Handle newlines in the middle of a name

ynr/apps/sopn_parsing/helpers/parse_tables.py

@@ -153,7 +153,18 @@ def clean_name(name):
     - Build a string to represent the other names by looking for all words not in all caps
     - Strip whitespace in case last_names is empty and return string titleized
     """
-    name = name.replace("\n", " ")
+
+    if "\n" in name and "-" not in name:
+        if name.count("\n") == 1 and name.find("\n") != -1:
+            name = name.replace("\n", " ")
+        #  # if the newline is in the middle of a word, then replace with an empty string
+        elif name.count("\n") > 1 or name.find("\n") == -1:
+            name = name.replace("\n", "")
+        else:
+            name = name.replace("\n", " ")
+    elif "\n" in name and "-" in name:
+        name = name.replace("\n", " ")
+
     name = name.replace("`", "'")
     name = name.replace("\u2013", "\u002d")
     # remove multiple whitespaces

ynr/apps/sopn_parsing/tests/test_parse_tables.py

@@ -319,10 +319,12 @@ def test_clean_name_replaces_backticks(self):
         assert "'" in name
 
     def test_clean_name_replaces_newlines(self):
-        name = parse_tables.clean_name(
-            "A Very Long Name That Splits \nOver Lines"
-        )
+        name = parse_tables.clean_name("D\nICKIE \nPaul")
+        assert "\n" not in name
+        assert name == "Paul Dickie"
+        name = parse_tables.clean_name("M\nARSH-PRITCHARD Thomas")
         assert "\n" not in name
+        assert name == "Thomas Marsh-Pritchard"
 
     def test_clean_name_capitalized_last_and_titalized(self):
         name = parse_tables.clean_name("SMITH John")