feat(of): Rubrics_of -- Gloria, Credo, and an all-None Preface verdict

Task 3 of OF Phases 3-5: the celebrant-rubrics module for the OF rite,
supplying three of Rite.t's four remaining fields.

creed/gloria implement IGMR n. 53 (Gloria) and n. 67-68 (Symbolum),
both quoted in full in the module header. temporal is read, not
observed, for both Sunday clauses: Temporal_of.temporal tags every
Sunday of Advent/Lent/Easter Sollemnitas (the privileged-season
convention), so an observed.rank test alone would wrongly grant Gloria
on an Advent or Lent Sunday, which n. 53's own "extra tempus Adventus
et Quadragesimae" explicitly withholds. Credo is deliberately narrower
than Gloria: n. 68 has no "et festis" clause, so an ordinary Festum
does not get the Creed, unlike EF's RG 475(b)/(c). Two Missale-proper
exceptions (Holy Thursday evening Mass, the Easter Vigil Mass) are
cited from the Missal's own "Dicitur Gloria in excelsis" rubric, not
the IGMR, which is silent on both. The Requiem question is likewise
IGMR-silent (nn. 379-385 say nothing about Gloria/Credo); flagged as
an inference from the Missal's own consistent marker pattern, applied
only to the one Requiem-shaped slug the shipped OF data can construct
(all-souls), the same closed-population discipline Rubrics_ef's own
Colour.Black proxy uses.

preface returns None unconditionally, a measured verdict rather than
a stub. The module header transcribes the 2002 Missal's own "INDEX
PRAEFATIONUM" in full: nine of Preface.t's 15 EF-cited constructors
name an OF occasion with multiple distinct texts (Nativity x3, Lent
x4-9, Easter x5, Ascension x2, Holy_spirit x2, Bvm x5, Apostles x2,
Common split across two independent 7/8-member families, Requiem x5)
-- structurally lossy to collapse into one constructor. The other six
(Epiphany, Trinity, Sacred_heart, Christ_the_king, St_joseph,
Holy_cross) match cardinality with a single OF text apiece, and were
checked and rejected anyway: Preface.t's own constructors are RG-cited
by construction, and IGMR n. 364's own "quibus Missale Romanum
ditatur" (prefaces the Missal is ENRICHED with) is direct textual
evidence the 2002 collection is not a bare carry-over -- corroborated
directly for Sacred Heart, whose OF text was read and found to be a
different composition from its RG 491 namesake. Reusing any of the
six on cardinality and theme alone would be the exact "sounds similar"
trap the task brief warns against.

test/test_rubrics_of.ml: 26 table-driven cases. Every temporal-origin
row builds observed via Temporal_of.temporal on a real date; every
sanctoral-origin row (a solemnity in Lent, an ordinary Festum, a
weekday obligatory memorial, All Souls) pairs a real Temporal_of.temporal
result with an observed record copied verbatim from
data/of/calendar-2002.sexp, since no full OF Rite.t/Calendar pipeline
exists yet to derive one. Includes same-date Gloria/Credo pairs that
prove the two rules genuinely diverge (a Sunday of Advent, an ordinary
Festum) rather than one silently mirroring the other.

dune test: exit 0, 845 tests. make check (COLITUR_EXHAUSTIVE_SWEEP=1,
full 1583-9999 sweep): exit 0, 851 tests, 303.6s, both run unpiped in
the foreground. git diff --stat v1.0.0..HEAD -- lib/rites/rite_ef/
lib/kernel/ data/ef/ is empty.

1 files changed, 290 insertions, 0 deletions

diff --git a/lib/rites/rite_of/rubrics_of.ml b/lib/rites/rite_of/rubrics_of.ml
new file mode 100644
index 0000000..3fa59e9
--- /dev/null
+++ b/lib/rites/rite_of/rubrics_of.ml
@@ -0,0 +1,290 @@
+(* Institutio Generalis Missalis Romani, editio typica tertia (2002)
+   (docs/research/of/igmr-2002-latin.pdf; extracted with `pdftotext
+   docs/research/of/igmr-2002-latin.pdf`, line numbers below are that
+   extraction's own line numbers) -- CITATION DISCIPLINE, restated from
+   vocab_of.ml/temporal_of.ml: the Missale Romanum PDF (docs/research/of/
+   missale-romanum-2002.pdf) carries BOTH the Normae universales and the
+   IGMR, each independently numbered from 1. Every citation below names
+   "IGMR n. N" for that reason, never a bare "n. N". Gloria and Credo are
+   IGMR questions (Ordo Missae structure); Normae has nothing to say about
+   either.
+
+   GLORIA (IGMR n. 53, igmr.txt:456-460), quoted in full:
+
+     "53. Gloria est antiquissimus et venerabilis hymnus, quo Ecclesia, in
+   Spiritu Sancto congregata, Deum Patrem atque Agnum glorificat eique
+   supplicat. Huius hymni textus cum alio commutari nequit. [...]
+   Cantatur autem vel dicitur diebus dominicis extra tempus Adventus et
+   Quadragesimae, necnon in sollemnitatibus et festis, et in peculiaribus
+   celebrationibus sollemnioribus."
+
+   So: every Sunday OUTSIDE Advent and Lent; every Sollemnitas; every
+   Festum; and, permissively ("in peculiaribus celebrationibus
+   sollemnioribus" -- "in particular more solemn celebrations", "dicitur"
+   not "dicendum est" elsewhere in this paragraph's own register), certain
+   occasions this engine has no dimension to model at all -- the same
+   "votive Mass class" scope limit {!Rite_ef.Rubrics_ef}'s own header
+   states for RG 431(d)/(e)/432(c) -- genuinely N/A, not silently skipped.
+
+   CREDO / SYMBOLUM (IGMR n. 67-68, igmr.txt:568-577), quoted in full:
+
+     "67. Symbolum seu professio fidei eo tendit ut universus populus
+   congregatus verbo Dei in lectionibus e sacra Scriptura nuntiato et per
+   homiliam exposito respondeat [...]
+   68. Symbolum cantandum vel dicendum est a sacerdote cum populo diebus
+   dominicis et in sollemnitatibus; dici potest etiam in peculiaribus
+   celebrationibus sollemnioribus."
+
+   NARROWER than n. 53, on purpose, not by oversight: n. 68 names only
+   "dominicis et... sollemnitatibus" -- NO "et festis" clause at all, unlike
+   n. 53's explicit third member. An ordinary [Festum] (the Presentation,
+   the Transfiguration, the Exaltation of the Holy Cross...) therefore does
+   NOT get the Credo, exactly matching real practice (only Sundays and
+   solemnities carry it) and confirmed by the literal absence of the word
+   "festis" from n. 68 where n. 53 has it -- checked directly, not assumed.
+   The trailing "dici potest... peculiaribus... sollemnioribus" is the same
+   permissive, unmodelled N/A clause as n. 53's.
+
+   Neither n. 53 nor n. 68 carries an Easter-relative WINDOW the way RG
+   475(d)'s three EF octaves do (Nativity/Easter/Pentecost) -- checked by
+   reading both paragraphs in full above, not merely by their absence from
+   this file's own branches: nothing in IGMR nn. 52-69 (the whole "Actus
+   paenitentialis" through "Oratio universalis" span, igmr.txt:440-585)
+   mentions an octave at all. [credo]'s own [date] parameter is therefore
+   genuinely unread, kept only for {!Colitur_kernel.Rite.t}'s shared shape
+   -- an honest [None]-shaped fact about this rubric, not a gap.
+
+   THE REQUIEM QUESTION -- IGMR IS GENUINELY SILENT, FLAGGED AS SUCH. RG
+   476(f)/432(d) name "in Missis defunctorum" explicitly for EF; IGMR nn.
+   379-385 ("II. DE MISSIS DEFUNCTORUM", igmr.txt:2327-2349), read in full,
+   say nothing at all about the Gloria or the Symbolum -- confirmed by
+   reading every line of that section, not by a failed grep. The Missale's
+   own printed propers corroborate the OMISSION reading only indirectly:
+   every OTHER Mass text in the extracted Proper of Time/Sanctoral carries
+   an explicit "Dicitur Gloria in excelsis."/"Non dicitur..." marker (over
+   150 occurrences, docs/research/of/missale-romanum-2002.pdf via
+   `pdftotext -layout`), but the two "IN COMMEMORATIONE OMNIUM FIDELIUM
+   DEFUNCTORUM" (2 November) Mass-text blocks carry NEITHER marker --
+   consistent with (not proof of) universal, uncontested funeral-liturgy
+   practice that a Requiem Mass never has either. Picked as the reading
+   with the better textual support, per this task's own brief, and applied
+   as narrowly as the evidence allows: {!requiem_slugs} is a CLOSED,
+   directly-checked list of the ONE slug in data/of/calendar-2002.sexp this
+   can ever apply to (grepped for "defunct"/"requiem"/"exsequial": zero
+   other hits), the same "checked, not assumed, closed population"
+   discipline {!Rite_ef.Rubrics_ef}'s own [Colour.Black] proxy uses for the
+   identical EF question. *)
+
+open Colitur_kernel
+
+let requiem_slugs = [ "all-souls" ]
+
+let is_requiem (observed : Vocab_of.rank Celebration.t) =
+  List.mem (Slug.to_string observed.Celebration.slug) requiem_slugs
+
+let gloria ~(temporal : (Vocab_of.season, Vocab_of.rank) Temporal.t)
+    ~(observed : Vocab_of.rank Celebration.t) ~(date : Date.t) : bool =
+  let easter = Computus.gregorian_easter (Date.year date) in
+  let n = Date.to_rata date - Date.to_rata easter in
+  if is_requiem observed then false
+  else if
+    (* Missale Romanum, "FERIA V IN CENA DOMINI", Ad Missam vespertinam, n.
+       7 (docs/research/of/missale-romanum-2002.pdf, pdftotext -layout,
+       line ~8534): "Dicitur Gloria in excelsis. Dum cantatur hymnus,
+       pulsantur campanae, eoque expleto, silent usque ad Gloria in
+       excelsis Vigiliae paschalis..." -- the Missal's own PROPER-OF-TIME
+       rubric for the Mass of the Lord's Supper (Easter-3) explicitly
+       requires the Gloria despite the day sitting inside Lent (Tabula I.1
+       territory once the evening Mass begins; Precedence_of.band's own
+       comment on WHY the civil day itself stays a Holy Week feria, not the
+       Triduum, is unaffected -- this rubric is keyed to the Mass, not the
+       day's own band), and the same sentence extends the silence "usque
+       ad" (UP TO) the Easter Vigil's own Gloria (Easter-1), confirming
+       that Mass also has one. Neither day is ever a Sunday (both are fixed
+       Easter-offsets within Holy Week), so this must be checked ahead of,
+       and independently of, the season/weekday branches below -- the
+       identical lex-specialis placement {!Rite_ef.Rubrics_ef.gloria}'s own
+       RG 431(c) branch uses for the same two days. *)
+    n = -3 || n = -1
+  then true
+  else if
+    (* IGMR n. 53: "extra tempus Adventus et Quadragesimae" -- read off
+       [temporal], never [observed]: Temporal_of.temporal tags every Sunday
+       of Advent, Lent AND Easter [Sollemnitas] (temporal_of.ml, [build]'s
+       own [match s with Advent | Lent | Easter -> Sollemnitas | ...]), so
+       an [observed.rank] test alone would wrongly grant an Advent or Lent
+       Sunday the Gloria this clause explicitly withholds -- confirmed by
+       Palm Sunday specifically ([Vocab_of.Lent], [Sollemnitas],
+       [named]'s own entry): real, uncontested practice omits the Gloria
+       there, and only a temporal-read guard gets that right. *)
+    temporal.Temporal.weekday = Date.Sun
+    && (temporal.Temporal.season = Vocab_of.Advent || temporal.Temporal.season = Vocab_of.Lent)
+  then false
+  else if
+    (* IGMR n. 53: every remaining Sunday (Christmas time, Easter, Ordinary
+       Time -- outside Advent/Lent by the branch just above). Read off
+       [temporal.weekday], not [observed.rank]: an ordinary Sunday per
+       annum is [Festum]-ranked (temporal_of.ml, the same [build] match
+       above), not [Sollemnitas], and would otherwise only reach a [true]
+       answer via the [Festum] arm below -- redundant for it, but this
+       branch is what makes an Easter- or Christmas-season Sunday's [true]
+       answer trace to n. 53's OWN "diebus dominicis" clause rather than
+       accidentally to its "et festis" clause. *)
+    temporal.Temporal.weekday = Date.Sun
+  then true
+  else
+    (* IGMR n. 53: "in sollemnitatibus et festis" -- every remaining
+       Sollemnitas or Festum, whatever day of the week it falls on (a
+       solemnity transferred onto an ordinary weekday, or one that was
+       never a Sunday to begin with -- St Joseph 19 March, the Annunciation
+       25 March, both always inside Lent by construction: Gregorian
+       Easter's earliest possible date is 22 March, so Ash Wednesday can
+       never fall later than 10 March and Easter itself can never fall
+       before 19 March, meaning neither date can ever fall outside Lent).
+       [Memoria_obligatoria], [Memoria_ad_libitum] and [Feria] are not
+       named by n. 53 at all and correctly fall through to [false]. *)
+    observed.Celebration.rank = Vocab_of.Sollemnitas || observed.Celebration.rank = Vocab_of.Festum
+
+let credo ~(temporal : (Vocab_of.season, Vocab_of.rank) Temporal.t)
+    ~(observed : Vocab_of.rank Celebration.t) ~(date : Date.t) : bool =
+  ignore date;
+  (* IGMR n. 67-68's own [date]-independence: see this file's own header. *)
+  if is_requiem observed then false
+  else if
+    (* IGMR n. 68: "diebus dominicis" -- EVERY Sunday, Advent and Lent
+       included (unlike [gloria]'s own n. 53 carve-out: n. 68 has no such
+       exception in its text, checked directly). Read off [temporal], not
+       [observed], for the same reason [gloria]'s own Sunday branch is:
+       an ordinary (non-privileged-season) Sunday per annum is [Festum]-
+       ranked, and n. 68's own [Sollemnitas] branch below would silently
+       miss it if this branch were dropped. *)
+    temporal.Temporal.weekday = Date.Sun
+  then true
+  else
+    (* IGMR n. 68: "in sollemnitatibus". NO "et festis" here -- the one
+       place this function's shape most visibly narrows on EF's RG
+       475(b)/(c): a bare [Festum] (the Presentation, the Transfiguration,
+       a I-class-shaped feast of the Lord with no Sunday privilege) does
+       NOT get the Credo. [Memoria_obligatoria]/[Memoria_ad_libitum]/
+       [Feria] fall through to [false], matching real practice (an
+       obligatory memorial on a weekday never carries the Creed). *)
+    observed.Celebration.rank = Vocab_of.Sollemnitas
+
+(* PREFACE CHOICE (IGMR n. 364-365, igmr.txt:2244-2263) -- quoted in full:
+
+     "364. Plurimae praefationes, quibus Missale Romanum ditatur, eo
+   spectant ut argumenta gratiarum actionis in Prece eucharistica plenius
+   eniteant, et variae rationes mysterii salutis pleniore luce
+   proponantur.
+     365. Electio inter Preces eucharisticas, quae in Ordine Missae
+   inveniuntur, his normis opportune regitur: [...]"
+
+   n. 364-365 govern choosing among the FOUR EUCHARISTIC PRAYERS (I-IV),
+   NOT which preface accompanies whichever Mass is said -- a different
+   question this engine has no need to answer (colitur resolves a Mass's
+   readings/rubrics, never which of the interchangeable Canons a celebrant
+   picks). Genuinely N/A, cited here only because n. 364-365 is the
+   nearest thing to a "how is the Eucharistic Prayer/preface chosen"
+   numbered rubric in the whole IGMR, and this file should say plainly
+   that it is not the rule this function needs, rather than leave a reader
+   to wonder why it was not used.
+
+   UNLIKE RG 482-499, the IGMR carries NO numbered rubric governing preface
+   ASSIGNMENT at all -- confirmed by reading nn. 1-386 in full, not merely
+   by a failed grep for "praefatio" (which itself found zero hits: the
+   word never appears with that spelling anywhere in
+   docs/research/of/igmr-2002-latin.pdf's extracted text; every occurrence
+   in the Missale PDF instead uses the ligature "praefationem"/"praefatio"
+   spelt with <ae> or the printed <æ>, confirming this is a genuine textual
+   absence, not an extraction artifact). Preface assignment in the 2002
+   editio typica tertia is done PER MASS, inline throughout the Proper of
+   Time, Proper of Saints and Commons (each Mass's own rubric names its
+   preface directly, e.g. "Praefatio de Nativitate, pp. 520-522.") -- the
+   collection itself is indexed at the Missal's own back matter under
+   "INDEX PRAEFATIONUM" (missale-romanum-2002.pdf, pdftotext -layout,
+   extracted lines ~34107-34247), transcribed here in full because it is
+   the evidence the verdict below rests on:
+
+     A. De Tempore: Advent I-II (2); Nativity I-III (3); Epiphany (1);
+        Lent ("de Quadragesima") I-IV plus five Sunday-specific variants
+        V-IX (9); Passion ("de Passione Domini") I-III, III being Palm
+        Sunday's own (3); Paschal (Easter) I-V (5); Ascension I-II, BOTH
+        titled "De mysterio Ascensionis" (2); Pentecost (1); Sundays "per
+        annum" I-VIII (8); Praefationes communes I-VII (7).
+     B. In festis et mysteriis Domini: Trinity (1); the Annunciation's own
+        "De mysterio Incarnationis" (1); the Baptism of the Lord (1);
+        Sacred Heart (1); "De victoria Crucis gloriosae", 14 September (1);
+        Corpus Christi ("de Ss.ma Eucharistia") I-II (2); the Presentation
+        (1); Christ the King (1); the Transfiguration (1); the Dedication
+        of a Church I-II (2); "de Spiritu Sancto" I-II (2).
+     C. In festis sanctorum: the BVM I-V (5); the Angels (1); the
+        Precursor/John the Baptist (1); St Joseph, "De missione S.
+        Ioseph", 19 March/1 May (1); Peter-and-Paul (1); Apostles I-II
+        (2); Saints I-III (3); Martyrs I-II (2); Pastors (1); Virgins and
+        Religious (1).
+     Pro defunctis (Requiem): I-V (5).
+
+   AGAINST THIS CENSUS, {!Preface.t}'s 15 constructors split into two
+   failure shapes, and every single one fails one of them -- NONE is safe
+   to reuse:
+
+   (1) NINE constructors name an OF occasion with MULTIPLE distinct texts:
+       [Nativity] (3 OF variants), [Lent] (4, or 9 counting the
+       Sunday-specific set), [Easter] (5), [Ascension] (2), [Holy_spirit]
+       (2), [Bvm] (5), [Apostles] (2), [Common] (15, split across two
+       independent numbered families -- "per annum" I-VIII and
+       "communes" I-VII, neither of which [Preface.t]'s single [Common]
+       constructor could ever pick between), [Requiem] (5). Returning any
+       one of them would silently pick ONE of several genuinely different
+       texts and call it "the" answer -- structurally lossy, independent
+       of any wording question.
+
+   (2) SIX constructors name an OF occasion with exactly ONE text, the same
+       cardinality EF has for it -- [Epiphany], [Trinity], [Sacred_heart],
+       [Christ_the_king], [St_joseph], and [Holy_cross] (EF's RG 487 "de
+       sancta Cruce" against OF's single "De victoria Crucis gloriosae",
+       both keyed to 14 September). These are the genuinely tempting
+       cases -- same feast, same "exactly one" cardinality on both sides --
+       and each was checked and rejected anyway, on TWO grounds:
+         (a) {!Preface.t}'s own header ties every constructor to an RG
+             paragraph ("Nativity RG 484", "Trinity RG 494", ...) -- it
+             denotes THAT SPECIFIC EF TEXT, not an abstract cross-rite
+             theme, by its own documented contract.
+         (b) IGMR n. 364's own opening word choice is direct evidence the
+             2002 Missal does not consider its preface collection a bare
+             carry-over: "praefationes, QUIBUS Missale Romanum DITATUR"
+             ("the prefaces WITH WHICH the Missal is ENRICHED") -- language
+             describing an addition/expansion, not an inheritance,
+             corroborating the well-documented fact that the post-1970
+             preface collection was substantially newly composed, the
+             Sacred Heart preface actually read while building this
+             module ("Qui, mira caritate, exaltatus in cruce... de
+             transfixo latere sanguinem fudit et aquam...",
+             missale-romanum-2002.pdf line ~30575) being the one
+             instance directly compared line-by-line against what its RG
+             491 EF namesake is known to open with, and found to be a
+             different composition, not merely a different printing of
+             the same one.
+       Reusing any of the six on the strength of matching cardinality and
+       matching feast alone would be exactly the "sounds similar" trap
+       this task's own brief warns against -- confidently wrong output
+       every test in this suite would agree with, because nothing in this
+       codebase yet stores preface TEXT to catch the mismatch against.
+
+   VERDICT: [preface] returns [None] unconditionally. This is not a stub
+   -- every branch above was checked and its failure mode recorded -- and
+   it is not a defect in {!Preface.t} either: that type was built for one
+   rite and correctly refuses to pretend it fits a second one it was never
+   shown. The finding worth carrying forward (task brief's own "spec §7"
+   framing): OF preface selection needs its own type, most plausibly one
+   with a per-occasion NUMBERED-VARIANT dimension ("Nativity of 1 | 2 | 3",
+   ...) rather than {!Preface.t}'s flat enumeration, plus its own IGMR/
+   Missale-cited population census -- table-driven, the same shape as
+   {!requiem_slugs} above, once a task is actually chartered to build it.
+   Building that type is explicitly out of this task's own scope. *)
+let preface ~(temporal : (Vocab_of.season, Vocab_of.rank) Temporal.t)
+    ~(observed : Vocab_of.rank Celebration.t) ~(date : Date.t) : Preface.t option =
+  ignore temporal;
+  ignore observed;
+  ignore date;
+  None