HtmlAgilityPack Drops Option End Tags

Question

I am using HtmlAgilityPack. I create an HtmlDocument and LoadHtml with the following string:

<select id="foo_Bar" name="foo.Bar"><option selected="selected" value="1">One</option><option value="2">Two</option></select> 

This does some unexpected things. First, it gives two parser errors, EndTagNotRequired. Second, the select node has 4 children - two for the option tags and two more for the inner text of the option tags. Last, the OuterHtml is like this:

<select id="foo_Bar" name="foo.Bar"><option selected="selected" value="1">One<option value="2">Two</select> 

So basically it is deciding for me to drop the closing tags on the options. Let's leave aside for a moment whether it is proper and desirable to do that. I am using HtmlAgilityPack to test HTML generation code, so I don't want it to make any decision for me or give any errors unless the HTML is truly malformed. Is there some way to make it behave how I want? I tried setting some of the options for HtmlDocument, specifically:

 doc.OptionAutoCloseOnEnd = false;  doc.OptionCheckSyntax = false;  doc.OptionFixNestedTags = false; 

This is not working. If HtmlAgilityPack cannot do what I want, can you recommend something that can?

Answer 1

It seems that there is some reason not to parse the Option tag as a "generic" tag, for XHTML compliance, however this can be a real pain in the neck.

My suggestion is to do a whole-string-replace and change all "option" tags to "my_option" tags, that way you:

1. Don't have to modify the source of the library (and can upgrade it later).
2. Can parse as you usually would.

The original post on HtmlAgilityPack forum can be found at: http://htmlagilitypack.codeplex.com/Thread/View.aspx?ThreadId=14982

Answer 2

The exact same error is reported on the HAP home page's discussion, but it looks like no meaningful fixes have been made to the project in a few years. Not encouraging.

A quick browse of the source suggests the error might be fixable by commenting out line 92 of HtmlNode.cs:

// they sometimes contain, and sometimes they don 't... ElementsFlags.Add("option", HtmlElementFlag.Empty); 

(Actually no, they always contain label text, although a blank string would also be valid text. A careless author might omit the end-tag, but then that's true of any element.)

ADD

An equivalent solution is calling HtmlNode.ElementsFlags.Remove("option"); before any use of liberary (without need to modify the liberary source code)