diff options
| author | Alexander_Kabui <alexanderkabua@gmail.com> | 2026-10-07 07:30:24 +0300 |
|---|---|---|
| committer | Alexander_Kabui <alexanderkabua@gmail.com> | 2026-10-07 07:30:24 +0300 |
| commit | d563910a0fc7f61e2a77cd7cc43643e2fe493939 (patch) | |
| tree | 4ab046a9c4169540d57da6f0781b81919fb2e86d /test/python/test_version.py | |
| parent | 95e9e7e920be5ccd4bf037b51bef6ab077b59c9a (diff) | |
| download | genenetwork4-d563910a0fc7f61e2a77cd7cc43643e2fe493939.tar.gz | |
src/search/native.rs (native engine): register the probabilistic prefixes and aliases (C3.3).
Port search.py:229-233: description=XD, rif=XRF, wiki=XWK, plus the uppercase aliases RIF=XRF and WIKI=XWK. With STEM_ALL_Z the parsed field term is Z+<prefix>+stem (e.g. wiki:diabetes -> ZXWKdiabet, matching the TermGenerator emission the production indexer writes), pinned hermetically by prefix in native.rs tests. tests/search_native_shapes.rs gains the C3.3 list (Q21-Q25, Q45 with per-shape parity.rs citations). Dev-index reference counts recorded, never compared to production: description:insulin=0, rif:insulin=0, RIF:insulin=0, wiki:diabetes=0 (phenotype type filter: XWK text lives only on gene documents, d.v. index-genenetwork:432-443), WIKI:diabetes=0, symbol:insr OR rif:insulin=1. The 0s are faithful, not registration failures: xapian-delve confirms the fields are populated on the dev index (term ZXWKdiabet: 24 gene docs; ZXRFdiabet exists) while ZXRFinsulin simply does not occur in the crawled RIF text and the phenotype-typed wiki shapes are filtered to an empty set exactly as gn3 would do on this index. Evidence: logs/c3-3.log — build EXIT=0, lib tests green (incl. the probabilistic-prefix assertions), FULL suite EXIT=0, SHAPE-CLASS GATE 33/33 PASS vs the dev index (cumulative C3.1+C3.2+C3.3); delve probe: /home/kabui/zxrf-all.txt, delve-result.txt.
Diffstat (limited to 'test/python/test_version.py')
0 files changed, 0 insertions, 0 deletions
