feat: SCHWED-5481 read gzipped SDF

modules/io/doc/structure_formats.rst

@@ -54,5 +54,5 @@ SDF - Structured Data File
 Chemical-data file format.
 
 *Recognized File Extensions*
-  .sdf
+  .sdf, .sdf.gz
   

modules/io/src/mol/entity_io_sdf_handler.cc

@@ -69,7 +69,7 @@ bool sdf_handler_is_responsible_for(const boost::filesystem::path& loc,
   if(type=="auto") {
 	String match_suf_string=loc.string();
     std::transform(match_suf_string.begin(),match_suf_string.end(),match_suf_string.begin(),tolower);
-    if(detail::FilenameEndsWith(match_suf_string,".sdf")) {
+    if(detail::FilenameEndsWith(match_suf_string,".sdf") || detail::FilenameEndsWith(match_suf_string,".sdf.gz")) {
       return true;
     }
 

modules/io/src/mol/sdf_reader.cc

@@ -21,7 +21,9 @@
  */
 
 #include <boost/algorithm/string.hpp>
+#include <boost/filesystem/convenience.hpp>
 #include <boost/format.hpp>
+#include <boost/iostreams/filter/gzip.hpp>
 #include <boost/lexical_cast.hpp>
 #include <ost/mol/bond_handle.hh>
 #include <ost/conop/conop.hh>
@@ -58,7 +60,7 @@ void SDFReader::Import(mol::EntityHandle& ent)
 {
   String line;
   mol::XCSEditor editor=ent.EditXCS(mol::BUFFERED_EDIT);
-  while (std::getline(instream_,line)) {
+  while (std::getline(in_,line)) {
     ++line_num;
 
     // std::getline removes EOL character but may leave a DOS CR (\r) in Unix
@@ -87,7 +89,7 @@ void SDFReader::Import(mol::EntityHandle& ent)
         throw IOException(str(format(msg) % line_num));
       }
       String data_value="";
-      while(std::getline(instream_,line) && !boost::iequals(line, "")) {
+      while(std::getline(in_,line) && !boost::iequals(line, "")) {
         data_value.append(line);
       }
       curr_chain_.SetStringProp(data_header, data_value);
@@ -103,6 +105,10 @@ void SDFReader::Import(mol::EntityHandle& ent)
 
 void SDFReader::ClearState(const boost::filesystem::path& loc)
 {
+  if (boost::iequals(".gz", boost::filesystem::extension(loc))) {
+    in_.push(boost::iostreams::gzip_decompressor());
+  }
+  in_.push(instream_);
   if(!infile_) throw IOException("could not open "+loc.string());
   curr_chain_=mol::ChainHandle();
   curr_residue_=mol::ResidueHandle();

modules/io/src/mol/sdf_reader.hh

@@ -22,6 +22,7 @@
 #ifndef OST_IO_SDF_READER_HH
 #define OST_IO_SDF_READER_HH
 
+#include <boost/iostreams/filtering_stream.hpp>
 #include <boost/filesystem/fstream.hpp>
 #include <ost/mol/chain_handle.hh>
 #include <ost/mol/residue_handle.hh>
@@ -61,6 +62,7 @@ private:
   int line_num;
   boost::filesystem::ifstream infile_;
   std::istream& instream_;
+  boost::iostreams::filtering_stream<boost::iostreams::input>  in_;
 };
 
 }}

modules/io/tests/CMakeLists.txt

@@ -2,6 +2,7 @@ set(OST_IO_UNIT_TESTS
   test_io_pdb.py
   test_io_mmcif.py
   test_io_omf.py
+  test_io_sdf.py
   test_clustal.cc
   test_io_pdb.cc
   test_io_crd.cc