pirl-unc / tsarina / 27153087900

Build # 27153087900

Build Type

push

github

Committed by web-flow

Commit Message

reference_data: delegate cache dir + download to datacache (like pyensembl) (#106)

Per the "why not datacache?" question. hitlist and tsarina each hand-rolled
cache-dir resolution + download, while pyensembl (same openvax ecosystem)
already uses datacache for exactly this. reference_data now does too:

- cache_dir() -> datacache.get_data_dir(subdir="tsarina",
  envkey="TSARINA_DATA_DIR") + ensure_dir. Platform-appropriate via appdirs
  (e.g. ~/Library/Caches/tsarina) when the env var is unset, instead of the
  bespoke repo-.cache / ~/.cache fallback.
- download() -> datacache.download._download_and_decompress_if_necessary
  (the same private helper pyensembl's download_cache calls), replacing the
  hand-rolled urllib fetch + zipfile/_extract_to. Handles .zip and .gz.

The value-add layer datacache does NOT provide stays here: the dataset
registry, HPA version pinning, and the sha256/version/timestamp manifest.

datacache added as a direct dependency (it was only transitive via the
optional pyensembl extra). Verified end-to-end: `reference fetch
hpa_rna_consensus` downloads + decompresses to the identical 37,949,495-byte
TSV. Cache location moves to the appdirs dir (behavior change, intended).

Tests rewritten to stub datacache's downloader instead of urllib. 424 pass.

Coverage Stats

2682 of 3625 relevant lines covered (73.99%)

0.74 hits per line

Source File

Press 'n' to go to next uncovered line, 'b' for previous

Source Not Available