python.checkMetadata Namespace Reference

Classes
class	ValidateMetadataAlg

Functions
	validateInputMetadata (infile)
	checkFileMetaData (file_names)

Variables
	flags = initConfigFlags()
	EventPrintoutInterval
	metadata = defaultdict(list)
	cfg = MainServicesCfg(flags)
	sequenceName
	sc = cfg.run()

Function Documentation

checkFileMetaData()

python.checkMetadata.checkFileMetaData

(

file_names

)

Check if FileMetaData is in all files

Definition at line 130 of file checkMetadata.py.

def checkFileMetaData(file_names):
    """Check if FileMetaData is in all files"""
 
    return all(
        [
            fmd_items
            for file_name in file_names
            for fmd_items in [
                value
                for _, value in GetFileMD(file_name)["metadata_items"].items()
                if "FileMetaData" in value
            ]
        ]
    )
 
 

validateInputMetadata()

python.checkMetadata.validateInputMetadata

(

infile

)

Definition at line 62 of file checkMetadata.py.

def validateInputMetadata(infile):
    logging.info(f"Using input file {infile}")
 
    current_file = ROOT.TFile(infile)
    md = read_metadata(infile, mode="full", unique_tag_info_values=False)
 
    nevts_esi = next(
        (
            md[infile][key]["numberOfEvents"]
            for key, value in md[infile].items()
            if isinstance(value, dict) and "numberOfEvents" in value
        ),
        None,
    )
 
    dataHeaderTree = current_file.Get(PoolOpts.TTreeNames.DataHeader)
    if isinstance(dataHeaderTree, ROOT.TTree):
        nevts_dh = dataHeaderTree.GetEntriesFast()
    else:
        if (
            current_file.GetListOfKeys().Contains(PoolOpts.RNTupleNames.DataHeader)
            and ROOT.gROOT.GetVersionInt() < 63100
        ):
            raise RuntimeError(
                "ROOT ver. 6.31/01 or greater needed to read RNTuple files"
            )
        dataHeaderRNT = current_file.Get(PoolOpts.RNTupleNames.DataHeader)
        if isRNTuple( dataHeaderRNT ):
            nevts_dh = ROOT.Experimental.RNTupleReader.Open(dataHeaderRNT).GetNEntries()
        else:
            nevts_dh = None
 
    if not (md[infile]["nentries"] == nevts_esi == nevts_dh):
        logging.error(
            "Number of events from EventStreamInfo inconsistent with number of entries in DataHeader"
        )
        return 1
 
    tag_info = md[infile]["/TagInfo"]
    if "project_name" in tag_info and isinstance(tag_info["project_name"], list):
        if "IS_SIMULATION" in tag_info["project_name"] and any(
            [item for item in tag_info["project_name"] if item.startswith("data")]
        ):
            logging.error("/TagInfo contains values reserved for both MC and data")
            return 1
        if (
            len(
                set(
                    item[5:6]
                    for item in tag_info["project_name"]
                    if item.startswith("data")
                )
            )
            > 1
        ):
            logging.error("/TagInfo contains values from different data taking periods")
            return 1
    if (
        "data_year" in tag_info
        and isinstance(tag_info["data_year"], list)
        and len(set(tag_info["data_year"])) > 1
    ):
        logging.error("/TagInfo contains values from different data taking periods")
        return 1
 
    return 0
 
 

Variable Documentation

cfg

python.checkMetadata.cfg = MainServicesCfg(flags)

Definition at line 184 of file checkMetadata.py.

EventPrintoutInterval

python.checkMetadata.EventPrintoutInterval

Definition at line 158 of file checkMetadata.py.

flags

python.checkMetadata.flags = initConfigFlags()

Definition at line 157 of file checkMetadata.py.

metadata

python.checkMetadata.metadata = defaultdict(list)

Definition at line 175 of file checkMetadata.py.

sc

python.checkMetadata.sc = cfg.run()

Definition at line 198 of file checkMetadata.py.

sequenceName

python.checkMetadata.sequenceName

Definition at line 194 of file checkMetadata.py.