Applies [csfmt_parse] to every value column (everything not in the structural schema) and returns a catalog: one row per column with its parsed components. This makes a dataset self-describing – generic tooling (QC, collapse, presentation) routes on the catalog instead of hardcoding column names.
Value
A data.table: `column, measure, denom, role, q, level, per, suffix, interpretable` (the last TRUE when a role/quantile/level coordinate was found).
See also
Neither package vignette covers this function. It is the dataset-wide
form of csfmt_parse, and is what
compare_results uses to find the value columns it should diff.
Other naming grammar functions:
csfmt_parse(),
csfmt_var(),
q_label(),
q_value()
Examples
d <- data.table::data.table(
isoyearweek = "2023-01",
numerator_nowcasted_q50x0 = 42,
numerator_nowcasted_vs_denominator_nowcasted_pr100_q50x0 = 8.4,
numerator_nowcasted_status_prob_high = 0.3,
a_column_outside_the_grammar = 1
)
# isoyearweek is structural, so it is not a value column at all; the last
# column is a value column the grammar cannot read (interpretable = FALSE)
csfmt_interpret(d)
#> column
#> <char>
#> 1: numerator_nowcasted_q50x0
#> 2: numerator_nowcasted_vs_denominator_nowcasted_pr100_q50x0
#> 3: numerator_nowcasted_status_prob_high
#> 4: a_column_outside_the_grammar
#> measure denom role q level per suffix
#> <char> <char> <char> <num> <char> <int> <char>
#> 1: numerator <NA> nowcasted 0.5 <NA> NA <NA>
#> 2: numerator_nowcasted denominator nowcasted 0.5 <NA> 100 <NA>
#> 3: numerator_nowcasted <NA> status NA high NA <NA>
#> 4: a_column_outside_the_grammar <NA> <NA> NA <NA> NA <NA>
#> interpretable
#> <lgcl>
#> 1: TRUE
#> 2: TRUE
#> 3: TRUE
#> 4: FALSE