Fixed compression level detection for tables and datasets

DavidAce · Oct 10, 2020 · accd0f8 · accd0f8
1 parent 84d97ab
commit accd0f8
Show file tree

Hide file tree

Showing 4 changed files with 33 additions and 9 deletions.
diff --git a/h5pp/include/h5pp/details/h5ppFile.h b/h5pp/include/h5pp/details/h5ppFile.h
@@ -874,16 +874,16 @@ namespace h5pp {
                                    hsize_t                           numRecordsToCopy,
                                    std::string_view                  tgtTablePath,
                                    hsize_t                           tgtStartIdx,
-                                   const OptDimsType &               chunkDims        = std::nullopt,
-                                   const std::optional<unsigned int> compressionLevel = std::nullopt) {
+                                   const OptDimsType &               chunkDims  = std::nullopt,
+                                   const std::optional<unsigned int> compression = std::nullopt) {
             Options options;
             options.linkPath      = h5pp::util::safe_str(tgtTablePath);
             options.dsetDimsChunk = chunkDims;
-            options.compression   = compressionLevel;
+            options.compression   = getCompressionLevel(compression);
             auto tgtInfo          = h5pp::scan::readTableInfo(openFileHandle(), options, plists);
             if(not tgtInfo.tableExists or not tgtInfo.tableExists.value())
                 tgtInfo = createTable(
-                    srcInfo.tableType.value(), tgtInfo.tablePath.value(), srcInfo.tableTitle.value(), chunkDims, compressionLevel);
+                    srcInfo.tableType.value(), tgtInfo.tablePath.value(), srcInfo.tableTitle.value(), chunkDims, compression);
 
             copyTableRecords(srcInfo, srcStartIdx, numRecordsToCopy, tgtInfo, tgtStartIdx);
             return tgtInfo;
@@ -896,11 +896,11 @@ namespace h5pp {
                                    std::string_view                  tgtTablePath,
                                    hsize_t                           tgtStartIdx,
                                    const std::optional<hsize_t>      chunkDims        = std::nullopt,
-                                   const std::optional<unsigned int> compressionLevel = std::nullopt) {
+                                   const std::optional<unsigned int> compression = std::nullopt) {
             Options options;
             options.linkPath = h5pp::util::safe_str(srcTablePath);
             auto srcInfo     = h5pp::scan::readTableInfo(srcLocation, options, plists);
-            return copyTableRecords(srcInfo, srcTableSelection, tgtTablePath, tgtStartIdx, chunkDims, compressionLevel);
+            return copyTableRecords(srcInfo, srcTableSelection, tgtTablePath, tgtStartIdx, chunkDims, compression);
         }
 
         template<typename h5x_src, typename = h5pp::type::sfinae::enable_if_is_h5_loc<h5x_src>>

diff --git a/h5pp/include/h5pp/details/h5ppHdf5.h b/h5pp/include/h5pp/details/h5ppHdf5.h
@@ -118,6 +118,21 @@ namespace h5pp::hdf5 {
         return getChunkDimensions(dcpl);
     }
 
+    [[nodiscard]] inline int getCompressionLevel(const hid::h5p &dsetCreatePropertyList) {
+        auto nfilter = H5Pget_nfilters(dsetCreatePropertyList);
+        H5Z_filter_t filter = H5Z_FILTER_NONE;
+        std::array<unsigned int,1> cdval = {0};
+        std::array<unsigned long,1> cdelm = {0};
+        for(int idx=0; idx < nfilter;idx++) {
+            constexpr size_t size = 10;
+            filter = H5Pget_filter(dsetCreatePropertyList, idx, nullptr, cdelm.data(), cdval.data(), 0, nullptr, nullptr);
+            if(filter != H5Z_FILTER_DEFLATE) continue;
+            H5Pget_filter_by_id(dsetCreatePropertyList, filter, nullptr, cdelm.data(), cdval.data(),  0, nullptr, nullptr);
+         }
+         return cdval[0];
+    }
+
+
     [[nodiscard]] inline std::optional<std::vector<hsize_t>> getMaxDimensions(const hid::h5s &space, H5D_layout_t layout) {
         if(layout != H5D_CHUNKED) return std::nullopt;
         if(H5Sget_simple_extent_type(space) != H5S_SIMPLE) return std::nullopt;

diff --git a/h5pp/include/h5pp/details/h5ppScan.h b/h5pp/include/h5pp/details/h5ppScan.h
@@ -59,7 +59,10 @@ namespace h5pp::scan {
             else
                 info.resizeMode = h5pp::ResizeMode::RESIZE_TO_FIT;
         }
-
+        if(not info.compression){
+            hid::h5p plist    = H5Dget_create_plist(info.h5Dset.value());
+            info.compression = h5pp::hdf5::getCompressionLevel(plist);
+        }
         // Get c++ properties
         if(not info.cppTypeIndex or not info.cppTypeName or not info.cppTypeSize)
             std::tie(info.cppTypeIndex, info.cppTypeName, info.cppTypeSize) = h5pp::hdf5::getCppType(info.h5Type.value());
@@ -534,6 +537,11 @@ namespace h5pp::scan {
             if(chunkVec and not chunkVec->empty()) info.chunkSize = chunkVec.value()[0];
         }
 
+        if(not info.compressionLevel){
+            hid::h5p plist    = H5Dget_create_plist(info.tableDset.value());
+            info.compressionLevel = h5pp::hdf5::getCompressionLevel(plist);
+        }
+
         if(not info.cppTypeIndex or not info.cppTypeName or not info.cppTypeSize) {
             // Get c++ type information
             info.cppTypeIndex = std::vector<std::type_index>();
@@ -569,7 +577,7 @@ namespace h5pp::scan {
         if(options.dsetDimsChunk and not options.dsetDimsChunk->empty()) info.chunkSize = options.dsetDimsChunk.value()[0];
 
         if(not info.chunkSize) info.chunkSize = h5pp::util::getChunkDimensions(info.recordBytes.value(), {1}, std::nullopt, H5D_layout_t::H5D_CHUNKED).value()[0];
-        if(not info.compressionLevel) info.compressionLevel = h5pp::hdf5::getValidCompressionLevel(info.compressionLevel);
+        if(not info.compressionLevel) info.compressionLevel = h5pp::hdf5::getValidCompressionLevel();
 
         info.fieldTypes   = std::vector<h5pp::hid::h5t>();
         info.fieldOffsets = std::vector<size_t>();

diff --git a/tests/test-readWriteTables.cpp b/tests/test-readWriteTables.cpp
@@ -29,6 +29,7 @@ h5pp::hid::h5t MY_HDF5_PARTICLE_TYPE;
 
 TEST_CASE("Test reading columns from table", "[Table fields]") {
     SECTION("Initialize file") {
+        file.setCompressionLevel(6);
         // Create a type for the char array from the template H5T_C_S1
         // The template describes a string with a single char.
         // Set the size with H5Tset_size.
@@ -52,7 +53,7 @@ TEST_CASE("Test reading columns from table", "[Table fields]") {
     }
 
     SECTION("Create table") {
-        auto tableInfo = file.createTable(MY_HDF5_PARTICLE_TYPE, "somegroup/particleTable", "particleTable");
+        auto tableInfo = file.createTable(MY_HDF5_PARTICLE_TYPE, "somegroup/particleTable", "particleTable",std::nullopt,6);
         CHECK(tableInfo.tableTitle.value() == "particleTable");
         CHECK(tableInfo.numRecords.value() == 0);
         CHECK(tableInfo.recordBytes.value() == sizeof(Particle));