Module

pixano.datasets.io.api

The shared Python entry points behind the CLI and the REST API (spec §3.1). ``analyze`` and ``import_dataset`` are the whole public surface: the CLI renders the returned :class:`ImportPlan` and calls ``import_dataset`` with a tqdm sink; the REST layer does the same with a job-store sink.

Functions

def

 analyze 
(source, spec, limits, importer)
  → ImportPlan

Analyze a source without side effects and return its import plan.

Parameter	Type	Default
`source`	`str \| Path`	—
`spec`	`ImportSpec \| None`	`None`
`limits`	`AnalyzeLimits \| None`	`None`
`importer`	`DatasetImporter \| None`	`None`

def

 export_dataset 
(dataset, destination, format, media, options)
  → Path

Export a dataset to a data format (spec §10). ``pixano_jsonl`` emits exactly the import grammar with explicit ids, so import → export → import is id-equal. Other formats arrive with their exporters (plan P3).

Parameter	Type	Default
`dataset`	`'Dataset \| Path \| str'`	—
`destination`	`str \| Path`	—
`format`	`str`	`'pixano_jsonl'`
`media`	`str`	`'files'`
`options`	`dict[str, Any] \| None`	`None`

def

 import_dataset 
(source, data_dir, spec, plan, info, importer, sinks, engine, job_id, resume_cursor)
  → ImportResult

Import a source into the data directory's library (analyze → validate → ingest). Args: source: Source directory/file, or ``hub://org/repo``. data_dir: The Pixano data directory (holds ``library/``). spec: Declarative import spec; defaults apply when omitted. plan: A previously computed plan; re-analyzed when omitted. info: Target dataset schema; derived from the workspace preset when omitted. importer: Explicit importer instance (advanced; bypasses the registry). sinks: Progress sinks (e.g. a tqdm sink). engine: Preconfigured engine (jobs wire checkpoints/cancellation through it). job_id: External job id (names the staging dir so crashed builds resume). resume_cursor: Last committed cursor; re-enters the importer past done work. Returns: The import result (dataset id/path, per-table row counts).

Parameter	Type	Default
`source`	`str \| Path`	—
`data_dir`	`str \| Path`	—
`spec`	`ImportSpec \| None`	`None`
`plan`	`ImportPlan \| None`	`None`
`info`	`DatasetInfo \| None`	`None`
`importer`	`DatasetImporter \| None`	`None`
`sinks`	`Sequence[ProgressSink]`	`()`
`engine`	`ImportEngine \| None`	`None`
`job_id`	`str \| None`	`None`
`resume_cursor`	`dict \| None`	`None`