fix #126, don't count as prefix when first name

derek73 · derek73 · commit 2a3edaef456e · 2022-01-03T15:01:12.000-08:00
update initials processing to only exclude conjunctions and prefixes when it is not a first name.
diff --git a/nameparser/parser.py b/nameparser/parser.py
@@ -191,10 +191,17 @@ def as_dict(self, include_empty=True):
 
     def __process_initial__(self, name_part):
         """
-            Name parts may include prefixes or conjuctions. This function filters these from the name.
-        """
-        parsed = " ".join([split for split in name_part.split(" ") if len(split) and not (self.is_prefix(split) or self.is_conjunction(split))])
-        return parsed[0] if len(parsed) else ""
+            Name parts may include prefixes or conjuctions. This function filters these from the name unless it is
+            a first name, since first names cannot be conjunctions or prefixes.
+        """
+        parts = name_part.split(" ")
+        parsed = ""
+        if len(parts) and not (name_part == 'first' and (self.is_prefix(parts) or self.is_conjunction(parts))):
+            parsed = " ".join(parts)
+        if len(parsed) > 0:
+            return parsed[0]
+        else:
+            return self.C.empty_attribute_default
 
     def initials_list(self):
         """
@@ -855,6 +862,9 @@ def join_on_conjunctions(self, pieces, additional_parts_count=0):
                 # join everything after the prefix until the next prefix or suffix
 
                 try:
+                    if i == 0 and total_length >= 1:
+                        # If it's the first piece and there are more than 1 rootnames, assume it's a first name
+                        continue
                     next_prefix = next(iter(filter(self.is_prefix, pieces[i + 1:])))
                     j = pieces.index(next_prefix)
                     if j == i + 1:
diff --git a/tests.py b/tests.py
@@ -1605,6 +1605,17 @@ def test_prefix_before_two_part_last_name(self):
         self.m(hn.first, "pennie", hn)
         self.m(hn.last, "von bergen wessels", hn)
 
+    def test_prefix_is_first_name(self):
+        hn = HumanName("Van Johnson")
+        self.m(hn.first, "Van", hn)
+        self.m(hn.last, "Johnson", hn)
+
+    def test_prefix_is_first_name_with_middle_name(self):
+        hn = HumanName("Van Jeremy Johnson")
+        self.m(hn.first, "Van", hn)
+        self.m(hn.middle, "Jeremy", hn)
+        self.m(hn.last, "Johnson", hn)
+
     def test_prefix_before_two_part_last_name_with_suffix(self):
         hn = HumanName("pennie von bergen wessels III")
         self.m(hn.first, "pennie", hn)
@@ -2313,6 +2324,10 @@ def test_initials_list_complex_name(self):
         hn = HumanName("Doe, John A. Kenneth, Jr.")
         self.m(hn.initials_list(), ["J", "A", "K", "D"], hn)
 
+    def test_initials_with_prefix_firstname(self):
+        hn = HumanName("Van Jeremy Johnson")
+        self.m(hn.initials_list(), ["V", "J", "J"], hn)
+
 
 TEST_NAMES = (
     "John Doe",
