Spaces:
Sleeping
Sleeping
Update src/streamlit_app.py
Browse files- src/streamlit_app.py +217 -254
src/streamlit_app.py
CHANGED
|
@@ -4,8 +4,7 @@ import json
|
|
| 4 |
from typing import Dict, List, Optional
|
| 5 |
import re
|
| 6 |
from urllib.parse import quote
|
| 7 |
-
import
|
| 8 |
-
import aiohttp
|
| 9 |
|
| 10 |
# Configure page
|
| 11 |
st.set_page_config(
|
|
@@ -41,48 +40,38 @@ class OllamaLLM:
|
|
| 41 |
def check_connection(self) -> bool:
|
| 42 |
"""Check if Ollama is running"""
|
| 43 |
try:
|
| 44 |
-
response = requests.get(self.models_url, timeout=
|
| 45 |
return response.status_code == 200
|
| 46 |
-
except:
|
| 47 |
return False
|
| 48 |
|
| 49 |
def get_available_models(self) -> List[str]:
|
| 50 |
"""Get list of available models"""
|
| 51 |
try:
|
| 52 |
-
response = requests.get(self.models_url, timeout=
|
| 53 |
if response.status_code == 200:
|
| 54 |
data = response.json()
|
| 55 |
-
|
|
|
|
| 56 |
return []
|
| 57 |
-
except:
|
| 58 |
return []
|
| 59 |
|
| 60 |
def generate_summary(self, text: str, model: str = "llama3.2", language: str = "English",
|
| 61 |
summary_type: str = "concise") -> str:
|
| 62 |
"""Generate AI summary using local LLM"""
|
| 63 |
try:
|
| 64 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
| 65 |
if summary_type == "concise":
|
| 66 |
-
prompt = f"
|
| 67 |
-
Make it clear and informative:
|
| 68 |
-
|
| 69 |
-
{text}
|
| 70 |
-
|
| 71 |
-
Summary:"""
|
| 72 |
elif summary_type == "detailed":
|
| 73 |
-
prompt = f"
|
| 74 |
-
Include key points, important facts, and context:
|
| 75 |
-
|
| 76 |
-
{text}
|
| 77 |
-
|
| 78 |
-
Detailed Summary:"""
|
| 79 |
else: # explanatory
|
| 80 |
-
prompt = f"
|
| 81 |
-
easy-to-understand way as if explaining to someone unfamiliar with the topic:
|
| 82 |
-
|
| 83 |
-
{text}
|
| 84 |
-
|
| 85 |
-
Explanation:"""
|
| 86 |
|
| 87 |
# Request to Ollama
|
| 88 |
payload = {
|
|
@@ -91,7 +80,7 @@ class OllamaLLM:
|
|
| 91 |
"stream": False,
|
| 92 |
"options": {
|
| 93 |
"temperature": 0.7,
|
| 94 |
-
"num_predict":
|
| 95 |
}
|
| 96 |
}
|
| 97 |
|
|
@@ -99,22 +88,20 @@ class OllamaLLM:
|
|
| 99 |
|
| 100 |
if response.status_code == 200:
|
| 101 |
data = response.json()
|
| 102 |
-
|
|
|
|
| 103 |
else:
|
| 104 |
-
return f"Error: {response.status_code}"
|
| 105 |
|
|
|
|
|
|
|
| 106 |
except Exception as e:
|
| 107 |
-
return f"
|
| 108 |
|
| 109 |
def translate_text(self, text: str, target_language: str, model: str = "llama3.2") -> str:
|
| 110 |
"""Translate text using local LLM"""
|
| 111 |
try:
|
| 112 |
-
prompt = f"
|
| 113 |
-
Provide only the translation, no additional text:
|
| 114 |
-
|
| 115 |
-
{text}
|
| 116 |
-
|
| 117 |
-
Translation:"""
|
| 118 |
|
| 119 |
payload = {
|
| 120 |
"model": model,
|
|
@@ -122,7 +109,7 @@ class OllamaLLM:
|
|
| 122 |
"stream": False,
|
| 123 |
"options": {
|
| 124 |
"temperature": 0.3,
|
| 125 |
-
"num_predict":
|
| 126 |
}
|
| 127 |
}
|
| 128 |
|
|
@@ -130,11 +117,12 @@ class OllamaLLM:
|
|
| 130 |
|
| 131 |
if response.status_code == 200:
|
| 132 |
data = response.json()
|
| 133 |
-
|
|
|
|
| 134 |
else:
|
| 135 |
-
return text
|
| 136 |
|
| 137 |
-
except Exception
|
| 138 |
return text
|
| 139 |
|
| 140 |
class WikipediaAPI:
|
|
@@ -160,6 +148,7 @@ class WikipediaAPI:
|
|
| 160 |
|
| 161 |
data = response.json()
|
| 162 |
return data.get("query", {}).get("search", [])
|
|
|
|
| 163 |
except Exception as e:
|
| 164 |
st.error(f"Search error: {str(e)}")
|
| 165 |
return []
|
|
@@ -174,11 +163,11 @@ class WikipediaAPI:
|
|
| 174 |
response.raise_for_status()
|
| 175 |
|
| 176 |
return response.json()
|
|
|
|
| 177 |
except Exception as e:
|
| 178 |
-
st.error(f"Summary error: {str(e)}")
|
| 179 |
return None
|
| 180 |
|
| 181 |
-
def get_page_content(self, title: str, lang: str = "en", char_limit: int =
|
| 182 |
"""Get page content sections"""
|
| 183 |
try:
|
| 184 |
params = {
|
|
@@ -204,23 +193,34 @@ class WikipediaAPI:
|
|
| 204 |
return page_data["extract"]
|
| 205 |
|
| 206 |
return None
|
| 207 |
-
|
| 208 |
-
|
| 209 |
return None
|
| 210 |
|
| 211 |
def clean_html(text: str) -> str:
|
| 212 |
"""Remove HTML tags from text"""
|
|
|
|
|
|
|
| 213 |
clean = re.compile('<.*?>')
|
| 214 |
return re.sub(clean, '', text)
|
| 215 |
|
| 216 |
def simple_summarize(text: str, max_sentences: int = 3) -> str:
|
| 217 |
"""Fallback simple text summarization"""
|
|
|
|
|
|
|
|
|
|
| 218 |
sentences = text.split('. ')
|
|
|
|
|
|
|
|
|
|
| 219 |
summary_sentences = sentences[:max_sentences]
|
| 220 |
-
|
|
|
|
|
|
|
|
|
|
| 221 |
|
| 222 |
def main():
|
| 223 |
-
# Custom CSS
|
| 224 |
st.markdown("""
|
| 225 |
<style>
|
| 226 |
.main-header {
|
|
@@ -230,13 +230,13 @@ def main():
|
|
| 230 |
}
|
| 231 |
.search-container {
|
| 232 |
background-color: #f8f9fa;
|
| 233 |
-
padding:
|
| 234 |
border-radius: 10px;
|
| 235 |
margin-bottom: 1rem;
|
| 236 |
}
|
| 237 |
.result-card {
|
| 238 |
background-color: white;
|
| 239 |
-
padding:
|
| 240 |
border-radius: 8px;
|
| 241 |
border: 1px solid #dee2e6;
|
| 242 |
margin-bottom: 1rem;
|
|
@@ -245,13 +245,14 @@ def main():
|
|
| 245 |
.article-title {
|
| 246 |
color: #007bff;
|
| 247 |
font-weight: bold;
|
|
|
|
| 248 |
margin-bottom: 0.5rem;
|
| 249 |
}
|
| 250 |
-
.
|
| 251 |
-
padding: 0.
|
| 252 |
-
border-radius:
|
| 253 |
margin-bottom: 1rem;
|
| 254 |
-
font-
|
| 255 |
}
|
| 256 |
.status-connected {
|
| 257 |
background-color: #d4edda;
|
|
@@ -270,51 +271,51 @@ def main():
|
|
| 270 |
border-left: 4px solid #007bff;
|
| 271 |
margin: 1rem 0;
|
| 272 |
}
|
| 273 |
-
@media (max-width: 768px) {
|
| 274 |
-
.stSelectbox, .stTextInput {
|
| 275 |
-
font-size: 16px;
|
| 276 |
-
}
|
| 277 |
-
}
|
| 278 |
</style>
|
| 279 |
""", unsafe_allow_html=True)
|
| 280 |
|
| 281 |
# Header
|
| 282 |
-
st.markdown("<h1 class='main-header'>π€ WikiBot - AI-Powered
|
| 283 |
-
st.markdown("<p style='text-align: center; color: #666;'>
|
|
|
|
|
|
|
|
|
|
|
|
|
| 284 |
|
| 285 |
# Initialize APIs
|
| 286 |
wiki_api = WikipediaAPI()
|
| 287 |
llm = OllamaLLM()
|
| 288 |
|
| 289 |
# Check LLM connection
|
| 290 |
-
|
| 291 |
-
|
|
|
|
| 292 |
|
| 293 |
-
#
|
| 294 |
if llm_connected:
|
| 295 |
st.markdown(f"""
|
| 296 |
-
<div class='
|
| 297 |
-
β
|
| 298 |
</div>
|
| 299 |
""", unsafe_allow_html=True)
|
| 300 |
else:
|
| 301 |
st.markdown("""
|
| 302 |
-
<div class='
|
| 303 |
-
β
|
| 304 |
</div>
|
| 305 |
""", unsafe_allow_html=True)
|
| 306 |
-
st.info("To enable AI features: Install Ollama from https://ollama.ai and run `ollama serve`")
|
| 307 |
|
| 308 |
-
#
|
| 309 |
st.markdown("<div class='search-container'>", unsafe_allow_html=True)
|
| 310 |
|
|
|
|
| 311 |
col1, col2 = st.columns([3, 1])
|
| 312 |
|
| 313 |
with col1:
|
| 314 |
query = st.text_input(
|
| 315 |
-
"π Search
|
| 316 |
-
placeholder="e.g., '
|
| 317 |
-
help="Enter your search query
|
| 318 |
)
|
| 319 |
|
| 320 |
with col2:
|
|
@@ -324,218 +325,180 @@ def main():
|
|
| 324 |
index=0
|
| 325 |
)
|
| 326 |
|
| 327 |
-
#
|
| 328 |
-
|
| 329 |
-
|
| 330 |
-
|
| 331 |
-
|
| 332 |
-
|
| 333 |
-
|
| 334 |
-
with col2:
|
| 335 |
-
if llm_connected:
|
| 336 |
-
summary_mode = st.selectbox(
|
| 337 |
-
"AI Summary Type",
|
| 338 |
-
["concise", "detailed", "explanatory"],
|
| 339 |
-
index=0
|
| 340 |
-
)
|
| 341 |
-
else:
|
| 342 |
-
summary_mode = st.selectbox(
|
| 343 |
-
"Summary Type",
|
| 344 |
-
["short", "medium", "long"],
|
| 345 |
-
index=1
|
| 346 |
-
)
|
| 347 |
-
|
| 348 |
-
with col3:
|
| 349 |
-
if llm_connected and available_models:
|
| 350 |
-
selected_model = st.selectbox(
|
| 351 |
-
"LLM Model",
|
| 352 |
-
options=available_models,
|
| 353 |
-
index=0
|
| 354 |
-
)
|
| 355 |
-
else:
|
| 356 |
-
st.info("No models available")
|
| 357 |
-
selected_model = None
|
| 358 |
-
|
| 359 |
-
# Translation options
|
| 360 |
if llm_connected:
|
| 361 |
-
|
| 362 |
-
|
| 363 |
-
|
| 364 |
-
|
| 365 |
-
|
| 366 |
-
|
| 367 |
-
|
| 368 |
-
|
| 369 |
-
|
| 370 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 371 |
|
| 372 |
st.markdown("</div>", unsafe_allow_html=True)
|
| 373 |
|
| 374 |
# Search button
|
| 375 |
-
|
| 376 |
-
|
| 377 |
-
|
| 378 |
-
|
| 379 |
-
|
| 380 |
-
|
| 381 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 382 |
|
| 383 |
-
|
| 384 |
-
|
| 385 |
-
|
| 386 |
-
|
| 387 |
-
|
| 388 |
-
|
| 389 |
-
|
| 390 |
-
|
| 391 |
-
|
| 392 |
-
|
| 393 |
-
|
| 394 |
-
|
| 395 |
-
|
| 396 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 397 |
|
| 398 |
-
if
|
| 399 |
-
|
| 400 |
-
|
| 401 |
-
|
| 402 |
-
content,
|
| 403 |
-
selected_model,
|
| 404 |
-
selected_lang,
|
| 405 |
-
summary_mode
|
| 406 |
-
)
|
| 407 |
|
| 408 |
-
|
| 409 |
-
|
| 410 |
-
st.
|
| 411 |
-
|
| 412 |
-
|
| 413 |
-
|
| 414 |
-
|
| 415 |
-
with st.spinner(f"Translating to {target_lang}..."):
|
| 416 |
-
translated = llm.translate_text(ai_summary, target_lang, selected_model)
|
| 417 |
-
if translated != ai_summary:
|
| 418 |
-
st.markdown(f"**π Translation to {target_lang}:**")
|
| 419 |
-
st.write(translated)
|
| 420 |
-
|
| 421 |
-
st.markdown("</div>", unsafe_allow_html=True)
|
| 422 |
-
else:
|
| 423 |
-
# Fallback to simple summary
|
| 424 |
-
st.warning("AI summary failed, using fallback")
|
| 425 |
-
fallback_summary = simple_summarize(content, 3)
|
| 426 |
-
st.write(fallback_summary)
|
| 427 |
-
|
| 428 |
-
elif summary_data:
|
| 429 |
-
# Standard Wikipedia summary
|
| 430 |
-
summary_text = summary_data.get("extract", "")
|
| 431 |
-
if not llm_connected:
|
| 432 |
-
if summary_mode == "short":
|
| 433 |
-
summary_text = simple_summarize(summary_text, 2)
|
| 434 |
-
elif summary_mode == "medium":
|
| 435 |
-
summary_text = simple_summarize(summary_text, 4)
|
| 436 |
|
| 437 |
-
st.
|
| 438 |
-
|
| 439 |
else:
|
| 440 |
-
|
| 441 |
-
|
| 442 |
-
|
| 443 |
-
|
| 444 |
-
|
| 445 |
-
|
| 446 |
-
|
| 447 |
-
|
| 448 |
-
|
| 449 |
-
|
| 450 |
-
|
| 451 |
-
|
| 452 |
-
|
| 453 |
-
|
| 454 |
-
|
| 455 |
-
|
| 456 |
-
|
| 457 |
-
|
| 458 |
-
|
| 459 |
-
height=300,
|
| 460 |
-
key=f"content_{idx}"
|
| 461 |
-
)
|
| 462 |
-
else:
|
| 463 |
-
st.warning("Detailed content not available")
|
| 464 |
-
|
| 465 |
-
st.markdown("</div>", unsafe_allow_html=True)
|
| 466 |
-
st.markdown("---")
|
| 467 |
|
| 468 |
-
|
| 469 |
-
|
| 470 |
-
|
| 471 |
-
|
| 472 |
-
|
| 473 |
-
|
|
|
|
| 474 |
|
| 475 |
-
|
|
|
|
|
|
|
|
|
|
| 476 |
st.markdown("---")
|
| 477 |
col1, col2, col3, col4 = st.columns(4)
|
| 478 |
|
| 479 |
with col1:
|
| 480 |
st.metric("π Languages", len(LANGUAGES))
|
| 481 |
-
|
| 482 |
with col2:
|
| 483 |
-
st.metric("π€ LLM
|
| 484 |
-
|
| 485 |
with col3:
|
| 486 |
st.metric("π Models", len(available_models))
|
| 487 |
-
|
| 488 |
with col4:
|
| 489 |
-
st.metric("
|
| 490 |
|
| 491 |
-
# Setup
|
| 492 |
-
with st.expander("π οΈ Setup
|
| 493 |
st.markdown("""
|
| 494 |
-
###
|
| 495 |
|
| 496 |
-
1.
|
| 497 |
-
|
| 498 |
-
|
| 499 |
-
|
| 500 |
-
|
| 501 |
-
# Windows - Download from https://ollama.ai
|
| 502 |
-
```
|
| 503 |
|
| 504 |
-
|
| 505 |
-
|
| 506 |
-
ollama pull llama3.2
|
| 507 |
-
# or
|
| 508 |
-
ollama pull mistral
|
| 509 |
-
ollama pull codellama
|
| 510 |
-
```
|
| 511 |
|
| 512 |
-
|
| 513 |
-
|
| 514 |
-
|
| 515 |
-
|
| 516 |
|
| 517 |
-
|
|
|
|
|
|
|
|
|
|
| 518 |
|
| 519 |
-
|
| 520 |
-
- **llama3.2** - Great for general summarization
|
| 521 |
-
- **mistral** - Fast and efficient
|
| 522 |
-
- **codellama** - Good for technical content
|
| 523 |
-
""")
|
| 524 |
-
|
| 525 |
-
# Usage examples
|
| 526 |
-
with st.expander("π‘ Usage Examples"):
|
| 527 |
-
st.markdown("""
|
| 528 |
-
**Try these example queries:**
|
| 529 |
-
- "Explain Kargil War in Telugu" β AI generates Telugu explanation
|
| 530 |
-
- "Machine Learning" β Detailed AI summary with translation
|
| 531 |
-
- "Climate Change" β AI explanatory summary
|
| 532 |
-
- "Quantum Computing" β Technical AI analysis
|
| 533 |
|
| 534 |
-
|
| 535 |
-
-
|
| 536 |
-
-
|
| 537 |
-
-
|
| 538 |
-
- π Enhanced content understanding
|
| 539 |
""")
|
| 540 |
|
| 541 |
if __name__ == "__main__":
|
|
|
|
| 4 |
from typing import Dict, List, Optional
|
| 5 |
import re
|
| 6 |
from urllib.parse import quote
|
| 7 |
+
import time
|
|
|
|
| 8 |
|
| 9 |
# Configure page
|
| 10 |
st.set_page_config(
|
|
|
|
| 40 |
def check_connection(self) -> bool:
|
| 41 |
"""Check if Ollama is running"""
|
| 42 |
try:
|
| 43 |
+
response = requests.get(self.models_url, timeout=3)
|
| 44 |
return response.status_code == 200
|
| 45 |
+
except Exception:
|
| 46 |
return False
|
| 47 |
|
| 48 |
def get_available_models(self) -> List[str]:
|
| 49 |
"""Get list of available models"""
|
| 50 |
try:
|
| 51 |
+
response = requests.get(self.models_url, timeout=5)
|
| 52 |
if response.status_code == 200:
|
| 53 |
data = response.json()
|
| 54 |
+
models = data.get("models", [])
|
| 55 |
+
return [model["name"] for model in models]
|
| 56 |
return []
|
| 57 |
+
except Exception:
|
| 58 |
return []
|
| 59 |
|
| 60 |
def generate_summary(self, text: str, model: str = "llama3.2", language: str = "English",
|
| 61 |
summary_type: str = "concise") -> str:
|
| 62 |
"""Generate AI summary using local LLM"""
|
| 63 |
try:
|
| 64 |
+
# Truncate text if too long
|
| 65 |
+
if len(text) > 2000:
|
| 66 |
+
text = text[:2000] + "..."
|
| 67 |
+
|
| 68 |
+
# Craft prompt based on summary type
|
| 69 |
if summary_type == "concise":
|
| 70 |
+
prompt = f"Summarize this Wikipedia content in {language} in 2-3 clear sentences:\n\n{text}\n\nSummary:"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
elif summary_type == "detailed":
|
| 72 |
+
prompt = f"Provide a detailed summary of this Wikipedia content in {language}. Include key points and important facts:\n\n{text}\n\nDetailed Summary:"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 73 |
else: # explanatory
|
| 74 |
+
prompt = f"Explain this Wikipedia content in {language} in simple terms that anyone can understand:\n\n{text}\n\nExplanation:"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
|
| 76 |
# Request to Ollama
|
| 77 |
payload = {
|
|
|
|
| 80 |
"stream": False,
|
| 81 |
"options": {
|
| 82 |
"temperature": 0.7,
|
| 83 |
+
"num_predict": 300 if summary_type == "detailed" else 150
|
| 84 |
}
|
| 85 |
}
|
| 86 |
|
|
|
|
| 88 |
|
| 89 |
if response.status_code == 200:
|
| 90 |
data = response.json()
|
| 91 |
+
summary = data.get("response", "").strip()
|
| 92 |
+
return summary if summary else "No summary generated"
|
| 93 |
else:
|
| 94 |
+
return f"Error: Status {response.status_code}"
|
| 95 |
|
| 96 |
+
except requests.exceptions.Timeout:
|
| 97 |
+
return "Error: Request timeout - try a smaller text"
|
| 98 |
except Exception as e:
|
| 99 |
+
return f"Error: {str(e)}"
|
| 100 |
|
| 101 |
def translate_text(self, text: str, target_language: str, model: str = "llama3.2") -> str:
|
| 102 |
"""Translate text using local LLM"""
|
| 103 |
try:
|
| 104 |
+
prompt = f"Translate this text to {target_language}. Only provide the translation:\n\n{text}\n\nTranslation:"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 105 |
|
| 106 |
payload = {
|
| 107 |
"model": model,
|
|
|
|
| 109 |
"stream": False,
|
| 110 |
"options": {
|
| 111 |
"temperature": 0.3,
|
| 112 |
+
"num_predict": 200
|
| 113 |
}
|
| 114 |
}
|
| 115 |
|
|
|
|
| 117 |
|
| 118 |
if response.status_code == 200:
|
| 119 |
data = response.json()
|
| 120 |
+
translation = data.get("response", "").strip()
|
| 121 |
+
return translation if translation else text
|
| 122 |
else:
|
| 123 |
+
return text
|
| 124 |
|
| 125 |
+
except Exception:
|
| 126 |
return text
|
| 127 |
|
| 128 |
class WikipediaAPI:
|
|
|
|
| 148 |
|
| 149 |
data = response.json()
|
| 150 |
return data.get("query", {}).get("search", [])
|
| 151 |
+
|
| 152 |
except Exception as e:
|
| 153 |
st.error(f"Search error: {str(e)}")
|
| 154 |
return []
|
|
|
|
| 163 |
response.raise_for_status()
|
| 164 |
|
| 165 |
return response.json()
|
| 166 |
+
|
| 167 |
except Exception as e:
|
|
|
|
| 168 |
return None
|
| 169 |
|
| 170 |
+
def get_page_content(self, title: str, lang: str = "en", char_limit: int = 2000) -> Optional[str]:
|
| 171 |
"""Get page content sections"""
|
| 172 |
try:
|
| 173 |
params = {
|
|
|
|
| 193 |
return page_data["extract"]
|
| 194 |
|
| 195 |
return None
|
| 196 |
+
|
| 197 |
+
except Exception:
|
| 198 |
return None
|
| 199 |
|
| 200 |
def clean_html(text: str) -> str:
|
| 201 |
"""Remove HTML tags from text"""
|
| 202 |
+
if not text:
|
| 203 |
+
return ""
|
| 204 |
clean = re.compile('<.*?>')
|
| 205 |
return re.sub(clean, '', text)
|
| 206 |
|
| 207 |
def simple_summarize(text: str, max_sentences: int = 3) -> str:
|
| 208 |
"""Fallback simple text summarization"""
|
| 209 |
+
if not text:
|
| 210 |
+
return "No content available"
|
| 211 |
+
|
| 212 |
sentences = text.split('. ')
|
| 213 |
+
if len(sentences) <= max_sentences:
|
| 214 |
+
return text
|
| 215 |
+
|
| 216 |
summary_sentences = sentences[:max_sentences]
|
| 217 |
+
result = '. '.join(summary_sentences)
|
| 218 |
+
if not result.endswith('.'):
|
| 219 |
+
result += '.'
|
| 220 |
+
return result
|
| 221 |
|
| 222 |
def main():
|
| 223 |
+
# Custom CSS
|
| 224 |
st.markdown("""
|
| 225 |
<style>
|
| 226 |
.main-header {
|
|
|
|
| 230 |
}
|
| 231 |
.search-container {
|
| 232 |
background-color: #f8f9fa;
|
| 233 |
+
padding: 1.5rem;
|
| 234 |
border-radius: 10px;
|
| 235 |
margin-bottom: 1rem;
|
| 236 |
}
|
| 237 |
.result-card {
|
| 238 |
background-color: white;
|
| 239 |
+
padding: 1.5rem;
|
| 240 |
border-radius: 8px;
|
| 241 |
border: 1px solid #dee2e6;
|
| 242 |
margin-bottom: 1rem;
|
|
|
|
| 245 |
.article-title {
|
| 246 |
color: #007bff;
|
| 247 |
font-weight: bold;
|
| 248 |
+
font-size: 1.2rem;
|
| 249 |
margin-bottom: 0.5rem;
|
| 250 |
}
|
| 251 |
+
.status-box {
|
| 252 |
+
padding: 0.8rem;
|
| 253 |
+
border-radius: 8px;
|
| 254 |
margin-bottom: 1rem;
|
| 255 |
+
font-weight: bold;
|
| 256 |
}
|
| 257 |
.status-connected {
|
| 258 |
background-color: #d4edda;
|
|
|
|
| 271 |
border-left: 4px solid #007bff;
|
| 272 |
margin: 1rem 0;
|
| 273 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 274 |
</style>
|
| 275 |
""", unsafe_allow_html=True)
|
| 276 |
|
| 277 |
# Header
|
| 278 |
+
st.markdown("<h1 class='main-header'>π€ WikiBot - AI-Powered Assistant</h1>", unsafe_allow_html=True)
|
| 279 |
+
st.markdown("<p style='text-align: center; color: #666;'>Wikipedia + Local LLM Intelligence</p>", unsafe_allow_html=True)
|
| 280 |
+
|
| 281 |
+
# Initialize session state
|
| 282 |
+
if 'search_results' not in st.session_state:
|
| 283 |
+
st.session_state.search_results = []
|
| 284 |
|
| 285 |
# Initialize APIs
|
| 286 |
wiki_api = WikipediaAPI()
|
| 287 |
llm = OllamaLLM()
|
| 288 |
|
| 289 |
# Check LLM connection
|
| 290 |
+
with st.spinner("Checking Ollama connection..."):
|
| 291 |
+
llm_connected = llm.check_connection()
|
| 292 |
+
available_models = llm.get_available_models() if llm_connected else []
|
| 293 |
|
| 294 |
+
# Status display
|
| 295 |
if llm_connected:
|
| 296 |
st.markdown(f"""
|
| 297 |
+
<div class='status-box status-connected'>
|
| 298 |
+
β
Ollama Connected - {len(available_models)} models available
|
| 299 |
</div>
|
| 300 |
""", unsafe_allow_html=True)
|
| 301 |
else:
|
| 302 |
st.markdown("""
|
| 303 |
+
<div class='status-box status-disconnected'>
|
| 304 |
+
β Ollama Offline - Basic mode only
|
| 305 |
</div>
|
| 306 |
""", unsafe_allow_html=True)
|
|
|
|
| 307 |
|
| 308 |
+
# Main search interface
|
| 309 |
st.markdown("<div class='search-container'>", unsafe_allow_html=True)
|
| 310 |
|
| 311 |
+
# Search inputs
|
| 312 |
col1, col2 = st.columns([3, 1])
|
| 313 |
|
| 314 |
with col1:
|
| 315 |
query = st.text_input(
|
| 316 |
+
"π Search Query",
|
| 317 |
+
placeholder="e.g., 'Artificial Intelligence', 'Kargil War'",
|
| 318 |
+
help="Enter your Wikipedia search query"
|
| 319 |
)
|
| 320 |
|
| 321 |
with col2:
|
|
|
|
| 325 |
index=0
|
| 326 |
)
|
| 327 |
|
| 328 |
+
# Options
|
| 329 |
+
col1, col2, col3 = st.columns(3)
|
| 330 |
+
|
| 331 |
+
with col1:
|
| 332 |
+
num_results = st.slider("Results", 1, 8, 3)
|
| 333 |
+
|
| 334 |
+
with col2:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 335 |
if llm_connected:
|
| 336 |
+
summary_type = st.selectbox(
|
| 337 |
+
"AI Summary",
|
| 338 |
+
["concise", "detailed", "explanatory"]
|
| 339 |
+
)
|
| 340 |
+
else:
|
| 341 |
+
summary_type = st.selectbox(
|
| 342 |
+
"Summary",
|
| 343 |
+
["short", "medium", "long"]
|
| 344 |
+
)
|
| 345 |
+
|
| 346 |
+
with col3:
|
| 347 |
+
if llm_connected and available_models:
|
| 348 |
+
selected_model = st.selectbox("Model", available_models)
|
| 349 |
+
else:
|
| 350 |
+
selected_model = None
|
| 351 |
+
st.info("No models")
|
| 352 |
+
|
| 353 |
+
# Translation option
|
| 354 |
+
if llm_connected:
|
| 355 |
+
enable_translation = st.checkbox("π Enable Translation")
|
| 356 |
+
if enable_translation:
|
| 357 |
+
target_lang = st.selectbox(
|
| 358 |
+
"Translate to",
|
| 359 |
+
[lang for lang in LANGUAGES.keys() if lang != selected_lang]
|
| 360 |
+
)
|
| 361 |
|
| 362 |
st.markdown("</div>", unsafe_allow_html=True)
|
| 363 |
|
| 364 |
# Search button
|
| 365 |
+
search_clicked = st.button("π Search", type="primary", use_container_width=True)
|
| 366 |
+
|
| 367 |
+
if search_clicked and query:
|
| 368 |
+
lang_code = LANGUAGES[selected_lang]
|
| 369 |
+
|
| 370 |
+
with st.spinner("Searching Wikipedia..."):
|
| 371 |
+
search_results = wiki_api.search_articles(query, lang_code, num_results)
|
| 372 |
+
st.session_state.search_results = search_results
|
| 373 |
+
|
| 374 |
+
# Display results
|
| 375 |
+
if st.session_state.search_results:
|
| 376 |
+
st.success(f"Found {len(st.session_state.search_results)} results")
|
| 377 |
+
|
| 378 |
+
for idx, result in enumerate(st.session_state.search_results):
|
| 379 |
+
with st.container():
|
| 380 |
+
st.markdown("<div class='result-card'>", unsafe_allow_html=True)
|
| 381 |
|
| 382 |
+
# Title
|
| 383 |
+
title = result.get("title", "")
|
| 384 |
+
st.markdown(f"<div class='article-title'>{idx+1}. {title}</div>", unsafe_allow_html=True)
|
| 385 |
+
|
| 386 |
+
# Get content
|
| 387 |
+
lang_code = LANGUAGES[selected_lang]
|
| 388 |
+
summary_data = wiki_api.get_page_summary(title, lang_code)
|
| 389 |
+
|
| 390 |
+
# Show thumbnail
|
| 391 |
+
if summary_data and "thumbnail" in summary_data:
|
| 392 |
+
col1, col2 = st.columns([1, 4])
|
| 393 |
+
with col1:
|
| 394 |
+
st.image(summary_data["thumbnail"]["source"], width=100)
|
| 395 |
+
content_col = col2
|
| 396 |
+
else:
|
| 397 |
+
content_col = st
|
| 398 |
+
|
| 399 |
+
with content_col:
|
| 400 |
+
# AI Summary
|
| 401 |
+
if llm_connected and selected_model:
|
| 402 |
+
# Get detailed content for AI
|
| 403 |
+
detailed_content = wiki_api.get_page_content(title, lang_code)
|
| 404 |
+
|
| 405 |
+
if detailed_content:
|
| 406 |
+
with st.spinner("Generating AI summary..."):
|
| 407 |
+
ai_summary = llm.generate_summary(
|
| 408 |
+
detailed_content,
|
| 409 |
+
selected_model,
|
| 410 |
+
selected_lang,
|
| 411 |
+
summary_type
|
| 412 |
+
)
|
| 413 |
|
| 414 |
+
if ai_summary and not ai_summary.startswith("Error"):
|
| 415 |
+
st.markdown("<div class='ai-summary'>", unsafe_allow_html=True)
|
| 416 |
+
st.markdown("**π€ AI Summary:**")
|
| 417 |
+
st.write(ai_summary)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 418 |
|
| 419 |
+
# Translation
|
| 420 |
+
if 'enable_translation' in locals() and enable_translation:
|
| 421 |
+
with st.spinner("Translating..."):
|
| 422 |
+
translated = llm.translate_text(ai_summary, target_lang, selected_model)
|
| 423 |
+
if translated != ai_summary:
|
| 424 |
+
st.markdown(f"**π {target_lang}:**")
|
| 425 |
+
st.write(translated)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 426 |
|
| 427 |
+
st.markdown("</div>", unsafe_allow_html=True)
|
|
|
|
| 428 |
else:
|
| 429 |
+
st.warning("AI summary failed")
|
| 430 |
+
if summary_data:
|
| 431 |
+
basic_summary = summary_data.get("extract", "")
|
| 432 |
+
st.write(simple_summarize(basic_summary, 3))
|
| 433 |
+
else:
|
| 434 |
+
st.warning("Could not fetch detailed content")
|
| 435 |
+
|
| 436 |
+
else:
|
| 437 |
+
# Basic summary
|
| 438 |
+
if summary_data:
|
| 439 |
+
basic_summary = summary_data.get("extract", "")
|
| 440 |
+
if summary_type == "short":
|
| 441 |
+
basic_summary = simple_summarize(basic_summary, 2)
|
| 442 |
+
elif summary_type == "medium":
|
| 443 |
+
basic_summary = simple_summarize(basic_summary, 4)
|
| 444 |
+
st.write(basic_summary)
|
| 445 |
+
else:
|
| 446 |
+
snippet = clean_html(result.get("snippet", ""))
|
| 447 |
+
st.write(snippet)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 448 |
|
| 449 |
+
# Wikipedia link
|
| 450 |
+
if summary_data and "content_urls" in summary_data:
|
| 451 |
+
wiki_url = summary_data["content_urls"]["desktop"]["page"]
|
| 452 |
+
st.markdown(f"[π Read on Wikipedia]({wiki_url})")
|
| 453 |
+
|
| 454 |
+
st.markdown("</div>", unsafe_allow_html=True)
|
| 455 |
+
st.markdown("---")
|
| 456 |
|
| 457 |
+
elif search_clicked and not query:
|
| 458 |
+
st.warning("Please enter a search query")
|
| 459 |
+
|
| 460 |
+
# Footer stats
|
| 461 |
st.markdown("---")
|
| 462 |
col1, col2, col3, col4 = st.columns(4)
|
| 463 |
|
| 464 |
with col1:
|
| 465 |
st.metric("π Languages", len(LANGUAGES))
|
|
|
|
| 466 |
with col2:
|
| 467 |
+
st.metric("π€ LLM", "ON" if llm_connected else "OFF")
|
|
|
|
| 468 |
with col3:
|
| 469 |
st.metric("π Models", len(available_models))
|
|
|
|
| 470 |
with col4:
|
| 471 |
+
st.metric("π Results", len(st.session_state.search_results))
|
| 472 |
|
| 473 |
+
# Setup guide
|
| 474 |
+
with st.expander("π οΈ Ollama Setup Guide"):
|
| 475 |
st.markdown("""
|
| 476 |
+
### Quick Setup:
|
| 477 |
|
| 478 |
+
**1. Install Ollama:**
|
| 479 |
+
```bash
|
| 480 |
+
# macOS/Linux
|
| 481 |
+
curl -fsSL https://ollama.ai/install.sh | sh
|
|
|
|
|
|
|
|
|
|
| 482 |
|
| 483 |
+
# Windows: Download from https://ollama.ai
|
| 484 |
+
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 485 |
|
| 486 |
+
**2. Pull a model:**
|
| 487 |
+
```bash
|
| 488 |
+
ollama pull llama3.2
|
| 489 |
+
```
|
| 490 |
|
| 491 |
+
**3. Start server:**
|
| 492 |
+
```bash
|
| 493 |
+
ollama serve
|
| 494 |
+
```
|
| 495 |
|
| 496 |
+
**4. Refresh this page!**
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 497 |
|
| 498 |
+
### Recommended Models:
|
| 499 |
+
- `llama3.2` - Best overall performance
|
| 500 |
+
- `mistral` - Fast and efficient
|
| 501 |
+
- `qwen2` - Good for multilingual content
|
|
|
|
| 502 |
""")
|
| 503 |
|
| 504 |
if __name__ == "__main__":
|