derek73 · Nov 13, 2026 · Nov 13, 2026 · Nov 13, 2026 · Nov 13, 2026 · Nov 13, 2026
diff --git a/AGENTS.md b/AGENTS.md
@@ -290,7 +290,7 @@ The 2.0 rewrite landed as underscore-private modules beside the v1 code, which 2
 - **The segmenter contract**: the optional `Parser(segmenter=...)` hook is parse-totality's ONE exception (locales spec section 4). Everything inside that exception is a bug in USER CODE, never a fact about the name, so it is surfaced rather than absorbed: the segmenter's own exceptions propagate, and the two protocol violations the stage can detect for itself — an answer of the wrong type, and one cutting at or past the end of the token it was handed — raise `TypeError`/`ValueError` from `_script_segment` for the same reason. The line to hold when adding a check there: a protocol violation by the segmenter's AUTHOR raises, while an adapter's defense against its own third-party library (`locales/ja.py`'s repertoire, length, kanji, reconstruction and score guards) declines with `None`, because what those catch is a fact about the content.
 - **Pickling**: v2 types must round-trip (`Parser` is picklable by construction, and it holds a `Lexicon`; the one qualifier is that a `Parser` pickles iff its segmenter does — see the segmenter bullet above). Every frozen type assigns `_guarded_getstate`/`_guarded_setstate` (`_types.py`) in its class body (`@dataclass(slots=True)` would override inherited pickle methods) — unpickling fails at the LOAD site on field-layout skew, with one widening: a field added after its type shipped carries `metadata={_types.ADDED_IN: "<release>"}` and a default, and an older pickle missing it loads with the default (`Ambiguity.alternative`, 2.7; `Token.reading`, `Reading.languages`, `Reading.second_surname`, `Reading.first_surname`, `ParsedName.order`, `Policy.two_given_names`, `PolicyPatch.two_given_names`, `Policy.lone_name_word`, `PolicyPatch.lone_name_word` and `Locale.languages`, 3.0.0; `Policy.initials_as_family`, `PolicyPatch.initials_as_family`, `Policy.capitals_as_family` and `PolicyPatch.capitals_as_family`, 3.1.0) -- a missing field without the marker, default or not, an unknown field and a renamed one are still refused, each a row in `tests/v2/test_types.py::_REFUSED_SKEWS`, so a new field on a pickled type takes the marker or breaks every older pickle; the shared guard deliberately does NOT re-validate values (pickle is not a security boundary; canonical state comes from a validated instance). `Lexicon` and `LexiconReading` keep their own copy of the guard (layering); a new unpicklable slot type needs the same treatment plus a round-trip test. Since 3.0's 4i-b a `Lexicon` pickles its lists, its masks, its edit log, its choice of languages (`_languages` and `_strict`, 4i-c, marked `ADDED_IN` so a 4i-b pickle loads choosing nothing) and its base as a SNAPSHOT of the reading index's rows, so a pickle parses as it did where it was written, a snapshot equal to the running release's index loading as the by-reference default again (decisions/vocabulary-model, 2026-11-02 #754, item 5); its private `_sets`, `_cap_map`, `_same_as`, `_same_as_named`, `_familiar`, `_chosen` and `_files` fields are built at construction and never pickled. `Lexicon`'s copy of the guard admits the `ADDED_IN` marker, read under `_lexicon._ADDED_IN`, the same string, since layering keeps `_lexicon` from importing `_types` (`test_lexicon_marks_an_added_field_with_the_shared_key` holds the two equal), so a new `Lexicon` field takes the marker and a default or breaks older pickles. A 2.x state holds the compiled sets instead, and is refused since 3.1 with a `ValueError` (`_lexicon._refused_2x`) that tells the layout 3.0 loaded, 2.7's and 2.8's, to re-pickle under 3.0, and any older one to rebuild — that layout a frozen record (`_LAYOUT_3_0_LOADED`), never `_VOCAB_FIELDS`, which a new field joins and every 2.8 pickle would then lack, the translation 3.0 loaded it by being deleted (decisions/vocabulary-model, 2026-11-08 #1003); the 2.8 `Lexicon`, `Parser` and locale-pack pickles in `tests/v2/data/lexicon_v28.pickle` are the refusal test's input. And unlike the shared guard it DOES re-validate: a load goes through the constructor, which checks the base, refuses an edit not in canonical form, and refuses a list or mask entry that is not already normalized, because a foreign blob would otherwise load a different lexicon in silence (a snapshot base of a kind this release does not know, an edit `add` would not have written, a list entry the fold changes).
 - **One sanctioned global**: the cached default `Parser` (`_parser._default_parser`, which `parse()` and `ParsedName.matches` use). Lazily cached FROZEN singletons (`Lexicon.default()`'s `functools.cache`, the default parser's) are constants, not state; any second piece of module-level MUTABLE state requires amending the conventions doc, on purpose, in review.
-- **Tests**: all v2 tests live in the `tests/v2/` package, one test module per source module plus the cross-cutting ones (`test_reprs.py`, `test_layering.py`, `test_contracts.py`, `test_properties.py`, `test_benchmark.py`, the `cases.py`/`test_cases.py` table, `test_regex_sync.py`, which pins render's hand-copied patterns to the pipeline's, and `test_snapshot.py`, which holds every corpus name's reading to `tools/differential/snapshot.jsonl` — sharing `_differential_fixtures.py` with `test_differential.py` and `test_corpora.py` for the handles on `tools/differential/`), names stating behavior; the 1.x suite's parsing tests stand ported to `parse()` in `tests/*.py`. Never assert `Lexicon.default()` contents; the narrow sourcing spot-checks in `test_default_sources_v1_vocabulary` that pin the v1→v2 migration contract (e.g. the flipped `particles_ambiguous` model) are the sanctioned exception.
+- **Tests**: all v2 tests live in the `tests/v2/` package, one test module per source module plus the cross-cutting ones (`test_reprs.py`, `test_layering.py`, `test_contracts.py`, `test_properties.py`, `test_benchmark.py`, the `cases.py`/`test_cases.py` table, `test_regex_sync.py`, which pins render's hand-copied patterns to the pipeline's, `test_script_properties.py`, which checks each row of the script-property table through the readers of each column that take a word, and `test_snapshot.py`, which holds every corpus name's reading to `tools/differential/snapshot.jsonl` — sharing `_differential_fixtures.py` with `test_differential.py` and `test_corpora.py` for the handles on `tools/differential/`), names stating behavior; the 1.x suite's parsing tests stand ported to `parse()` in `tests/*.py`. Never assert `Lexicon.default()` contents; the narrow sourcing spot-checks in `test_default_sources_v1_vocabulary` that pin the v1→v2 migration contract (e.g. the flipped `particles_ambiguous` model) are the sanctioned exception.
 
 ## Extension Patterns
 
@@ -328,7 +328,7 @@ The 2.0 rewrite landed as underscore-private modules beside the v1 code, which 2
 
 **Cyrillic suffix regexes need `re.I` even when the pattern is suffix-only** — a Latin title-cased word (`Ivanovich`) keeps its suffix lowercase, so `re.I` seemed skippable; but an irregular Cyrillic suffix can be nearly the whole word (`ильич`), so title-casing capitalizes into the suffix itself (`Ильич`). `east_slavic_patronymic_cyrillic` (v1's name for what is `_vocab._EAST_SLAVIC_CYR` since 3.0 removed `config/regexes.py`, #1008) shipped without `re.I` on the Latin reasoning and silently failed on capitalized irregular forms — don't assume Latin's title-case safety transfers to Cyrillic. (#185)
 
-**Suffix words vs the initial veto** — single-letter roman numeral suffixes (`i`, `v`) are suffix words but also initial-shaped, so `_vocab.is_suffix_strict` (v1's `is_suffix`) vetoes them through `_vocab.is_initial`. `_vocab.is_suffix_lenient` accepts suffix words without the veto and is only safe in unambiguous positions, after a comma (#136, #144). **The veto is Latin-only since #320**: `is_initial` ANDs the bare shape test (`is_initial_shaped`) with "not in `_policy._NO_INITIALS`", the four scripts that have no initials (Han, Hangul, Hiragana, Katakana), so the single-character CJK suffix words (씨, 様, 氏, 군, 양, 님, 殿) written with a period read as suffixes: `"김민준 씨."` gives suffix `씨.`. `V.`/`I.` still lose to the veto and still need the lenient test.
+**Suffix words vs the initial veto** — single-letter roman numeral suffixes (`i`, `v`) are suffix words but also initial-shaped, so `_vocab.is_suffix_strict` (v1's `is_suffix`) vetoes them through `_vocab.is_initial`. `_vocab.is_suffix_lenient` accepts suffix words without the veto and is only safe in unambiguous positions, after a comma (#136, #144). **The veto is Latin-only since #320**: `is_initial` ANDs the bare shape test (`is_initial_shaped`) with "not in a script that writes no initials", the rows of `_policy._SCRIPT_PROPERTIES` whose `writes_initials` is false (Han, Hangul, Hiragana, Katakana, and since #1256 the Myanmar and Thai scripts; decisions/script-properties), so the single-character CJK suffix words (씨, 様, 氏, 군, 양, 님, 殿) written with a period read as suffixes: `"김민준 씨."` gives suffix `씨.`. `V.`/`I.` still lose to the veto and still need the lenient test.
 
 **Comparing against v1 needs `PYTHONSAFEPATH=1` AND a directory outside the worktree** — `uv run --isolated --no-project --with 'nameparser==1.4.0'` still puts the checkout's `nameparser/` ahead of the pinned wheel on `sys.path`, so the "v1" side silently imports the branch and every comparison reports parity. Run it as `cd <scratch dir> && PYTHONSAFEPATH=1 uv run --isolated --no-project --with 'nameparser==1.4.0' python -c "..."`. This produces false confidence rather than an error, so it invalidates results without ever looking wrong. **A `tests/v2/cases.py` row's `parity` means 1.4.0, and nothing checks it** -- the table runs every row against the tree, so a row labelled `parity` that only matches master passes for good. #614 added eight such rows, compared against master; #620 found them by running the wheel this way and relabelled each with the change its reading came from, found by `git bisect run` in a scratch worktree (`git worktree add --detach`, never this one). Run the wheel before writing `parity`.
 

diff --git a/docs/design/decisions/H2-leading-abbreviation-title.md b/docs/design/decisions/H2-leading-abbreviation-title.md
@@ -5,7 +5,7 @@
 - An abbreviation of two or more letters ending in its one period, opening the part that carries the given name (the whole name, or the part after a family comma), is a title even unlisted and even where it is suffix vocabulary, and a single-letter initial is not: `Insp. Jane Morse` → title 'Insp.', `Morse, Det. Insp. Jane` → title 'Det. Insp.', `Esq. Smith` → title 'Esq.', `J. Smith` → given 'J.' — 2026-06-30 #288.
 - The shape is an unbroken run of letters, so an unlisted abugida abbreviation written with a vowel sign stays a name word and only vocabulary recognizes it — 2026-06-30 #288 (the run was the whole cause until 2026-11-08 #1191, below, found the word with no sign).
 - The shape declines a period-marked word carrying a letter of the ten Brahmic scripts from Devanagari to Sinhala (U+0900 to U+0DFF), an Except of rules/H2, so a Latin initial spelled there with no vowel sign is no title: `एम. एस. वर्मा` → given 'एम.', middle 'एस.' — 2026-11-08 #1191.
-- The shape declines a period-marked word carrying a script with no initials: `田中. 太郎` → family '田中.', where `Smith. John` keeps title 'Smith.' — 2026-09-10.
+- The shape declines a period-marked word carrying a script that writes no period abbreviations (Han, kana, hangul; Thai and Burmese write no initials but are not among them, decisions/script-properties): `田中. 太郎` → family '田中.', where `Smith. John` keeps title 'Smith.' — 2026-09-10.
 - C13 wins for a credential opening the part after a family comma (`Smith, Esq. John` → suffix 'Esq. John'), and the shape still reads a title there for a word that starts no S8 run (`Smith, Insp. John`, `Smith, Do. John` → title) — 2026-10-20 #761; its `Smith, Ma. John` example moved to the given name on 2026-10-26 #789.
 - A given name written short outranks the shape: a word the vocabulary lists with a `given` reading whose `same_as` is the full name, period-marked where the shape would fire, is the given name and the name reads as it would with the full name there (`Chas. Smith` → given 'Chas.', `Mrs Ma. Liza Cruz` → title 'Mrs', given 'Ma.'). The shipped set is the traditional English abbreviations, the Filipino `ma` and `jo`, and since #1115 Muhammad, Achmad and Francisco written short (`mohd`, `mhd`, `muh`, `moh`, `moch`, `ach`, `fco`; decisions/H6's 2026-11-06 entry); a listed title still wins, and the reading acts nowhere else — 2026-10-26 #789.
 - A transliteration digraph declines the shape, an Except of rules/H2: the Latin spellings of one Cyrillic letter that are no title in any corpus read, `Yu`, `Ya`, `Ye`, `Yo`, `Zh`, `Kh`, `Shch` and `Dzh`, in any case, so `Yu. A. Chibrikova` → given 'Yu.', middle 'A.', as `Ю. А. Чибрикова` reads; `Ts`, `Ch` and `Sh`, each also a title written short, keep the shape, and a digraph of another alphabet or a Latin given name written short (`Th.`) is outside the set — 2026-11-09 #1136.