Skip to contents

Changelog

Source: NEWS.md

gao 0.5.0

  • CRAN preparation release.
  • Bundled RDS shrunk from 6.4 MB to 3.7 MB by storing only 14 core columns with xz compression. The 82 indicator columns are now computed on the fly by gao_links() and cached in memory.
  • New gao_update_data() function downloads the latest data from GitHub Releases using base R download.file() — no curl-impersonate needed. gao_links() checks for user-local cached data before the bundled copy.
  • auto_download() now offers to check for updated data in interactive sessions before proceeding.
  • Daily CI workflow now uploads the RDS to a pinned GitHub Release (data-latest) for gao_update_data() to fetch.
  • Fixed missing lifecycle-deprecated.svg badge referenced by extract_pdf_links.Rd.
  • Updated CITATION.cff to match current version and license.

gao 0.4.0

  • New extract_text() function for extracting text from downloaded PDFs. Requires pdftools (added to Suggests).
  • Bundled dataset now includes page_count, topics, and subject_terms columns. gao_links() returns a 9-column data.frame.
  • Full metadata backfill: title, published, released are now 100% populated across all 56,000+ reports. summary at 97.5%.
  • Missing report_id values filled from URL slugs (now 100% complete).
  • Page counts extracted from 55,000+ PDF archive and matched to metadata via URL slug and report ID (80.7% coverage).
  • Daily CI workflow now backfills page_count for newly added reports.
  • Fixed update_links() column mismatch when bundled data has columns that new scrape results lack.
  • License changed from MIT to GPL (>= 3).

gao 0.3.0

  • Breaking: gao_links() now returns a data.frame with columns url, title, report_id, published, released, and summary instead of a character vector.
  • Bundled dataset switched from text (.txt) to RDS (.rds) for compression with rich metadata.
  • Year filtering in auto_download() now uses published date and fiscal year calculation instead of regex on report IDs, fixing ~29% of reports with legacy ID formats that previously yielded NA years.
  • extract_links() and update_links() now return data.frames with full report metadata.
  • Fixed R-CMD-check GitHub Action syntax error (args parameter).

gao 0.2.0

  • Added auto_download() convenience wrapper that handles the full pipeline (load links, filter by year, download as PDF/HTML) in one call.
  • PDF URLs are now constructed directly from report IDs, avoiding one HTTP request per report compared to extract_pdf_links().
  • Interactive prompts for format and year range when arguments are omitted.
  • Non-interactive safety: confirm = TRUE errors unless explicitly set to FALSE, preventing accidental mass downloads.

gao 0.1.0

  • Initial release.
  • Bundled dataset of ~55,000 GAO report URLs (1921–present).
  • gao_links() to access bundled report URLs.
  • update_links() to scrape newly published reports.
  • extract_links() to build the full link list from scratch.
  • extract_pdf_links() to find PDF download links from report pages.
  • download_pdfs() and download_htmls() for batch downloading.
  • Requires ‘curl-impersonate’ for TLS fingerprint compatibility.