Add a tool to merge several podio files into a single one (#681)

* Add a tool to merge several podio files into a single one

* Generate a metadata frame if it doesn't exist

* Add configuration for the metadata parameter name

* Hardcode the metadata parameters

---------

Co-authored-by: jmcarcell <[email protected]>

tools/CMakeLists.txt

@@ -3,6 +3,7 @@ install(PROGRAMS ${CMAKE_CURRENT_LIST_DIR}/podio-vis DESTINATION ${CMAKE_INSTALL
 if(ENABLE_RNTUPLE)
   install(PROGRAMS ${CMAKE_CURRENT_LIST_DIR}/podio-ttree-to-rntuple DESTINATION ${CMAKE_INSTALL_BINDIR})
 endif()
+install(PROGRAMS ${CMAKE_CURRENT_LIST_DIR}/podio-merge-files DESTINATION ${CMAKE_INSTALL_BINDIR})
 
 # Add a very basic test of podio-vis
 if(BUILD_TESTING)

tools/podio-merge-files

@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+"""podio-merge-files tool to merge any number of podio files into one"""
+
+import argparse
+import sys
+import podio
+import podio.root_io
+from podio import reading
+
+parser = argparse.ArgumentParser(
+    description="Merge any number of podio files into one, can merge TTree and RNTuple files"
+)
+
+parser.add_argument("--output-file", help="name of the output file", required=True)
+parser.add_argument("files", nargs="+", help="which files to merge")
+parser.add_argument(
+    "--metadata",
+    choices=["none", "all", "first"],
+    default="first",
+    help="metadata to include in the output file, default: "
+    "only the one from the first file, other options: all files, none",
+)
+args = parser.parse_args()
+
+all_files = set()
+for f in args.files:
+    if f in all_files:
+        raise ValueError(f"File {f} is present more than once in the input list")
+    all_files.add(f)
+
+ROOT_FORMAT = reading._determine_root_format(args.files[0])  # pylint: disable=protected-access
+if ROOT_FORMAT == reading.RootFileFormat.TTREE:
+    reader = podio.root_io.Reader(args.files)
+    writer = podio.root_io.Writer(args.output_file)
+elif ROOT_FORMAT == reading.RootFileFormat.RNTUPLE:
+    reader = podio.root_io.RNTupleReader(args.files)
+    writer = podio.root_io.RNTupleWriter(args.output_file)
+else:
+    raise ValueError(f"Input file {args.files[0]} is not a TTree or RNTuple file")
+
+categories = list(reader.categories)
+is_metadata_available = True  # pylint: disable=invalid-name
+try:
+    # All frames will be copied as they are except the metadata ones
+    categories.remove("metadata")
+except ValueError:
+    is_metadata_available = False  # pylint: disable=invalid-name
+
+for category in categories:
+    all_frames = reader.get(category)
+    for frame in all_frames:
+        writer.write_frame(frame, category)
+
+if args.metadata == "none":
+    sys.exit(0)
+
+if not is_metadata_available:
+    print("Warning: metadata category 'metadata' not found in the input files, it will be created")
+    all_frames = [podio.Frame()]
+else:
+    if args.metadata == "first":
+        all_frames = [reader.get("metadata")[0]]
+    else:
+        all_frames = reader.get("metadata")
+for frame in all_frames:
+    frame.put_parameter("MergedInputFiles", args.files)
+    writer.write_frame(frame, "metadata")