fix: Fix missing usage metadata in GoogleAIGeminiChatGenerator (#1195)

* fix: Fix missing usage metadata in GoogleAIGeminiChatGenerator * small fixes + test --------- Co-authored-by: anakin87 <[email protected]>
deepset-ai · Nov 19, 2024 · d00e7d9 · d00e7d9
1 parent 3ed8dfb
commit d00e7d9
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 2 deletions.
diff --git a/...ations/google_ai/src/haystack_integrations/components/generators/google_ai/chat/gemini.py b/...ations/google_ai/src/haystack_integrations/components/generators/google_ai/chat/gemini.py
@@ -313,9 +313,24 @@ def _get_response(self, response_body: GenerateContentResponse) -> List[ChatMess
         """
         replies: List[ChatMessage] = []
         metadata = response_body.to_dict()
+
+        # currently Google only supports one candidate and usage metadata reflects this
+        # this should be refactored when multiple candidates are supported
+        usage_metadata_openai_format = {}
+
+        usage_metadata = metadata.get("usage_metadata")
+        if usage_metadata:
+            usage_metadata_openai_format = {
+                "prompt_tokens": usage_metadata["prompt_token_count"],
+                "completion_tokens": usage_metadata["candidates_token_count"],
+                "total_tokens": usage_metadata["total_token_count"],
+            }
+
         for idx, candidate in enumerate(response_body.candidates):
             candidate_metadata = metadata["candidates"][idx]
             candidate_metadata.pop("content", None)  # we remove content from the metadata
+            if usage_metadata_openai_format:
+                candidate_metadata["usage"] = usage_metadata_openai_format
 
             for part in candidate.content.parts:
                 if part.text != "":

diff --git a/integrations/google_ai/tests/generators/chat/test_chat_gemini.py b/integrations/google_ai/tests/generators/chat/test_chat_gemini.py
@@ -295,5 +295,11 @@ def test_past_conversation():
     ]
     response = gemini_chat.run(messages=messages)
     assert "replies" in response
-    assert len(response["replies"]) > 0
-    assert all(reply.role == ChatRole.ASSISTANT for reply in response["replies"])
+    replies = response["replies"]
+    assert len(replies) > 0
+    assert all(reply.role == ChatRole.ASSISTANT for reply in replies)
+
+    assert all("usage" in reply.meta for reply in replies)
+    assert all("prompt_tokens" in reply.meta["usage"] for reply in replies)
+    assert all("completion_tokens" in reply.meta["usage"] for reply in replies)
+    assert all("total_tokens" in reply.meta["usage"] for reply in replies)