Locking a corpus prevents mutation by sm_refresh() and other modifying
operations. This is useful for archival or when you want to guarantee
reproducibility after generating a sm_certificate().
sm_lock() sets metadata$is_locked = TRUE with an optional reason.
sm_unlock() reverses the lock but requires explicit confirmation to
prevent accidental unlocking of archived corpora.
Usage
sm_lock(corpus, reason = NULL, call = rlang::caller_env())
sm_unlock(corpus, confirm = FALSE, call = rlang::caller_env())Arguments
- corpus
An
sm_corpusobject.- reason
Optional character string documenting why the corpus was locked (e.g.,
"submitted for review","certificate generated").- call
Caller environment for error reporting.
- confirm
Logical. Must be
TRUEto actually unlock. This safeguard prevents accidental unlocking of archived corpora.
Details
Merging, deduplication, materialisation, citation maturity, enrichment, embedding changes, clustering, labelling, affiliation matching and screening refuse locked inputs, as do refreshing and input registration. Subsets remain locked but lose their stored lock hash. Merged corpora are unlocked; unlock the inputs explicitly before merging.
The lock records a content hash. sm_certificate() warns and
sm_validate() reports when direct edits changed that content.
Printing shows only the lock status and does not hash the corpus.
Earlier locked corpora without a stored hash are not checked.
See also
Other refresh:
sm_refresh(),
sm_staleness()
Examples
corpus <- sm_example_corpus()
locked <- sm_lock(corpus, reason = "archival snapshot")
#> ✔ Corpus locked.
#> ℹ Reason: archival snapshot
locked$metadata$is_locked
#> [1] TRUE
unlocked <- sm_unlock(locked, confirm = TRUE)
#> ✔ Corpus unlocked.
unlocked$metadata$is_locked
#> [1] FALSE