Suggest Column Tags From Header Names — suggest_column

Heuristically guesses a tag type for each column from its (already janitor::clean_names()-normalized) header name. This powers the "suggest, don't auto-apply" pre-fill in the Tag Columns step: the returned suggestions seed the dropdowns, but the user always reviews and applies.

Usage

suggest_column_tags(col_names)

Arguments

col_names: Character vector of column names (post-clean_names()).

Value

A named list, one element per input column (1:1, order preserved), whose value is a tag type (e.g. "CASRN", "Name", "Result") or "" when no confident match is found. Returns an empty named list for zero-length input. Returned as a list (not an atomic vector) so callers can safely use suggestions[[col]] %||% "" without a subscript-out-of-bounds error on missing keys.

Details

The matcher is name-only (no cell-value sampling) and precision-first: it matches on whole tokens, prefers the most specific (longest) keyword phrase, drops dangerous generic tokens (bare name, date, value, id, exposure), and emits at most one suggestion for the singular chemical tags (Name, CASRN).

Every emitted value is a member of the classify_tags() taxonomy (R/tag_helpers.R); unmatched columns return "".

Examples

suggest_column_tags(c("cas_number", "chemical_name", "supplier_name"))
#> $cas_number
#> [1] "CASRN"
#> 
#> $chemical_name
#> [1] "Name"
#> 
#> $supplier_name
#> [1] ""
#> 
# $cas_number    -> "CASRN"
# $chemical_name -> "Name"
# $supplier_name -> ""   (generic 'name' without a chemical qualifier)