Skip to contents

Changelog

Source: NEWS.md

VasGBIF 3.5.2

Website

  • Added search and dark/light mode toggle to the pkgdown navbar.
  • Simplified navbar labels: “Get started” → “Start”, “Reference” → “Functions”, “Articles” → “Manuals”.
  • Reorganized reference index: detect_native_status() and restore_duplicates() moved under Utilities.
  • Workflow diagram converted from JPG to PNG.

Documentation

  • Fixed Unicode characters in set_vouchers() documentation (✓, ✗, −) that caused the PDF manual to fail building under LaTeX.
  • Updated Distributions data source URL from http to https.
  • Added @seealso reference in WorldLandMap documentation.
  • README: updated codecov and R-CMD-check badge links, added GBIF DOI citation, rephrased performance statement.
  • Vignette Example.Rmd: all code chunks set to eval = FALSE to avoid network-dependent failures during R CMD check.
  • Vignette Application.Rmd: fixed citation format and added missing letsR reference.
  • Vignette GetRecords.Rmd: fixed table column alignment.
  • Reported retention rate updated from ~35% to ~30% across package docs and vignettes.

Testing

  • Added testthat snapshot files for set_threads() and import_records() error messages.
  • Added comprehensive test suite across five core functions (81 new tests, 177 total):
    • set_vouchers() (54 tests): return structure, verbatim_quality scoring (0–9), geospatial_quality scoring (0 to -9), moreInformativeRecord calculation, non-groupable record handling, voucher selection, tie-breaking, coordinate propagation, taxonomic consensus (identified / divergent / unidentified), and final classification (usable / duplicate / unusable).
    • get_collections() (26 tests): return structure, key format taxon|date|lat|lon, precision-controlled coordinate rounding, complete vs. incomplete key counting, precision parameter validation, and eventDate parsing (full dates, dates with time, year-month only).
    • restore_duplicates() (26 tests): return structure, restoration of all eight metadata fields (eventDate, year, month, day, identifiedBy, countryCode, stateProvince, locality), integer coercion for year/month/day, "NA" string treated as missing, skipping of values >10,000 characters, special character stripping, and first-available-duplicate selection.
    • detect_native_status() (14 tests): return structure, unknown status for taxa absent from Distributions, known-native classification (e.g. Rosa canina in Denmark), independent multi-taxon processing, and verification of the fcase priority logic (location_doubtful > introduced > extinct > native > unknown).
    • refine_records() (15 tests): return structure, native_status column integration, valid coordinate pass-through, zero-coordinate flagging by the "zeros" test, equal lat/lon flagging by "equal", empty tests parameter behaviour, and exclusion of "unusable" records by restore_duplicates().

VasGBIF 3.5.1

Bug Fixes

  • check_taxon(): fixed merge failure for names containing commas. The TNRS internally converts commas to spaces in submitted scientific names, but the merge key on the package side used the original comma-containing name, causing records to fail to match their TNRS resolution results. A comma-stripped key column (wcvp_searchedName2) is now created for the merge, while the original wcvp_searchedName is preserved unchanged.
  • refine_records(): removed erroneous pre-filter before restore_duplicates(). The call restore_duplicates(voucher$occ_digital_voucher[VasGBIF_dataset_result != "unusable"]) stripped all unusable records from the input.

VasGBIF 3.5.0

Major release: package renamed from UltraGBIF to VasGBIF with a substantially simplified pipeline.

Breaking Changes

Package Rename

  • The package has been renamed from UltraGBIF to VasGBIF. All function names, class names, and column prefixes have been updated accordingly (e.g., UltraGBIF_dataset_result → VasGBIF_dataset_result).

Removed Functions

  • check_collectors() and get_collectors_name() have been removed. The collector-based duplicate detection stage has been replaced by a coordinate- and date-based collection-event key system in get_collections().
  • check_occ_taxon() replaced by check_taxon().
  • set_collection_mark() replaced by get_collections().
  • set_digital_voucher() replaced by set_vouchers().

Removed Data

  • occ_import, ref_wcvp_names, seas_ref, and wcvp_distributions datasets have been removed from the package. The internal datasets Distributions, WGSRPD3, WorldLandMap, and EnumOccurrenceIssue remain available.

Function Signature Changes

  • import_records(): parameter GBIF_file renamed to path; only_PRESERVED_SPECIMEN removed; new remove_tempfile parameter.
  • refine_records(): parameter export_path removed (use export_records() instead); save_path removed; return value is now a refined object with elements all_records, CoordinateProblematic, and runtime.
  • check_taxon() replaces check_occ_taxon(): default sources changed from c("wcvp", "wfo") to "wcvp" only.

New Functions

  • detect_native_status() detects native status by matching validated coordinates to WGSRPD Level 3 polygons and WCVP distribution data. Classification uses a priority-based scheme: location_doubtful > introduced > extinct > native > unknown.
  • export_records() writes refined records to disk as three gzip-compressed CSV files: all usable records, the native subset, and records that failed coordinate validation.
  • restore_duplicates() is a new internal helper that fills missing metadata fields (eventDate, year, month, day, identifiedBy, countryCode, stateProvince, locality) in usable voucher records using data from duplicate records sharing the same collection key.

Improvements

  • Simplified pipeline: The workflow has been reduced from 8 modules (with collector checking) to 7 core functions across 4 stages. The collector name standardization stage has been removed entirely; duplicate detection now uses composite keys of taxon_name|eventDate|latitude|longitude via get_collections().
  • refine_records() has been modularised into a three-step internal pipeline: restore_duplicates() → CoordinateCleaner → detect_native_status().
  • get_collections() builds collection-event keys from resolved taxon names, event dates, and rounded coordinates, with user-controlled spatial precision.
  • set_vouchers() implements a refined quality scoring system (metadata completeness + geospatial penalty) and returns a vouchers object.
  • restore_duplicates() now restores month and day in addition to the previously restored fields.
  • set_threads() is now exported for normalising thread counts.

Dependency Changes

  • Removed tokenizers from Imports (no longer needed after collector module removal).
  • Added lubridate to Imports.
  • Added @import bit64 to NAMESPACE.

Documentation

  • All roxygen2 documentation rewritten in Markdown-first style (roxygen2 ≥ 8.0.0).
  • RoxygenNote replaced by Config/roxygen2/version: 8.0.0 in DESCRIPTION.
  • Pipeline overview in package documentation and README updated to reflect the seven-function, four-stage workflow.
  • Tutorial_of_UltraGBIF.Rmd vignette replaced by Application.Rmd and Example.Rmd.
  • utils::globalVariables() expanded to suppress data.table NSE notes in R CMD check.
  • Author field de-anonymised with ORCID.

VasGBIF 3.4.1

Bug Fixes

  • fix some big integers with bit64
  • fix function refine_records with identifiedBy

VasGBIF 3.4.0

Improvements

  • change Record Completeness Score calculation in set_digital_voucher(), replace fieledNotes with identifiedBy.
  • test the package on R 4.6.0

VasGBIF 3.3.2

Improvements

  • fix some help documents.

VasGBIF 3.3.1

Improvements

  • update lots of help documents.
  • remove functions about richness.

VasGBIF 3.3.0

Improvements

  • check_occ_taxon Implemented timeout and automatic redial logic for data chunks to resolve connection hangs during TNRS queries, ensuring robust processing of large datasets under unstable network conditions.

VasGBIF 3.2.9

Bug Fixes

  • debug check_occ_taxon

Improvements

  • perf(check_occ_taxon): Optimize TNRS query performance by processing large datasets in chunks

VasGBIF 3.2.8

Improvements

  • Enhanced TNRS error handling: Improved the retry mechanism in check_occ_taxon() with proper error catching using tryCatch() instead of try(). The function now provides informative error messages and gracefully handles network failures when the TNRS API is unreachable.
  • Added LazyData compression: Added LazyDataCompression: gzip to DESCRIPTION to reduce package size and meet CRAN requirements.

Bug Fixes

  • Fixed check_occ_taxon() retry logic crash: Resolved an issue where TNRS API failures caused the function to hang indefinitely. Changed initialization from NULL to empty data.frame() and added proper error handling to ensure nrow() operations work correctly during retry attempts.

Documentation

  • Standardized DOI citation format: Updated all R documentation files to use \doi{} macro instead of \url{https://doi.org/...} for academic reference formatting across VasGBIF-package.R, import_records.R, check_occ_taxon.R, refine_records.R, plot_richness.R, and data.R.
  • Updated README.md: Refreshed the README with the new four-stage workflow diagram, updated installation instructions, and refined descriptions of all modules.
  • Complete tutorial restructuring: Rewrote Tutorial_of_VasGBIF.Rmd following a 4-stage, 8-module structure with academic English, clear organization, and comprehensive workflow explanations. Added summary section for core modules with statistics table explaining data reduction process.

VasGBIF 3.2.7

Major Improvements

Dependency Restructuring

  • Removed dependency on rWCVP and rWCVPdata: VasGBIF now fully relies on the Taxonomic Name Resolution Service (TNRS) for taxonomic name resolution instead of the discontinued rWCVP packages. This critical change enables VasGBIF to meet CRAN submission requirements, as packages published on CRAN may only depend on other CRAN-hosted packages. Users can now install VasGBIF with a single install.packages("VasGBIF") call.
  • Streamlined taxonomic databases: The required WCVP databases are now bundled within the package, eliminating external data dependencies and ensuring consistent behavior across installations.

Taxonomic Name Resolution

  • Integrated TNRS (Taxonomic Name Resolution Service): Implemented a more mature and widely adopted name correction scheme that fully replaces the original complex scripts. The TNRS queries the World Checklist of Vascular Plants (WCVP) to resolve synonyms, correct misspellings, and standardize plant scientific names.
  • Enhanced taxonomic workflow: The check_occ_taxon() function now provides detailed TNRS workflow documentation, including the four-step resolution process (Parse → Match → Correct → Select Best Match), making the taxonomic standardization process transparent and understandable for users.

Performance Optimization

  • Vectorized Set Digital Voucher: For the most time-consuming step in the “Set Digital Voucher” stage—the “Process taxonomic information for groupable records” phase—vectorization techniques have been fully leveraged, resulting in a further 40% speedup of the set_digital_voucher() function.
  • Optimized richness calculation: Deeply integrated and optimized the richness calculation and heatmap plotting functionality inspired by lets.presab.points and plot.PresenceAbsence from the letsR package (Vilela & Villalobos, 2015). The implementation fully leverages vectorized terra operations to avoid explicit looping when filling large presence-absence matrices, achieving nearly a hundredfold speedup over traditional approaches.
  • Efficient collection event key generation: Optimized the algorithm for generating collection event keys, reducing computational overhead in the set_collection_mark() function.

Functionality Enhancements

  • Refactored collector name standardization: Completely restructured the collector name normalization and collection event mark generation functionality. The new check_collectors() function simplifies the workflow by using tokenizers::tokenize_words() for robust word tokenization, replacing the previous complex string splitting logic with multiple parameters.

  • New S3 class system: Introduced R’s S3 class system for VasGBIF objects, providing a clearer workflow structure with well-defined return types:

    • VasGBIF_import class for import_records() output
    • VasGBIF_taxa_checked class for check_occ_taxon() output
    • VasGBIF_collection_key class for set_collection_mark() output
    • VasGBIF_voucher class for set_digital_voucher() output
    • VasGBIF_refined class for refine_records() output

  • Deleted deprecated ref_dictionary data: Removed the built-in collector name reference dictionary (ref_dictionary) and related functions, reducing package size and simplifying dependencies.

Bug Fixes

  • Fixed plot_richness.R raster value extraction error: Resolved an intermittent error (invalid name(s)) in the terra::values() operation by adding cell ID validation and using direct raster object access instead of the unstable layer name accessor (ras_richness$richness).

Documentation

  • Comprehensive roxygen2 documentation: Systematically optimized and enriched help documentation across all 8 R modules. Every exported function now includes:

    • Structured @description with workflow explanation using \itemize
    • Detailed @param specifications with type and default value information
    • Multi-section @details with \strong subsection headers
    • Comprehensive @return descriptions
    • Academic references where applicable

  • Package-level documentation: Added complete package-level documentation in VasGBIF-package.R, providing:

    • Three-stage workflow overview (Data Acquisition → Duplicate Removal → Refine Records)
    • Eight-module function reference with cross-links
    • Quick start example code demonstrating the complete workflow
    • Performance optimization technical details (C/C++ backend integration, vectorization, SIMD exploitation, memory-efficient design, chunk-based parallelization)
    • Numbered reference list for all cited literature

  • Updated README.md: Refreshed the README with the new four-stage workflow diagram, updated installation instructions, and refined descriptions of all modules.

  • Unified reference formatting: Standardized all citations across documentation files to use \enumerate for numbered reference lists.

Code Quality

  • Cleaned code formatting: Removed redundant imports and standardized code style across all R files.
  • Removed deprecated functions: Deleted obsolete functions including check_occ_name(), prepare_collectors_dictionary(), generate_collection_mark(), and usecores(), replacing them with their more streamlined successors.