diff options
| author | Alexander_Kabui <alexanderkabua@gmail.com> | 2026-10-07 07:05:06 +0300 |
|---|---|---|
| committer | Alexander_Kabui <alexanderkabua@gmail.com> | 2026-10-07 07:05:06 +0300 |
| commit | 13a7cb117deadd7a8253e218ede7d6cd430c9da7 (patch) | |
| tree | 5484d665fc03b7a078be6787b90f29b4e82aef12 /test/python/test_version.py | |
| parent | a5c40fc78f6b3cb32fba3daeb9f2a8e95558ba09 (diff) | |
| download | genenetwork4-13a7cb117deadd7a8253e218ede7d6cd430c9da7.tar.gz | |
src/search/native.rs (native engine): register the boolean prefixes (C3.2).
Port search.py:220-228: author=A, species=XS, group=XG (through the lowercasing FieldProcessor, search.py:204-209/222-223 — xapian does not stem boolean prefixed fields when the value starts with a capital, so gn3 forces the lowercase term; the r1-test LowercaseGroupProcessor pattern), tissue=XI, dataset=XDS, symbol=XY, chr=XC, peakchr=XPC. The upstream cases rule (uppercase values produce XDS:BXDPublish-style terms — binding-report R2: gn3-inherited xapian behavior, not a registration bug) is documented, not pinned: no graded shape relies on it. tests/search_native_shapes.rs gains the C3.2 list (Q14-Q20, Q26, Q27, Q43, Q46 with per-shape parity.rs citations); native.rs gains a hermetic parser test pinning each registration to its exact index term (group:B6 -> XGbxd, not XGBXD). Evidence: logs/c3-2.log — build EXIT=0, lib tests green (incl. the new prefix assertions), FULL suite EXIT=0, SHAPE-CLASS GATE 27/27 PASS vs the dev index (cumulative C3.1+C3.2; dev counts recorded as reference: group:B6 = 3, symbol:insr = 1, peakchr:1 = 113, etc.).
Diffstat (limited to 'test/python/test_version.py')
0 files changed, 0 insertions, 0 deletions
