Spaces removed when decoding strings with escaped elements

[mxcl]

I am getting this kind of XML from an ancient SOAP server:

<aResponse>&lt;uesb2b:response xmlns:uesb2b=&quot;http://services.b2b.ues.ut.uhg.com/types/plans/&quot; xmlns=&quot;http://services.b2b.ues.ut.uhg.com/types/plans/&quot;&gt;&#xd;
  &lt;uesb2b:st cd=&quot;GA&quot; /&gt;&#xd;
  &lt;uesb2b:obligId val=&quot;01&quot; /&gt;&#xd;
  &lt;uesb2b:shrArrangementId val=&quot;00&quot; /&gt;&#xd;
  &lt;uesb2b:busInsType val=&quot;CG&quot; /&gt;&#xd;
  &lt;uesb2b:metalPlans typ=&quot;Array&quot;
…

When I decode with a struct like:

struct Response: Decodable {
     let aResponse: String
}

Using:

let decoder = XMLDecoder()
decoder.shouldProcessNamespaces = true
let rsp = try decoder.decode(Response.self, from: data)
print(rsp.aResponse)

I get:

<uesb2b:response xmlns:uesb2b="http://services.b2b.ues.ut.uhg.com/types/plans/"xmlns="http://services.b2b.ues.ut.uhg.com/types/plans/">
<uesb2b:st cd="GA"/>
<uesb2b:obligId val="01"/>
<uesb2b:shrArrangementId val="00"/>
<uesb2b:busInsType val="CG"/>
<uesb2b:metalPlans typ="Array"arrayTyp="metalPlan[110]">
<uesb2b:metalPlan cd="AUWJ"rx="286A"level="S"min="0"max="0"/>
<uesb2b:metalPlan cd="AUWK"rx="286A"level="G"min="0"max="0"/><uesb2b:metalPlan …

(Newlines added for legibility). You can see the spaces on either side of the "s in the attribute heavy nodes are removed (last two lines).

This makes the inner XML invalid.

Happy to fix the bug, just point me at the right code, thanks.

[mxcl]

Ping on a pointer. This is a real problem for us.

[MaxDesiatov]

Hi @mxcl, sorry for the delay. Does setting trimValueWhitespaces to false on XMLDecoder instance resolve the issue for you?

[MaxDesiatov]

I've added a test for this in #137, but it didn't require any changes in the library code, just disabling the trimValueWhitespaces flag. Please let me know if anything's missing.

[mxcl]

I'll check, I apologize that I didn't see this property before.

[mxcl]

Can confirm this fixes it thanks.

Not sure if the feature is working as intended, since it was trimming from the middle of my strings.

[MaxDesiatov]

We trim any string chunk that Foundation's XMLParser passes us in its delegate function, which usually is just whole XML element content. Oddly enough, XMLParser chunks strings on every escaped element, and I agree this is unexpected behavior. Will keep the issue open for a bit until I make it accumulate those chunks and trim whole element content as expected.

[ethan-kusters]

Looks like this is already being addressed, but I just noticed the same issue with single quote values as well.

For example, the value:

was returned as:

What does'being verified' actually mean?

Setting trimValueWhitespaces to false fixed the problem for me as well. Thank you so much for all your work here!