Add documentation for usage of "extra_types"

[ollie-bell]

The parse function docstring mentions documentation for extra_types argument, but I cannot find it. Can it be added?

The default types are not sufficient for my needs. e.g. I would like to be able to enforce letters and numbers ONLY in a field of the format string (currently there is one for letters, numbers and underscores which doesn't work for me, as my format strings are all underscore delimited). I also have datetime strings of format %Y%m%d which doesn't seem to be supported for datetime parsing.

[jenisys]

The module docstring indirectly provides examples how to use it, in the @parse.with_pattern decorator section (but it does not mention the parameter extra_types by name, it just uses the parameter in function calls).

SEE ALSO:

• parse.py:326
• tests: test_parse.py (extra_types is explicitly used as named parameter)

OTHERWISE:
Provide your own types / type-converters via extra_types.

[jenisys]

@ollie-bell
The ti type (ISO date or ISO timestamp, like: 2022-11-26 or 2022-11-26T12:04:23) comes pretty close to what you need.

[ollie-bell]

@jenisys thanks for pointing that out to me. Tbh I'm not really experienced with string manipulation and regex etc. so I don't understand the usage very well. I'm hoping someone can just help me out with a question :D

I have formats like

"{variable}_{domain}_{GCMsource}_{scenario}_{member}_{RCMsource}_{RCMversion}_{frequency}_{start:d}-{end:d}.nc"
"{variable}_{frequency}_{source}_{scenario}_{member}_{grid}_{start:d}-{end:d}.nc"

and strings to parse like

"pr_EUR-11_CNRM-CERFACS-CNRM-CM5_rcp45_r1i1p1_CLMcom-CCLM4-8-17_v1_day_20060101-20101231.nc"
"pr_day_MIROC-ESM-CHEM_rcp45_r1i1p1_gr1_20310101-20310101.nc"

As you can see, each {field} in the format is made up of a mix of letters, numbers and dashes (-). Underscores are always used to separate fields and never within the fields themselves.

I'd like to add a format specification within each {field} which constrains it to letters, numbers and dashes only. Curently the :w specification does letters, numbers and underscore.

Thanks!