Support Non-Standard Formatting in ParaFrame

[rndsrc]

In EHT simulation libraries, the spin values are usually labeled by a-0.94, a-0.5, a0, a+0.5, and a+0.94. Unfortunately, this is not standard python number formatting because

f'{0.5:g}' -> '0.5'
f'{0.0:g}' -> '0'
f'{-.5:g}' -> '-0.5'
f'{0.5:+g}' -> '+0.5'
f'{0.0:+g}' -> '+0'
f'{-.5:+g}' -> '-0.5'

While creating the EHT convention programmatically is not too complicated (with condition statements), parsing with hallmark.ParaFrame is only supported if the spin values are treated as strings, which is annoying for filtering.

It will be great if we can create a pythonic mechanism to pass in special formatting functions.

[rndsrc]

To make this pythonic, it is likely that we will use python string Formatter. Note that we want the same mechanism to work for both converting list of parameters to string (and hence file paths) and vice versa to parse file paths back to parameters. This is because in hallmark-next, we hope to be able to reorganize data on disks.

[rndsrc]

We may address this issue by thinking about what python features we may take advantage of, and how the configuration file may look like.

For python features, I think we can support at least dict-based map, i.e.,

Feature 1: use dict to map strings to strings, e.g.,

fmt:{mag}a{aspin:+g:spin_UIUC}_w{win:d}/img_s{snapshot:d}_Rh{Rhigh:d}_i{inc:d}.h5
formatter:
    spin_UIUC: {'+0':'0'} # => turn into spin_UIUC['+0']

and regular expression based map, i.e.,

Feature 2: use re to substitute sub-string based on regular expression,

fmt:{mag}a{aspin:+g:spin_UIUC}_w{win:d}/img_s{snapshot:d}_Rh{Rhigh:d}_i{inc:d}.h5
formatter:
    spin_UIUC: {pattern:repl} # => turn into re.sub(pattern, repl, ...)

see https://docs.python.org/3/library/re.html#re.sub

These can be described in the YAML file as above, or use a more structure way:

fmt:{mag}a{aspin:+g:spin_UIUC}_w{win:d}/img_s{snapshot:d}_Rh{Rhigh:d}_i{inc:d}.h5
formatter:
    - name: spin_UIUC
      method: dict
      map: {'+0':'0'} # => turn into spin_UIUC['+0']
    - name: spin_UIUC_re
      method: re
      map: {pattern:repl}

Noted that we want the same formatter syntax works for both reading files and writing files. I.e., when use use fmt as a f-string, or when we use fmt with the package parse.