File size: 14,521 Bytes
8fccc59
6b26bdf
 
 
29345ee
6b26bdf
 
 
 
29345ee
6b26bdf
 
 
 
 
 
29345ee
6b26bdf
 
29345ee
6b26bdf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ac29b87
6b26bdf
 
 
 
 
 
 
 
 
 
 
 
 
ab0d0f7
6b26bdf
 
 
 
 
 
 
 
 
 
 
 
1d4a770
6b26bdf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ab0d0f7
6b26bdf
 
 
 
1d4a770
6b26bdf
 
 
 
 
 
1d4a770
6b26bdf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
29345ee
6b26bdf
 
 
 
29345ee
6b26bdf
 
 
 
 
 
 
9f59c2f
 
6b26bdf
 
 
 
 
 
9f59c2f
 
6b26bdf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9f59c2f
6b26bdf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8fccc59
6b26bdf
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
import gradio as gr
from huggingface_hub import InferenceClient
from datetime import datetime
import re

# Import NLP and Math Matching frameworks
import nltk
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity

# Initialize NLTK tokenization assets safely on basic CPU bootup
try:
    nltk.download('punkt', quiet=True)
    nltk.download('punkt_tab', quiet=True)
except Exception:
    pass  # Fallback to standard regex splitting if environment is strictly offline

# Initialize the Hugging Face Serverless client
client = InferenceClient("Qwen/Qwen2.5-7B-Instruct")

# 1. Collect FAQs related to topics/products into a structured corpus
FAQ_DATA = [
    {
        "q": "What are your shipping times and policies?",
        "a": "Shipping times vary by location. Typically, local retail deliveries take 2-3 business days, while international dropshipping options take 7-14 business days. All tracking codes are sent via email."
    },
    {
        "q": "What is the return policy for products?",
        "a": "We offer a flexible 30-day return policy. Items must be unworn, in their original packaging layout, and accompanied by the checkout receipt."
    },
    {
        "q": "What are your service project timelines and quotes?",
        "a": "Standard design and consulting project timelines range between 2 to 6 weeks. You can request a personalized pricing quote directly via our booking panel."
    },
    {
        "q": "What are the differences between your software subscription tiers?",
        "a": "We offer Free, Pro, and Enterprise tiers. Pro unlocks comprehensive API documentation access, and Enterprise unlocks custom third-party platform integrations."
    },
    {
        "q": "How do I unlock my course completion certificate?",
        "a": "Completion certificates unlock automatically on your student profile dashboard as soon as all video modules and the final evaluation quiz are marked 100% complete."
    },
    {
        "q": "What insurance providers do you accept at the clinic?",
        "a": "We accept major national medical insurance providers. Please contact our front office coordinator to verify your specific policy layout before scheduling appointments."
    },
    {
        "q": "What are your standard check-in and check-out times?",
        "a": "Standard hospitality check-in opens at 3:00 PM, and checkout must be finalized by 11:00 AM. Early check-in requests can be accommodated based on live room availability."
    }
]

def perform_nlp_cosine_matching(user_query):
    """
    Preprocesses text using tokenization and applies Cosine Similarity 
    matching over TF-IDF vectors to find the best matching FAQ.
    """
    try:
        tokens = nltk.word_tokenize(user_query.lower())
        cleaned_query = " ".join([w for w in tokens if re.match(r'^\w+$', w)])
    except Exception:
        cleaned_query = " ".join(re.findall(r'\w+', user_query.lower()))
        
    if not cleaned_query.strip():
        return None, 0.0

    questions_corpus = [faq["q"] for faq in FAQ_DATA]
    
    vectorizer = TfidfVectorizer(stop_words='english')
    tfidf_matrix = vectorizer.fit_transform(questions_corpus)
    
    query_vector = vectorizer.transform([cleaned_query])
    similarity_scores = cosine_similarity(query_vector, tfidf_matrix)[0]
    
    best_match_idx = similarity_scores.argmax()
    highest_score = similarity_scores[best_match_idx]
    
    return FAQ_DATA[best_match_idx], float(highest_score)

def get_thematic_datetime():
    """Generates precise, real-time updated date and running seconds strings."""
    now = datetime.now()
    day_int = now.day
    if 11 <= day_int <= 13:
        suffix = "th"
    else:
        suffix = {1: "st", 2: "nd", 3: "rd"}.get(day_int % 10, "th")
        
    formatted_date = f"{now.strftime('%d')}{suffix} {now.strftime('%B %Y')} | Time: {now.strftime('%I:%M:%S %p').lower()}"
    return f"<div style='text-align: right; font-weight: 600; font-size: 14px; font-family: monospace; letter-spacing: 0.5px;'>{formatted_date}</div>"

def extract_text(content):
    """Safely extracts plain string text from any Gradio 6 message structure."""
    if isinstance(content, str):
        return content
    if isinstance(content, list):
        extracted_parts = []
        for block in content:
            if isinstance(block, dict) and "text" in block:
                extracted_parts.append(block["text"])
            elif isinstance(block, str):
                extracted_parts.append(block)
        return "".join(extracted_parts)
    if isinstance(content, dict):
        if "text" in content: return content["text"]
        if "content" in content: return content["content"]
    return str(content)

def user_side(user_message, history):
    """Handles incoming user messages and sanitizes state history format."""
    if not user_message.strip():
        return "", history
    
    clean_history = []
    for msg in history:
        if isinstance(msg, dict):
            clean_history.append({"role": msg["role"], "content": extract_text(msg["content"])})
        elif hasattr(msg, "role") and hasattr(msg, "content"):
            clean_history.append({"role": msg.role, "content": extract_text(msg.content)})
            
    clean_history.append({"role": "user", "content": user_message})
    return "", clean_history

def bot_side(history, system_message, max_tokens, temperature, top_p):
    """Streams responses out, embedding the NLP similarity match directly into context."""
    if not history:
        yield history
        return

    user_query = extract_text(history[-1]["content"])
    matched_faq, match_score = perform_nlp_cosine_matching(user_query)
    
    faq_context_injection = ""
    
    # RECTIFIED LOGIC: The metadata line only outlines telemetry metrics. It doesn't dump the duplicate answer text block.
    if matched_faq and match_score > 0.20:
        nlp_metadata_prefix = f"πŸ” *NLP Intent Match: \"{matched_faq['q']}\" (Confidence: {match_score*100:.1f}%)*\n\n"
        faq_context_injection = (
            f"The user's query closely matches this curated FAQ entry:\n"
            f"Question: {matched_faq['q']}\n"
            f"Answer: {matched_faq['a']}\n"
            f"Please output this answer clearly and conversationally. Do not wrap it in quotes."
        )
    else:
        nlp_metadata_prefix = "πŸ” *NLP Search Notice: No close database FAQ match found. Processing via general knowledge base...*\n\n"
        faq_context_injection = "No matching database entry found. Answer the question directly using standard general system knowledge."

    full_system_prompt = (
        f"{system_message}\n\n"
        f"{faq_context_injection}"
    )
    
    messages = [{"role": "system", "content": full_system_prompt}]
    clean_history = []
    
    for msg in history[:-1]:
        if isinstance(msg, dict):
            role, content = msg.get("role"), extract_text(msg.get("content"))
        else:
            role, content = getattr(msg, "role", None), extract_text(getattr(msg, "content", ""))
        if role and content:
            messages.append({"role": role, "content": content})
            clean_history.append({"role": role, "content": content})

    messages.append({"role": "user", "content": user_query})
    clean_history.append({"role": "user", "content": user_query})
    
    # Pre-populate the visualization header block securely
    clean_history.append({"role": "assistant", "content": nlp_metadata_prefix})

    try:
        for message_chunk in client.chat_completion(
            messages,
            max_tokens=int(max_tokens),
            stream=True,
            temperature=float(temperature),
            top_p=float(top_p),
        ):
            token = message_chunk.choices[0].delta.content
            if token is not None:
                clean_history[-1]["content"] += token
                yield clean_history
    except Exception as e:
        clean_history[-1]["content"] += f"\nAn error occurred while connecting to the model API: {str(e)}"
        yield clean_history

# Custom UI Theme Stylesheet
custom_css = """
#header-container {
    background-color: #eef2f5 !important;
    border-radius: 12px !important;
    padding: 18px !important;
    margin-bottom: 20px !important;
    border: 1px solid #dcdfe6 !important;
}
body.dark #header-container {
    background-color: #1f2937 !important;
    border-color: #374151 !important;
}
#header-container * { color: #1a1a1a !important; }
body.dark #header-container * { color: #f9fafb !important; }
#right-meta-card {
    background-color: #005f53 !important;
    border-radius: 12px !important;
    padding: 24px !important;
    color: white !important;
    text-align: center;
}
#right-meta-card * { color: white !important; }
#send-btn {
    background-color: #ff6600 !important;
    color: white !important;
    border: none !important;
    font-weight: bold !important;
}
#toggle-mode-btn {
    background-color: #4b5563 !important;
    color: white !important;
    border-radius: 20px !important;
    cursor: pointer;
    font-size: 13px !important;
}
.meta-status-text {
    text-align: left; 
    font-size: 13px; 
    line-height: 1.8; 
    margin-top: 20px;
    border-top: 1px solid rgba(255,255,255,0.2);
    padding-top: 15px;
}
"""

with gr.Blocks(css=custom_css) as demo:
    
    # 1. Header Section
    with gr.Row(elem_id="header-container"):
        with gr.Column(scale=3):
            gr.Markdown(
                """
                ### About Us
                At FAQ Agentic Hub, we are passionate about the power of structured, NLP-driven automation and its ability to transform customer workflows. This engine utilizes real-time Tokenization and Cosine Similarity computations.
                """
            )
        with gr.Column(scale=1):
            time_display = gr.HTML(value=get_thematic_datetime())
            toggle_btn = gr.Button("β˜€οΈ/πŸŒ™ βž” Change Mode (Day / Night)", elem_id="toggle-mode-btn")
            toggle_btn.click(None, None, None, js="() => document.body.classList.toggle('dark')")

    # 2. Workspace Layout
    with gr.Row():
        with gr.Column(scale=2):
            chatbot = gr.Chatbot(
                label="Lane NLP AI Support Stream", 
                value=[{"role": "assistant", "content": "Hello! I'm Lane, your upgraded NLP FAQ Support Agent. Ask me questions about shipping, returns, or certificates, and watch the cosine-similarity system extract matches in real-time."}]
            )
            
            with gr.Row():
                msg_input = gr.Textbox(
                    placeholder="Ask Lane anything regarding automated FAQ policies...", 
                    show_label=False, 
                    scale=4
                )
                send_btn = gr.Button("Send", variant="primary", scale=1, elem_id="send-btn")
            
            with gr.Row():
                suggest_btn1 = gr.Button("What are your shipping times and policies?")
                suggest_btn2 = gr.Button("How do I unlock my course completion certificate?")

        with gr.Column(scale=1):
            with gr.Group(elem_id="right-meta-card"):
                gr.HTML(
                    """
                    <div style='display: flex; justify-content: center; margin-bottom: 12px;'>
                        <div style='background: white; border-radius: 50%; padding: 12px; width: 110px; height: 110px; display: flex; align-items: center; justify-content: center; box-shadow: 0 4px 6px rgba(0,0,0,0.1);'>
                            <img src='https://fonts.gstatic.com/s/i/short-term/release/googlesymbols/robot/default/48px.svg' style='width: 70px; filter: invert(27%) sepia(91%) saturate(2853%) hue-rotate(152deg) brightness(93%) contrast(101%);'>
                        </div>
                    </div>
                    <h2 style='margin: 5px 0 2px 0; font-size: 24px;'>Lane <span style='font-size: 11px; background: rgba(255,255,255,0.25); padding: 2px 7px; border-radius: 4px; vertical-align: middle;'>NLP AI</span></h2>
                    <p style='margin: 0; opacity: 0.85; font-size: 14px;'>Similarity-Augmented Assistant</p>
                    
                    <div class='meta-status-text'>
                        <p>🟒 NLP Match Engine Online</p>
                        <p><strong>Vectorization Model:</strong> Text TF-IDF Vectorizer</p>
                        <p><strong>Processing Hardware:</strong> Base CPU Node</p>
                    </div>
                    """
                )
                
            with gr.Accordion("Advanced Agent Parameter Configurations", open=False):
                system_msg = gr.Textbox(
                    value="You are an expert AI FAQ Agent assistant. Answer user queries concisely, expanding intelligently upon matched context summaries when available.", 
                    label="System Prompt Context Injection"
                )
                max_tokens = gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens")
                temperature = gr.Slider(minimum=0.1, maximum=4.0, value=0.4, step=0.1, label="Temperature")
                top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (Nucleus Sampling)")

    # 3. Components Event Wiring
    send_btn.click(
        user_side, inputs=[msg_input, chatbot], outputs=[msg_input, chatbot], queue=False
    ).then(
        bot_side, inputs=[chatbot, system_msg, max_tokens, temperature, top_p], outputs=[chatbot]
    )
    
    msg_input.submit(
        user_side, inputs=[msg_input, chatbot], outputs=[msg_input, chatbot], queue=False
    ).then(
        bot_side, inputs=[chatbot, system_msg, max_tokens, temperature, top_p], outputs=[chatbot]
    )

    suggest_btn1.click(
        user_side, inputs=[suggest_btn1, chatbot], outputs=[msg_input, chatbot], queue=False
    ).then(
        bot_side, inputs=[chatbot, system_msg, max_tokens, temperature, top_p], outputs=[chatbot]
    )
    
    suggest_btn2.click(
        user_side, inputs=[suggest_btn2, chatbot], outputs=[msg_input, chatbot], queue=False
    ).then(
        bot_side, inputs=[chatbot, system_msg, max_tokens, temperature, top_p], outputs=[chatbot]
    )

    # Running digital time clock update loop
    clock_timer = gr.Timer(1.0)
    clock_timer.tick(fn=get_thematic_datetime, outputs=time_display, queue=False)

if __name__ == "__main__":
    demo.launch(theme=gr.themes.Default())