using podio::DataSource

HEP-FCC · Aug 28, 2024 · c9da963 · c9da963
1 parent 62f833b
commit c9da963
Show file tree

Hide file tree

Showing 7 changed files with 66 additions and 39 deletions.
diff --git a/analyzers/dataframe/CMakeLists.txt b/analyzers/dataframe/CMakeLists.txt
@@ -69,6 +69,8 @@ target_link_libraries(FCCAnalyses PUBLIC
                       EDM4HEP::edm4hep
                       EDM4HEP::edm4hepDict
                       podio::podio
+                      podio::podioRootIO
+                      podio::podioDataSource
                       ${ADDONS_LIBRARIES}
                       ${DELPHES_LIBRARY}
                       gfortran # todo: why necessary?

diff --git a/examples/data_source/standalone.py b/examples/data_source/standalone.py
@@ -18,9 +18,9 @@ def main():
         print('----> Debug: Found FCCAnalyses library.')
     print("----> Info: Loading analyzers from libFCCAnalyses... ",)
 
-    if ROOT.podio.ROOTReader():
-        print('----> Debug: Found Podio ROOT I/O.')
-    print('----> Info: Loading events through EDM4hep RDataSource...')
+    if ROOT.podio.DataSource:
+        print('----> Debug: Found Podio ROOT DataSource.')
+    print('----> Info: Loading events through podio::DataSource...')
 
     try:
         dframe = ROOT.podio.CreateDataFrame(input_list)

diff --git a/python/run_analysis.py b/python/run_analysis.py
@@ -316,18 +316,19 @@ def run_rdf(rdf_module,
     Create RDataFrame and snapshot it.
     '''
     if args.use_data_source:
-        if ROOT.podio.ROOTReader():
-            LOGGER.debug('Found Podio ROOT I/O.')
+        if ROOT.podio.DataSource:
+            LOGGER.debug('Found PODIO ROOT DataSource.')
         else:
-            LOGGER.error('Podio ROOT I/O library not found!\nAborting...')
+            LOGGER.error('PODIO ROOT DataSource library not found!\n'
+                         'Aborting...')
             sys.exit(3)
-        LOGGER.info('Loading events through podio::ROOTDataSource...')
+        LOGGER.info('Loading events through podio::DataSource...')
 
         try:
             dframe = ROOT.podio.CreateDataFrame(input_list)
         except TypeError as excp:
-            LOGGER.error('Unable to build dataframe using'
-                         'podio::RDataSource!\n%s', excp)
+            LOGGER.error('Unable to build dataframe using '
+                         'podio::DataSource!\n%s', excp)
             sys.exit(3)
     else:
         dframe = ROOT.RDataFrame("events", input_list)
@@ -382,7 +383,7 @@ def send_to_batch(rdf_module, chunk_list, process, anapath: str):
                                 stderr=subprocess.DEVNULL
                                 )
     except subprocess.CalledProcessError:
-        LOGGER.error('The FCCanalyses libraries are not properly build and '
+        LOGGER.error('The FCCAnalyses libraries are not properly build and '
                      'installed!\nAborting job submission...')
         sys.exit(3)
 
@@ -538,7 +539,7 @@ def run_local(rdf_module, infile_list, args):
     start_time = time.time()
     inn, outn = run_rdf(rdf_module, file_list, outfile_path, args)
     elapsed_time = time.time() - start_time
-    
+
     # replace nevents_local by inn = the amount of processed events
 
     info_msg = f"{' SUMMARY ':=^80}\n"
@@ -611,6 +612,10 @@ def run_stages(args, rdf_module, anapath):
     if not os.path.exists(output_dir_eos) and output_dir_eos:
         os.system(f'mkdir -p {output_dir_eos}')
 
+    # Check whether to use PODIO ROOT DataSource to load the events
+    if get_element(rdf_module, "useDataSource", False):
+        args.use_data_source = True
+
     # Check if test mode is specified, and if so run the analysis on it (this
     # will exit after)
     if args.test:
@@ -733,6 +738,10 @@ def run_histmaker(args, rdf_module, anapath):
     do_scale = get_element(rdf_module, "doScale", True)
     int_lumi = get_element(rdf_module, "intLumi", True)
 
+    # Check whether to use PODIO ROOT DataSource to load the events
+    if get_element(rdf_module, "useDataSource", False):
+        args.use_data_source = True
+
     # check if the process list is specified, and create graphs for them
     process_list = get_element(rdf_module, "processList")
     graph_function = getattr(rdf_module, "build_graph")
@@ -804,18 +813,19 @@ def run_histmaker(args, rdf_module, anapath):
         LOGGER.info(info_msg)
 
         if args.use_data_source:
-            if ROOT.podio.ROOTReader():
-                LOGGER.debug('Found Podio ROOT I/O.')
+            if ROOT.podio.DataSource:
+                LOGGER.debug('Found Podio ROOT DataSource.')
             else:
-                LOGGER.error('Podio ROOT I/O library not found!\nAborting...')
+                LOGGER.error('Podio ROOT DataSource library not found!'
+                             '\nAborting...')
                 sys.exit(3)
-            LOGGER.info('Loading events through podio::ROOTDataSource...')
+            LOGGER.info('Loading events through podio::DataSource...')
 
             try:
                 dframe = ROOT.podio.CreateDataFrame(file_list_root)
             except TypeError as excp:
-                LOGGER.error('Unable to build dataframe using EDM4hep '
-                             'RDataSource!\n%s', excp)
+                LOGGER.error('Unable to build dataframe using '
+                             'podio::DataSource!\n%s', excp)
                 sys.exit(3)
         else:
             dframe = ROOT.ROOT.RDataFrame("events", file_list_root)
@@ -998,14 +1008,8 @@ def run(parser):
                      err)
         sys.exit(3)
 
-    # Merge command line arguments with anascript
-    # Check whether to use RDataSource to load the events
-    use_data_source = get_element(rdf_module, "useDataSource")
-    if use_data_source:
-        args.use_data_source = True
-
     # Merge configuration from analysis script file with command line arguments
-    if get_element(rdf_module, 'graph'):
+    if get_element(rdf_module, 'graph', False):
         args.graph = True
 
     if get_element(rdf_module, 'graphPath') != '':

diff --git a/python/run_fccanalysis.py b/python/run_fccanalysis.py
@@ -309,7 +309,23 @@ def run_rdf(args,
     Run the analysis ROOTDataFrame and snapshot it.
     '''
     # Create initial dataframe
-    dframe = ROOT.RDataFrame("events", input_list)
+    if args.use_data_source:
+        if ROOT.podio.DataSource:
+            LOGGER.debug('Found PODIO ROOT DataSource.')
+        else:
+            LOGGER.error('PODIO ROOT DataSource library not found!\n'
+                         'Aborting...')
+            sys.exit(3)
+        LOGGER.info('Loading events through podio::DataSource...')
+
+        try:
+            dframe = ROOT.podio.CreateDataFrame(input_list)
+        except TypeError as excp:
+            LOGGER.error('Unable to build dataframe using '
+                         'podio::DataSource!\n%s', excp)
+            sys.exit(3)
+    else:
+        dframe = ROOT.RDataFrame("events", input_list)
 
     # Limit number of events processed
     if args.nevents > 0:
@@ -469,7 +485,8 @@ def run_local(args, analysis, infile_list):
     nevents_local = 0
     for filepath in infile_list:
 
-        filepath = apply_filepath_rewrites(filepath)
+        if not args.use_data_source:
+            filepath = apply_filepath_rewrites(filepath)
 
         file_list.push_back(filepath)
         info_msg += f'- {filepath}\t\n'
@@ -590,6 +607,10 @@ def run_fccanalysis(args, analysis_module):
     if output_dir_eos is not None and not os.path.exists(output_dir_eos):
         os.system(f'mkdir -p {output_dir_eos}')
 
+    # Check whether to use PODIO ROOT DataSource to load the events
+    if get_element(analysis, "use_data_source", False):
+        args.use_data_source = True
+
     # Check if test mode is specified, and if so run the analysis on it (this
     # will exit after)
     if args.test:

diff --git a/tests/CMakeLists.txt b/tests/CMakeLists.txt
@@ -8,44 +8,44 @@ add_integration_test("examples/FCCee/test/jet_constituents.py")
 add_integration_test("examples/FCCee/vertex_lcfiplus/analysis_V0.py")
 
 if(WITH_SOURCE)
-  add_integration_test("e4hsource/test/histmaker_source.py")
-  add_integration_test("e4hsource/test/stages_source.py")
-  add_integration_test("e4hsource/test/analysis_stage1.py")
-  add_standalone_test("e4hsource/test/standalone.py")
+  add_integration_test("examples/data_source/histmaker_source.py")
+  add_integration_test("examples/data_source/stages_source.py")
+  add_integration_test("examples/data_source/analysis_stage1.py")
+  add_standalone_test("examples/data_source/standalone.py")
 endif()
 
 add_standalone_test("examples/FCCee/fullSim/caloNtupleizer/analysis.py")
 
 # C++ Analyses
 
-# podio::ROOTDataSource
+# podio::DataSource
 add_executable(test_data_source ${CMAKE_SOURCE_DIR}/tests/data_source/test_source.cxx)
 target_include_directories(test_data_source PRIVATE ${CMAKE_SOURCE_DIR}/include)
-target_link_libraries(test_data_source podio::podioRootIO
+target_link_libraries(test_data_source podio::podioDataSource
                                        EDM4HEP::edm4hep
 )
 
 add_executable(test_source_associations
                ${CMAKE_SOURCE_DIR}/tests/data_source/test_source_associations.cxx)
 target_include_directories(test_source_associations PRIVATE
                            ${CMAKE_SOURCE_DIR}/include)
-target_link_libraries(test_source_associations podio::podioRootIO
+target_link_libraries(test_source_associations podio::podioDataSource
                                                EDM4HEP::edm4hep
                                                FCCAnalyses
 )
 
 # Low level access
 add_executable(test_low_level ${CMAKE_SOURCE_DIR}/tests/data_source/test_low_level.cxx)
 target_link_libraries(test_low_level ROOT::ROOTDataFrame
-                                     podio::podioRootIO
+                                     podio::podioDataSource
                                      EDM4HEP::edm4hep
                                      EDM4HEP::edm4hepDict
 )
 
 add_executable(test_low_level_associations
                ${CMAKE_SOURCE_DIR}/tests/data_source/test_low_level_associations.cxx)
 target_link_libraries(test_low_level_associations ROOT::ROOTDataFrame
-                                                  podio::podioRootIO
+                                                  podio::podioDataSource
                                                   EDM4HEP::edm4hep
                                                   EDM4HEP::edm4hepDict
                                                   FCCAnalyses

diff --git a/tests/data_source/test_source.cxx b/tests/data_source/test_source.cxx
@@ -5,7 +5,7 @@
 #include <TCanvas.h>
 
 // PODIO
-#include <podio/ROOTDataSource.h>
+#include <podio/DataSource.h>
 
 // EDM4hep
 #include <edm4hep/ReconstructedParticle.h>
@@ -125,7 +125,7 @@ int main(int argc, char *argv[]) {
 
   ROOT::EnableImplicitMT(nCPU);
 
-  ROOT::RDataFrame rdf(std::make_unique<podio::ROOTDataSource>(filePathList));
+  ROOT::RDataFrame rdf(std::make_unique<podio::DataSource>(filePathList));
 
   // rdf.Describe().Print();
   std::cout << std::endl;

diff --git a/tests/data_source/test_source_associations.cxx b/tests/data_source/test_source_associations.cxx
@@ -14,7 +14,7 @@
 #include <TMatrixDSym.h>
 
 // PODIO
-#include <podio/ROOTDataSource.h>
+#include <podio/DataSource.h>
 
 // EDM4hep
 #include <edm4hep/MCParticle.h>
@@ -256,7 +256,7 @@ int main(int argc, char *argv[]) {
 
   ROOT::EnableImplicitMT(nCPU);
 
-  ROOT::RDataFrame rdf(std::make_unique<podio::ROOTDataSource>(filePathList));
+  ROOT::RDataFrame rdf(std::make_unique<podio::DataSource>(filePathList));
 
   // rdf.Describe().Print();
   std::cout << std::endl;