Work around states w dots, move "United States" code

* Work around the appearance of periods in state abbreviations, like
  "S.C." which GeoNames will not resolve.
* Move the code that normalizes "United States" and its variants into
  geocode.set_name().

lib/akamod/geocode.py

@@ -168,10 +168,6 @@ def geocode_place(self, place):
         are already present, will attempt to validate them before accepting
         the new place.
         """
-        if place.name and re.search(ur" *(United States(?!-)|États-Unis|USA)",
-                                    place.name):
-            place.name = 'United States'
-
         place.set_name()
         params = {}
 
@@ -457,10 +453,20 @@ def map_fields(self):
 
     def set_name(self):
         """
-        Returns the name property. If none is set, sets it to the smallest
-        available geographic division label.
+        Modify and return our name property, after cleaning it up. If none is
+        set, initialize it to the smallest available geographic division label.
         """
         if self.name:
+            if re.search(ur" *(United States(?!-)|États-Unis|USA)", self.name):
+                self.name = 'United States'
+            # Kludge state abbreviations with periods (e.g. "S.C." or "CA.")
+            # into their equivalent official post-office abbreviations (e.g.
+            # "SC" or "CA"). GeoNames just doesn't resolve the ones with
+            # periods.  We can get rid of this when we switch to Twofishes.
+            self.name = re.sub(r'([ACDFGHIKLMNOPRSTUVW])\.?'
+                               r'([KLRZAOTEIDNSYCHJMVXV])\.',
+                               r'\1\2',
+                               self.name)
             return self.name
 
         prop_order = ["city", "county", "state", "country", "region"]

test/test_geocode.py

@@ -590,6 +590,42 @@ def test_geocode_geonames_name_search_context():
     assert resp.status == 200
     assert_same_jsons(EXPECTED, json.loads(content))
 
+@attr(travis_exclude='yes')
+def test_geocode_works_with_dotted_abbreviations():
+    """Resolves something like "Greenville (S.C.)" as well as "SC" """
+    # Note when retrofitting Twofishes later: Twofishes handles "(S.C.)" just
+    # fine, so most of this test's assertion should be kept, but the code that
+    # works around this syntax should be altered.  When we use Twofishes,
+    # we're going to be able to preserve the "S.C." spelling in the "name"
+    # property, and when we do this for Ingestion 3 with MAPv4 we'll be able
+    # to preserve that spelling in the providedLabel property.
+    INPUT =  {
+        "_id": "foo",
+        "sourceResource": {
+            "spatial": {
+                "name": "Greenville (S.C.)"
+            }
+        }
+    }
+    EXPECTED = {
+        "_id": "foo",
+        "sourceResource": {
+            "spatial": [
+                {
+                    "county": "Greenville County",
+                    "country": "United States",
+                    "state": "South Carolina",
+                    "name": "Greenville (SC)",
+                    "coordinates": "34.85262, -82.39401"
+                }
+            ]
+        }
+    }
+    url = server() + "geocode"
+    resp, content = H.request(url, "POST", body=json.dumps(INPUT))
+    assert resp.status == 200
+    assert_same_jsons(EXPECTED, json.loads(content))
+
 @attr(travis_exclude='yes')
 def test_geocode_geonames_name_search_failure():
     """Shouldn't fall down when nothing is returned.