void sort(const DOM::Node<std::string>& node, Arabica::XPath::NodeSet<std::string>& nodes, ExecutionContext& context) const
{
if(!sort_)
{
if(!nodes.forward())
nodes.to_document_order();
return;
}
sort_->set_context(node, context);
std::stable_sort(nodes.begin(), nodes.end(), SortP(*sort_));
} // sort
Arabica::XPath::NodeSet<std::string> nodeSetUnion(const std::string& keyClarkName,
const Arabica::XPath::NodeSet<std::string> nodes,
const Arabica::XPath::ExecutionContext<std::string>& executionContext) const
{
Arabica::XPath::NodeSet<std::string> results;
for(Arabica::XPath::NodeSet<std::string>::const_iterator n = nodes.begin(), ne = nodes.end(); n != ne; ++n)
{
std::string id = Arabica::XPath::impl::nodeStringValue<std::string, Arabica::default_string_adaptor<std::string> >(*n);
results.push_back(keys_.lookup(keyClarkName, id, executionContext));
} // for ...
results.to_document_order();
return results;
} // nodeSetUnion
NistXmlTestset::NistXmlTestset(const std::string &file)
: logger_(logkw::channel = "NistXmlTestset") {
Arabica::SAX2DOM::Parser<std::string> domParser;
Arabica::SAX::InputSource<std::string> is(file);
Arabica::SAX::CatchErrorHandler<std::string> errh;
domParser.setErrorHandler(errh);
domParser.parse(is);
if(errh.errorsReported())
BOOST_LOG_SEV(logger_, error) << errh.errors();
Arabica::DOM::Document<std::string> doc = domParser.getDocument();
if(doc == 0) {
BOOST_LOG_SEV(logger_, error) << "Error parsing input file: " << file;
exit(1);
}
doc.getDocumentElement().normalize();
Arabica::XPath::XPath<std::string> xp;
Arabica::XPath::NodeSet<std::string> docnodes =
xp.compile("/mteval/srcset/doc").evaluateAsNodeSet(doc.getDocumentElement());
docnodes.to_document_order();
BOOST_FOREACH(Arabica::DOM::Node<std::string> n, docnodes)
documents_.push_back(boost::make_shared<NistXmlDocument>(n));
outdoc_ = static_cast<Arabica::DOM::Document<std::string> >(doc.cloneNode(true));
Arabica::DOM::Element<std::string> srcset =
static_cast<Arabica::DOM::Element<std::string> >(
xp.compile("/mteval/srcset").evaluateAsNodeSet(outdoc_.getDocumentElement())[0]);
Arabica::DOM::Element<std::string> tstset = outdoc_.createElement("tstset");
int docno = 0;
while(srcset.hasChildNodes()) {
Arabica::DOM::Node<std::string> n = srcset.removeChild(srcset.getFirstChild());
tstset.appendChild(n);
if(n.getNodeType() == Arabica::DOM::Node<std::string>::ELEMENT_NODE &&
n.getNodeName() == "doc")
documents_[docno++]->setOutputNode(n);
}
tstset.setAttribute("setid", srcset.getAttribute("setid"));
tstset.setAttribute("srclang", srcset.getAttribute("srclang"));
tstset.setAttribute("trglang", "TRGLANG");
tstset.setAttribute("sysid", "SYSID");
srcset.getParentNode().replaceChild(tstset, srcset);
}