From 8d6ef523d98b736b1229dda91bcd95e6c499ca7b Mon Sep 17 00:00:00 2001
From: Matt VanEseltine <matvan@umich.edu>
Date: Thu, 20 Jun 2019 16:26:16 -0400
Subject: [PATCH] Prevent false nicknames due to multiple quotes

Certain Anglicized names such as those from some Hawaiian, Samoan,
and Kenyan traditions, include multiple single quotation marks.
This adjusts the quoted_word regex to only capture single quote
marks that are not inside words. Without this fix, false nicknames
are extracted from inside names like Ng'ang'a and Kawai'ae'a.
Tests are included to cover; existing Benjamin 'Ben' Franklin test
assures that the typical nickname case is unchanged.
---
 nameparser/config/regexes.py |  2 +-
 nameparser/parser.py         |  6 +++---
 tests.py                     | 28 ++++++++++++++++++++++++++++
 3 files changed, 32 insertions(+), 4 deletions(-)

diff --git a/nameparser/config/regexes.py b/nameparser/config/regexes.py
index beac95f..bd4b320 100644
--- a/nameparser/config/regexes.py
+++ b/nameparser/config/regexes.py
@@ -23,7 +23,7 @@
     ("word", re.compile(r"(\w|\.)+", re.U)),
     ("mac", re.compile(r'^(ma?c)(\w{2,})', re.I | re.U)),
     ("initial", re.compile(r'^(\w\.|[A-Z])?$', re.U)),
-    ("quoted_word", re.compile(r'\'([^\s]*?)\'', re.U)),
+    ("quoted_word", re.compile(r'(?<!\w)\'([^\s]*?)\'(?!\w)', re.U)),
     ("double_quotes", re.compile(r'\"(.*?)\"', re.U)),
     ("parenthesis", re.compile(r'\((.*?)\)', re.U)),
     ("roman_numeral", re.compile(r'^(X|IX|IV|V?I{0,3})$', re.I | re.U)),
diff --git a/nameparser/parser.py b/nameparser/parser.py
index 8c10c8f..5aa754d 100644
--- a/nameparser/parser.py
+++ b/nameparser/parser.py
@@ -413,9 +413,9 @@ def parse_nicknames(self):
         The content of parenthesis or quotes in the name will be added to the 
         nicknames list. This happens before any other processing of the name.
         
-        Single quotes cannot span white space characters to allow for single
-        quotes in names like O'Connor. Double quotes and parenthesis can span
-        white space.
+        Single quotes cannot span white space characters and must border
+        white space to allow for quotes in names like O'Connor and Kawai'ae'a.
+        Double quotes and parenthesis can span white space.
         
         Loops through 3 :py:data:`~nameparser.config.regexes.REGEXES`; 
         `quoted_word`, `double_quotes` and `parenthesis`.
diff --git a/tests.py b/tests.py
index 0ee7d75..b764955 100644
--- a/tests.py
+++ b/tests.py
@@ -1431,6 +1431,34 @@ def test_single_quotes_on_end_of_last_name_not_treated_as_nickname(self):
         self.m(hn.last, "Aube'", hn)
         self.m(hn.nickname, "", hn)
     
+    def test_okina_inside_name_not_treated_as_nickname(self):
+        hn = HumanName("Harrieta Keōpūolani Nāhiʻenaʻena")
+        self.m(hn.first, "Harrieta", hn)
+        self.m(hn.middle, "Keōpūolani", hn)
+        self.m(hn.last, "Nāhiʻenaʻena", hn)
+        self.m(hn.nickname, "", hn)
+
+    def test_single_quotes_not_treated_as_nickname_Hawaiian_example(self):
+        hn = HumanName("Harietta Keopuolani Nahi'ena'ena")
+        self.m(hn.first, "Harietta", hn)
+        self.m(hn.middle, "Keopuolani", hn)
+        self.m(hn.last, "Nahi'ena'ena", hn)
+        self.m(hn.nickname, "", hn)
+
+    def test_single_quotes_not_treated_as_nickname_Kenyan_example(self):
+        hn = HumanName("Naomi Wambui Ng'ang'a")
+        self.m(hn.first, "Naomi", hn)
+        self.m(hn.middle, "Wambui", hn)
+        self.m(hn.last, "Ng'ang'a", hn)
+        self.m(hn.nickname, "", hn)
+
+    def test_single_quotes_not_treated_as_nickname_Samoan_example(self):
+        hn = HumanName("Va'apu'u Vitale")
+        self.m(hn.first, "Va'apu'u", hn)
+        self.m(hn.middle, "", hn)
+        self.m(hn.last, "Vitale", hn)
+        self.m(hn.nickname, "", hn)
+
     # http://code.google.com/p/python-nameparser/issues/detail?id=17
     def test_parenthesis_are_removed_from_name(self):
         hn = HumanName("John Jones (Unknown)")