R bindings for libcozip. Read a Cloud-Optimized ZIP archive's manifest over HTTP range requests, or write one from an arrow Table.
libzip, zlib and karu are vendored under src/. Building the reader needs
a C++23 compiler, pkg-config, libcurl 7.83 or newer and OpenSSL 3. On Debian
and Ubuntu, install libcurl4-openssl-dev and libssl-dev.
install.packages("cozip", repos = "https://asterisk-labs.r-universe.dev")
library(cozip)
library(arrow)
tbl <- arrow_table(
name = c("a.txt", "b.bin"),
path = c("/path/to/a.txt", "/path/to/b.bin")
)
create("out.zip", tbl)
name is how each file appears inside the archive. path is where it lives on disk, used at write time and dropped from the manifest. Any extra columns ride along into __metadata__ and become queryable on read.
Archive names must be ASCII. Writer source and output paths may contain
Unicode; paths and URLs passed to cozip::read() are currently ASCII-only.
Empty files are not valid cozip entries.
tbl <- arrow_table(
name = c("a.tif", "b.tif"),
path = c("/path/to/a.tif", "/path/to/b.tif"),
cloud_pct = c(12.3, 45.1)
)
create("out.zip", tbl)
library(cozip)
manifest <- read("https://example.com/dataset.zip")
train <- manifest[manifest$split == "train", ]
manifest is a tibble: name, offset, size, cozip:location, and the writer's extras. Local path or URL, same call. It fetches the byte-0 index and __metadata__, never the payloads.
columns = c(...) picks extras. location = FALSE drops cozip:location.
cozip::read() supports only Flat-profile archives (profile = 1). TACO
archives (profile = 2) and all other profiles are rejected.
read() fetches the manifest through libcozip; DuckDB decodes Parquet without
loading extensions.
Remote reads use environment or standard credential files. For SQL over stored
files, use the optional cozip_reader
extension.
The R package version tracks the C library exactly and is copied from the
repo-root VERSION file by make sync.
packageVersion("cozip") # "2026.9.4"
See SPEC.md for the on-disk format.
MIT. See LICENSE.