Check consistency of data request content

[rigoudyg]

Hi @CMIP-IPO @JamesAnstey

I have written a script to check consistency of data request content.
It is in the following branch (no pull request for now at it is writtent on the optimized branch not yet merged): https://github.com/rigoudyg/CMIP7_DReq_Software/tree/gr_check_attributes

It checked that, for all variables derived on the same physical parameters, that some attributes have the same values and are not missing.
The outputs are the following:
check_attributes.json
check_attributes.log
Some attributes can be left aside (frequencies, names, modelling realms) but are present because they give information on the variable which have the issues.
For others, such as description or title, they can give information on inconsistencies (case issues). Maybe, creating template for values which must be consistent but with only a change in frequency or area could be useful).

I have an other related question.
Should variables have a single value for attributes 'cell_measures' and 'structure_title' or could those attributes have several values for a single variable?

Best,
Gaëlle

[rigoudyg]

In the log file, you can look for:

• "the following attributes have missing values": to find missing values
• "the following attributes have different values": to find attributes with several (possibly inconsistent) values

[JamesAnstey]

Thanks @rigoudyg, this should be useful in cleaning up the request content for v1.2.1.

For title/description consistency is good but I wonder how much can be automatically checked as these are (I think) meant just to be human-readable info. But even just flagging inconsistencies could be useful if it leads to us manually correcting some.

For cell_measures I think we can have something like "cell_measures": "area: areacello volume: volcello", is that what you mean by multiple values?

[rigoudyg]

Update of logs and issues

check_attributes.json
check_attributes.log

[rigoudyg]

Hi,

The script is now available in main branch.
Bugfixes will come with #95
Here are the check attributes outputs for both raw and release exports and version 1.2.

check_attributes_release.json
check_attributes_release.log
check_attributes_raw.json
check_attributes_raw.log

Best,

[rigoudyg]

Update with v1.2.1

check_attributes_release.json
check_attributes_release.log

[rigoudyg]

Update with draft of v1.2.2

check_attributes_release.json
check_attributes_release.log