# Load and inspect a structure

With *httk-atomistic* installed, `httk.core.load()` provides a one-call
atomistic loading experience through a format adapter registered during
registry discovery:

```python
from httk.core import load

structure = load("example.cif")
# POSCAR, CONTCAR, and compressed variants such as "CONTCAR.bz2" work too.
# CIF loads return ASUStructure; POSCAR/CONTCAR loads return UnitcellStructure.

print("Formula:", structure.formula)
print("Volume:", float(structure.cell.volume))
print("Species at sites:", structure.species_at_sites)
print("Reduced coordinates:", structure.sites.reduced_coords)
```

Views provide the lazy form: construction checks the source configuration, and parsing
waits until the first property is accessed.

```python
from httk.atomistic import ASUStructureView, UnitcellStructureView

structure = UnitcellStructureView("example.cif")
asu = ASUStructureView("example.cif")
```

For a remote source, pass an explicit `httk.core.DatastreamURL` consent token.

The split is architectural: *httk-atomistic*'s I/O layer (`httk.atomistic.io`)
parses file formats and returns neutral payloads, *httk-atomistic* owns
`UnitcellStructure` and `ASUStructure`, and *httk-core* dispatches between
them. The adapter registration makes the one-call domain-loading experience
work when *httk-atomistic* is installed. Code that needs the neutral reader
result can use the explicit escape hatch `load("example.cif", raw=True)`.
Composition and formula properties are part of the structure itself. Providers
reuse the same domain projection when exposing a structure through OPTIMADE, so
serving or storing it does not require a separate formula calculation. Geometry
remains exact until the explicit `float(...)` used for display.

See the full *httk-atomistic* loading example in the versioned module
documentation listed by the {doc}`module directory <../modules>`.

See also the {doc}`/structures` topic page for the current structure vocabulary.