scripts/mcp_server.py

"""MCP server (stdio, stdlib only) exposing the SCUT-SS RAG library to AI coding tools.



Lets an external/cloud model (CodeBuddy, Cursor, Claude Desktop, Cherry Studio, ...) call

local retrieval directly — no local model needed on the student's machine.



Register it in your client's MCP config, e.g.:



    {

      "mcpServers": {

        "scut-ss": {

          "command": "python",

          "args": ["D:/Projects/其他项目/创作空间/MySpace/SCUT-SS/scripts/mcp_server.py"]

        }

      }

    }



Manual smoke test:

    (echo {"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}; echo {...}) | python scripts/mcp_server.py

"""

import json
import sys
from pathlib import Path

sys.path.insert(0, str(Path(__file__).resolve().parent))
from vectorstore import Index  # noqa: E402

ROOT = Path(__file__).resolve().parent.parent
INDEX_DIR = ROOT / "vectorstore"
PAGES = ROOT / "data" / "dataset" / "pages.jsonl"

SERVER_INFO = {"name": "scut-ss-rag", "version": "0.2.0"}
PROTOCOL_VERSION = "2025-06-18"

TOOLS = [
    {
        "name": "search_materials",
        "description": (
            "检索华南理工软件学院课程复习资料（软件体系结构/软件测试/数据库系统/软件项目管理/C++/线性代数/历年卷等）。"
            "提问前先调用它拿到真实资料，再基于资料回答；不要凭记忆编造考点。"
        ),
        "inputSchema": {
            "type": "object",
            "properties": {
                "query": {"type": "string", "description": "检索问题或关键词"},
                "k": {"type": "integer", "description": "返回条数，默认 5", "default": 5},
                "course": {"type": "string", "description": "可选，限定课程关键词，如 软件测试 / 数据库系统", "default": ""},
            },
            "required": ["query"],
        },
    },
    {
        "name": "list_courses",
        "description": "列出资料库包含的课程/分类及各自的资料篇数。",
        "inputSchema": {"type": "object", "properties": {}},
    },
    {
        "name": "get_material",
        "description": "按标题关键词取回一篇完整资料（页面级全文，超出长度会截断）。",
        "inputSchema": {
            "type": "object",
            "properties": {
                "title": {"type": "string", "description": "资料标题关键词"},
                "max_chars": {"type": "integer", "description": "最大字符数，默认 8000", "default": 8000},
            },
            "required": ["title"],
        },
    },
]

_state = {}


def get_index():
    if "index" not in _state:
        _state["index"] = Index(INDEX_DIR)
    return _state["index"]


def load_pages():
    if "pages" not in _state:
        pages = []
        if PAGES.exists():
            for line in PAGES.open(encoding="utf-8"):
                if line.strip():
                    pages.append(json.loads(line))
        _state["pages"] = pages
    return _state["pages"]


def tool_search(args):
    query = args.get("query", "").strip()
    if not query:
        return "请提供检索关键词。"
    k = int(args.get("k") or 5)
    course = (args.get("course") or "").strip()
    index = get_index()
    hits = index.search(query, k=max(k * 6, 20))
    if course:
        filtered = [h for h in hits if course in (h[0].get("course") or "") or course in (h[0].get("path") or "")]
        hits = filtered or hits
    hits = hits[:k]
    if not hits:
        return f"资料库中没有检索到与「{query}」相关的内容。"
    out = [f"检索到 {len(hits)} 条资料（backend={index.meta['backend']}）：\n"]
    for i, (chunk, score) in enumerate(hits, 1):
        out.append(
            f"[{i}] {chunk['path']}"
            + (f" · {chunk['section']}" if chunk.get("section") else "")
            + f"\n来源URL: {chunk.get('url', '')}\n相关度: {score:.3f}\n{chunk['text']}\n"
        )
    return "\n".join(out)


def tool_courses(_args):
    pages = load_pages()
    counts = {}
    for p in pages:
        counts.setdefault(p["course"], 0)
        counts[p["course"]] += 1
    lines = [f"资料库共 {len(pages)} 篇资料，{sum(len(p['text']) for p in pages)} 字符。\n"]
    for course, n in sorted(counts.items(), key=lambda kv: -kv[1]):
        lines.append(f"- {course}：{n} 篇")
    return "\n".join(lines)


def tool_get(args):
    title = (args.get("title") or "").strip()
    max_chars = int(args.get("max_chars") or 8000)
    pages = load_pages()
    hits = [p for p in pages if title in p["title"] or title in (p.get("path") or "")]
    if not hits:
        return f"没有找到标题包含「{title}」的资料，可先用 search_materials 检索。"
    p = max(hits, key=lambda x: len(x["text"]))
    text = p["text"][:max_chars]
    tail = "\n\n...（已截断，完整内容见 data/dataset/pages.jsonl）" if len(p["text"]) > max_chars else ""
    return f"# {p['title']}\n路径: {p['path']}\nURL: {p.get('url','')}\n\n{text}{tail}"


HANDLERS = {
    "search_materials": tool_search,
    "list_courses": tool_courses,
    "get_material": tool_get,
}


def handle(req):
    method = req.get("method")
    params = req.get("params") or {}
    if method == "initialize":
        return {
            "protocolVersion": PROTOCOL_VERSION,
            "capabilities": {"tools": {"listChanged": False}},
            "serverInfo": SERVER_INFO,
        }
    if method == "tools/list":
        return {"tools": TOOLS}
    if method == "tools/call":
        name = params.get("name")
        handler = HANDLERS.get(name)
        if not handler:
            return {"content": [{"type": "text", "text": f"unknown tool: {name}"}], "isError": True}
        try:
            text = handler(params.get("arguments") or {})
        except Exception as exc:  # noqa: BLE001
            return {"content": [{"type": "text", "text": f"tool error: {exc}"}], "isError": True}
        return {"content": [{"type": "text", "text": text}]}
    if method in ("ping", "notifications/initialized"):
        return {}
    return {}


def main():
    for _s in (sys.stdout, sys.stderr):
        try:
            _s.reconfigure(encoding="utf-8", errors="replace")
        except Exception:  # noqa: BLE001
            pass
    for line in sys.stdin:
        line = line.strip()
        if not line:
            continue
        try:
            req = json.loads(line)
        except json.JSONDecodeError:
            continue
        if "id" not in req:  # notification
            continue
        resp = {"jsonrpc": "2.0", "id": req["id"], "result": handle(req)}
        sys.stdout.write(json.dumps(resp, ensure_ascii=False) + "\n")
        sys.stdout.flush()


if __name__ == "__main__":
    main()