Conform a data frame to a registered canonical schema

Validates that every required canonical column is present, then coerces every declared column to its canonical type. Returns the conformed tibble.

Usage

crt_schema_conform(df, source, file_name)

Arguments

df

A data frame.

source

Character. Source family code (e.g. "nge").

file_name

Character. Logical file name (e.g. "track_sessions").

Value

A tibble conformed to the canonical schema for that entry.

Details

This is the entry point for data crate did not read. crt_ingest() reads a registered CSV and dispatches to a handler before conforming; a caller that already has a data frame — from a database, a spatial file, an API — comes here directly. Both paths run the same validation and typing, because crt_ingest() calls this function.

Use crt_files() to discover what (source, file_name) pairs are registered.

Presence is checked before types are applied, because a missing required column would otherwise become a column of NA at coercion and the failure would surface far from its cause.

What this does not do is reject columns the schema does not declare — extra columns pass through untouched. A caller that wants the canonical set to be exhaustive should assert that itself; the schema declares a floor, not a ceiling.

Throws on an unknown (source, file_name) pair, on a missing required column, and on a schema declaring a type crate does not support.

See also

crt_ingest() to read and conform a registered file in one call, crt_files() to list registered entries.

Examples

# A frame that is already in canonical shape passes through, typed.
path <- system.file(
  "extdata/examples/bcfp/user_habitat_classification_wide.csv",
  package = "crate"
)
raw <- readr::read_csv(path, show_col_types = FALSE)
conformed <- crt_schema_conform(raw, "bcfp", "user_habitat_classification")
conformed
#> # A tibble: 6 × 11
#>   blue_line_key downstream_route_measure upstream_route_measure
#>           <int>                    <dbl>                  <dbl>
#> 1     356385867                        0                    208
#> 2     356385867                        0                    208
#> 3     356400111                      500                   1500
#> 4     356400111                      500                   1500
#> 5     356500222                        0                    800
#> 6     356600333                        0                    300
#> # ℹ 8 more variables: watershed_group_code <chr>, species_code <chr>,
#> #   spawning <int>, rearing <int>, reviewer_name <chr>, review_date <chr>,
#> #   source <chr>, notes <chr>

# The schema declares blue_line_key as integer; readr guessed double.
class(raw$blue_line_key)
#> [1] "numeric"
class(conformed$blue_line_key)
#> [1] "integer"