Your data is still not well-formed XML. The only pre-defined entities in XML are <, &, >, " and '. Numerical entities, &#nb; or &#xnb; are also allowed. Everything else needs to be declared.

So you have several options here:

• you can just turn all those entities into their numerical equivalent (for example ≤ is≤); this will be a pain, error-prone, and will not take advantage of XML power (which is a mortal sin in certain circles!),
• you can include all of the entity definitions you need in the document:

  <!DOCTYPE rec [
  <!ENTITY le "&#8804; >
  <!-- many more entity declarations-->
  ]>
  <rec> &le;</rec>
  [download]

  this has the advantage that your documents are standalone (they don't depend on external files) but an empty document might be quite big,
• you can also go all the (XML) way and include the entity declarations through... entities, the way it's done in the HTML DTD:

  <!DOCTYPE rec [
  <!ENTITY % HTMLlat1 PUBLIC
     "-//W3C//ENTITIES Latin 1 for XHTML//EN"
     "xhtml-lat1.ent">
  %HTMLlat1;
  
  <!ENTITY % HTMLsymbol PUBLIC
     "-//W3C//ENTITIES Symbols for XHTML//EN"
     "xhtml-symbol.ent">
  %HTMLsymbol;
  
  <!ENTITY % HTMLspecial PUBLIC
     "-//W3C//ENTITIES Special for XHTML//EN"
     "xhtml-special.ent">
  %HTMLspecial;
  ]>
  <rec> &le;</rec>
  [download]

  you can get the entity declaration files from the W3C.

You could also internally use the entity declaration files and pre-process the XML to convert them to numerical entities. I can't think of an easy way to do this right now (except using xmlwf -p -d result_dir file.xml but then the output is in utf-8) but I'll have a look at it.

Update: fixed typo in doctype

Just to get this 100% clear in my not very XMLized head.

An valid XML file with no includes/inline entity definitions may contain:

• Valid unicode utf-8 characters
• <, &, >, ", '.
• Numerical entities, &#nb; or &#xnb;
and nothing else?

The good news is that I do in fact control the source data file so I can do further mungeing. It looks like the best option is to utf-8 the file including convertig to utf-8 the entities that are not defined. Then, since the characters are in fact all valid latin-1 doing my favourite pack/unpack trick to convert UTF-8 back to latin-1 for the display

sub utf8toNative() {
    my $c = pack("C*",unpack("U*",$_[0]));
    return ((length($c)==length($u))?$_[0]:$c);
[download]

(You have to return the string unchanged if the lengths are the same as new string may be incorrect in such cases)

Dingus

---

Enter any 47-digit prime number to continue.

utf-8-ing everything will indeed save you some headache. That will be playing along with the "XML Way", instead of fighting it. Just to be complete though: you can use an other encoding if you specify it in the xml declaration (<?xml version="1.0" encoding="ISO-8859-1"?>. XML::Parser based modules will nevertheless convert the input to utf-8 before passing it to your code.

The easiest way I have found to resolve the entities, replacing them by the numerical entity, and to drop the now useledd DTD is to use xmllint, from libxml2:

xmllint --noent --dropdtd <file.xml>