Changes in version 0.3.0.9000                      

                 Changes in version 0.3.0 (2026-08-24)                  

Major changes

  - Unified column rename mechanism: RSF periodically renames columns in
    their exports (e.g., 2025–2026: "Score" →
    "Score 2025"/"Score 2026"). All such renames now flow through a
    single, transparent override system
    (inst/extdata/period3_column_overrides.csv + three
    validation/override functions) instead of special-cased detection
    logic. This makes future RSF format changes maintainable without
    code edits — just append a row to the CSV. Includes three new safety
    net functions for per-column validation, override loading, and
    mapping patching.

  - Improved encoding robustness: Each raw RSF CSV is now scanned
    per-file for encoding (UTF-8 vs. ISO-8859-1 inherited from RSF's
    silent 2025 switch), preventing mojibake that would otherwise
    propagate silently into RDS datasets. Detection happens at download
    time; data are converted to UTF-8 during cleaning.

Internal refactoring

  - Removed detect_score_column() function (functionality moved to
    override mechanism, making score renames no longer special).
  - Simplified normalize_column_names() signature: dropped unused period
    and year parameters. Function now does pure mechanical
    rename/reorder; all period/year-specific logic moved to caller. This
    removes ~22 lines of code and clarifies the API boundary.
  - Test harness: suppressed console output from validation failure-case
    tests, so only "All validation checks passed!" appears in R CMD
    check output, avoiding false suspicion of package issues during CRAN
    review.

Compliance

  - Addressed all feedback from CRAN v0.2.0 rejection (Aug 8, 2026):
    DESCRIPTION URL, print method documentation, removal of unexported
    function examples, elimination of \dontrun{}, replacement of cat()
    with message(), removal of package-relative path defaults from
    internal functions.

                        Changes in version 0.2.0                        

Breaking changes

  - ETL/update functions (download_rwb_data(), clean_rwb_single(),
    clean_all_rwb_years(), combine_cleaned_periods(),
    standardize_rwb_countries(), update_rwb_data(), and related internal
    helpers) are no longer exported. They rely on a development checkout
    (here::here() paths, git operations) and were never usable from an
    installed copy of the package. The public API is now just the
    rwb_standardized dataset. Use pressfreedom.data:::fun() if you still
    need direct access.

  - zone now uses English region names instead of RSF's original French
    labels, e.g. "MENA" -> "Middle East & North Africa", "Afrique" ->
    "Africa". Update any code that filters or joins on the old French
    values (see ?rwb_standardized for the full mapping). This also fixes
    a recurring zone mojibake/encoding issue.

  - Fixed score_n_1 for 2023-2026 (720 rows), which was about 100x too
    large (e.g. 9265 instead of 92.65) due to a missing decimal-scaling
    step. It's now on the same 0-100 scale as score. Remove any manual
    rescaling workaround for these years.

Documentation

  - Clarified the rwb_ naming convention (RSF vs. RWB) in
    ?rwb_standardized.
  - De-duplicated content between README.md and
    vignette("getting-started"); consolidated citation info into
    inst/CITATION.
  - Added a \value tag to print.rwb_update()'s documentation and cited
    the RSF web service in the Description: field, addressing CRAN
    feedback on the initial 0.2.0 submission.

Internal (CRAN feedback, 0.2.0 resubmission)

  - Removed \examples from the unexported helpers download_rwb_data(),
    get_period(), and get_years_to_download() rather than exporting
    them, eliminating the package's only \dontrun{} usage.
  - download_rwb_data() now uses message() instead of unconditional
    cat() for progress/summary output, so it can be suppressed with
    suppressMessages().
  - Removed package-relative default paths (e.g. "inst/extdata",
    here::here("data", "cleaned")) from download_rwb_data(),
    get_years_to_download(), clean_rwb_single(), clean_all_rwb_years(),
    combine_cleaned_periods(), and standardize_rwb_countries(). These
    arguments are now required with no default, so the functions never
    write to (or read from) a package or home directory implicitly.

Other

  - New hex sticker logo, redesigned to be visually distinct from the
    companion pressfreedom Shiny app's logo.