[FR] add document classifier to operation details (Azure#29732)

Author: kristapratico

sdk/formrecognizer/azure-ai-formrecognizer/CHANGELOG.md

@@ -13,7 +13,7 @@
   `list_document_classifiers`, `get_document_classifier`, and `delete_document_classifier`.
 - Added support for classifying documents on `DocumentAnalysisClient`: `begin_classify_document` and `begin_classify_document_from_url`.
 - Added model `QuotaDetails` and property `custom_neural_document_model_builds` on `ResourceDetails`.
-- Added kind `documentClassifierBuild` to `OperationSummary`.
+- Added kind `documentClassifierBuild` to `OperationSummary` and `OperationDetails`.
 - Added property `expires_on` to `DocumentModelDetails` and `DocumentModelSummary`.
 - Added kind `formulaBlock` to `DocumentParagraph`.
 - Added property `common_name` to `DocumentKeyValuePair`.

sdk/formrecognizer/azure-ai-formrecognizer/assets.json

@@ -2,5 +2,5 @@
   "AssetsRepo": "Azure/azure-sdk-assets",
   "AssetsRepoPrefixPath": "python",
   "TagPrefix": "python/formrecognizer/azure-ai-formrecognizer",
-  "Tag": "python/formrecognizer/azure-ai-formrecognizer_390e54ecb9"
+  "Tag": "python/formrecognizer/azure-ai-formrecognizer_311c87465f"
 }

sdk/formrecognizer/azure-ai-formrecognizer/azure/ai/formrecognizer/_models.py

@@ -10,7 +10,11 @@
 from enum import Enum
 from collections import namedtuple
 from azure.core import CaseInsensitiveEnumMeta
-from ._generated.v2023_02_28_preview.models import DocumentModelDetails as ModelDetails, Error
+from ._generated.v2023_02_28_preview.models import (
+    DocumentModelDetails as ModelDetails,
+    DocumentClassifierDetails as ClassifierDetails,
+    Error
+)
 from ._generated.models import ClassifierDocumentTypeDetails
 from ._helpers import (
     adjust_value_type,
@@ -3700,7 +3704,8 @@ class OperationSummary:
     created, and more.
 
     Note that operation information only persists for 24 hours. If the operation was successful,
-    the model can be accessed using the :func:`~get_document_model` or :func:`~list_document_models` APIs.
+    the model can be accessed using the :func:`~get_document_model`, :func:`~list_document_models`,
+    :func:`~get_document_classifier`, :func:`~list_document_classifiers` APIs.
     To find out why an operation failed, use :func:`~get_operation` and provide the `operation_id`.
 
     .. versionadded:: 2023-02-28-preview
@@ -3800,10 +3805,11 @@ class OperationDetails(OperationSummary):
     error of the operation if it has completed.
 
     Note that operation information only persists for 24 hours. If the operation was successful,
-    the model can also be accessed using the :func:`~get_document_model` or :func:`~list_document_models` APIs.
+    the model can also be accessed using the :func:`~get_document_model`, :func:`~list_document_models`,
+    :func:`~get_document_classifier`, :func:`~list_document_classifiers` APIs.
 
     .. versionadded:: 2023-02-28-preview
-        The `documentClassifierBuild` kind.
+        The `documentClassifierBuild` kind and `DocumentClassifierDetails` result.
     """
     operation_id: str
     """Operation ID."""
@@ -3824,10 +3830,9 @@ class OperationDetails(OperationSummary):
     error: Optional[DocumentAnalysisError]
     """Encountered error, includes the error code, message, and details for why
         the operation failed."""
-    result: Optional[DocumentModelDetails]
-    """Operation result upon success. Returns a DocumentModelDetails which contains
-        all information about the model including the doc types
-        and fields it can analyze from documents."""
+    result: Optional[Union[DocumentModelDetails, DocumentClassifierDetails]]
+    """Operation result upon success. Returns a DocumentModelDetails or DocumentClassifierDetails
+        which contains all the information about the model."""
     api_version: Optional[str]
     """API version used to create this operation."""
     tags: Optional[Dict[str, str]]
@@ -3871,6 +3876,14 @@ def from_dict(cls, data: Dict) -> "OperationDetails":
         :return: OperationDetails
         :rtype: OperationDetails
         """
+
+        kind = data.get("kind", None)
+        if kind == "documentClassifierBuild":
+            result = \
+                DocumentClassifierDetails.from_dict(data.get("result")) if data.get("result") else None  # type: ignore
+        else:
+            result = \
+                DocumentModelDetails.from_dict(data.get("result")) if data.get("result") else None  # type: ignore
         return cls(
             operation_id=data.get("operation_id", None),
             status=data.get("status", None),
@@ -3879,7 +3892,7 @@ def from_dict(cls, data: Dict) -> "OperationDetails":
             last_updated_on=data.get("last_updated_on", None),
             kind=data.get("kind", None),
             resource_location=data.get("resource_location", None),
-            result=DocumentModelDetails.from_dict(data.get("result")) if data.get("result") else None,  # type: ignore
+            result=result,
             error=DocumentAnalysisError.from_dict(data.get("error")) if data.get("error") else None,  # type: ignore
             api_version=data.get("api_version", None),
             tags=data.get("tags", {}),
@@ -3888,6 +3901,12 @@ def from_dict(cls, data: Dict) -> "OperationDetails":
     @classmethod
     def _from_generated(cls, op, api_version):  # pylint: disable=arguments-differ
         deserialize = _get_deserialize(api_version)
+        if op.kind == "documentClassifierBuild":
+            result = DocumentClassifierDetails._from_generated(deserialize(ClassifierDetails, op.result)) \
+                if op.result else None
+        else:
+            result = DocumentModelDetails._from_generated(deserialize(ModelDetails, op.result)) \
+                if op.result else None
         return cls(
             operation_id=op.operation_id,
             status=op.status,
@@ -3896,8 +3915,7 @@ def _from_generated(cls, op, api_version):  # pylint: disable=arguments-differ
             last_updated_on=op.last_updated_date_time,
             kind=op.kind,
             resource_location=op.resource_location,
-            result=DocumentModelDetails._from_generated(deserialize(ModelDetails, op.result))
-            if op.result else None,
+            result=result,
             error=DocumentAnalysisError._from_generated(deserialize(Error, op.error))
             if op.error else None,
             api_version=op.api_version,

sdk/formrecognizer/azure-ai-formrecognizer/samples/v3.2/async_samples/sample_get_operations_async.py

@@ -62,7 +62,10 @@ async def sample_get_operations_async():
                 print("My {} operation is completed.".format(operation_info.kind))
                 result = operation_info.result
                 if result is not None:
-                    print("Model ID: {}".format(result.model_id))
+                    if operation_info.kind == "documentClassifierBuild":
+                        print(f"Classifier ID: {result.classifier_id}")
+                    else:
+                        print("Model ID: {}".format(result.model_id))
             elif operation_info.status == "failed":
                 print("My {} operation failed.".format(operation_info.kind))
                 error = operation_info.error

sdk/formrecognizer/azure-ai-formrecognizer/samples/v3.2/sample_get_operations.py

@@ -58,7 +58,10 @@ def sample_get_operations():
             print("My {} operation is completed.".format(operation_info.kind))
             result = operation_info.result
             if result is not None:
-                print("Model ID: {}".format(result.model_id))
+                if operation_info.kind == "documentClassifierBuild":
+                    print(f"Classifier ID: {result.classifier_id}")
+                else:
+                    print("Model ID: {}".format(result.model_id))
         elif operation_info.status == "failed":
             print("My {} operation failed.".format(operation_info.kind))
             error = operation_info.error

sdk/formrecognizer/azure-ai-formrecognizer/tests/test_dmac_mgmt.py

@@ -157,11 +157,14 @@ def test_get_list_operations(self, client, **kwargs):
             assert op.kind
             assert op.resource_location
             if op.status == "succeeded":
-                successful_op = op
+                if op.kind != "documentClassifierBuild":
+                    successful_op = op
+                else:
+                    successful_classifier_op = op
             if op.status == "failed":
                 failed_op = op
 
-        # check successful op
+        # check successful document model op
         if successful_op:
             op = client.get_operation(successful_op.operation_id)
             # TODO not seeing this returned at the operation level
@@ -182,8 +185,25 @@ def test_get_list_operations(self, client, **kwargs):
                 for key, field in doc_type.field_schema.items():
                     assert key
                     assert field["type"]
+                    if doc_type.build_mode == "neural":
+                        continue  # neural models don't have field confidence
                     assert doc_type.field_confidence[key] is not None
 
+        # check successful classifier model op
+        if successful_classifier_op:
+            op = client.get_operation(successful_classifier_op.operation_id)
+            # test to/from dict
+            op_dict = op.to_dict()
+            op = OperationDetails.from_dict(op_dict)
+            classifier = op.result
+            assert classifier.api_version
+            assert classifier.classifier_id
+            assert classifier.created_on
+            assert classifier.expires_on
+            for doc_type, source in classifier.doc_types.items():
+                assert doc_type
+                assert source.azure_blob_source or source.azure_blob_file_list_source
+
         # check failed op
         if failed_op:
             op = client.get_operation(failed_op.operation_id)

sdk/formrecognizer/azure-ai-formrecognizer/tests/test_dmac_mgmt_async.py

@@ -166,7 +166,10 @@ async def test_get_list_operations(self, client):
                 assert op.kind
                 assert op.resource_location
                 if op.status == "succeeded":
-                    successful_op = op
+                    if op.kind != "documentClassifierBuild":
+                        successful_op = op
+                    else:
+                        successful_classifier_op = op
                 if op.status == "failed":
                     failed_op = op
 
@@ -188,8 +191,25 @@ async def test_get_list_operations(self, client):
                     for key, field in doc_type.field_schema.items():
                         assert key
                         assert field["type"]
+                        if doc_type.build_mode == "neural":
+                            continue  # neural models don't have field confidence
                         assert doc_type.field_confidence[key] is not None
 
+            # check successful classifier model op
+            if successful_classifier_op:
+                op = await client.get_operation(successful_classifier_op.operation_id)
+                # test to/from dict
+                op_dict = op.to_dict()
+                op = OperationDetails.from_dict(op_dict)
+                classifier = op.result
+                assert classifier.api_version
+                assert classifier.classifier_id
+                assert classifier.created_on
+                assert classifier.expires_on
+                for doc_type, source in classifier.doc_types.items():
+                    assert doc_type
+                    assert source.azure_blob_source or source.azure_blob_file_list_source
+
             # check failed op
             if failed_op:
                 op = await client.get_operation(failed_op.operation_id)