Add German/Dutch prefixes and German title/degree suffixes

Closes #18. Adds prefixes (aan, aen, auf, dem, freiherr, freiherrin,
heer, het, op, te, tho, thoe, vande, vd) and titles/suffixes (Dipl.-Ing.,
FH-Prof., Gräfin, Me., PD, Priv.-Doz., RA, Univ.Prof., WP, ba, bsc, meng,
stb, MdB/MdL/MdEP/MdA/MdHB/MdBB) that don't collide with existing
English-language parsing.

Also fixes join_on_conjunctions() to register a conjunction-merged piece
(e.g. "von" + "und" + "zu") as a prefix too, mirroring the existing
title-handling, so multi-word prefix chains like German "von und zu"
bridge correctly into the last name instead of getting stranded in the
middle name.

Deliberately left out short, high-frequency English words (to, in, an,
then, ten) that collide with common Korean/Vietnamese given-name
syllables in the middle-token position, and bare "v" as a prefix, which
collides with ordinary Western middle initials.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

Author: derek73

nameparser/config/prefixes.py

@@ -9,9 +9,12 @@
 #: correct parsing of the last name "von bergen wessels".
 PREFIXES = set([
     "'t",
+    'aan',
+    'aen',
     'abu',
     'af',
     'al',
+    'auf',
     'av',
     'bar',
     'bat',
@@ -30,25 +33,36 @@
     'delle',
     'delli',
     'dello',
+    'dem',
     'den',
     'der',
     'di',
     'dí',
     'do',
     'dos',
     'du',
+    'freiherr',
+    'freiherrin',
+    'heer',
+    'het',
     'ibn',
     'la',
     'le',
     'mac',
     'mc',
+    'op',
     'san',
     'santa',
     'st',
     'ste',
+    'te',
     'ter',
+    'tho',
+    'thoe',
     'van',
+    'vande',
     'vander',
+    'vd',
     'vel',
     'vom',
     'von',

nameparser/config/suffixes.py

@@ -97,6 +97,7 @@
     'asp',
     'atc',
     'awb',
+    'ba',
     'bca',
     'bcl',
     'bcss',
@@ -107,6 +108,7 @@
     'bpe',
     'bpi',
     'bpt',
+    'bsc',
     'bt',
     'btcs',
     'bts',
@@ -507,8 +509,15 @@
     'mcse',
     'mct',
     'md',
+    'mda',
+    'mdb',
+    'mdbb',
+    'mdep',
+    'mdhb',
     'mdiv',
+    'mdl',
     'mem',
+    'meng',
     'mfa',
     'micp',
     'mieee',
@@ -653,6 +662,7 @@
     'sphr',
     'sra',
     'sscp',
+    'stb',
     'stmieee',
     'tbr-ct',
     'td',

nameparser/config/titles.py

@@ -231,6 +231,7 @@
     'detective',
     'developer',
     'dhr',
+    'dipl.-ing',
     'diplomat',
     'dir',
     'director',
@@ -282,6 +283,7 @@
     'fadm',
     'family',
     'federal',
+    'fh-prof',
     'field',
     'film',
     'financial',
@@ -312,6 +314,7 @@
     'goodwife',
     'governor',
     'graf',
+    'gräfin',
     'grand',
     'großfürst',
     'group',
@@ -412,6 +415,7 @@
     'mcpoc',
     'mcpon',
     'md',
+    'me',
     'member',
     'memoirist',
     'merchant',
@@ -470,6 +474,7 @@
     'paleontologist',
     'pastor',
     'patriarch',
+    'pd',
     'pediatrician',
     'personality',
     'petty',
@@ -512,6 +517,7 @@
     'printer',
     'printmaker',
     'prinz',
+    'priv.-doz',
     'prior',
     'private',
     'pro',
@@ -526,6 +532,7 @@
     'pursuivant',
     'pv2',
     'pvt',
+    'ra',
     'rabbi',
     'radio',
     'radm',
@@ -648,6 +655,7 @@
     'tsgt',
     'uk',
     'united',
+    'univ.prof',
     'us',
     'vadm',
     'vardapet',
@@ -665,6 +673,7 @@
     'warrant',
     'wing',
     'wm',
+    'wp',
     'wo-1',
     'wo1',
     'wo2',

nameparser/parser.py

@@ -1149,6 +1149,11 @@ def join_on_conjunctions(self, pieces: list[str], additional_parts_count: int =
                 if self.is_title(pieces[i+1]):
                     # when joining to a title, make new_piece a title too
                     self.C.titles.add(new_piece)
+                if self.is_prefix(pieces[i+1]):
+                    # when joining to a prefix, make new_piece a prefix too, so
+                    # e.g. "von" + "und" bridges into "von und" and can still
+                    # chain onto a following prefix/lastname (see "von und zu")
+                    self.C.prefixes.add(new_piece)
                 pieces[i] = new_piece
                 pieces.pop(i+1)
                 # subtract 1 from the index of all the remaining conjunctions
@@ -1161,6 +1166,11 @@ def join_on_conjunctions(self, pieces: list[str], additional_parts_count: int =
                 if self.is_title(pieces[i-1]):
                     # when joining to a title, make new_piece a title too
                     self.C.titles.add(new_piece)
+                if self.is_prefix(pieces[i-1]):
+                    # when joining to a prefix, make new_piece a prefix too, so
+                    # e.g. "von" + "und" bridges into "von und" and can still
+                    # chain onto a following prefix/lastname (see "von und zu")
+                    self.C.prefixes.add(new_piece)
                 pieces[i-1] = new_piece
                 pieces.pop(i)
                 rm_count = 2

tests/test_conjunctions.py

@@ -223,3 +223,34 @@ def test_conjunction_in_an_address_with_a_first_name_title(self) -> None:
     def test_name_is_conjunctions(self) -> None:
         hn = HumanName("e and e")
         self.m(hn.first, "e and e", hn)
+
+    def test_conjunction_bridges_prefix_chain(self) -> None:
+        # "von" and "zu" are both prefixes, but "und" between them is only a
+        # conjunction. join_on_conjunctions() merges "von und zu" into one
+        # piece before the prefix-joining step runs, so without registering
+        # that merged piece as a prefix too, it's stranded in the middle name
+        # instead of joining to the last name. See German nobility styles
+        # like "von und zu".
+        hn = HumanName("Alois von und zu Liechtenstein")
+        self.m(hn.first, "Alois", hn)
+        self.m(hn.middle, "", hn)
+        self.m(hn.last, "von und zu Liechtenstein", hn)
+
+    def test_conjunction_bridges_prefix_chain_with_leading_title(self) -> None:
+        # Same bridging, but with extra prefix words on both sides of the
+        # conjunction and a leading title-like word ("Freiherrin") that is
+        # itself a prefix, confirming the chain still joins fully into last.
+        hn = HumanName("Annette Charlotte Freiherrin von und zu der Tann-Rathsamhausen")
+        self.m(hn.first, "Annette", hn)
+        self.m(hn.middle, "Charlotte", hn)
+        self.m(hn.last, "Freiherrin von und zu der Tann-Rathsamhausen", hn)
+
+    def test_conjunction_prefix_merge_at_start_stays_first_name(self) -> None:
+        # Guards the i == 0 branch of the same fix: when the conjunction is
+        # merged with a following prefix at the very start of the name, the
+        # existing leading-prefix rule (a lone prefix opening the name is
+        # treated as part of the first name, not joined to last) must still
+        # apply to the merged piece.
+        hn = HumanName("and van Buren")
+        self.m(hn.first, "and van", hn)
+        self.m(hn.last, "Buren", hn)