Version 3.03

Improved smart deduping, now finds and eliminates more duplicated games. Sorry I made the V3.03 improvement in June 2021 and didn't fully commit it and make a new release here until February 2022 even though the code and the executable I used (which identifies itself as V3.03) was unchanged over that 8 months. Only pgn2line.exe has changed, the other programs are unchanged.

Version 3.02

Bug fix, the white list feature now repaired - it seems I never tested it as although games from white listed tournaments were being included as intended, games from other tournaments were not being excluded, rendering the feature rather moot!

Also, I have extended tests for the UTF8 BOM to all the ancillary programs.

Version 3.01

Fix bug: If there was no month discontinuity (i.e. if there were games every month for years) then the buffer used for refinement sort was not periodically flushed and could grow to gigabytes, potentially causing a malloc exception and an apparent crash.

Version 3.0

Three important improvements;

1. The tie breaker field remains very useful in improving final sort order BUT it was severely limiting the usefulness of lpgn files for comparison purposes (and slightly reducing their utility for general game management by making the game prefix a less cogent game summary). Solution - keep the tie breaker during intermediate operations, but strip it out of the final lpgn. Best of both worlds, brilliant!

2. Smart de-duping wasn't very smart. It was basically declaring every game with identical meta-data to be a dup, way too many false positives with that approach. Basically smart deduping was useless - it threw away good games. The fix is to check for identical meta data PLUS identical main line moves (It turns out it's not too difficult to parse out the main line moves - eliminating annotations, comments and variations and then convert to binary moves for a precise comparison using our well established THC chess library). Smart de-duping now very useful!

3. If a UTF8 BOM is detected as the start of input - write one at the start of output.

Version 2.0

Improved sorting. A new sort tie breaker field before White players's surname added to prefix. The idea is that if everything ahead of that point in prefix matches (tournament/date/round) then ordering will respect original source rather than the (rather arbitrary) alphabetic ordering of White player's surname.

Also round field is now multiple (not just two) subfields separated by period characters. All subfields three digit which is sadly incompatible with V1.x where first of the two subfields was two digit. The sort tie breaker also disturbs V1.x compatibility, sorry. The other programs in the suite (apart from pgn2line itself) are unaffected and unchanged.

Added -p option, a very useful way to improve workflow, it automatically generates a PGN file (does line2pgn internally) as the last step of (pgn2line). PGN file is named by appending ".pgn" to output name.

Bug fixed: refinement sort handles edge cases better.

Bad bug fixed: we were actually always doing smart de-dup even if the flag wasn't specified.

Refinements, bug fixes, new features.

The main bug fixed is that the "wordsearch" program had an off by one bug that meant for example that searching for player "Ker" would also give you games by "Kerr". Important TODOs have been knocked off, notably a smart
de-duplication feature has been added (optionally games can now be considered duplicates if their game summary meta data matches - allows annotated game say to match bare game - annotated game kept (it's a dumb algorithm though - annotated game only kept because it's textually longer). Also Player names can now be added to the fixup file. Another new feature is the "players" program which reports on players in a similar way as the existing "tournaments" reports on tournaments.

Bugfix plus new wordsearch program added to suite

Edward Bauman has been using the program and he reported the first bug, fixed here. PGN format does require a blank line at the end of the file to terminate the last game, but it is nice to not rely on that. Bug fixed in v1.1. Also a new wordsearch.exe program has been added, for Windows users who cannot necessarily go grep -w. See README.md for an explanation

Initial release

This is the very first release, but I have been using and refining the program for sometime before putting it on Github, so it is already very useful.