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);
}
int main(int argc, char** argv) {
factory_ = Arabica::SimpleDOM::DOMImplementation<string_type, string_adaptor>::getDOMImplementation();
document_ = factory_.createDocument(SA::construct_from_utf8(""), SA::construct_from_utf8(""), 0);
root_ = document_.createElement("root");
document_.appendChild(root_);
assert(root_);
element1_ = document_.createElement(SA::construct_from_utf8("child1"));
element2_ = document_.createElement(SA::construct_from_utf8("child2"));
element3_ = document_.createElement(SA::construct_from_utf8("child3"));
element1_.setAttribute(SA::construct_from_utf8("one"), SA::construct_from_utf8("1"));
element2_.setAttribute(SA::construct_from_utf8("one"), SA::construct_from_utf8("1"));
element2_.setAttribute(SA::construct_from_utf8("two"), SA::construct_from_utf8("1"));
element2_.setAttribute(SA::construct_from_utf8("three"), SA::construct_from_utf8("1"));
element2_.setAttribute(SA::construct_from_utf8("four"), SA::construct_from_utf8("1"));
text_ = document_.createTextNode(SA::construct_from_utf8("data"));
comment_ = document_.createComment(SA::construct_from_utf8("comment"));
processingInstruction_ = document_.createProcessingInstruction(SA::construct_from_utf8("target"), SA::construct_from_utf8("data"));
element2_.appendChild(text_);
spinkle_ = document_.createElement(SA::construct_from_utf8("spinkle"));
element2_.appendChild(spinkle_);
element2_.appendChild(comment_);
element2_.appendChild(processingInstruction_);
attr_ = element1_.getAttributeNode(SA::construct_from_utf8("one"));
root_.appendChild(element1_);
root_.appendChild(element2_);
root_.appendChild(element3_);
chapters_ = factory_.createDocument(SA::construct_from_utf8(""), SA::construct_from_utf8(""), 0);
chapters_.appendChild(chapters_.createElement(SA::construct_from_utf8("document")));
chapters_.getFirstChild().appendChild(chapters_.createElement(SA::construct_from_utf8("chapter"))).appendChild(chapters_.createTextNode(SA::construct_from_utf8("one")));
chapters_.getFirstChild().appendChild(chapters_.createElement(SA::construct_from_utf8("chapter"))).appendChild(chapters_.createTextNode(SA::construct_from_utf8("two")));
chapters_.getFirstChild().appendChild(chapters_.createElement(SA::construct_from_utf8("chapter"))).appendChild(chapters_.createTextNode(SA::construct_from_utf8("three")));
chapters_.getFirstChild().appendChild(chapters_.createElement(SA::construct_from_utf8("chapter"))).appendChild(chapters_.createTextNode(SA::construct_from_utf8("four")));
chapters_.getFirstChild().appendChild(chapters_.createElement(SA::construct_from_utf8("chapter"))).appendChild(chapters_.createTextNode(SA::construct_from_utf8("five")));
numbers_ = factory_.createDocument(SA::construct_from_utf8(""), SA::construct_from_utf8(""), 0);
numbers_.appendChild(numbers_.createElement(SA::construct_from_utf8("doc")));
numbers_.getFirstChild().appendChild(numbers_.createElement(SA::construct_from_utf8("number"))).appendChild(numbers_.createTextNode(SA::construct_from_utf8("1")));
numbers_.getFirstChild().appendChild(numbers_.createElement(SA::construct_from_utf8("number"))).appendChild(numbers_.createTextNode(SA::construct_from_utf8("2")));
numbers_.getFirstChild().appendChild(numbers_.createElement(SA::construct_from_utf8("number"))).appendChild(numbers_.createTextNode(SA::construct_from_utf8("3")));
numbers_.getFirstChild().appendChild(numbers_.createElement(SA::construct_from_utf8("number"))).appendChild(numbers_.createTextNode(SA::construct_from_utf8("4")));
numbers_.getFirstChild().appendChild(numbers_.createElement(SA::construct_from_utf8("number"))).appendChild(numbers_.createTextNode(SA::construct_from_utf8("5")));
numbers_.getFirstChild().appendChild(numbers_.createElement(SA::construct_from_utf8("number"))).appendChild(numbers_.createTextNode(SA::construct_from_utf8("6")));
numbers_.getFirstChild().appendChild(numbers_.createElement(SA::construct_from_utf8("number"))).appendChild(numbers_.createTextNode(SA::construct_from_utf8("7")));
numbers_.getFirstChild().appendChild(numbers_.createElement(SA::construct_from_utf8("number"))).appendChild(numbers_.createTextNode(SA::construct_from_utf8("8")));
numbers_.getFirstChild().appendChild(numbers_.createElement(SA::construct_from_utf8("number"))).appendChild(numbers_.createTextNode(SA::construct_from_utf8("9")));
std::cout << document_ << std::endl;
std::cout << numbers_ << std::endl;
std::cout << chapters_ << std::endl;
if (true) {
using namespace Arabica::XPath;
using namespace Arabica::DOM;
XPathValue<string_type, string_adaptor> result = parser.evaluate(SA::construct_from_utf8("//*"), document_);
for(int i = 0; i < result.asNodeSet().size(); i++) {
Node<string_type, string_adaptor> node = result.asNodeSet()[i];
std::string xpathExpr = uscxml::DOMUtils::xPathForNode(node);
if (xpathExpr.size()) {
XPathValue<string_type, string_adaptor> innerResult = parser.evaluate(xpathExpr, document_);
assert(innerResult.asNodeSet().size() > 0);
assert(innerResult.asNodeSet().size() == 1);
assert(innerResult.asNodeSet()[0] == node);
} else {
assert(node.getNodeType() != Node_base::ELEMENT_NODE);
}
}
}
if (false) {
using namespace Arabica::XPath;
StringVariableResolver svr;
svr.setVariable(SA::construct_from_utf8("index"), SA::construct_from_utf8("1"));
parser.setVariableResolver(svr);
XPathValue<string_type, string_adaptor> result = parser.evaluate(SA::construct_from_utf8("/root/*[@two = $index]"), document_);
assert(NODE_SET == result.type());
assert(element2_ == result.asNodeSet()[0]);
parser.resetVariableResolver();
} // test18
if (false) {
using namespace Arabica::XPath;
XPathExpression<string_type, string_adaptor> xpath = parser.compile(SA::construct_from_utf8("root/*[position() = 2]"));
XPathValue<string_type, string_adaptor> result = xpath.evaluate(document_);
assert(NODE_SET == result.type());
assert(1 == result.asNodeSet().size());
Arabica::DOM::Node<string_type, string_adaptor> n = result.asNodeSet()[0];
assert(element2_ == n);
} // test19
if (false) {
using namespace Arabica::XPath;
Arabica::DOM::DocumentFragment<string_type, string_adaptor> frag = document_.createDocumentFragment();
frag.appendChild(document_.createElement(SA::construct_from_utf8("foo")));
NodeSetVariableResolver svr;
NodeSet<string_type, string_adaptor> ns;
ns.push_back(frag);
svr.setVariable(SA::construct_from_utf8("fruit"), ns);
parser.setVariableResolver(svr);
XPathValue<string_type, string_adaptor> result = parser.evaluate_expr(SA::construct_from_utf8("$fruit/foo|/root/child3"), document_);
assert(NODE_SET == result.type());
assert(2 == result.asNodeSet().size());
assert(element3_ == result.asNodeSet()[0]);
} // testUnion11
if (false) {
using namespace Arabica::XPath;
XPathValue<string_type, string_adaptor> result = parser.evaluate_expr(SA::construct_from_utf8("local-name(/root)"), document_);
assert(STRING == result.type());
assert(SA::construct_from_utf8("root") == result.asString());
} // testLocalNameFn1
if (0) {
using namespace Arabica::XPath;
Arabica::DOM::DocumentFragment<std::string> frag = document_.createDocumentFragment();
frag.appendChild(document_.createElement("foo"));
NodeSetVariableResolver svr;
NodeSet<string_type, string_adaptor> ns;
ns.push_back(frag);
svr.setVariable("fruit", ns);
parser.setVariableResolver(svr);
XPathValue<string_type, string_adaptor> result = parser.evaluate(SA::construct_from_utf8("local-name($fruit/foo) == 'foo'"), document_);
std::cout << result.asBool() << std::endl;
}
}
Copy_base()
{
Arabica::XPath::XPath<std::string> compiler;
namespace_select_ = compiler.compile("namespace::node()");
} // Copy_base