Added features ldep and rdep

transition_machine/src/FeatureBank.cpp

@@ -32,6 +32,50 @@ FeatureModel::FeatureValue access(Config & config, int bufferIndex, const std::s
   return {dict, featName, tape[bufferIndex-config.getHead()], policy};
 }
 
+int getLeftMostDep(Config & config, int index)
+{
+  auto & govs = config.getTape("GOV");
+  auto & eos = config.getTape(ProgramParameters::sequenceDelimiterTape);
+  int b0 = config.getHead();
+
+  if(index < 0 || index >= govs.hypSize())
+    return -1;
+
+  int candidate = -1;
+  unsigned int maxDist = 50;
+
+  for (int i = std::max<int>(index-1, 0); index - i <= (int)maxDist && i >= 0 && eos[i-b0] != ProgramParameters::sequenceDelimiter; i--)
+  {
+    int dist = index - i;
+    if(govs[i-b0] == std::to_string(dist))
+      candidate = i;
+  }
+
+  return candidate;
+}
+
+int getRightMostDep(Config & config, int index)
+{
+  auto & govs = config.getTape("GOV");
+  auto & eos = config.getTape(ProgramParameters::sequenceDelimiterTape);
+  int b0 = config.getHead();
+
+  if(index < 0 || index >= govs.hypSize())
+    return -1;
+
+  int candidate = -1;
+  unsigned int maxDist = 50;
+
+  for (int i = index+1; i - index <= (int)maxDist && i < eos.hypSize() && eos[i-1-b0] != ProgramParameters::sequenceDelimiter; i++)
+  {
+    int dist = index - i;
+    if(govs[i-b0] == std::to_string(dist))
+      candidate = i;
+  }
+
+  return candidate;
+}
+
 std::function<FeatureModel::FeatureValue(Config &)> FeatureBank::str2func(const std::string & s)
 {
   auto splited = split(s, '#');
@@ -105,6 +149,18 @@ std::function<FeatureModel::FeatureValue(Config &)> FeatureBank::str2func(const
   std::vector<std::function<int(Config&, int)> > contextFuncs;
   char object = splited[0][0];
   contextFuncs.emplace_back([relativeIndex, object](Config & c, int){return getBufferIndex(c, object, relativeIndex);});
+  for (unsigned int i = 2; i < splited.size(); i++)
+  {
+    if (splited[i] == "ldep")
+      contextFuncs.emplace_back([](Config & c, int currentIndex){return getLeftMostDep(c, currentIndex);});
+    else if (splited[i] == "rdep")
+      contextFuncs.emplace_back([](Config & c, int currentIndex){return getRightMostDep(c, currentIndex);});
+    else
+    {
+      fprintf(stderr, "ERROR (%s) : invalid feature format \'%s\', unknown context \'%s\'. Aborting.\n", ERRINFO, s.c_str(), splited[i].c_str());
+      exit(1);
+    }
+  }
 
   auto getContext = [contextFuncs](Config & c)
     {