Struct Corpus 

pub struct Corpus { /* private fields */ }

Handle for accessing cached test-image datasets.

Create with Corpus::new() (default cache location) or Corpus::with_cache_root() (explicit path). Then call Corpus::get() to obtain the local path to a dataset — downloading it if necessary.

Implementations

impl Corpus

pub fn new() -> Result<Self, Error>

Initialize with default cache location.

Resolves cache root via CODEC_CORPUS_CACHE env var, then dirs::cache_dir(). Performs no I/O beyond directory creation.

pub fn with_cache_root(path: impl Into<PathBuf>) -> Result<Self, Error>

Initialize with explicit cache root. Overrides the environment variable.

Files will live at {path}/codec-corpus/v{major}/.

pub fn get(&self, path: &str) -> Result<PathBuf, Error>

Get the local path to a corpus subdirectory, downloading if needed.

path is any path into the repository (e.g. "webp-conformance", "webp-conformance/valid", "clic2025/training"). The top-level folder is the download unit — requesting any path under it fetches the entire folder recursively.

Examples

let corpus = codec_corpus::Corpus::new()?;
let webp = corpus.get("webp-conformance")?;
let valid = corpus.get("webp-conformance/valid")?;
let training = corpus.get("clic2025/training")?;

pub fn is_cached(&self, path: &str) -> bool

Check if a path is already cached locally, without downloading.

pub fn list_cached(&self) -> Vec<String>

List datasets currently cached on disk.

Returns directory names under the cache root, excluding internal files (.version, .lock, .tmp-*).