// Converts a SyntaxNode tree to a Moses::GHKM::ParseTree.
std::auto_ptr<ParseTree> XmlTreeParser::ConvertTree(
const SyntaxNode &tree,
const std::vector<std::string> &words)
{
std::auto_ptr<ParseTree> root(new ParseTree(tree.GetLabel()));
const std::vector<SyntaxNode*> &children = tree.GetChildren();
if (children.empty()) {
if (tree.GetStart() != tree.GetEnd()) {
std::ostringstream msg;
msg << "leaf node covers multiple words (" << tree.GetStart()
<< "-" << tree.GetEnd() << "): this is currently unsupported";
throw Exception(msg.str());
}
std::auto_ptr<ParseTree> leaf(new ParseTree(words[tree.GetStart()]));
leaf->SetParent(root.get());
root->AddChild(leaf.release());
} else {
for (std::vector<SyntaxNode*>::const_iterator p = children.begin();
p != children.end(); ++p) {
assert(*p);
std::auto_ptr<ParseTree> child = ConvertTree(**p, words);
child->SetParent(root.get());
root->AddChild(child.release());
}
}
return root;
}
void SyntaxTree::ConnectNodes()
{
typedef SyntaxTreeIndex2::const_reverse_iterator InnerIterator;
SyntaxNode *prev = 0;
// Iterate over all start indices from lowest to highest.
for (SyntaxTreeIndexIterator p = m_index.begin(); p != m_index.end(); ++p) {
const SyntaxTreeIndex2 &inner = p->second;
// Iterate over all end indices from highest to lowest.
for (InnerIterator q = inner.rbegin(); q != inner.rend(); ++q) {
const std::vector<SyntaxNode*> &nodes = q->second;
// Iterate over all nodes that cover the same span in order of tree
// depth, top-most first.
for (std::vector<SyntaxNode*>::const_reverse_iterator r = nodes.rbegin();
r != nodes.rend(); ++r) {
SyntaxNode *node = *r;
if (!prev) {
// node is the root.
m_top = node;
node->SetParent(0);
} else if (prev->GetStart() == node->GetStart()) {
// prev is the parent of node.
assert(prev->GetEnd() >= node->GetEnd());
node->SetParent(prev);
prev->AddChild(node);
} else {
// prev is a descendant of node's parent. The lowest common
// ancestor of prev and node will be node's parent.
SyntaxNode *ancestor = prev->GetParent();
while (ancestor->GetEnd() < node->GetEnd()) {
ancestor = ancestor->GetParent();
}
assert(ancestor);
node->SetParent(ancestor);
ancestor->AddChild(node);
}
prev = node;
}
}
}
}