Keep the titles 2.6.0 listed in v1's Constants().titles (#752)

docs/design/decisions/3-0-reevaluations.md

@@ -63,3 +63,4 @@ Open: #753 (phase 3, language packs, where the three pack bullets land).
 - (A) 2026-09-20 #397 — A SECOND INSTANCE OF THE SAME PER-LOCALE QUESTION, cross-referenced rather than parked again. Catalan/Polish `i` joins `conjunctions_ambiguous` beside `e`, so the marked subset now carries two letters and the 2026-08-18 bullet above — which counted six single-letter conjunctions and named the per-word-policy requirement — describes a set of seven, `i` included. Nothing in that bullet's argument moves: the subset IS the per-word policy the bullet asked for, made configurable without packs, and the pack-side half is still what the 2026-09-13 SHOULD A LOCALE PACK BE ABLE TO REMOVE BASE VOCABULARY? (A) bullet above asks (this pointer named only its date until 2026-10-22, which two bullets share), since a `ca` or `pl` caller still has to write `Lexicon.default().remove(conjunctions_ambiguous={"i"})` by hand. See `decisions/P3`'s 2026-09-20 bullet.
 - (B) 2026-09-23 #459 — THE SHIM RAISES on a v1 config it once accepted: a `capitalization_exceptions` value that does not spell its key's letters and digits raises `ValueError` at the first parse (decisions/R4, 2026-09-23). Cited against the shim's never-raise rule and taken anyway, so it is recorded here: the raise stands on its own — a case mask has no reading for a value that re-spells its word, and the core `Lexicon` raises the same way with no shim involved — and only where it LANDS for a v1 caller, at the first parse rather than at construction, is shim-era. 3.0 has nothing left to decide.
 - (B) 2026-10-02 #541 — THE SHIM DROPS, with a warning, every v1 `Constants` entry v1 could never match (decisions/R4, 2026-10-02). Cited from 1.4 parity, and it dies with the shim: the core `Lexicon` normalizes edge whitespace away and has no v1 reading to reproduce, so 3.0 has nothing left to decide.
+- (B) 2026-10-26 #752 — THE SHIM KEEPS 2.6.0'S TITLE LIST as v1's default `titles`, frozen in `nameparser/config/_titles_2_6_0.py`, so a title a later release made phrase-only (decisions/H9, 2026-10-26, the v1 title list) is still listed and removable. Cited from v1 parity, and it dies with the shim: the core `Lexicon` keeps the two sets apart and has no v1 list to reproduce, so 3.0 deletes the file with the facade and decides nothing.

docs/design/decisions/H9-title-phrase.md

@@ -7,7 +7,7 @@
 - `grace` (a style) and `marshal` (a rank) are titles that may be given names, `sergeant` attaches to the word `major` and `major` to `general` (`attaches` naming a word where a kind would reach too far): `His Grace John Smith`, `Field Marshal Bernard Montgomery` and `Sergeant Major John Smith` read the title; `Mr Justice Smith` keeps given 'Justice', no exception behind a salutation — 2026-10-26 #752, phase 2i (Derek).
 - Its partner is a modifier or a style before it, a title before it that attaches to it or a title after it that it attaches to (`attaches`), or `of` after it; a salutation is never one, nor any other title before it (#853's `Mr Dean Smith`), nor a title after it it does not attach to (`SMITH, DEAN MR.`) — 2026-10-25 #752.
 - A modifier in the leading run of titles before the first comma is a title only with a head title after it in that run, `of` after it, or in the object of `of` or `the` there; outside that run (ending the name, H5's, or after a comma) it reads as any title does, the given part after a family comma included — 2026-10-25 #752; save a modifier that may be a given name, which the run's test reads in every part (`Lau, Wing Kai` → given 'Wing', `John Smith Wing.` → family 'Wing.') — 2026-10-26 #752, phase 2j, `wing`.
-- A title that may be a given name which a caller lists in `titles` is a title wherever it stands, which keeps the 1.4 recipe `c.titles.add("dean")`; a v1 caller who empties `titles` turns the phrase reading off with them, while one who replaces the list keeps it (`c.titles.clear(); c.titles.add('dr')` still reads `Dean of Students` as the title), v1 holding no knob for the set — 2026-10-25 #752.
+- A title that may be a given name which a caller lists in `titles` is a title wherever it stands, which keeps the 1.4 recipe `c.titles.add("dean")`; a v1 caller who empties `titles` turns the phrase reading off with them, while one who replaces the list keeps it (`c.titles.clear(); c.titles.add('dr')` still reads `Dean of Students` as the title), v1 holding no knob for the set — 2026-10-25 #752. A title 2.6.0 listed that a later batch made a title only inside a phrase stays in v1's default `titles`, read by the phrase while it is there and as no title once removed, so `c.titles.remove('king')` works as in 2.6.0; listing it in v1 cannot make it a title everywhere, which takes `Lexicon.default().add(titles={'king'})` — 2026-10-26 #752, the v1 title list.
 
 ## Log
 
@@ -59,7 +59,9 @@
 - **2026-10-26 #752, phase 2j, `sultana` and `sultan` (Derek) — THE FIFTH AND SIXTH TITLE BATCHES, COMBINED.** The batches decisions/vocabulary-model's 2026-10-24 #752 (Derek) entry put fifth and sixth; measured apart and together, `sultana` costing nothing and `sultan` only the royal class accepted for `king` (the entry above), Derek combined them on 2026-10-26. Each gains a dated `given` reading (`sultan`'s on its tr.toml entry, the other two files' readings deriving the same marker) and is a title only inside a title phrase; no listed phrase holds either, so nothing is reported. `Sultana Begum`, `Mrs Sultana Begum`, `Sultan Khan`, `Mr Sultan Khan`, `Khan, Sultan` and `Sultan Bin Zayed` read the given name, where 1.4.0 through 2.6.0 read the title (and `Mrs Sultana`, `Mr Sultan`). A style keeps the title (`His Majesty Sultan Hassanal Bolkiah`), as does `of` (`Sultan of Brunei`), and the Ottoman writing after the name was already a name word (`Hurrem Sultan`). The other sites read as for `king`.
   - THE COUNTS, recorded on the words (`uv run python tools/name_usage.py Sultana Sultan`, 2026-10-26): `sultana` given US SSA 275 births, Wikidata 19 (Bangladesh 5), nam_dict band 3 in Romania and Arabia/Persia, family Census 3,471, Wikidata 34; `sultan` given US SSA 2,396, Wikidata 151 (Turkey 28, Saudi Arabia 13), nam_dict band 7 in Turkey, family Census 3,363, Wikidata 106 (Ottoman Empire 26, the title written after the name).
   - THE COST, `sultan`'s only: a sultan written with the bare title in front reads it as the given name (`Sultan Hassanal Bolkiah` → given 'Sultan', where 1.4.0 through 2.6.0 read title 'Sultan', given 'Hassanal'; so `Sultan Qaboos bin Said`), the class accepted for `king`; in PSC one filer typed `Sultan` in the title box before a name, the one broken record. Sized afterwards, at the review's prompting, in the news text of jrc-names: 135 persons (266 spellings) open with `Sultan` or `Sultana`, by a reading of each about 40 rulers (mostly Ottoman sultans and today's Gulf and Malay rulers), about 80 people whose given name is Sultan, and 13 mosques and other places, a cost about four times `king`'s 10 there.
-  - FOUND BY THE REVIEW, NOT FIXED HERE: A v1 `remove` OF A WORD 2.6.0 LISTED NOW RAISES. A title that gains a given reading compiles out of `TITLES` (the 2026-10-25 #752, phase 2i entry of decisions/vocabulary-model), so `Constants().titles` no longer holds it and v1's strict `remove` raises `KeyError`: `c.titles.remove('sultan')`, which 2.6.0 accepted (and which then read `Sultan Khan` with first 'Sultan'), raises on this tree, and so does it for every word phase 2j moved out of the titles 2.6.0 listed (`wing`, `mahdi`, `bard`, `buddha`, `siddha`, `gyani`, `king`, `sultana`); phase 2i's own words were never in 2.6.0's titles, so removing them raised before too. The shim is not to raise on a config v1 accepted (AGENTS.md), so this is a defect of the batches, left for its own change before 2.7.0 ships.
+  - FOUND BY THE REVIEW, NOT FIXED HERE (fixed by the 2026-10-26 #752, the v1 title list entry below): A v1 `remove` OF A WORD 2.6.0 LISTED NOW RAISES. A title that gains a given reading compiles out of `TITLES` (the 2026-10-25 #752, phase 2i entry of decisions/vocabulary-model), so `Constants().titles` no longer holds it and v1's strict `remove` raises `KeyError`: `c.titles.remove('sultan')`, which 2.6.0 accepted (and which then read `Sultan Khan` with first 'Sultan'), raises on this tree, and so does it for every word phase 2j moved out of the titles 2.6.0 listed (`wing`, `mahdi`, `bard`, `buddha`, `siddha`, `gyani`, `king`, `sultana`); phase 2i's own words were never in 2.6.0's titles, so removing them raised before too. The shim is not to raise on a config v1 accepted (AGENTS.md), so this is a defect of the batches, left for its own change before 2.7.0 ships.
   - NOT REACHED, for `sheikh`'s batch. `sheikh` is a style, so under rules/H9 it partners a contested word behind it: `Sheikh Sultan Khan` keeps title 'Sheikh Sultan', as master reads it and as `Sheikh Dean Smith` reads, where in Gulf names the word is the person's given name.
   - MEASURED (2026-10-26, against master `6ba78d0d`, py3.11, every local check with `--base`). Both words: PSC (1,039,174 scored) 111 records fixed, 1 broken (the title box above), 6 changed, each still wrong for another reason, mostly a two-word surname box; ctgov-officials 41 moved; fec-individuals 21; tse-donors 1; spelling pairs (mechanisms.md#PAIRED-SPELLINGS-PER-CHANGE) jrc-names 9 persons and nepal-candidates 6 pairs now alike, none apart. `sultana` alone: PSC 28 fixed, none broken, 1 changed (a two-word surname box); ctgov 7, every one a fix, 6 of them `Aa Sultana, CRED` names whose credential master read as the given name (`Nasreen Sultana, MBBS` → given 'Nasreen', family 'Sultana', suffix 'MBBS', reported `suffix-or-name`); FEC 5 (`AA, SULTANA`); tse 1; nepal 1. The committed corpora and CLDR and GND agreement do not move.
   Recompute: `uv run python tools/differential/local_agreement.py psc-individuals --base <checkout of this commit's parent>`, and `uv run python tools/differential/snapshot.py --local ctgov-officials --base <the same checkout>` (and `fec-individuals`, `tse-donors`).
+
+- **2026-10-26 #752, the v1 title list — A v1 `Constants().titles` HOLDS EVERY TITLE 2.6.0 LISTED THAT IS STILL A TITLE OF EITHER KIND.** The review of phase 2j's fifth and sixth batches found that a word a batch moved from `titles` to `titles_ambiguous` also left v1's default `titles`, which the shim copied from TITLES: `Constants().titles.remove('king')` raised `KeyError` and `'king' in Constants().titles` was False, where 2.6.0 accepted the removal and answered True. Measured against the 2.6.0 wheel (`PYTHONSAFEPATH=1`, `nameparser.__file__` asserted on both sides), word by word over all nine v1 set attributes and the `capitalization_exceptions` keys: `titles` alone had lost words (`bard`, `buddha`, `gyani`, `king`, `mahdi`, `siddha`, `wing`, and `sultan` and `sultana` once the batch above landed), and `suffix_not_acronyms` had gained the seven agnomens and `fils` of phase 2h, a vocabulary addition its release bullet records. `suffix_acronyms_ambiguous` (derived in phase 2a), `given_name_abbreviations` (2k) and the phase 2h sets have no v1 manager that lost a word. The translation: `nameparser/config/_titles_2_6_0.py` freezes 2.6.0's TITLES, generated from the wheel and never edited, and v1's default `titles` is TITLES with every 2.6.0 title now in `titles_ambiguous` added back (`moved`). `Constants._build_snapshot` passes a moved word to `titles_ambiguous` while the caller keeps it, so a default `Constants` still translates to `Lexicon.default()`, and passes it to neither once removed, which turns its modifier, style and phrase readings off with it, as removing any title does: `c.titles.remove('king')` then reads `Hon. King Smith` as title 'Hon.', given 'King', 2.6.0's reading after the same removal. A word 2.6.0 never listed (`dean`, `justice`, `grace`, `major`, `marshal`) is not in `moved`, so `c.titles.add('dean')` still makes it a title everywhere. The gap, accepted because v1's one bit per word cannot carry three readings: adding a moved word in v1 restores the default, the phrase reading, and not 2.6.0's title everywhere, which a caller gets from the 2.0 API (`Lexicon.default().add(titles={'king'})`). Keyed on the compiled sets rather than a hand list, so a later batch that moves another 2.6.0 title (`sheikh`) is covered with no edit, as `sultan` and `sultana` were when the batch above merged in, and `tests/v2/test_config_shim.py` sweeps every moved word by the same derivation: membership, set operators, `remove`, `discard`, re-adding, a replaced list, pickle, deepcopy and copy, 37 of its 39 rows failing on master `6ba78d0d` (the two that pass are the non-empty guard and `dean`'s recipe), with rows added in review for a spelling Lexicon folds onto a moved word (`king。`, `wing。`: the moved test and the modifier, style and phrase intersections compare folded spellings, #582's rule, the second review finding that folding the first alone lost `Wing Commander`) and for a replaced list's unlisted words. The default parse does not move: the snapshot, the CLDR, GND and native agreement records and the facade case rows pass unchanged.

docs/release_log.rst

@@ -13,7 +13,7 @@ Release Log
     - **Fix a split Ph. D. in the middle of a name stopping short of the words after it.** ``John Smith Ph. D. Jones`` gives suffix ``Ph. D. Jones``, as ``John Smith PhD Jones`` does, where 1.4.0 through 2.6.0 gave last ``Jones``, suffix ``Ph. D.``; ``John Smith V Ph. D.`` gives last ``V``, as ``John Smith V PhD`` does, where 1.4.0 through 2.6.0 read ``V`` as a suffix. :issue:`827`
     - **Read the Brazilian agnomens and French fils as suffixes where the name has a word to spare.** ``Pedro Alves Filho`` and ``Alexandre Dumas fils`` give last ``Alves`` and ``Dumas``, suffix ``Filho`` and ``fils``, where 1.4.0 through 2.6.0 gave middle ``Alves`` and ``Dumas``, last ``Filho`` and ``fils``; ``Neto``, ``Netto``, ``Sobrinho`` and the feminines ``Filha``, ``Neta``, ``Sobrinha`` read the same way. Each is a surname too, so a name of two words keeps it as the last name (``Ana Neto``, ``PEDRO FILHO``) and ``Fils`` in Title case in a mixed-case name stays the last name (``Marie Fils``), the parse reports the choice as a ``suffix-or-name`` ambiguity. ``Lexicon.suffix_words_ambiguous`` lists the words and ``Lexicon.agnomens`` the ones read by the word count alone. :issue:`843`, :issue:`752`
     - **Read a credential that is also a name, written in capitals in a mixed-case name, as settled.** ``John Smith MA`` reports no ``suffix-or-name`` ambiguity, where 2.0.0 through 2.6.0 reported one, and ``Smith, John MA Jones`` gives suffix ``MA Jones``, where 2.0.0 through 2.6.0 gave middle ``MA Jones`` and 1.4.0 middle ``Jones``, suffix ``MA``. A name written in one case still reports (``JOHN SMITH MA``), and with no comma a name word behind the capitals keeps them a name word, since they may be initials (``Maria Ana MA Silva`` keeps last ``Silva``). :issue:`785`, :issue:`752`
-    - **Read a title that is also a given name as a title only inside a title phrase.** ``Chief Justice John Roberts`` gives title ``Chief Justice``, first ``John``, where 1.4.0 through 2.6.0 gave title ``Chief``, first ``Justice``; ``Field Marshal Bernard Montgomery``, ``Dean of Students Jane Smith`` and ``Major General John Smith`` give the phrase as the title too. Outside a phrase the word is still a name, so ``Dean Smith`` and ``Grace Smith`` give first ``Dean`` and ``Grace``; ``Dean Smith`` now reports a ``title-or-name`` ambiguity unless a comma has marked the given name (``Smith, Dean``), while ``Grace``, a title only in a phrase such as ``His Grace``, reports nothing. Listing the word as a title still makes it one everywhere (``Lexicon.default().add(titles={"dean"})``). :issue:`752`, :issue:`853`
+    - **Read a title that is also a given name as a title only inside a title phrase.** ``Chief Justice John Roberts`` gives title ``Chief Justice``, first ``John``, where 1.4.0 through 2.6.0 gave title ``Chief``, first ``Justice``; ``Field Marshal Bernard Montgomery``, ``Dean of Students Jane Smith`` and ``Major General John Smith`` give the phrase as the title too. Outside a phrase the word is still a name, so ``Dean Smith`` and ``Grace Smith`` give first ``Dean`` and ``Grace``; ``Dean Smith`` now reports a ``title-or-name`` ambiguity unless a comma has marked the given name (``Smith, Dean``), while ``Grace``, a title only in a phrase such as ``His Grace``, reports nothing. Listing the word as a title still makes it one everywhere (``Lexicon.default().add(titles={"dean"})``). A v1 ``Constants().titles`` still lists ``King``, ``Mahdi`` and the other titles 2.6.0 listed, and removing one reads it as a name everywhere, but adding one there no longer makes it a title outside a phrase; that takes the ``Lexicon`` spelling. :issue:`752`, :issue:`853`
     - **Fix the given names Sultan and Sultana reading as titles.** ``Sultan Khan`` and ``Sultana Begum`` give first ``Sultan`` and ``Sultana``, where 1.4.0 through 2.6.0 gave title ``Sultan`` and ``Sultana`` with no first name; ``Mr Sultan Khan`` and ``Khan, Sultan`` read the same. A sultan written with the bare title reads it as a first name too (``Sultan Hassanal Bolkiah`` gives first ``Sultan``), where a style keeps it (``His Majesty Sultan Hassanal Bolkiah``). :issue:`752`, :issue:`853`
     - **Fix the given name King reading as a title.** ``King Curtis`` and ``Mr King Smith`` give first ``King``, where 1.4.0 through 2.6.0 gave title ``King`` and ``Mr King``; ``SMITH, KING`` reads the same. A monarch written with the bare title reads it as a first name too (``King Henry`` gives first ``King``, last ``Henry``), where a style keeps it (``His Majesty King Henry``). :issue:`752`, :issue:`348`
     - **Fix the Nepali given names Buddha, Siddha and Gyani reading as titles.** ``Buddha Bahadur Tamang`` and ``Gyani Maya Gurung`` give first ``Buddha`` and ``Gyani``, middle ``Bahadur`` and ``Maya``, where 1.4.0 through 2.6.0 gave title ``Buddha`` and ``Gyani``, first ``Bahadur`` and ``Maya``; ``Siddha Raj Joshi`` and ``Tamang, Buddha Bahadur`` read the same. ``Guru`` is still a title. The Sikh title written ``Gyani`` reads as a given name too (``Gyani Zail Singh``); its usual spelling ``Giani`` is still a title. :issue:`869`, :issue:`752`