Decouple coordgen and maeparser integrations (#3286)

* decouple the maeparser and coordgen support

* further untangling of coordgen and maeparser

* more tweaks

* be explicit in the azure devops builds

* forgot one file

* get win64_dll builds working

.azure-pipelines/linux_build.yml

@@ -25,6 +25,7 @@ steps:
     -DRDK_BUILD_PYTHON_WRAPPERS=ON \
     -DRDK_USE_BOOST_REGEX=ON \
     -DRDK_BUILD_COORDGEN_SUPPORT=ON \
+    -DRDK_BUILD_MAEPARSER_SUPPORT=ON \
     -DRDK_OPTIMIZE_POPCNT=ON \
     -DRDK_BUILD_TEST_GZIP=ON \
     -DRDK_BUILD_FREESASA_SUPPORT=ON \

.azure-pipelines/linux_build_cartridge.yml

@@ -15,6 +15,7 @@ steps:
     -DRDK_BUILD_PYTHON_WRAPPERS=OFF \
     -DRDK_USE_BOOST_REGEX=OFF \
     -DRDK_BUILD_COORDGEN_SUPPORT=ON \
+    -DRDK_BUILD_MAEPARSER_SUPPORT=ON \
     -DRDK_OPTIMIZE_POPCNT=ON \
     -DRDK_BUILD_TEST_GZIP=ON \
     -DRDK_BUILD_AVALON_SUPPORT=ON \

.azure-pipelines/linux_build_java.yml

@@ -23,6 +23,7 @@ steps:
     -DRDK_BUILD_PYTHON_WRAPPERS=OFF \
     -DRDK_USE_BOOST_REGEX=OFF \
     -DRDK_BUILD_COORDGEN_SUPPORT=ON \
+    -DRDK_BUILD_MAEPARSER_SUPPORT=ON \
     -DRDK_OPTIMIZE_POPCNT=ON \
     -DRDK_BUILD_TEST_GZIP=OFF \
     -DRDK_BUILD_FREESASA_SUPPORT=ON \

.azure-pipelines/linux_build_py37.yml

@@ -25,6 +25,7 @@ steps:
     -DRDK_BUILD_PYTHON_WRAPPERS=ON \
     -DRDK_USE_BOOST_REGEX=ON \
     -DRDK_BUILD_COORDGEN_SUPPORT=ON \
+    -DRDK_BUILD_MAEPARSER_SUPPORT=ON \
     -DRDK_OPTIMIZE_POPCNT=ON \
     -DRDK_BUILD_TEST_GZIP=ON \
     -DRDK_BUILD_FREESASA_SUPPORT=ON \

.azure-pipelines/mac_build.yml

@@ -37,6 +37,7 @@ steps:
     -DRDK_BUILD_PYTHON_WRAPPERS=ON \
     -DRDK_USE_BOOST_REGEX=ON \
     -DRDK_BUILD_COORDGEN_SUPPORT=ON \
+    -DRDK_BUILD_MAEPARSER_SUPPORT=ON \
     -DRDK_OPTIMIZE_POPCNT=ON \
     -DRDK_BUILD_TEST_GZIP=ON \
     -DRDK_BUILD_FREESASA_SUPPORT=ON \

.azure-pipelines/mac_build_java.yml

@@ -36,6 +36,7 @@ steps:
     -DRDK_BUILD_PYTHON_WRAPPERS=OFF \
     -DRDK_USE_BOOST_REGEX=OFF \
     -DRDK_BUILD_COORDGEN_SUPPORT=ON \
+    -DRDK_BUILD_MAEPARSER_SUPPORT=ON \
     -DRDK_OPTIMIZE_POPCNT=ON \
     -DRDK_BUILD_TEST_GZIP=OFF \
     -DRDK_BUILD_FREESASA_SUPPORT=ON \

.azure-pipelines/vs_build.yml

@@ -23,6 +23,7 @@ steps:
     -DRDK_BUILD_PYTHON_WRAPPERS=ON ^
     -DRDK_USE_BOOST_REGEX=ON ^
     -DRDK_BUILD_COORDGEN_SUPPORT=ON ^
+    -DRDK_BUILD_MAEPARSER_SUPPORT=ON ^
     -DRDK_OPTIMIZE_POPCNT=ON ^
     -DRDK_BUILD_TEST_GZIP=ON ^
     -DRDK_BUILD_FREESASA_SUPPORT=OFF ^

.azure-pipelines/vs_build_dll.yml

@@ -22,6 +22,7 @@ steps:
     -DRDK_BUILD_CPP_TESTS=ON ^
     -DRDK_USE_BOOST_REGEX=ON ^
     -DRDK_BUILD_COORDGEN_SUPPORT=ON ^
+    -DRDK_BUILD_MAEPARSER_SUPPORT=ON ^
     -DRDK_OPTIMIZE_POPCNT=ON ^
     -DRDK_BUILD_TEST_GZIP=ON ^
     -DRDK_BUILD_FREESASA_SUPPORT=OFF ^

CMakeLists.txt

@@ -51,6 +51,7 @@ option(RDK_USE_STRICT_ROTOR_DEFINITION "Use the most strict rotatable bond defin
 option(RDK_BUILD_DESCRIPTORS3D "Build the 3D descriptors calculators, requires Eigen3 to be installed" ON)
 option(RDK_BUILD_FREESASA_SUPPORT "build the rdkit freesasa wrapper" OFF )
 option(RDK_BUILD_COORDGEN_SUPPORT "build the rdkit coordgen wrapper" ON )
+option(RDK_BUILD_MAEPARSER_SUPPORT "build the rdkit MAE parser wrapper" ON )
 option(RDK_BUILD_MOLINTERCHANGE_SUPPORT "build in support for CommonChem molecule interchange" ON )
 option(RDK_BUILD_YAEHMOP_SUPPORT "build support for the YAeHMOP wrapper" OFF)
 option(RDK_BUILD_STRUCTCHECKER_SUPPORT "build in support for the StructChecker alpha (not recommended, use the MolVS integration instead)" OFF )
@@ -178,6 +179,11 @@ if(RDK_BUILD_COORDGEN_SUPPORT)
     include_directories(${RDKit_ExternalDir})
 endif()
 
+if(RDK_BUILD_MAEPARSER_SUPPORT)
+    add_definitions(-DRDK_BUILD_MAEPARSER_SUPPORT)
+    include_directories(${RDKit_ExternalDir})
+endif()
+
 if(RDK_USE_URF)
   ADD_DEFINITIONS("-DRDK_USE_URF")
   include_directories(${RDKit_ExternalDir}/RingFamilies/RingDecomposerLib/src/RingDecomposerLib)

Code/GraphMol/FileParsers/CMakeLists.txt

@@ -20,7 +20,7 @@ endif()
 ADD_DEFINITIONS("-DRDK_USE_BOOST_IOSTREAMS")
 endif(RDK_USE_BOOST_IOSTREAMS)
 
-if(RDK_BUILD_COORDGEN_SUPPORT)
+if(RDK_BUILD_MAEPARSER_SUPPORT)
     include_directories(${maeparser_INCLUDE_DIRS})
     set (maesupplier MaeMolSupplier.cpp)
 endif()
@@ -45,7 +45,7 @@ rdkit_library(FileParsers
               ProximityBonds.cpp
               SequenceParsers.cpp SequenceWriters.cpp
               SVGParser.cpp
-              LINK_LIBRARIES Depictor SmilesParse GraphMol ${RDK_COORDGEN_LIBS} ${regex_lib} ${link_iostreams})
+              LINK_LIBRARIES Depictor SmilesParse GraphMol ${RDK_MAEPARSER_LIBS} ${regex_lib} ${link_iostreams})
 
 rdkit_headers(FileParsers.h
               FileParserUtils.h

Code/GraphMol/FileParsers/MolSupplier.h

@@ -22,14 +22,14 @@
 #include <GraphMol/ROMol.h>
 #include <RDGeneral/BadFileException.h>
 
-#ifdef RDK_BUILD_COORDGEN_SUPPORT
+#ifdef RDK_BUILD_MAEPARSER_SUPPORT
 namespace schrodinger {
 namespace mae {
 class Reader;
 class Block;
 }  // namespace mae
 }  // namespace schrodinger
-#endif  // RDK_BUILD_COORDGEN_SUPPORT
+#endif  // RDK_BUILD_MAEPARSER_SUPPORT
 
 namespace RDKit {
 RDKIT_FILEPARSERS_EXPORT std::string strip(const std::string &orig);
@@ -398,7 +398,7 @@ class RDKIT_FILEPARSERS_EXPORT PDBMolSupplier : public MolSupplier {
   bool df_sanitize, df_removeHs, df_proximityBonding;
   unsigned int d_flavor;
 };
-#ifdef RDK_BUILD_COORDGEN_SUPPORT
+#ifdef RDK_BUILD_MAEPARSER_SUPPORT
 //! lazy file parser for MAE files
 class RDKIT_FILEPARSERS_EXPORT MaeMolSupplier : public MolSupplier {
   /**
@@ -435,7 +435,7 @@ class RDKIT_FILEPARSERS_EXPORT MaeMolSupplier : public MolSupplier {
   std::shared_ptr<std::istream> dp_sInStream;
   std::string d_stored_exc;
 };
-#endif  // RDK_BUILD_COORDGEN_SUPPORT
+#endif  // RDK_BUILD_MAEPARSER_SUPPORT
 }  // namespace RDKit
 
 #endif

Code/GraphMol/FileParsers/testMolSupplier.cpp

@@ -103,7 +103,7 @@ int testMolSup() {
     }
     TEST_ASSERT(i == 16);
   }
-#ifdef RDK_BUILD_COORDGEN_SUPPORT
+#ifdef RDK_BUILD_MAEPARSER_SUPPORT
   {  // Test reading properties
     fname = rdbase + "/Code/GraphMol/FileParsers/test_data/props_test.mae";
 
@@ -283,7 +283,7 @@ int testMolSup() {
     TEST_ASSERT(info->getChainId() == "A");
     TEST_ASSERT(info->getResidueNumber() == 5);
   }
-#endif  // RDK_BUILD_COORDGEN_SUPPORT
+#endif  // RDK_BUILD_MAEPARSER_SUPPORT
   return 1;
 }
 
@@ -2281,7 +2281,7 @@ int testForwardSDSupplier() {
   }
 #endif
 
-#ifdef RDK_BUILD_COORDGEN_SUPPORT
+#ifdef RDK_BUILD_MAEPARSER_SUPPORT
   // Now test that Maestro parsing of gz files works
   std::string maefname =
       rdbase + "/Code/GraphMol/FileParsers/test_data/NCI_aids_few.mae";
@@ -2335,7 +2335,7 @@ int testForwardSDSupplier() {
     }
     TEST_ASSERT(i == 16);
   }
-#endif  // RDK_BUILD_COORDGEN_SUPPORT
+#endif  // RDK_BUILD_MAEPARSER_SUPPORT
 
   return 1;
 }
@@ -2685,7 +2685,7 @@ M  END
   }
 }
 
-#ifdef RDK_BUILD_COORDGEN_SUPPORT
+#ifdef RDK_BUILD_MAEPARSER_SUPPORT
 void testGitHub2881() {
   std::string data = R"DATA(f_m_ct { 
  s_m_title

Code/GraphMol/Wrap/CMakeLists.txt

@@ -21,7 +21,7 @@ rdkit_python_extension(rdqueries
                        LINK_LIBRARIES
                        GraphMol  )
 
-if(RDK_BUILD_COORDGEN_SUPPORT)
+if(RDK_BUILD_MAEPARSER_SUPPORT)
     set (maesupplier MaeMolSupplier.cpp)
     include_directories(${maeparser_INCLUDE_DIRS})
 endif()

Code/GraphMol/Wrap/rdmolfiles.cpp

@@ -402,7 +402,7 @@ void wrap_sdsupplier();
 void wrap_forwardsdsupplier();
 void wrap_tdtsupplier();
 void wrap_smisupplier();
-#ifdef RDK_BUILD_COORDGEN_SUPPORT
+#ifdef RDK_BUILD_MAEPARSER_SUPPORT
 void wrap_maesupplier();
 #endif
 
@@ -1456,7 +1456,7 @@ BOOST_PYTHON_MODULE(rdmolfiles) {
   wrap_forwardsdsupplier();
   wrap_tdtsupplier();
   wrap_smisupplier();
-#ifdef RDK_BUILD_COORDGEN_SUPPORT
+#ifdef RDK_BUILD_MAEPARSER_SUPPORT
   wrap_maesupplier();
 #endif
   // wrap_pdbsupplier();

Code/GraphMol/Wrap/rough_test.py

@@ -5923,12 +5923,13 @@ M  END
         Chem.SDMolSupplier,
         Chem.TDTMolSupplier,
         #Chem.CompressedSDMolSupplier,
-        Chem.MaeMolSupplier
     ]:
       print("supplier:", supplier)
       with self.assertRaises(OSError):
         suppl = supplier(d)
-
+    if hasattr(Chem, 'MaeMolSupplier'):
+      with self.assertRaises(OSError):
+        suppl = Chem.MaeMolSupplier(d)
     os.rmdir(d)
 
   def testRandomSmilesVect(self):

Code/RDGeneral/RDConfig.h.cmake

@@ -34,6 +34,8 @@
 
 #cmakedefine RDK_BUILD_COORDGEN_SUPPORT
 
+#cmakedefine RDK_BUILD_MAEPARSER_SUPPORT
+
 #cmakedefine RDK_BUILD_AVALON_SUPPORT
 
 #cmakedefine RDK_BUILD_INCHI_SUPPORT

External/CoordGen/CMakeLists.txt

@@ -1,14 +1,10 @@
 add_custom_target(coordgen_support ALL)
+include(RDKitUtils)
 
-if(RDK_BUILD_COORDGEN_SUPPORT)
-  include(RDKitUtils)
-
-  if(MSVC AND (NOT RDK_INSTALL_DLLS_MSVC))
-    add_definitions(-DSTATIC_COORDGEN)
-  else(MSVC AND (NOT RDK_INSTALL_DLLS_MSVC))
+if(RDK_BUILD_MAEPARSER_SUPPORT)
+  if((NOT MSVC) OR RDK_INSTALL_DLLS_MSVC)
     add_definitions(-DIN_MAEPARSER)
-    add_definitions(-DIN_COORDGEN)
-  endif(MSVC AND (NOT RDK_INSTALL_DLLS_MSVC))
+  endif()
 
   find_package(Boost ${RDK_BOOST_VERSION} COMPONENTS system iostreams REQUIRED)
   set (link_iostreams ${Boost_LIBRARIES})
@@ -61,9 +57,26 @@ if(RDK_BUILD_COORDGEN_SUPPORT)
 
   endif(MAEPARSER_FORCE_BUILD OR (NOT maeparser_FOUND))
 
+  include_directories(${maeparser_INCLUDE_DIRS})
+
+  set(RDK_MAEPARSER_LIBS ${maeparser_LIBRARIES} ${Boost_LIBRARIES}
+      CACHE STRING "the external libraries" FORCE)
+
+else (RDK_BUILD_MAEPARSER_SUPPORT)
+
+  set(RDK_MAEPARSER_LIBS CACHE STRING "the external libraries" FORCE)
+  
+endif(RDK_BUILD_MAEPARSER_SUPPORT)
+
+if(RDK_BUILD_COORDGEN_SUPPORT)
+  if(MSVC AND (NOT RDK_INSTALL_DLLS_MSVC))
+    add_definitions(-DSTATIC_COORDGEN)
+  else(MSVC AND (NOT RDK_INSTALL_DLLS_MSVC))
+    add_definitions(-DIN_COORDGEN)
+  endif(MSVC AND (NOT RDK_INSTALL_DLLS_MSVC))
+
   find_package(coordgen MODULE)
   if(COORDGEN_FORCE_BUILD OR (NOT coordgen_FOUND))
-
     if(NOT DEFINED COORDGEN_DIR)
       set(COORDGEN_DIR "${CMAKE_CURRENT_SOURCE_DIR}/coordgen")
     endif()
@@ -85,19 +98,15 @@ if(RDK_BUILD_COORDGEN_SUPPORT)
     set(coordgen_INCLUDE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}
         CACHE STRING "CoordGen Include File" FORCE)
     file(GLOB CGSOURCES "${COORDGEN_DIR}/*.cpp")
-    rdkit_library(coordgen ${CGSOURCES} SHARED LINK_LIBRARIES ${maeparser_LIBRARIES})
+    rdkit_library(coordgen ${CGSOURCES} SHARED )
     install(TARGETS coordgen DESTINATION ${RDKit_LibDir})
     set(coordgen_LIBRARIES coordgen)
 
   endif(COORDGEN_FORCE_BUILD OR (NOT coordgen_FOUND))
 
-  include_directories(${maeparser_INCLUDE_DIRS})
   include_directories(${coordgen_INCLUDE_DIRS})
 
-  install(FILES DESTINATION ${RDKit_ShareDir}/Data
-          COMPONENT data)
-
-  set(RDK_COORDGEN_LIBS MolAlign ${coordgen_LIBRARIES} ${maeparser_LIBRARIES} ${Boost_LIBRARIES}
+  set(RDK_COORDGEN_LIBS MolAlign ${coordgen_LIBRARIES} ${RDK_MAEPARSER_LIBS} ${Boost_LIBRARIES}
       CACHE STRING "the external libraries" FORCE)
   rdkit_headers(CoordGen.h DEST GraphMol)