Changelog • tidygedcom

tidygedcom 0.2.0

readGedcom(parse_dates = TRUE) now retains dates that are precise only to the month or the year. Previously these became NA, which discarded most of the dates in a typical historical pedigree. Partial dates are completed with the midpoint of the known interval (the 15th of a known month, 15 June for a known year); set the new impute_partial_dates = FALSE to restore the stricter behavior and keep only dates that specify a day.
Date qualifiers written with a trailing period (Abt., Aft.), as Ancestry.com exports them, are now stripped. Previously only the specification’s unpunctuated forms (ABT, AFT) were recognized, so "Aft. Oct 1896" failed to parse.

readGedcom() no longer leaves a stray / in the name column when a name suffix follows the surname (e.g. 1 NAME William Pitt /Waugh/ Jr parsed as William Pitt Waugh/ Jr). Both surname delimiters are now removed.
readGedcom() no longer errors with “missing value where TRUE/FALSE needed” when a spouse record has no SEX line. Individuals with unknown sex are skipped during parent mapping rather than compared.

The “Getting Started” vignette now walks through exporting a GEDCOM file from a genealogy service, including the instability of person IDs across exports and a note on the privacy implications of sharing an export containing living people.
Added two example GEDCOM files, installed under inst/extdata, drawn from the W. Henderson Waugh Family Tree and covering eight individuals (all deceased) across four families:
- waugh.ged — a clean excerpt for documentation examples.
- waugh_messy.ged — the same individuals retaining source-export defects (conflicting duplicate BIRT blocks, competing PLAC lines, an uncertain surname, a missing SEX value, and an individual with no birth record).

Reorganized R source files for clearer separation of concerns: parsing logic split across parseIndividuals.R, parseEvents.R, parseLines.R, parseFamily.R, and postProcessGedcom.R; readGedcom.R is now a slim orchestration entry point only.
Extracted four constructor functions (make_event_fields(), make_name_piece_mappings(), make_attribute_mappings(), make_relationship_mappings()) that build static tag-mapping tables. These are called once in readGedcom() before the per-individual lapply, and passed in as a mappings argument to parseIndividualBlock(), avoiding redundant reconstruction on every block.
Replaced the processEventLine() if/else chain with a data-driven dispatch over the event_fields lookup table, reducing the cost of adding new event types to a single table entry.
Pre-allocated the blocks list in splitIndividuals() (vector("list", n)) rather than growing it element-by-element.
Split the monolithic test file into focused files: test-parseEvents.R, test-convertCoords.R, test-postProcessGedcom.R, and test-readGedcom.R.
Added Roxygen documentation for all new internal functions.

Optimized GEDCOM reader for speed and memory usage, with a focus on large pedigrees.
Fixed bug in GEDCOM reader that resulted in document records being added to the final person in the pedigree.
Added more unit tests for GEDCOM reader and data parser.
Several improvements to GEDCOM parsing, focusing on more robust and flexible event parsing, better support for different GEDCOM versions, and enhanced usability.