track spans in t2s translation

3 files changed, 26 insertions, 2 deletions

diff --git a/decoder/tree2string_translator.cc b/decoder/tree2string_translator.cc
index c9c91a37..adc8dc89 100644
--- a/decoder/tree2string_translator.cc
+++ b/decoder/tree2string_translator.cc
@@ -332,7 +332,9 @@ struct Tree2StringTranslatorImpl {
             assert(tail.size() == r->Arity());
             HG::Edge* new_edge = hg.AddEdge(r, tail);
             new_edge->feature_values_ = r->GetFeatureValues();
-            // TODO: set i and j
+            auto& inspan = input_tree.nodes[s.task.input_node_idx].span;
+            new_edge->i_ = inspan.first;
+            new_edge->j_ = inspan.second;
             hg.ConnectEdgeToHeadNode(new_edge, &hg.nodes_[node_id]);
           }
           for (const auto& n : s.future_work) {
diff --git a/decoder/tree_fragment.cc b/decoder/tree_fragment.cc
index aad0b2c4..42f7793a 100644
--- a/decoder/tree_fragment.cc
+++ b/decoder/tree_fragment.cc
@@ -28,12 +28,14 @@ TreeFragment::TreeFragment(const StringPiece& tree, bool allow_frontier_sites) {
   unsigned cp = 0, symp = 0, np = 0;
   ParseRec(tree, allow_frontier_sites, cp, symp, np, &cp, &symp, &np);
   root = nodes.back().lhs;
+  if (!allow_frontier_sites) SetupSpansRec(open - 1, 0);
   //cerr << "ROOT: " << TD::Convert(root & ALL_MASK) << endl;
   //DebugRec(open - 1, &cerr); cerr << "\n";
 }
 
 void TreeFragment::DebugRec(unsigned cur, ostream* out) const {
   *out << '(' << TD::Convert(nodes[cur].lhs & ALL_MASK);
+  // *out << "_{" << nodes[cur].span.first << ',' << nodes[cur].span.second << '}';
   for (auto& x : nodes[cur].rhs) {
     *out << ' ';
     if (IsFrontier(x)) {
@@ -47,6 +49,21 @@ void TreeFragment::DebugRec(unsigned cur, ostream* out) const {
   *out << ')';
 }
 
+// returns left + the number of terminals rooted at NT cur,
+int TreeFragment::SetupSpansRec(unsigned cur, int left) {
+  int right = left;
+  for (auto& x : nodes[cur].rhs) {
+    if (IsRHS(x)) {
+      right = SetupSpansRec(x & ALL_MASK, right);
+    } else {
+      ++right;
+    }
+  }
+  nodes[cur].span.first = left;
+  nodes[cur].span.second = right;
+  return right;
+}
+
 // cp is the character index in the tree
 // np keeps track of the nodes (nonterminals) that have been built
 // symp keeps track of the terminal symbols that have been built
diff --git a/decoder/tree_fragment.h b/decoder/tree_fragment.h
index 8bb7251a..6b4842ee 100644
--- a/decoder/tree_fragment.h
+++ b/decoder/tree_fragment.h
@@ -43,9 +43,10 @@ inline bool IsTerminal(unsigned x) {
 
 struct TreeFragmentProduction {
   TreeFragmentProduction() {}
-  TreeFragmentProduction(int nttype, const std::vector<unsigned>& r) : lhs(nttype), rhs(r) {}
+  TreeFragmentProduction(int nttype, const std::vector<unsigned>& r) : lhs(nttype), rhs(r), span(std::make_pair<short,short>(-1,-1)) {}
   unsigned lhs;
   std::vector<unsigned> rhs;
+  std::pair<short, short> span; // the span of the node (in input, or not set for rules)
 };
 
 // this data structure represents a tree or forest
@@ -76,6 +77,10 @@ class TreeFragment {
   // np keeps track of the nodes (nonterminals) that have been built
   // symp keeps track of the terminal symbols that have been built
   void ParseRec(const StringPiece& tree, bool afs, unsigned cp, unsigned symp, unsigned np, unsigned* pcp, unsigned* psymp, unsigned* pnp);
+
+  // used by constructor to set up span indices for logging/alignment purposes
+  int SetupSpansRec(unsigned cur, int left);
+
  public:
   unsigned root;
   unsigned char frontier_sites;