Extending SPECTROview¶

SPECTROview does not have a plugin/entry-point framework — there is no registry to hook custom code into the application. Instead, extensibility works through two file-based mechanisms that the API exposes directly. If you're looking for "how do I add a custom X to SPECTROview," one of these is almost always the answer.

Custom Baseline Methods¶

Baseline evaluation (spectroview.api.preprocessing.subtract_baseline, and every workspace's set_baseline) dispatches on config["mode"] by string name. Beyond the built-in None, "Linear", "Polynomial", and "arpls" modes, any pybaselines Baseline method name works out of the box — no registration step:

from spectroview.api import preprocessing

# "asls", "airpls", "modpoly", "snip", ... any pybaselines.Baseline method name
Y_corrected, Y_baseline = preprocessing.subtract_baseline(x, Y, {"mode": "asls", "lam": 1e6})

If pybaselines adds a new method in a future release, it becomes available to SPECTROview immediately with no code change on either side.

Fit-Model & Plot Templates¶

Both fit models and plot configurations are plain JSON files in a folder you control — SPECTROview (GUI or API) just scans that folder. This means any external tool, script, or version-controlled repository of templates can populate them:

from spectroview.api import fitting, graphs, settings

# Fit-model templates live under the SPECTROview Working Folder. Point it at a
# shared location — set_working_folder() creates fit_model/plot_recipe/plot_style
# subfolders under it:
settings.set_working_folder("./shared_spectroview")
# ...then any *.json dropped into the fit_model folder (by this API, the GUI, or
# hand-written) is immediately visible:
names = fitting.list_fit_model_templates(settings.get_fit_model_folder())

# Plot recipes work the same way, via a folder-backed store:
graphs.save_plot_recipe("./shared_spectroview/plot_recipe", "QC Dashboard", configs=[...])

A fit-model template's JSON shape is {"0": {"peak_labels": [...], "peak_models": {...}, "baseline": {...}, "range_min": ..., "range_max": ..., "fit_params": {...}}} — write one directly if you're generating templates from an external system, or build one with fitting.build_fit_model() / fitting.save_fit_model_template() and inspect the output.

Custom Peak Shapes¶

The batched peak-shape registry (spectroview.fit_engine.models.BATCHED_MODELS) is a plain Python dict mapping shape name to (eval_fn, jacobian_fn, param_names). It is not part of the public API surface (no stability guarantee across releases), but if you need a shape SPECTROview doesn't ship, that registry — plus spectroview.fit_engine.scalar_models.PEAK_MODEL_REGISTRY for the non-batched fallback path — is where it would need to be added. This requires a source change, not a runtime extension; there is currently no way to register a new peak shape from a script without modifying the installed package.