Detect a change in a rich text field's value in SPItemEventReceiver?

I currently have an Event Receiver that is attached to a custom list. My current requirement is to implement column level security for a Rich Text field (Multiple lines of text with enhanced rich text).

According to this post_[webarchive], I can get the field's before and after values like so:

object oBefore = properties.ListItem[f.InternalName];
object oAfter = properties.AfterProperties[f.InternalName];

The problem is that I'm running to issues comparing these two values, which lead to false positives (code is detecting a change when there wasn't one).

Exhibit A: Using ToString on both objects

oBefore.ToString()

<div class=ExternalClass271E860C95FF42C6902BE21043F01572>
<p class=MsoNormal style="margin:0in 0in 0pt">Text. 
</div>

oAfter.ToString()

<DIV class=ExternalClass271E860C95FF42C6902BE21043F01572>
<P class=MsoNormal style="MARGIN: 0in 0in 0pt">开发者_开发知识库;Text. 
</DIV>

Problems?

• HTML tags are capitalized
• Random spaces (see the additional space after margin:)

Using GetFieldValueForEdit or GetFieldValueAsHTML seem to result in the same values.

"OK," you say, so lets just compare the plain text values.

Exhibit B: Using GetFieldValueAsText

Fortunately, this method strips all of the HTML tags out of the value and only plain text is displayed. However, using this method led me to discover additional issues with whitespace characters:

In the before value:

1. Sometimes there are additional newline characters.
2. Sometimes spaces are displayed as non-breaking spaces (ASCII char code 160)

Question:

How can I detect if the user changed a rich text field in an event receiver?

1. [Ideal] Detect any change to HTML or text or white space
2. [Acceptable] Detect changes to text or white space
3. [Not so good] Detect changes to text characters only (strip all non-alphanumeric characters)

What happens if you set the ListItem field with the new value and read it back out? Does that give the same formatting?

object oBefore = properties.ListItem[f.InternalName];
properties.ListItem[f.InternalName] = properties.AfterProperties[f.InternalName]
object oAfter = properties.ListItem[f.InternalName];
//dont update
properties.ListItem[f.InternalName] = oBefore;

I would probably try something between choices 2 and 3:

bool changed =
    valueAsTextBefore != valueAsTextAfter ||
    0 != string.Compare(
        oBefore.ToString().Replace(" ", ""), 
        oAfter.ToString().Replace(" ", ""), 
        true);

The left half checks if the text (including case) has changed while the right half checks if the tags or attributes have changed. Very kludgy, but should fit your case.

The only other thing I can think of is to run an XML transform on the HTML in order to standardize on case and spacing. But not only does that seem like overkill, but it assumes the HTML will always be well formed.

I'm currently testing a combination approach: GetFieldValueAsText and then stripping out all characters except alphanumeric/punctuation:

static string GetRichTextValue(string value)
{
    if (null == value)
    {
         return string.Empty;
    }
    StringBuilder sb = new StringBuilder(value.Length);
    foreach (char c in value)
    {
         if (char.IsLetterOrDigit(c) || char.IsPunctuation(c))
         {
              sb.Append(c);
         }
    }
    return sb.ToString();
}

This only detects changes to the text of a rich text field but seems to work consistently.