[Search] Add skillset validation (Azure#20669)

* Add client-side validation and test.

* Add skillset validation test.

* Make code more Pythonic.

Author: tjprescott

sdk/search/azure-search-documents/CHANGELOG.md

@@ -11,6 +11,7 @@
 
 - Renamed `SearchClient.speller` to `SearchClient.query_speller`.
 - Removed keyword arguments from `SearchClient`: `answers` and `captions`.
+- `SentimentSkill`, `EntityRecognitionSkill`: added client-side validation to prevent sending unsupported parameters.
 
 ### Bugs Fixed
 

sdk/search/azure-search-documents/azure/search/documents/indexes/_search_indexer_client.py

@@ -15,7 +15,11 @@
     get_access_conditions,
     normalize_endpoint,
 )
-from .models import SearchIndexerDataSourceConnection
+from .models import (
+    EntityRecognitionSkillVersion,
+    SearchIndexerDataSourceConnection,
+    SentimentSkillVersion
+)
 from .._api_versions import DEFAULT_VERSION
 from .._headers_mixin import HeadersMixin
 from .._utils import get_authentication_policy
@@ -564,7 +568,9 @@ def create_skillset(self, skillset, **kwargs):
 
         """
         kwargs["headers"] = self._merge_client_headers(kwargs.get("headers"))
+        _validate_skillset(skillset)
         skillset = skillset._to_generated() if hasattr(skillset, '_to_generated') else skillset # pylint:disable=protected-access
+
         result = self._client.skillsets.create(skillset, **kwargs)
         return SearchIndexerSkillset._from_generated(result) # pylint:disable=protected-access
 
@@ -587,6 +593,7 @@ def create_or_update_skillset(self, skillset, **kwargs):
             skillset, kwargs.pop("match_condition", MatchConditions.Unconditionally)
         )
         kwargs.update(access_condition)
+        _validate_skillset(skillset)
         skillset = skillset._to_generated() if hasattr(skillset, '_to_generated') else skillset # pylint:disable=protected-access
 
         result = self._client.skillsets.create_or_update(
@@ -596,3 +603,44 @@ def create_or_update_skillset(self, skillset, **kwargs):
             **kwargs
         )
         return SearchIndexerSkillset._from_generated(result) # pylint:disable=protected-access
+
+def _validate_skillset(skillset):
+    """Validates any multi-version skills in the skillset to verify that unsupported
+    parameters are not supplied by the user.
+    """
+    skills = getattr(skillset, 'skills', None)
+    if not skills:
+        return
+
+    error_strings = []
+    for skill in skills:
+        try:
+            skill_version = skill.get('skill_version')
+        except AttributeError:
+            skill_version = getattr(skill, 'skill_version', None)
+        if not skill_version:
+            continue
+
+        if skill_version == SentimentSkillVersion.V1:
+            unsupported = ['model_version', 'include_opinion_mining']
+        elif skill_version == SentimentSkillVersion.V3:
+            unsupported = []
+        elif skill_version == EntityRecognitionSkillVersion.V1:
+            unsupported = ['model_version']
+        elif skill_version == EntityRecognitionSkillVersion.V3:
+            unsupported = ['include_typeless_entities']
+
+        errors = []
+        for item in unsupported:
+            try:
+                if skill.get(item, None):
+                    errors.append(item)
+            except AttributeError:
+                if skill.__dict__.get(item, None):
+                    errors.append(item)
+        if errors:
+            error_strings.append("Unsupported parameters for skill version {}: {}".format(
+                skill_version, ", ".join(errors))
+            )
+    if error_strings:
+        raise ValueError("\n".join(error_strings))

sdk/search/azure-search-documents/azure/search/documents/indexes/models/_models.py

@@ -181,6 +181,7 @@ class EntityRecognitionSkill(SearchIndexerSkill):
         'include_typeless_entities': {'key': 'includeTypelessEntities', 'type': 'bool'},
         'minimum_precision': {'key': 'minimumPrecision', 'type': 'float'},
         'model_version': {'key': 'modelVersion', 'type': 'str'},
+        'skill_version': {'key': 'skillVersion', 'type': 'str'}
     }
 
     def __init__(
@@ -210,8 +211,7 @@ def _to_generated(self):
                 categories=self.categories,
                 default_language_code=self.default_language_code,
                 include_typeless_entities=self.include_typeless_entities,
-                minimum_precision=self.minimum_precision,
-                model_version=self.model_version
+                minimum_precision=self.minimum_precision
             )
         if self.skill_version in [EntityRecognitionSkillVersion.V3, EntityRecognitionSkillVersion.LATEST]:
             return _EntityRecognitionSkillV3(
@@ -221,7 +221,6 @@ def _to_generated(self):
                 odata_type=self.odata_type,
                 categories=self.categories,
                 default_language_code=self.default_language_code,
-                include_typeless_entities=self.include_typeless_entities,
                 minimum_precision=self.minimum_precision,
                 model_version=self.model_version
             )
@@ -318,6 +317,7 @@ class SentimentSkill(SearchIndexerSkill):
         'default_language_code': {'key': 'defaultLanguageCode', 'type': 'str'},
         'include_opinion_mining': {'key': 'includeOpinionMining', 'type': 'bool'},
         'model_version': {'key': 'modelVersion', 'type': 'str'},
+        'skill_version': {'key': 'skillVersion', 'type': 'str'}
     }
 
     def __init__(
@@ -331,7 +331,7 @@ def __init__(
         self.skill_version = skill_version
         self.odata_type = self.skill_version  # type: str
         self.default_language_code = kwargs.get('default_language_code', None)
-        self.include_opinion_mining = kwargs.get('include_opinion_mining', False)
+        self.include_opinion_mining = kwargs.get('include_opinion_mining', None)
         self.model_version = kwargs.get('model_version', None)
 
     def _to_generated(self):
@@ -341,10 +341,8 @@ def _to_generated(self):
                 outputs=self.outputs,
                 name=self.name,
                 odata_type=self.odata_type,
-                default_language_code=self.default_language_code,
-                include_opinion_mining=self.include_opinion_mining,
-                model_version=self.model_version
-            )
+                default_language_code=self.default_language_code
+        )
         if self.skill_version in [SentimentSkillVersion.V3, SentimentSkillVersion.LATEST]:
             return _SentimentSkillV3(
                 inputs=self.inputs,

sdk/search/azure-search-documents/tests/recordings/test_search_index_client_skillset_live.test_create_or_update_skillset.yaml

@@ -18,12 +18,12 @@ interactions:
       Prefer:
       - return=representation
       User-Agent:
-      - azsdk-python-search-documents/11.3.0b3 Python/3.9.2 (Windows-10-10.0.19041-SP0)
+      - azsdk-python-search-documents/11.3.0b4 Python/3.9.2 (Windows-10-10.0.19041-SP0)
     method: PUT
     uri: https://searche2bf1c71.search.windows.net/skillsets('test-ss')?api-version=2021-04-30-Preview
   response:
     body:
-      string: '{"@odata.context":"https://searche2bf1c71.search.windows.net/$metadata#skillsets/$entity","@odata.etag":"\"0x8D96E6871C89346\"","name":"test-ss","description":"desc1","skills":[{"@odata.type":"#Microsoft.Skills.Text.EntityRecognitionSkill","name":null,"description":null,"context":null,"categories":[],"defaultLanguageCode":null,"minimumPrecision":null,"includeTypelessEntities":null,"inputs":[{"name":"text","source":"/document/content","sourceContext":null,"inputs":[]}],"outputs":[{"name":"organizations","targetName":"organizations"}]}],"cognitiveServices":null,"knowledgeStore":null,"encryptionKey":null}'
+      string: '{"@odata.context":"https://searche2bf1c71.search.windows.net/$metadata#skillsets/$entity","@odata.etag":"\"0x8D977BF242A141A\"","name":"test-ss","description":"desc1","skills":[{"@odata.type":"#Microsoft.Skills.Text.EntityRecognitionSkill","name":null,"description":null,"context":null,"categories":[],"defaultLanguageCode":null,"minimumPrecision":null,"includeTypelessEntities":null,"inputs":[{"name":"text","source":"/document/content","sourceContext":null,"inputs":[]}],"outputs":[{"name":"organizations","targetName":"organizations"}]}],"cognitiveServices":null,"knowledgeStore":null,"encryptionKey":null}'
     headers:
       cache-control:
       - no-cache
@@ -32,11 +32,11 @@ interactions:
       content-type:
       - application/json; odata.metadata=minimal
       date:
-      - Thu, 02 Sep 2021 23:21:51 GMT
+      - Tue, 14 Sep 2021 20:35:07 GMT
       elapsed-time:
-      - '1433'
+      - '2124'
       etag:
-      - W/"0x8D96E6871C89346"
+      - W/"0x8D977BF242A141A"
       expires:
       - '-1'
       location:
@@ -48,7 +48,7 @@ interactions:
       preference-applied:
       - odata.include-annotations="*"
       request-id:
-      - 8ded272a-0c44-11ec-ac43-74c63bed1137
+      - 3ff59952-159b-11ec-bda5-74c63bed1137
       strict-transport-security:
       - max-age=15724800; includeSubDomains
     status:
@@ -73,12 +73,12 @@ interactions:
       Prefer:
       - return=representation
       User-Agent:
-      - azsdk-python-search-documents/11.3.0b3 Python/3.9.2 (Windows-10-10.0.19041-SP0)
+      - azsdk-python-search-documents/11.3.0b4 Python/3.9.2 (Windows-10-10.0.19041-SP0)
     method: PUT
     uri: https://searche2bf1c71.search.windows.net/skillsets('test-ss')?api-version=2021-04-30-Preview
   response:
     body:
-      string: '{"@odata.context":"https://searche2bf1c71.search.windows.net/$metadata#skillsets/$entity","@odata.etag":"\"0x8D96E6871DB3571\"","name":"test-ss","description":"desc2","skills":[{"@odata.type":"#Microsoft.Skills.Text.EntityRecognitionSkill","name":null,"description":null,"context":null,"categories":[],"defaultLanguageCode":null,"minimumPrecision":null,"includeTypelessEntities":null,"inputs":[{"name":"text","source":"/document/content","sourceContext":null,"inputs":[]}],"outputs":[{"name":"organizations","targetName":"organizations"}]}],"cognitiveServices":null,"knowledgeStore":null,"encryptionKey":null}'
+      string: '{"@odata.context":"https://searche2bf1c71.search.windows.net/$metadata#skillsets/$entity","@odata.etag":"\"0x8D977BF243E8A49\"","name":"test-ss","description":"desc2","skills":[{"@odata.type":"#Microsoft.Skills.Text.EntityRecognitionSkill","name":null,"description":null,"context":null,"categories":[],"defaultLanguageCode":null,"minimumPrecision":null,"includeTypelessEntities":null,"inputs":[{"name":"text","source":"/document/content","sourceContext":null,"inputs":[]}],"outputs":[{"name":"organizations","targetName":"organizations"}]}],"cognitiveServices":null,"knowledgeStore":null,"encryptionKey":null}'
     headers:
       cache-control:
       - no-cache
@@ -87,11 +87,11 @@ interactions:
       content-type:
       - application/json; odata.metadata=minimal
       date:
-      - Thu, 02 Sep 2021 23:21:51 GMT
+      - Tue, 14 Sep 2021 20:35:07 GMT
       elapsed-time:
-      - '46'
+      - '85'
       etag:
-      - W/"0x8D96E6871DB3571"
+      - W/"0x8D977BF243E8A49"
       expires:
       - '-1'
       odata-version:
@@ -101,7 +101,7 @@ interactions:
       preference-applied:
       - odata.include-annotations="*"
       request-id:
-      - 8ef2919c-0c44-11ec-a88c-74c63bed1137
+      - 41558274-159b-11ec-bc0f-74c63bed1137
       strict-transport-security:
       - max-age=15724800; includeSubDomains
       vary:
@@ -119,12 +119,12 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - azsdk-python-search-documents/11.3.0b3 Python/3.9.2 (Windows-10-10.0.19041-SP0)
+      - azsdk-python-search-documents/11.3.0b4 Python/3.9.2 (Windows-10-10.0.19041-SP0)
     method: GET
     uri: https://searche2bf1c71.search.windows.net/skillsets?api-version=2021-04-30-Preview
   response:
     body:
-      string: '{"@odata.context":"https://searche2bf1c71.search.windows.net/$metadata#skillsets","value":[{"@odata.etag":"\"0x8D96E6871DB3571\"","name":"test-ss","description":"desc2","skills":[{"@odata.type":"#Microsoft.Skills.Text.EntityRecognitionSkill","name":"#1","description":null,"context":"/document","categories":["Person","Quantity","Organization","URL","Email","Location","DateTime"],"defaultLanguageCode":"en","minimumPrecision":null,"includeTypelessEntities":null,"inputs":[{"name":"text","source":"/document/content","sourceContext":null,"inputs":[]}],"outputs":[{"name":"organizations","targetName":"organizations"}]}],"cognitiveServices":null,"knowledgeStore":null,"encryptionKey":null}]}'
+      string: '{"@odata.context":"https://searche2bf1c71.search.windows.net/$metadata#skillsets","value":[{"@odata.etag":"\"0x8D977BF243E8A49\"","name":"test-ss","description":"desc2","skills":[{"@odata.type":"#Microsoft.Skills.Text.EntityRecognitionSkill","name":"#1","description":null,"context":"/document","categories":["Person","Quantity","Organization","URL","Email","Location","DateTime"],"defaultLanguageCode":"en","minimumPrecision":null,"includeTypelessEntities":null,"inputs":[{"name":"text","source":"/document/content","sourceContext":null,"inputs":[]}],"outputs":[{"name":"organizations","targetName":"organizations"}]}],"cognitiveServices":null,"knowledgeStore":null,"encryptionKey":null}]}'
     headers:
       cache-control:
       - no-cache
@@ -133,9 +133,9 @@ interactions:
       content-type:
       - application/json; odata.metadata=minimal
       date:
-      - Thu, 02 Sep 2021 23:21:51 GMT
+      - Tue, 14 Sep 2021 20:35:07 GMT
       elapsed-time:
-      - '139'
+      - '69'
       expires:
       - '-1'
       odata-version:
@@ -145,7 +145,7 @@ interactions:
       preference-applied:
       - odata.include-annotations="*"
       request-id:
-      - 8f03c3cb-0c44-11ec-94e8-74c63bed1137
+      - 4169c877-159b-11ec-9f5a-74c63bed1137
       strict-transport-security:
       - max-age=15724800; includeSubDomains
       vary:
@@ -163,12 +163,12 @@ interactions:
       Connection:
       - keep-alive
       User-Agent:
-      - azsdk-python-search-documents/11.3.0b3 Python/3.9.2 (Windows-10-10.0.19041-SP0)
+      - azsdk-python-search-documents/11.3.0b4 Python/3.9.2 (Windows-10-10.0.19041-SP0)
     method: GET
     uri: https://searche2bf1c71.search.windows.net/skillsets('test-ss')?api-version=2021-04-30-Preview
   response:
     body:
-      string: '{"@odata.context":"https://searche2bf1c71.search.windows.net/$metadata#skillsets/$entity","@odata.etag":"\"0x8D96E6871DB3571\"","name":"test-ss","description":"desc2","skills":[{"@odata.type":"#Microsoft.Skills.Text.EntityRecognitionSkill","name":"#1","description":null,"context":"/document","categories":["Person","Quantity","Organization","URL","Email","Location","DateTime"],"defaultLanguageCode":"en","minimumPrecision":null,"includeTypelessEntities":null,"inputs":[{"name":"text","source":"/document/content","sourceContext":null,"inputs":[]}],"outputs":[{"name":"organizations","targetName":"organizations"}]}],"cognitiveServices":null,"knowledgeStore":null,"encryptionKey":null}'
+      string: '{"@odata.context":"https://searche2bf1c71.search.windows.net/$metadata#skillsets/$entity","@odata.etag":"\"0x8D977BF243E8A49\"","name":"test-ss","description":"desc2","skills":[{"@odata.type":"#Microsoft.Skills.Text.EntityRecognitionSkill","name":"#1","description":null,"context":"/document","categories":["Person","Quantity","Organization","URL","Email","Location","DateTime"],"defaultLanguageCode":"en","minimumPrecision":null,"includeTypelessEntities":null,"inputs":[{"name":"text","source":"/document/content","sourceContext":null,"inputs":[]}],"outputs":[{"name":"organizations","targetName":"organizations"}]}],"cognitiveServices":null,"knowledgeStore":null,"encryptionKey":null}'
     headers:
       cache-control:
       - no-cache
@@ -177,11 +177,11 @@ interactions:
       content-type:
       - application/json; odata.metadata=minimal
       date:
-      - Thu, 02 Sep 2021 23:21:51 GMT
+      - Tue, 14 Sep 2021 20:35:07 GMT
       elapsed-time:
-      - '13'
+      - '15'
       etag:
-      - W/"0x8D96E6871DB3571"
+      - W/"0x8D977BF243E8A49"
       expires:
       - '-1'
       odata-version:
@@ -191,7 +191,7 @@ interactions:
       preference-applied:
       - odata.include-annotations="*"
       request-id:
-      - 8f22fc82-0c44-11ec-8c37-74c63bed1137
+      - 417b5016-159b-11ec-acf0-74c63bed1137
       strict-transport-security:
       - max-age=15724800; includeSubDomains
       vary: