H5 Writer: allow double -> half

H5 Writer: allow double -> half

PhysicsAnalysis/AnalysisCommon/HDF5Utils/HDF5Utils/CompressedTypes.h

@@ -16,16 +16,26 @@
 namespace H5Utils {
 
   namespace internal {
+
+    H5::DataType halfPrecisionFloat(int ebias = 15);
+
     template <typename T>
     H5::DataType getCompressedType(Compression comp) {
+      if constexpr (std::is_floating_point<T>::value) {
+        switch (comp) {
+          case Compression::STANDARD: return H5Traits<T>::type;
+          case Compression::HALF_PRECISION: return halfPrecisionFloat();
+          case Compression::HALF_PRECISION_LARGE: return halfPrecisionFloat(5);
+          default: throw std::logic_error("unknown float compression");
+        }
+      }
       if (comp != Compression::STANDARD) {
         throw std::logic_error("compression not supported for this type");
       }
       return H5Traits<T>::type;
     }
-    template <>
-    H5::DataType getCompressedType<float>(Compression comp);
-  }
-}
+
+  } // end internal
+}   // end H5Utils
 
 #endif

PhysicsAnalysis/AnalysisCommon/HDF5Utils/Root/CompressedTypes.cxx

@@ -5,33 +5,24 @@
 #include "HDF5Utils/CompressedTypes.h"
 
 namespace {
-  H5::DataType halfPrecisionFloat(int ebias = 15) {
-    // start with native float
-    H5::FloatType type(H5Tcopy(H5::PredType::NATIVE_FLOAT.getId()));
-
-    // These definitions are copied from h5py, see:
-    //
-    //  https://github.com/h5py/h5py/blob/596748d52c351258c851bb56c8df1c25d3673110/h5py/h5t.pyx#L212-L217
-    //
-    type.setFields(15, 10, 5, 0, 10);
-    type.setSize(2);
-    type.setEbias(ebias);
-    return type;
-  }
 }
 
 namespace H5Utils {
 
   namespace internal {
 
-    template <>
-    H5::DataType getCompressedType<float>(Compression comp) {
-      switch (comp) {
-      case Compression::STANDARD: return H5Traits<float>::type;
-      case Compression::HALF_PRECISION: return halfPrecisionFloat();
-      case Compression::HALF_PRECISION_LARGE: return halfPrecisionFloat(5);
-      default: throw std::logic_error("unknown float compression");
-      }
+    H5::DataType halfPrecisionFloat(int ebias) {
+      // start with native float
+      H5::FloatType type(H5Tcopy(H5::PredType::NATIVE_FLOAT.getId()));
+
+      // These definitions are copied from h5py, see:
+      //
+      //  https://github.com/h5py/h5py/blob/596748d52c351258c851bb56c8df1c25d3673110/h5py/h5t.pyx#L212-L217
+      //
+      type.setFields(15, 10, 5, 0, 10);
+      type.setSize(2);
+      type.setEbias(ebias);
+      return type;
     }
 
   }

PhysicsAnalysis/AnalysisCommon/HDF5Utils/util/test-hdf5-writer.cxx

@@ -11,10 +11,12 @@ struct out_t
   double dtype;
   float ftype;
   char ctype;
+  short stype;
   int itype;
   long ltype;
   long long lltype;
   unsigned char uctype;
+  unsigned short ustype;
   unsigned int uitype;
   unsigned long ultype;
   unsigned long long ulltype;
@@ -24,20 +26,20 @@ using consumer_t = H5Utils::Consumers<const out_t&>;
 
 consumer_t getConsumers() {
   consumer_t consumers;
-  consumers.add(
-    "half",
-    [](const out_t& o) -> float { return o.ftype; },
-    0,
-    H5Utils::Compression::HALF_PRECISION);
+  auto half = H5Utils::Compression::HALF_PRECISION;
+  consumers.add("half" , [](const out_t& o) { return o.ftype; }, 0, half);
+  consumers.add("dhalf", [](const out_t& o) { return o.dtype; }, 0, half);
 #define ADD(NAME) consumers.add(#NAME, [](const out_t& o){ return o.NAME;}, 0)
   ADD(ftype);
   ADD(dtype);
   ADD(btype);
   ADD(ctype);
+  ADD(stype);
   ADD(itype);
   ADD(ltype);
   ADD(lltype);
   ADD(uctype);
+  ADD(ustype);
   ADD(uitype);
   ADD(ultype);
   ADD(ulltype);
@@ -57,10 +59,12 @@ std::vector<out_t> getOutputs(int offset, size_t length, float factor) {
     out.dtype = factored;
     out.ftype = factored;
     out.ctype = shifted;
+    out.stype = shifted;
     out.itype = shifted;
     out.ltype = shifted;
     out.lltype = shifted;
     out.uctype = shifted;
+    out.ustype = shifted;
     out.uitype = shifted;
     out.ultype = shifted;
     out.ulltype = shifted;