Which Epi version are you using and how are you getting the XhtmlString property value that you use in the API result?

We are using episerver CMS 8.2

And this is the code I am using to pull the Xhtmlstring value

message =
MemoryCacheManager.Instance.Retrieve<dynamic>(CacheKeys.ConsentPageHtml)
.[ContentBlockName].ToString()
.Replace("\n", "")
.Replace(baseAddress, "..")
.Replace("60",
[ValuetoReplacePartofTheStringWith].ToString(CultureInfo.InvariantCulture));

To give more clarity in to what I am seeing, sometimes, the value to replace in the string will be "60" so I will replace "60" in the string with "60". In this situation the link works fine. But sometimes the [ValuetoReplacePartofTheStringWith] is "59" or "57" or something, so "60" gets replaced with "59". In this case the link is broken.

Here is an example of the strings"

Content Block in CMS, un manipulated:

"You are 60 days from x, please go to [link] to find out more"

Manipulated string with a replace value of "60":

"You are 60 days from x, please go to [link] to find out more" - Link works fine

Manipulated string with a replace value of "59":

"You are 59 dats from x, please go to [link] to find out more" - Link is broken

It seems if the content around the link is in any way different then the link becomes broken.

I've never seen people fetching HTML off a block in that manor (using CacheManager).

If you instead use the defined property you can do ToHtmlString() on it and then the URLs will be rewritten and consistent.

For example:

message = yourBlock.YourXhtmlStringProp.ToHtmlString().Replace(etc..)