add ElementDecls to an internal subset

[flavorjones]

The use case is transforming this XML:

<?xml version="1.0"?>
<!DOCTYPE arbitrary SYSTEM "https://127.0.0.1/some.dtd" [

    <!ENTITY test "Dereferenced!">

]>
<arbitrary>
  <child>&test;</child>
</arbitrary>

into this XML (note the additional element definition within the DTD):

<?xml version="1.0"?>
<!DOCTYPE arbitrary SYSTEM "https://127.0.0.1/some.dtd" [

    <!ENTITY test "Dereferenced!">
    <!ELEMENT new_element ANY>

]>
<arbitrary>
  <child>&test;</child>
</arbitrary>

Nokogiri doesn't have any functionality to create or manipulate element declarations. The ElementDecl class is pretty much read-only, meaning that if your XML has one, you can parse and inspect it.

See original nokogiri-talk thread for more details.

[tisba]

Pardon me, if this makes no sense, but this would also allow something like this?

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE mydoc SYSTEM "mydoc.dtd" [
  <!ENTITY foo SYSTEM "./foo.xml">
]>
<root>&foo;</root>
]>

I'm having a hard time to understand how to get this using Nokogiri::XML::Builder 😢

[flavorjones]

Similarly for AttributeDecl ... Nokogiri offers no way to dynamically modify the document DTD to add these declarations.

Side note: libxml2 does not support adding EntityDecl to the DTD at runtime.

As a workaround for anyone who comes across this limitation: a way to work around this, suggested in this SO post is to create a new doc by parsing the desired DTD, and then build the desired DOM tree within that document.

[flavorjones]

libxml2 API links:

• https://gnome.pages.gitlab.gnome.org/libxml2/devhelp/libxml2-valid.html#xmlAddElementDecl
• https://gnome.pages.gitlab.gnome.org/libxml2/devhelp/libxml2-valid.html#xmlAddAttributeDecl

[flavorjones]

Interesting side note: the JRuby implementation's parser, Xerces, does not support modifying the doctype!

https://xerces.apache.org/xerces-j/apiDocs/org/w3c/dom/Document.html#getDoctype()

So this is a feature that can only be implemented in CRuby, as far as I know.