create a tidy dictionary from a data.frame (sf supported). infer a simple type per column, attach english labels (optionally overridden by a csv map), and report stats: missing %, unique count, example values, and min/max for numeric/date/datetime. optionally add a translated label column.
Usage
build_dictionary(
data,
labels_path = base::getOption("snt.labels_en_path", NULL),
language = NULL,
max_levels = 50L,
n_examples = 3L,
trans_cache_path = NULL,
override_yaml = FALSE,
verbose = TRUE
)Arguments
- data
data.frame or tibble; sf columns allowed.
- labels_path
optional csv with columns like
name,labelto override english labels.- language
optional iso code (e.g., "fr") to add
label_<language>.- max_levels
max factor levels to summarize in notes. default 50.
- n_examples
number of example values to show. default 3.
- trans_cache_path
optional cache dir for translate_text_vec().
- override_yaml
logical; if TRUE, CSV labels override YAML labels. default FALSE (YAML takes precedence).
- verbose
logical; if TRUE (default), prints info messages such as fuzzy match notifications.
Value
tibble with:
variable, type, label_en, n, n_missing, pct_missing, n_unique,
example_values, min, max, notes, and optionally label_<language>.
Details
english labels are merged as: internal defaults, then csv overrides. unknown variables fall back to their column name.
performance: the snt variable tree is cached in a package environment on first use. subsequent calls reuse the flattened tree. the cache automatically refreshes when the tree version changes (tracked via _meta$last_updated).
Examples
dd <- build_dictionary(dplyr::as_tibble(iris))
dd |> dplyr::select(variable, type, label_en) |> utils::head()
#> # A tibble: 5 × 3
#> variable type label_en
#> <chr> <chr> <chr>
#> 1 Sepal.Length double Sepal.Length
#> 2 Sepal.Width double Sepal.Width
#> 3 Petal.Length double Petal.Length
#> 4 Petal.Width double Petal.Width
#> 5 Species factor Species