Validate human-readable gifter database sources — validate_gifter

Performs structural validation before a SQLite database is compiled. Any duplicate identifiers, broken references, missing hierarchy levels, invalid directions, malformed anchor boundaries, graph cycles, or inconsistent release metadata are reported as build errors.

Usage

validate_gifter_sources(
  source_dir,
  stop_on_error = TRUE,
  reference_dir = NULL,
  marker_links = NULL
)

Arguments

source_dir: Directory containing the gifter TSV source tables.
stop_on_error: If `TRUE`, stop when structural errors are found.
reference_dir: Optional directory holding the pinned reference snapshot written by `data-raw/reference_snapshot.R`. `NULL` runs the structural checks only.
marker_links: Optional paths to marker link extracts, used to recompute the basis of marker assignments. A namespace with no extract is reported as not recomputed.

Value

A list with `valid`, `errors`, `warnings`, and table row counts.

Details

Structural validation cannot tell a real accession from a plausible one. When `reference_dir` names a pinned reference snapshot, the tables are also compared with Rhea, ChEBI and KEGG: every external accession must exist, a Rhea identifier must be a master, imported equations and names must match, and each declared anchor must take part in the chemistry of its routes.