Refactor python code into "reprostim" library to be shipped from pypi

[yarikoptic]

and also packaged for Debian / NeuroDebian (@yarikoptic would do that part).

ATM we have Python libraries and scripts scattered around and without pip installable library so others could

• embed QR codes into their psychopy (or other, e.g. eventually even in matlab - psychtoolbox-3) stimuli scripts at any particular time point -- e.g. beginning of the experiment, specific trial, or end of experiment

• detect and decode QR codes from a provided video file

• similarly embed/decode data from audio stream (WiP, e.g. Research on how to best encode data within audio stream #115 )

• similarly to us run their own "time synchronization" procedures regularly

• library which would provide

  • easy "low level" interfaces to aforementioned functionality (produce QR code images etc)
  • PsychoPy constructs so people could even more easily embed into their experiments

• script to run "synchronization" paradigm

  • presenting QR codes so it could be executed e.g. during regular MRI QA: https://github.com/ReproNim/reprostim/blob/master/tools/reprostim-timesync-stimuli which would be refactored to use our aforementioned higher level constructs.
  • potentially synchronization utilities which we have in https://github.com/ReproNim/reproflow-data-sync/tree/master/code
    • Develop a script to "slice" videos to "match" DICOM series timing #14

Sample projects to use and follow in organization of the "library" and overall project

• https://hatch.pypa.io/latest/ as overall framework.
  • mih of datalad I believe started to use for all things datalad. e.g. https://hub.datalad.org/datalad/datasalad
  • E.g. @candleindark used for https://github.com/dandi/dandisets-linkml-status-tools so you could ask him
• for CLI can use click .
  • may be place all "click" interfaces under reprostim/cli/ subpackage since they cannot be used directly from python. See e.g. how we did it in https://github.com/dandi/dandi-cli/tree/master/dandi/cli -- those are typically lean wrappers for interfaces within library itself
• all the tools/` should just become entrypoints within the library.
• make sure to centralize logging . In each file do lgr = logging.getLogger(__name__), and then for CLI in global context handler for click provide tune up of formatting etc. @jwodder could you please point to "optimal" setup for setup of logging so that python library is not effected.

Stages

• prepare and implement design doc (alike to the ones we are preparing elsewhere) listing on what Python packages and subpackages we would like to see, and which CLI interfaces (based on scripts we already have).
  • WiP: Refactor codebase into installable "reprostim" Python module (+ reprostim-videocapture) package #124
• upload to pypi.org and verify that it installs/functions correctly, e.g. via
  d=venv-reprostim; python3 -m venv $d && source $d/bin/activate && pip install reprostim && reprostim --help
• Add versioning : we use versioneer in many projects e.g. dandi-cli, heudiconv, but some times use @jwodder's versioningit whenever we require more tricky cases, e.g. in http://github.com/dandi/dandischema where we tag for two things (library and schema itself)
• Automate releasing e.g. via intuit's auto (we do that in dandi, heudiconv, etc)

[jwodder]

@yarikoptic

could you please point to "optimal" setup for setup of logging so that python library is not effected.

The main thing is to only call logging.basicConfig() or similar inside the CLI's main() function; it should never be called by library code.