LO HTML to PDF conversion, character ending with ‍ not converted properly

We are using LO 5.4 for converting HTML to PDF via UNO API. It works fine for most languages but one particular language i.e. Malayalam has some characters (called chillu) which had a representation ending with ‍ (Zero Width Joiner).

  1. When we convert these characters where we apply Malayalam font-family explicitly then the conversion is fine.
    e.g. <span style="font-family:Meera">ന്&zwj;</span>
Here the character is converted as expected i.e  **ന്‍**
  1. But when we the content has no font-family and inherits the base family from body which is English (e.g. Verdana) then the character is not converted properly.

e.g. <span>ന്&zwj;</span>

In this case the output is **ന്** instead of **ന്‍** eventhough the font embedded to display this word is same one i.e. **Meera**

Can somebody throw some light on how this work.