cc2go 0.7.3__py3-none-any.whl

cc2go-0.7.3.dist-info/METADATA

@@ -0,0 +1,194 @@
+Metadata-Version: 2.4
+Name: cc2go
+Version: 0.7.3
+Summary: Claude Code to OpenCode Go adapter — route Anthropic format to OpenAI models
+Author: lzg14
+License: MIT
+Project-URL: Homepage, https://github.com/lzg14/cc2go
+Project-URL: Source, https://github.com/lzg14/cc2go
+Project-URL: BugTracker, https://github.com/lzg14/cc2go/issues
+Keywords: claude-code,opencode,llm-proxy,ai-adapter,anthropic,openai
+Classifier: Development Status :: 4 - Beta
+Classifier: Environment :: Web Environment
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Internet :: Proxy Servers
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: fastapi
+Requires-Dist: uvicorn
+Requires-Dist: httpx
+Requires-Dist: python-dotenv
+Requires-Dist: pystray>=0.19.0
+Requires-Dist: Pillow>=10.0.0
+Dynamic: license-file
+
+# cc2go <small>v0.7.3</small>
+
+<p align="center">
+  <b>Claude Code → OpenCode Go Adapter</b>
+  <br>
+  A lightweight proxy that lets Claude Code use any OpenAI-compatible model
+</p>
+
+<p align="center">
+  <img src="https://img.shields.io/badge/python-3.9%2B-blue" alt="Python">
+  <img src="https://img.shields.io/github/license/lzg14/cc2go" alt="License">
+  <img src="https://img.shields.io/github/v/release/lzg14/cc2go" alt="Release">
+  <img src="https://img.shields.io/github/stars/lzg14/cc2go" alt="Stars">
+  <img src="https://img.shields.io/github/actions/workflow/status/lzg14/cc2go/ci.yml?branch=master" alt="CI">
+  <img src="https://img.shields.io/pypi/v/cc2go" alt="PyPI">
+</p>
+
+<p align="center">
+  <img src="static/screenshot.png" alt="cc2go Web UI" width="800">
+  <br>
+  <em>Web admin page — switch models, add endpoints, view logs</em>
+</p>
+
+## Features
+
+- 🔄 **Protocol translation** — Converts Anthropic Messages API ↔ OpenAI Chat Completions
+- 🌐 **Web UI** — Built-in admin page at `http://localhost:4001`, click to switch models
+- 🎯 **Model switching** — Click to switch models, auto-syncs to Claude Code settings
+- ➕ **Custom models** — Add your own endpoints with independent API keys and URLs
+- 🖼️ **Image support** — Converts Anthropic image blocks to OpenAI image_url format
+- ⚡ **Streaming** — Real-time SSE streaming conversion (OpenAI → Anthropic format)
+- 🔁 **Adaptive retry** — Error classification with exponential backoff, max 3 retries
+- 📋 **Log management** — Built-in log viewer with rotation (5MB per file, 3 backups)
+- 🖥️ **System tray** — Tray icon for opening admin page and quitting; auto-opens admin on start
+- 💰 **Token saving** — Strips `<system-reminder>`, `[思考过程]` reasoning, and `thinking` blocks before forwarding upstream
+- 💾 **Config backup** — Auto-backup original Claude Code config on first model switch; one-click restore from admin UI
+- 🔑 **Security** — Defaults to 127.0.0.1 (local-only), Bearer Token authentication on proxy endpoints
+- 🔧 **Tool name sanitization** — Replaces special characters in tool names to prevent 400 errors
+- 🧹 **Schema cleaning** — Recursively removes incompatible JSON Schema fields for broader model compatibility
+
+---
+
+## Getting Started (For Non-Technical Users)
+
+### Step 1: Download
+
+Download the latest release from [GitHub Releases](https://github.com/lzg14/cc2go/releases).
+Extract the ZIP file to any folder (not Program Files).
+
+### Step 2: Configure API Key
+
+1. Open the extracted folder
+2. Copy `.env.example` to `.env` (or create a new file named `.env`)
+3. Open `.env` with a text editor (like Notepad) and add your OpenCode Go API key:
+   ```
+   OPENCODE_API_KEY=your_api_key_here
+   ```
+
+### Step 3: Run
+
+Double-click `start_bg.bat` in the folder.
+
+A system tray icon will appear. The admin page will open automatically in your browser.
+
+### Step 4: Open in Claude Code
+
+In Claude Code's settings, configure:
+
+| Setting | Value |
+|---------|-------|
+| Base URL | `http://localhost:4001` |
+| API Key | `sk-cc2go-local` |
+
+That's it! Use Claude Code as normal — select models from the web admin page at `http://localhost:4001`.
+
+### How to Quit
+
+Right-click the system tray icon → click "Exit".
+
+---
+
+## For Developers
+
+```bash
+pip install -r requirements.txt
+cp .env.example .env   # Edit .env with your API key
+python src/router.py
+```
+
+### Scripts (Windows)
+
+```
+scripts\start_bg.bat    # Background mode (system tray, no terminal)
+scripts\stop.bat        # Stop background process
+```
+
+---
+
+## Docker
+
+```bash
+docker build -t cc2go .
+docker run -d -p 4001:4001 --env-file .env cc2go
+```
+
+Access at `http://localhost:4001`.
+
+## Linux Service (systemd)
+
+For permanent deployment on Linux:
+
+```bash
+# Copy service file
+sudo cp cc2go.service /etc/systemd/system/
+sudo systemctl daemon-reload
+sudo systemctl enable cc2go
+sudo systemctl start cc2go
+```
+
+Requires a `cc2go` user and `/opt/cc2go` installation directory.
+
+---
+
+## API Endpoints
+
+| Endpoint | Method | Description |
+|----------|--------|-------------|
+| `/` | GET | Web admin UI |
+| `/v1/messages` | POST | Claude format entry (Anthropic → OpenAI) |
+| `/v1/chat/completions` | POST | OpenAI format passthrough |
+| `/v1/models` | GET | List available models |
+| `/health` | GET | Health check |
+| `/api/config` | GET/PUT | Configuration management |
+| `/api/custom-models` | GET/PUT | Custom model management |
+| `/api/logs` | GET | Recent log entries |
+
+---
+
+## Custom Model Routing
+
+| Endpoint | Behavior |
+|----------|----------|
+| `/v1/messages` | Anthropic format passthrough (no conversion, thinking blocks preserved) |
+| `/v1/chat/completions` | OpenAI format conversion (tool_calls gets reasoning_content="" if missing) |
+
+---
+
+## Configuration
+
+All configuration via Web UI (`http://localhost:4001`):
+- **Connection** — OpenCode Go base URL and API key
+- **Service** — Host, port, master key (auto-syncs to Claude Code)
+- **Custom Models** — Add/edit/remove custom model endpoints
+- **Logs** — View logs, set log level, toggle detailed logging
+
+---
+
+## License
+
+MIT
+
+> Technical details in [ARCHITECTURE.md](ARCHITECTURE.md)

cc2go-0.7.3.dist-info/RECORD

@@ -0,0 +1,11 @@
+error_handler.py,sha256=-96Xz4YbxlnIpIo3QFp7zkA_WRYMHX-vm2OGqsP9_i8,4854
+mcp_bypass.py,sha256=S8JQuNu7K24d-Fk2Vdl07uNsL5EsjHEEYQn5M6KqcZU,7015
+router.py,sha256=3RLgkAlEGY8dopbLBnnuIovKhTfjfg-XNi1-yTKUX4A,54950
+streaming.py,sha256=UqY49q-Z-OoikOSLdj5hZK71KLSbK4ezr2WOF7bMspM,6822
+tray.py,sha256=1xrXQf9yNEuzLcammlLJydIW56SGR-fc9FZoNU5dNbM,5825
+cc2go-0.7.3.dist-info/licenses/LICENSE,sha256=uAClLHfIrJJKFrvTQTb-KHf0d--sHqoa968MNnA9iWY,1077
+cc2go-0.7.3.dist-info/METADATA,sha256=qaZYe-Zy6PATGyzzf-TiPcFeE_bPmosDVZ6j6qc4zzc,6597
+cc2go-0.7.3.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
+cc2go-0.7.3.dist-info/entry_points.txt,sha256=A4BMVruTUktAGpSWnJjHgrm5lJS5BetQ4bYM42gStBo,38
+cc2go-0.7.3.dist-info/top_level.txt,sha256=54MV57l6iulpEPmpuR9EcDNEKNnNebv1TePNnwY7tCI,47
+cc2go-0.7.3.dist-info/RECORD,,

cc2go-0.7.3.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (82.0.1)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

cc2go-0.7.3.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+cc2go = router:main

cc2go-0.7.3.dist-info/licenses/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

cc2go-0.7.3.dist-info/top_level.txt

@@ -0,0 +1,5 @@
+error_handler
+mcp_bypass
+router
+streaming
+tray

error_handler.py

@@ -0,0 +1,150 @@
+"""
+错误自适应处理器
+提供错误分类、指数退避重试、模型切换 fallback
+"""
+
+import json
+import logging
+import random
+import threading
+import time
+from enum import Enum
+from typing import Tuple, Optional
+
+logger = logging.getLogger("llm_router")
+
+
+class ErrorType(Enum):
+    RATE_LIMIT = "rate_limit"       # 429
+    SERVER_ERROR = "server_error"  # 500/502/503
+    AUTH_ERROR = "auth_error"      # 401/403
+    CLIENT_ERROR = "client_error"  # 400
+    UNKNOWN = "unknown"
+
+
+class RetryStrategy(Enum):
+    RETRY_WITH_BACKOFF = "retry_with_backoff"  # 指数退避重试
+    SWITCH_MODEL = "switch_model"               # 切换模型
+    FAIL_FAST = "fail_fast"                     # 直接失败
+
+
+def classify_error(status_code: int) -> ErrorType:
+    """根据状态码分类错误类型"""
+    if status_code == 429:
+        return ErrorType.RATE_LIMIT
+    elif status_code in (500, 502, 503, 504):
+        return ErrorType.SERVER_ERROR
+    elif status_code in (401, 403):
+        return ErrorType.AUTH_ERROR
+    elif status_code == 400:
+        return ErrorType.CLIENT_ERROR
+    else:
+        return ErrorType.UNKNOWN
+
+
+def get_backoff_delay(attempt: int, base: float = 1.0, max_delay: float = 60.0, jitter: bool = True) -> float:
+    """计算指数退避延迟"""
+    delay = base * (2 ** attempt)
+    delay = min(delay, max_delay)
+    if jitter:
+        delay = delay * (0.5 + random.random() * 0.5)
+    return delay
+
+
+def parse_upstream_error(response_body) -> str:
+    """从上游响应中提取人类可读的错误信息"""
+    if isinstance(response_body, dict):
+        if "error" in response_body:
+            return response_body["error"].get("message", str(response_body["error"]))
+        if "message" in response_body:
+            return response_body["message"]
+        return str(response_body)
+    elif isinstance(response_body, str):
+        try:
+            return parse_upstream_error(json.loads(response_body))
+        except Exception:
+            return response_body[:500]
+    return "Unknown error"
+
+
+def classify_and_suggest_action(
+    status_code: int,
+    response_body,
+    attempt: int,
+    max_retry: int
+) -> Tuple[RetryStrategy, str, Optional[str]]:
+    """
+    分析错误并建议处理动作
+    Returns: (strategy, log_message, fallback_hint)
+    fallback_hint: None | "try_next_available"
+    """
+    error_type = classify_error(status_code)
+    error_msg = parse_upstream_error(response_body)
+
+    if error_type == ErrorType.RATE_LIMIT:
+        if attempt < max_retry - 1:
+            delay = get_backoff_delay(attempt)
+            return (
+                RetryStrategy.RETRY_WITH_BACKOFF,
+                f"[RateLimit] 429, 退避 {delay:.1f}s 后重试 (attempt {attempt + 1}/{max_retry})",
+                None
+            )
+        else:
+            return (
+                RetryStrategy.SWITCH_MODEL,
+                "[RateLimit] 多次 429，建议切换模型",
+                "try_next_available"
+            )
+    elif error_type == ErrorType.SERVER_ERROR:
+        if attempt < max_retry - 1:
+            delay = get_backoff_delay(attempt, base=2.0)
+            return (
+                RetryStrategy.RETRY_WITH_BACKOFF,
+                f"[ServerError] {status_code}, 退避 {delay:.1f}s 后重试 (attempt {attempt + 1}/{max_retry})",
+                None
+            )
+        else:
+            return (
+                RetryStrategy.SWITCH_MODEL,
+                f"[ServerError] 多次 {status_code}，建议切换模型",
+                "try_next_available"
+            )
+    elif error_type == ErrorType.AUTH_ERROR:
+        return (
+            RetryStrategy.FAIL_FAST,
+            f"[AuthError] {status_code} — 认证失败，请检查 API Key 配置",
+            None
+        )
+    else:
+        return (
+            RetryStrategy.FAIL_FAST,
+            f"[Error] {status_code}: {error_msg[:200]}",
+            None
+        )
+
+
+# ============ 归档限速 ============
+class ErrorArchiveRateLimiter:
+    """错误归档限速：window_seconds 内最多归档 1 次"""
+
+    def __init__(self, window_seconds: float = 30.0):
+        self.window = window_seconds
+        self._last_archive: float = 0.0
+        self._lock = threading.Lock()
+
+    def update(self, window_seconds: float):
+        with self._lock:
+            self.window = window_seconds
+
+    def archive(self) -> bool:
+        now = time.time()
+        with self._lock:
+            if now - self._last_archive >= self.window:
+                self._last_archive = now
+                return True
+            logger.debug("[ArchiveRateLimit] 限速跳过归档")
+            return False
+
+
+# 模块级限速器
+_archive_limiter = ErrorArchiveRateLimiter()

mcp_bypass.py

@@ -0,0 +1,176 @@
+"""
+MCP 工具短路模块
+检测特定工具调用并直接处理，避免绕道 LLM 后端
+"""
+
+import asyncio
+import json
+import logging
+import shutil
+import subprocess
+import sys
+import time
+from typing import Dict, List, Optional, Tuple
+
+logger = logging.getLogger("llm_router")
+
+if sys.platform == "win32":
+    MMX_PATH = shutil.which("mmx") or shutil.which("mmx.cmd") or "mmx"
+else:
+    MMX_PATH = "mmx"
+
+# 可短路的工具及其处理器
+BYPASS_TOOLS = {
+    "web_search": {"type": "mmx", "args": ["search", "query"]},
+    "mcp__MiniMax__web_search": {"type": "mmx", "args": ["search", "query"]},
+}
+
+
+def should_bypass(body: Dict) -> Tuple[bool, Optional[str]]:
+    """
+    判断请求是否应短路
+    仅当消息中模型已实际调用了 bypass 工具（tool_use）才触发，
+    不因 tools 参数中有该工具定义就短路。
+    Returns: (should_bypass, tool_name)
+    """
+    messages = body.get("messages", [])
+    if not messages:
+        logger.debug("[Bypass] no messages, skip")
+        return False, None
+
+    # 详细调试日志：打印每条消息的结构
+    for i, msg in enumerate(messages):
+        role = msg.get("role", "?")
+        content = msg.get("content", [])
+        content_types = []
+        if isinstance(content, list):
+            content_types = [b.get("type", "?") if isinstance(b, dict) else str(type(b)) for b in content]
+        elif isinstance(content, str):
+            content_types = ["str"]
+        logger.debug(f"[Bypass] msg[{i}] role={role}, content_types={content_types}")
+
+        if not isinstance(content, list):
+            continue
+        for j, block in enumerate(content):
+            if not isinstance(block, dict):
+                continue
+            block_type = block.get("type", "?")
+            block_name = block.get("name", "")
+            logger.debug(f"[Bypass]   msg[{i}] block[{j}] type={block_type}, name={repr(block_name)}")
+            if block_type in ("tool_use", "tool_use_block"):
+                if not block_name:
+                    logger.debug(f"[Bypass]   → tool_use block[{j}] has no name, skip")
+                    continue
+                # MCP 格式: mcp__Provider__tool_name → 归一化到 base
+                if block_name.startswith("mcp__"):
+                    base = block_name.split("__", 2)[-1]
+                    logger.debug(f"[Bypass]   → MCP format, base={base}, BYPASS_TOOLS={list(BYPASS_TOOLS.keys())}")
+                    if base in BYPASS_TOOLS:
+                        logger.info(f"[Bypass] HIT! name={block_name} → base={base}")
+                        return True, base
+                if block_name in BYPASS_TOOLS:
+                    logger.info(f"[Bypass] HIT! name={block_name}")
+                    return True, block_name
+                else:
+                    logger.debug(f"[Bypass]   → tool_use[{j}] name={block_name!r} NOT in BYPASS_TOOLS")
+    logger.debug(f"[Bypass] no tool_use found, BYPASS_TOOLS={list(BYPASS_TOOLS.keys())}")
+    return False, None
+
+
+def extract_query(messages: List[Dict]) -> str:
+    """从消息列表中提取用户查询文本"""
+    for msg in reversed(messages):
+        if msg.get("role") == "user":
+            content = msg.get("content", "")
+            if isinstance(content, str):
+                return content.strip()
+            elif isinstance(content, list):
+                for part in content:
+                    if isinstance(part, dict) and part.get("type") == "text":
+                        txt = part.get("text", "").strip()
+                        if txt:
+                            return txt
+    # fallback: 最后一条消息
+    if messages:
+        content = messages[-1].get("content", "")
+        if isinstance(content, str):
+            return content.strip()
+    return ""
+
+
+async def handle_bypass(tool_name: str, query: str) -> Dict:
+    """
+    执行短路处理，返回 Anthropic 格式的响应
+    """
+    logger.info(f"[Bypass] handle_bypass called: tool={tool_name}, query={query!r}")
+    handler = BYPASS_TOOLS.get(tool_name)
+    if not handler:
+        raise ValueError(f"Unknown bypass tool: {tool_name}")
+
+    if handler["type"] == "mmx":
+        result = await handle_mmx_search(tool_name, query)
+        logger.debug(f"[Bypass] mmx result: type={result['type']}, content_len={len(result.get('content',[]))}")
+        return result
+
+    raise ValueError(f"Unknown handler type: {handler['type']}")
+
+
+async def handle_mmx_search(tool_name: str, query: str) -> Dict:
+    """通过 mmx CLI 执行搜索并返回 Anthropic 格式"""
+    logger.info(f"[Bypass/mmx] starting search: query={query!r}, mmx_path={MMX_PATH}")
+    result = {
+        "type": "message",
+        "id": f"bypass-{int(time.time() * 1000)}",
+        "role": "assistant",
+        "content": [],
+        "model": "bypass",
+        "stop_reason": "end_turn",
+        "usage": {"input_tokens": 0, "output_tokens": 0}
+    }
+
+    if not query:
+        result["content"] = [{"type": "text", "text": "No query provided"}]
+        return result
+
+    try:
+        proc = await asyncio.create_subprocess_exec(
+            MMX_PATH, "search", "query", query,
+            "--output", "json",
+            stdout=subprocess.PIPE,
+            stderr=subprocess.PIPE
+        )
+        try:
+            stdout, stderr = await asyncio.wait_for(
+                proc.communicate(), timeout=30.0
+            )
+        except asyncio.TimeoutError:
+            proc.kill()
+            result["content"] = [{"type": "text", "text": "Search timed out after 30s"}]
+            return result
+
+        if proc.returncode != 0:
+            err_msg = stderr.decode("utf-8", errors="replace").strip()
+            result["content"] = [{"type": "text", "text": f"Search failed: {err_msg}"}]
+        else:
+            output = stdout.decode("utf-8", errors="replace")
+            try:
+                data = json.loads(output)
+                lines = []
+                for i, item in enumerate(data.get("organic", [])[:5]):
+                    title = item.get("title", "")
+                    link = item.get("link", "")
+                    snippet = item.get("snippet", "")
+                    snippet_short = snippet[:100] + "..." if len(snippet) > 100 else snippet
+                    lines.append(f"{i+1}. {title}\n   {link}")
+                    if snippet_short:
+                        lines.append(f"   {snippet_short}")
+                if not lines:
+                    result["content"] = [{"type": "text", "text": f"搜索 [{query}] 无结果"}]
+                else:
+                    result["content"] = [{"type": "text", "text": f"搜索 [{query}]:\n\n" + "\n\n".join(lines)}]
+            except json.JSONDecodeError:
+                result["content"] = [{"type": "text", "text": output[:2000]}]
+    except Exception as e:
+        result["content"] = [{"type": "text", "text": f"Search error: {str(e)}"}]
+
+    return result