Fails to parse JMDict which has a large !DOCTYPE section

[udoprog]

File can be downloaded from here from here: https://www.edrdg.org/jmdict/j_jmdict.html.

I'm not entirely sure what's going on, a large chunk of the DOCTYPE section is processed, then it stops in the middle and emits the following error:

Note that I've added the byte offset:

Only Comment (`--`), CDATA (`[CDATA[`) and DOCTYPE (`DOCTYPE`) nodes can start with a '!', but symbol `B` found (at 8466)

This is the following element (inside of the !DOCTYPE section):

<!ELEMENT field (#PCDATA)>
        <!-- Information about the field of application of the entry/sense.
        When absent, general application is implied. Entity coding for
        specific fields of application. -->

[Mingun]

Likely duplicate of #590