API stability
This page records the frozen public contract for padelpy2 on the v0.x line through v1.0 unless a major version bump says otherwise. Additive APIs are welcome when they preserve the defaults below.
See also API Documentation (autodoc) and Architecture.
Frozen top-level exports
Importable from padelpy2 (package __all__):
CalculatorPaDELConfig__version__descriptors,descriptors_2d,descriptors_3dfingerprintspadeldescriptor
Catalog types Descriptor and Fingerprint, and individual catalog
singletons (for example Weight, MACCSFingerprinter), are imported from
submodules. padelpy2.utils is semi-internal (empty __all__).
Frozen Calculator behavior
Calculator(descriptors, config=None)— high-level calculator over the stock Yap JAR.Calculator.__call__(mols)returns apandas.DataFrameand drops the engineNamecolumn by default.Additive optional keyword-only arguments (defaults preserve v0.1 behavior):
chunk_size=None— optional batchingretain_names=False— drop engineNameunlessTrueon_error="raise"— fail fast;"nan"isolates invalid molecules / failed chunks as NaN rows when a column schema is available
Default
PaDELConfigfield values andpadeldescriptorkeyword names used by existing callers are stable.
Default catalog column counts
These shapes are guarded in CI (metadata counts and live Calculator runs):
Catalog |
Columns |
|---|---|
|
1444 |
|
431 |
|
1875 |
Each fingerprint |
|
Stock-JAR golden oracles under tests/fixtures/oracles_v1/ pin representative
values for these catalogs.
Excluded from default lists
The following descriptors exist as importable singletons but are not in the
default descriptors / descriptors_2d lists:
AminoAcidCountIPMolecularLearningKierHallSmarts
Stability checklist (do not break without a major version)
Signatures of
Calculator.__init__andCalculator.__call__(required arguments)Default contents of
descriptors/descriptors_2d/descriptors_3d/fingerprintsDefault dropping of the
Namecolumnpadeldescriptorkeyword names used by existing callersDefault
PaDELConfigfield values