Added psShouldSkipAttributeReferenceParsing to ParseSettings

[jaunruh]

Referencing: #177
Instead of replacing the removed characters in the render method using the RenderSettings I adapted the ParseSettings to allow skipping of reference rendering in attributes.

[k0ral]

Is there a reason why you chose to work around #177 at parser-level, rather than at renderer-level as suggested in that issue ?

Indeed, my understanding is that the present change violates the standard (XML references aren't resolved in attribute values), and I'm under the impression that a workaround at renderer-level would be equally simple yet still standard compliant.

Besides, it seems to me that this change, as is, will parse newline references into ContentText "
", which then will get rendered into 
, which isn't even what you're looking for, or am I mistaken ?

[jaunruh]

Honestly didn't think about that. But indeed, you are correct:

Text.XML.renderLBS def $ Text.XML.parseLBS_ customParse $ Data.ByteString.Lazy.Char8.pack "<someElement id=\"1\" name=\"Picture 1\" descr=\"A picture containing window, indoor&#xA;&#xA;Description automatically generated\"/>"
"<?xml version=\"1.0\" encoding=\"UTF-8\"?><someElement descr=\"A picture containing window, indoor&amp;#xA;&amp;#xA;Description automatically generated\" id=\"1\" name=\"Picture 1\"/>"

My reasoning was that changing the parsing-level instead render-level would make sure all references are retained in the standard case. With rsEncodeCharacters :: Set Char in RenderSettings like you suggested in #177 one would always have to supply the Set Char in order to receive the original input. So basically "why break something to repair it again".

Could you maybe refer me to the XML reference that you are referring to, please?

So I guess I am going to implement the render-level suggestion from #177. You can either close this pull request and let me open a new one or we rename this one.

[k0ral]

Could you maybe refer me to the XML reference that you are referring to, please?

The relevant parts of the standard I was referring to are:

• https://www.w3.org/TR/xml/#NT-AttValue

  AttValue    ::=    '"' ([^<&"] | Reference)* '"'    |  "'" ([^<&'] | Reference)* "'"
  

• https://www.w3.org/TR/xml/#dt-entref

  Reference    ::=    EntityRef | CharRef 
  

• https://www.w3.org/TR/xml/#NT-CharRef

  CharRef    ::=    '&#' [0-9]+ ';'    | '&#x' [0-9a-fA-F]+ ';'
  

My understanding is that this essentially means &#xA must be interpreted as a character reference for the newline character.

You can either close this pull request and let me open a new one or we rename this one.

I'd rather close this pull-request, you're welcome to open a new one for an implementation in the rendering logic :) .