Plotmap aims at replacing operator plot, without relying on Ncl. It uses matplotlib, cartopy and geocat-viz.
This PR brings its code and its doc (which includes two notebooks). It doesn't yet include specific tests for plotmap in the test suite
Also, a new module alllows to use intake catalogs for accessing project's data, for CMIP5, CMIP6, PMIP3, CORDEX
It also includes the changes in V3.1 (w.r.t. to master) as of 20040318. Unfortunately, due to a few heavy commits, rebasing on V3.1 proved intractable
It also includes a series of fixes for running on Jean-Zay at IDRIS
Test suite is OK on spirit and IDRIS
GitHub actions were updated
Here is the content of doc/news.rst :
On Spirit, use intake catalogs, as available at IPSL for some MIPs. First use implies reading the catalog, which can last up to 30s for CMIP6; but next requests are much quicker. Can be deactivated by setting env.environment.projects_using_intake list to []. Current list includes CMIP5, CMIP6, and CORDEX. Project PMIP3 was also included as a test case for introducing a new project natively managed with intake. See ‘internals’ below for details.
Add operator plotmap, a replacement for operator plot, albeit only for maps. It is based on Matplotlib, Cartopy and GeoCat Viz; see :doc:scripts/plotmap`
Function cshow now calls display(Image()) if called from a notebook. This applies to default function cshow (the one in climaf.api)
Function ds() has a new argument 'check' for light to full check of datafiles associated with the dataset. See :py:func:~climaf.classes.ds
Various fixes for running at IDRIS
Fixes errors in : cdu(), iplot_members()
Dataset's method :py:meth:~climaf.classes.cdataset.glob has new argument ensure_period (default is True)
Internals :
New module projects/intake_search.py handles intake catalogs, thanks to:
variables in env.environment.py :
the list of target projects: projects_using_intake and
the catalog path: intake_catalog
for each target project:
a dict translate_facet for matching CliMAF facet names to intake catalog facet names
a string period_pattern which allows to extract the period from the filename; this is needed only because IPSL catalogs do not provide relevant values for that, yet
When calling an external script, operator arguments values are dumped in json format; so, called scripts must decode json format when they accept complex arguments; and must also interpret strings 'true' and 'false' as logical values
Add value 'show' to operator's output format possibe values; in that case, CliMAF doesn't handle any output; the value is forwarded to the script
a number of data samples have been added in examples/data : Nemo, Aladin, (uas, vas)
test suite on GitHub was updated for dependencies (ipython, pyproj, geocat-viz) and for cache and miniconda versions (v4)
Plotmap aims at replacing operator plot, without relying on Ncl. It uses matplotlib, cartopy and geocat-viz. This PR brings its code and its doc (which includes two notebooks). It doesn't yet include specific tests for plotmap in the test suite Also, a new module alllows to use intake catalogs for accessing project's data, for CMIP5, CMIP6, PMIP3, CORDEX
It also includes the changes in V3.1 (w.r.t. to master) as of 20040318. Unfortunately, due to a few heavy commits, rebasing on V3.1 proved intractable
It also includes a series of fixes for running on Jean-Zay at IDRIS
Test suite is OK on spirit and IDRIS GitHub actions were updated
Here is the content of doc/news.rst :
On Spirit, use intake catalogs, as available at IPSL for some MIPs. First use implies reading the catalog, which can last up to 30s for CMIP6; but next requests are much quicker. Can be deactivated by setting env.environment.projects_using_intake list to []. Current list includes CMIP5, CMIP6, and CORDEX. Project PMIP3 was also included as a test case for introducing a new project natively managed with intake. See ‘internals’ below for details.
Add operator plotmap, a replacement for operator plot
, albeit only for maps. It is based on Matplotlib, Cartopy and GeoCat Viz; see :doc:scripts/plotmap`Function cshow now calls display(Image()) if called from a notebook. This applies to default function cshow (the one in climaf.api)
Function ds() has a new argument 'check' for light to full check of datafiles associated with the dataset. See :py:func:
~climaf.classes.dsVarious fixes for running at IDRIS
Fixes errors in : cdu(), iplot_members()
Dataset's method :py:meth:
~climaf.classes.cdataset.globhas new argument ensure_period (default is True)Internals :
New module projects/intake_search.py handles intake catalogs, thanks to:
variables in env.environment.py :
projects_using_intakeandintake_catalogfor each target project:
translate_facetfor matching CliMAF facet names to intake catalog facet namesperiod_patternwhich allows to extract the period from the filename; this is needed only because IPSL catalogs do not provide relevant values for that, yetWhen calling an external script, operator arguments values are dumped in json format; so, called scripts must decode json format when they accept complex arguments; and must also interpret strings 'true' and 'false' as logical values
Add value 'show' to operator's output format possibe values; in that case, CliMAF doesn't handle any output; the value is forwarded to the script
a number of data samples have been added in examples/data : Nemo, Aladin, (uas, vas)
test suite on GitHub was updated for dependencies (ipython, pyproj, geocat-viz) and for cache and miniconda versions (v4)
add function period.build_date_rexep()
fix concurrency issue in makedirs(tmpdir)
create tests/reference_data/test_data_plot/idris_20230611_V3.0_IPSL2