refactor(bedrock): remove dead BedrockLLM invoke path and its tests

The Feb 2025 refactor routed Bedrock Invoke through AmazonInvokeConfig and
base_llm_http_handler, leaving BedrockLLM.completion and its transitive
helpers (process_response, convert_messages_to_prompt, async_completion,
async_streaming, _async_anthropic_messages_completion) unreachable, along
with AmazonAnthropicClaudeConfig.async_transform_request. Delete them and
the unit tests that exercised only that dead path. The class is kept as a
thin shell because get_bedrock_invoke_provider is still referenced through
it, and the live make_call/make_sync_call dispatch and stream decoders stay
put.

litellm/llms/bedrock/chat/invoke_transformations/anthropic_claude3_transformation.py

@@ -154,34 +154,6 @@ class AmazonAnthropicClaudeConfig(AmazonInvokeConfig, AnthropicConfig):
 
         return _anthropic_request
 
-    async def async_transform_request(
-        self,
-        model: str,
-        messages: List[AllMessageValues],
-        optional_params: dict,
-        litellm_params: dict,
-        headers: dict,
-    ) -> dict:
-        _anthropic_request = self._build_bedrock_anthropic_request_base(
-            model=model,
-            messages=messages,
-            optional_params=optional_params,
-            litellm_params=litellm_params,
-            headers=headers,
-        )
-
-        await self._async_convert_document_url_sources_to_base64(_anthropic_request)
-        beta_list = self._compute_bedrock_invoke_beta_headers(
-            model=model,
-            messages=messages,
-            optional_params=optional_params,
-            headers=headers,
-        )
-        if beta_list:
-            _anthropic_request["anthropic_beta"] = beta_list
-
-        return _anthropic_request
-
     def _build_bedrock_anthropic_request_base(
         self,
         model: str,

tests/llm_translation/test_bedrock_completion.py

@@ -3559,89 +3559,6 @@ def test_bedrock_openai_model_id_extraction():
     print(f"✓ Model ID extracted and encoded: {model_id}")
 
 
-def test_bedrock_openai_convert_messages_to_prompt():
-    """
-    Test that convert_messages_to_prompt returns empty string for OpenAI models.
-    """
-    from litellm.llms.bedrock.chat.invoke_handler import BedrockLLM
-
-    bedrock_llm = BedrockLLM()
-    messages = [
-        {"role": "system", "content": "You are helpful"},
-        {"role": "user", "content": "Hello"},
-    ]
-
-    prompt, chat_history = bedrock_llm.convert_messages_to_prompt(
-        model="test-model", messages=messages, provider="openai", custom_prompt_dict={}
-    )
-
-    # OpenAI models use messages directly, no prompt conversion
-    assert prompt == ""
-    assert chat_history is None
-    print("✓ convert_messages_to_prompt returns empty for OpenAI")
-
-
-def test_bedrock_openai_response_parsing():
-    """
-    Test that OpenAI responses are correctly parsed.
-    """
-    from litellm.llms.bedrock.chat.invoke_handler import BedrockLLM
-    from litellm import ModelResponse
-    from unittest.mock import Mock
-    import json
-
-    bedrock_llm = BedrockLLM()
-
-    # Mock OpenAI-style response
-    openai_response = {
-        "choices": [
-            {
-                "message": {
-                    "content": "The capital of France is Paris.",
-                    "role": "assistant",
-                },
-                "finish_reason": "stop",
-                "index": 0,
-            }
-        ],
-        "usage": {"prompt_tokens": 10, "completion_tokens": 8, "total_tokens": 18},
-    }
-
-    mock_response = Mock()
-    mock_response.json.return_value = openai_response
-    mock_response.text = json.dumps(openai_response)
-    mock_response.status_code = 200
-    mock_response.headers = {}
-
-    model_response = ModelResponse()
-    mock_logging = Mock()
-
-    result = bedrock_llm.process_response(
-        model="openai/arn:aws:bedrock:us-east-1:123:imported-model/test",
-        response=mock_response,
-        model_response=model_response,
-        stream=False,
-        logging_obj=mock_logging,
-        optional_params={},
-        api_key="",
-        data={},
-        messages=[{"role": "user", "content": "What is the capital of France?"}],
-        print_verbose=lambda x: None,
-        encoding=None,
-    )
-
-    # Verify response content
-    assert result.choices[0].message.content == "The capital of France is Paris."
-    assert result.choices[0].finish_reason == "stop"
-
-    # Verify usage
-    assert result.usage.prompt_tokens == 10
-    assert result.usage.completion_tokens == 8
-    assert result.usage.total_tokens == 18
-
-    print("✓ OpenAI response parsing works correctly")
-
-
 def test_bedrock_openai_request_transformation():
     """
     Test that the request is correctly transformed for OpenAI models.
@@ -3831,46 +3748,6 @@ def test_bedrock_openai_multiple_message_types():
             print("✓ Multiple message types handled correctly")
 
 
-def test_bedrock_openai_error_handling():
-    """
-    Test that errors from OpenAI models are properly handled.
-    """
-    from litellm.llms.bedrock.chat.invoke_handler import BedrockLLM
-    from litellm import ModelResponse
-    from litellm.llms.bedrock.common_utils import BedrockError
-    from unittest.mock import Mock
-    import json
-
-    bedrock_llm = BedrockLLM()
-
-    # Mock error response
-    mock_response = Mock()
-    mock_response.json.side_effect = Exception("Invalid JSON")
-    mock_response.text = "Invalid response"
-    mock_response.status_code = 422
-
-    model_response = ModelResponse()
-    mock_logging = Mock()
-
-    with pytest.raises(BedrockError) as exc_info:
-        bedrock_llm.process_response(
-            model="openai/arn:aws:bedrock:us-east-1:123:imported-model/test",
-            response=mock_response,
-            model_response=model_response,
-            stream=False,
-            logging_obj=mock_logging,
-            optional_params={},
-            api_key="",
-            data={},
-            messages=[],
-            print_verbose=lambda x: None,
-            encoding=None,
-        )
-
-    assert exc_info.value.status_code == 422
-    print("✓ Error handling works correctly")
-
-
 # ============================================================================
 # Nova Grounding (web_search_options) Unit Tests (Mocked)
 # ============================================================================

tests/test_litellm/llms/bedrock/chat/test_invoke_handler.py

@@ -8,14 +8,11 @@ sys.path.insert(
     0, os.path.abspath("../../../../..")
 )  # Adds the parent directory to the system path
 
-import litellm
 from litellm.llms.bedrock.chat.invoke_handler import (
     AWSEventStreamDecoder,
-    BedrockLLM,
     make_call,
     make_sync_call,
 )
-from litellm.llms.custom_httpx.http_handler import HTTPHandler
 
 
 def test_transform_thinking_blocks_with_redacted_content():
@@ -336,34 +333,3 @@ def test_make_sync_call_guards_against_leaked_control_param():
         )
 
     client.post.assert_not_called()
-
-
-def test_legacy_bedrock_llm_streaming_does_not_rechunk_by_default():
-    mock_response = MagicMock()
-    mock_response.status_code = 200
-    mock_response.iter_bytes = MagicMock(return_value=iter([]))
-    client = HTTPHandler()
-    client.post = MagicMock(return_value=mock_response)
-
-    BedrockLLM().completion(
-        model="cohere.command-text-v14",
-        messages=[{"role": "user", "content": "hi"}],
-        api_base=None,
-        custom_prompt_dict={},
-        model_response=litellm.ModelResponse(),
-        print_verbose=lambda *args, **kwargs: None,
-        encoding=litellm.encoding,
-        logging_obj=MagicMock(),
-        optional_params={
-            "stream": True,
-            "aws_access_key_id": "fake",
-            "aws_secret_access_key": "fake",
-            "aws_region_name": "us-east-1",
-        },
-        acompletion=False,
-        timeout=None,
-        litellm_params={},
-        client=client,
-    )
-
-    mock_response.iter_bytes.assert_called_once_with(chunk_size=None)