really fixed binarization. test

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

1 files changed, 56 insertions, 24 deletions

diff --git a/decoder/cfg_test.cc b/decoder/cfg_test.cc
index 81efa768..cde4706c 100755
--- a/decoder/cfg_test.cc
+++ b/decoder/cfg_test.cc
@@ -1,23 +1,34 @@
+#include <boost/tuple/tuple.hpp>
 #include <gtest/gtest.h>
 #include "cfg.h"
 #include "hg_test.h"
 #include "cfg_options.h"
 
-#define CSHOW_V 1
+/* TODO: easiest way to get meaningful confirmations that things work: implement conversion back to hg, and compare viterbi/inside etc. stats for equality to original hg.  or you can define CSHOW_V and see lots of output */
+
+using namespace boost;
+
+#define CSHOW_V 0
+
 #if CSHOW_V
-# define CSHOWDO(x) x
+# define CSHOWDO(x) x;
 #else
 # define CSHOWDO(x)
 #endif
 #define CSHOW(x) CSHOWDO(cerr<<#x<<'='<<x<<endl;)
 
-struct CFGTest : public HGSetup {
-  CFGTest() {  }
-  ~CFGTest() {  }
-  static void JsonFN(Hypergraph hg,CFG &cfg,std::string file
+typedef std::pair<string,string> HgW; // hg file,weights
+
+struct CFGTest : public TestWithParam<HgW> {
+  string hgfile;
+  Hypergraph hg;
+  CFG cfg;
+  CFGFormat form;
+  FeatureVector weights;
+
+  static void JsonFN(Hypergraph &hg,CFG &cfg,FeatureVector &featw,std::string file
                      ,std::string const& wts="Model_0 1 EgivenF 1 f1 1")
   {
-    FeatureVector featw;
     istringstream ws(wts);
     EXPECT_TRUE(ws>>featw);
     CSHOW(featw)
@@ -25,35 +36,56 @@ struct CFGTest : public HGSetup {
     hg.Reweight(featw);
     cfg.Init(hg,true,true,false);
   }
-
   static void SetUpTestCase() {
   }
   static void TearDownTestCase() {
   }
+  CFGTest() {
+    hgfile=GetParam().first;
+    JsonFN(hg,cfg,weights,hgfile,GetParam().second);
+    CSHOWDO(cerr<<"\nCFG Test: ")
+    CSHOW(hgfile);
+    form.nt_span=true;
+    form.comma_nt=false;
+  }
+  ~CFGTest() {  }
 };
 
-TEST_F(CFGTest,Binarize) {
-  Hypergraph hg;
-  CFG cfg;
-  JsonFN(hg,cfg,perro_json,perro_wts);
-  CSHOW("\nCFG Test.\n");
+TEST_P(CFGTest,Binarize) {
   CFGBinarize b;
-  CFGFormat form;
-  form.nt_span=true;
-  for (int i=-1;i<16;++i) {
-    b.bin_l2r=i>=0;
-    b.bin_unary=i&1;
-    b.bin_name_nts=i&2;
-    b.bin_uniq=i&4;
-    b.bin_topo=i&8;
-    CFG cc=cfg;
-    EXPECT_EQ(cc,cfg);
-    CSHOW("\nBinarizing: "<<b);
+  b.bin_name_nts=1;
+  CFG cfgu=cfg;
+  EXPECT_EQ(cfgu,cfg);
+  int nrules=cfg.rules.size();
+  CSHOWDO(cerr<<"\nUniqing: "<<nrules<<"\n");
+  int nrem=cfgu.UniqRules();
+  cerr<<"\nCFG "<<hgfile<<" Uniqed - remaining: "<<nrem<<" of "<<nrules<<"\n";
+  if (nrem==nrules)
+    EXPECT_EQ(cfgu,cfg);
+  for (int i=-1;i<8;++i) {
+    bool uniq;
+    if (i>=0) {
+      int f=i<<1;
+      b.bin_l2r=1;
+      b.bin_unary=(f>>=1)&1;
+      b.bin_topo=(f>>=1)&1;
+      uniq=(f>>=1)&1;
+    } else
+      b.bin_l2r=0;
+    CFG cc=uniq?cfgu:cfg;
+    CSHOW("\nBinarizing "<<(uniq?"uniqued ":"")<<": "<<i<<" "<<b);
     cc.Binarize(b);
     CSHOWDO(cc.Print(cerr,form);cerr<<"\n\n";);
   }
 }
 
+INSTANTIATE_TEST_CASE_P(HypergraphsWeights,CFGTest,
+                        Values(
+                          HgW(perro_json,perro_wts)
+                          , HgW(small_json,small_wts)
+                            ,HgW(urdu_json,urdu_wts)
+                          ));
+
 int main(int argc, char **argv) {
   testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();