diff options
Diffstat (limited to 'tools')
| -rw-r--r-- | tools/extract_fiuv_ordo.ml | 21 |
1 files changed, 21 insertions, 0 deletions
diff --git a/tools/extract_fiuv_ordo.ml b/tools/extract_fiuv_ordo.ml index 42d3b96..c9688a7 100644 --- a/tools/extract_fiuv_ordo.ml +++ b/tools/extract_fiuv_ordo.ml @@ -292,7 +292,28 @@ let extract_te_deum full_text = | None -> String.length full_text in let span = String.sub full_text mat_start (laudes_start - mat_start) in + (* TWO negative phrasings the source actually uses, found live + (celebrant-rubrics-phase1 Phase 2, 2026-08-22): "non dicitur Te + Deum" (e.g. ordinary Time-after-Pentecost ferias, 27/28 November) + AND, separately, "sine Te Deum" (e.g. every Sunday, every Ember + day, every privileged Lenten/Passiontide feria carrying a + commemoration -- Advent I, 30 November: "...3a de homilia (cum + suo R), sine Te Deum."). The ORIGINAL version of this function + checked only the first phrasing, so "sine Te Deum" fell through + to the bare "Te Deum" substring test and was wrongly read as a + POSITIVE hit -- confirmed by grepping the raw pdftotext dump + directly against a first, uncorrected run's own output: EVERY + date this bug affected showed "sine Te Deum" in the source and + [Some true] in the fixture, a 100% correlation, not a handful of + coincidences. This is the SAME "does the source negate the + hymn's own name with a DIFFERENT word than the one this parser + already checks for" shape [find_word]'s own "Gloria Patri" + substring trap already documents for Gloria -- this trap simply + went unnoticed until Phase 2 actually compared the extracted + values against colitur's own output and against the raw text by + hand, rather than only checking coverage counts. *) if contains span ~sub:"non dicitur Te Deum" then Some false + else if contains span ~sub:"sine Te Deum" then Some false else if contains span ~sub:"Te Deum" then Some true else None) |