check_structural_metadata.py can be used to perform basic validation of the structural metadata in Classification.csv, Category.csv and Category_Mapping.csv.
It is not a full QA suite and only performs limited checks.
Primarily it checks the consistency of codes and labels for mappings between a parent classification and a derived classification.
It is run by invoking the script with python3 (some systems may require the use of python instead),
and supplying an input directory with the -i option:
python3 check_structural_metadata.py -i <input_directory>
The repository contains some sample source files that are used in testing. To run the script against these files:
python3 check_structural_metadata.py -i test/data/bad
The output provides descriptions of the checks being performed and detailed information about any errors detected.
All classification codes are strings, although many codes are string representations of numbers.
This means that 001 is a different code from 1.
Sometimes issues can arise where usage of leading zeros is inconsistent.
A --zeros flag can be used to ignore these inconsistencies
i.e. 001 and 1 will be treated as being equivalent.
This flag is only intended for debug purposes as each code must be specified with the correct format wherever it is used.
To run checks with leading zeros ignored:
python3 check_structural_metadata.py -i <input_directory> --zeros
The repository contains some simple tests that can be used to validate that the checks behave as expected. To run:
python3 -m unittest -v