File size: 15,660 Bytes
33cfdca
 
 
 
 
b1235ef
33cfdca
 
 
 
 
 
 
 
7b03579
b2bf24a
 
33cfdca
 
 
9b167e4
 
90302f8
9b167e4
33cfdca
 
 
b2bf24a
33cfdca
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7b03579
 
 
 
33cfdca
 
 
b1235ef
 
33cfdca
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7b03579
33cfdca
 
 
b2bf24a
33cfdca
 
 
 
 
 
 
 
 
b08e1ce
33cfdca
 
 
 
 
 
 
 
b1235ef
 
 
 
 
 
 
33cfdca
b2bf24a
33cfdca
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c078c66
7b03579
c078c66
 
 
 
 
8721a61
b2bf24a
57c78ad
33cfdca
 
 
 
 
 
b2bf24a
33cfdca
 
 
c078c66
57c78ad
c078c66
b2bf24a
c078c66
 
 
 
 
 
 
b2bf24a
95a8be8
 
b1235ef
33cfdca
c078c66
33cfdca
7b03579
33cfdca
 
 
 
 
 
 
 
b2bf24a
8721a61
 
 
 
 
 
 
 
 
33cfdca
 
 
 
8721a61
33cfdca
7b03579
33cfdca
6e5a889
33cfdca
 
b1235ef
33cfdca
 
 
 
 
 
 
 
b2bf24a
33cfdca
 
 
 
7b03579
 
 
 
 
33cfdca
 
 
 
 
 
b1235ef
33cfdca
 
 
 
 
7b03579
33cfdca
 
 
 
 
 
 
 
b2bf24a
8721a61
 
 
 
 
 
 
 
 
33cfdca
 
 
 
8721a61
33cfdca
7b03579
33cfdca
 
 
 
 
b2bf24a
33cfdca
 
 
 
 
 
8721a61
33cfdca
 
b1235ef
7b03579
 
 
 
 
 
 
 
 
 
 
 
 
 
b2bf24a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7b03579
 
e7e9bc6
b2bf24a
fb9cbfc
b2bf24a
 
 
7b03579
 
 
 
b2bf24a
7b03579
b2bf24a
 
 
 
 
7b03579
 
 
 
 
 
 
 
 
b2bf24a
 
7b03579
 
 
 
b2bf24a
7b03579
 
b2bf24a
 
 
 
 
 
 
 
 
 
 
7b03579
b2bf24a
 
 
 
 
 
 
7b03579
bc117da
b2bf24a
 
 
 
 
 
 
 
 
 
 
 
 
bc117da
b2bf24a
bc117da
b2bf24a
bc117da
b2bf24a
 
bc117da
90302f8
bc117da
90302f8
 
bc117da
 
 
b1235ef
bc117da
 
 
 
 
b1235ef
bc117da
 
 
 
 
b1235ef
90302f8
bc117da
 
 
7b03579
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
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
import os
import sqlite3
from datetime import datetime
import gradio as gr
from huggingface_hub import InferenceClient
from datasets import load_dataset

# ---------------------------
# Config
# ---------------------------
MODELS = {
    "Meta LLaMA 3.1 (8B Instruct)": "meta-llama/Llama-3.1-8B-Instruct",
    "Mistral 7B Instruct": "mistralai/Mistral-7B-Instruct-v0.3",
}

DATASETS = ["The Stack", "CodeXGLUE"]  # Dropdown for dataset selection
HF_TOKEN = os.getenv("HF_TOKEN")  # Set in your Space's Secrets
DB_PATH = "history.db"

SYSTEM_DEFAULT = (
    "Specializes in databases, APIs, auth, CRUD."
    "Provides complete backend code scaffolds."
    "Provide full backend code scaffolds with files, paths, and commands. "
    "Declines frontend-heavy requests."
)

# ---------------------------
# DB Setup
# ---------------------------
def db():
    conn = sqlite3.connect(DB_PATH)
    conn.execute("PRAGMA journal_mode=WAL;")
    return conn

def init_db():
    conn = db()
    cur = conn.cursor()
    cur.execute("""
        CREATE TABLE IF NOT EXISTS sessions (
            id INTEGER PRIMARY KEY AUTOINCREMENT,
            title TEXT NOT NULL,
            created_at TEXT NOT NULL
        )
    """)
    cur.execute("""
        CREATE TABLE IF NOT EXISTS messages (
            id INTEGER PRIMARY KEY AUTOINCREMENT,
            session_id INTEGER NOT NULL,
            role TEXT NOT NULL,
            content TEXT NOT NULL,
            created_at TEXT NOT NULL,
            FOREIGN KEY(session_id) REFERENCES sessions(id) ON DELETE CASCADE
        )
    """)
    conn.commit()
    conn.close()

def create_session(title: str = "New chat") -> int:
    conn = db()
    cur = conn.cursor()
    cur.execute(
        "INSERT INTO sessions (title, created_at) VALUES (?, ?)",
        (title, datetime.utcnow().isoformat())
    )
    session_id = cur.lastrowid
    conn.commit()
    conn.close()
    return session_id

def delete_session(session_id: int):
    conn = db()
    cur = conn.cursor()
    cur.execute("DELETE FROM messages WHERE session_id = ?", (session_id,))
    cur.execute("DELETE FROM sessions WHERE id = ?", (session_id,))
    conn.commit()
    conn.close()

def list_sessions():
    conn = db()
    cur = conn.cursor()
    cur.execute("SELECT id, title FROM sessions ORDER BY id DESC")
    rows = cur.fetchall()
    conn.close()
    labels = [f"{r[0]}{r[1]}" for r in rows]
    return labels, rows

def get_messages(session_id: int):
    conn = db()
    cur = conn.cursor()
    cur.execute("""
        SELECT role, content
        FROM messages
        WHERE session_id = ?
        ORDER BY id ASC
    """, (session_id,))
    rows = cur.fetchall()
    conn.close()
    msgs = [{"role": role, "content": content} for (role, content) in rows]
    return msgs

def add_message(session_id: int, role: str, content: str):
    conn = db()
    cur = conn.cursor()
    cur.execute(
        "INSERT INTO messages (session_id, role, content, created_at) VALUES (?, ?, ?, ?)",
        (session_id, role, content, datetime.utcnow().isoformat())
    )
    conn.commit()
    conn.close()

def update_session_title_if_needed(session_id: int, first_user_text: str):
    conn = db()
    cur = conn.cursor()
    cur.execute("SELECT COUNT(*) FROM messages WHERE session_id=? AND role='user'", (session_id,))
    count_users = cur.fetchone()[0]
    if count_users == 1:
        title = first_user_text.strip().split("\n")[0]
        title = (title[:50] + "…") if len(title) > 50 else title
        cur.execute("UPDATE sessions SET title=? WHERE id=?", (title or "New chat", session_id))
        conn.commit()
    conn.close()

# ---------------------------
# Helpers
# ---------------------------
def label_to_id(label: str | None) -> int | None:
    if not label:
        return None
    try:
        return int(label.split("•", 1)[0].strip())
    except Exception:
        return None

def build_api_messages(session_id: int, system_message: str):
    msgs = [{"role": "system", "content": system_message.strip()}]
    msgs.extend(get_messages(session_id))
    return msgs

def get_client(model_choice: str):
    model_id = MODELS.get(model_choice, list(MODELS.values())[0])
    return InferenceClient(model_id, token=HF_TOKEN)

def load_dataset_by_name(name: str):
    if name == "The Stack":
        return load_dataset("bigcode/the-stack", split="train")
    elif name == "CodeXGLUE":
        return load_dataset("google/code_x_glue_cc_code_to_code_trans", split="train")
    return None

# ---------------------------
# Gradio Callbacks
# ---------------------------
def refresh_sessions_cb():
    labels, _ = list_sessions()
    selected = labels[0] if labels else None
    return gr.update(choices=labels, value=selected)

def new_chat_cb():
    sid = create_session("New chat")
    labels, _ = list_sessions()
    selected = next((lbl for lbl in labels if lbl.startswith(f"{sid} ")), None)
    return (gr.update(choices=labels, value=selected), [], "")

def load_session_cb(selected_label):
    sid = label_to_id(selected_label)
    if not sid:
        return []
    return get_messages(sid)

def delete_chat_cb(selected_label):
    sid = label_to_id(selected_label)
    if sid:
        delete_session(sid)
    labels, _ = list_sessions()
    selected = labels[0] if labels else None
    return gr.update(choices=labels, value=selected), []

FRONTEND_KEYWORDS = [
    "react", "vue", "angular", "html", "css", "javascript", "tailwind", "recharts", "typescript"
]

def is_frontend_request(user_text: str) -> bool:
    text_lower = user_text.lower()
    return any(kw in text_lower for kw in FRONTEND_KEYWORDS)

# --- Fixed send_cb to show user message ---
def send_cb(user_text, selected_label, chatbot_msgs, system_message, max_tokens, temperature, top_p, model_choice, dataset_choice, *args):
    sid = label_to_id(selected_label)
    if sid is None:
        sid = create_session("New chat")
        labels, _ = list_sessions()
        selected_label = next((lbl for lbl in labels if lbl.startswith(f"{sid} ")), None)

    # Save user message
    add_message(sid, "user", user_text)
    update_session_title_if_needed(sid, user_text)

    display_msgs = chatbot_msgs[:]
    display_msgs.append({"role": "user", "content": user_text})

    # Check for frontend-heavy request
    if is_frontend_request(user_text):
        apology = "⚠️ I'm a backend-focused assistant and cannot provide frontend code."
        display_msgs.append({"role": "assistant", "content": apology})
        add_message(sid, "assistant", apology)
        yield (display_msgs, "", selected_label)
        return

    # Normal backend response
    display_msgs.append({"role": "assistant", "content": "…"})
    yield (display_msgs, "", selected_label)

    client = get_client(model_choice)
    api_messages = build_api_messages(sid, system_message)
    partial = ""

    try:
        for chunk in client.chat_completion(
            messages=api_messages,
            max_tokens=int(max_tokens),
            temperature=float(temperature),
            top_p=float(top_p),
            stream=True,
        ):
            # --- FIX: handle models that send empty chunks or use message instead of delta ---
            if not hasattr(chunk, "choices") or not chunk.choices:
                continue
            choice = chunk.choices[0]
            delta = ""
            if hasattr(choice, "delta") and choice.delta and getattr(choice.delta, "content", None) is not None:
                delta = choice.delta.content
            elif hasattr(choice, "message") and getattr(choice.message, "content", None) is not None:
                delta = choice.message.content

            if delta:
                partial += delta
                display_msgs[-1]["content"] = partial
                yield (display_msgs, "", selected_label)

        add_message(sid, "assistant", partial)

    except Exception as e:
        display_msgs[-1]["content"] = f"⚠️ Error: {str(e)}"
        yield (display_msgs, "", selected_label)

def regenerate_cb(selected_label, system_message, max_tokens, temperature, top_p, model_choice, dataset_choice):
    sid = label_to_id(selected_label)
    if sid is None:
        return [], ""

    msgs = get_messages(sid)
    if not msgs:
        return [], ""

    # Remove the last assistant message if it exists (to regenerate it)
    if msgs and msgs[-1]["role"] == "assistant":
        conn = db()
        cur = conn.cursor()
        cur.execute("""
            DELETE FROM messages
            WHERE id = (
                SELECT id FROM messages
                WHERE session_id=?
                ORDER BY id DESC LIMIT 1
            )
        """, (sid,))
        conn.commit()
        conn.close()
        msgs = get_messages(sid)

    dataset = load_dataset_by_name(dataset_choice)
    api_messages = [{"role": "system", "content": system_message.strip()}] + msgs
    display_msgs = msgs + [{"role": "assistant", "content": ""}]

    client = get_client(model_choice)
    partial = ""

    try:
        for chunk in client.chat_completion(
            messages=api_messages,
            max_tokens=int(max_tokens),
            temperature=float(temperature),
            top_p=float(top_p),
            stream=True,
        ):
            # --- FIX: handle models that send empty chunks or use message instead of delta ---
            if not hasattr(chunk, "choices") or not chunk.choices:
                continue
            choice = chunk.choices[0]
            delta = ""
            if hasattr(choice, "delta") and choice.delta and getattr(choice.delta, "content", None) is not None:
                delta = choice.delta.content
            elif hasattr(choice, "message") and getattr(choice.message, "content", None) is not None:
                delta = choice.message.content

            if delta:
                partial += delta
                display_msgs[-1]["content"] = partial
                yield display_msgs

        add_message(sid, "assistant", partial)

    except Exception as e:
        display_msgs[-1]["content"] = f"⚠️ Error: {str(e)}"
        yield display_msgs

# ---------------------------
# App UI
# ---------------------------
init_db()
labels, _ = list_sessions()
if not labels:
    first_sid = create_session("New chat")
    labels, _ = list_sessions()

default_selected = labels[0] if labels else None

with gr.Blocks(title="Backend-Focused LLaMA/Mistral CRUD Assistant", theme=gr.themes.Soft()) as demo:
    gr.HTML("""
    <style>
    button {
        background-color: #22c55e !important;
        color: #ffffff !important;
        border: none !important;
    }
    button:hover {
        background-color: #16a34a !important;
    }
    button:focus {
        outline: 2px solid #166534 !important;
        outline-offset: 2px;
    }
    .compact-sliders .gr-form {
        margin-bottom: 0.5rem !important;
    }
    .compact-sliders .gr-slider {
        margin-bottom: 0.5rem !important;
    }
    .compact-sliders .gr-number {
        margin-bottom: 0.5rem !important;
    }
    .system-message-compact .gr-form {
        margin-bottom: 0.5rem !important;
    }
    .system-message-compact .gr-textarea {
        min-height: 80px !important;
        max-height: 120px !important;
    }
    </style>
    """)

    gr.Markdown("## 🗄️ LLaMA & Mistral Backend-Focused CRUD Automation — with Persistent History")

    with gr.Row(equal_height=True):
        with gr.Column(scale=1, min_width=260):
            gr.Markdown("### 📁 Sessions")
            session_list = gr.Radio(
                choices=labels,
                value=default_selected,
                label="Your chats",
                interactive=True
            )
            # -----------------------
            # Editable title input
            # -----------------------
            edit_title_box = gr.Textbox(label="✏️ Rename Chat", placeholder="Edit selected chat title...")
            rename_btn = gr.Button("💾 Save Title")
            
            def rename_session_cb(new_title, selected_label):
                sid = label_to_id(selected_label)
                if sid and new_title.strip():
                    conn = db()
                    cur = conn.cursor()
                    cur.execute("UPDATE sessions SET title=? WHERE id=?", (new_title.strip(), sid))
                    conn.commit()
                    conn.close()
            
                # Refresh the session list and keep the same one selected
                labels, _ = list_sessions()
                new_selected = next((lbl for lbl in labels if lbl.startswith(f"{sid} ")), None)
                return gr.update(choices=labels, value=new_selected)
            
            # Connect button to callback
            rename_btn.click(rename_session_cb, inputs=[edit_title_box, session_list], outputs=session_list)

            with gr.Row():
                new_btn = gr.Button("➕ New Chat", variant="primary")
                del_btn = gr.Button("🗑️ Delete", variant="stop")
                refresh_btn = gr.Button("🔄 Refresh", variant="secondary")

            gr.Markdown("### 🤖 Model Selection")
            model_choice = gr.Dropdown(
                choices=list(MODELS.keys()),
                value=list(MODELS.keys())[0],
                label="Choose a model",
                interactive=True
            )

            gr.Markdown("### 📚 Dataset Selection")
            dataset_choice = gr.Dropdown(
                choices=DATASETS,
                value=DATASETS[0],
                label="Select a dataset",
                interactive=True
            )

            gr.Markdown("### ⚙️ Generation Settings")
            with gr.Group(elem_classes="system-message-compact"):
                system_box = gr.Textbox(
                    value=SYSTEM_DEFAULT,
                    label="System message",
                    lines=3
                )
            
            with gr.Group(elem_classes="compact-sliders"):
                max_tokens = gr.Slider(256, 4096, value=1200, step=16, label="Max tokens")
                temperature = gr.Slider(0.0, 2.0, value=0.25, step=0.05, label="Temperature")
                top_p = gr.Slider(0.1, 1.0, value=0.9, step=0.05, label="Top-p")

        with gr.Column(scale=3):
            chatbot = gr.Chatbot(label="Assistant", height=600, type="messages")
            with gr.Row():
                user_box = gr.Textbox(placeholder="Describe your CRUD/backend task…", lines=3, scale=5)
            with gr.Row():
                send_btn = gr.Button("Send ▶️", variant="primary")
                regen_btn = gr.Button("Regenerate 🔁", variant="secondary")

    refresh_btn.click(refresh_sessions_cb, outputs=session_list)
    new_btn.click(new_chat_cb, outputs=[session_list, chatbot, user_box])
    del_btn.click(delete_chat_cb, inputs=session_list, outputs=[session_list, chatbot])
    session_list.change(load_session_cb, inputs=session_list, outputs=chatbot)

    send_btn.click(
        send_cb,
        inputs=[user_box, session_list, chatbot, system_box, max_tokens, temperature, top_p, model_choice, dataset_choice],
        outputs=[chatbot, user_box, session_list]
    )

    user_box.submit(
        send_cb,
        inputs=[user_box, session_list, chatbot, system_box, max_tokens, temperature, top_p, model_choice, dataset_choice],
        outputs=[chatbot, user_box, session_list]
    )

    regen_btn.click(
        regenerate_cb,
        inputs=[session_list, system_box, max_tokens, temperature, top_p, model_choice, dataset_choice],
        outputs=chatbot
    )

if __name__ == "__main__":
    demo.launch()