Have brought in a set of CVs dictionaries for required fields -- process followed was

(First commit)

• took "CMIP6Plus_CV.json" file produced when dividing CMIP6 CVs into generic_CVs for Prototype mip tables PCMDI/mip-cmor-tables#3, which included all fields that did not go into the generic_CV.json file
• Split that file out into its components, one file per term using src/Deconstruction.ipynb notebook
  • note that the original CMIP6Plus_CV file is not in this repository
• wrote src/construct_cmip6plus_cv.py to reconstruct the CMIP6Plus_CV.json file from the individual JSON files. Includes a list of all MD5 checksums (see note below)

(second commit)

• The experiment id and source id information in the CV file embedded in PCMDI/cmip6-cmor-tables have cut down versions of the information in the WCRP-CMIP/CMIP6_CVs. Replaced the CMIP6Plus experiment_id and source_id files from the CMIP6_CV repository and added md5 checksums.
• modified src/construct_cmip6plus_cv.py to correctly construct source_id and experiment_id entries in the CV JSON file.
• Reconstructed the CMIP6Plus_CV.json file

The construction script should now be able to take any updated JSON files for the individual terms and reconstruct the part needed to drive CMOR.

In the src/ directory there is a cv_checksums module that contains functions for setting and validating the md5 checksums.

To be clear: The functions to handle MD5 checksums have been copied from other tools. The functions look at the dictionaries they are given, remove the original checksum entry, compute the checksum of the JSON string representing the dictionary and then adds this to the dictionary. As such these are a reasonably robust method of monitoring for manual changes, which is what I've used it for in the past elsewhere when I don't want a user manually editing carefully controlled information.