Malformed characters in contents

[darkgeek]

There are many malformed characters in lookup results, especially <pr></pr> blocks, just like this:

In fact, its entity name is aitalic_, but unfortunately, it's mapped to &#xFFFD, which is not a valid Unicode character.
A possible solution is to digging into the official gcide repo (http://git.savannah.gnu.org/r/gcide.git), find the respective word definition in CIDE.* files, get the symbol name (in the previous example, aitalic_ has the symbol name <ait/), and then find its real character (in this case, 𝘢) in symbols.png.