Skip to contents

Build a reproducible analysis dataset

entropia_analysis_dataset.Rd

[Experimental]

Usage

entropia_analysis_dataset(con, ..., name = NULL)

Arguments

con

A connection returned by entropia_connect().

...

Filter expressions applied to the corpus, e.g. asset_type == "image". Column names resolve against the lazy corpus (see entropia_corpus() for the full column set). Must be unnamed.

name

Optional human-readable label stored in the provenance.

Value

A tibble::tibble() of class entropia_dataset, one row per asset, with the entropia_prov attribute.

Details

The dataset boundary of the package: assembles the lazy corpus (entropia_corpus()), applies any filter expressions passed in ..., and materialises the result with a deterministic row order (arranged by asset_id). The returned tibble carries class entropia_dataset and an entropia_prov attribute recording everything needed to reconstruct the dataset:

  • name: the human label passed to name (NULL for unnamed);

  • schema_version: the database schema head (e.g. "0029_rag_chunks");

  • content_hash: the connection's schema content hash (see entropia_connect());

  • source_path: the database file the dataset was built from;

  • filters: the deparsed filter expressions captured from ...;

  • package_version: the entropiaR version used;

  • built_at: the build timestamp (ISO-8601, UTC);

  • r_version: the R version used.

Building the same dataset twice against an unchanged database yields byte-identical rows (deterministic ordering) and identical provenance apart from built_at. Read the stamp with entropia_provenance() and persist it as a JSON sidecar with entropia_write_provenance().

Examples

con <- entropia_connect(system.file("extdata", "entropia-example.sqlite",
  package = "entropiaR"
))
ds <- entropia_analysis_dataset(con, asset_type == "pdf", name = "PDF corpus")
entropia_provenance(ds)
#> entropiaR dataset provenance
#>   name:           PDF corpus
#>   schema version: 0029_rag_chunks
#>   content hash:   09d4c603b66d68c4c0cef0f51ff09a04fb30a49fe200907ef69693d11dd25732
#>   source path:    /home/runner/work/_temp/Library/entropiaR/extdata/entropia-example.sqlite
#>   filters:        asset_type == "pdf"
#>   package:        0.0.0.9000
#>   built at:       2026-08-24T01:10:02.465Z
#>   R version:      R version 4.6.1 (2026-06-24)
entropia_disconnect(con)