From addc7291bd44db96a01f399f8ee4efbe226245e2 Mon Sep 17 00:00:00 2001
From: graehl <graehl@ec762483-ff6d-05da-a07a-a48fb63a330f>
Date: Tue, 17 Aug 2010 22:05:59 +0000
Subject: vest generate / map vector print / read compatability

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@585 ec762483-ff6d-05da-a07a-a48fb63a330f
---
 decoder/cdec.cc        |  2 +-
 decoder/cfg.cc         | 12 +++++++++++-
 decoder/cfg.h          |  3 ++-
 decoder/cfg_binarize.h |  4 ++--
 utils/sparse_vector.h  | 11 ++++++-----
 5 files changed, 22 insertions(+), 10 deletions(-)

diff --git a/decoder/cdec.cc b/decoder/cdec.cc
index 0a02801e..b156c268 100644
--- a/decoder/cdec.cc
+++ b/decoder/cdec.cc
@@ -379,7 +379,7 @@ void maybe_prune(Hypergraph &forest,po::variables_map const& conf,string nbeam,s
       }
       forest.PruneInsideOutside(beam_prune,density_prune,pm,false,1,conf["promise_power"].as<double>());
       if (!forestname.empty()) forestname=" "+forestname;
-      forest_stats(forest,"  Pruned "+forestname+" forest",false,false,false);
+      forest_stats(forest,"  Pruned "+forestname+" forest",false,false,0,false);
       cerr << "  Pruned "<<forestname<<" forest portion of edges kept: "<<forest.edges_.size()/presize<<endl;
     }
 }
diff --git a/decoder/cfg.cc b/decoder/cfg.cc
index f51da9bf..c02f46ec 100755
--- a/decoder/cfg.cc
+++ b/decoder/cfg.cc
@@ -10,7 +10,7 @@
 #include "show.h"
 
 #define DUNIQ(x) x
-#define DBIN(x) x
+#define DBIN(x)
 #define DSP(x) x
 //SP:binarize by splitting.
 #define DCFG(x) IF_CFG_DEBUG(x)
@@ -464,6 +464,8 @@ void CFG::BinarizeSplit(CFGBinarize const& b) {
 void CFG::Binarize(CFGBinarize const& b) {
   if (!b.Binarizing()) return;
   cerr << "Binarizing "<<b<<endl;
+  if (b.bin_thresh>0)
+    BinarizeThresh(b);
   if (b.bin_split)
     BinarizeSplit(b);
   if (b.bin_l2r)
@@ -473,6 +475,14 @@ void CFG::Binarize(CFGBinarize const& b) {
 
 }
 
+namespace {
+}
+
+void CFG::BinarizeThresh(CFGBinarize const& b) {
+  throw runtime_error("TODO: some fancy linked list thing - see NOTES.partial.binarize");
+}
+
+
 void CFG::BinarizeL2R(bool bin_unary,bool name) {
   add_virtual_rules<BinRhs> v(*this,name);
 cerr << "Binarizing left->right " << (bin_unary?"real to unary":"stop at binary") <<endl;
diff --git a/decoder/cfg.h b/decoder/cfg.h
index 5a418234..9be0926d 100755
--- a/decoder/cfg.h
+++ b/decoder/cfg.h
@@ -306,7 +306,8 @@ struct CFG {
 
   void BinarizeL2R(bool bin_unary=false,bool name_nts=false);
   void Binarize(CFGBinarize const& binarize_options); // see cfg_binarize.h for docs
-  void BinarizeSplit(CFGBinarize const& binarize_options); // there may be many options affecting split.
+  void BinarizeSplit(CFGBinarize const& binarize_options);
+  void BinarizeThresh(CFGBinarize const& binarize_options); // maybe unbundle opts later
 
   typedef std::vector<NT> NTs;
   NTs nts;
diff --git a/decoder/cfg_binarize.h b/decoder/cfg_binarize.h
index 41eba11b..ae06f8bf 100755
--- a/decoder/cfg_binarize.h
+++ b/decoder/cfg_binarize.h
@@ -37,8 +37,8 @@ struct CFGBinarize {
   }
   void Validate() {
     if (bin_thresh>0&&!bin_l2r) {
-      std::cerr<<"\nWARNING: greedy binarization not yet supported; using l2r (right branching) instead.\n";
-      bin_l2r=true;
+//      std::cerr<<"\nWARNING: greedy binarization not yet supported; using l2r (right branching) instead.\n";
+//      bin_l2r=true;
     }
     if (false && bin_l2r && bin_split) { // actually, split may be slightly incomplete due to finite number of passes.
       std::cerr<<"\nWARNING: l2r and split are both complete binarization and redundant.  Using split.\n";
diff --git a/utils/sparse_vector.h b/utils/sparse_vector.h
index e3904403..5d0dac27 100644
--- a/utils/sparse_vector.h
+++ b/utils/sparse_vector.h
@@ -95,7 +95,7 @@ public:
 
   typedef char const* Str;
   template <class O>
-  void print(O &o,Str pre="",Str post="",Str kvsep="=",Str pairsep=" ") const {
+  void print(O &o,Str kvsep="=",Str pairsep=" ",Str pre="",Str post="") const {
     o << pre;
     bool first=true;
     for (const_iterator i=values_.begin(),e=values_.end();i!=e;++i) {
@@ -121,7 +121,7 @@ public:
 
   // either key val alternating whitespace sep, or key=val (kvsep char is '=').  end at eof or terminator (non-ws) char
   template <class S>
-  void read(S &s,DupPolicy dp=NO_DUPS,bool use_kvsep=true,char kvsep='=',bool stop_at_terminator=false,char terminator=')') {
+  void read(S &s,DupPolicy dp=NO_DUPS,bool use_kvsep=true,char kvsep='=',bool use_pairsep=true,char optional_pairsep=';',bool stop_at_terminator=false,char terminator=')') {
     values_.clear();
     std::string id;
     WordID k;
@@ -130,11 +130,12 @@ public:
 #define SPARSE_MUST_READ(x) if (!(x)) error(#x);
     int ki;
     while (s) {
-      if (stop_at_terminator) {
+      if (stop_at_terminator||use_pairsep) {
         char c;
         if (!(s>>c)) goto eof;
-        s.unget();
-        if (c==terminator) return;
+        if (stop_at_terminator && c==terminator) return;
+        if (!use_pairsep || c!=optional_pairsep)
+          s.unget();
       }
       if (!(s>>id)) goto eof;
       if (use_kvsep && (ki=id.find(kvsep))!=std::string::npos) {
-- 
cgit v1.2.3