Skip to content

v0.6.0

@retospect retospect tagged this 20 Apr 12:55
Sidecar overrides can now clear stale s2_id/arxiv_id and bypass verify.

- Add s2_id and arxiv_id to _SIDECAR_OVERRIDE_KEYS so a .meta.json can
  clear a wrong identifier left over from the pre-sidecar Crossref/S2
  lookup. An explicit JSON null clears the field; empty string remains a
  no-op for backward compatibility.

- Recognize "verified": true in the sidecar as an attestation that the
  user has manually confirmed the metadata matches the PDF. When set,
  verify_metadata is skipped entirely. Fixes papers whose real title is
  not prominent on page 1 (Elsevier header strips, ACS production PDFs,
  abstract collections) and was scored below the fuzzy-match threshold.

- Tests: TestSidecarOverrideKeys and TestSidecarVerifiedOptIn covering
  both fixes. README updated with the new keys and clearing semantics.
Assets 2
Loading