ENH: template indicators with jinja

[abel-bzz]

Pull Request Checklist:

• This PR addresses an already opened issue (for bug fixes / features)
  • This PR fixes Conditional string formatting in indicator metadata fields #1174
  • This PR fixes per-gridcell heat wave indice calculations? #1093
  • This PR fixes Inconsistent usage of "degC" and "°C" in Indicator metadata #890
  • This PR fixes #XYZ
• Tests for the changes have been added (for bug fixes / features)
  • (If applicable) Documentation has been added / updated (for bug fixes / features)
• HISTORY.rst has been updated (with summary of main changes)
  • Link to issue (:issue:number) and pull request (:pull:number) has been added
• The relevant author information has been added to .zenodo.json

What kind of change does this PR introduce?

Follow up to discussions in here: #1093
This pr is a proof of concept to get an idea of how jinja can be used to format indicator descriptions.

Does this PR introduce a breaking change?

No, it should not.

Other information:

To illustrate the results with heat_wave_frequency:

ds = xr.open_dataset("/Users/aoun/workspace/xclim/climpact.sampledata.gridded.1991-2010.nc")
res = xclim.atmos.heat_wave_frequency(tasmin = ds.tmin,tasmax=ds.tmax)
print(res.attrs["description"])

prints
"Ys number of heat wave events over a given period. an event occurs when the minimum and maximum daily temperature both exceeds specific thresholds : (tmin > 22.0 degc and tmax > 30 degc)over a minimum number of days (3)."

[abel-bzz]

A bit ugly, maybe there is a better way to add numpy in jinja rendering namespace

[abel-bzz]

Here is the template for heat_wave_frequency

[abel-bzz]

I tried to mitigate the changes by applying both AttrFormatter and jinja templating.
The effects are

• jinja control structures are interpreted, ignoring values surrounded by single curly brackets.
• values surrounded by single curly brackets are formatted as usual. (e.g. {freq} -> "Annual")

So, this

res = xclim.atmos.heat_wave_frequency(tasmin=ds.tmin,
                                      tasmax=ds.tmax,
                                      thresh_tasmin=ds.tmin.isel(time=0),
                                      thresh_tasmax=ds.tmax.isel(time=0))
print(res.attrs["description"])

prints

Annual number of heat wave events over a given period. an event occurs when the minimum and maximum daily 
temperature both exceeds specific thresholds : 
(tmin > per_gridcell_tmin_thresholds and tmax > per_gridcell_tmax_thresholds) over a minimum number of days (3).

[coveralls]

Coverage increased (+0.006%) to 91.377% when pulling 1a596f7 on enh/add_jinja_for_indicator_templating into 4d5dd03 on master.

[abel-bzz]

For the record, I'm working (in icclim) on generic indicators which would fully utilize jinja templating to generate rich metadata.
I hope it won't be too hard for me to upstream this work into xclim once it's in a stable state.

[abel-bzz]

Unfortunately, I don't think I will be able to work much more on this PR (or generic indicators in xclim), my contract at cerfacs ends soon and I don't know yet if I will continue working on climate indices afterward.
So feel free to take over or close this PR!
I could present you what I did in icc in case you want port it to xclim though.

[huard]

Hi Abel, sorry to hear that, we really enjoyed your contributions to the project.

Yes, it would be useful for us to get a sense of where things stand so we can work on the xclim port. Let us know when is a good time for you.