ccproxy-api 0.1.7__py3-none-any.whl → 0.2.0a4__py3-none-any.whl

ccproxy/llms/formatters/openai_to_openai/responses.py

@@ -0,0 +1,594 @@
+"""Response conversion entry points for OpenAI↔OpenAI adapters."""
+
+from __future__ import annotations
+
+import contextlib
+import json
+import time
+from typing import Any, Literal
+
+import ccproxy.core.logging
+from ccproxy.llms.formatters.common import (
+    THINKING_PATTERN,
+    ThinkingSegment,
+    convert_openai_completion_usage_to_responses_usage,
+    convert_openai_responses_usage_to_completion_usage,
+    merge_thinking_segments,
+)
+from ccproxy.llms.models import openai as openai_models
+
+from ._helpers import (
+    _get_attr,
+)
+
+
+logger = ccproxy.core.logging.get_logger(__name__)
+
+
+def convert__openai_responses_usage_to_openai_completion__usage(
+    usage: openai_models.ResponseUsage,
+) -> openai_models.CompletionUsage:
+    return convert_openai_responses_usage_to_completion_usage(usage)
+
+
+def convert__openai_completion_usage_to_openai_responses__usage(
+    usage: openai_models.CompletionUsage,
+) -> openai_models.ResponseUsage:
+    return convert_openai_completion_usage_to_responses_usage(usage)
+
+
+def _adopt_summary_entry(entry: Any) -> dict[str, Any] | None:
+    """Conversion of arbitrary summary nodes to dicts."""
+
+    if isinstance(entry, dict):
+        return entry
+
+    if hasattr(entry, "model_dump"):
+        with contextlib.suppress(Exception):
+            data = entry.model_dump(mode="json", exclude_none=True)
+            if isinstance(data, dict):
+                return data
+        with contextlib.suppress(Exception):
+            data = entry.model_dump()
+            if isinstance(data, dict):
+                return data
+
+    if hasattr(entry, "__dict__"):
+        dict_data: dict[str, Any] = {}
+        for key in (
+            "type",
+            "text",
+            "content",
+            "signature",
+            "summary",
+            "value",
+            "delta",
+            "reasoning",
+        ):
+            if hasattr(entry, key):
+                value = getattr(entry, key)
+                if value is not None:
+                    dict_data[key] = value
+        if dict_data:
+            return dict_data
+    return None
+
+
+def _collect_reasoning_segments(source: Any) -> list[ThinkingSegment]:
+    if source is None:
+        return []
+
+    segments: list[ThinkingSegment] = []
+    visited: set[int] = set()
+
+    def _walk(node: Any, inherited_signature: str | None) -> None:
+        if node is None:
+            return
+
+        # if isinstance(node, str):
+        #     if node:
+        #         normalized = node.strip().lower()
+        #         if normalized and normalized not in _REASONING_SUMMARY_MODES:
+        #             segments.append(
+        #                 ThinkingSegment(thinking=node, signature=inherited_signature)
+        #             )
+        #     return
+
+        if isinstance(node, bytes | bytearray):
+            try:
+                decoded = node.decode()
+            except UnicodeDecodeError:
+                return
+            if decoded:
+                segments.append(
+                    ThinkingSegment(thinking=decoded, signature=inherited_signature)
+                )
+            return
+
+        if isinstance(node, list | tuple | set):
+            node_id = id(node)
+            if node_id in visited:
+                return
+            visited.add(node_id)
+            start_idx = len(segments)
+            current_signature = inherited_signature
+            for child in node:
+                child_data = _adopt_summary_entry(child)
+                child_type = (
+                    child_data.get("type")
+                    if isinstance(child_data, dict)
+                    else _get_attr(child, "type")
+                )
+                if child_type == "signature":
+                    candidate = None
+                    if isinstance(child_data, dict):
+                        candidate = child_data.get("text") or child_data.get(
+                            "signature"
+                        )
+                    else:
+                        candidate = _get_attr(child, "text") or _get_attr(
+                            child, "signature"
+                        )
+                    if isinstance(candidate, str) and candidate:
+                        current_signature = candidate
+                        for idx in range(start_idx, len(segments)):
+                            segments[idx] = ThinkingSegment(
+                                thinking=segments[idx].thinking,
+                                signature=current_signature,
+                            )
+                        start_idx = len(segments)
+                _walk(child, current_signature)
+            return
+
+        if isinstance(node, dict) or hasattr(node, "__dict__"):
+            current_node_id = id(node)
+            if current_node_id in visited:
+                return
+            visited.add(current_node_id)
+
+        data = _adopt_summary_entry(node)
+        if data is None:
+            text_attr = _get_attr(node, "text")
+            signature_attr = _get_attr(node, "signature")
+            type_attr = _get_attr(node, "type")
+            next_signature = inherited_signature
+            if isinstance(signature_attr, str) and signature_attr:
+                next_signature = signature_attr
+            if type_attr == "signature" and isinstance(text_attr, str) and text_attr:
+                next_signature = text_attr
+                text_attr = None
+            if isinstance(text_attr, str) and text_attr:
+                segments.append(
+                    ThinkingSegment(thinking=text_attr, signature=next_signature)
+                )
+
+            for key in ("summary", "content"):
+                nested = _get_attr(node, key)
+                if isinstance(nested, list | tuple | set):
+                    for child in nested:
+                        _walk(child, next_signature)
+                elif isinstance(nested, dict):
+                    _walk(nested, next_signature)
+            return
+
+        node_type = data.get("type")
+        text_value = data.get("text")
+        signature_value = data.get("signature")
+        content_value = data.get("content")
+        summary_value = data.get("summary")
+        reasoning_value = data.get("reasoning")
+
+        next_signature = inherited_signature
+        if isinstance(signature_value, str) and signature_value:
+            next_signature = signature_value
+
+        if node_type == "signature":
+            if isinstance(text_value, str) and text_value:
+                next_signature = text_value
+            if isinstance(content_value, list | tuple | set):
+                for child in content_value:
+                    _walk(child, next_signature)
+            return
+
+        if node_type in {"summary_group", "group"}:
+            if isinstance(content_value, list | tuple | set):
+                start_idx = len(segments)
+                current_signature = next_signature
+                for child in content_value:
+                    child_data = _adopt_summary_entry(child)
+                    child_type = (
+                        child_data.get("type")
+                        if isinstance(child_data, dict)
+                        else _get_attr(child, "type")
+                    )
+                    if child_type == "signature":
+                        candidate = None
+                        if isinstance(child_data, dict):
+                            candidate = child_data.get("text") or child_data.get(
+                                "signature"
+                            )
+                        else:
+                            candidate = _get_attr(child, "text") or _get_attr(
+                                child, "signature"
+                            )
+                        if isinstance(candidate, str) and candidate:
+                            current_signature = candidate
+                            for idx in range(start_idx, len(segments)):
+                                segments[idx] = ThinkingSegment(
+                                    thinking=segments[idx].thinking,
+                                    signature=current_signature,
+                                )
+                            start_idx = len(segments)
+                    _walk(child, current_signature)
+            return
+
+        emitted = False
+        if node_type in {"summary_text", "text", "reasoning_text"}:
+            if isinstance(text_value, str) and text_value:
+                segments.append(
+                    ThinkingSegment(thinking=text_value, signature=next_signature)
+                )
+                emitted = True
+        elif (
+            isinstance(text_value, str)
+            and text_value
+            and node_type not in {"signature"}
+        ):
+            segments.append(
+                ThinkingSegment(thinking=text_value, signature=next_signature)
+            )
+            emitted = True
+
+        value_value = data.get("value")
+        if not emitted and isinstance(value_value, str) and value_value:
+            segments.append(
+                ThinkingSegment(thinking=value_value, signature=next_signature)
+            )
+            emitted = True
+
+        if isinstance(summary_value, list | tuple | set):
+            for child in summary_value:
+                _walk(child, next_signature)
+        elif isinstance(summary_value, dict):
+            _walk(summary_value, next_signature)
+
+        if isinstance(content_value, list | tuple | set):
+            for child in content_value:
+                _walk(child, next_signature)
+        elif isinstance(content_value, dict):
+            _walk(content_value, next_signature)
+
+        if isinstance(reasoning_value, list | tuple | set | dict):
+            _walk(reasoning_value, next_signature)
+
+    _walk(source, None)
+    return merge_thinking_segments(segments)
+
+
+def _wrap_thinking(signature: str | None, text: str) -> str:
+    """Serialize a reasoning block into <thinking> XML."""
+    return ThinkingSegment(thinking=text, signature=signature).to_xml()
+
+
+def _extract_reasoning_blocks(payload: Any) -> list[ThinkingSegment]:
+    """Extract reasoning blocks from a response output payload."""
+
+    if not payload:
+        return []
+
+    summary = _get_attr(payload, "summary")
+    segments = _collect_reasoning_segments(summary)
+    if segments:
+        return segments
+
+    if isinstance(payload, list | tuple | set):
+        segments = _collect_reasoning_segments(payload)
+        if segments:
+            return segments
+
+    text_value = _get_attr(payload, "text")
+    if isinstance(text_value, str) and text_value:
+        return [ThinkingSegment(thinking=text_value)]
+
+    if isinstance(payload, dict):
+        raw = payload.get("reasoning")
+        if raw:
+            return _extract_reasoning_blocks(raw)
+
+    return []
+
+
+def _split_content_segments(content: str) -> list[tuple[str, Any]]:
+    """Split mixed assistant content into ordered text vs thinking blocks."""
+
+    if not content:
+        return []
+
+    segments: list[tuple[str, Any]] = []
+    last_idx = 0
+    for match in THINKING_PATTERN.finditer(content):
+        start, end = match.span()
+        if start > last_idx:
+            text_segment = content[last_idx:start]
+            if text_segment:
+                segments.append(("text", text_segment))
+        signature = match.group(1) or None
+        thinking_text = match.group(2) or ""
+        segments.append(
+            ("thinking", ThinkingSegment.from_xml(signature, thinking_text))
+        )
+        last_idx = end
+
+    if last_idx < len(content):
+        tail = content[last_idx:]
+        if tail:
+            segments.append(("text", tail))
+
+    if not segments:
+        segments.append(("text", content))
+    return segments
+
+
+def convert__openai_responses_to_openai_chat__response(
+    response: openai_models.ResponseObject,
+) -> openai_models.ChatCompletionResponse:
+    """Convert an OpenAI ResponseObject to a ChatCompletionResponse."""
+    text_segments: list[str] = []
+    added_reasoning: set[tuple[str, str]] = set()
+    tool_calls: list[openai_models.ToolCall] = []
+
+    for item in response.output or []:
+        logger.debug(
+            "convert_responses_to_chat_response_item", item_type=_get_attr(item, "type")
+        )
+        item_type = _get_attr(item, "type")
+        if item_type == "reasoning":
+            for segment in _extract_reasoning_blocks(item):
+                signature = segment.signature
+                thinking_text = segment.thinking
+                logger.debug(
+                    "convert_responses_to_chat_reasoning_block",
+                    signature=signature,
+                    text_snippet=(thinking_text[:30] + "...")
+                    if thinking_text and len(thinking_text) > 30
+                    else thinking_text,
+                )
+                if thinking_text:
+                    key = (signature or "", thinking_text)
+                    if key not in added_reasoning:
+                        text_segments.append(_wrap_thinking(signature, thinking_text))
+                        added_reasoning.add(key)
+        elif item_type == "message":
+            parts: list[str] = []
+            content_list = _get_attr(item, "content")
+            if isinstance(content_list, list):
+                for part in content_list:
+                    part_type = _get_attr(part, "type")
+                    if part_type == "output_text":
+                        text_val = _get_attr(part, "text")
+                        if isinstance(text_val, str):
+                            parts.append(text_val)
+                    elif isinstance(part, str):
+                        parts.append(part)
+            elif isinstance(content_list, str):
+                parts.append(content_list)
+            if parts:
+                text_segments.append("".join(parts))
+        elif item_type == "function_call":
+            function_block = _get_attr(item, "function")
+            name = _get_attr(function_block, "name") or _get_attr(item, "name")
+            arguments_value: Any = _get_attr(item, "arguments")
+            if arguments_value is None and isinstance(function_block, dict):
+                arguments_value = function_block.get("arguments")
+
+            if not isinstance(name, str) or not name:
+                continue
+
+            if isinstance(arguments_value, dict):
+                arguments_str = json.dumps(arguments_value)
+            elif isinstance(arguments_value, str):
+                arguments_str = arguments_value
+            else:
+                arguments_str = json.dumps(arguments_value or {})
+
+            tool_calls.append(
+                openai_models.ToolCall(
+                    id=_get_attr(item, "id")
+                    or _get_attr(item, "call_id")
+                    or f"call_{len(tool_calls)}",
+                    type="function",
+                    function=openai_models.FunctionCall(
+                        name=name,
+                        arguments=arguments_str,
+                    ),
+                )
+            )
+
+    text_content = "".join(text_segments)
+
+    usage = None
+    if response.usage:
+        usage = convert__openai_responses_usage_to_openai_completion__usage(
+            response.usage
+        )
+
+    finish_reason: Literal["stop", "length", "tool_calls", "content_filter"] = (
+        "tool_calls" if tool_calls else "stop"
+    )
+
+    return openai_models.ChatCompletionResponse(
+        id=response.id or "chatcmpl-resp",
+        choices=[
+            openai_models.Choice(
+                index=0,
+                message=openai_models.ResponseMessage(
+                    role="assistant",
+                    content=text_content,
+                    tool_calls=tool_calls or None,
+                ),
+                finish_reason=finish_reason,
+            )
+        ],
+        created=0,
+        model=response.model or "",
+        object="chat.completion",
+        usage=usage
+        or openai_models.CompletionUsage(
+            prompt_tokens=0, completion_tokens=0, total_tokens=0
+        ),
+    )
+
+
+async def convert__openai_chat_to_openai_responses__response(
+    chat_response: openai_models.ChatCompletionResponse,
+) -> openai_models.ResponseObject:
+    content_text = ""
+    tool_calls: list[Any] = []
+    if chat_response.choices:
+        first_choice = chat_response.choices[0]
+        if first_choice.message:
+            content = first_choice.message.content
+            if content:
+                if isinstance(content, str):
+                    content_text = content
+                elif isinstance(content, list):
+                    # Handle list content - convert to string
+                    content_text = str(content)
+                else:
+                    content_text = str(content)
+            if first_choice.message.tool_calls:
+                tool_calls = list(first_choice.message.tool_calls)
+
+    segments = _split_content_segments(content_text)
+
+    outputs: list[Any] = []
+    reasoning_entries: list[Any] = []
+    message_buffer: list[str] = []
+    message_counter = 0
+
+    def flush_message() -> None:
+        nonlocal message_buffer, message_counter
+        if not message_buffer:
+            return
+        message_text = "".join(message_buffer)
+        message_buffer = []
+        message_id = f"msg_{chat_response.id or 'unknown'}_{message_counter}"
+        message_counter += 1
+        outputs.append(
+            openai_models.MessageOutput(
+                type="message",
+                role="assistant",
+                id=message_id,
+                status="completed",
+                content=[
+                    openai_models.OutputTextContent(
+                        type="output_text", text=message_text
+                    )
+                ],
+            )
+        )
+
+    for segment in segments or [("text", "")]:
+        if not segment:
+            continue
+        kind = segment[0]
+        if kind == "text":
+            text_part = segment[1]
+            if isinstance(text_part, str) and text_part:
+                message_buffer.append(text_part)
+        elif kind == "thinking":
+            segment_value = segment[1]
+            if isinstance(segment_value, ThinkingSegment):
+                signature = segment_value.signature
+                thinking_text = segment_value.thinking
+            else:
+                signature = None
+                thinking_text = ""
+            flush_message()
+            summary_entry: dict[str, Any] = {
+                "type": "summary_text",
+                "text": thinking_text,
+            }
+            if signature:
+                summary_entry["signature"] = signature
+            reasoning_id = (
+                f"reasoning_{chat_response.id or 'unknown'}_{len(reasoning_entries)}"
+            )
+            reasoning_output = openai_models.ReasoningOutput(
+                type="reasoning",
+                id=reasoning_id,
+                status="completed",
+                summary=[summary_entry],
+            )
+            outputs.append(reasoning_output)
+            reasoning_entries.append(reasoning_output)
+
+    # Flush any remaining assistant text
+    flush_message()
+
+    if not outputs:
+        outputs.append(
+            openai_models.MessageOutput(
+                type="message",
+                role="assistant",
+                id=f"msg_{chat_response.id or 'unknown'}_0",
+                status="completed",
+                content=[openai_models.OutputTextContent(type="output_text", text="")],
+            )
+        )
+
+    if tool_calls:
+        for idx, tool_call in enumerate(tool_calls):
+            fn = getattr(tool_call, "function", None)
+            name = _get_attr(fn, "name") or _get_attr(tool_call, "name") or ""
+            arguments = _get_attr(fn, "arguments") or _get_attr(tool_call, "arguments")
+            if isinstance(arguments, dict):
+                arguments_value: str | dict[str, Any] | None = arguments
+            else:
+                arguments_value = str(arguments) if arguments is not None else None
+            outputs.append(
+                openai_models.FunctionCallOutput(
+                    type="function_call",
+                    id=getattr(tool_call, "id", f"call_{idx}"),
+                    status="completed",
+                    name=name,
+                    call_id=getattr(tool_call, "id", None),
+                    arguments=arguments_value,
+                )
+            )
+
+    reasoning_summary = []
+    for entry in reasoning_entries:
+        summary_list = _get_attr(entry, "summary")
+        if isinstance(summary_list, list):
+            reasoning_summary.extend(summary_list)
+
+    usage: openai_models.ResponseUsage | None = None
+    if chat_response.usage:
+        usage = convert__openai_completion_usage_to_openai_responses__usage(
+            chat_response.usage
+        )
+
+    return openai_models.ResponseObject(
+        id=chat_response.id or "resp-unknown",
+        object="response",
+        created_at=int(time.time()),
+        model=chat_response.model or "",
+        status="completed",
+        output=outputs,
+        parallel_tool_calls=False,
+        usage=usage,
+        reasoning=(
+            openai_models.Reasoning(summary=reasoning_summary)
+            if reasoning_summary
+            else None
+        ),
+    )
+
+
+__all__ = [
+    "convert__openai_chat_to_openai_responses__response",
+    "convert__openai_completion_usage_to_openai_responses__usage",
+    "convert__openai_responses_to_openai_chat__response",
+    "convert__openai_responses_usage_to_openai_completion__usage",
+]