protected override string Replace(Element el)
{
if (el is BeginTag)
{
BeginTag beginTag = (BeginTag)el;
if (beginTag.NameEquals("a"))
{
Attr href = beginTag.GetAttribute("href");
if (href != null && href.Value != null)
{
href.Value = ConvertUrl(href.Value);
return(beginTag.ToString());
}
}
else if (beginTag.NameEquals("img"))
{
Attr src = beginTag.GetAttribute("src");
if (src != null && src.Value != null)
{
src.Value = ConvertUrl(src.Value);
return(beginTag.ToString());
}
}
}
return(base.Replace(el));
}
public string ScanAndPreserve(string html)
{
StringBuilder sb = new StringBuilder(html.Length);
SimpleHtmlParser p = new SimpleHtmlParser(html);
Element e;
while (null != (e = p.Next()))
{
if (!(e is BeginTag))
{
sb.Append(html, e.Offset, e.Length);
continue;
}
BeginTag bt = (BeginTag)e;
if (bt.NameEquals("div"))
{
switch (bt.GetAttributeValue("class"))
{
case ContentSourceManager.EDITABLE_SMART_CONTENT:
case ContentSourceManager.SMART_CONTENT:
sb.Append(html, e.Offset, e.Length);
sb.Append(p.CollectHtmlUntil("div"));
sb.Append("</div>");
continue;
}
}
if (!(bt.NameEquals("object") ||
bt.NameEquals("embed") ||
bt.NameEquals("noembed") ||
bt.NameEquals("script")))
{
sb.Append(html, e.Offset, e.Length);
continue;
}
else
{
string collected = p.CollectHtmlUntil(bt.Name);
string preserve = bt.RawText + collected + "</" + bt.Name + ">";
string preserveId = Guid.NewGuid().ToString("N");
preserved[preserveId] = preserve;
sb.AppendFormat("<span id=\"preserve{0}\" class=\"{1}\">", preserveId, PRESERVE_CLASS);
sb.Append(preserve);
sb.Append("</span>");
}
}
return(sb.ToString());
}
public bool IsMatch(Element e)
{
BeginTag tag = e as BeginTag;
if (tag == null)
{
return(false);
}
if (!tag.NameEquals("meta"))
{
return(false);
}
if (tag.GetAttributeValue("name") != "generator")
{
return(false);
}
string generator = tag.GetAttributeValue("content");
if (generator == null || CaseInsensitiveComparer.DefaultInvariant.Compare("blogger", generator) != 0)
{
return(false);
}
return(true);
}
protected override void OnBeginTag(BeginTag tag)
{
if (FlagIsSet(Flag.RemovePartialTags) && tag.Unterminated)
{
return;
}
//remove all illegal attributes from the tag
foreach (Attr attr in tag.Attributes)
{
if (IsIllegalAttribute(attr))
{
attr.Value = string.Empty;
}
}
if (tag.NameEquals("script"))
{
Debug.WriteLine("Script tag");
}
if (IsRegexMatch(IllegalTagTreeName, tag.Name))
{
suspendTagDepth++;
}
else if (!IsIllegalTag(tag) && suspendTagDepth == 0)
{
PushStartTag(tag.Name);
base.OnBeginTag(tag);
}
}
/// <summary>
/// Is the tag a meaningless tag such as <p></p> or <a href="..."></a> or <a href="..."> </a>
/// </summary>
/// <param name="htmlParser"></param>
/// <param name="bt"></param>
/// <returns></returns>
private static bool RemoveMeaninglessTags(SimpleHtmlParser htmlParser, BeginTag bt)
{
// Look to see if the tag is a <p> without any attributes
if ((bt.NameEquals("p") && bt.Attributes.Length == 0 && !bt.HasResidue))
{
Element e = htmlParser.Peek(0);
// Look to see if thereis a matching end tag to the element we are looking at
if (e != null && e is EndTag && ((EndTag)e).NameEquals("p"))
{
// eat up the end tag
htmlParser.Next();
return(true);
}
}
// Look to see if the tag is an <a> without a style/id/name attribute, but has an href... meaning the link is not useful
if ((bt.NameEquals("a") && bt.GetAttribute("name") == null && bt.GetAttributeValue("style") == null && bt.GetAttributeValue("id") == null && bt.GetAttributeValue("href") != null))
{
bool hadWhiteSpaceText = false;
Element e = htmlParser.Peek(0);
// Look to see if the a just has whitespace inside of it
if (e is Text && HtmlUtils.UnEscapeEntities(e.RawText, HtmlUtils.UnEscapeMode.NonMarkupText).Trim().Length == 0)
{
e = htmlParser.Peek(1);
hadWhiteSpaceText = true;
}
// Look to see if thereis a matching end tag to the element we are looking at
if (e != null && e is EndTag && ((EndTag)e).NameEquals("a"))
{
// if this was an <a> with whitespace in the middle eat it up
if (hadWhiteSpaceText)
{
htmlParser.Next();
}
// eat up the end tag
htmlParser.Next();
return(true);
}
}
return(false);
}
protected override void OnBeginTag(BeginTag tag)
{
if (tag.NameEquals(HTMLTokens.Body))
{
bodyBeginTag = tag;
}
base.OnBeginTag(tag);
}
protected override void OnBeginTag(BeginTag tag)
{
if (tag.NameEquals(HTMLTokens.Ul))
{
unorderedListLevel++;
}
else if (tag.NameEquals(HTMLTokens.Ol))
{
orderedListLevel++;
}
else if ((unorderedListLevel < 1) &&
(orderedListLevel < 1) &&
(tag.NameEquals(HTMLTokens.Li)))
{
hasIncompleteList = true;
}
base.OnBeginTag(tag);
}
private bool TagPermittedAboveBody(BeginTag tag)
{
foreach (string permittedAboveBody in _permittedBeforeBody)
{
if (tag.NameEquals(permittedAboveBody))
{
return(true);
}
}
return(false);
}
public static string Trim(string html, bool onlyTrimParagraphs)
{
Element[] els = Elements(html);
int pos;
// First, go backwards over the list, deleting
// all <br> and whitespace. Stop as soon as
// significant content is encountered.
if (onlyTrimParagraphs)
{
pos = 1 + FindCleanupIndexForParagraphTrim(els);
}
else
{
pos = 1 + FindLastVisibleElementAndRemoveWhitespace(els);
}
// pos now points to the index where whitespace cleanup should begin
// Remove empty pairs of invisible tags, e.g. <b></b>. Each time
// a pair is removed, start over, because the removal
// of an empty pair may create another empty pair, e.g. <p><i></i></p>
while (FindAndRemoveEmptyTag(pos, els))
{
}
// Remove extra unmatched <p> begin tags.
for (int i = pos; i < els.Length; i++)
{
BeginTag bt = els[i] as BeginTag;
if (bt != null && bt.NameEquals("p"))
{
els[i] = null;
}
}
// Concatenate all the elements that are left.
StringBuilder output = new StringBuilder(html.Length);
foreach (Element el in els)
{
if (el != null)
{
output.Append(el.RawText);
}
}
return(output.ToString());
}
protected override void OnBeginTag(BeginTag tag)
{
if (tag.NameEquals(HTMLTokens.Title) && !tag.Complete)
{
_inTitle = true;
}
if (TagsToPreserve.Contains(tag.Name.ToUpper(CultureInfo.InvariantCulture)))
{
EmitTagAndAttributes(tag.Name, tag);
}
else if (ReplaceTags.ContainsKey(tag.Name.ToUpper(CultureInfo.InvariantCulture)))
{
EmitTagAndAttributes((string)ReplaceTags[tag.Name.ToUpper(CultureInfo.InvariantCulture)], tag);
}
}
private bool IsIllegalTag(BeginTag tag)
{
if (IsRegexMatch(IllegalTagName, tag.Name))
{
return(true);
}
else if (FlagIsSet(Flag.RemoveStyles) && tag.NameEquals("link"))
{
//if this link element is a stylesheet, it is illegal
Attr relAttr = tag.GetAttribute("rel");
if (relAttr != null && relAttr.Value != null && relAttr.Value.ToUpperInvariant().Trim() == "STYLESHEET")
{
return(true);
}
}
return(false);
}
public HtmlForm NextForm()
{
Element el;
while (null != (el = parser.Next()))
{
BeginTag tag = el as BeginTag;
if (tag == null)
{
continue;
}
if (tag.NameEquals("form"))
{
return(HandleForm(tag));
}
}
return(null);
}
protected override void OnBeginTag(BeginTag tag)
{
if (tag != null && LightWeightHTMLDocument.AllUrlElements.ContainsKey(tag.Name.ToUpper(CultureInfo.InvariantCulture)))
{
Attr attr = tag.GetAttribute((string)LightWeightHTMLDocument.AllUrlElements[tag.Name.ToUpper(CultureInfo.InvariantCulture)]);
if (attr != null)
{
string url = attr.Value;
if (!UrlHelper.IsUrl(url) && ShouldEscapeRelativeUrl(url))
{
attr.Value = UrlHelper.EscapeRelativeURL(BaseUrl, url);
}
}
}
// Special case params
if (tag != null && tag.NameEquals(HTMLTokens.Param))
{
// Handle Params
foreach (string paramValue in LightWeightHTMLDocument.ParamsUrlElements)
{
Attr attr = tag.GetAttribute(HTMLTokens.Name);
if (attr != null)
{
if (attr.Value.ToUpper(CultureInfo.InvariantCulture) == paramValue)
{
Attr valueAttr = tag.GetAttribute(HTMLTokens.Value);
if (valueAttr != null)
{
string url = valueAttr.Value;
if (!UrlHelper.IsUrl(url))
{
valueAttr.Value = UrlHelper.EscapeRelativeURL(BaseUrl, url);
}
}
}
}
}
}
base.OnBeginTag(tag);
}
private void HandleSelect(HtmlForm parentForm, BeginTag selectTag)
{
string name = selectTag.GetAttributeValue("name");
int dummy;
bool multiple = selectTag.GetAttribute("multiple", true, 0, out dummy) != null;
ArrayList optionInfos = new ArrayList();
Element el = parser.Next();
while (el != null)
{
BeginTag tag = el as BeginTag;
if (tag != null && tag.NameEquals("option"))
{
string value = tag.GetAttributeValue("value");
bool isSelected = tag.GetAttribute("selected", true, 0, out dummy) != null;
string label = string.Empty;
el = parser.Next();
if (el != null && el is Text)
{
label = HtmlUtils.UnEscapeEntities(el.ToString(), HtmlUtils.UnEscapeMode.NonMarkupText).TrimEnd(' ', '\r', '\n', '\t');
el = parser.Next();
}
optionInfos.Add(new OptionInfo(value, label, isSelected));
continue;
}
if (el is EndTag && ((EndTag)el).NameEquals("select"))
{
new Select(parentForm, name, multiple, (OptionInfo[])optionInfos.ToArray(typeof(OptionInfo)));
return;
}
el = parser.Next();
}
}
protected override void OnBeginTag(BeginTag tag)
{
if (tag != null)
{
// Reset any frame urls
// This is done because the HTML that is often in this document may have
// incorrect urls for frames. The frames enumeration is accurate, so if the
// name from the frames enumeration is the same as this frame, we should fix its
// url up.
if (tag.NameEquals(HTMLTokens.Frame))
{
Attr name = tag.GetAttribute(HTMLTokens.Name);
if (name != null && this._frames != null)
{
LightWeightHTMLDocument frameDoc = GetFrameDocumentByName(name.Value);
if (frameDoc != null)
{
Attr src = tag.GetAttribute(HTMLTokens.Src);
if (src != null && src.Value != frameDoc.Url)
{
Generator.AddSubstitionUrl(new UrlToReplace(src.Value, frameDoc.Url));
}
}
}
}
LightWeightTag currentTag = new LightWeightTag(tag);
// The key we'll use for the table
string key = tag.Name.ToUpper(CultureInfo.InvariantCulture);
if (!_tagTable.ContainsKey(key))
{
_tagTable[key] = new LightWeightTag[0];
}
LightWeightTag[] currentTags = (LightWeightTag[])_tagTable[key];
LightWeightTag[] grownTags = new LightWeightTag[currentTags.Length + 1];
currentTags.CopyTo(grownTags, 0);
grownTags[currentTags.Length] = currentTag;
_tagTable[key] = grownTags;
// Accumulate the title text
if (tag.NameEquals(HTMLTokens.Title) && !tag.Complete)
{
_nextTextIsTitleText = true;
}
else if (tag.NameEquals(HTMLTokens.A) && !tag.Complete && tag.GetAttribute(HTMLTokens.Href) != null)
{
if (_collectingForTag != null)
{
if (tag.NameEquals(HTMLTokens.A))
{
_collectingForTagDepth++;
}
}
else
{
_collectingForTag = currentTag;
}
}
}
base.OnBeginTag(tag);
}
protected override void OnBeginTag(BeginTag tag)
{
if (tag == null)
{
return;
}
if (_firstTag)
{
if (!tag.NameEquals(HTMLTokens.Html))
{
EmitTag(HTMLTokens.Html);
}
_firstTag = false;
}
if (!_seenHead && !TagPermittedAboveBody(tag))
{
Emit("<head>");
EmitAdditionalMetaData();
Emit("</head>");
_seenHead = true;
}
if (tag.NameEquals(HTMLTokens.Script))
{
if (!tag.Complete)
{
_scriptDepth++;
}
return;
}
if (tag.NameEquals(HTMLTokens.Head))
{
_seenHead = true;
}
else if (!_seenBody && !tag.NameEquals(HTMLTokens.Body))
{
if (!TagPermittedAboveBody(tag))
{
EmitTag(HTMLTokens.Body);
_seenBody = true;
}
}
else if (!_seenBody && tag.NameEquals(HTMLTokens.Body))
{
_seenBody = true;
}
if (tag.NameEquals(HTMLTokens.Base))
{
if (_metaData == null || _metaData.Base == null)
{
return;
}
else
{
Attr href = tag.GetAttribute(HTMLTokens.Href);
if (href != null)
{
href.Value = _metaData.Base;
}
}
_emittedMetaData.Add(HTMLTokens.Base);
}
if (tag.NameEquals(HTMLTokens.Meta))
{
ModifyMetaDataAsNecessary(tag);
}
foreach (Attr attr in tag.Attributes)
{
if (attr != null)
{
if (IsScriptAttribute(attr))
{
tag.RemoveAttribute(attr.Name);
}
else
{
attr.Value = ReplaceValue(attr.Value);
}
}
}
Emit(tag.ToString());
base.OnBeginTag(tag);
}