Quickstart

This takes one config through the three things the package does most: check it, explain what is wrong, and repair it.

A config with a typo

Save this as my_config.yaml. It describes a small CCD exposure, with one mistake: the dark current model is spelled dark_curent.

exposure:
  readout:
    times: [1.0]

ccd_detector:
  geometry:
    row: 64
    col: 64
    total_thickness: 40.0
    pixel_vert_size: 10.0
    pixel_horz_size: 10.0
  environment:
    temperature: 150.0
  characteristics:
    quantum_efficiency: 0.9
    charge_to_volt:
      value: 3.0e-6
    pre_amplification: 100.0
    adc_bit_resolution: 16
    adc_voltage_range: [0.0, 10.0]

pipeline:
  photon_collection:
    - func: pyxel.models.photon_collection.illumination
      name: illumination
      enabled: true
      arguments:
        level: 1000.0
  charge_generation:
    - func: pyxel.models.charge_generation.simple_conversion
      name: photoelectric_effect
      enabled: true
    - func: pyxel.models.charge_generation.dark_curent
      name: dark_current
      enabled: true
      arguments:
        figure_of_merit: 1.0
  charge_collection:
    - func: pyxel.models.charge_collection.simple_collection
      name: simple_collection
      enabled: true
  charge_measurement:
    - func: pyxel.models.charge_measurement.simple_measurement
      name: simple_measurement
      enabled: true
  readout_electronics:
    - func: pyxel.models.readout_electronics.simple_adc
      name: simple_adc
      enabled: true

Check it

from pyxel_config_core import validate_config

result = validate_config(open("my_config.yaml").read())
print(result.valid)
for d in result.diagnostics:
    print(d.severity, d.path)
    print(d.message)

False
error /pipeline/charge_generation/1/func
Unknown model 'dark_curent' in stage 'charge_generation'. Did you mean 'dark_current'?

Each finding is a Diagnostic: how much it matters, where it is (a JSON pointer into the config), what is wrong, and for a mistyped model, the suggestion it most likely meant. result.pyxel_version names the Pyxel release whose schema judged the config.

Ask for an explanation

explain_config() runs the same checks and adds a summary and one instruction per finding:

from pyxel_config_core import explain_config

explanation = explain_config(open("my_config.yaml").read())
print(explanation.summary)
print(explanation.diagnostics[0].next_step)

This CCD exposure config has 1 problem in the charge_generation stage.
Rename func to 'pyxel.models.charge_generation.dark_current'.

Repair it

autofix() renames mistyped models and moves misplaced ones. It returns a corrected copy and leaves the original alone:

from pyxel_config_core import autofix, validate_config

fix = autofix(open("my_config.yaml").read())
for change in fix.changes:
    print(change.action, change.before, "->", change.after)
print(validate_config(fix.config).valid)

rename_func pyxel.models.charge_generation.dark_curent -> pyxel.models.charge_generation.dark_current
True

The same from the command line

pyxel-config validate my_config.yaml           # exits with 1: the config has errors
pyxel-config explain my_config.yaml            # adds the summary and next steps
pyxel-config fix my_config.yaml -o fixed.yaml  # writes the corrected copy
pyxel-config validate fixed.yaml               # exits with 0: valid

The exit code is the verdict, so these lines work in CI as they are. See the command line for report formats and a ready-made GitLab CI job.

Going further

• validate_config(config, deep=True), or pyxel-config validate --deep, also hands the config to Pyxel’s own loader. It needs Pyxel installed; see the deep check.

• How validation works lists every kind of check, and what is deliberately left to Pyxel.