ccproxy-api 0.1.7__py3-none-any.whl → 0.2.0__py3-none-any.whl

ccproxy/llms/formatters/anthropic_to_openai/streams.py

@@ -0,0 +1,1546 @@
+"""Anthropic→OpenAI streaming conversion entry points."""
+
+from __future__ import annotations
+
+import json
+import time
+from collections.abc import AsyncGenerator, AsyncIterator
+from typing import Any, Literal, cast
+
+from pydantic import ValidationError
+
+import ccproxy.core.logging
+from ccproxy.llms.formatters.common import (
+    IndexedToolCallTracker,
+    ObfuscationTokenFactory,
+    ToolCallState,
+    ensure_identifier,
+)
+from ccproxy.llms.formatters.constants import ANTHROPIC_TO_OPENAI_FINISH_REASON
+from ccproxy.llms.formatters.context import (
+    get_last_instructions,
+    get_last_request,
+    register_request,
+)
+from ccproxy.llms.formatters.utils import anthropic_usage_snapshot
+from ccproxy.llms.models import anthropic as anthropic_models
+from ccproxy.llms.models import openai as openai_models
+from ccproxy.llms.streaming.accumulators import ClaudeAccumulator
+
+from .requests import _build_responses_payload_from_anthropic_request
+from .responses import (
+    convert__anthropic_usage_to_openai_responses__usage,
+)
+
+
+logger = ccproxy.core.logging.get_logger(__name__)
+
+FinishReason = Literal["stop", "length", "tool_calls"]
+
+
+def _normalize_anthropic_stream_event(
+    event: Any,
+) -> tuple[str | None, dict[str, Any]]:
+    """Return a (type, payload) tuple for mixed dict/model stream events."""
+
+    if isinstance(event, dict):
+        event_type = event.get("type") or event.get("event")
+        return (cast(str | None, event_type), event)
+
+    event_type = getattr(event, "type", None)
+    if event_type is None:
+        return None, {}
+
+    if hasattr(event, "model_dump"):
+        payload = cast(dict[str, Any], event.model_dump(mode="json"))
+    elif hasattr(event, "dict"):
+        payload = cast(dict[str, Any], event.dict())
+    else:
+        payload = {}
+
+    return cast(str | None, event_type), payload
+
+
+def _anthropic_delta_to_text(
+    accumulator: ClaudeAccumulator,
+    block_index: int,
+    delta: dict[str, Any] | None,
+) -> str | None:
+    if not isinstance(delta, dict):
+        return None
+
+    block_info = accumulator.get_block_info(block_index)
+    block_meta = block_info[1] if block_info else {}
+    block_type = block_meta.get("type")
+
+    if block_type == "thinking":
+        thinking_text = delta.get("thinking")
+        if not isinstance(thinking_text, str) or not thinking_text:
+            return None
+        signature = block_meta.get("signature")
+        if isinstance(signature, str) and signature:
+            return f'<thinking signature="{signature}">{thinking_text}</thinking>'
+        return f"<thinking>{thinking_text}</thinking>"
+
+    text_val = delta.get("text")
+    if isinstance(text_val, str) and text_val:
+        return text_val
+
+    return None
+
+
+def _build_openai_tool_call(
+    accumulator: ClaudeAccumulator,
+    block_index: int,
+) -> openai_models.ToolCall | None:
+    for tool_call in accumulator.get_complete_tool_calls():
+        if tool_call.get("index") != block_index:
+            continue
+
+        function_payload = (
+            tool_call.get("function", {}) if isinstance(tool_call, dict) else {}
+        )
+        name = function_payload.get("name") or tool_call.get("name") or "function"
+        arguments = function_payload.get("arguments")
+        if not isinstance(arguments, str) or not arguments:
+            try:
+                arguments = json.dumps(tool_call.get("input", {}), ensure_ascii=False)
+            except Exception:
+                arguments = json.dumps(tool_call.get("input", {}))
+
+        tool_id = tool_call.get("id") or f"call_{block_index}"
+
+        return openai_models.ToolCall(
+            id=str(tool_id),
+            function=openai_models.FunctionCall(
+                name=str(name),
+                arguments=str(arguments),
+            ),
+        )
+
+    return None
+
+
+class AnthropicToOpenAIResponsesStreamAdapter:
+    """Stateful adapter for Anthropic → OpenAI Responses streaming."""
+
+    async def run(
+        self,
+        stream: AsyncIterator[anthropic_models.MessageStreamEvent],
+    ) -> AsyncGenerator[openai_models.StreamEventType, None]:
+        async for event in self._convert_responses_stream(stream):
+            yield event
+
+    async def _convert_responses_stream(
+        self,
+        stream: AsyncIterator[anthropic_models.MessageStreamEvent],
+    ) -> AsyncGenerator[openai_models.StreamEventType, None]:
+        """Convert Anthropic MessageStreamEvents into OpenAI Responses stream events."""
+
+        accumulator = ClaudeAccumulator()
+        sequence_counter = -1
+        model_id = ""
+        response_id = ""
+        id_suffix: str | None = None
+        message_item_id = ""
+        message_output_index: int | None = None
+        next_output_index = 0
+        content_index = 0
+        message_item_added = False
+        message_content_part_added = False
+        text_buffer: list[str] = []
+        message_last_logprobs: Any | None = None
+        message_text_done_emitted = False
+        message_part_done_emitted = False
+        message_item_done_emitted = False
+        message_completed_entry: tuple[int, openai_models.MessageOutput] | None = None
+        latest_usage_model: openai_models.ResponseUsage | None = None
+        final_stop_reason: str | None = None
+        stream_completed = False
+
+        reasoning_item_id = ""
+        reasoning_output_index: int | None = None
+        reasoning_item_added = False
+        reasoning_output_done = False
+        reasoning_summary_indices: dict[str, int] = {}
+        reasoning_summary_added: set[int] = set()
+        reasoning_summary_text_fragments: dict[int, list[str]] = {}
+        reasoning_summary_text_done: set[int] = set()
+        reasoning_summary_part_done: set[int] = set()
+        reasoning_completed_entry: tuple[int, openai_models.ReasoningOutput] | None = (
+            None
+        )
+        next_reasoning_summary_index = 0
+        reasoning_summary_signatures: dict[int, str | None] = {}
+        created_at_value: int | None = None
+
+        instructions_text = get_last_instructions()
+        if not instructions_text:
+            try:
+                from ccproxy.core.request_context import RequestContext
+
+                ctx = RequestContext.get_current()
+                if ctx is not None:
+                    instr = ctx.metadata.get("instructions")
+                    if isinstance(instr, str) and instr.strip():
+                        instructions_text = instr.strip()
+            except Exception:
+                pass
+
+        instructions_value = instructions_text or None
+
+        envelope_base_kwargs: dict[str, Any] = {
+            "id": "",
+            "object": "response",
+            "created_at": 0,
+            "instructions": instructions_value,
+        }
+        reasoning_summary_payload: list[dict[str, Any]] | None = None
+
+        last_request = get_last_request()
+        anthropic_request: anthropic_models.CreateMessageRequest | None = None
+        if isinstance(last_request, anthropic_models.CreateMessageRequest):
+            anthropic_request = last_request
+        elif isinstance(last_request, dict):
+            try:
+                anthropic_request = (
+                    anthropic_models.CreateMessageRequest.model_validate(last_request)
+                )
+            except ValidationError:
+                anthropic_request = None
+
+        base_parallel_tool_calls = True
+        text_payload: dict[str, Any] | None = None
+
+        if anthropic_request is not None:
+            payload_data, _ = _build_responses_payload_from_anthropic_request(
+                anthropic_request
+            )
+            base_parallel_tool_calls = bool(
+                payload_data.get("parallel_tool_calls", True)
+            )
+            background_value = payload_data.get("background", None)
+            envelope_base_kwargs["background"] = (
+                bool(background_value) if background_value is not None else None
+            )
+            for key in (
+                "max_output_tokens",
+                "tool_choice",
+                "tools",
+                "service_tier",
+                "temperature",
+                "prompt_cache_key",
+                "top_p",
+                "metadata",
+            ):
+                if key in payload_data:
+                    envelope_base_kwargs[key] = payload_data[key]
+            text_payload = payload_data.get("text")
+        else:
+            envelope_base_kwargs["background"] = None
+
+        if text_payload is None:
+            text_payload = {"format": {"type": "text"}}
+        else:
+            text_payload = dict(text_payload)
+        text_payload.setdefault("verbosity", "low")
+        envelope_base_kwargs["text"] = text_payload
+
+        if "store" not in envelope_base_kwargs:
+            envelope_base_kwargs["store"] = True
+
+        if "temperature" not in envelope_base_kwargs:
+            temp_value = None
+            if anthropic_request is not None:
+                temp_value = anthropic_request.temperature
+            envelope_base_kwargs["temperature"] = (
+                temp_value if temp_value is not None else 1.0
+            )
+
+        if "service_tier" not in envelope_base_kwargs:
+            service_value = None
+            if anthropic_request is not None:
+                service_value = anthropic_request.service_tier
+            envelope_base_kwargs["service_tier"] = service_value or "auto"
+
+        if "top_p" not in envelope_base_kwargs:
+            top_p_value = None
+            if anthropic_request is not None:
+                top_p_value = anthropic_request.top_p
+            envelope_base_kwargs["top_p"] = (
+                top_p_value if top_p_value is not None else 1.0
+            )
+
+        if "metadata" not in envelope_base_kwargs:
+            envelope_base_kwargs["metadata"] = {}
+
+        reasoning_effort = None
+        if anthropic_request is not None:
+            thinking_cfg = getattr(anthropic_request, "thinking", None)
+            if getattr(thinking_cfg, "type", None) == "enabled":
+                reasoning_effort = "medium"
+        envelope_base_kwargs["reasoning"] = openai_models.Reasoning(
+            effort=reasoning_effort,
+            summary=None,
+        )
+
+        if "tool_choice" not in envelope_base_kwargs:
+            envelope_base_kwargs["tool_choice"] = "auto"
+        if "tools" not in envelope_base_kwargs:
+            envelope_base_kwargs["tools"] = []
+
+        parallel_setting_initial = bool(base_parallel_tool_calls)
+        envelope_base_kwargs["parallel_tool_calls"] = parallel_setting_initial
+
+        tool_states = IndexedToolCallTracker()
+        obfuscation_factory = ObfuscationTokenFactory(
+            lambda: id_suffix or response_id or "stream"
+        )
+
+        def ensure_message_output_item() -> list[openai_models.StreamEventType]:
+            nonlocal message_item_added, message_output_index, next_output_index
+            events: list[openai_models.StreamEventType] = []
+            if message_output_index is None:
+                message_output_index = next_output_index
+                next_output_index += 1
+            if not message_item_added:
+                message_item_added = True
+                nonlocal sequence_counter
+                sequence_counter += 1
+                events.append(
+                    openai_models.ResponseOutputItemAddedEvent(
+                        type="response.output_item.added",
+                        sequence_number=sequence_counter,
+                        output_index=message_output_index,
+                        item=openai_models.OutputItem(
+                            id=message_item_id,
+                            type="message",
+                            role="assistant",
+                            status="in_progress",
+                            content=[],
+                        ),
+                    )
+                )
+            return events
+
+        def ensure_message_content_part() -> list[openai_models.StreamEventType]:
+            events = ensure_message_output_item()
+            nonlocal message_content_part_added, sequence_counter
+            if not message_content_part_added and message_output_index is not None:
+                message_content_part_added = True
+                sequence_counter += 1
+                events.append(
+                    openai_models.ResponseContentPartAddedEvent(
+                        type="response.content_part.added",
+                        sequence_number=sequence_counter,
+                        item_id=message_item_id,
+                        output_index=message_output_index,
+                        content_index=content_index,
+                        part=openai_models.ContentPart(
+                            type="output_text",
+                            text="",
+                            annotations=[],
+                        ),
+                    )
+                )
+            return events
+
+        def emit_message_text_delta(
+            text_delta: str,
+            *,
+            logprobs: Any | None = None,
+            obfuscation: str | None = None,
+        ) -> list[openai_models.StreamEventType]:
+            if not isinstance(text_delta, str) or not text_delta:
+                return []
+
+            nonlocal sequence_counter, message_last_logprobs, message_item_done_emitted
+            if message_item_done_emitted:
+                return []
+
+            events = ensure_message_content_part()
+            sequence_counter += 1
+            event_sequence = sequence_counter
+            logprobs_value: Any = [] if logprobs is None else logprobs
+            events.append(
+                openai_models.ResponseOutputTextDeltaEvent(
+                    type="response.output_text.delta",
+                    sequence_number=event_sequence,
+                    item_id=message_item_id,
+                    output_index=message_output_index or 0,
+                    content_index=content_index,
+                    delta=text_delta,
+                    logprobs=logprobs_value,
+                )
+            )
+            text_buffer.append(text_delta)
+            message_last_logprobs = logprobs_value
+            return events
+
+        def _reasoning_key(signature: str | None) -> str:
+            if isinstance(signature, str) and signature.strip():
+                return signature.strip()
+            return "__default__"
+
+        def get_reasoning_summary_index(signature: str | None) -> int:
+            nonlocal next_reasoning_summary_index
+            key = _reasoning_key(signature)
+            existing = reasoning_summary_indices.get(key)
+            if existing is not None:
+                return existing
+            reasoning_summary_indices[key] = next_reasoning_summary_index
+            reasoning_summary_signatures[next_reasoning_summary_index] = signature
+            next_reasoning_summary_index += 1
+            return reasoning_summary_indices[key]
+
+        def ensure_reasoning_output_item() -> (
+            openai_models.ResponseOutputItemAddedEvent | None
+        ):
+            nonlocal reasoning_item_added, reasoning_output_index
+            nonlocal sequence_counter, next_output_index
+            if reasoning_output_index is None:
+                reasoning_output_index = next_output_index
+                next_output_index += 1
+            if not reasoning_item_added:
+                reasoning_item_added = True
+                sequence_counter += 1
+                return openai_models.ResponseOutputItemAddedEvent(
+                    type="response.output_item.added",
+                    sequence_number=sequence_counter,
+                    output_index=reasoning_output_index,
+                    item=openai_models.OutputItem(
+                        id=reasoning_item_id,
+                        type="reasoning",
+                        status="in_progress",
+                        summary=[],
+                    ),
+                )
+            return None
+
+        def ensure_reasoning_summary_part(
+            summary_index: int,
+        ) -> openai_models.ReasoningSummaryPartAddedEvent | None:
+            nonlocal sequence_counter
+            if reasoning_output_index is None:
+                return None
+            if summary_index in reasoning_summary_added:
+                return None
+            reasoning_summary_added.add(summary_index)
+            sequence_counter += 1
+            return openai_models.ReasoningSummaryPartAddedEvent(
+                type="response.reasoning_summary_part.added",
+                sequence_number=sequence_counter,
+                item_id=reasoning_item_id,
+                output_index=reasoning_output_index,
+                summary_index=summary_index,
+                part=openai_models.ReasoningSummaryPart(
+                    type="summary_text",
+                    text="",
+                ),
+            )
+
+        def emit_reasoning_text_delta(
+            text_delta: str,
+            signature: str | None,
+        ) -> list[openai_models.StreamEventType]:
+            if not isinstance(text_delta, str) or not text_delta:
+                return []
+
+            events: list[openai_models.StreamEventType] = []
+            output_event = ensure_reasoning_output_item()
+            if output_event is not None:
+                events.append(output_event)
+
+            summary_index = get_reasoning_summary_index(signature)
+            part_event = ensure_reasoning_summary_part(summary_index)
+            if part_event is not None:
+                events.append(part_event)
+
+            fragments = reasoning_summary_text_fragments.setdefault(summary_index, [])
+            fragments.append(text_delta)
+            if summary_index not in reasoning_summary_signatures:
+                reasoning_summary_signatures[summary_index] = signature
+
+            nonlocal sequence_counter
+            sequence_counter += 1
+            event_sequence = sequence_counter
+            events.append(
+                openai_models.ReasoningSummaryTextDeltaEvent(
+                    type="response.reasoning_summary_text.delta",
+                    sequence_number=event_sequence,
+                    item_id=reasoning_item_id,
+                    output_index=reasoning_output_index or 0,
+                    summary_index=summary_index,
+                    delta=text_delta,
+                )
+            )
+            return events
+
+        def finalize_reasoning() -> list[openai_models.StreamEventType]:
+            nonlocal reasoning_output_done, reasoning_completed_entry
+            nonlocal reasoning_summary_payload, sequence_counter
+            if not reasoning_item_added or reasoning_output_index is None:
+                return []
+
+            events: list[openai_models.StreamEventType] = []
+            summary_entries: list[dict[str, Any]] = []
+
+            for summary_index in sorted(reasoning_summary_text_fragments):
+                text_value = "".join(
+                    reasoning_summary_text_fragments.get(summary_index, [])
+                )
+                if summary_index not in reasoning_summary_text_done:
+                    sequence_counter += 1
+                    events.append(
+                        openai_models.ReasoningSummaryTextDoneEvent(
+                            type="response.reasoning_summary_text.done",
+                            sequence_number=sequence_counter,
+                            item_id=reasoning_item_id,
+                            output_index=reasoning_output_index,
+                            summary_index=summary_index,
+                            text=text_value,
+                        )
+                    )
+                    reasoning_summary_text_done.add(summary_index)
+                if summary_index not in reasoning_summary_part_done:
+                    sequence_counter += 1
+                    events.append(
+                        openai_models.ReasoningSummaryPartDoneEvent(
+                            type="response.reasoning_summary_part.done",
+                            sequence_number=sequence_counter,
+                            item_id=reasoning_item_id,
+                            output_index=reasoning_output_index,
+                            summary_index=summary_index,
+                            part=openai_models.ReasoningSummaryPart(
+                                type="summary_text",
+                                text=text_value,
+                            ),
+                        )
+                    )
+                    reasoning_summary_part_done.add(summary_index)
+                summary_entry: dict[str, Any] = {
+                    "type": "summary_text",
+                    "text": text_value,
+                }
+                signature_value = reasoning_summary_signatures.get(summary_index)
+                if signature_value:
+                    summary_entry["signature"] = signature_value
+                summary_entries.append(summary_entry)
+
+            reasoning_summary_payload = summary_entries
+
+            if not reasoning_output_done:
+                sequence_counter += 1
+                events.append(
+                    openai_models.ResponseOutputItemDoneEvent(
+                        type="response.output_item.done",
+                        sequence_number=sequence_counter,
+                        output_index=reasoning_output_index,
+                        item=openai_models.OutputItem(
+                            id=reasoning_item_id,
+                            type="reasoning",
+                            status="completed",
+                            summary=summary_entries,
+                        ),
+                    )
+                )
+                reasoning_output_done = True
+                reasoning_completed_entry = (
+                    reasoning_output_index,
+                    openai_models.ReasoningOutput(
+                        type="reasoning",
+                        id=reasoning_item_id,
+                        status="completed",
+                        summary=summary_entries,
+                    ),
+                )
+
+            return events
+
+        def ensure_tool_state(block_index: int) -> ToolCallState:
+            nonlocal next_output_index
+            state = tool_states.ensure(block_index)
+            if state.output_index < 0:
+                state.output_index = next_output_index
+                next_output_index += 1
+            return state
+
+        def emit_tool_item_added(
+            block_index: int, state: ToolCallState
+        ) -> list[openai_models.StreamEventType]:
+            events: list[openai_models.StreamEventType] = []
+            if state.added_emitted:
+                return events
+
+            tool_entry = accumulator.get_tool_entry(block_index)
+            if tool_entry:
+                if not state.name:
+                    state.name = tool_entry.get("function", {}).get(
+                        "name"
+                    ) or tool_entry.get("name")
+                if not state.call_id:
+                    state.call_id = tool_entry.get("id")
+
+            item_id = state.item_id or state.call_id or f"call_{state.index}"
+            state.item_id = item_id
+
+            name = state.name or "function"
+
+            nonlocal sequence_counter
+            sequence_counter += 1
+            events.append(
+                openai_models.ResponseOutputItemAddedEvent(
+                    type="response.output_item.added",
+                    sequence_number=sequence_counter,
+                    output_index=state.output_index,
+                    item=openai_models.OutputItem(
+                        id=str(item_id),
+                        type="function_call",
+                        status="in_progress",
+                        name=str(name),
+                        arguments="",
+                        call_id=state.call_id,
+                    ),
+                )
+            )
+            state.added_emitted = True
+            return events
+
+        def emit_tool_arguments_delta(
+            state: ToolCallState, delta_text: str
+        ) -> openai_models.StreamEventType:
+            nonlocal sequence_counter
+            sequence_counter += 1
+            event_sequence = sequence_counter
+            state.add_arguments_part(delta_text)
+            item_identifier = str(state.item_id or f"call_{state.index}")
+            return openai_models.ResponseFunctionCallArgumentsDeltaEvent(
+                type="response.function_call_arguments.delta",
+                sequence_number=event_sequence,
+                item_id=item_identifier,
+                output_index=state.output_index,
+                delta=delta_text,
+            )
+
+        def emit_tool_finalize(
+            block_index: int, state: ToolCallState
+        ) -> list[openai_models.StreamEventType]:
+            events: list[openai_models.StreamEventType] = []
+            tool_entry = accumulator.get_tool_entry(block_index)
+
+            if tool_entry:
+                if not state.name:
+                    state.name = tool_entry.get("function", {}).get(
+                        "name"
+                    ) or tool_entry.get("name")
+                if not state.call_id:
+                    state.call_id = tool_entry.get("id")
+                if not state.item_id:
+                    state.item_id = tool_entry.get("id")
+
+            item_id = state.item_id or state.call_id or f"call_{state.index}"
+            state.item_id = item_id
+            name = state.name or "function"
+
+            args_str = "".join(state.arguments_parts)
+            if not args_str and tool_entry:
+                try:
+                    args_str = json.dumps(
+                        tool_entry.get("input", {}), ensure_ascii=False
+                    )
+                except Exception:
+                    args_str = json.dumps(tool_entry.get("input", {}))
+
+            nonlocal sequence_counter
+            if not state.added_emitted:
+                events.extend(emit_tool_item_added(block_index, state))
+
+            if not state.arguments_done_emitted:
+                sequence_counter += 1
+                events.append(
+                    openai_models.ResponseFunctionCallArgumentsDoneEvent(
+                        type="response.function_call_arguments.done",
+                        sequence_number=sequence_counter,
+                        item_id=str(item_id),
+                        output_index=state.output_index,
+                        arguments=args_str,
+                    )
+                )
+                state.arguments_done_emitted = True
+
+            if not state.item_done_emitted:
+                sequence_counter += 1
+                events.append(
+                    openai_models.ResponseOutputItemDoneEvent(
+                        type="response.output_item.done",
+                        sequence_number=sequence_counter,
+                        output_index=state.output_index,
+                        item=openai_models.OutputItem(
+                            id=str(item_id),
+                            type="function_call",
+                            status="completed",
+                            name=str(name),
+                            arguments=args_str,
+                            call_id=state.call_id,
+                        ),
+                    )
+                )
+                state.item_done_emitted = True
+                state.final_arguments = args_str
+
+            return events
+
+        def finalize_message() -> list[openai_models.StreamEventType]:
+            nonlocal sequence_counter
+            nonlocal message_text_done_emitted, message_part_done_emitted
+            nonlocal message_item_done_emitted, message_completed_entry
+            nonlocal message_last_logprobs
+            nonlocal accumulator
+
+            if not message_item_added or message_output_index is None:
+                return []
+
+            events: list[openai_models.StreamEventType] = []
+            final_text = "".join(text_buffer)
+            logprobs_value: Any
+            if message_last_logprobs is None:
+                logprobs_value = []
+            else:
+                logprobs_value = message_last_logprobs
+
+            primary_text_part: openai_models.OutputTextContent | None = None
+            tool_and_aux_blocks: list[Any] = []
+
+            if accumulator.content_blocks:
+                sorted_blocks = sorted(
+                    accumulator.content_blocks, key=lambda block: block.get("index", 0)
+                )
+                for block in sorted_blocks:
+                    block_type = block.get("type")
+                    if block_type == "text":
+                        text_value = block.get("text", "")
+                        part = openai_models.OutputTextContent(
+                            type="output_text",
+                            text=text_value,
+                            annotations=[],
+                            logprobs=logprobs_value if text_value else [],
+                        )
+                        if primary_text_part is None and text_value:
+                            primary_text_part = part
+                        tool_and_aux_blocks.append(part)
+                    else:
+                        block_payload = {k: v for k, v in block.items() if k != "index"}
+                        if block_payload.get("type") == "tool_use":
+                            tool_input = block_payload.get("input")
+                            if tool_input is not None:
+                                block_payload.setdefault("arguments", tool_input)
+                        tool_and_aux_blocks.append(block_payload)
+
+            if primary_text_part is None and final_text:
+                primary_text_part = openai_models.OutputTextContent(
+                    type="output_text",
+                    text=final_text,
+                    annotations=[],
+                    logprobs=logprobs_value if final_text else [],
+                )
+                tool_and_aux_blocks.insert(0, primary_text_part)
+
+            if message_content_part_added and not message_text_done_emitted:
+                sequence_counter += 1
+                event_sequence = sequence_counter
+                events.append(
+                    openai_models.ResponseOutputTextDoneEvent(
+                        type="response.output_text.done",
+                        sequence_number=event_sequence,
+                        item_id=message_item_id,
+                        output_index=message_output_index,
+                        content_index=content_index,
+                        text=final_text,
+                        logprobs=logprobs_value,
+                    )
+                )
+                message_text_done_emitted = True
+
+            if message_content_part_added and not message_part_done_emitted:
+                sequence_counter += 1
+                event_sequence = sequence_counter
+                events.append(
+                    openai_models.ResponseContentPartDoneEvent(
+                        type="response.content_part.done",
+                        sequence_number=event_sequence,
+                        item_id=message_item_id,
+                        output_index=message_output_index,
+                        content_index=content_index,
+                        part=openai_models.ContentPart(
+                            type="output_text",
+                            text=final_text,
+                            annotations=[],
+                        ),
+                    )
+                )
+                message_part_done_emitted = True
+
+            if not message_item_done_emitted:
+                sequence_counter += 1
+                event_sequence = sequence_counter
+                if primary_text_part is None:
+                    primary_text_part = openai_models.OutputTextContent(
+                        type="output_text",
+                        text=final_text,
+                        annotations=[],
+                        logprobs=logprobs_value if logprobs_value != [] else [],
+                    )
+                    tool_and_aux_blocks.insert(0, primary_text_part)
+                message_output = openai_models.MessageOutput(
+                    type="message",
+                    id=message_item_id,
+                    status="completed",
+                    role="assistant",
+                    content=tool_and_aux_blocks,
+                )
+                message_completed_entry = (message_output_index, message_output)
+                events.append(
+                    openai_models.ResponseOutputItemDoneEvent(
+                        type="response.output_item.done",
+                        sequence_number=event_sequence,
+                        output_index=message_output_index,
+                        item=openai_models.OutputItem(
+                            id=message_item_id,
+                            type="message",
+                            role="assistant",
+                            status="completed",
+                            content=[
+                                part.model_dump()
+                                if hasattr(part, "model_dump")
+                                else part
+                                for part in tool_and_aux_blocks
+                            ],
+                            text=final_text or None,
+                        ),
+                    )
+                )
+                message_item_done_emitted = True
+            else:
+                if primary_text_part is None and final_text:
+                    primary_text_part = openai_models.OutputTextContent(
+                        type="output_text",
+                        text=final_text,
+                        annotations=[],
+                        logprobs=logprobs_value if logprobs_value != [] else [],
+                    )
+                    tool_and_aux_blocks.insert(0, primary_text_part)
+                message_completed_entry = (
+                    message_output_index,
+                    openai_models.MessageOutput(
+                        type="message",
+                        id=message_item_id,
+                        status="completed",
+                        role="assistant",
+                        content=tool_and_aux_blocks,
+                    ),
+                )
+
+            return events
+
+        def make_response_object(
+            *,
+            status: str,
+            model: str | None,
+            usage: openai_models.ResponseUsage | None = None,
+            output: list[Any] | None = None,
+            parallel_override: bool | None = None,
+            reasoning_summary: list[dict[str, Any]] | None = None,
+            extra: dict[str, Any] | None = None,
+        ) -> openai_models.ResponseObject:
+            payload = dict(envelope_base_kwargs)
+            payload["status"] = status
+            payload["model"] = model or payload.get("model") or ""
+            payload["output"] = output or []
+            payload["usage"] = usage
+            payload.setdefault("object", "response")
+            payload.setdefault("created_at", int(time.time()))
+            if parallel_override is not None:
+                payload["parallel_tool_calls"] = parallel_override
+            if reasoning_summary is not None:
+                reasoning_entry = payload.get("reasoning")
+                if isinstance(reasoning_entry, openai_models.Reasoning):
+                    payload["reasoning"] = reasoning_entry.model_copy(
+                        update={"summary": reasoning_summary}
+                    )
+                elif isinstance(reasoning_entry, dict):
+                    payload["reasoning"] = openai_models.Reasoning(
+                        effort=reasoning_entry.get("effort"),
+                        summary=reasoning_summary,
+                    )
+                else:
+                    payload["reasoning"] = openai_models.Reasoning(
+                        effort=None,
+                        summary=reasoning_summary,
+                    )
+            if extra:
+                payload.update(extra)
+            return openai_models.ResponseObject(**payload)
+
+        try:
+            async for raw_event in stream:
+                event_type, event_payload = _normalize_anthropic_stream_event(raw_event)
+                if not event_type:
+                    continue
+
+                accumulator.accumulate(event_type, event_payload)
+
+                if event_type == "ping":
+                    continue
+
+                if event_type == "error":
+                    continue
+
+                if event_type == "message_start":
+                    message = (
+                        event_payload.get("message", {})
+                        if isinstance(event_payload, dict)
+                        else {}
+                    )
+                    model_id = str(message.get("model", ""))
+                    response_id, id_suffix = ensure_identifier(
+                        "resp", message.get("id")
+                    )
+                    envelope_base_kwargs["id"] = response_id
+                    envelope_base_kwargs.setdefault("object", "response")
+                    if model_id:
+                        envelope_base_kwargs["model"] = model_id
+                    if not message_item_id:
+                        message_item_id = f"msg_{id_suffix}"
+                    if not reasoning_item_id:
+                        reasoning_item_id = f"rs_{id_suffix}"
+
+                    created_at_value = (
+                        message.get("created_at")
+                        or message.get("created")
+                        or int(time.time())
+                    )
+                    envelope_base_kwargs["created_at"] = int(created_at_value)
+
+                    sequence_counter += 1
+                    yield openai_models.ResponseCreatedEvent(
+                        type="response.created",
+                        sequence_number=sequence_counter,
+                        response=make_response_object(
+                            status="in_progress",
+                            model=model_id,
+                            usage=None,
+                            output=[],
+                            parallel_override=parallel_setting_initial,
+                        ),
+                    )
+                    sequence_counter += 1
+                    yield openai_models.ResponseInProgressEvent(
+                        type="response.in_progress",
+                        sequence_number=sequence_counter,
+                        response=make_response_object(
+                            status="in_progress",
+                            model=model_id,
+                            usage=latest_usage_model,
+                            output=[],
+                            parallel_override=parallel_setting_initial,
+                        ),
+                    )
+                    continue
+
+                if event_type == "content_block_start":
+                    block_index = int(event_payload.get("index", 0))
+                    content_block = (
+                        event_payload.get("content_block", {})
+                        if isinstance(event_payload, dict)
+                        else {}
+                    )
+                    if (
+                        isinstance(content_block, dict)
+                        and content_block.get("type") == "tool_use"
+                    ):
+                        state = ensure_tool_state(block_index)
+                        name_value = content_block.get("name")
+                        if isinstance(name_value, str) and name_value:
+                            state.name = state.name or name_value
+                        block_id = content_block.get("id")
+                        if isinstance(block_id, str) and block_id:
+                            if not state.call_id:
+                                state.call_id = block_id
+                            if not state.item_id:
+                                state.item_id = block_id
+                        for event in finalize_message():
+                            yield event
+                        for event in emit_tool_item_added(block_index, state):
+                            yield event
+                    continue
+
+                if event_type == "content_block_delta":
+                    block_index = int(event_payload.get("index", 0))
+                    block_info = accumulator.get_block_info(block_index)
+                    if not block_info:
+                        continue
+                    _, block_meta = block_info
+                    delta_payload = event_payload.get("delta")
+
+                    block_type = block_meta.get("type")
+
+                    if block_type == "thinking" and isinstance(delta_payload, dict):
+                        thinking_text = delta_payload.get("thinking")
+                        if isinstance(thinking_text, str) and thinking_text:
+                            signature = block_meta.get("signature")
+                            for event in emit_reasoning_text_delta(
+                                thinking_text, signature
+                            ):
+                                yield event
+                        continue
+
+                    if block_type == "text" and isinstance(delta_payload, dict):
+                        text_delta = delta_payload.get("text")
+                        if isinstance(text_delta, str) and text_delta:
+                            for event in emit_message_text_delta(
+                                text_delta,
+                                logprobs=delta_payload.get("logprobs"),
+                                obfuscation=delta_payload.get("obfuscation")
+                                or delta_payload.get("obfuscated"),
+                            ):
+                                yield event
+                        continue
+
+                    if block_type == "tool_use" and isinstance(delta_payload, dict):
+                        partial = delta_payload.get("partial_json") or ""
+                        if partial:
+                            state = ensure_tool_state(block_index)
+                            for event in finalize_message():
+                                yield event
+                            for event in emit_tool_item_added(block_index, state):
+                                yield event
+                            yield emit_tool_arguments_delta(
+                                state,
+                                str(partial),
+                            )
+                    continue
+
+                if event_type == "content_block_stop":
+                    block_index = int(event_payload.get("index", 0))
+                    block_info = accumulator.get_block_info(block_index)
+                    if block_info and block_info[1].get("type") == "tool_use":
+                        state = ensure_tool_state(block_index)
+                        for event in emit_tool_finalize(block_index, state):
+                            yield event
+                    continue
+
+                if event_type == "message_delta":
+                    delta_payload = (
+                        event_payload.get("delta", {})
+                        if isinstance(event_payload, dict)
+                        else {}
+                    )
+                    stop_reason = (
+                        delta_payload.get("stop_reason")
+                        if isinstance(delta_payload, dict)
+                        else None
+                    )
+                    if isinstance(stop_reason, str):
+                        final_stop_reason = stop_reason
+
+                    usage_payload = (
+                        event_payload.get("usage")
+                        if isinstance(event_payload, dict)
+                        else None
+                    )
+                    usage_model: anthropic_models.Usage | None = None
+                    if usage_payload:
+                        try:
+                            usage_model = anthropic_models.Usage.model_validate(
+                                usage_payload
+                            )
+                        except ValidationError:
+                            usage_model = anthropic_models.Usage(
+                                input_tokens=usage_payload.get("input_tokens", 0),
+                                output_tokens=usage_payload.get("output_tokens", 0),
+                            )
+                    elif hasattr(raw_event, "usage") and raw_event.usage is not None:
+                        usage_model = raw_event.usage
+
+                    if usage_model is not None:
+                        latest_usage_model = (
+                            convert__anthropic_usage_to_openai_responses__usage(
+                                usage_model
+                            )
+                        )
+
+                    sequence_counter += 1
+                    yield openai_models.ResponseInProgressEvent(
+                        type="response.in_progress",
+                        sequence_number=sequence_counter,
+                        response=make_response_object(
+                            status="in_progress",
+                            model=model_id,
+                            usage=latest_usage_model,
+                            output=[],
+                            parallel_override=parallel_setting_initial,
+                        ),
+                    )
+                    continue
+
+                if event_type == "message_stop":
+                    for event in finalize_reasoning():
+                        yield event
+
+                    for event in finalize_message():
+                        yield event
+
+                    for index, state in list(tool_states.items()):
+                        for event in emit_tool_finalize(index, state):
+                            yield event
+
+                    first_completed_entries: list[tuple[int, Any]] = []
+                    if reasoning_completed_entry is not None:
+                        first_completed_entries.append(reasoning_completed_entry)
+                    if message_completed_entry is not None:
+                        first_completed_entries.append(message_completed_entry)
+
+                    for index, state in tool_states.items():
+                        tool_entry = accumulator.get_tool_entry(index)
+                        if state.name is None and tool_entry is not None:
+                            state.name = tool_entry.get("name") or tool_entry.get(
+                                "function", {}
+                            ).get("name")
+                        if state.call_id is None and tool_entry is not None:
+                            state.call_id = tool_entry.get("id")
+                        if not state.item_id:
+                            state.item_id = state.call_id or f"call_{state.index}"
+
+                        final_args = state.final_arguments
+                        if final_args is None:
+                            combined = "".join(state.arguments_parts)
+                            if not combined and tool_entry is not None:
+                                input_payload = tool_entry.get("input", {}) or {}
+                                try:
+                                    combined = json.dumps(
+                                        input_payload, ensure_ascii=False
+                                    )
+                                except Exception:
+                                    combined = json.dumps(input_payload)
+                            final_args = combined or ""
+                        state.final_arguments = final_args
+
+                        first_completed_entries.append(
+                            (
+                                state.output_index,
+                                openai_models.FunctionCallOutput(
+                                    type="function_call",
+                                    id=state.item_id,
+                                    status="completed",
+                                    name=state.name,
+                                    call_id=state.call_id,
+                                    arguments=final_args,
+                                ),
+                            )
+                        )
+
+                    first_completed_entries.sort(key=lambda item: item[0])
+                    completed_outputs = [entry for _, entry in first_completed_entries]
+
+                    complete_tool_calls_payload = accumulator.get_complete_tool_calls()
+                    parallel_final = parallel_setting_initial or len(tool_states) > 1
+
+                    extra_fields: dict[str, Any] | None = None
+                    if complete_tool_calls_payload:
+                        extra_fields = {"tool_calls": complete_tool_calls_payload}
+
+                    status_value = "completed"
+                    if final_stop_reason == "max_tokens":
+                        status_value = "incomplete"
+
+                    completed_response = make_response_object(
+                        status=status_value,
+                        model=model_id,
+                        usage=latest_usage_model,
+                        output=completed_outputs,
+                        parallel_override=parallel_final,
+                        reasoning_summary=reasoning_summary_payload,
+                        extra=extra_fields,
+                    )
+
+                    sequence_counter += 1
+                    yield openai_models.ResponseCompletedEvent(
+                        type="response.completed",
+                        sequence_number=sequence_counter,
+                        response=completed_response,
+                    )
+                    stream_completed = True
+                    break
+
+            if not stream_completed:
+                for event in finalize_reasoning():
+                    yield event
+
+                for event in finalize_message():
+                    yield event
+
+                for index, state in list(tool_states.items()):
+                    for event in emit_tool_finalize(index, state):
+                        yield event
+
+                if (
+                    message_completed_entry is None
+                    and message_item_added
+                    and message_output_index is not None
+                ):
+                    final_text = "".join(text_buffer)
+                    logprobs_value: Any
+                    if message_last_logprobs is None:
+                        logprobs_value = []
+                    else:
+                        logprobs_value = message_last_logprobs
+                    content_blocks: list[Any] = []
+                    if accumulator.content_blocks:
+                        sorted_blocks = sorted(
+                            accumulator.content_blocks,
+                            key=lambda block: block.get("index", 0),
+                        )
+                        for block in sorted_blocks:
+                            block_type = block.get("type")
+                            if block_type == "text":
+                                text_value = block.get("text", "")
+                                content_blocks.append(
+                                    openai_models.OutputTextContent(
+                                        type="output_text",
+                                        text=text_value,
+                                        annotations=[],
+                                        logprobs=logprobs_value if text_value else [],
+                                    )
+                                )
+                            else:
+                                payload = {
+                                    k: v for k, v in block.items() if k != "index"
+                                }
+                                if payload.get("type") == "tool_use":
+                                    tool_input = payload.get("input")
+                                    if tool_input is not None:
+                                        payload.setdefault("arguments", tool_input)
+                                content_blocks.append(payload)
+                    else:
+                        if final_text:
+                            content_blocks.append(
+                                openai_models.OutputTextContent(
+                                    type="output_text",
+                                    text=final_text,
+                                    annotations=[],
+                                    logprobs=logprobs_value
+                                    if logprobs_value != []
+                                    else [],
+                                )
+                            )
+
+                    message_completed_entry = (
+                        message_output_index,
+                        openai_models.MessageOutput(
+                            type="message",
+                            id=message_item_id,
+                            status="completed",
+                            role="assistant",
+                            content=content_blocks,
+                        ),
+                    )
+
+                final_completed_entries: list[tuple[int, Any]] = []
+                if reasoning_completed_entry is not None:
+                    final_completed_entries.append(reasoning_completed_entry)
+                if message_completed_entry is not None:
+                    final_completed_entries.append(message_completed_entry)
+
+                for index, state in tool_states.items():
+                    tool_entry = accumulator.get_tool_entry(index)
+                    if state.name is None and tool_entry is not None:
+                        state.name = tool_entry.get("name") or tool_entry.get(
+                            "function", {}
+                        ).get("name")
+                    if state.call_id is None and tool_entry is not None:
+                        state.call_id = tool_entry.get("id")
+                    if not state.item_id:
+                        state.item_id = state.call_id or f"call_{state.index}"
+                    final_args = state.final_arguments
+                    if final_args is None:
+                        combined = "".join(state.arguments_parts)
+                        if not combined and tool_entry is not None:
+                            input_payload = tool_entry.get("input", {}) or {}
+                            try:
+                                combined = json.dumps(input_payload, ensure_ascii=False)
+                            except Exception:
+                                combined = json.dumps(input_payload)
+                        final_args = combined or ""
+                    state.final_arguments = final_args
+                    final_completed_entries.append(
+                        (
+                            state.output_index,
+                            openai_models.FunctionCallOutput(
+                                type="function_call",
+                                id=state.item_id,
+                                status="completed",
+                                name=state.name,
+                                call_id=state.call_id,
+                                arguments=final_args,
+                            ),
+                        )
+                    )
+
+                final_completed_entries.sort(key=lambda item: item[0])
+                completed_outputs = [entry for _, entry in final_completed_entries]
+
+                complete_tool_calls_payload = accumulator.get_complete_tool_calls()
+                parallel_final = parallel_setting_initial or len(tool_states) > 1
+
+                final_extra_fields: dict[str, Any] | None = None
+                if complete_tool_calls_payload:
+                    final_extra_fields = {"tool_calls": complete_tool_calls_payload}
+
+                fallback_response = make_response_object(
+                    status="completed",
+                    model=model_id,
+                    usage=latest_usage_model,
+                    output=completed_outputs,
+                    parallel_override=parallel_final,
+                    reasoning_summary=reasoning_summary_payload,
+                    extra=final_extra_fields,
+                )
+
+                sequence_counter += 1
+                yield openai_models.ResponseCompletedEvent(
+                    type="response.completed",
+                    sequence_number=sequence_counter,
+                    response=fallback_response,
+                )
+
+        finally:
+            register_request(None)
+
+
+class AnthropicToOpenAIChatStreamAdapter:
+    """Stateful adapter for Anthropic → OpenAI Chat streaming."""
+
+    async def run(
+        self,
+        stream: AsyncIterator[anthropic_models.MessageStreamEvent],
+    ) -> AsyncGenerator[openai_models.ChatCompletionChunk, None]:
+        async for chunk in self._convert_chat_stream(stream):
+            yield chunk
+
+    def _convert_chat_stream(
+        self,
+        stream: AsyncIterator[anthropic_models.MessageStreamEvent],
+    ) -> AsyncGenerator[openai_models.ChatCompletionChunk, None]:
+        """Convert Anthropic stream to OpenAI stream using ClaudeAccumulator."""
+
+        async def generator() -> AsyncGenerator[
+            openai_models.ChatCompletionChunk, None
+        ]:
+            accumulator = ClaudeAccumulator()
+            model_id = ""
+            finish_reason: FinishReason = "stop"
+            usage_prompt = 0
+            usage_completion = 0
+            message_started = False
+            emitted_tool_indices: set[int] = set()
+
+            async for raw_event in stream:
+                event_type, event_payload = _normalize_anthropic_stream_event(raw_event)
+                if not event_type:
+                    continue
+
+                accumulator.accumulate(event_type, event_payload)
+
+                if event_type == "ping":
+                    continue
+
+                if event_type == "error":
+                    # Error events are handled elsewhere by callers.
+                    continue
+
+                if event_type == "message_start":
+                    message_data = (
+                        event_payload.get("message", {})
+                        if isinstance(event_payload, dict)
+                        else {}
+                    )
+                    model_id = str(message_data.get("model", ""))
+                    message_started = True
+                    yield openai_models.ChatCompletionChunk(
+                        id="chatcmpl-stream",
+                        object="chat.completion.chunk",
+                        created=0,
+                        model=model_id,
+                        choices=[
+                            openai_models.StreamingChoice(
+                                index=0,
+                                delta=openai_models.DeltaMessage(
+                                    role="assistant", content=""
+                                ),
+                                finish_reason=None,
+                            )
+                        ],
+                    )
+                    continue
+
+                if not message_started:
+                    continue
+
+                if event_type == "content_block_delta":
+                    block_index = int(event_payload.get("index", 0))
+                    text_delta = _anthropic_delta_to_text(
+                        accumulator,
+                        block_index,
+                        cast(dict[str, Any] | None, event_payload.get("delta")),
+                    )
+                    if text_delta:
+                        yield openai_models.ChatCompletionChunk(
+                            id="chatcmpl-stream",
+                            object="chat.completion.chunk",
+                            created=0,
+                            model=model_id,
+                            choices=[
+                                openai_models.StreamingChoice(
+                                    index=0,
+                                    delta=openai_models.DeltaMessage(
+                                        role="assistant", content=text_delta
+                                    ),
+                                    finish_reason=None,
+                                )
+                            ],
+                        )
+                    continue
+
+                if event_type == "content_block_stop":
+                    block_index = int(event_payload.get("index", 0))
+                    block_info = accumulator.get_block_info(block_index)
+                    if not block_info:
+                        continue
+                    _, block_meta = block_info
+                    if block_meta.get("type") != "tool_use":
+                        continue
+                    if block_index in emitted_tool_indices:
+                        continue
+                    tool_call = _build_openai_tool_call(accumulator, block_index)
+                    if tool_call is None:
+                        continue
+                    emitted_tool_indices.add(block_index)
+                    yield openai_models.ChatCompletionChunk(
+                        id="chatcmpl-stream",
+                        object="chat.completion.chunk",
+                        created=0,
+                        model=model_id,
+                        choices=[
+                            openai_models.StreamingChoice(
+                                index=0,
+                                delta=openai_models.DeltaMessage(
+                                    role="assistant", tool_calls=[tool_call]
+                                ),
+                                finish_reason=None,
+                            )
+                        ],
+                    )
+                    continue
+
+                if event_type == "message_delta":
+                    delta_payload = (
+                        event_payload.get("delta", {})
+                        if isinstance(event_payload, dict)
+                        else {}
+                    )
+                    stop_reason = (
+                        delta_payload.get("stop_reason")
+                        if isinstance(delta_payload, dict)
+                        else None
+                    )
+                    if isinstance(stop_reason, str):
+                        finish_reason = cast(
+                            FinishReason,
+                            ANTHROPIC_TO_OPENAI_FINISH_REASON.get(stop_reason, "stop"),
+                        )
+
+                    usage_payload = (
+                        event_payload.get("usage")
+                        if isinstance(event_payload, dict)
+                        else None
+                    )
+                    if usage_payload:
+                        snapshot = anthropic_usage_snapshot(usage_payload)
+                        usage_prompt = snapshot.input_tokens
+                        usage_completion = snapshot.output_tokens
+                    elif hasattr(raw_event, "usage") and raw_event.usage is not None:
+                        snapshot = anthropic_usage_snapshot(raw_event.usage)
+                        usage_prompt = snapshot.input_tokens
+                        usage_completion = snapshot.output_tokens
+                    continue
+
+                if event_type == "message_stop":
+                    usage = None
+                    if usage_prompt or usage_completion:
+                        usage = openai_models.CompletionUsage(
+                            prompt_tokens=usage_prompt,
+                            completion_tokens=usage_completion,
+                            total_tokens=usage_prompt + usage_completion,
+                        )
+
+                    yield openai_models.ChatCompletionChunk(
+                        id="chatcmpl-stream",
+                        object="chat.completion.chunk",
+                        created=0,
+                        model=model_id,
+                        choices=[
+                            openai_models.StreamingChoice(
+                                index=0,
+                                delta=openai_models.DeltaMessage(),
+                                finish_reason=finish_reason,
+                            )
+                        ],
+                        usage=usage,
+                    )
+                    break
+
+            else:
+                if message_started:
+                    yield openai_models.ChatCompletionChunk(
+                        id="chatcmpl-stream",
+                        object="chat.completion.chunk",
+                        created=0,
+                        model=model_id,
+                        choices=[
+                            openai_models.StreamingChoice(
+                                index=0,
+                                delta=openai_models.DeltaMessage(),
+                                finish_reason=finish_reason,
+                            )
+                        ],
+                    )
+
+        return generator()
+
+
+async def convert__anthropic_message_to_openai_responses__stream(
+    stream: AsyncIterator[anthropic_models.MessageStreamEvent],
+) -> AsyncGenerator[openai_models.StreamEventType, None]:
+    """Convert Anthropic MessageStreamEvents into OpenAI Responses stream events."""
+
+    adapter = AnthropicToOpenAIResponsesStreamAdapter()
+    async for event in adapter.run(stream):
+        yield event
+
+
+async def convert__anthropic_message_to_openai_chat__stream(
+    stream: AsyncIterator[anthropic_models.MessageStreamEvent],
+) -> AsyncGenerator[openai_models.ChatCompletionChunk, None]:
+    """Convert Anthropic stream to OpenAI stream using ClaudeAccumulator."""
+
+    adapter = AnthropicToOpenAIChatStreamAdapter()
+    async for chunk in adapter.run(stream):
+        yield chunk
+
+
+__all__ = [
+    "AnthropicToOpenAIChatStreamAdapter",
+    "AnthropicToOpenAIResponsesStreamAdapter",
+    "convert__anthropic_message_to_openai_chat__stream",
+    "convert__anthropic_message_to_openai_responses__stream",
+]