Hugging Face's logo

Spaces:
NujacsaintS
/
CorporateChatAPI
Runtime error

App Files Community
NujacsaintS commited on
Commit
31decff
·
verified ·
1 Parent(s): d522740

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +14 -141
app.py CHANGED
@@ -3,6 +3,8 @@ from fastapi.responses import HTMLResponse, JSONResponse
3
  from pydantic import BaseModel
4
  from typing import Dict
5
  import time
 
 
6
 
7
  app = FastAPI(title="Tessai LLM Bridge", version="0.1.0")
8
 
@@ -62,152 +64,23 @@ def count_active_sessions(window_seconds: int = 300) -> int:
62
 
63
 
64
  # -----------------------------
65
- # LLM integration stub
66
- # -----------------------------
67
-
68
- def call_llm(message: str, context: Dict[str, str] | None = None) -> tuple[str, int]:
69
- """
70
- Replace this with your real LLM call.
71
-
72
- For now, it just echoes the message and pretends each character is a token.
73
- This keeps the server functional while you wire in the real backend.
74
- """
75
- reply = f"Echo: {message}"
76
- tokens_used = len(message)
77
- return reply, tokens_used
78
-
79
-
80
- # -----------------------------
81
- # API endpoints
82
  # -----------------------------
 
 
 
83
 
84
 
85
- @app.post("/v1/chat", response_model=ChatResponse)
86
- async def chat_endpoint(payload: ChatRequest):
87
- reply, tokens_used = call_llm(payload.message, payload.context)
88
- record_request(payload.session_id, tokens_used)
89
-
90
- return ChatResponse(
91
- session_id=payload.session_id,
92
- reply=reply,
93
- tokens_used=tokens_used,
94
- )
95
 
96
 
97
- @app.get("/health")
98
- async def health():
99
- return JSONResponse(
100
- {
101
- "status": "ok",
102
- "total_requests": metrics["total_requests"],
103
- "total_tokens": metrics["total_tokens"],
104
- "active_sessions_5m": count_active_sessions(300),
105
- }
106
- )
107
-
108
-
109
- @app.get("/admin", response_class=HTMLResponse)
110
- async def admin_dashboard():
111
- active_5m = count_active_sessions(300)
112
- rows = []
113
-
114
- for sid, s in metrics["sessions"].items():
115
- last_seen = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(s["last_seen"]))
116
- rows.append(
117
- f"<tr>"
118
- f"<td>{sid}</td>"
119
- f"<td>{s['requests']}</td>"
120
- f"<td>{s['tokens']}</td>"
121
- f"<td>{last_seen}</td>"
122
- f"</tr>"
123
- )
124
-
125
- rows_html = "\n".join(rows) if rows else "<tr><td colspan='4'>No sessions yet</td></tr>"
126
-
127
- html = f"""
128
- <!doctype html>
129
- <html>
130
- <head>
131
- <title>Tessai LLM Admin</title>
132
- <meta charset="utf-8" />
133
- <style>
134
- body {{
135
- font-family: system-ui, sans-serif;
136
- margin: 20px;
137
- }}
138
- h1, h2 {{
139
- margin-bottom: 0.2rem;
140
- }}
141
- .metrics {{
142
- display: flex;
143
- gap: 1.5rem;
144
- margin-bottom: 1.5rem;
145
- }}
146
- .metric-card {{
147
- padding: 1rem 1.5rem;
148
- border-radius: 8px;
149
- border: 1px solid #ddd;
150
- box-shadow: 0 1px 3px rgba(0,0,0,0.05);
151
- }}
152
- table {{
153
- border-collapse: collapse;
154
- width: 100%;
155
- }}
156
- th, td {{
157
- border: 1px solid #ddd;
158
- padding: 8px;
159
- font-size: 0.9rem;
160
- }}
161
- th {{
162
- background: #f4f4f4;
163
- text-align: left;
164
- }}
165
- </style>
166
- </head>
167
- <body>
168
- <h1>Tessai LLM Bridge</h1>
169
- <p>Simple meter board for current traffic.</p>
170
-
171
- <div class="metrics">
172
- <div class="metric-card">
173
- <h2>Total requests</h2>
174
- <p>{metrics["total_requests"]}</p>
175
- </div>
176
- <div class="metric-card">
177
- <h2>Total tokens (approx)</h2>
178
- <p>{metrics["total_tokens"]}</p>
179
- </div>
180
- <div class="metric-card">
181
- <h2>Active sessions (last 5 min)</h2>
182
- <p>{active_5m}</p>
183
- </div>
184
- </div>
185
-
186
- <h2>Sessions</h2>
187
- <table>
188
- <thead>
189
- <tr>
190
- <th>Session ID</th>
191
- <th>Requests</th>
192
- <th>Tokens</th>
193
- <th>Last seen</th>
194
- </tr>
195
- </thead>
196
- <tbody>
197
- {rows_html}
198
- </tbody>
199
- </table>
200
- </body>
201
- </html>
202
  """
203
- return HTMLResponse(content=html)
204
-
205
 
206
- # -----------------------------
207
- # Local dev entrypoint
208
- # -----------------------------
209
-
210
- if __name__ == "__main__":
211
- import uvicorn
212
 
213
- uvicorn.run("app:app", host="0.0.0.0", port=8000, reload=True)
 
3
  from pydantic import BaseModel
4
  from typing import Dict
5
  import time
6
+ import os
7
+ import httpx
8
 
9
  app = FastAPI(title="Tessai LLM Bridge", version="0.1.0")
10
 
 
64
 
65
 
66
  # -----------------------------
67
+ # LLM integration
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
68
  # -----------------------------
69
+ # Configure via environment variables in Hugging Face:
70
+ # TESSAI_LLM_URL = full URL of the model endpoint
71
+ # TESSAI_LLM_KEY = auth token (if needed)
72
 
73
 
74
+ LLM_URL = os.getenv("TESSAI_LLM_URL", "").strip()
75
+ LLM_KEY = os.getenv("TESSAI_LLM_KEY", "").strip()
 
 
 
 
 
 
 
 
76
 
77
 
78
+ async def call_llm(message: str, context: Dict[str, str] | None = None) -> tuple[str, int]:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
79
  """
80
+ Replace this with whatever target you want.
 
81
 
82
+ This implementation is ready for a typical Hugging Face Inference API
83
+ text-generation endpoint. If LLM_URL is not set, it falls back to a local echo.
84
+ """
 
 
 
85
 
86
+ # Fallback so the b