Azure OpenAI: fix streaming completions deserialization (Azure#34486)

* Azure OpenAI: fix streaming completions deserialization

* Refresh release date/changelog for retry after yesterday's pipeline infra issues

sdk/openai/Azure.AI.OpenAI/CHANGELOG.md

@@ -1,14 +1,9 @@
 # Release History
 
-## 1.0.0-beta.4 (Unreleased)
+## 1.0.0-beta.4 (2023-02-23)
 
-### Features Added
-
-### Breaking Changes
-
-### Bugs Fixed
-
-### Other Changes
+### Bugs fixed
+- Addressed issues that sometimes caused `beta.3`'s new `GetStreamingCompletions` method to execute indefinitely
 
 ## 1.0.0-beta.3 (2023-02-17)
 

sdk/openai/Azure.AI.OpenAI/src/Custom/CompletionsUsage.Serialization.cs

@@ -0,0 +1,54 @@
+// Copyright (c) Microsoft Corporation. All rights reserved.
+// Licensed under the MIT License.
+
+// <auto-generated/>
+
+#nullable disable
+
+using System.Text.Json;
+
+namespace Azure.AI.OpenAI
+{
+    public partial class CompletionsUsage
+    {
+        internal static CompletionsUsage DeserializeCompletionsUsage(JsonElement element)
+        {
+            if (element.ValueKind == JsonValueKind.Null)
+            {
+                // Handle deserialization of "usage":null in payloads
+                return null;
+            }
+
+            int completionTokens = default;
+            int promptTokens = default;
+            int totalTokens = default;
+            foreach (var property in element.EnumerateObject())
+            {
+                if (property.NameEquals("completion_tokens"u8))
+                {
+                    completionTokens = property.Value.GetInt32();
+                    continue;
+                }
+                if (property.NameEquals("prompt_tokens"u8))
+                {
+                    promptTokens = property.Value.GetInt32();
+                    continue;
+                }
+                if (property.NameEquals("total_tokens"u8))
+                {
+                    totalTokens = property.Value.GetInt32();
+                    continue;
+                }
+            }
+            return new CompletionsUsage(completionTokens, promptTokens, totalTokens);
+        }
+
+        /// <summary> Deserializes the model from a raw response. </summary>
+        /// <param name="response"> The response to deserialize the model from. </param>
+        internal static CompletionsUsage FromResponse(Response response)
+        {
+            using var document = JsonDocument.Parse(response.Content);
+            return DeserializeCompletionsUsage(document.RootElement);
+        }
+    }
+}

sdk/openai/Azure.AI.OpenAI/src/Custom/StreamingCompletions.cs

@@ -118,7 +118,7 @@ public async IAsyncEnumerable<StreamingChoice> GetChoicesStreaming(
                     lock (_streamingChoicesLock)
                     {
                         doneWaiting = _streamingTaskComplete || i < _streamingChoices.Count;
-                        isFinalIndex = _streamingTaskComplete && i == _streamingChoices.Count - 1;
+                        isFinalIndex = _streamingTaskComplete && i >= _streamingChoices.Count - 1;
                     }
 
                     if (!doneWaiting)