Markdown formatting returned by API can't be reliably decoded

I am trying to read formatted text from a rich text field via API. However, sometimes different text may result in the same Markdown representation.

For example, each of the following four text snippets:

~a~
~a~
~a~
~a~

is represented in an API response as ~~~a~~~.

Is it possible to get the contents of a rich text field that can be decoded unambiguously?