Fetches, by version id, the objects the index lists for each key's run (the
run its tracked record names, so a checkout restores what its tables were
made from) and their evidence build, decompresses the grade table, and
checks every file's md5 against the index's source_md5 (and a grade
table against its record). Each file is written aside and moved in once
checked; a file already there with the same md5 is left, and one with
another md5 refuses. The evidence and the bundles' ignored files come by
default; grade tables, gigabytes each, only with grades = TRUE. Grade
tables are restored before segments, whose mtime scripts/0330 holds to
be later.
Usage
knw_archive_pull(
keys,
dest = ".",
grades = FALSE,
index = "data/archive_index.csv",
bucket = .knw_archive_bucket,
batch = NULL
)Arguments
- keys
the keys whose run to restore.
- dest
the tree to restore into (its
data/).- grades
TRUEto restore the grade tables too.- index
the index path.
- bucket
the bucket.
- batch
NULL, or a named character vector of batch ids by key, for a record that several archives hold with other curves or segments (the checkout's own files, and then the published batch, decide otherwise).