@scemoon/cdh 1.0.6 → 1.0.8

package/onecode/models/providers/minimaxi_provider.py

@@ -1,417 +0,0 @@
-from __future__ import annotations
-
-import json
-import logging
-import time
-from datetime import datetime
-from pathlib import Path
-from typing import AsyncIterator, Callable, Optional
-
-import httpx
-
-from onecode.models.errors import (
-    ProviderError, TransientProviderError, retry_after_seconds, safe_error_msg,
-)
-from onecode.models.provider import ChatResponse, Message, ModelResponse, Provider, ProviderRegistry
-
-logger = logging.getLogger("onecode.provider.minimaxi")
-
-LOG_DIR = Path.home() / ".cdh" / "logs"
-LOG_DIR.mkdir(parents=True, exist_ok=True)
-
-
-def _raise_for_status(resp: httpx.Response, body_text: str = "") -> None:
-    """Raise a :class:`ProviderError` subclass if ``resp.status_code`` is not 2xx.
-
-    Falls back to :class:`TransientProviderError` for 5xx, and to the
-    specific :class:`RateLimitError` / :class:`AuthError` /
-    :class:`ContextLengthError` subclasses when the body matches.
-    Callers *must* provide ``body_text`` for streaming responses (e.g.
-    after calling ``await resp.aread()``).
-    """
-    if 200 <= resp.status_code < 300:
-        return
-    if not body_text:
-        body_text = resp.text
-    raise Provider.classify_http_error(
-        resp.status_code,
-        body_text,
-        retry_after=retry_after_seconds(resp),
-    )
-
-
-def _accumulate_tool_call_delta(
-    slot: dict, tcd: dict
-) -> None:
-    """Fold a single OpenAI-style ``delta.tool_calls`` entry into ``slot``.
-
-    Extracted so the streaming parser and unit tests can share one
-    implementation.  Mutates ``slot`` in place.
-    """
-    if tcd.get("id"):
-        slot["id"] = tcd["id"]
-    fn = tcd.get("function", {}) or {}
-    if fn.get("name"):
-        slot["name"] = fn["name"]
-    slot["arguments"] += fn.get("arguments", "") or ""
-
-
-def _finalize_stream_tool_calls(
-    stream_tool_calls: dict[int, dict],
-) -> list[dict]:
-    """Convert accumulated OpenAI-style tool-call deltas into engine ``tool_uses``.
-
-    Each entry becomes ``{id, name, input}`` where ``input`` is the
-    ``arguments`` JSON string parsed back to a dict.  Entries without
-    a ``name`` are dropped (incomplete deltas at stream end).
-    """
-    tool_uses: list[dict] = []
-    for slot in stream_tool_calls.values():
-        name = slot.get("name") or ""
-        if not name:
-            continue
-        raw_args = slot.get("arguments") or ""
-        try:
-            inp: dict = json.loads(raw_args) if raw_args else {}
-        except (json.JSONDecodeError, TypeError):
-            inp = {"raw": raw_args}
-        if not isinstance(inp, dict):
-            inp = {"raw": inp}
-        tool_uses.append(
-            {"id": slot.get("id") or "", "name": name, "input": inp}
-        )
-    return tool_uses
-
-
-
-def _log_request(model: str, messages: list, response_data: dict = None, error: str = None):
-    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S_%f")
-    log_file = LOG_DIR / f"minimaxi_{timestamp}.json"
-    log_entry = {
-        "timestamp": timestamp,
-        "model": model,
-        "messages": [{"role": m.role, "content": m.to_api_content()} for m in messages],
-        "response": response_data,
-        "error": error,
-    }
-    try:
-        with open(log_file, "w") as f:
-            json.dump(log_entry, f, indent=2, ensure_ascii=False)
-        logger.info(f"Request logged to {log_file}")
-    except Exception as e:
-        logger.error(f"Failed to write log: {e}")
-    return log_file
-
-
-class MiniMaxiProvider(Provider):
-    name = "minimaxi"
-
-    def __init__(self, api_key: str = "", endpoint: str = "", **kwargs):
-        self.api_key = api_key
-        self._endpoint = endpoint or "https://api.minimaxi.com/v1"
-        self._request_count = 0
-
-    def _pick_default_model(self, requested: str) -> str:
-        """Resolve the effective default model.
-
-        ``cn-cp-`` prefixed keys belong to the ``minimax-cn-coding-plan``
-        tier and must use ``MiniMax-M3``.  Any other key keeps the
-        requested model (or the historical ``MiniMax-M2.7`` fallback).
-        """
-        if requested and requested != "MiniMax-M2.7":
-            return requested
-        key = (self.api_key or "").strip()
-        if key.startswith("cn-cp-"):
-            return "MiniMax-M3"
-        return requested or "MiniMax-M2.7"
-
-    def is_anthropic_style(self) -> bool:
-        return False
-
-    def supports_native_tools(self) -> bool:
-        return False
-
-    async def chat(
-        self, messages: list[Message], model: str = "", **kwargs
-    ) -> ModelResponse:
-        model = self._pick_default_model(model)
-        self._request_count += 1
-        req_id = f"req_{self._request_count}_{int(time.time())}"
-        logger.info(f"[{req_id}] chat() called with model={model}, messages={len(messages)}")
-
-        key = self.resolve_api_key(self.api_key)
-        logger.info(f"[{req_id}] API key resolved: {'Yes' if key else 'No'}, key_prefix: '{key[:15]}...' if key else 'None'")
-
-        if not key:
-            error_msg = "API key not configured. Set MINMAXI_API_KEY."
-            _log_request(model, messages, error=error_msg)
-            raise ProviderError(error_msg)
-
-        start_time = time.time()
-        try:
-            async with httpx.AsyncClient(timeout=120) as client:
-                logger.info(f"[{req_id}] Sending request to {self._endpoint}/chat/completions")
-                outgoing_payload = {
-                    "model": model,
-                    "messages": self.prepare_messages(messages),
-                    **kwargs,
-                }
-                # Debug-only: log the post-`prepare_messages` wire payload
-                # so we can see exactly which messages reach the gateway
-                # when the upstream returns (2013) or any other 400.
-                # Disabled at INFO/WARN — set log_level: debug to enable.
-                logger.debug(
-                    "[%s] OUTGOING PAYLOAD (chat):\n%s",
-                    req_id,
-                    json.dumps(outgoing_payload, ensure_ascii=False, indent=2),
-                )
-                resp = await client.post(
-                    f"{self._endpoint}/chat/completions",
-                    headers={
-                        "Authorization": f"Bearer {key}",
-                        "content-type": "application/json",
-                    },
-                    json=outgoing_payload,
-                )
-                elapsed = time.time() - start_time
-                logger.info(f"[{req_id}] Response status: {resp.status_code}, elapsed: {elapsed:.2f}s")
-
-                if resp.status_code != 200:
-                    error_text = resp.text
-                    logger.error(f"[{req_id}] Non-2xx HTTP {resp.status_code}: {error_text[:500]}")
-                    _log_request(model, messages, error=f"HTTP {resp.status_code}: {error_text[:500]}")
-                    classified = Provider.classify_http_error(
-                        resp.status_code, error_text,
-                        retry_after=retry_after_seconds(resp),
-                    )
-                    body_excerpt = (error_text or "").strip()[:200]
-                    if body_excerpt and classified.__class__.__name__ == "ProviderError":
-                        classified.args = (f"{classified.args[0]}: {body_excerpt}",)
-                    raise classified
-
-                data = resp.json()
-                choice = data.get("choices", [{}])[0]
-                content = choice.get("message", {}).get("content", "")
-
-                _log_request(model, messages, response_data=data)
-                logger.info(f"[{req_id}] Response content length: {len(content)}")
-
-                return ModelResponse(
-                    content=[{"type": "text", "text": content}],
-                    model=model,
-                    usage=data.get("usage", {}),
-                    raw=data,
-                )
-        except ProviderError:
-            raise
-        except httpx.ConnectError as e:
-            error_msg = f"Connection error: {e}"
-            logger.error(f"[{req_id}] {error_msg}")
-            _log_request(model, messages, error=error_msg)
-            raise TransientProviderError(error_msg) from e
-        except Exception as e:
-            error_msg = f"Error: {safe_error_msg(e)}"
-            logger.exception(f"[{req_id}] {error_msg}")
-            _log_request(model, messages, error=error_msg)
-            raise TransientProviderError(error_msg) from e
-
-    async def chat_stream_response(
-        self,
-        messages: list[Message],
-        model: str,
-        on_text_chunk: Optional[Callable[[str], None]] = None,
-        on_tool_call_delta: Optional[Callable[[str, str, str], None]] = None,
-        **kwargs,
-    ) -> ChatResponse:
-        self._request_count += 1
-        req_id = f"req_{self._request_count}_{int(time.time())}"
-        logger.info(f"[{req_id}] chat_stream_response() called with model={model}, messages={len(messages)}")
-
-        key = self.resolve_api_key(self.api_key)
-        if not key:
-            raise ProviderError("API key not configured.")
-
-        start_time = time.time()
-        content_parts: list[str] = []
-        # OpenAI-compatible tool calls are streamed as a sequence of
-        # ``delta.tool_calls`` deltas.  Accumulate by ``index`` so the
-        # final ``arguments`` string can be parsed back into a dict
-        # that the engine passes to ``_build_tool_call_content``.
-        stream_tool_calls: dict[int, dict] = {}
-
-        try:
-            async with httpx.AsyncClient(timeout=300) as client:
-                outgoing_payload = {
-                    "model": model,
-                    "messages": self.prepare_messages(messages),
-                    "stream": True,
-                    **kwargs,
-                }
-                # Debug-only: log the post-`prepare_messages` wire payload
-                # so we can see exactly which messages reach the gateway
-                # when the upstream returns (2013) or any other 400.
-                # Disabled at INFO/WARN — set log_level: debug to enable.
-                logger.debug(
-                    "[%s] OUTGOING PAYLOAD (chat_stream_response):\n%s",
-                    req_id,
-                    json.dumps(outgoing_payload, ensure_ascii=False, indent=2),
-                )
-                async with client.stream(
-                    "POST",
-                    f"{self._endpoint}/chat/completions",
-                    headers={
-                        "Authorization": f"Bearer {key}",
-                        "Content-Type": "application/json",
-                    },
-                    json=outgoing_payload,
-                ) as resp:
-                    if resp.status_code >= 400:
-                        error_body = (await resp.aread()).decode("utf-8", errors="replace")
-                        _raise_for_status(resp, body_text=error_body)
-                    async for line in resp.aiter_lines():
-                        if line.startswith("data: ") and line[6:] != "[DONE]":
-                            try:
-                                chunk = json.loads(line[6:])
-                                delta = chunk.get("choices", [{}])[0].get("delta", {})
-                                content = delta.get("content", "")
-                                if content:
-                                    content_parts.append(content)
-                                    if on_text_chunk:
-                                        on_text_chunk(content)
-                                for tcd in delta.get("tool_calls", []) or []:
-                                    idx = tcd.get("index", 0)
-                                    slot = stream_tool_calls.setdefault(
-                                        idx,
-                                        {"id": "", "name": "", "arguments": ""},
-                                    )
-                                    _accumulate_tool_call_delta(slot, tcd)
-                                    if on_tool_call_delta:
-                                        call_id = slot.get("id", "")
-                                        name = slot.get("name", "")
-                                        args_delta = tcd.get("function", {}).get("arguments", "") or ""
-                                        on_tool_call_delta(call_id, name, args_delta)
-                            except json.JSONDecodeError:
-                                continue
-                        elif line.startswith("data: [DONE]"):
-                            pass
-
-            elapsed = time.time() - start_time
-            logger.info(
-                f"[{req_id}] Stream complete, elapsed: {elapsed:.2f}s, "
-                f"chars: {sum(len(p) for p in content_parts)}, "
-                f"tool_calls: {len(stream_tool_calls)}"
-            )
-        except ProviderError:
-            # Already classified — propagate verbatim.
-            _log_request(model, messages, error="provider error (see exception)")
-            raise
-        except httpx.ConnectError as e:
-            error_msg = f"Connection error: {e}"
-            logger.error(f"[{req_id}] {error_msg}")
-            _log_request(model, messages, error=error_msg)
-            raise TransientProviderError(error_msg) from e
-        except Exception as e:
-            error_msg = f"Error: {safe_error_msg(e)}"
-            logger.exception(f"[{req_id}] {error_msg}")
-            _log_request(model, messages, error=error_msg)
-            raise TransientProviderError(error_msg) from e
-
-        tool_uses = _finalize_stream_tool_calls(stream_tool_calls)
-
-        _log_request(model, messages, response_data={"stream": True})
-        return ChatResponse(content="".join(content_parts), tool_uses=tool_uses)
-
-    async def chat_stream(
-        self, messages: list[Message], model: str = "", **kwargs
-    ) -> AsyncIterator[str]:
-        model = self._pick_default_model(model)
-        self._request_count += 1
-        req_id = f"req_{self._request_count}_{int(time.time())}"
-        logger.info(f"[{req_id}] chat_stream() called with model={model}, messages={len(messages)}")
-
-        key = self.resolve_api_key(self.api_key)
-        logger.info(f"[{req_id}] API key resolved: {'Yes' if key else 'No'}, key_prefix: '{key[:15]}...' if key else 'None'")
-
-        if not key:
-            error_msg = "API key not configured."
-            _log_request(model, messages, error=error_msg)
-            raise ProviderError(error_msg)
-
-        start_time = time.time()
-        full_content = []
-        response_data = None
-
-        try:
-            async with httpx.AsyncClient(timeout=300) as client:
-                logger.info(f"[{req_id}] Sending stream request to {self._endpoint}/chat/completions")
-                logger.info(f"[{req_id}] Request: model={model}, messages_count={len(messages)}")
-
-                outgoing_payload = {
-                    "model": model,
-                    "messages": self.prepare_messages(messages),
-                    "stream": True,
-                }
-                # Debug-only: same wire-payload log as chat() — see comment above.
-                logger.debug(
-                    "[%s] OUTGOING PAYLOAD (chat_stream):\n%s",
-                    req_id,
-                    json.dumps(outgoing_payload, ensure_ascii=False, indent=2),
-                )
-                async with client.stream(
-                    "POST",
-                    f"{self._endpoint}/chat/completions",
-                    headers={
-                        "Authorization": f"Bearer {key}",
-                        "Content-Type": "application/json",
-                    },
-                    json=outgoing_payload,
-                ) as resp:
-                    logger.info(f"[{req_id}] Stream response status: {resp.status_code}")
-                    if resp.status_code != 200:
-                        error_body = await resp.aread()
-                        error_text = error_body.decode("utf-8", errors="replace")
-                        logger.error(f"[{req_id}] Stream error HTTP {resp.status_code}: {error_text[:1000]}")
-                        _log_request(model, messages, error=f"HTTP {resp.status_code}: {error_text[:500]}")
-                        classified = Provider.classify_http_error(
-                            resp.status_code, error_text,
-                            retry_after=retry_after_seconds(resp),
-                        )
-                        body_excerpt = error_text.strip()[:200]
-                        if body_excerpt and classified.__class__.__name__ == "ProviderError":
-                            classified.args = (f"{classified.args[0]}: {body_excerpt}",)
-                        raise classified
-
-                    async for line in resp.aiter_lines():
-                        if line.startswith("data: ") and line[6:] != "[DONE]":
-                            try:
-                                chunk = json.loads(line[6:])
-                                delta = chunk.get("choices", [{}])[0].get("delta", {})
-                                content = delta.get("content", "")
-                                if content:
-                                    full_content.append(content)
-                                    yield content
-                            except json.JSONDecodeError as e:
-                                logger.warning(f"[{req_id}] JSON decode error: {e}, line: {line[:100]}")
-                        elif line.startswith("data: [DONE]"):
-                            pass
-
-                elapsed = time.time() - start_time
-                logger.info(f"[{req_id}] Stream complete, elapsed: {elapsed:.2f}s, chars: {len(full_content)}")
-                response_data = {"stream": True, "full_content_length": len(full_content)}
-        except ProviderError:
-            raise
-        except httpx.ConnectError as e:
-            error_msg = f"Connection error: {e}"
-            logger.error(f"[{req_id}] {error_msg}")
-            _log_request(model, messages, error=error_msg)
-            raise TransientProviderError(error_msg) from e
-        except Exception as e:
-            error_msg = f"Error: {safe_error_msg(e)}"
-            logger.exception(f"[{req_id}] {error_msg}")
-            _log_request(model, messages, error=error_msg)
-            raise TransientProviderError(error_msg) from e
-
-        _log_request(model, messages, response_data=response_data)
-
-
-ProviderRegistry.register("minimaxi", MiniMaxiProvider)

package/onecode/models/providers/ollama_provider.py

@@ -1,100 +0,0 @@
-from __future__ import annotations
-
-import json
-from typing import AsyncIterator, Callable, Optional
-
-import httpx
-
-from onecode.models.errors import safe_error_msg
-from onecode.models.provider import ChatResponse, Message, ModelResponse, Provider, ProviderRegistry
-
-
-class OllamaProvider(Provider):
-    name = "ollama"
-
-    def __init__(self, endpoint: str = "http://localhost:11434", **kwargs):
-        self.endpoint = endpoint.rstrip("/")
-
-    def is_anthropic_style(self) -> bool:
-        return False
-
-    def supports_native_tools(self) -> bool:
-        return False
-
-    async def chat(
-        self, messages: list[Message], model: str = "llama2", **kwargs
-    ) -> ModelResponse:
-        async with httpx.AsyncClient(timeout=120) as client:
-            resp = await client.post(
-                f"{self.endpoint}/api/chat",
-                json={
-                    "model": model,
-                    "messages": self.prepare_messages(messages),
-                    "stream": False,
-                },
-            )
-            data = resp.json()
-            return ModelResponse(
-                content=[{"type": "text", "text": data.get("message", {}).get("content", "")}],
-                model=model,
-                usage={"total_tokens": data.get("eval_count", 0)},
-                raw=data,
-            )
-
-    async def chat_stream(
-        self, messages: list[Message], model: str = "llama2", **kwargs
-    ) -> AsyncIterator[str]:
-        async with httpx.AsyncClient(timeout=300) as client:
-            async with client.stream(
-                "POST",
-                f"{self.endpoint}/api/chat",
-                json={
-                    "model": model,
-                    "messages": self.prepare_messages(messages),
-                    "stream": True,
-                },
-            ) as resp:
-                async for line in resp.aiter_lines():
-                    if line.strip():
-                        import json
-                        chunk = json.loads(line)
-                        yield chunk.get("message", {}).get("content", "")
-                        if chunk.get("done"):
-                            break
-
-    async def chat_stream_response(
-        self,
-        messages: list[Message],
-        model: str,
-        on_text_chunk: Optional[Callable[[str], None]] = None,
-        on_tool_call_delta: Optional[Callable[[str, str, str], None]] = None,
-        **kwargs,
-    ) -> ChatResponse:
-        content_parts: list[str] = []
-        try:
-            async with httpx.AsyncClient(timeout=300) as client:
-                async with client.stream(
-                    "POST",
-                    f"{self.endpoint}/api/chat",
-                    json={
-                        "model": model,
-                        "messages": self.prepare_messages(messages),
-                        "stream": True,
-                    },
-                ) as resp:
-                    async for line in resp.aiter_lines():
-                        if line.strip():
-                            chunk = json.loads(line)
-                            text = chunk.get("message", {}).get("content", "")
-                            if text:
-                                content_parts.append(text)
-                                if on_text_chunk:
-                                    on_text_chunk(text)
-                            if chunk.get("done"):
-                                break
-        except Exception as e:
-            return ChatResponse(content=f"Error: {safe_error_msg(e)}")
-        return ChatResponse(content="".join(content_parts))
-
-
-ProviderRegistry.register("ollama", OllamaProvider)