fix(titles): swami, guru, baba, lama join the given-name titles

The four renunciate titles were in the TITLES-only block, so rules.md#H1 familied the one name word behind them: "Swami Vivekananda" read family 'Vivekananda' with no given name at every release since 1.4.0. rules.md#H's Background has stated the criteria since the docs landed -- membership in the given-name-title list follows HOW THE TITLE ADDRESSES, and whether an empty family is correct output follows SURNAME RETENTION -- and renunciation answers both at once. So they move, and "Swami Vivekananda" reads given 'Vivekananda', family ''. rabbi and imam stay out: those traditions keep surnames, and the control row surname_retaining_title_keeps_the_family pins "Rabbi Cohen" at family 'Cohen'. venerable stays in TITLES only -- Buddhist monastics drop the family name, Anglican archdeacons and Catholic candidates keep it, and this set has no way to say "sometimes". Four corpus names move, all from the radar-tier corpus_issues.jsonl and all the same way: 'Swami Vivekananda', 'Guru Nanak', 'Baba Ramdev' and 'Lama Zopa'. The boundary holds and is pinned by the ledger's _MUST_NOT_MATCH probes -- 'Swami Vivekananda Saraswati' and 'Guru Gobind Singh' keep their family names, H1's fold needing exactly one name word -- and 'Md Abdul Karim' does not move. TITLES membership is unchanged: TITLES is GIVEN_NAME_TITLES | {...}, so the words only changed which side of the union they enter from. Gate: 364 / 258 / 170 / 32 intentional at 1.4.0 / 2.0.0 / 2.1.0 / 2.2.0, 0 unexplained, 0 radar unclassified. Closes #346 Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
derek73 · Sep 7, 2026 · Sep 7, 2026 · Sep 7, 2026 · Sep 7, 2026 · Sep 7, 2026
commit 3263a8d583a0a507eae821cbd4b1a9bfa5b96710
diff --git a/nameparser/config/titles.py b/nameparser/config/titles.py
@@ -3,17 +3,33 @@
 GIVEN_NAME_TITLES = frozenset({
     'aunt',
     'auntie',
+    # #346: the renunciate class -- 'baba' here, and 'guru', 'lama',
+    # 'swami' in their own alphabetical places below. All four were in
+    # the TITLES-only block until 2026-09-06. Two separate criteria
+    # both say yes (rules.md#H Background): the title addresses by the
+    # GIVEN name, which is what membership here means, and the
+    # traditions renounce the surname, which is what makes the empty
+    # family the right output for "Swami Vivekananda". 'rabbi' and
+    # 'imam' pass the first and fail the second, so they stay out;
+    # 'venerable' stays in TITLES only, because the traditions using it
+    # split on surname retention (Buddhist monastics drop the family
+    # name, Anglican archdeacons keep it) and this set has no way to
+    # say "sometimes".
+    'baba',
     'brother',
     'dame',
     'father',
+    'guru',
     'king',
+    'lama',
     'maid',
     'master',
     'mother',
     'pope',
     'queen',
     'sir',
     'sister',
+    'swami',
     'uncle',
     'sheikh',
     'sheik',
@@ -142,7 +158,6 @@
     'author',
     'award-winning',
     'ayatollah',
-    'baba',
     'bailiff',
     'ballet',
     'bandleader',
@@ -355,7 +370,6 @@
     'großfürst',
     'group',
     'guitarist',
-    'guru',
     'gyani',
     'gysgt',
     'hajji',
@@ -400,7 +414,6 @@
     'kingdom',
     'knowledge',
     'lady',
-    'lama',
     'lamido',
     'law',
     'lawyer',
@@ -664,7 +677,6 @@
     'superior',
     'supreme',
     'surgeon',
-    'swami',
     'swordbearer',
     'sysselmann',
     'tax',

diff --git a/tests/v2/cases.py b/tests/v2/cases.py
@@ -2552,6 +2552,31 @@ def _check_cjk_shape_purity(self) -> None:
          {"family": "Smith", "suffix": "MD"},
          classification="fix(#296)",
          notes="the third leg of 'position decides' for 'md'"),
+    # -- #346: the renunciate class. A title of religious renunciation
+    # addresses the GIVEN name AND belongs to a tradition that
+    # abolishes the surname, so rules.md#H1's given-name-title fold is
+    # the right reading and family="" is the right output. The two
+    # criteria are separate (rules.md#H Background) and the control
+    # row below is what keeps them separate.
+    Case("renunciate_title_keeps_the_given_name", "Swami Vivekananda",
+         {"title": "Swami", "given": "Vivekananda"},
+         classification="fix(#346)",
+         notes="pins the #346 membership change the ledger "
+               "classifies: 'swami' moved from the TITLES-only block "
+               "into GIVEN_NAME_TITLES, so H1's fold fires and the "
+               "family is empty where 1.4.0 through 2.2.0 read family "
+               "'Vivekananda'. The fold itself is already pinned by "
+               "the Sir John rows; this is the name the release note "
+               "advertises"),
+    Case("surname_retaining_title_keeps_the_family",
+         "Rabbi Cohen",
+         {"title": "Rabbi", "family": "Cohen"},
+         notes="makes rules.md#H Background's 'Rabbi Cohen' sentence "
+               "executable -- no other row parses a Rabbi. 'rabbi' "
+               "addresses by title and the tradition keeps surnames, "
+               "so it stays out of GIVEN_NAME_TITLES and H1 families "
+               "the one name word. The control that keeps 'addresses "
+               "by the given name' and 'has no surname' separate"),
     Case("audit_do_leading_is_a_name", "Do Nguyen",
          {"given": "Do", "family": "Nguyen"},
          ambiguities=("particle-or-given",),

diff --git a/tests/v2/test_ledger_guards.py b/tests/v2/test_ledger_guards.py
@@ -883,6 +883,15 @@ def test_case_shape_ids_exist_in_the_inventory() -> None:
     "fix(#436/#437) the space-separated post-nominal run compounds with the bound-given reserve":
         ("abdul Smith V", "abdul Smith Jr Ma", "abdul Smith Jr",
          "Smith Jr V"),
+    # #346's rule is a literal alternation of four names, so a
+    # widening reaching only names the corpora lack would leave
+    # _CORPUS_CLAIMS unmoved. These probes are the wall, and each is a
+    # name that must not move: two name words after the title (H1 does
+    # not fold), a surname-retaining title, and the bare renunciate
+    # title standing alone.
+    "fix(#346) a renunciate title and one name word leave the name a given name":
+        ("Swami Vivekananda Saraswati", "Guru Gobind Singh",
+         "Rabbi Cohen", "Swami"),
 }
 
 
@@ -1508,6 +1517,15 @@ class _LatinCopy(NamedTuple):
                "The Rt Hon Kenneth Clarke QC MP, HMG",
                "Washington Jr\\. MD, Franklin", "abdul Smith Jr Ma",
                "abdul Smith Jr V"}),
+    # #346's movers, one corpus name per alternative -- a list of
+    # names, not a copy of any wordlist, so there is no vocabulary for
+    # it to drift from. The rule's subject is a SHAPE the vocabulary
+    # participates in (a renunciate title plus exactly one name word),
+    # and a member copying GIVEN_NAME_TITLES would reach 'Swami
+    # Vivekananda Saraswati' and 'Guru Gobind Singh', which do not
+    # move. One set, identical in all four ledgers.
+    frozenset({"Baba Ramdev", "Guru Nanak", "Lama Zopa",
+               "Swami Vivekananda"}),
     # fix(#445)'s movers, one corpus name per alternative -- a list of
     # names, not a copy of any wordlist, so there is no vocabulary for
     # it to drift from. Two sets because the ledgers group the nine
@@ -1994,6 +2012,12 @@ def _claim(rule: dict) -> _Claim:
         # change the row here before it reached the gate.
         "fix(#436/#437) a space-separated post-nominal run renders with spaces, not commas":
             _Claim(10, ('suffix',), "30f5314a2662", None),
+        # #346's alternation. Four corpus names, `family` and
+        # `given` together: the fold moves both roles at once, so a
+        # widening taking one alone would change the roles here
+        # before it reached the gate.
+        "fix(#346) a renunciate title and one name word leave the name a given name":
+            _Claim(4, ('family', 'given'), "a3399ee7b21e", None),
         "fix(A2) content-free input names nobody, so every role empties":
             _Claim(5, ('given',), "1af8d718688b", None),
         "fix(#335) a marker-led clause leaves the one name word its bare reading":
@@ -2202,6 +2226,12 @@ def _claim(rule: dict) -> _Claim:
         # change the row here before it reached the gate.
         "fix(#436/#437) a space-separated post-nominal run renders with spaces, not commas":
             _Claim(10, ('suffix',), "30f5314a2662", None),
+        # #346's alternation. Four corpus names, `family` and
+        # `given` together: the fold moves both roles at once, so a
+        # widening taking one alone would change the roles here
+        # before it reached the gate.
+        "fix(#346) a renunciate title and one name word leave the name a given name":
+            _Claim(4, ('family', 'given'), "a3399ee7b21e", None),
         # The compound rule, at the two baselines where 'abdul Smith
         # Jr V' already diffs {family, given} under fix(#401) and the
         # widened diff leaves that rule's `fields`. Three roles here
@@ -2371,6 +2401,12 @@ def _claim(rule: dict) -> _Claim:
         # change the row here before it reached the gate.
         "fix(#436/#437) a space-separated post-nominal run renders with spaces, not commas":
             _Claim(10, ('suffix',), "30f5314a2662", None),
+        # #346's alternation. Four corpus names, `family` and
+        # `given` together: the fold moves both roles at once, so a
+        # widening taking one alone would change the roles here
+        # before it reached the gate.
+        "fix(#346) a renunciate title and one name word leave the name a given name":
+            _Claim(4, ('family', 'given'), "a3399ee7b21e", None),
         # The four one-name CJK rules, literal-anchored, at the two
         # baselines where the render is the whole of what moved. A
         # reach of 1 is one _CORPUS_CLAIMS cannot police on its own --
@@ -2394,6 +2430,12 @@ def _claim(rule: dict) -> _Claim:
         # change the row here before it reached the gate.
         "fix(#436/#437) a space-separated post-nominal run renders with spaces, not commas":
             _Claim(10, ('suffix',), "30f5314a2662", None),
+        # #346's alternation. Four corpus names, `family` and
+        # `given` together: the fold moves both roles at once, so a
+        # widening taking one alone would change the roles here
+        # before it reached the gate.
+        "fix(#346) a renunciate title and one name word leave the name a given name":
+            _Claim(4, ('family', 'given'), "a3399ee7b21e", None),
         # The four one-name CJK rules, literal-anchored, at the two
         # baselines where the render is the whole of what moved. A
         # reach of 1 is one _CORPUS_CLAIMS cannot police on its own --

diff --git a/tools/differential/expected_since_1.4.0.toml b/tools/differential/expected_since_1.4.0.toml
@@ -84,6 +84,35 @@ issue = "fix(#436/#437) a space-separated post-nominal run renders with spaces,
 name_regex = "^(?:JOHN DOE PHD MD|John Doe MD PhD|John Smith MD PhD|John Smith Mc V|Kenneth Clarke QC MP|Smith, John PhD I\\.|The Rt Hon Kenneth Clarke QC MP, HMG|Washington Jr\\. MD, Franklin|abdul Smith Jr Ma|abdul Smith Jr V)$"
 fields = ["suffix"]
 
+# #346: swami, guru, baba and lama moved from the TITLES-only block
+# into GIVEN_NAME_TITLES on 2026-09-06. rules.md#H1's Accepted clause
+# -- "a given-name title plus one name word leaves the family empty"
+# -- now reaches them, so the one name word behind the title is the
+# GIVEN name and the family is empty, where every baseline read it as
+# the family. Both roles move together and `fields` declares both: a
+# name whose TITLE moved would arrive UNEXPLAINED, as it should.
+#
+# Four corpus names, all from the radar-tier corpus_issues.jsonl and
+# all moving the same way. The two- and three-word spellings are the
+# boundary and are _MUST_NOT_MATCH probes: 'Swami Vivekananda
+# Saraswati' and 'Guru Gobind Singh' keep their family names, because
+# H1's fold requires exactly one name word, and a rule reaching them
+# would stand ready to explain the very regression that boundary
+# exists to catch.
+#
+# An anchored alternation of the four NAMES rather than a shape,
+# under the ledger discipline expected_since_1.4.0.toml's header
+# states for all four files: keep each entry's `name_regex` as tight
+# as the diff allows. Four names moved, so a four-name list is the
+# tightest expression of that diff; anything shape-shaped is wider
+# than what was measured. It copies no wordlist either, which is what
+# _NOT_A_VOCABULARY_COPY in tests/v2/test_ledger_guards.py records;
+# _CORPUS_CLAIMS pins the reach at 4 with its digest.
+[[change]]
+issue = "fix(#346) a renunciate title and one name word leave the name a given name"
+name_regex = "^(?:Baba Ramdev|Guru Nanak|Lama Zopa|Swami Vivekananda)$"
+fields = ["family", "given"]
+
 [[change]]
 issue = "fix(#271/#272/#298) native-script CJK: family-first order, hangul segmentation, the kana license and the dots"
 # '毛 泽东', '김민준': script_orders flips first/last for a name written

diff --git a/tools/differential/expected_since_2.0.0.toml b/tools/differential/expected_since_2.0.0.toml
@@ -57,6 +57,35 @@ issue = "fix(#436/#437) a space-separated post-nominal run renders with spaces,
 name_regex = "^(?:JOHN DOE PHD MD|John Doe MD PhD|John Smith MD PhD|John Smith Mc V|Kenneth Clarke QC MP|Smith, John PhD I\\.|The Rt Hon Kenneth Clarke QC MP, HMG|Washington Jr\\. MD, Franklin|abdul Smith Jr Ma|abdul Smith Jr V)$"
 fields = ["suffix"]
 
+# #346: swami, guru, baba and lama moved from the TITLES-only block
+# into GIVEN_NAME_TITLES on 2026-09-06. rules.md#H1's Accepted clause
+# -- "a given-name title plus one name word leaves the family empty"
+# -- now reaches them, so the one name word behind the title is the
+# GIVEN name and the family is empty, where every baseline read it as
+# the family. Both roles move together and `fields` declares both: a
+# name whose TITLE moved would arrive UNEXPLAINED, as it should.
+#
+# Four corpus names, all from the radar-tier corpus_issues.jsonl and
+# all moving the same way. The two- and three-word spellings are the
+# boundary and are _MUST_NOT_MATCH probes: 'Swami Vivekananda
+# Saraswati' and 'Guru Gobind Singh' keep their family names, because
+# H1's fold requires exactly one name word, and a rule reaching them
+# would stand ready to explain the very regression that boundary
+# exists to catch.
+#
+# An anchored alternation of the four NAMES rather than a shape,
+# under the ledger discipline expected_since_1.4.0.toml's header
+# states for all four files: keep each entry's `name_regex` as tight
+# as the diff allows. Four names moved, so a four-name list is the
+# tightest expression of that diff; anything shape-shaped is wider
+# than what was measured. It copies no wordlist either, which is what
+# _NOT_A_VOCABULARY_COPY in tests/v2/test_ledger_guards.py records;
+# _CORPUS_CLAIMS pins the reach at 4 with its digest.
+[[change]]
+issue = "fix(#346) a renunciate title and one name word leave the name a given name"
+name_regex = "^(?:Baba Ramdev|Guru Nanak|Lama Zopa|Swami Vivekananda)$"
+fields = ["family", "given"]
+
 [[change]]
 issue = "fix(#271/#272/#298) native-script CJK: family-first order, hangul segmentation, the kana license and the dots"
 # The 2.1 East Asian defaults, in the fields they move name pieces

diff --git a/tools/differential/expected_since_2.1.0.toml b/tools/differential/expected_since_2.1.0.toml
@@ -115,6 +115,35 @@ issue = "fix(#436/#437) the glued honorific and the generational suffix are one
 name_regex = "^김민준씨 Jr\\.$"
 fields = ["suffix"]
 
+# #346: swami, guru, baba and lama moved from the TITLES-only block
+# into GIVEN_NAME_TITLES on 2026-09-06. rules.md#H1's Accepted clause
+# -- "a given-name title plus one name word leaves the family empty"
+# -- now reaches them, so the one name word behind the title is the
+# GIVEN name and the family is empty, where every baseline read it as
+# the family. Both roles move together and `fields` declares both: a
+# name whose TITLE moved would arrive UNEXPLAINED, as it should.
+#
+# Four corpus names, all from the radar-tier corpus_issues.jsonl and
+# all moving the same way. The two- and three-word spellings are the
+# boundary and are _MUST_NOT_MATCH probes: 'Swami Vivekananda
+# Saraswati' and 'Guru Gobind Singh' keep their family names, because
+# H1's fold requires exactly one name word, and a rule reaching them
+# would stand ready to explain the very regression that boundary
+# exists to catch.
+#
+# An anchored alternation of the four NAMES rather than a shape,
+# under the ledger discipline expected_since_1.4.0.toml's header
+# states for all four files: keep each entry's `name_regex` as tight
+# as the diff allows. Four names moved, so a four-name list is the
+# tightest expression of that diff; anything shape-shaped is wider
+# than what was measured. It copies no wordlist either, which is what
+# _NOT_A_VOCABULARY_COPY in tests/v2/test_ledger_guards.py records;
+# _CORPUS_CLAIMS pins the reach at 4 with its digest.
+[[change]]
+issue = "fix(#346) a renunciate title and one name word leave the name a given name"
+name_regex = "^(?:Baba Ramdev|Guru Nanak|Lama Zopa|Swami Vivekananda)$"
+fields = ["family", "given"]
+
 [[change]]
 issue = "fix(#369) a given-name title licenses the bound given-name join with one word to spare"
 # 'Sheik abdul salam': rules.md#P5 -- "or after a given-name title,

diff --git a/tools/differential/expected_since_2.2.0.toml b/tools/differential/expected_since_2.2.0.toml
@@ -104,6 +104,35 @@ issue = "fix(#436/#437) the glued honorific and the generational suffix are one
 name_regex = "^김민준씨 Jr\\.$"
 fields = ["suffix"]
 
+# #346: swami, guru, baba and lama moved from the TITLES-only block
+# into GIVEN_NAME_TITLES on 2026-09-06. rules.md#H1's Accepted clause
+# -- "a given-name title plus one name word leaves the family empty"
+# -- now reaches them, so the one name word behind the title is the
+# GIVEN name and the family is empty, where every baseline read it as
+# the family. Both roles move together and `fields` declares both: a
+# name whose TITLE moved would arrive UNEXPLAINED, as it should.
+#
+# Four corpus names, all from the radar-tier corpus_issues.jsonl and
+# all moving the same way. The two- and three-word spellings are the
+# boundary and are _MUST_NOT_MATCH probes: 'Swami Vivekananda
+# Saraswati' and 'Guru Gobind Singh' keep their family names, because
+# H1's fold requires exactly one name word, and a rule reaching them
+# would stand ready to explain the very regression that boundary
+# exists to catch.
+#
+# An anchored alternation of the four NAMES rather than a shape,
+# under the ledger discipline expected_since_1.4.0.toml's header
+# states for all four files: keep each entry's `name_regex` as tight
+# as the diff allows. Four names moved, so a four-name list is the
+# tightest expression of that diff; anything shape-shaped is wider
+# than what was measured. It copies no wordlist either, which is what
+# _NOT_A_VOCABULARY_COPY in tests/v2/test_ledger_guards.py records;
+# _CORPUS_CLAIMS pins the reach at 4 with its digest.
+[[change]]
+issue = "fix(#346) a renunciate title and one name word leave the name a given name"
+name_regex = "^(?:Baba Ramdev|Guru Nanak|Lama Zopa|Swami Vivekananda)$"
+fields = ["family", "given"]
+
 [[change]]
 issue = "fix(#462) the facade keeps an initial-shaped conjunction letter"
 # 'Scott E. Werner', 'Amy E Maid', 'Jose E. Maria Santos':