really fixed binarization. test

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@555 ec762483-ff6d-05da-a07a-a48fb63a330f

1 files changed, 24 insertions, 14 deletions

diff --git a/decoder/cfg_options.h b/decoder/cfg_options.h
index 5dca168e..331363d2 100755
--- a/decoder/cfg_options.h
+++ b/decoder/cfg_options.h
@@ -10,11 +10,14 @@ struct CFGOptions {
   CFGFormat format;
   CFGBinarize binarize;
   std::string out,source_out,unbin_out;
+  bool uniq;
   void set_defaults() {
     format.set_defaults();
     binarize.set_defaults();
-    out="";
+    out=source_out=unbin_out="";
+    uniq=false;
   }
+
   CFGOptions() { set_defaults(); }
   template <class Opts> // template to support both printable_opts and boost nonprintable
   void AddOptions(Opts *opts) {
@@ -22,6 +25,8 @@ struct CFGOptions {
       ("cfg_output", defaulted_value(&out),"write final target CFG (before FSA rescoring) to this file")
       ("source_cfg_output", defaulted_value(&source_out),"write source CFG (after prelm-scoring, prelm-prune) to this file")
       ("cfg_unbin_output", defaulted_value(&unbin_out),"write pre-binarization CFG to this file") //TODO:
+      ("cfg_uniq", defaulted_value(&uniq),"in case of duplicate rules, keep only the one with highest prob")
+
     ;
     binarize.AddOptions(opts);
     format.AddOptions(opts);
@@ -29,10 +34,6 @@ struct CFGOptions {
   void Validate() {
     format.Validate();
     binarize.Validate();
-//    if (out.empty()) binarize.bin_name_nts=false;
-  }
-  char const* description() const {
-    return "CFG output options";
   }
   void maybe_output_source(Hypergraph const& hg) {
     if (source_out.empty()) return;
@@ -41,24 +42,33 @@ struct CFGOptions {
     CFG cfg(hg,false,format.features,format.goal_nt());
     cfg.Print(o.get(),format);
   }
-  void maybe_print(CFG &cfg,std::string cfg_output,char const* desc=" unbinarized") {
-      WriteFile o(cfg_output);
-      std::cerr<<"Printing target"<<desc<<" CFG to "<<cfg_output<<": "<<format<<'\n';
-      cfg.Print(o.get(),format);
-  }
-
-  void maybe_output(HgCFG &hgcfg) {
+  // executes all options except source_cfg_output, building target hgcfg
+  void prepare(HgCFG &hgcfg) {
     if (out.empty() && unbin_out.empty()) return;
     CFG &cfg=hgcfg.GetCFG();
     maybe_print(cfg,unbin_out);
+    maybe_uniq(hgcfg);
     maybe_binarize(hgcfg);
     maybe_print(cfg,out,"");
   }
 
+  char const* description() const {
+    return "CFG output options";
+  }
+  void maybe_print(CFG &cfg,std::string cfg_output,char const* desc=" unbinarized") {
+    WriteFile o(cfg_output);
+    std::cerr<<"Printing target"<<desc<<" CFG to "<<cfg_output<<": "<<format<<'\n';
+    cfg.Print(o.get(),format);
+  }
+
+  void maybe_uniq(HgCFG &hgcfg) {
+    if (hgcfg.uniqed) return;
+    hgcfg.GetCFG().UniqRules();
+    hgcfg.uniqed=true;
+  }
   void maybe_binarize(HgCFG &hgcfg) {
     if (hgcfg.binarized) return;
-    CFG &cfg=hgcfg.GetCFG();
-    cfg.Binarize(binarize);
+    hgcfg.GetCFG().Binarize(binarize);
     hgcfg.binarized=true;
   }
 };