CiteSource 1.0.0

First stable release, following acceptance on CRAN.

Bug fixes

• compare_sources() now honors include_references: when TRUE, the bibliographic columns are joined back onto the source/label/string indicator matrix, matching the behavior of count_unique(). Previously the argument was silently ignored and only indicator columns were returned.
• calculate_detailed_records(): the Total row's Distinct Records now respects labels_to_include. It previously used the unfiltered input, making the Total inconsistent with the filtered per-source rows and overstating distinct records when labels were restricted.

Improvements

• create_detailed_record_table() and create_precision_sensitivity_table() redesigned for readability: related columns are grouped under spanner headers, the contribution/uniqueness percentages have clearer labels, Precision and Recall are formatted as percentages, and the Total row is visually distinguished as a summary.

Documentation

• README now includes CRAN status and download badges.

Full Changelog: v0.2.1...v1.0.0

CiteSource 0.2.1 (CRAN)

CiteSource 0.2.1 — now on CRAN.

New features

• Incremental deduplication: dedup_citations_add_sources() adds new citations to a previously deduplicated set and deduplicates across both, preserving prior automatic and manual merge decisions and the original record_ids provenance. Exposed in the Shiny app.
• Deferred manual deduplication: run automatic dedup now and complete manual review later via export_dedup_candidates() / reimport_dedup_candidates(), with a manual_dedup_complete flag in export_csv() / reimport_csv().
• Shiny app: re-importing a deduplicated set shows a read-only source overview before adding more; the re-upload input accepts a candidate-pairs CSV and several files at once.

Bug fixes

• reimport_csv() now reads all columns as character, matching the canonical types produced by dedup_citations().
• Shiny app: the re-upload input no longer errors when more than one file is selected.

Documentation

• In-app User Guide and README updated to document incremental and deferred deduplication.

CiteSource 0.2.0

CiteSource 0.2.0

Breaking changes

• Requires R >= 4.1.0 (native pipe and across() syntax used throughout).

New features

• read_citations() now warns when cite_label values are outside the standard
  vocabulary (search, screened, final), since phase-analysis functions
  depend on those exact strings.

Bug fixes

• calculate_phase_records(): n_distinct() called as a terminal pipe no
  longer throws an error; fixed by replacing with summarise() |> pull().
• calculate_initial_records() / calculate_detailed_records(): separator
  regex in separate_rows() changed from "," to ",\\s*" so sources with
  spaces after commas are split correctly.
• generate_apa_reference(): NA DOI values no longer cause an error in
  str_detect(); handled with case_when(is.na(doi) ~ NA_character_, ...).
• Shiny app — summaryPrecTab: n_unique was referenced without the rv$
  prefix, causing a scope error; fixed.
• Shiny app — manual_dedup_dt: columnDefs targets now computed as
  0-based integer indices rather than column name strings, matching the DT API.
• count_unique() / compare_sources(): added !is.na() guard alongside
  != "" so NA values in cite_source, cite_label, cite_string no longer
  cause filter or pivot errors.
• Shiny app — rv$pairs_to_check[,1:36] subsetting removed; full data frame
  passed directly.

Performance

• generate_apa_citation() and generate_apa_reference(): all rowwise()
  calls replaced with purrr::map2_chr() and purrr::pmap_chr(), avoiding
  per-row dplyr group overhead.
• Shiny app: compare_sources() is now computed once in a shared reactive and
  consumed by both the heatmap and upset plot, halving the work on each render.
• Shiny app: multi-value column filtering replaced from per-row sapply +
  str_split to vectorized strsplit + vapply via a file-scope helper
  .filter_multivalue_col.

CRAN / dependency changes

• ASySD deduplication functions vendored directly into R/asys_dedup.R
  (GPL-3, with attribution to CAMARADES / Kaitlyn Hair); the GitHub-only
  ASySD package is no longer a dependency.
• Remotes: field removed from DESCRIPTION.
• New imports: cli, igraph, parallelly, RecordLinkage, utf8.
• Removed from imports: ASySD, plogr.
• Added to Suggests: bslib.
• .onLoad side effects removed; shiny.maxRequestSize is no longer set
  at package load time.
• %>% re-export removed; use the native |> pipe.

Deprecations

• record_counts() → calculate_initial_records()
• calculate_record_counts() → calculate_detailed_records()
• calculate_phase_count() → calculate_phase_records()
• record_counts_table() → create_initial_record_table()
• record_summary_table() → create_summary_record_table()
• precision_sensitivity_table() → create_phase_record_table()

All deprecated functions remain callable with a .Deprecated() warning
pointing to their replacements.

Internal

• Google Analytics injection simplified: URL-path reactive detection replaced
  by a file-scope CITESOURCE_ENV environment variable check evaluated once at
  app startup.