Ilia Denisov
74ff3aefb6
build / dawg (pull_request) Successful in 6m4s
The list leaked abbreviations two ways: an OpenCorpora noun tagged Abbr without Fixd (the declinable acronym ОГВЗ) slipped past the Abbr+Fixd seed filter, and the РАН «(сокр.)» words (вуз, нэп, колхоз, гост, …) were kept on purpose by the old policy. Drop every abbreviation instead. A final pass in ru_stage2.py removes a word when the orthographic note marks it «(сокр.)» (is_abbrev_note) or when every OpenCorpora noun reading carries Abbr (oc_abbr_only — also the declinable огвз/заз the seed filter missed). Homographs with a plain-noun reading survive (ваза «ВАЗ», рис). Lexicalised words are rescued via manual_confirm.txt (колхоз, совхоз, рация, спецназ, эсминец, дзот, дот, прораб, токамак, тэн, … and под — a real noun OC knows only as ПОД); the bound prefix гос is vetoed in manual_reject.txt. 108 words removed from ru_scrabble / ru_erudit; en_sowpods unchanged. tools/README documents the policy and that libmorph is required for a faithful rebuild.
scrabble_ru source
scrabble.txt — Russian Scrabble common nouns (nominative singular), produced by the prep
pipeline under ../../tools/ from the Russian academic orthographic
dictionary, cross-checked against OpenCorpora and libmorph. manual_confirm.txt holds the
hand-reviewed additions the pipeline merges in. Built to dawg/ru_scrabble.dawg (make dawg-ru).
The pdftotext source of truth (orfo_dict_2025.txt) and the Stage-1 base (all.txt) are
committed here; only the debug dumps (undefined.txt, adjectives.txt, verbs.txt,
singulars.txt, fate.tsv) are regenerated locally and git-ignored.