Improve the session caches of as.ab(), as.mo() and all mo_*() functions (#313) (#314)

- One cache design for as.ab() and as.mo() in R/aa_session_cache.R: lazily built
  lookup indices (ATC codes, drug synonyms, SNOMED codes) and coercion caches with
  the same structure (key, input, value), keyed on input plus outcome-relevant settings
- Caches are invalidated whenever the lookup tables change (custom antimicrobials or
  microorganisms, set_mo_source())
- mo_*() functions use a fast path for valid <mo> input and work on unique values only
- synonym_mo_to_accepted_mo() is memoised per code
- Fix as.mo() cache keys that never matched again, fast_mode results of as.ab() being
  reused by regular calls, and ab_reset_session() clearing as.mo() uncertainties

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Matthijs BerendsandClaude Opus 5.5 authored and GitHub committed 2026-10-07 14:29:48 +02:00
1 parent 0ca43df459
commit 891d248fab
14 files changed
+462 -124

No files matched your search

+2
View File
@@ -113,6 +113,8 @@ There are three helper functions that can be run after using the \code{\link[=as
\item Use \code{\link[=mo_failures]{mo_failures()}} to get a \link{character} \link{vector} with all values that could not be coerced to a valid value.
\item Use \code{\link[=mo_renamed]{mo_renamed()}} to get a \link{data.frame} with all values that could be coerced based on outdated, previously accepted taxonomic names.
}
Results of non-exact input are remembered during the session, so that the same input is not matched again. This only applies to the same settings of \code{minimum_matching_score}, \code{only_fungi} and \code{cleaning_regex}. Use \code{\link[=mo_reset_session]{mo_reset_session()}} to forget these results. They are also forgotten when the taxonomy changes, e.g. after \code{\link[=add_custom_microorganisms]{add_custom_microorganisms()}}.
}
\subsection{For Mycologists}{