Merge pull request #106 from rosette-api/rcb-620-name-and-address-similarity-parameters-overrides

RCB-620:  Name and Address match parameter overrides.

Author: seth-mg

CI.Jenkinsfile

@@ -1,6 +1,6 @@
 
 
-def versions = [3.11, 3.10, 3.9, 3.8, 3.7]
+def versions = [3.7, 3.8, 3.9, 3.10, 3.11]
 
 def runSonnarForPythonVersion(sourceDir, ver){
     mySonarOpts="-Dsonar.sources=/source -Dsonar.host.url=${env.SONAR_HOST_URL} -Dsonar.login=${env.SONAR_AUTH_TOKEN}"
@@ -13,21 +13,27 @@ def runSonnarForPythonVersion(sourceDir, ver){
         mySonarOpts="$mySonarOpts -Dsonar.pullrequest.base=${env.CHANGE_TARGET} -Dsonar.pullrequest.branch=${env.CHANGE_BRANCH}"
     }
 
-    // TODO: find a way to skip the Sonar scan for all those version, but one (maybe the latest?).
+    // Only run Sonar once.
+    if(ver == 3.11) {
+        sonarExec="cd /root/ && \
+                   wget -q https://binaries.sonarsource.com/Distribution/sonar-scanner-cli/sonar-scanner-cli-4.8.0.2856-linux.zip && \
+                   unzip -q sonar-scanner-cli-4.8.0.2856-linux.zip && \
+                   cd /source && \
+                   /root/sonar-scanner-4.8.0.2856-linux/bin/sonar-scanner ${mySonarOpts}"
+    } else {
+        sonarExec="echo Skipping Sonar for this version."
+    }
+
     sh "docker run \
             --pull always \
             --rm --volume ${sourceDir}:/source \
             python:${ver}-slim \
             bash -c \"apt-get update && \
             apt-get install -y wget unzip && \
             pip3 install tox && \
-            cd /root/ && \
-            wget -q https://binaries.sonarsource.com/Distribution/sonar-scanner-cli/sonar-scanner-cli-4.8.0.2856-linux.zip && \
-            unzip -q sonar-scanner-cli-4.8.0.2856-linux.zip && \
             cd /source && \
             tox && \
-            /root/sonar-scanner-4.8.0.2856-linux/bin/sonar-scanner \
-            ${mySonarOpts}\""
+            ${sonarExec}\""
 }
 
 node ("docker-light") {

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

@@ -113,7 +113,7 @@ def serialize(self, options):
         """serialize keys with values"""
         self.validate()
         values = {}
-        for (key, val) in self.__params.items():
+        for key, val in self.__params.items():
             if val is None:
                 continue
             else:
@@ -212,7 +212,7 @@ class NameTranslationParameters(_DocumentParamSetBase):
 
     C{targetLangauge} The language into which the name is to be translated.
 
-    C{entityType} The entity type (TBD) of the name.
+    C{entityType} The entity type of the name.  PERSON (default), LOCATION, or ORGANIZATION
 
     C{sourceLanguageOfOrigin} The language of origin of the name.
 
@@ -240,7 +240,7 @@ def __init__(self):
 
     def validate(self):
         """Internal. Do not use."""
-        for option in ("name", "targetLanguage"):  # required
+        for option in "name", "targetLanguage":  # required
             if self[option] is None:
                 raise RosetteException(
                     "missingParameter",
@@ -250,23 +250,32 @@ 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."""
-        for option in ("address1", "address2"):  # required
+        for option in "address1", "address2":  # required
             if self[option] is None:
                 raise RosetteException(
                     "missingParameter",
@@ -276,30 +285,39 @@ 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."""
-        for option in ("name1", "name2"):  # required
+        for option in "name1", "name2":  # required
             if self[option] is None:
                 raise RosetteException(
                     "missingParameter",

sonar-project.properties

@@ -1,4 +1,5 @@
 sonar.projectKey=rosette-api-python-binding
 sonar.sources=rosette
+sonar.exclusions=**/tests/**,**/docs/**,**/examples/**
 sonar.python.coverage.reportPaths=coverage.xml
 #sonar.branch.name=RCB-596-pool-size

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",
@@ -546,7 +589,7 @@ def test_for_content_and_contentUri(api, json_response, doc_params):
     httpretty.register_uri(httpretty.POST, "https://api.rosette.com/rest/v1/entities",
                            body=json_response, status=200, content_type="application/json")
 
-    doc_params['contentUri'] = 'http://google.com'
+    doc_params['contentUri'] = 'https://example.com'
     with pytest.raises(RosetteException) as e_rosette:
         api.entities(doc_params)
 
@@ -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
 
 

tox.ini

@@ -1,19 +1,21 @@
-# Tox (http://tox.testrun.org/) is a tool for running tests
-# in multiple virtualenvs. This configuration file will run the
-# test suite on all supported python versions. To use it, "pip install tox"
-# and then run "tox" from this directory.
-
 [tox]
-skipsdist = True
 envlist = py3
+skipsdist = True
 
 [testenv]
-commands =
-    {envpython} setup.py install
-    {envbindir}/py.test
 deps =
     pytest
     pep8
     httpretty
     epydoc
     requests
+    coverage
+    build
+
+commands =
+    python -m build
+    coverage run -m pytest
+    coverage xml
+
+[coverage:run]
+relative_files = True