Blind per-date source-file reading (Sancti/SanctiM MM-DD.txt) turned out
unreliable for collect authorship: those files are era-merged, carrying
centuries of accumulated saints behind rubric-conditional gating, so a
naive read often returned a saint canonized long after 1617 instead of
the Monastic-Tridentine-era one actually on this calendar (St. Didacus
instead of All Saints of the Benedictine Order, St. Rose of Lima instead
of Ss. Felix and Adauctus, etc).
Fixed by pulling from web/www/Tabulae/Kalendaria/'s own era-specific
calendar tables (1570.txt Tridentine base + M1617.txt's Monastic diffs)
instead, cross-validated against the pre-1955/Divino Afflatu chain
(1888->1906->1939->1954) as a corroborating signal. Cross-referencing all
205 existing saint records against this table confirmed 204/205 were
already correctly placed -- the earlier live-query verification work was
sound, only blind file-reading was flawed -- so no sanctoral-calendar.yml
corrections were needed, just the collect content itself.
Two collect shapes came out of the 94: 34 proper collects (own YAML file,
Latin verified against source, English translated+draft where the source
lacked it) and 60 resolved via a new Common-collect templating mechanism
-- SaintRecord.collectCommon/collectName plus substituteName() in
resolve-common.ts, filling a template's {N} placeholders positionally
from the saint's own declined name, across 21 new common/collect-c*.yml
templates named after the source's own C-numbering.
St. Michael's Sep 29 propers resolved via its whole-file @Sancti/05-08
inheritance (Apparition of St. Michael, May 8). St. Martha's plural-
template default was overridden to the singular collect-c6a since she's
one person, not a pair.
Also fixes a latent bug this surfaced: getDayCollect looked up
`${collectCommon}-collect`, but collect-c*.yml files are id'd as
`collect-c2` etc with no such suffix -- caught by a smoke test where
every one of the 93 non-Michael saints resolved as missing.
4 pre-existing tests updated: they asserted specific saints (St.
Benedict, St. Romanus Abbot, St. Anthony Abbot) had no authored collect,
which is no longer true; resolve-common.test.ts's generic "missing" case
now uses a synthetic day/winner since no real unauthored saint remains
as an example.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Every antiphon that goes through the incipit/full split -- which is
every antiphon across every hour -- lost its real status the moment it
passed through here: splitNamedAntiphon hardcoded `verified` on its
output via verifiedText(), regardless of what the source's actual
status was. Concretely, this meant St. Scholastica's antiphon (I
translated it myself, no English existed in the source), Matthias's and
several other apostles' reworded English, and Gregory the Great's
translated antiphon all silently rendered as if fully verified --
losing the "unverified draft text" UI marker (src/ui/styles.css's
.text-draft, a dashed underline + tooltip) that's supposed to
distinguish "copied straight from the source" from "Claude's own
rendering."
Fixed by having splitNamedAntiphon accept a real ResolvedText and carry
its per-language status through unchanged to both outputs, instead of
silently upgrading everything to verified. Callers whose antiphon never
had a status field to begin with (plain weekday-default antiphons and
Lauds psalmody overrides -- neither type tracks status at all, always
implicitly verified by the source file's own live-checked convention)
now wrap with verifiedText() explicitly at the call site, rather than
that assumption being buried inside splitNamedAntiphon itself.
resolveMinorHourAntiphon (Prime/Terce/Sext/None's override lookup) now
returns a full ResolvedText for the same reason.
Verified end-to-end with St. Scholastica's real draft-English antiphon,
both via a direct unit test and visually in the browser (the dashed
underline now shows correctly under her Benedictus antiphon).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
11 collects, 4 antiphons -- including backfilling the three pre-existing
August entries (Lawrence, Assumption, Augustine) now that the pulling
methodology exists. Also updates three pre-existing tests whose dates
collided with this month's new real content: the vigil-of-St.-Lawrence
backward-transfer case now runs into a real feast at its landing day
instead of an empty one (confirmed correct against the live reference
engine, including the surviving commemoration on the impeded Sunday
itself), and two other tests moved off Aug 9/10 to dates unaffected by
the new data.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Terce/Sext/None's Oratio is the day's own Proper-of-season/-saints
collect, not a fixed text — getDayCollect() resolves it from the
LiturgicalDay's winner (temporal id or sanctoral saint's propers),
falling through to an honest "missing" status for saints that don't
have authored propers yet.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
will