RCB-620: Name and Address match parameter overrides.

Author: seth-mg

examples/address_similarity.py

@@ -18,6 +18,7 @@ def run(key, alt_url='https://api.rosette.com/rest/v1/'):
     params = AddressSimilarityParameters()
     params["address1"] = {"houseNumber": "1600", "road": "Pennsylvania Ave NW", "city": "Washington", "state": "DC", "postCode": "20500"}
     params["address2"] = "160 Pennsilvana Avenue, Washington, D.C., 20500"
+    #params["parameters"] = {"houseNumberAddressFieldWeight": "0.9"}
 
     try:
         return api.address_similarity(params)

examples/name_similarity.py

@@ -20,6 +20,8 @@ def run(key, alt_url='https://api.rosette.com/rest/v1/'):
     params = NameSimilarityParameters()
     params["name1"] = {"text": matched_name_data1, "language": "eng", "entityType": "PERSON"}
     params["name2"] = {"text": matched_name_data2, "entityType": "PERSON"}
+    #params["parameters"] = {"conflictScore": "0.9", "deletionScore": "0.2"}
+
     try:
         return api.name_similarity(params)
     except RosetteException as exception:

rosette/api.py

@@ -250,19 +250,28 @@ def validate(self):
 
 class AddressSimilarityParameters(_DocumentParamSetBase):
     """Parameter object for C{address-similarity} endpoint.
-    All are required.
+
+    C{address1} and C{address2} are required.
+
+    `parameters` is optional.
 
     C{address1} The address to be matched, a C{address} object or address string.
 
     C{address2} The address to be matched, a C{address} object or address string.
 
     The C{address} object contains these optional fields:
       city, island, district, stateDistrict, state, countryRegion, country, worldRegion, postCode, poBox
+
+    `parameters` is a dictionary listing any parameter overrides to include.  For example, `postCodeAddressFieldWeight`.
+    Setting `parameters` is not cumulative.  Define all overrides at once.  If defined multiple times, only the
+    final declaration is used.
+
+    See `examples/address_similarity.py`
     """
 
     def __init__(self):
         self.use_multipart = False
-        _DocumentParamSetBase.__init__(self, ("address1", "address2"))
+        _DocumentParamSetBase.__init__(self, ("address1", "address2", "parameters"))
 
     def validate(self):
         """Internal. Do not use."""
@@ -276,26 +285,35 @@ def validate(self):
 
 class NameSimilarityParameters(_DocumentParamSetBase):
     """Parameter object for C{name-similarity} endpoint.
-    All are required.
+
+    C{name1} and C{name2} are required.
+
+    `parameters` is optional.
 
     C{name1} The name to be matched, a C{name} object.
 
     C{name2} The name to be matched, a C{name} object.
 
     The C{name} object contains these fields:
 
-    C{text} Text of the name, required.
+        C{text} Text of the name, required.
+
+        C{language} Language of the name in ISO639 three-letter code, optional.
+
+        C{script} The ISO15924 code of the name, optional.
 
-    C{language} Language of the name in ISO639 three-letter code, optional.
+        C{entityType} The entity type, can be "PERSON", "LOCATION" or "ORGANIZATION", optional.
 
-    C{script} The ISO15924 code of the name, optional.
+    `parameters` is a dictionary listing any parameter overrides to include.  For example, `deletionScore`.
+    Setting `parameters` is not cumulative.  Define all overrides at once.  If defined multiple times, only the
+    final declaration is used.
 
-    C{entityType} The entity type, can be "PERSON", "LOCATION" or "ORGANIZATION", optional.
+    See `examples/name_similarity.py`
     """
 
     def __init__(self):
         self.use_multipart = False
-        _DocumentParamSetBase.__init__(self, ("name1", "name2"))
+        _DocumentParamSetBase.__init__(self, ("name1", "name2", "parameters"))
 
     def validate(self):
         """Internal. Do not use."""

tests/test_rosette_api.py

@@ -427,6 +427,48 @@ def test_the_name_translation_endpoint(api, json_response):
 # Test the name similarity endpoint
 
 
+def test_the_name_similarity_single_parameters(api, json_response):
+    """Test the name similarity parameters"""
+    httpretty.enable()
+    httpretty.register_uri(httpretty.POST, "https://api.rosette.com/rest/v1/info",
+                           body=json_response, status=200, content_type="application/json")
+    httpretty.register_uri(httpretty.POST, "https://api.rosette.com/rest/v1/name-similarity",
+                           body=json_response, status=200, content_type="application/json")
+
+    matched_name_data1 = "John Mike Smith"
+    matched_name_data2 = "John Joe Smith"
+    params = NameSimilarityParameters()
+    params["name1"] = {"text": matched_name_data1}
+    params["name2"] = {"text": matched_name_data2}
+    params["parameters"] = {"conflictScore": "0.9"}
+
+    result = api.name_similarity(params)
+    assert result["name"] == "Rosette"
+    httpretty.disable()
+    httpretty.reset()
+
+
+def test_the_name_similarity_multiple_parameters(api, json_response):
+    """Test the name similarity parameters"""
+    httpretty.enable()
+    httpretty.register_uri(httpretty.POST, "https://api.rosette.com/rest/v1/info",
+                           body=json_response, status=200, content_type="application/json")
+    httpretty.register_uri(httpretty.POST, "https://api.rosette.com/rest/v1/name-similarity",
+                           body=json_response, status=200, content_type="application/json")
+
+    matched_name_data1 = "John Mike Smith"
+    matched_name_data2 = "John Joe Smith"
+    params = NameSimilarityParameters()
+    params["name1"] = {"text": matched_name_data1}
+    params["name2"] = {"text": matched_name_data2}
+    params["parameters"] = {"conflictScore": "0.9", "deletionScore": "0.5"}
+
+    result = api.name_similarity(params)
+    assert result["name"] == "Rosette"
+    httpretty.disable()
+    httpretty.reset()
+
+
 def test_the_name_similarity_endpoint(api, json_response):
     """Test the name similarity endpoint"""
     httpretty.enable()
@@ -449,10 +491,11 @@ def test_the_name_similarity_endpoint(api, json_response):
     httpretty.disable()
     httpretty.reset()
 
+
 # Test the name deduplication endpoint
 
 
-def test_name_deduplicatation_parameters(api, json_response):
+def test_name_deduplication_parameters(api, json_response):
     """Test the Name Deduplication Parameters"""
     httpretty.enable()
     httpretty.register_uri(httpretty.POST, "https://api.rosette.com/rest/v1/info",
@@ -612,6 +655,32 @@ def test_for_address_similarity_required_parameters(api, json_response):
     httpretty.reset()
 
 
+def test_for_address_similarity_optional_parameters(api, json_response):
+    """Test address similarity parameters"""
+    httpretty.enable()
+    httpretty.register_uri(httpretty.POST, "https://api.rosette.com/rest/v1/info",
+                           body=json_response, status=200, content_type="application/json")
+    httpretty.register_uri(httpretty.POST, "https://api.rosette.com/rest/v1/address-similarity",
+                           body=json_response, status=200, content_type="application/json")
+
+    params = AddressSimilarityParameters()
+
+    params["address1"] = {"houseNumber": "1600",
+                          "road": "Pennsylvania Ave NW",
+                          "city": "Washington",
+                          "state": "DC",
+                          "postCode": "20500"}
+
+    params["address2"] = {"text": "160 Pennsilvana Avenue, Washington, D.C., 20500"}
+
+    params["parameters"] = {"houseNumberAddressFieldWeight": "0.9"}
+
+    result = api.address_similarity(params)
+    assert result["name"] == "Rosette"
+    httpretty.disable()
+    httpretty.reset()
+
+
 # Test for required Name Similarity parameters