Add util for querying on-disk pt

2 files changed, 32 insertions, 1 deletions

diff --git a/contrib/queryOnDiskPt/compile.sh b/contrib/queryOnDiskPt/compile.sh
index 8d1660b17..1643e27d1 100755
--- a/contrib/queryOnDiskPt/compile.sh
+++ b/contrib/queryOnDiskPt/compile.sh
@@ -1,6 +1,6 @@
 SRI=/Users/hieuhoang/workspace/srilm
 IRST=/Users/hieuhoang/workspace/irstlm/trunk
 
-g++ queryOnDiskPt.cpp ../../moses/src/PhraseDictionary.cpp -I../../moses/src/ -I../../ -L../../dist/lib/ -I../../OnDiskPt -lmert_lib -ldynsa -lz -lmoses_internal -lOnDiskPt -lLM -lkenlm -lkenutil -lRuleTable -lCYKPlusParser -lScope3Parser -L$SRI/lib/macosx/ -ldstruct -lflm -llattice -lmisc -loolm -L/opt/local/lib -lboost_thread-mt -L$IRST/lib -lirstlm
+g++ -o queryOnDiskPt queryOnDiskPt.cpp ../../moses/src/PhraseDictionary.cpp -I../../moses/src/ -I../../ -L../../dist/lib/ -I../../OnDiskPt -lmert_lib -ldynsa -lz -lmoses_internal -lOnDiskPt -lLM -lkenlm -lkenutil -lRuleTable -lCYKPlusParser -lScope3Parser -L$SRI/lib/macosx/ -ldstruct -lflm -llattice -lmisc -loolm -L/opt/local/lib -lboost_thread-mt -L$IRST/lib -lirstlm
 
 
diff --git a/contrib/queryOnDiskPt/queryOnDiskPt.cpp b/contrib/queryOnDiskPt/queryOnDiskPt.cpp
index fb2667adf..afcd1b46b 100644
--- a/contrib/queryOnDiskPt/queryOnDiskPt.cpp
+++ b/contrib/queryOnDiskPt/queryOnDiskPt.cpp
@@ -17,6 +17,8 @@ void usage();
 
 typedef unsigned int uint;
 
+#define TABLE_LIMIT 20
+
 void Tokenize(OnDiskPt::Phrase &phrase
               , const std::string &token, bool addSourceNonTerm, bool addTargetNonTerm
               , OnDiskPt::OnDiskWrapper &onDiskWrapper)
@@ -94,6 +96,8 @@ int main(int argc, char **argv)
   bool retDb = onDiskWrapper.BeginLoad(ttable);
 	CHECK(retDb);
 	
+	cerr << "Ready..." << endl;
+	
   std::string line;
   while(getline(std::cin, line)) {
     std::vector<std::string> tokens;
@@ -116,10 +120,37 @@ int main(int argc, char **argv)
 			}
 		}
 
+    const PhraseNode *node = &onDiskWrapper.GetRootSourceNode();
+    assert(node);
+    
+    for (size_t pos = 0; pos < sourcePhrase.GetSize(); ++pos)
+		{
+		  const Word &word = sourcePhrase.GetWord(pos);
+		  node = node->GetChild(word, onDiskWrapper);
+		  
+		  if (node == NULL)
+		  {
+		    break;
+		  }
+		}
+    
+    if (node)
+    { // source phrase points to something
+      const TargetPhraseCollection *coll = node->GetTargetPhraseCollection(TABLE_LIMIT, onDiskWrapper);
+      string str = coll->GetDebugStr();
+      cout << "Found" << str << endl;
+    }
+    else
+    {
+      cout << "Not found" << endl;
+    }
     
     std::cout << '\n';
     std::cout.flush();
   }
+  
+  cerr << "Finished." << endl;
+	
 }
 
 void usage()