Target-side only output format

3 files changed, 33 insertions, 0 deletions

diff --git a/decoder/decoder.cc b/decoder/decoder.cc
index 3b53fd6b..3394e0b8 100644
--- a/decoder/decoder.cc
+++ b/decoder/decoder.cc
@@ -408,6 +408,7 @@ DecoderImpl::DecoderImpl(po::variables_map& conf, int argc, char** argv, istream
         ("show_partition,z", "Compute and show the partition (inside score)")
         ("show_conditional_prob", "Output the conditional log prob to STDOUT instead of a translation")
         ("show_cfg_search_space", "Show the search space as a CFG")
+        ("show_target_graph", "Output the target hypergraph")
         ("coarse_to_fine_beam_prune", po::value<double>(), "Prune paths from coarse parse forest before fine parse, keeping paths within exp(alpha>=0)")
         ("ctf_beam_widen", po::value<double>()->default_value(2.0), "Expand coarse pass beam by this factor if no fine parse is found")
         ("ctf_num_widenings", po::value<int>()->default_value(2), "Widen coarse beam this many times before backing off to full parse")
@@ -1017,6 +1018,8 @@ bool DecoderImpl::Decode(const string& input, DecoderObserver* o) {
   }
   if (conf.count("show_cfg_search_space"))
     HypergraphIO::WriteAsCFG(forest);
+  if (conf.count("show_target_graph"))
+    HypergraphIO::WriteTarget(forest);
   if (has_ref) {
     if (HG::Intersect(ref, &forest)) {
 //      if (crf_uniform_empirical) {
diff --git a/decoder/hg_io.cc b/decoder/hg_io.cc
index c1c93933..0283ec3c 100644
--- a/decoder/hg_io.cc
+++ b/decoder/hg_io.cc
@@ -624,3 +624,30 @@ void HypergraphIO::WriteAsCFG(const Hypergraph& hg) {
   }
 }
 
+/* Output format:
+ * #vertices
+ * for each vertex in bottom-up topological order:
+ *   #downward_edges
+ *   for each downward edge:
+ *     RHS with [vertex_index] for NTs ||| scores
+ */
+void HypergraphIO::WriteTarget(const Hypergraph& hg) {
+  cout << hg.nodes_.size() << '\n';
+  for (unsigned int i = 0; i < hg.nodes_.size(); ++i) {
+    const Hypergraph::EdgesVector &edges = hg.nodes_[i].in_edges_;
+    cout << edges.size() << '\n';
+    for (unsigned int j = 0; j < edges.size(); ++j) {
+      const Hypergraph::Edge &edge = hg.edges_[edges[j]];
+      const std::vector<WordID> &e = edge.rule_->e();
+      for (std::vector<WordID>::const_iterator word = e.begin(); word != e.end(); ++word) {
+        if (*word <= 0) {
+          cout << '[' << edge.tail_nodes_[-*word] << "] ";
+        } else {
+          cout << TD::Convert(*word) << ' ';
+        }
+      }
+      cout << "||| " << edge.rule_->scores_ << '\n';
+    }
+  }
+}
+
diff --git a/decoder/hg_io.h b/decoder/hg_io.h
index 082489d8..44817157 100644
--- a/decoder/hg_io.h
+++ b/decoder/hg_io.h
@@ -23,6 +23,9 @@ struct HypergraphIO {
 
   static void WriteAsCFG(const Hypergraph& hg);
 
+  // Write only the target size information in bottom-up order.  
+  static void WriteTarget(const Hypergraph& hg);
+
   // serialization utils
   static void ReadFromPLF(const std::string& in, Hypergraph* out, int line = 0);
   // return PLF string representation (undefined behavior on non-lattices)