plaid.viewer.models¶

Reference to a PLAID sample.

Attributes:

• dataset_id (str) –

  Identifier of the dataset (typically the dataset directory name for local datasets, or the org/repo id for Hub datasets).

• split (str | None) –

  Optional split name ("train", "test", ...). None when the dataset is not split.

• sample_id (str) –

  Identifier of the sample within the split. For disk-backed datasets this is the zero-based index rendered as a string.

encode()

32
33
34
35

def encode(self) -> str:
    """Return a URL-safe string identifier usable as a route parameter."""
    split = self.split if self.split is not None else "_"
    return f"{self.dataset_id}:{split}:{self.sample_id}"

decode(value)

37
38
39
40
41
42
43
44
45
46
47
48

@classmethod
def decode(cls, value: str) -> "SampleRef":
    """Parse a string produced by :meth:`encode`."""
    parts = value.split(":")
    if len(parts) != 3:
        raise ValueError(f"Invalid sample reference: {value!r}")
    dataset_id, split, sample_id = parts
    return cls(
        dataset_id=dataset_id,
        split=None if split == "_" else split,
        sample_id=sample_id,
    )

A ParaView-readable artifact produced from a PLAID sample.

For time-dependent samples, cgns_path points to a .cgns.series sidecar file that groups multiple CGNS files into a single time sequence. For single-timestep samples, it points to the single CGNS file directly.

Attributes:

• artifact_id (str) –

  Stable identifier used in API routes. Derived from the cache key.

• cgns_path (Path) –

  Path to the file ParaView should open. Either a .cgns.series sidecar (multi-time) or a .cgns file.

• state_path (Path | None) –

  Optional ParaView state file (.pvsm) providing a reasonable default scene.

• metadata_path (Path | None) –

  Optional JSON metadata file describing the artifact.

• cache_key (str) –

  Deterministic SHA256 key over the artifact inputs.

• created (bool) –

  True if the artifact was newly created, False if it was already present in the cache.

Summary information about an available dataset.

is_streaming distinguishes datasets opened with :func:plaid.storage.init_from_disk (False) from Hugging Face repositories streamed through :func:plaid.storage.init_streaming_from_hub (True). Streamed datasets do not always expose a total sample count and may need to be navigated sequentially through a streaming cursor.

Full detail view of a dataset.

splits maps each split name to its sample count. The count is None for streaming datasets where the total is unknown.

Serializable form of :class:SampleRef used by the API.

from_ref(ref)

123
124
125
126
127
128
129
130
131

@classmethod
def from_ref(cls, ref: SampleRef) -> "SampleRefDTO":
    """Build the DTO from a :class:`SampleRef`."""
    return cls(
        dataset_id=ref.dataset_id,
        split=ref.split,
        sample_id=ref.sample_id,
        encoded=ref.encode(),
    )

Minimal metadata describing a PLAID sample.

Validation outcome for a PLAID sample.

Public view of a :class:ParaviewArtifact.

Response model for the viewer-url endpoint.