Fold a leading non-first-name prefix into the surname (closes #121)

Author: derek73

nameparser/parser.py

@@ -599,6 +599,11 @@ def is_first_name_prefix(self, piece: str) -> bool:
         """Lowercased, leading/trailing-periods-stripped version of piece is in :py:attr:`~nameparser.config.Constants.first_name_prefixes`."""
         return lc(piece) in self.C.first_name_prefixes
 
+    def is_non_first_name_prefix(self, piece: str) -> bool:
+        """Lowercased, leading/trailing-periods-stripped version of piece is in
+        :py:attr:`~nameparser.config.Constants.non_first_name_prefixes`."""
+        return lc(piece) in self.C.non_first_name_prefixes
+
     def _join_first_name_prefix(self, pieces: list[str], reserve_last: bool) -> list[str]:
         """Join a first-name prefix to its following piece.
 
@@ -821,6 +826,21 @@ def handle_turkic_patronymic_name_order(self) -> None:
                 self.first_list,
             )
 
+    def handle_non_first_name_prefix(self) -> None:
+        """
+        A leading prefix that is never a first name means the whole name is a
+        surname -- fold first (and any middle) into last. Keys on the parsed
+        first name, so a non-leading particle ("Jean de Mesnil") is untouched
+        and title/suffix are preserved. The middle_list/last_list guard leaves a
+        degenerate bare "de" as first="de" rather than inventing a surname.
+        """
+        if (len(self.first_list) == 1
+                and self.is_non_first_name_prefix(self.first_list[0])
+                and (self.middle_list or self.last_list)):
+            self.last_list = self.first_list + self.middle_list + self.last_list
+            self.first_list = []
+            self.middle_list = []
+
     def handle_middle_name_as_last(self) -> None:
         """
         When middle_name_as_last is enabled, fold middle_list into last_list
@@ -837,6 +857,7 @@ def post_process(self) -> None:
         and :py:func:`handle_capitalization`.
         """
         self.handle_firstnames()
+        self.handle_non_first_name_prefix()
         if self.C.patronymic_name_order:
             self.handle_east_slavic_patronymic_name_order()
             self.handle_turkic_patronymic_name_order()

tests/test_prefixes.py

@@ -278,3 +278,76 @@ def test_case_insensitive_prefix_detection(self) -> None:
         hn = HumanName("VINCENT VAN GOGH")
         self.m(hn.last_prefixes, "VAN", hn)
         self.m(hn.last_base, "GOGH", hn)
+
+    # --- targets: leading non-first-name prefix becomes the surname ---
+
+    def test_leading_non_first_name_prefix_de(self) -> None:
+        hn = HumanName("de Mesnil")
+        self.m(hn.first, "", hn)
+        self.m(hn.middle, "", hn)
+        self.m(hn.last, "de Mesnil", hn)
+
+    def test_leading_non_first_name_prefix_dos(self) -> None:
+        hn = HumanName("dos Santos")
+        self.m(hn.first, "", hn)
+        self.m(hn.last, "dos Santos", hn)
+
+    def test_leading_non_first_name_prefix_chain(self) -> None:
+        hn = HumanName("de la Vega")
+        self.m(hn.first, "", hn)
+        self.m(hn.last, "de la Vega", hn)
+
+    def test_leading_non_first_name_prefix_derived_props(self) -> None:
+        hn = HumanName("de Mesnil")
+        self.m(hn.last_prefixes, "de", hn)
+        self.m(hn.last_base, "Mesnil", hn)
+
+    def test_non_first_name_prefix_with_custom_title(self) -> None:
+        # 'Gunny' is NOT a default title -> genuinely exercises the custom-title
+        # path. Title is consumed first (first_list == ['']), so the fold does
+        # not fire and the surname is already correct; asserts we don't corrupt
+        # the title case.
+        CONSTANTS.titles.add('gunny')
+        hn = HumanName("Gunny de Mesnil")
+        self.m(hn.title, "Gunny", hn)
+        self.m(hn.first, "", hn)
+        self.m(hn.last, "de Mesnil", hn)
+
+    # --- safety: excluded / ambiguous particles are unchanged ---
+
+    def test_leading_von_is_unchanged(self) -> None:
+        hn = HumanName("von Braun")
+        self.m(hn.first, "von", hn)
+        self.m(hn.last, "Braun", hn)
+
+    def test_leading_van_is_unchanged(self) -> None:
+        hn = HumanName("Van Johnson")
+        self.m(hn.first, "Van", hn)
+        self.m(hn.last, "Johnson", hn)
+
+    def test_leading_della_is_unchanged(self) -> None:
+        hn = HumanName("Della Reese")
+        self.m(hn.first, "Della", hn)
+        self.m(hn.last, "Reese", hn)
+
+    def test_leading_di_is_unchanged(self) -> None:
+        hn = HumanName("Di Caprio")
+        self.m(hn.first, "Di", hn)
+        self.m(hn.last, "Caprio", hn)
+
+    def test_leading_del_is_unchanged(self) -> None:
+        hn = HumanName("Del Toro")
+        self.m(hn.first, "Del", hn)
+        self.m(hn.last, "Toro", hn)
+
+    def test_non_leading_prefix_is_unchanged(self) -> None:
+        hn = HumanName("Jean de Mesnil")
+        self.m(hn.first, "Jean", hn)
+        self.m(hn.last, "de Mesnil", hn)
+
+    # --- guard: bare particle with nothing to attach to ---
+
+    def test_bare_non_first_name_prefix_guard(self) -> None:
+        hn = HumanName("de")
+        self.m(hn.first, "de", hn)
+        self.m(hn.last, "", hn)