From 32154b45828f05add1db7c89752ef4220c0fdf16 Mon Sep 17 00:00:00 2001
From: "graehl@gmail.com"
 <graehl@gmail.com@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Tue, 10 Aug 2010 10:02:04 +0000
Subject: cdec --cfg_output=-

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@499 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 decoder/cfg_format.h | 74 +++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 73 insertions(+), 1 deletion(-)

(limited to 'decoder/cfg_format.h')

diff --git a/decoder/cfg_format.h b/decoder/cfg_format.h
index 1bce3d06..169632a6 100755
--- a/decoder/cfg_format.h
+++ b/decoder/cfg_format.h
@@ -3,9 +3,19 @@
 
 #include <program_options.h>
 #include <string>
+#include "wordid.h"
+#include "feature_vector.h"
 
 struct CFGFormat {
-  bool identity_scfg;bool features;bool logprob_feat;bool cfg_comma_nt;std::string goal_nt_name;std::string nt_prefix;
+  bool identity_scfg;
+  bool features;
+  bool logprob_feat;
+  bool cfg_comma_nt;
+  bool nt_span;
+  std::string goal_nt_name;
+  std::string nt_prefix;
+  std::string logprob_feat_name;
+  std::string partsep;
   template <class Opts> // template to support both printable_opts and boost nonprintable
   void AddOptions(Opts *opts) {
     using namespace boost::program_options;
@@ -14,19 +24,81 @@ struct CFGFormat {
       ("identity_scfg",defaulted_value(&identity_scfg),"output an identity SCFG: add an identity target side - '[X12] ||| [X13,1] a ||| [1] a ||| feat= ...' - the redundant target '[1] a |||' is omitted otherwise.")
       ("features",defaulted_value(&features),"print the CFG feature vector")
       ("logprob_feat",defaulted_value(&logprob_feat),"print a LogProb=-1.5 feature irrespective of --features.")
+      ("logprob_feat_name",defaulted_value(&logprob_feat_name),"alternate name for the LogProb feature")
       ("cfg_comma_nt",defaulted_value(&cfg_comma_nt),"if false, omit the usual [NP,1] ',1' variable index in the source side")
       ("goal_nt_name",defaulted_value(&goal_nt_name),"if nonempty, the first production will be '[goal_nt_name] ||| [x123] ||| LogProb=y' where x123 is the actual goal nt, and y is the pushed prob, if any")
       ("nt_prefix",defaulted_value(&nt_prefix),"NTs are [<nt_prefix>123] where 123 is the node number starting at 0, and the highest node (last in file) is the goal node in an acyclic hypergraph")
+      ("nt_span",defaulted_value(&nt_span),"prefix A(i,j) for NT coming from hypergraph node with category A on span [i,j).  this is after --nt_prefix if any")
       ;
   }
+
+  template<class CFG>
+  void print_source_nt(std::ostream &o,CFG const&cfg,int id,int position=1) const {
+    o<<'[';
+    print_nt_name(o,cfg,id);
+    if (cfg_comma_nt) o<<','<<position;
+    o<<']';
+  }
+
+  template <class CFG>
+  void print_nt_name(std::ostream &o,CFG const& cfg,int id) const {
+    o<<nt_prefix;
+    cfg.print_nt_name(o,id);
+    o<<id;
+  }
+
+  template <class CFG>
+  void print_lhs(std::ostream &o,CFG const& cfg,int id) const {
+    o<<'[';
+    print_nt_name(o,cfg,id);
+    o<<']';
+  }
+
+  template <class CFG,class Iter>
+  void print_rhs(std::ostream &o,CFG const&cfg,Iter begin,Iter end) const {
+    o<<partsep;
+    int pos=0;
+    for (Iter i=begin;i!=end;++i) {
+      WordID w=*i;
+      if (i!=begin) o<<' ';
+      if (w>0) o << TD::Convert(w);
+      else print_source_nt(o,cfg,-w,++pos);
+    }
+    if (identity_scfg) {
+      o<<partsep;
+      int pos=0;
+      for (Iter i=begin;i!=end;++i) {
+        WordID w=*i;
+        if (i!=begin) o<<' ';
+        if (w>0) o << TD::Convert(w);
+        else o << '['<<++pos<<']';
+      }
+    }
+  }
+
+  void print_features(std::ostream &o,prob_t p,FeatureVector const& fv=FeatureVector()) const {
+    bool logp=(logprob_feat && p!=1);
+    if (features || logp) {
+      o << partsep;
+      if (logp)
+        o << logprob_feat_name<<'='<<log(p)<<' ';
+      if (features)
+        o << fv;
+    }
+  }
+
   void set_defaults() {
     identity_scfg=false;
     features=true;
     logprob_feat=true;
     cfg_comma_nt=true;
     goal_nt_name="S";
+    logprob_feat_name="LogProb";
     nt_prefix="";
+    partsep=" ||| ";
+    nt_span=true;
   }
+
   CFGFormat() {
     set_defaults();
   }
-- 
cgit v1.2.3