Changes in version 0.3.0 (2026-11-08)                  

First CRAN release.

Breaking changes

  - Remove the GitHub-only dataexcept dependency so that every
    dependency is available from CRAN. Request failures are still
    signalled as abstract_extract_request_error conditions with the
    endpoint in $source and the original error in $parent, but they no
    longer inherit from dataexcept_error or serialise to a dataexcept
    JSON envelope.

Documentation

  - Generate all help pages from roxygen2 comments, so that every
    exported function documents its return value and has a runnable
    offline example.

                        Changes in version 0.2.0                        

Internal development release (not published on CRAN). It consolidates
the changes listed below: native Crossref, OpenAlex and Scopus search;
DOI reconciliation; quality diagnostics; reproducible SearchSnapshot
evidence; SHA-256 research archives; longitudinal metadata comparisons;
offline integration and clean-install tests.

                        Changes in version 0.1.0                        

Initial development release.

  - Implement Scopus search with cursor pagination and article
    normalization.
  - Support Crossref DOI lookups and bibliographic article discovery.
  - Enrich missing abstracts using reusable in-memory DOI caching.
  - Add bounded retry policies and configurable request intervals.
  - Export normalized article records as data frames, CSV, BibTeX, and
    RIS.
  - Provide offline examples, unit tests, and full HTML/PDF
    documentation checks.

This package has not yet been submitted to CRAN.

Development additions (unreleased)

  - Add deterministic DOI normalization, article deduplication and an
    audit trail for conflicting metadata across Crossref and Scopus.

  - CRAN submission is deferred until the package matures.

  - Add metadata completeness and structural validity scoring with
    customizable field weights, DOI-duplicate flags and issue reports
    (unreleased).

  - Add OpenAlex works search, paginated retrieval, abstract
    reconstruction, and normalized article metadata (unreleased).

  - Use the pinned GitHub dataexcept R package for structured request
    failures, preserving the existing request error class and cause
    (unreleased).

  - Add bounded multi-provider search with provenance and error
    diagnostics (unreleased).

  - Add versioned, credential-free search plans and reproducible RDS
    snapshots containing results, provenance, conflicts, and UTC
    timestamps (unreleased).

  - Add JSON search manifests and DOI-based comparisons between saved
    search snapshots (unreleased).

  - Add chronological search histories and safe CSV metadata-drift
    reports (unreleased).

  - Add explicit on-demand search refresh with non-overwriting evidence
    bundles and drift reports (unreleased).

  - Add local append-only research archives with chronological history
    (unreleased).

  - Add local archive checksums, cooperative writer locks and explicit
    interrupted-write recovery (unreleased).

  - Introduce SHA-256 integrity v2 with protected-run policy and
    backward compatibility for previous MD5 archive runs (unreleased).

  - Add manifest-based archive inventory and validated per-run selective
    access for large archives (unreleased).

  - Add bounded and selective archive comparison without loading all RDS
    snapshots (unreleased).

  - Speed up archive append by avoiding full historical RDS loading and
    add optional recent-only refresh results (unreleased).

  - Add per-run DOI indexes and historical DOI observation tracing
    (unreleased).

  - Add DOI-level longitudinal metadata timelines with abstract
    fingerprints and observed gaps (unreleased).

  - Align refresh documentation with the API and qualify stats helper
    imports; use a warning-free development check independently of CRAN
    eligibility.

  - Add representative frozen JSON contract fixtures for three providers
    and an explicitly invoked live smoke-test script (unreleased).

  - Add bounded pagination stress tests and reject unusable provider
    pages; redact contact email in provider error reports (unreleased).

  - Reduce repeated allocations in DOI reconciliation, metadata-quality
    issue reporting and manifest indexing; add larger offline regression
    datasets.

  - Add deterministic end-to-end research workflow regression tests
    covering search, reconciliation, quality assessment, archive refresh
    and exports.

  - Consolidate primary research workflows in getting-started and API
    documentation, with a regression test of the documented offline path
    (unreleased).

  - Add clean source-tarball installation smoke tests and R 4.2.3
    optional compatibility checks.

  - Audit development-release validation evidence and correct outdated
    archive locking and checksum documentation (unreleased).

  - Add an opt-in manual GitLab job for live provider connectivity and
    normalization checks, isolated from routine offline CI (unreleased).