File size: 4,613 Bytes
4af70f5
 
2eaccce
bedf4dd
4af70f5
 
8d0f69f
ecbc1ec
bedf4dd
4af70f5
 
 
 
 
 
 
 
 
 
 
 
bedf4dd
 
 
ceb6d8a
 
bedf4dd
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ecbc1ec
4af70f5
2eaccce
 
 
ecbc1ec
2eaccce
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4af70f5
ecbc1ec
 
 
 
 
 
 
 
 
186a0ec
 
 
ecbc1ec
 
186a0ec
 
ecbc1ec
 
 
2eaccce
4af70f5
ecbc1ec
 
 
4af70f5
 
2eaccce
4af70f5
ecbc1ec
 
 
 
4af70f5
 
2eaccce
4af70f5
ecbc1ec
 
 
 
 
 
 
4af70f5
 
2eaccce
4af70f5
ecbc1ec
 
 
 
4af70f5
 
ecbc1ec
 
 
2eaccce
4af70f5
f87d282
7737dc2
 
bedf4dd
4af70f5
7737dc2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
import os
import uvicorn
import inspect
import httpx

from mcp.server.fastmcp import FastMCP
from starlette.requests import Request
from starlette.responses import PlainTextResponse, JSONResponse
from starlette.middleware.base import BaseHTTPMiddleware

from langchain_community.utilities import SQLDatabase
from langchain_community.tools.sql_database.tool import QuerySQLCheckerTool
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    api_key=os.environ.get('OPENAI_API_KEY', None),
    base_url=os.environ['OPENAI_BASE_URL'],
    model='gpt-4o-mini',
    temperature=0
)

# Hugging Face Token Auth Middleware
class HuggingFaceTokenAuthMiddleware(BaseHTTPMiddleware):
    async def dispatch(self, request: Request, call_next):
        # Allow "/" and "/tools" to be public, protect everything else
        if request.url.path in ["/", "/tools"]:
            return await call_next(request)
        # Check Authorization header
        auth = request.headers.get("authorization")
        if not auth or not auth.lower().startswith("bearer "):
            return PlainTextResponse("Missing or invalid Authorization header (expected Bearer token)", status_code=401)
        token = auth.split(" ", 1)[1].strip()
        # Validate token with Hugging Face API
        async with httpx.AsyncClient() as client:
            resp = await client.get(
                "https://huggingface.co/api/whoami-v2",
                headers={"Authorization": f"Bearer {token}"}
            )
            if resp.status_code != 200:
                return PlainTextResponse("Invalid or expired Hugging Face token", status_code=401)
            hf_user_info = resp.json()
        request.state.hf_user = hf_user_info
        return await call_next(request)

# Create an MCP server and the tool registry
mcp = FastMCP("Credit Card Database Server")
tool_registry = []

def register_tool(fn):
    """Decorator to register tool metadata and with MCP."""
    mcp.tool()(fn)
    sig = inspect.signature(fn)
    params = [
        {
            "name": param.name,
            "type": str(param.annotation) if param.annotation is not inspect._empty else "Any",
            "default": param.default if param.default is not inspect._empty else None,
        }
        for param in sig.parameters.values()
    ]
    tool_registry.append({
        "name": fn.__name__,
        "description": fn.__doc__.strip() if fn.__doc__ else "",
        "parameters": params,
    })
    return fn

credit_card_db = SQLDatabase.from_uri(r"sqlite:///data/ccms.db")
query_checker_tool = QuerySQLCheckerTool(db=credit_card_db, llm=llm)

@mcp.custom_route("/", methods=["GET"])
async def home(request: Request) -> PlainTextResponse:
    return PlainTextResponse(
        """
        Credit Card Database MCP Server
        ----
        This server needs a HuggingFace token for access
        
        Use the following URL to connect your client with this server
        https://pgurazada1-credit-card-database-mcp-server.hf.space/mcp/
        
        Access the following public URL for a list of tools and their documentation.
        https://pgurazada1-credit-card-database-mcp-server.hf.space/tools
        """
    )

@register_tool
def sql_db_list_tables():
    """
    Returns a comma-separated list of table names in the database.
    """
    return credit_card_db.get_usable_table_names()

@register_tool
def sql_db_schema(table_names: list[str]) -> str:
    """
    Input 'table_names_str' is a comma-separated string of table names.
    Returns the DDL SQL schema for these tables.
    """
    return credit_card_db.get_table_info(table_names)

@register_tool
def sql_db_query_checker(query: str) -> str:
    """
    Input 'query' is a SQL query string.
    Checks if the query is valid.
    If the query is valid, it returns the original query.
    If the query is not valid, it returns the corrected query.
    This tool is used to ensure the query is valid before executing it.
    """
    return query_checker_tool.run(query)

@register_tool
def sql_db_query(query: str) -> str:
    """
    Input 'query' is a SQL query string.
    Executes the query (SELECT only) and returns the result.
    """
    return credit_card_db.run(query)

@mcp.custom_route("/tools", methods=["GET"])
async def list_tools(request: Request) -> JSONResponse:
    """Return all registered tool metadata as JSON."""
    return JSONResponse(tool_registry)

# --- Build the app and add middleware ---
app = mcp.streamable_http_app()
app.add_middleware(HuggingFaceTokenAuthMiddleware)

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=8000)