rite(ef): RG 91 Table of Precedence

Precedence_ef.band transcribes RG 91's 28-entry Table of Precedence
(rules-register.md §4) for the EF rite: given a day's context and a
candidate celebration, returns the table's own entry number, 1-28
(I class 1-13, II class 14-21, III class 22-26, IV class 27-28); lower
wins. Every branch carries its entry number and register citation in a
comment, checked in the table's own numeric order.

Two entries are transcribed as the register states them even though
they invert the pattern the rest of the table follows: at III class,
23 (particular calendars) outranks 24 (universal), the reverse of how
11/12 and 14/16/19/20 rank a universal feast ahead of a proper one at
I and II class.

Sanctoral-origin, layer-decided entries (11-13, 14/16/19/20, 23/24)
follow the brief's structural insight: a celebration whose layer is
not the universal base is an overlay -- proper, or indult if its
layer id also carries the indult prefix. Neither the universal-layer
id nor the indult prefix is an RG citation; both are colitur's own
data-modelling convention, exposed from the module so whichever task
loads the real EF sanctoral overlays can align to them.

Vigils (21, 26) are read off the temporal cycle's own -vigil slug
suffix rather than gated on origin, since a II/III-class vigil can be
either temporal-origin (Ascension, already produced by temporal_ef) or
sanctoral-origin (a saint's vigil, no task has loaded yet); Ember days
(part of entry 18) are read off temporal_ef's own ember slug prefixes
rather than re-derived, since the September anchor is independently
flagged there as one of the more contested dates in the calendar.

A candidate shape the table has no row for (e.g. a Class1 vigil that
is not Nativity or Pentecost, or a Class4 candidate marked as a vigil
-- RG 91 has no IV-class vigil either) returns a dedicated unclassified
sentinel (max_int) rather than being folded into a same-rank entry it
does not belong to.

test_precedence_ef.ml is table-driven: one Alcotest.test_case per RG
91 entry (55 rows total, several entries covered by more than one
named day so a single missed offset cannot hide behind a passing
sibling), each date computed from Computus.gregorian_easter rather
than hand-typed, so an arithmetic slip cannot pass by accident.

1 files changed, 202 insertions, 0 deletions

diff --git a/test/test_precedence_ef.ml b/test/test_precedence_ef.ml
new file mode 100644
index 0000000..fb05582
--- /dev/null
+++ b/test/test_precedence_ef.ml
@@ -0,0 +1,202 @@
+(* RG 91's Table of Precedence, transcribed by Rite_ef.Precedence_ef.band.
+   Table-driven, one row (hence one Alcotest.test_case) per RG 91 entry, so a
+   misplaced or missing entry names itself in the failure output instead of
+   failing anonymously (docs/research/rules-register.md §4). Each row's date
+   is checked against the register to make sure it is not ALSO an instance of
+   some other entry at the same band (the vacuous-test trap this project has
+   caught before -- see the Advent-Ember-day note on entry 18 below). *)
+
+module P = Colitur_kernel.Precedence
+module Cel = Colitur_kernel.Celebration
+module S = Colitur_kernel.Slug
+module Col = Colitur_kernel.Colour
+module D = Colitur_kernel.Date
+module Sub = Colitur_kernel.Subject
+module Comp = Colitur_kernel.Computus
+module T = Rite_ef.Temporal_ef
+module V = Rite_ef.Vocab_ef
+module PE = Rite_ef.Precedence_ef
+
+let mk y m dd = match D.make ~year:y ~month:m ~day:dd with Ok t -> t | Error e -> failwith e
+
+(* [T.season] is the same function Calendar itself would use to build a
+   context, so a row's [season]/[weekday] are exactly what the real engine
+   would compute for that date, not a hand-picked value that might not
+   actually occur together with it. *)
+let ctx date = { P.date; season = T.season date; weekday = D.weekday date }
+
+let cand ?(origin = P.Temporal) ?(rank = V.Class1) ?(subject = Sub.Temporal) ?(layer = "temporal")
+    slug =
+  { P.cel = Cel.make ~slug:(S.of_string_exn slug) ~rank ~colour:Col.White ~subject ~layer ();
+    origin }
+
+(* Every Easter-relative date below is anchored to this single computed
+   Easter rather than a hand-typed calendar date, so an arithmetic slip in a
+   test date cannot silently pass by accident. *)
+let easter = Comp.gregorian_easter 2026
+let off n = D.add_days easter n
+
+(* (description, date, candidate, expected RG 91 entry). *)
+let cases =
+  [ (* Entry 1 -- register line 327: Nativity, Easter Sunday, Pentecost Sunday. *)
+    ("1 Nativity", mk 2026 12 25, cand "ef-nativity", 1);
+    ("1 Easter Sunday", off 0, cand "ef-easter-sunday", 1);
+    ("1 Pentecost Sunday", off 49, cand "ef-pentecost", 1);
+    (* Entry 2 -- register line 328: Sacred Triduum. Thu-Sat of Holy Week,
+       NOT entry 7 (which stops at Wednesday -- see entry 7 below). *)
+    ("2 Holy Thursday", off (-3), cand "ef-holy-thursday", 2);
+    ("2 Good Friday", off (-2), cand "ef-good-friday", 2);
+    ("2 Holy Saturday", off (-1), cand "ef-holy-saturday", 2);
+    (* Entry 3 -- register line 329. *)
+    ("3 Epiphany", mk 2026 1 6, cand "ef-epiphany", 3);
+    ("3 Ascension", off 39, cand "ef-ascension", 3);
+    ("3 Trinity", off 56, cand "ef-trinity", 3);
+    ("3 Corpus Christi", off 60, cand "ef-corpus-christi", 3);
+    ("3 Sacred Heart", off 68, cand "ef-sacred-heart", 3);
+    ("3 Christ the King", T.christ_the_king 2026, cand "ef-christ-the-king", 3);
+    (* Entry 4 -- register line 330. Sanctoral-origin: neither feast is part
+       of temporal_ef's movable cycle. *)
+    ( "4 Immaculate Conception", mk 2026 12 8,
+      cand ~origin:P.Sanctoral ~subject:Sub.Bvm ~layer:PE.universal_layer
+        "ef-immaculate-conception",
+      4 );
+    ("4 Assumption", mk 2026 8 15, cand ~origin:P.Sanctoral ~subject:Sub.Bvm ~layer:PE.universal_layer "ef-assumption", 4);
+    (* Entry 5 -- register line 331. *)
+    ("5 Nativity Vigil", mk 2026 12 24, cand "ef-nativity-vigil", 5);
+    ("5 Octave day (Circumcision)", mk 2026 1 1, cand "ef-circumcision", 5);
+    (* Entry 6 -- register line 332. *)
+    ("6 Advent Sunday", T.advent_start 2026, cand "ef-advent-sunday-1", 6);
+    ("6 Lent Sunday", off (-42), cand "ef-lent-sunday-1", 6);
+    ("6 Passion Sunday (I Passiontide)", off (-14), cand "ef-passion-sunday", 6);
+    ("6 Palm Sunday (II Passiontide)", off (-7), cand "ef-palm-sunday", 6);
+    ("6 Low Sunday", off 7, cand "ef-low-sunday", 6);
+    (* Entry 7 -- register line 333: Ash Wednesday and Mon/Tue/Wed of Holy
+       Week ONLY -- Thu-Sat are entry 2 above, not this entry. *)
+    ("7 Ash Wednesday", off (-46), cand "ef-ash-wednesday", 7);
+    ("7 Monday of Holy Week", off (-6), cand "ef-holy-monday", 7);
+    ("7 Tuesday of Holy Week", off (-5), cand "ef-holy-tuesday", 7);
+    ("7 Wednesday of Holy Week", off (-4), cand "ef-holy-wednesday", 7);
+    (* Entry 8 -- register line 334. *)
+    ("8 All Souls", mk 2026 11 2, cand ~origin:P.Sanctoral ~layer:PE.universal_layer "ef-all-souls", 8);
+    (* Entry 9 -- register line 335. *)
+    ("9 Pentecost Vigil", off 48, cand "ef-pentecost-vigil", 9);
+    (* Entry 10 -- register line 336: both range boundaries, to guard the
+       off-by-one an inclusive Easter-offset window invites. *)
+    ("10 Easter octave, day+1", off 1, cand "ef-easter-1-mon", 10);
+    ("10 Easter octave, day+6", off 6, cand "ef-easter-1-sat", 10);
+    ("10 Pentecost octave, day+50", off 50, cand "ef-pentecost-1-mon", 10);
+    ("10 Pentecost octave, day+55", off 55, cand "ef-pentecost-1-sat", 10);
+    (* Entry 11 -- register line 337. *)
+    ( "11 Universal I-class feast", mk 2026 6 29,
+      cand ~origin:P.Sanctoral ~subject:Sub.Saint ~layer:PE.universal_layer "ef-ss-peter-paul",
+      11 );
+    (* Entry 12 -- register line 338. The one non-base-layer case the brief
+       asks for explicitly: same date/rank/subject as 11, only the layer
+       differs, so this row isolates the layer test as the deciding factor. *)
+    ( "12 Proper I-class feast (non-base layer)", mk 2026 6 29,
+      cand ~origin:P.Sanctoral ~subject:Sub.Saint ~layer:"diocese-warsaw" "ef-local-patron",
+      12 );
+    (* Entry 13 -- register line 339. *)
+    ( "13 Indult I-class feast", mk 2026 6 29,
+      cand ~origin:P.Sanctoral ~subject:Sub.Saint ~layer:(PE.indult_prefix ^ "local-grant")
+        "ef-indult-feast-1",
+      13 );
+    (* Entry 14 -- register line 341. *)
+    ( "14 Feast of the Lord, II class", mk 2026 7 1,
+      cand ~origin:P.Sanctoral ~rank:V.Class2 ~subject:Sub.Lord ~layer:PE.universal_layer
+        "ef-precious-blood",
+      14 );
+    (* Entry 15 -- register line 342: an ordinary Sunday not named at entry 6
+       -- Septuagesima is II class (RG 11-12 names only Advent/Lent/
+       Passiontide/Easter/Low/Pentecost as I class). *)
+    ("15 II-class Sunday (Septuagesima)", off (-63), cand ~rank:V.Class2 "ef-septuagesima-sunday", 15);
+    (* Entry 16 -- register line 342. *)
+    ( "16 Universal II-class feast, not of the Lord", mk 2026 1 20,
+      cand ~origin:P.Sanctoral ~rank:V.Class2 ~subject:Sub.Saint ~layer:PE.universal_layer
+        "ef-some-saint",
+      16 );
+    (* Entry 17 -- register line 343: days WITHIN the Nativity octave (26-28
+       Dec are Stephen/John/Innocents -- sanctoral, not this entry; 1 Jan is
+       entry 5's Octave DAY, not this entry either). *)
+    ("17 Nativity octave, 29 Dec", mk 2026 12 29, cand ~rank:V.Class2 "ef-nativity-octave-day-5", 17);
+    ("17 Nativity octave, 31 Dec", mk 2026 12 31, cand ~rank:V.Class2 "ef-nativity-octave-day-7", 17);
+    (* Entry 18 -- register line 343-344: Advent 17-23 Dec ferias AND the
+       Ember days of Advent/Lent/September share this one entry. The second
+       row is deliberately a Lent date (season Lent, NOT Advent) to prove the
+       Ember-slug path fires on its own, not merely because it also happens
+       to fall in the Dec 17-23 window -- the exact trap the brief warns
+       about, worked the other way round: this Ember day must NOT be
+       mistaken for an ordinary entry-22 Lent feria either. *)
+    ("18 Advent 17-23 Dec feria", mk 2026 12 21, cand ~rank:V.Class2 "ef-advent-4-mon", 18);
+    ("18 Lent Ember Wednesday", off (-39), cand ~rank:V.Class2 "ef-lent-ember-wed", 18);
+    (* Entry 19 -- register line 344. *)
+    ( "19 Proper II-class feast", mk 2026 1 20,
+      cand ~origin:P.Sanctoral ~rank:V.Class2 ~subject:Sub.Saint ~layer:"diocese-warsaw"
+        "ef-local-saint-2",
+      19 );
+    (* Entry 20 -- register line 345. *)
+    ( "20 Indult II-class feast", mk 2026 1 20,
+      cand ~origin:P.Sanctoral ~rank:V.Class2 ~subject:Sub.Saint
+        ~layer:(PE.indult_prefix ^ "local-grant-2") "ef-indult-feast-2",
+      20 );
+    (* Entry 21 -- register line 345 (RG 28-34). Two rows: the Ascension
+       Vigil is the one II-class vigil temporal_ef already produces today
+       (temporal-origin); the Assumption Vigil stands in for the
+       sanctoral-origin case no task has loaded data for yet -- proving
+       [band] does not gate this entry on [origin] (see precedence_ef.ml's
+       file comment). *)
+    ("21 Ascension Vigil (temporal-origin)", off 38, cand ~rank:V.Class2 "ef-ascension-vigil", 21);
+    ( "21 Assumption Vigil (sanctoral-origin)", mk 2026 8 14,
+      cand ~origin:P.Sanctoral ~rank:V.Class2 ~layer:PE.universal_layer "ef-assumption-vigil",
+      21 );
+    (* Entry 22 -- register line 347-348 (corrected: ends at Palm Sunday, not
+       Passion Sunday). Both a Lent and a Passiontide feria, clear of Ash
+       Wednesday, Holy Week and the Ember days. *)
+    ("22 Lent feria", off (-41), cand ~rank:V.Class3 "ef-lent-1-mon", 22);
+    ("22 Passiontide feria", off (-12), cand ~rank:V.Class3 "ef-passiontide-1-tue", 22);
+    (* Entry 23 -- register line 349. NOTE the table's own order here is the
+       REVERSE of 11/12 and 14/16/19/20 above: entry 23 (particular
+       calendars) is numbered BELOW entry 24 (universal), so a proper
+       III-class feast outranks a universal one -- transcribed as the
+       register states it, not "corrected" to match the other classes. *)
+    ( "23 Proper III-class feast (non-base layer)", mk 2026 6 30,
+      cand ~origin:P.Sanctoral ~rank:V.Class3 ~layer:"diocese-warsaw" "ef-local-saint-3",
+      23 );
+    (* Entry 24 -- register line 349. *)
+    ( "24 Universal III-class feast", mk 2026 6 30,
+      cand ~origin:P.Sanctoral ~rank:V.Class3 ~layer:PE.universal_layer "ef-some-saint-3",
+      24 );
+    (* Entry 25 -- register line 350. *)
+    ("25 Advent feria to 16 Dec", mk 2026 12 1, cand ~rank:V.Class3 "ef-advent-1-tue", 25);
+    (* Entry 26 -- register line 350. *)
+    ( "26 III-class vigil", mk 2026 8 9,
+      cand ~origin:P.Sanctoral ~rank:V.Class3 ~layer:PE.universal_layer "ef-lawrence-vigil",
+      26 );
+    (* Entry 27 -- register line 352: an otherwise-unoccupied IV-class
+       Saturday. *)
+    ( "27 Office of the BVM on Saturday", off 62,
+      cand ~rank:V.Class4 "ef-time-after-pentecost-1-sat",
+      27 );
+    (* Entry 28 -- register line 352: the unqualified IV-class catch-all. *)
+    ("28 IV-class feria", off 65, cand ~rank:V.Class4 "ef-time-after-pentecost-1-tue", 28);
+    (* Not an RG 91 row at all: a I-class candidate marked as a vigil, which
+       is not the Nativity or Pentecost (entries 5/9, the only I-class
+       vigils the table names) and so has no entry to fall into. Proves the
+       documented fallback -- not entry 11/12/13, which the [not is_vigil]
+       guard exists specifically to keep this out of. *)
+    ( "unclassified: I-class vigil outside Nativity/Pentecost", mk 2026 3 10,
+      cand ~origin:P.Sanctoral ~layer:PE.universal_layer "ef-mystery-vigil",
+      PE.unclassified );
+    (* RG 91's own vigil list (register lines 381-384) stops at III class --
+       there is no IV-class vigil for entry 28's ferial catch-all to absorb. *)
+    ( "unclassified: IV-class candidate marked as a vigil", mk 2026 6 20,
+      cand ~rank:V.Class4 "ef-second-mystery-vigil", PE.unclassified )
+  ]
+
+let suite =
+  ( "Precedence_ef",
+    List.map
+      (fun (desc, date, c, expect) ->
+        Alcotest.test_case desc `Quick (fun () ->
+            Alcotest.(check int) desc expect (PE.band (ctx date) c)))
+      cases )