Fetch a capsule's source again and pin the new bytes

Downloads the source named in a capsule's data_provenance.json (resource$direct_url, else source$url; an https:// URL, or a local file inside the capsule), and when its bytes differ from the pin writes them as the capsule's data file and pins them in the provenance, in the form the analysis.R of usecpstmpl writes on its first run: source$sha256, and resource$filename, sha256, size_bytes and retrieved_utc (row_count_data_rows too when it was recorded). vrfcps then verifies the capsule against the new data. From the shell: rmbl refresh PATH.

Usage

cpsrfrsh(dir, allowfile = FALSE)

Arguments

dir: Character scalar, the capsule directory.
allowfile: FALSE (the default), TRUE to accept a file:// source inside dir, or an existing folder whose files may also be read as a source (a local path or a file:// URL).

Value

The mnfdiff of the provenance before and after: all "unchanged" when the source still matches its pin.

Details

Nothing is written when the bytes are unchanged. A download goes through fdl, with the provenance's wayback$csv_snapshot as its fallback (no other Wayback copy is looked up). A source is fetched over https:// only. A local path (relative to the capsule) or a file:// URL is read only when it resolves, symbolic links followed, to a file inside the capsule, inside the allowfile folder, or in the package's extdata; never under /proc, /sys or /dev. A gzip source may inflate to at most the option rmoriebricklayer.maxcsvbytes (1 GiB by default) when it is read. A capsule bundle is not signed again: that needs your key (cpsbndl). Side effects: writes the data file and data_provenance.json in dir. Errors: dir is not a directory or holds no provenance; the provenance names no source, or a TODO one, or a file:// one without allowfile; a URL that is not https://; a local source outside those folders; the source cannot be fetched, or inflates past the limit; resource$filename is not a plain relative path inside dir; data_provenance.json is a link out of dir; the new data fail the schema the provenance pins (valsch).

Examples

d <- usecpstmpl(tempfile("capsule-"), ex = TRUE)
cpsrfrsh(d)            # the first pin: the data file is added
#> ╭── Manifest diff ────────────────────────────────────────────────────────────╮
#> │   Resources  0 changed, 1 added, 0 removed, 0 unchanged                     │
#> │                                                                             │
#> │   Files below differ between the two manifests (old -> new).                │
#> │                                                                             │
#> │   added     otis_a01_individuals.csv sha256 - -> 6b1a0d2d11f6, - -> 462     │
#> │             bytes                                                           │
#> ╰─────────────────────────────────────────────────────────────────────────────╯
vrfcps(d)$ok
#> [1] TRUE
cpsrfrsh(d)$change     # the source has not changed since
#> [1] "unchanged"

try(cpsrfrsh(tempfile()))
#> Error : cpsrfrsh: `dir` is not an existing directory: /tmp/RtmpZpRdcE/file1e6a33a1c1c9