText;
}
if (text.Key >= 0)
{
element.Text(new HtmlString(value), element.Children[text.Key].UId);
}
else
{
element.Text(new HtmlString(value));
}
}
}
element.Attributes = attributes;
return(element);
}
public virtual IHtmlElement Create(HtmlAttributesCollection attributes, string html, IHtmlParserManager htmlParserManager)
{
IHtmlElement element = (HtmlSelfClosingTagElement)CreateInstance();
element.Attributes = attributes;
return(element);
}
public override IHtmlElement Create(HtmlAttributesCollection attributes, string html, int startContentIndex, int endContentIndex, IHtmlParserManager htmlParserManager)
{
HtmlScriptElement element = new HtmlScriptElement();
element.Attributes = attributes;
element.Text(html.SubStringToIndex(startContentIndex, endContentIndex - 1), true);
return(element);
}
public virtual IHtmlElement Create(HtmlAttributesCollection attributes, string html, int startContentIndex, int endContentIndex, IHtmlParserManager htmlParserManager)
{
HtmlPairTagsElement element = (HtmlPairTagsElement)CreateInstance();
element.Attributes = attributes;
element.Text(html.SubStringToIndex(startContentIndex, endContentIndex - 1));
return(element);
}
public IHtmlElementsCollection ParseElements(int startIndex, int endIndex)
{
IHtmlElementsCollection elements = new HtmlElementsCollection();
// get all matches in current range
List <Match> startTagMatchesInCurrentRange = _startTagsMathes.Where(x => x.Index >= startIndex && x.Index < endIndex).ToList();
List <Match> pairStartTagMatchesInCurrentRange = _pairStartTagsMathes.Where(x => x.Index >= startIndex && x.Index < endIndex).ToList();
List <Match> endTagMatchesInCurrentRange = _endTagsMathes.Where(x => x.Index > startIndex && x.Index < endIndex).ToList();
try
{
// get root start tags in current range
List <Match> rootStartTagMatches = startTagMatchesInCurrentRange
.Where(currentStartTag =>
pairStartTagMatchesInCurrentRange
.Count(startTag => startTag.Index < currentStartTag.Index) == endTagMatchesInCurrentRange.Count(endTag => endTag.Index <= currentStartTag.Index)
)
.ToList();
// get root end tags in current range
List <Match> rootEndTagMatches = endTagMatchesInCurrentRange
.Where(currentEndTag =>
pairStartTagMatchesInCurrentRange
.Count(startTag => startTag.Index < currentEndTag.Index) == endTagMatchesInCurrentRange.Count(endTag2 => endTag2.Index <= currentEndTag.Index)
)
.ToList();
// create IHtmlElement for every start tag match and add to collection
foreach (Match startTagMatch in rootStartTagMatches)
{
// get tag name
string tagName = _htmlHelper.ExtractTagNameFromStartTag(startTagMatch.Value);
// parse attributes
HtmlAttributesCollection attributes = ParseAttributes(startTagMatch.Value);
// if tag is self closing
if (_htmlHelper.IsSelfClosingHtmlTag(tagName))
{
// create elements factory
IHtmlSelfClosingTagElementFactory htmlElementsFactory;
switch (tagName.ToLower())
{
case "!doctype":
htmlElementsFactory = new HtmlDoctypeElementFactory();
break;
default:
htmlElementsFactory = new HtmlSelfClosingTagElementFactory(tagName);
break;
}
// create element
IHtmlElement element = htmlElementsFactory.Create(attributes, _encodedHtml, this);
// add
elements.Add(element);
}
// when tag have pair tags
else
{
// create elements factory
IHtmlPairTagsElementFactory htmlElementsFactory;
switch (tagName.ToLower())
{
case "noscript":
htmlElementsFactory = new HtmlNoScriptElementFactory();
break;
case "script":
htmlElementsFactory = new HtmlScriptElementFactory();
break;
case "style":
htmlElementsFactory = new HtmlStyleElementFactory();
break;
case "code":
htmlElementsFactory = new HtmlCodeElementFactory();
break;
default:
htmlElementsFactory = new HtmlNodeElementFactory(tagName);
break;
}
// find start content index
int startContentIndex = startTagMatch.Index + startTagMatch.Value.Length;
// find cloing tang on current star tag
Match endTagMatch = rootEndTagMatches.FirstOrDefault(x => x.Index > startTagMatch.Index);
// in html may have tags which should have end tag but he is not defined, in this case just skip him as set end index to be end index on current start tag
int endContentIndex = startTagMatch.Index + startTagMatch.Value.Length;
if (endTagMatch != null)
{
endContentIndex = endTagMatch.Index;
}
// create element
IHtmlElement element = htmlElementsFactory.Create(attributes, _encodedHtml, startContentIndex, endContentIndex, this);
// add
elements.Add(element);
}
}
}
catch (Exception ex)
{
return(elements);
}
return(elements);
}
/// <summary>
/// Create attributes collection including in the start tag string.
/// </summary>
/// <param name="startTag">Html start tag string. Example <span id="1">, <div class="div">, ect. </param>
/// <returns cref="HtmlAttributesCollection">HtmlAttributesCollection whit all found attributes in given start tag.</returns>
private HtmlAttributesCollection ParseAttributes(string startTag)
{
HtmlAttributesCollection attributes = new HtmlAttributesCollection();
// decode
startTag = HttpUtility.HtmlDecode(startTag);
// replace spacing between name-value
startTag = Regex.Replace(startTag, keyValueAttributeEqualSymbolSpacingRegex, "=");
// return if dont have attributes
int endTagNameIndex = startTag.IndexOf(' ');
if (startTag.IndexOf("\r\n") > 0)
{
if (startTag.IndexOf("\r\n") < endTagNameIndex || endTagNameIndex == -1)
{
endTagNameIndex = startTag.IndexOf("\r\n") + 2;
}
}
if (endTagNameIndex < 0)
{
return(attributes);
}
// remove tag name from string
startTag = startTag.Remove(0, endTagNameIndex);
// trim start
startTag = startTag.TrimEnd(new char[] { '>', '/' }).TrimEnd();
try
{
while (startTag.IsNotNullNorEmpty())
{
startTag = startTag.TrimStart();
// create attribute name
string attributeName = "";
for (int i = 0; i < startTag.Length; i++)
{
if (startTag[i] == ' ' || startTag[i] == '=')
{
break;
}
attributeName += startTag[i];
}
// create attributes factory
IHtmlAttributeFactory htmlAttributeFactory;
switch (attributeName)
{
case "style":
htmlAttributeFactory = new HtmlStyleAttributeFactory();
break;
default:
htmlAttributeFactory = new HtmlAttributeFactory(attributeName);
break;
}
IHtmlAttribute attribute = htmlAttributeFactory.Create(startTag);
attributes.Add(attribute);
var endAttributeIndex = startTag.IndexOf(attribute.Name) + attribute.Name.Length;
if (attribute.Value.IsNotNullNorEmpty())
{
endAttributeIndex = startTag.IndexOf(attribute.Value, endAttributeIndex) + attribute.Value.Length;
}
else if (startTag.TrimStart().StarstWithPattern(_htmlHelper.GetEmptyValueAttributeRegex(attribute.Name)))
{
endAttributeIndex += 3;
}
startTag = startTag.Remove(0, endAttributeIndex);
startTag = startTag.TrimStart().TrimStart(new char[] { '"', '\'' });
}
}
catch (Exception)
{
return(attributes);
}
return(attributes);
}
public HtmlElement()
{
Attributes = new HtmlAttributesCollection();
}
