bit more alignment stuff

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@686 ec762483-ff6d-05da-a07a-a48fb63a330f

3 files changed, 53 insertions, 0 deletions

diff --git a/decoder/cdec_ff.cc b/decoder/cdec_ff.cc
index 91646253..ca5334e9 100644
--- a/decoder/cdec_ff.cc
+++ b/decoder/cdec_ff.cc
@@ -55,6 +55,7 @@ void register_feature_functions() {
   ff_registry.Register("Tagger_BigramIdentity", new FFFactory<Tagger_BigramIdentity>);
   ff_registry.Register("LexicalPairIdentity", new FFFactory<LexicalPairIdentity>);
   ff_registry.Register("OutputIdentity", new FFFactory<OutputIdentity>);
+  ff_registry.Register("InputIdentity", new FFFactory<InputIdentity>);
   ff_registry.Register("LexicalTranslationTrigger", new FFFactory<LexicalTranslationTrigger>);
 }
 
diff --git a/decoder/ff_wordalign.cc b/decoder/ff_wordalign.cc
index b4981961..f8b8060c 100644
--- a/decoder/ff_wordalign.cc
+++ b/decoder/ff_wordalign.cc
@@ -574,3 +574,39 @@ void BlunsomSynchronousParseHack::TraversalFeaturesImpl(const SentenceMetadata&
   SetStateMask(it->second, it->second + yield.size(), state);
 }
 
+InputIdentity::InputIdentity(const std::string& param) {}
+
+void InputIdentity::FireFeature(WordID src,
+                                SparseVector<double>* features) const {
+  int& fid = fmap_[src];
+  if (!fid) {
+    static map<WordID, WordID> escape;
+    if (escape.empty()) {
+      escape[TD::Convert("=")] = TD::Convert("__EQ");
+      escape[TD::Convert(";")] = TD::Convert("__SC");
+      escape[TD::Convert(",")] = TD::Convert("__CO");
+    }
+    if (escape.count(src)) src = escape[src];
+    ostringstream os;
+    os << "S:" << TD::Convert(src);
+    fid = FD::Convert(os.str());
+  }
+  features->set_value(fid, 1.0);
+}
+
+void InputIdentity::TraversalFeaturesImpl(const SentenceMetadata& smeta,
+                                     const Hypergraph::Edge& edge,
+                                     const std::vector<const void*>& ant_contexts,
+                                     SparseVector<double>* features,
+                                     SparseVector<double>* estimated_features,
+                                     void* context) const {
+  const vector<WordID>& fw = edge.rule_->f_;
+  for (int i = 0; i < fw.size(); ++i) {
+    const WordID& f = fw[i];
+    if (f > 0) FireFeature(f, features);
+  }
+}
+
+
+
+
diff --git a/decoder/ff_wordalign.h b/decoder/ff_wordalign.h
index 0754d70e..30ddf7a1 100644
--- a/decoder/ff_wordalign.h
+++ b/decoder/ff_wordalign.h
@@ -234,4 +234,20 @@ class BlunsomSynchronousParseHack : public FeatureFunction {
   mutable std::vector<std::vector<WordID> > refs_;
 };
 
+class InputIdentity : public FeatureFunction {
+ public:
+  InputIdentity(const std::string& param);
+ protected:
+  virtual void TraversalFeaturesImpl(const SentenceMetadata& smeta,
+                                     const Hypergraph::Edge& edge,
+                                     const std::vector<const void*>& ant_contexts,
+                                     SparseVector<double>* features,
+                                     SparseVector<double>* estimated_features,
+                                     void* context) const;
+ private:
+  void FireFeature(WordID src,
+                   SparseVector<double>* features) const;
+  mutable Class2FID fmap_;
+};
+
 #endif