[formrecognizer] Code updates (Azure#25825)

* fix value checks in dac

* update test

* remove repeat import

* fix url stream test, remove recordings

* update test assertion, add code comments for future maintenance

Author: catalinaperalta

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

@@ -105,18 +105,28 @@ def begin_analyze_document(
                 :caption: Analyze a custom document. For more samples see the `samples` folder.
         """
 
-        if not model_id:
-            raise ValueError("model_id cannot be None or empty.")
-
         cls = kwargs.pop("cls", self._analyze_document_callback)
         continuation_token = kwargs.pop("continuation_token", None)
 
+        if continuation_token is not None:
+            return self._client.begin_analyze_document(  # type: ignore
+                model_id=model_id,
+                analyze_request=document,  # type: ignore
+                content_type="application/octet-stream",
+                string_index_type="unicodeCodePoint",
+                continuation_token=continuation_token,
+                cls=cls,
+                **kwargs
+            )
+
+        if not model_id:
+            raise ValueError("model_id cannot be None or empty.")
+
         return self._client.begin_analyze_document(  # type: ignore
             model_id=model_id,
             analyze_request=document,  # type: ignore
             content_type="application/octet-stream",
             string_index_type="unicodeCodePoint",
-            continuation_token=continuation_token,
             cls=cls,
             **kwargs
         )
@@ -154,22 +164,33 @@ def begin_analyze_document_from_url(
                 :caption: Analyze a receipt. For more samples see the `samples` folder.
         """
 
+        cls = kwargs.pop("cls", self._analyze_document_callback)
+        continuation_token = kwargs.pop("continuation_token", None)
+
+        # continuation token requests do not perform the same value checks as
+        # regular analysis requests
+        if continuation_token is not None:
+            return self._client.begin_analyze_document(  # type: ignore
+            model_id=model_id,
+            analyze_request={"urlSource": document_url},  # type: ignore
+            string_index_type="unicodeCodePoint",
+            continuation_token=continuation_token,
+            cls=cls,
+            **kwargs
+        )
+
         if not model_id:
             raise ValueError("model_id cannot be None or empty.")
 
         if not isinstance(document_url, str):
             raise ValueError(
-                "'document_url' needs to be of type 'str'."
+                "'document_url' needs to be of type 'str'. "
                 "Please see `begin_analyze_document()` to pass a byte stream.")
 
-        cls = kwargs.pop("cls", self._analyze_document_callback)
-        continuation_token = kwargs.pop("continuation_token", None)
-
         return self._client.begin_analyze_document(  # type: ignore
             model_id=model_id,
             analyze_request={"urlSource": document_url},  # type: ignore
             string_index_type="unicodeCodePoint",
-            continuation_token=continuation_token,
             cls=cls,
             **kwargs
         )

sdk/formrecognizer/azure-ai-formrecognizer/azure/ai/formrecognizer/aio/_document_analysis_client_async.py

@@ -108,18 +108,28 @@ async def begin_analyze_document(
                 :caption: Analyze a custom document. For more samples see the `samples` folder.
         """
 
-        if not model_id:
-            raise ValueError("model_id cannot be None or empty.")
-
         cls = kwargs.pop("cls", self._analyze_document_callback)
         continuation_token = kwargs.pop("continuation_token", None)
 
+        if continuation_token is not None:
+            return await self._client.begin_analyze_document(  # type: ignore
+                model_id=model_id,
+                analyze_request=document,  # type: ignore
+                content_type="application/octet-stream",
+                string_index_type="unicodeCodePoint",
+                continuation_token=continuation_token,
+                cls=cls,
+                **kwargs
+            )
+
+        if not model_id:
+            raise ValueError("model_id cannot be None or empty.")
+
         return await self._client.begin_analyze_document(  # type: ignore
             model_id=model_id,
             analyze_request=document,  # type: ignore
             content_type="application/octet-stream",
             string_index_type="unicodeCodePoint",
-            continuation_token=continuation_token,
             cls=cls,
             **kwargs
         )
@@ -157,17 +167,32 @@ async def begin_analyze_document_from_url(
                 :caption: Analyze a receipt. For more samples see the `samples` folder.
         """
 
+        cls = kwargs.pop("cls", self._analyze_document_callback)
+        continuation_token = kwargs.pop("continuation_token", None)
+
+        # continuation token requests do not perform the same value checks as
+        # regular analysis requests
+        if continuation_token is not None:
+            return await self._client.begin_analyze_document(  # type: ignore
+                model_id=model_id,
+                analyze_request={"urlSource": document_url},  # type: ignore
+                string_index_type="unicodeCodePoint",
+                continuation_token=continuation_token,
+                cls=cls,
+                **kwargs
+            )
         if not model_id:
             raise ValueError("model_id cannot be None or empty.")
 
-        cls = kwargs.pop("cls", self._analyze_document_callback)
-        continuation_token = kwargs.pop("continuation_token", None)
+        if not isinstance(document_url, str):
+            raise ValueError(
+                "'document_url' needs to be of type 'str'. "
+                "Please see `begin_analyze_document()` to pass a byte stream.")
 
         return await self._client.begin_analyze_document(  # type: ignore
             model_id=model_id,
             analyze_request={"urlSource": document_url},  # type: ignore
             string_index_type="unicodeCodePoint",
-            continuation_token=continuation_token,
             cls=cls,
             **kwargs
         )

sdk/formrecognizer/azure-ai-formrecognizer/tests/recordings/test_dac_analyze_prebuilts_from_url.pyTestDACAnalyzePrebuiltsFromUrltest_receipt_url_pass_stream.json

@@ -11,7 +11,6 @@
 from azure.core.credentials import AzureKeyCredential
 from azure.ai.formrecognizer._generated.v2022_08_31.models import AnalyzeResultOperation
 from azure.ai.formrecognizer.aio import DocumentAnalysisClient, DocumentModelAdministrationClient
-from azure.ai.formrecognizer._generated.v2022_08_31.models import AnalyzeResultOperation
 from azure.ai.formrecognizer import AnalyzeResult
 from preparers import FormRecognizerPreparer
 from asynctestcase import AsyncFormRecognizerTest

sdk/formrecognizer/azure-ai-formrecognizer/tests/recordings/test_dac_analyze_prebuilts_from_url_async.pyTestDACAnalyzePrebuiltsFromUrlAsynctest_receipt_url_pass_stream.json

@@ -218,7 +218,7 @@ def test_receipt_url_pass_stream(self, **kwargs):
         with open(self.receipt_png, "rb") as receipt:
             with pytest.raises(ValueError) as e:
                 poller = client.begin_analyze_document_from_url("prebuilt-receipt", receipt)
-                assert str(e) == "'document_url' needs to be of type 'str'. Please see `begin_analyze_document()` to pass a stream."
+            assert str(e.value) == "'document_url' needs to be of type 'str'. Please see `begin_analyze_document()` to pass a byte stream."
 
     @FormRecognizerPreparer()
     @DocumentAnalysisClientPreparer()
@@ -355,7 +355,7 @@ def test_receipt_continuation_token(self, **kwargs):
 
         initial_poller = client.begin_analyze_document_from_url("prebuilt-receipt", self.receipt_url_jpg)
         cont_token = initial_poller.continuation_token()
-        poller = client.begin_analyze_document_from_url("prebuilt-receipt", None, continuation_token=cont_token)
+        poller = client.begin_analyze_document_from_url(None, None, continuation_token=cont_token)
         result = poller.result()
         assert result is not None
         initial_poller.wait()  # necessary so azure-devtools doesn't throw assertion error

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

@@ -236,16 +236,16 @@ async def test_receipt_bad_url(self, client):
 
     @FormRecognizerPreparer()
     @DocumentAnalysisClientPreparer()
-    @recorded_by_proxy_async
-    async def test_receipt_url_pass_stream(self, client):
+    async def test_receipt_url_pass_stream(self, **kwargs):
+        client = kwargs.get("client", None)
 
         with open(self.receipt_png, "rb") as fd:
             receipt = fd.read(4)  # makes the recording smaller
 
-        with pytest.raises(HttpResponseError):
+        with pytest.raises(ValueError) as e:
             async with client:
                 poller = await client.begin_analyze_document_from_url("prebuilt-receipt", receipt)
-                result = await poller.result()
+        assert str(e.value) == "'document_url' needs to be of type 'str'. Please see `begin_analyze_document()` to pass a byte stream."
 
     @FormRecognizerPreparer()
     @DocumentAnalysisClientPreparer()