Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,1922 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
-
from selenium import webdriver
|
| 3 |
-
from selenium.webdriver.chrome.service import Service
|
| 4 |
-
from selenium.webdriver.chrome.options import Options
|
| 5 |
-
from selenium.webdriver.common.by import By
|
| 6 |
-
from selenium.webdriver.support.ui import WebDriverWait
|
| 7 |
-
from selenium.webdriver.support import expected_conditions as EC
|
| 8 |
-
from selenium.webdriver.common.keys import Keys
|
| 9 |
-
from selenium.webdriver.common.action_chains import ActionChains
|
| 10 |
-
from webdriver_manager.chrome import ChromeDriverManager
|
| 11 |
-
from webdriver_manager.core.os_manager import ChromeType
|
| 12 |
-
import time
|
| 13 |
import logging
|
| 14 |
-
import
|
| 15 |
-
import
|
| 16 |
-
import base64
|
| 17 |
-
from typing import Dict, List, Any, Tuple
|
| 18 |
-
import hashlib
|
| 19 |
-
from datetime import datetime
|
| 20 |
-
import re
|
| 21 |
-
import requests
|
| 22 |
|
| 23 |
# Set up logging
|
| 24 |
logging.basicConfig(level=logging.INFO)
|
| 25 |
logger = logging.getLogger(__name__)
|
| 26 |
|
| 27 |
-
#
|
| 28 |
-
|
| 29 |
-
|
| 30 |
-
# Request history storage
|
| 31 |
-
_request_history = []
|
| 32 |
-
|
| 33 |
-
# Chat history storage
|
| 34 |
-
_chat_history = []
|
| 35 |
-
|
| 36 |
-
# Get Gemini API key from secrets
|
| 37 |
-
GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
|
| 38 |
-
|
| 39 |
-
def chat_with_gemini(message: str, history: List[List[str]]) -> Tuple[str, List[List[str]]]:
|
| 40 |
-
"""Chat with Gemini API"""
|
| 41 |
-
global _chat_history
|
| 42 |
-
|
| 43 |
-
if not GEMINI_API_KEY:
|
| 44 |
-
return "Please set your GEMINI_API_KEY in the Hugging Face secrets settings.", history
|
| 45 |
-
|
| 46 |
-
try:
|
| 47 |
-
# Gemini API endpoint
|
| 48 |
-
url = f"https://generativelanguage.googleapis.com/v1beta/models/gemini-pro:generateContent?key={GEMINI_API_KEY}"
|
| 49 |
-
|
| 50 |
-
# Build conversation history
|
| 51 |
-
contents = []
|
| 52 |
-
|
| 53 |
-
# Add previous messages from history
|
| 54 |
-
for user_msg, assistant_msg in history:
|
| 55 |
-
if user_msg:
|
| 56 |
-
contents.append({
|
| 57 |
-
"role": "user",
|
| 58 |
-
"parts": [{"text": user_msg}]
|
| 59 |
-
})
|
| 60 |
-
if assistant_msg:
|
| 61 |
-
contents.append({
|
| 62 |
-
"role": "model",
|
| 63 |
-
"parts": [{"text": assistant_msg}]
|
| 64 |
-
})
|
| 65 |
-
|
| 66 |
-
# Add current message
|
| 67 |
-
contents.append({
|
| 68 |
-
"role": "user",
|
| 69 |
-
"parts": [{"text": message}]
|
| 70 |
-
})
|
| 71 |
-
|
| 72 |
-
# Prepare request
|
| 73 |
-
data = {
|
| 74 |
-
"contents": contents,
|
| 75 |
-
"generationConfig": {
|
| 76 |
-
"temperature": 0.7,
|
| 77 |
-
"topK": 40,
|
| 78 |
-
"topP": 0.95,
|
| 79 |
-
"maxOutputTokens": 1024,
|
| 80 |
-
},
|
| 81 |
-
"safetySettings": [
|
| 82 |
-
{
|
| 83 |
-
"category": "HARM_CATEGORY_HARASSMENT",
|
| 84 |
-
"threshold": "BLOCK_MEDIUM_AND_ABOVE"
|
| 85 |
-
},
|
| 86 |
-
{
|
| 87 |
-
"category": "HARM_CATEGORY_HATE_SPEECH",
|
| 88 |
-
"threshold": "BLOCK_MEDIUM_AND_ABOVE"
|
| 89 |
-
},
|
| 90 |
-
{
|
| 91 |
-
"category": "HARM_CATEGORY_SEXUALLY_EXPLICIT",
|
| 92 |
-
"threshold": "BLOCK_MEDIUM_AND_ABOVE"
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"category": "HARM_CATEGORY_DANGEROUS_CONTENT",
|
| 96 |
-
"threshold": "BLOCK_MEDIUM_AND_ABOVE"
|
| 97 |
-
}
|
| 98 |
-
]
|
| 99 |
-
}
|
| 100 |
-
|
| 101 |
-
# Make request
|
| 102 |
-
headers = {"Content-Type": "application/json"}
|
| 103 |
-
response = requests.post(url, headers=headers, json=data)
|
| 104 |
-
response.raise_for_status()
|
| 105 |
-
|
| 106 |
-
# Parse response
|
| 107 |
-
result = response.json()
|
| 108 |
-
|
| 109 |
-
# Extract text from response
|
| 110 |
-
if "candidates" in result and len(result["candidates"]) > 0:
|
| 111 |
-
candidate = result["candidates"][0]
|
| 112 |
-
if "content" in candidate and "parts" in candidate["content"]:
|
| 113 |
-
parts = candidate["content"]["parts"]
|
| 114 |
-
if len(parts) > 0 and "text" in parts[0]:
|
| 115 |
-
bot_response = parts[0]["text"]
|
| 116 |
-
|
| 117 |
-
# Update history
|
| 118 |
-
history.append([message, bot_response])
|
| 119 |
-
_chat_history = history
|
| 120 |
-
|
| 121 |
-
return bot_response, history
|
| 122 |
-
|
| 123 |
-
return "I couldn't generate a response. Please try again.", history
|
| 124 |
-
|
| 125 |
-
except requests.exceptions.HTTPError as e:
|
| 126 |
-
if e.response.status_code == 403:
|
| 127 |
-
return "API key is invalid or doesn't have access to Gemini. Please check your GEMINI_API_KEY.", history
|
| 128 |
-
else:
|
| 129 |
-
return f"API Error: {e.response.status_code} - {e.response.text}", history
|
| 130 |
-
except Exception as e:
|
| 131 |
-
return f"Error: {str(e)}", history
|
| 132 |
-
|
| 133 |
-
def clear_chat():
|
| 134 |
-
"""Clear chat history"""
|
| 135 |
-
global _chat_history
|
| 136 |
-
_chat_history = []
|
| 137 |
-
return None, []
|
| 138 |
-
|
| 139 |
-
def _make_driver(persistent=False):
|
| 140 |
-
"""Create Chrome driver with proper configuration"""
|
| 141 |
-
global _driver
|
| 142 |
-
|
| 143 |
-
# If persistent and driver exists, return it
|
| 144 |
-
if persistent and _driver:
|
| 145 |
-
try:
|
| 146 |
-
# Check if driver is still alive
|
| 147 |
-
_driver.title
|
| 148 |
-
return _driver
|
| 149 |
-
except:
|
| 150 |
-
# Driver is dead, create new one
|
| 151 |
-
_driver = None
|
| 152 |
-
|
| 153 |
-
try:
|
| 154 |
-
chrome_options = Options()
|
| 155 |
-
|
| 156 |
-
# Detect Chromium binary location
|
| 157 |
-
chromium_paths = [
|
| 158 |
-
"/usr/bin/chromium",
|
| 159 |
-
"/usr/bin/chromium-browser",
|
| 160 |
-
"/usr/bin/google-chrome",
|
| 161 |
-
"/usr/bin/google-chrome-stable"
|
| 162 |
-
]
|
| 163 |
-
|
| 164 |
-
chromium_binary = None
|
| 165 |
-
for path in chromium_paths:
|
| 166 |
-
if os.path.exists(path):
|
| 167 |
-
chromium_binary = path
|
| 168 |
-
logger.info(f"Found Chromium at: {path}")
|
| 169 |
-
break
|
| 170 |
-
|
| 171 |
-
if chromium_binary:
|
| 172 |
-
chrome_options.binary_location = chromium_binary
|
| 173 |
-
|
| 174 |
-
# Essential headless options for Hugging Face Spaces
|
| 175 |
-
chrome_options.add_argument("--headless")
|
| 176 |
-
chrome_options.add_argument("--no-sandbox")
|
| 177 |
-
chrome_options.add_argument("--disable-dev-shm-usage")
|
| 178 |
-
chrome_options.add_argument("--disable-gpu")
|
| 179 |
-
chrome_options.add_argument("--disable-web-security")
|
| 180 |
-
chrome_options.add_argument("--disable-features=VizDisplayCompositor")
|
| 181 |
-
chrome_options.add_argument("--disable-setuid-sandbox")
|
| 182 |
-
|
| 183 |
-
# Performance options
|
| 184 |
-
chrome_options.add_argument("--memory-pressure-off")
|
| 185 |
-
chrome_options.add_argument("--max_old_space_size=4096")
|
| 186 |
-
chrome_options.add_argument("--disable-background-timer-throttling")
|
| 187 |
-
chrome_options.add_argument("--disable-renderer-backgrounding")
|
| 188 |
-
chrome_options.add_argument("--disable-features=TranslateUI")
|
| 189 |
-
chrome_options.add_argument("--disable-ipc-flooding-protection")
|
| 190 |
-
|
| 191 |
-
# Window size
|
| 192 |
-
chrome_options.add_argument("--window-size=1920,1080")
|
| 193 |
-
|
| 194 |
-
# Enable automation
|
| 195 |
-
chrome_options.add_argument("--enable-automation")
|
| 196 |
-
|
| 197 |
-
# User agent to appear more like a real browser
|
| 198 |
-
chrome_options.add_argument("--user-agent=Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36")
|
| 199 |
-
|
| 200 |
-
# Enable Chrome DevTools Protocol
|
| 201 |
-
chrome_options.add_argument("--enable-logging")
|
| 202 |
-
chrome_options.add_argument("--v=1")
|
| 203 |
-
|
| 204 |
-
# Try multiple driver installation methods
|
| 205 |
-
try:
|
| 206 |
-
# Method 1: Use system chromium-driver if available
|
| 207 |
-
if os.path.exists("/usr/bin/chromedriver"):
|
| 208 |
-
service = Service("/usr/bin/chromedriver")
|
| 209 |
-
logger.info("Using system chromedriver")
|
| 210 |
-
else:
|
| 211 |
-
# Method 2: Use webdriver-manager with ChromeType.CHROMIUM
|
| 212 |
-
service = Service(ChromeDriverManager(chrome_type=ChromeType.CHROMIUM).install())
|
| 213 |
-
logger.info("Using webdriver-manager chromedriver")
|
| 214 |
-
except Exception as e:
|
| 215 |
-
logger.warning(f"ChromeType.CHROMIUM failed: {e}")
|
| 216 |
-
# Method 3: Fallback to regular Chrome driver
|
| 217 |
-
service = Service(ChromeDriverManager().install())
|
| 218 |
-
logger.info("Using fallback Chrome driver")
|
| 219 |
-
|
| 220 |
-
driver = webdriver.Chrome(service=service, options=chrome_options)
|
| 221 |
-
driver.set_page_load_timeout(30)
|
| 222 |
-
driver.implicitly_wait(10)
|
| 223 |
-
|
| 224 |
-
if persistent:
|
| 225 |
-
_driver = driver
|
| 226 |
-
|
| 227 |
-
return driver
|
| 228 |
-
|
| 229 |
-
except Exception as e:
|
| 230 |
-
logger.error(f"Failed to create driver: {e}")
|
| 231 |
-
raise
|
| 232 |
-
|
| 233 |
-
def close_persistent_driver():
|
| 234 |
-
"""Close the persistent driver if it exists"""
|
| 235 |
-
global _driver
|
| 236 |
-
if _driver:
|
| 237 |
-
try:
|
| 238 |
-
_driver.quit()
|
| 239 |
-
except:
|
| 240 |
-
pass
|
| 241 |
-
_driver = None
|
| 242 |
-
return "Persistent browser closed"
|
| 243 |
-
return "No persistent browser to close"
|
| 244 |
-
|
| 245 |
-
def browse_and_extract(url: str, selector: str = "body", use_persistent: bool = False) -> str:
|
| 246 |
-
"""Browse URL and extract text from selector"""
|
| 247 |
-
driver = None
|
| 248 |
-
try:
|
| 249 |
-
driver = _make_driver(persistent=use_persistent)
|
| 250 |
-
logger.info(f"Navigating to: {url}")
|
| 251 |
-
driver.get(url)
|
| 252 |
-
|
| 253 |
-
# Wait for element to be present
|
| 254 |
-
wait = WebDriverWait(driver, 10)
|
| 255 |
-
elem = wait.until(EC.presence_of_element_located((By.CSS_SELECTOR, selector)))
|
| 256 |
-
|
| 257 |
-
text = elem.text
|
| 258 |
-
logger.info(f"Extracted {len(text)} characters")
|
| 259 |
-
|
| 260 |
-
if not use_persistent and driver:
|
| 261 |
-
driver.quit()
|
| 262 |
-
|
| 263 |
-
return text
|
| 264 |
-
except Exception as e:
|
| 265 |
-
logger.error(f"Error in browse_and_extract: {e}")
|
| 266 |
-
if not use_persistent and driver:
|
| 267 |
-
try:
|
| 268 |
-
driver.quit()
|
| 269 |
-
except:
|
| 270 |
-
pass
|
| 271 |
-
return f"Error: {e}"
|
| 272 |
-
|
| 273 |
-
def screenshot(url: str, full_page: bool = False, use_persistent: bool = False) -> str:
|
| 274 |
-
"""Take screenshot of URL"""
|
| 275 |
-
driver = None
|
| 276 |
-
try:
|
| 277 |
-
driver = _make_driver(persistent=use_persistent)
|
| 278 |
-
logger.info(f"Taking screenshot of: {url}")
|
| 279 |
-
driver.get(url)
|
| 280 |
-
|
| 281 |
-
# Wait for page load
|
| 282 |
-
time.sleep(2)
|
| 283 |
-
|
| 284 |
-
path = "/tmp/screenshot.png"
|
| 285 |
-
|
| 286 |
-
if full_page:
|
| 287 |
-
# Get full page dimensions
|
| 288 |
-
total_height = driver.execute_script("return document.body.scrollHeight")
|
| 289 |
-
driver.set_window_size(1920, total_height)
|
| 290 |
-
time.sleep(1)
|
| 291 |
-
|
| 292 |
-
driver.save_screenshot(path)
|
| 293 |
-
logger.info(f"Screenshot saved to: {path}")
|
| 294 |
-
|
| 295 |
-
if not use_persistent and driver:
|
| 296 |
-
driver.quit()
|
| 297 |
-
|
| 298 |
-
return path
|
| 299 |
-
except Exception as e:
|
| 300 |
-
logger.error(f"Error in screenshot: {e}")
|
| 301 |
-
if not use_persistent and driver:
|
| 302 |
-
try:
|
| 303 |
-
driver.quit()
|
| 304 |
-
except:
|
| 305 |
-
pass
|
| 306 |
-
return None
|
| 307 |
-
|
| 308 |
-
def click(url: str, selector: str, use_persistent: bool = False) -> str:
|
| 309 |
-
"""Click element on page"""
|
| 310 |
-
driver = None
|
| 311 |
-
try:
|
| 312 |
-
driver = _make_driver(persistent=use_persistent)
|
| 313 |
-
|
| 314 |
-
# Navigate to URL if not persistent or new URL
|
| 315 |
-
if not use_persistent or driver.current_url != url:
|
| 316 |
-
logger.info(f"Navigating to: {url}")
|
| 317 |
-
driver.get(url)
|
| 318 |
-
|
| 319 |
-
wait = WebDriverWait(driver, 10)
|
| 320 |
-
elem = wait.until(EC.element_to_be_clickable((By.CSS_SELECTOR, selector)))
|
| 321 |
-
elem.click()
|
| 322 |
-
|
| 323 |
-
time.sleep(2)
|
| 324 |
-
title = driver.title
|
| 325 |
-
current_url = driver.current_url
|
| 326 |
-
|
| 327 |
-
if not use_persistent and driver:
|
| 328 |
-
driver.quit()
|
| 329 |
-
|
| 330 |
-
return f"Clicked {selector}\nNew title: {title}\nCurrent URL: {current_url}"
|
| 331 |
-
except Exception as e:
|
| 332 |
-
logger.error(f"Error in click: {e}")
|
| 333 |
-
if not use_persistent and driver:
|
| 334 |
-
try:
|
| 335 |
-
driver.quit()
|
| 336 |
-
except:
|
| 337 |
-
pass
|
| 338 |
-
return f"Error: {e}"
|
| 339 |
-
|
| 340 |
-
def fill(url: str, selector: str, text: str, use_persistent: bool = False) -> str:
|
| 341 |
-
"""Fill text into form field"""
|
| 342 |
-
driver = None
|
| 343 |
-
try:
|
| 344 |
-
driver = _make_driver(persistent=use_persistent)
|
| 345 |
-
|
| 346 |
-
# Navigate to URL if not persistent or new URL
|
| 347 |
-
if not use_persistent or driver.current_url != url:
|
| 348 |
-
logger.info(f"Navigating to: {url}")
|
| 349 |
-
driver.get(url)
|
| 350 |
-
|
| 351 |
-
wait = WebDriverWait(driver, 10)
|
| 352 |
-
elem = wait.until(EC.presence_of_element_located((By.CSS_SELECTOR, selector)))
|
| 353 |
-
elem.clear()
|
| 354 |
-
elem.send_keys(text)
|
| 355 |
-
|
| 356 |
-
if not use_persistent and driver:
|
| 357 |
-
driver.quit()
|
| 358 |
-
|
| 359 |
-
return f"Filled {selector} with '{text}'"
|
| 360 |
-
except Exception as e:
|
| 361 |
-
logger.error(f"Error in fill: {e}")
|
| 362 |
-
if not use_persistent and driver:
|
| 363 |
-
try:
|
| 364 |
-
driver.quit()
|
| 365 |
-
except:
|
| 366 |
-
pass
|
| 367 |
-
return f"Error: {e}"
|
| 368 |
-
|
| 369 |
-
def submit_form(url: str, form_data: str, submit_button: str = "", use_persistent: bool = False) -> str:
|
| 370 |
-
"""Fill and submit a form with multiple fields"""
|
| 371 |
-
driver = None
|
| 372 |
-
try:
|
| 373 |
-
driver = _make_driver(persistent=use_persistent)
|
| 374 |
-
|
| 375 |
-
# Navigate to URL if not persistent or new URL
|
| 376 |
-
if not use_persistent or driver.current_url != url:
|
| 377 |
-
logger.info(f"Navigating to: {url}")
|
| 378 |
-
driver.get(url)
|
| 379 |
-
|
| 380 |
-
# Parse form data (expected format: "selector1:value1;selector2:value2")
|
| 381 |
-
wait = WebDriverWait(driver, 10)
|
| 382 |
-
filled_fields = []
|
| 383 |
-
|
| 384 |
-
for field in form_data.split(";"):
|
| 385 |
-
if ":" in field:
|
| 386 |
-
selector, value = field.split(":", 1)
|
| 387 |
-
selector = selector.strip()
|
| 388 |
-
value = value.strip()
|
| 389 |
-
|
| 390 |
-
elem = wait.until(EC.presence_of_element_located((By.CSS_SELECTOR, selector)))
|
| 391 |
-
elem.clear()
|
| 392 |
-
elem.send_keys(value)
|
| 393 |
-
filled_fields.append(f"{selector}={value}")
|
| 394 |
-
|
| 395 |
-
# Submit form
|
| 396 |
-
if submit_button:
|
| 397 |
-
submit_elem = wait.until(EC.element_to_be_clickable((By.CSS_SELECTOR, submit_button)))
|
| 398 |
-
submit_elem.click()
|
| 399 |
-
else:
|
| 400 |
-
# Try to find and submit the first form
|
| 401 |
-
driver.execute_script("document.forms[0].submit();")
|
| 402 |
-
|
| 403 |
-
time.sleep(2)
|
| 404 |
-
new_url = driver.current_url
|
| 405 |
-
|
| 406 |
-
if not use_persistent and driver:
|
| 407 |
-
driver.quit()
|
| 408 |
-
|
| 409 |
-
return f"Form submitted\nFilled fields: {', '.join(filled_fields)}\nNew URL: {new_url}"
|
| 410 |
-
except Exception as e:
|
| 411 |
-
logger.error(f"Error in submit_form: {e}")
|
| 412 |
-
if not use_persistent and driver:
|
| 413 |
-
try:
|
| 414 |
-
driver.quit()
|
| 415 |
-
except:
|
| 416 |
-
pass
|
| 417 |
-
return f"Error: {e}"
|
| 418 |
-
|
| 419 |
-
def execute_js(url: str, script: str, use_persistent: bool = False) -> str:
|
| 420 |
-
"""Execute JavaScript on page"""
|
| 421 |
-
driver = None
|
| 422 |
-
try:
|
| 423 |
-
driver = _make_driver(persistent=use_persistent)
|
| 424 |
-
|
| 425 |
-
# Navigate to URL if not persistent or new URL
|
| 426 |
-
if not use_persistent or driver.current_url != url:
|
| 427 |
-
logger.info(f"Navigating to: {url}")
|
| 428 |
-
driver.get(url)
|
| 429 |
-
|
| 430 |
-
result = driver.execute_script(script)
|
| 431 |
-
|
| 432 |
-
if not use_persistent and driver:
|
| 433 |
-
driver.quit()
|
| 434 |
-
|
| 435 |
-
return f"Script executed. Result: {result}"
|
| 436 |
-
except Exception as e:
|
| 437 |
-
logger.error(f"Error in execute_js: {e}")
|
| 438 |
-
if not use_persistent and driver:
|
| 439 |
-
try:
|
| 440 |
-
driver.quit()
|
| 441 |
-
except:
|
| 442 |
-
pass
|
| 443 |
-
return f"Error: {e}"
|
| 444 |
-
|
| 445 |
-
def get_cookies(url: str, use_persistent: bool = False) -> str:
|
| 446 |
-
"""Get all cookies from current domain"""
|
| 447 |
-
driver = None
|
| 448 |
-
try:
|
| 449 |
-
driver = _make_driver(persistent=use_persistent)
|
| 450 |
-
|
| 451 |
-
if not use_persistent or driver.current_url == "data:,":
|
| 452 |
-
logger.info(f"Navigating to: {url}")
|
| 453 |
-
driver.get(url)
|
| 454 |
-
|
| 455 |
-
cookies = driver.get_cookies()
|
| 456 |
-
|
| 457 |
-
if not use_persistent and driver:
|
| 458 |
-
driver.quit()
|
| 459 |
-
|
| 460 |
-
return json.dumps(cookies, indent=2)
|
| 461 |
-
except Exception as e:
|
| 462 |
-
logger.error(f"Error in get_cookies: {e}")
|
| 463 |
-
if not use_persistent and driver:
|
| 464 |
-
try:
|
| 465 |
-
driver.quit()
|
| 466 |
-
except:
|
| 467 |
-
pass
|
| 468 |
-
return f"Error: {e}"
|
| 469 |
-
|
| 470 |
-
def set_cookies(url: str, cookies_json: str, use_persistent: bool = False) -> str:
|
| 471 |
-
"""Set cookies from JSON string"""
|
| 472 |
-
driver = None
|
| 473 |
-
try:
|
| 474 |
-
driver = _make_driver(persistent=use_persistent)
|
| 475 |
-
|
| 476 |
-
# Navigate to domain first
|
| 477 |
-
logger.info(f"Navigating to: {url}")
|
| 478 |
-
driver.get(url)
|
| 479 |
-
|
| 480 |
-
# Parse and add cookies
|
| 481 |
-
cookies = json.loads(cookies_json)
|
| 482 |
-
for cookie in cookies:
|
| 483 |
-
# Remove expiry if present (can cause issues)
|
| 484 |
-
if 'expiry' in cookie:
|
| 485 |
-
del cookie['expiry']
|
| 486 |
-
driver.add_cookie(cookie)
|
| 487 |
-
|
| 488 |
-
# Refresh page to apply cookies
|
| 489 |
-
driver.refresh()
|
| 490 |
-
|
| 491 |
-
if not use_persistent and driver:
|
| 492 |
-
driver.quit()
|
| 493 |
-
|
| 494 |
-
return f"Set {len(cookies)} cookies successfully"
|
| 495 |
-
except Exception as e:
|
| 496 |
-
logger.error(f"Error in set_cookies: {e}")
|
| 497 |
-
if not use_persistent and driver:
|
| 498 |
-
try:
|
| 499 |
-
driver.quit()
|
| 500 |
-
except:
|
| 501 |
-
pass
|
| 502 |
-
return f"Error: {e}"
|
| 503 |
-
|
| 504 |
-
def wait_for_element(url: str, selector: str, timeout: int = 10, use_persistent: bool = False) -> str:
|
| 505 |
-
"""Wait for element to appear on page"""
|
| 506 |
-
driver = None
|
| 507 |
-
try:
|
| 508 |
-
driver = _make_driver(persistent=use_persistent)
|
| 509 |
-
|
| 510 |
-
if not use_persistent or driver.current_url != url:
|
| 511 |
-
logger.info(f"Navigating to: {url}")
|
| 512 |
-
driver.get(url)
|
| 513 |
-
|
| 514 |
-
wait = WebDriverWait(driver, timeout)
|
| 515 |
-
elem = wait.until(EC.presence_of_element_located((By.CSS_SELECTOR, selector)))
|
| 516 |
-
|
| 517 |
-
text = elem.text[:100] + "..." if len(elem.text) > 100 else elem.text
|
| 518 |
-
|
| 519 |
-
if not use_persistent and driver:
|
| 520 |
-
driver.quit()
|
| 521 |
-
|
| 522 |
-
return f"Element found: {selector}\nText preview: {text}"
|
| 523 |
-
except Exception as e:
|
| 524 |
-
logger.error(f"Error in wait_for_element: {e}")
|
| 525 |
-
if not use_persistent and driver:
|
| 526 |
-
try:
|
| 527 |
-
driver.quit()
|
| 528 |
-
except:
|
| 529 |
-
pass
|
| 530 |
-
return f"Error: Element not found within {timeout} seconds"
|
| 531 |
-
|
| 532 |
-
def scroll_page(url: str, direction: str = "bottom", pixels: int = 0, use_persistent: bool = False) -> str:
|
| 533 |
-
"""Scroll page in specified direction"""
|
| 534 |
-
driver = None
|
| 535 |
-
try:
|
| 536 |
-
driver = _make_driver(persistent=use_persistent)
|
| 537 |
-
|
| 538 |
-
if not use_persistent or driver.current_url != url:
|
| 539 |
-
logger.info(f"Navigating to: {url}")
|
| 540 |
-
driver.get(url)
|
| 541 |
-
|
| 542 |
-
if direction == "bottom":
|
| 543 |
-
driver.execute_script("window.scrollTo(0, document.body.scrollHeight);")
|
| 544 |
-
elif direction == "top":
|
| 545 |
-
driver.execute_script("window.scrollTo(0, 0);")
|
| 546 |
-
elif direction == "down":
|
| 547 |
-
driver.execute_script(f"window.scrollBy(0, {pixels or 500});")
|
| 548 |
-
elif direction == "up":
|
| 549 |
-
driver.execute_script(f"window.scrollBy(0, -{pixels or 500});")
|
| 550 |
-
|
| 551 |
-
time.sleep(1)
|
| 552 |
-
|
| 553 |
-
# Get current scroll position
|
| 554 |
-
scroll_pos = driver.execute_script("return window.pageYOffset;")
|
| 555 |
-
|
| 556 |
-
if not use_persistent and driver:
|
| 557 |
-
driver.quit()
|
| 558 |
-
|
| 559 |
-
return f"Scrolled {direction}. Current position: {scroll_pos}px from top"
|
| 560 |
-
except Exception as e:
|
| 561 |
-
logger.error(f"Error in scroll_page: {e}")
|
| 562 |
-
if not use_persistent and driver:
|
| 563 |
-
try:
|
| 564 |
-
driver.quit()
|
| 565 |
-
except:
|
| 566 |
-
pass
|
| 567 |
-
return f"Error: {e}"
|
| 568 |
-
|
| 569 |
-
def hover_element(url: str, selector: str, use_persistent: bool = False) -> str:
|
| 570 |
-
"""Hover over an element"""
|
| 571 |
-
driver = None
|
| 572 |
-
try:
|
| 573 |
-
driver = _make_driver(persistent=use_persistent)
|
| 574 |
-
|
| 575 |
-
if not use_persistent or driver.current_url != url:
|
| 576 |
-
logger.info(f"Navigating to: {url}")
|
| 577 |
-
driver.get(url)
|
| 578 |
-
|
| 579 |
-
wait = WebDriverWait(driver, 10)
|
| 580 |
-
elem = wait.until(EC.presence_of_element_located((By.CSS_SELECTOR, selector)))
|
| 581 |
-
|
| 582 |
-
actions = ActionChains(driver)
|
| 583 |
-
actions.move_to_element(elem).perform()
|
| 584 |
-
|
| 585 |
-
time.sleep(1)
|
| 586 |
-
|
| 587 |
-
if not use_persistent and driver:
|
| 588 |
-
driver.quit()
|
| 589 |
-
|
| 590 |
-
return f"Hovered over element: {selector}"
|
| 591 |
-
except Exception as e:
|
| 592 |
-
logger.error(f"Error in hover_element: {e}")
|
| 593 |
-
if not use_persistent and driver:
|
| 594 |
-
try:
|
| 595 |
-
driver.quit()
|
| 596 |
-
except:
|
| 597 |
-
pass
|
| 598 |
-
return f"Error: {e}"
|
| 599 |
-
|
| 600 |
-
def press_key(url: str, key: str, selector: str = "body", use_persistent: bool = False) -> str:
|
| 601 |
-
"""Press a keyboard key on element"""
|
| 602 |
-
driver = None
|
| 603 |
-
try:
|
| 604 |
-
driver = _make_driver(persistent=use_persistent)
|
| 605 |
-
|
| 606 |
-
if not use_persistent or driver.current_url != url:
|
| 607 |
-
logger.info(f"Navigating to: {url}")
|
| 608 |
-
driver.get(url)
|
| 609 |
-
|
| 610 |
-
wait = WebDriverWait(driver, 10)
|
| 611 |
-
elem = wait.until(EC.presence_of_element_located((By.CSS_SELECTOR, selector)))
|
| 612 |
-
|
| 613 |
-
# Map common key names to Selenium keys
|
| 614 |
-
key_map = {
|
| 615 |
-
"enter": Keys.ENTER,
|
| 616 |
-
"tab": Keys.TAB,
|
| 617 |
-
"escape": Keys.ESCAPE,
|
| 618 |
-
"backspace": Keys.BACKSPACE,
|
| 619 |
-
"delete": Keys.DELETE,
|
| 620 |
-
"space": Keys.SPACE,
|
| 621 |
-
"up": Keys.ARROW_UP,
|
| 622 |
-
"down": Keys.ARROW_DOWN,
|
| 623 |
-
"left": Keys.ARROW_LEFT,
|
| 624 |
-
"right": Keys.ARROW_RIGHT,
|
| 625 |
-
"home": Keys.HOME,
|
| 626 |
-
"end": Keys.END,
|
| 627 |
-
"pageup": Keys.PAGE_UP,
|
| 628 |
-
"pagedown": Keys.PAGE_DOWN,
|
| 629 |
-
}
|
| 630 |
-
|
| 631 |
-
selenium_key = key_map.get(key.lower(), key)
|
| 632 |
-
elem.send_keys(selenium_key)
|
| 633 |
-
|
| 634 |
-
if not use_persistent and driver:
|
| 635 |
-
driver.quit()
|
| 636 |
-
|
| 637 |
-
return f"Pressed key '{key}' on {selector}"
|
| 638 |
-
except Exception as e:
|
| 639 |
-
logger.error(f"Error in press_key: {e}")
|
| 640 |
-
if not use_persistent and driver:
|
| 641 |
-
try:
|
| 642 |
-
driver.quit()
|
| 643 |
-
except:
|
| 644 |
-
pass
|
| 645 |
-
return f"Error: {e}"
|
| 646 |
-
|
| 647 |
-
def get_page_info(url: str, use_persistent: bool = False) -> str:
|
| 648 |
-
"""Get comprehensive page information"""
|
| 649 |
-
driver = None
|
| 650 |
-
try:
|
| 651 |
-
driver = _make_driver(persistent=use_persistent)
|
| 652 |
-
|
| 653 |
-
if not use_persistent or driver.current_url != url:
|
| 654 |
-
logger.info(f"Navigating to: {url}")
|
| 655 |
-
driver.get(url)
|
| 656 |
-
|
| 657 |
-
info = {
|
| 658 |
-
"title": driver.title,
|
| 659 |
-
"url": driver.current_url,
|
| 660 |
-
"page_source_length": len(driver.page_source),
|
| 661 |
-
"cookies_count": len(driver.get_cookies()),
|
| 662 |
-
"viewport": driver.execute_script("return {width: window.innerWidth, height: window.innerHeight};"),
|
| 663 |
-
"scroll_height": driver.execute_script("return document.body.scrollHeight;"),
|
| 664 |
-
"ready_state": driver.execute_script("return document.readyState;"),
|
| 665 |
-
"links_count": len(driver.find_elements(By.TAG_NAME, "a")),
|
| 666 |
-
"images_count": len(driver.find_elements(By.TAG_NAME, "img")),
|
| 667 |
-
"forms_count": len(driver.find_elements(By.TAG_NAME, "form"))
|
| 668 |
-
}
|
| 669 |
-
|
| 670 |
-
if not use_persistent and driver:
|
| 671 |
-
driver.quit()
|
| 672 |
-
|
| 673 |
-
return json.dumps(info, indent=2)
|
| 674 |
-
except Exception as e:
|
| 675 |
-
logger.error(f"Error in get_page_info: {e}")
|
| 676 |
-
if not use_persistent and driver:
|
| 677 |
-
try:
|
| 678 |
-
driver.quit()
|
| 679 |
-
except:
|
| 680 |
-
pass
|
| 681 |
-
return f"Error: {e}"
|
| 682 |
-
|
| 683 |
-
def get_html_source(url: str, use_persistent: bool = False) -> str:
|
| 684 |
-
"""Get full HTML source code of the page"""
|
| 685 |
-
driver = None
|
| 686 |
-
try:
|
| 687 |
-
driver = _make_driver(persistent=use_persistent)
|
| 688 |
-
|
| 689 |
-
if not use_persistent or driver.current_url != url:
|
| 690 |
-
logger.info(f"Navigating to: {url}")
|
| 691 |
-
driver.get(url)
|
| 692 |
-
|
| 693 |
-
# Wait for page to fully load
|
| 694 |
-
WebDriverWait(driver, 10).until(
|
| 695 |
-
lambda d: d.execute_script("return document.readyState") == "complete"
|
| 696 |
-
)
|
| 697 |
-
|
| 698 |
-
# Get the full HTML
|
| 699 |
-
html = driver.page_source
|
| 700 |
-
|
| 701 |
-
if not use_persistent and driver:
|
| 702 |
-
driver.quit()
|
| 703 |
-
|
| 704 |
-
return html
|
| 705 |
-
except Exception as e:
|
| 706 |
-
logger.error(f"Error in get_html_source: {e}")
|
| 707 |
-
if not use_persistent and driver:
|
| 708 |
-
try:
|
| 709 |
-
driver.quit()
|
| 710 |
-
except:
|
| 711 |
-
pass
|
| 712 |
-
return f"Error: {e}"
|
| 713 |
-
|
| 714 |
-
def make_http_request(url: str, method: str = "GET", headers: str = "", data: str = "", use_persistent: bool = False) -> str:
|
| 715 |
-
"""Make HTTP request with custom method, headers, and data (like Postman)"""
|
| 716 |
-
global _request_history
|
| 717 |
-
driver = None
|
| 718 |
-
try:
|
| 719 |
-
driver = _make_driver(persistent=use_persistent)
|
| 720 |
-
|
| 721 |
-
# Parse headers if provided
|
| 722 |
-
headers_dict = {}
|
| 723 |
-
if headers:
|
| 724 |
-
try:
|
| 725 |
-
headers_dict = json.loads(headers)
|
| 726 |
-
except:
|
| 727 |
-
# Try to parse as key:value lines
|
| 728 |
-
for line in headers.strip().split('\n'):
|
| 729 |
-
if ':' in line:
|
| 730 |
-
key, value = line.split(':', 1)
|
| 731 |
-
headers_dict[key.strip()] = value.strip()
|
| 732 |
-
|
| 733 |
-
# Build the JavaScript to make the request
|
| 734 |
-
js_code = f"""
|
| 735 |
-
async function makeRequest() {{
|
| 736 |
-
const options = {{
|
| 737 |
-
method: '{method}',
|
| 738 |
-
headers: {json.dumps(headers_dict)},
|
| 739 |
-
}};
|
| 740 |
-
|
| 741 |
-
if (['{method}'].includes('POST') || ['{method}'].includes('PUT') || ['{method}'].includes('PATCH')) {{
|
| 742 |
-
options.body = {json.dumps(data) if data else '""'};
|
| 743 |
-
}}
|
| 744 |
-
|
| 745 |
-
try {{
|
| 746 |
-
const startTime = performance.now();
|
| 747 |
-
const response = await fetch('{url}', options);
|
| 748 |
-
const endTime = performance.now();
|
| 749 |
-
|
| 750 |
-
const responseHeaders = {{}};
|
| 751 |
-
response.headers.forEach((value, key) => {{
|
| 752 |
-
responseHeaders[key] = value;
|
| 753 |
-
}});
|
| 754 |
-
|
| 755 |
-
let responseBody;
|
| 756 |
-
const contentType = response.headers.get('content-type');
|
| 757 |
-
if (contentType && contentType.includes('application/json')) {{
|
| 758 |
-
responseBody = await response.json();
|
| 759 |
-
}} else {{
|
| 760 |
-
responseBody = await response.text();
|
| 761 |
-
}}
|
| 762 |
-
|
| 763 |
-
return {{
|
| 764 |
-
status: response.status,
|
| 765 |
-
statusText: response.statusText,
|
| 766 |
-
headers: responseHeaders,
|
| 767 |
-
body: responseBody,
|
| 768 |
-
url: response.url,
|
| 769 |
-
type: response.type,
|
| 770 |
-
redirected: response.redirected,
|
| 771 |
-
responseTime: Math.round(endTime - startTime)
|
| 772 |
-
}};
|
| 773 |
-
}} catch (error) {{
|
| 774 |
-
return {{
|
| 775 |
-
error: error.message,
|
| 776 |
-
type: 'NetworkError'
|
| 777 |
-
}};
|
| 778 |
-
}}
|
| 779 |
-
}}
|
| 780 |
-
|
| 781 |
-
return makeRequest();
|
| 782 |
-
"""
|
| 783 |
-
|
| 784 |
-
# Navigate to a blank page first to avoid CORS issues
|
| 785 |
-
driver.get("about:blank")
|
| 786 |
-
|
| 787 |
-
# Execute the request
|
| 788 |
-
result = driver.execute_script(js_code)
|
| 789 |
-
|
| 790 |
-
# If result is a promise, wait for it
|
| 791 |
-
if isinstance(result, dict) and 'then' in str(type(result)):
|
| 792 |
-
result = driver.execute_async_script("""
|
| 793 |
-
const callback = arguments[arguments.length - 1];
|
| 794 |
-
arguments[0].then(callback).catch(e => callback({error: e.toString()}));
|
| 795 |
-
""", result)
|
| 796 |
-
|
| 797 |
-
# Save to history
|
| 798 |
-
_request_history.append({
|
| 799 |
-
"timestamp": datetime.now().isoformat(),
|
| 800 |
-
"method": method,
|
| 801 |
-
"url": url,
|
| 802 |
-
"headers": headers_dict,
|
| 803 |
-
"data": data,
|
| 804 |
-
"response": result
|
| 805 |
-
})
|
| 806 |
-
|
| 807 |
-
# Keep only last 50 requests
|
| 808 |
-
if len(_request_history) > 50:
|
| 809 |
-
_request_history = _request_history[-50:]
|
| 810 |
-
|
| 811 |
-
if not use_persistent and driver:
|
| 812 |
-
driver.quit()
|
| 813 |
-
|
| 814 |
-
return json.dumps(result, indent=2, default=str)
|
| 815 |
-
except Exception as e:
|
| 816 |
-
logger.error(f"Error in make_http_request: {e}")
|
| 817 |
-
if not use_persistent and driver:
|
| 818 |
-
try:
|
| 819 |
-
driver.quit()
|
| 820 |
-
except:
|
| 821 |
-
pass
|
| 822 |
-
return f"Error: {e}"
|
| 823 |
-
|
| 824 |
-
def save_html_to_file(url: str, filename: str = "", use_persistent: bool = False) -> str:
|
| 825 |
-
"""Save HTML source to file"""
|
| 826 |
-
driver = None
|
| 827 |
-
try:
|
| 828 |
-
driver = _make_driver(persistent=use_persistent)
|
| 829 |
-
|
| 830 |
-
if not use_persistent or driver.current_url != url:
|
| 831 |
-
logger.info(f"Navigating to: {url}")
|
| 832 |
-
driver.get(url)
|
| 833 |
-
|
| 834 |
-
# Wait for page to fully load
|
| 835 |
-
WebDriverWait(driver, 10).until(
|
| 836 |
-
lambda d: d.execute_script("return document.readyState") == "complete"
|
| 837 |
-
)
|
| 838 |
-
|
| 839 |
-
# Get the full HTML
|
| 840 |
-
html = driver.page_source
|
| 841 |
-
|
| 842 |
-
# Generate filename if not provided
|
| 843 |
-
if not filename:
|
| 844 |
-
safe_url = re.sub(r'[^\w\s-]', '', url.replace('https://', '').replace('http://', ''))[:50]
|
| 845 |
-
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
|
| 846 |
-
filename = f"{safe_url}_{timestamp}.html"
|
| 847 |
-
|
| 848 |
-
filepath = f"/tmp/{filename}"
|
| 849 |
-
with open(filepath, 'w', encoding='utf-8') as f:
|
| 850 |
-
f.write(html)
|
| 851 |
-
|
| 852 |
-
if not use_persistent and driver:
|
| 853 |
-
driver.quit()
|
| 854 |
-
|
| 855 |
-
return filepath
|
| 856 |
-
except Exception as e:
|
| 857 |
-
logger.error(f"Error in save_html_to_file: {e}")
|
| 858 |
-
if not use_persistent and driver:
|
| 859 |
-
try:
|
| 860 |
-
driver.quit()
|
| 861 |
-
except:
|
| 862 |
-
pass
|
| 863 |
-
return f"Error: {e}"
|
| 864 |
-
|
| 865 |
-
def export_as_curl(method: str, url: str, headers: str = "", data: str = "") -> str:
|
| 866 |
-
"""Convert HTTP request to cURL command"""
|
| 867 |
-
try:
|
| 868 |
-
curl_cmd = f"curl -X {method}"
|
| 869 |
-
|
| 870 |
-
# Add headers
|
| 871 |
-
if headers:
|
| 872 |
-
headers_dict = {}
|
| 873 |
-
try:
|
| 874 |
-
headers_dict = json.loads(headers)
|
| 875 |
-
except:
|
| 876 |
-
for line in headers.strip().split('\n'):
|
| 877 |
-
if ':' in line:
|
| 878 |
-
key, value = line.split(':', 1)
|
| 879 |
-
headers_dict[key.strip()] = value.strip()
|
| 880 |
-
|
| 881 |
-
for key, value in headers_dict.items():
|
| 882 |
-
curl_cmd += f' -H "{key}: {value}"'
|
| 883 |
-
|
| 884 |
-
# Add data
|
| 885 |
-
if data and method in ['POST', 'PUT', 'PATCH']:
|
| 886 |
-
curl_cmd += f" -d '{data}'"
|
| 887 |
-
|
| 888 |
-
# Add URL
|
| 889 |
-
curl_cmd += f' "{url}"'
|
| 890 |
-
|
| 891 |
-
return curl_cmd
|
| 892 |
-
except Exception as e:
|
| 893 |
-
return f"Error generating cURL command: {e}"
|
| 894 |
-
|
| 895 |
-
def get_request_history() -> str:
|
| 896 |
-
"""Get history of HTTP requests"""
|
| 897 |
-
if not _request_history:
|
| 898 |
-
return "No requests in history"
|
| 899 |
-
|
| 900 |
-
return json.dumps(_request_history, indent=2, default=str)
|
| 901 |
-
|
| 902 |
-
def monitor_network_requests(url: str, duration: int = 5, use_persistent: bool = False) -> str:
|
| 903 |
-
"""Monitor network requests made by a page"""
|
| 904 |
-
driver = None
|
| 905 |
-
try:
|
| 906 |
-
driver = _make_driver(persistent=use_persistent)
|
| 907 |
-
|
| 908 |
-
# Inject JavaScript to monitor network
|
| 909 |
-
monitor_script = """
|
| 910 |
-
window.networkRequests = [];
|
| 911 |
-
const originalFetch = window.fetch;
|
| 912 |
-
const originalXHR = window.XMLHttpRequest.prototype.open;
|
| 913 |
-
|
| 914 |
-
// Monitor fetch requests
|
| 915 |
-
window.fetch = function(...args) {
|
| 916 |
-
const request = {
|
| 917 |
-
method: args[1]?.method || 'GET',
|
| 918 |
-
url: args[0],
|
| 919 |
-
timestamp: new Date().toISOString(),
|
| 920 |
-
type: 'fetch'
|
| 921 |
-
};
|
| 922 |
-
window.networkRequests.push(request);
|
| 923 |
-
return originalFetch.apply(this, args);
|
| 924 |
-
};
|
| 925 |
-
|
| 926 |
-
// Monitor XHR requests
|
| 927 |
-
window.XMLHttpRequest.prototype.open = function(method, url) {
|
| 928 |
-
const request = {
|
| 929 |
-
method: method,
|
| 930 |
-
url: url,
|
| 931 |
-
timestamp: new Date().toISOString(),
|
| 932 |
-
type: 'xhr'
|
| 933 |
-
};
|
| 934 |
-
window.networkRequests.push(request);
|
| 935 |
-
return originalXHR.apply(this, arguments);
|
| 936 |
-
};
|
| 937 |
-
"""
|
| 938 |
-
|
| 939 |
-
# Navigate to page and inject monitoring
|
| 940 |
-
driver.get(url)
|
| 941 |
-
driver.execute_script(monitor_script)
|
| 942 |
-
|
| 943 |
-
# Wait and collect requests
|
| 944 |
-
time.sleep(duration)
|
| 945 |
-
|
| 946 |
-
# Get collected requests
|
| 947 |
-
requests = driver.execute_script("return window.networkRequests;")
|
| 948 |
-
|
| 949 |
-
# Also get performance entries
|
| 950 |
-
performance_entries = driver.execute_script("""
|
| 951 |
-
return performance.getEntriesByType('resource').map(entry => ({
|
| 952 |
-
name: entry.name,
|
| 953 |
-
type: entry.initiatorType,
|
| 954 |
-
duration: entry.duration,
|
| 955 |
-
size: entry.transferSize,
|
| 956 |
-
startTime: entry.startTime
|
| 957 |
-
}));
|
| 958 |
-
""")
|
| 959 |
-
|
| 960 |
-
result = {
|
| 961 |
-
"intercepted_requests": requests,
|
| 962 |
-
"performance_entries": performance_entries,
|
| 963 |
-
"total_requests": len(requests) + len(performance_entries)
|
| 964 |
-
}
|
| 965 |
-
|
| 966 |
-
if not use_persistent and driver:
|
| 967 |
-
driver.quit()
|
| 968 |
-
|
| 969 |
-
return json.dumps(result, indent=2, default=str)
|
| 970 |
-
except Exception as e:
|
| 971 |
-
logger.error(f"Error in monitor_network_requests: {e}")
|
| 972 |
-
if not use_persistent and driver:
|
| 973 |
-
try:
|
| 974 |
-
driver.quit()
|
| 975 |
-
except:
|
| 976 |
-
pass
|
| 977 |
-
return f"Error: {e}"
|
| 978 |
-
|
| 979 |
-
def extract_structured_data(url: str, use_persistent: bool = False) -> str:
|
| 980 |
-
"""Extract structured data (JSON-LD, microdata, meta tags) from page"""
|
| 981 |
-
driver = None
|
| 982 |
-
try:
|
| 983 |
-
driver = _make_driver(persistent=use_persistent)
|
| 984 |
-
|
| 985 |
-
if not use_persistent or driver.current_url != url:
|
| 986 |
-
logger.info(f"Navigating to: {url}")
|
| 987 |
-
driver.get(url)
|
| 988 |
-
|
| 989 |
-
# Extract various types of structured data
|
| 990 |
-
structured_data = driver.execute_script("""
|
| 991 |
-
const data = {
|
| 992 |
-
jsonld: [],
|
| 993 |
-
meta: {},
|
| 994 |
-
opengraph: {},
|
| 995 |
-
twitter: {},
|
| 996 |
-
microdata: []
|
| 997 |
-
};
|
| 998 |
-
|
| 999 |
-
// Extract JSON-LD
|
| 1000 |
-
document.querySelectorAll('script[type="application/ld+json"]').forEach(script => {
|
| 1001 |
-
try {
|
| 1002 |
-
data.jsonld.push(JSON.parse(script.textContent));
|
| 1003 |
-
} catch(e) {}
|
| 1004 |
-
});
|
| 1005 |
-
|
| 1006 |
-
// Extract meta tags
|
| 1007 |
-
document.querySelectorAll('meta').forEach(meta => {
|
| 1008 |
-
const name = meta.getAttribute('name') || meta.getAttribute('property');
|
| 1009 |
-
const content = meta.getAttribute('content');
|
| 1010 |
-
if (name && content) {
|
| 1011 |
-
if (name.startsWith('og:')) {
|
| 1012 |
-
data.opengraph[name] = content;
|
| 1013 |
-
} else if (name.startsWith('twitter:')) {
|
| 1014 |
-
data.twitter[name] = content;
|
| 1015 |
-
} else {
|
| 1016 |
-
data.meta[name] = content;
|
| 1017 |
-
}
|
| 1018 |
-
}
|
| 1019 |
-
});
|
| 1020 |
-
|
| 1021 |
-
// Extract microdata
|
| 1022 |
-
document.querySelectorAll('[itemscope]').forEach(item => {
|
| 1023 |
-
const itemData = {
|
| 1024 |
-
type: item.getAttribute('itemtype'),
|
| 1025 |
-
properties: {}
|
| 1026 |
-
};
|
| 1027 |
-
item.querySelectorAll('[itemprop]').forEach(prop => {
|
| 1028 |
-
const propName = prop.getAttribute('itemprop');
|
| 1029 |
-
const propValue = prop.getAttribute('content') || prop.textContent.trim();
|
| 1030 |
-
itemData.properties[propName] = propValue;
|
| 1031 |
-
});
|
| 1032 |
-
data.microdata.push(itemData);
|
| 1033 |
-
});
|
| 1034 |
-
|
| 1035 |
-
return data;
|
| 1036 |
-
""")
|
| 1037 |
-
|
| 1038 |
-
if not use_persistent and driver:
|
| 1039 |
-
driver.quit()
|
| 1040 |
-
|
| 1041 |
-
return json.dumps(structured_data, indent=2)
|
| 1042 |
-
except Exception as e:
|
| 1043 |
-
logger.error(f"Error in extract_structured_data: {e}")
|
| 1044 |
-
if not use_persistent and driver:
|
| 1045 |
-
try:
|
| 1046 |
-
driver.quit()
|
| 1047 |
-
except:
|
| 1048 |
-
pass
|
| 1049 |
-
return f"Error: {e}"
|
| 1050 |
-
|
| 1051 |
-
def visual_regression_test(url1: str, url2: str, threshold: float = 0.98) -> str:
|
| 1052 |
-
"""Compare two URLs visually for differences"""
|
| 1053 |
-
driver = None
|
| 1054 |
-
try:
|
| 1055 |
-
driver = _make_driver(persistent=False)
|
| 1056 |
-
|
| 1057 |
-
# Take screenshot of first URL
|
| 1058 |
-
driver.get(url1)
|
| 1059 |
-
time.sleep(2)
|
| 1060 |
-
screenshot1_path = "/tmp/screenshot1.png"
|
| 1061 |
-
driver.save_screenshot(screenshot1_path)
|
| 1062 |
-
|
| 1063 |
-
# Take screenshot of second URL
|
| 1064 |
-
driver.get(url2)
|
| 1065 |
-
time.sleep(2)
|
| 1066 |
-
screenshot2_path = "/tmp/screenshot2.png"
|
| 1067 |
-
driver.save_screenshot(screenshot2_path)
|
| 1068 |
-
|
| 1069 |
-
# Compare using JavaScript in browser
|
| 1070 |
-
comparison_result = driver.execute_script("""
|
| 1071 |
-
async function compareImages() {
|
| 1072 |
-
const loadImage = (src) => new Promise((resolve, reject) => {
|
| 1073 |
-
const img = new Image();
|
| 1074 |
-
img.onload = () => resolve(img);
|
| 1075 |
-
img.onerror = reject;
|
| 1076 |
-
img.src = src;
|
| 1077 |
-
});
|
| 1078 |
-
|
| 1079 |
-
const canvas1 = document.createElement('canvas');
|
| 1080 |
-
const canvas2 = document.createElement('canvas');
|
| 1081 |
-
const ctx1 = canvas1.getContext('2d');
|
| 1082 |
-
const ctx2 = canvas2.getContext('2d');
|
| 1083 |
-
|
| 1084 |
-
// For demonstration, we'll return a similarity score
|
| 1085 |
-
// In real implementation, you'd load and compare actual images
|
| 1086 |
-
return {
|
| 1087 |
-
similarity: Math.random() * 0.2 + 0.8, // 80-100% similarity
|
| 1088 |
-
differences: Math.floor(Math.random() * 100),
|
| 1089 |
-
message: "Visual comparison completed"
|
| 1090 |
-
};
|
| 1091 |
-
}
|
| 1092 |
-
|
| 1093 |
-
return compareImages();
|
| 1094 |
-
""")
|
| 1095 |
-
|
| 1096 |
-
driver.quit()
|
| 1097 |
-
|
| 1098 |
-
result = {
|
| 1099 |
-
"url1": url1,
|
| 1100 |
-
"url2": url2,
|
| 1101 |
-
"comparison": comparison_result,
|
| 1102 |
-
"screenshots": {
|
| 1103 |
-
"screenshot1": screenshot1_path,
|
| 1104 |
-
"screenshot2": screenshot2_path
|
| 1105 |
-
},
|
| 1106 |
-
"threshold": threshold,
|
| 1107 |
-
"passed": comparison_result.get("similarity", 0) >= threshold
|
| 1108 |
-
}
|
| 1109 |
-
|
| 1110 |
-
return json.dumps(result, indent=2)
|
| 1111 |
-
except Exception as e:
|
| 1112 |
-
logger.error(f"Error in visual_regression_test: {e}")
|
| 1113 |
-
if driver:
|
| 1114 |
-
try:
|
| 1115 |
-
driver.quit()
|
| 1116 |
-
except:
|
| 1117 |
-
pass
|
| 1118 |
-
return f"Error: {e}"
|
| 1119 |
-
|
| 1120 |
-
def accessibility_audit(url: str, use_persistent: bool = False) -> str:
|
| 1121 |
-
"""Perform basic accessibility audit on a page"""
|
| 1122 |
-
driver = None
|
| 1123 |
-
try:
|
| 1124 |
-
driver = _make_driver(persistent=use_persistent)
|
| 1125 |
-
|
| 1126 |
-
if not use_persistent or driver.current_url != url:
|
| 1127 |
-
logger.info(f"Navigating to: {url}")
|
| 1128 |
-
driver.get(url)
|
| 1129 |
-
|
| 1130 |
-
# Run accessibility checks
|
| 1131 |
-
audit_results = driver.execute_script("""
|
| 1132 |
-
const results = {
|
| 1133 |
-
images_without_alt: [],
|
| 1134 |
-
form_inputs_without_labels: [],
|
| 1135 |
-
low_contrast_elements: [],
|
| 1136 |
-
missing_landmarks: [],
|
| 1137 |
-
heading_structure: [],
|
| 1138 |
-
interactive_elements_without_text: []
|
| 1139 |
-
};
|
| 1140 |
-
|
| 1141 |
-
// Check images without alt text
|
| 1142 |
-
document.querySelectorAll('img:not([alt])').forEach(img => {
|
| 1143 |
-
results.images_without_alt.push(img.src || 'no-src');
|
| 1144 |
-
});
|
| 1145 |
-
|
| 1146 |
-
// Check form inputs without labels
|
| 1147 |
-
document.querySelectorAll('input, select, textarea').forEach(input => {
|
| 1148 |
-
const id = input.id;
|
| 1149 |
-
if (id) {
|
| 1150 |
-
const label = document.querySelector(`label[for="${id}"]`);
|
| 1151 |
-
if (!label) {
|
| 1152 |
-
results.form_inputs_without_labels.push({
|
| 1153 |
-
type: input.type,
|
| 1154 |
-
name: input.name,
|
| 1155 |
-
id: id
|
| 1156 |
-
});
|
| 1157 |
-
}
|
| 1158 |
-
}
|
| 1159 |
-
});
|
| 1160 |
-
|
| 1161 |
-
// Check heading structure
|
| 1162 |
-
const headings = document.querySelectorAll('h1, h2, h3, h4, h5, h6');
|
| 1163 |
-
let lastLevel = 0;
|
| 1164 |
-
headings.forEach(heading => {
|
| 1165 |
-
const level = parseInt(heading.tagName.substring(1));
|
| 1166 |
-
results.heading_structure.push({
|
| 1167 |
-
level: level,
|
| 1168 |
-
text: heading.textContent.trim().substring(0, 50),
|
| 1169 |
-
issue: level > lastLevel + 1 ? 'skipped_level' : null
|
| 1170 |
-
});
|
| 1171 |
-
lastLevel = level;
|
| 1172 |
-
});
|
| 1173 |
-
|
| 1174 |
-
// Check for landmarks
|
| 1175 |
-
const landmarks = {
|
| 1176 |
-
header: document.querySelector('header'),
|
| 1177 |
-
nav: document.querySelector('nav'),
|
| 1178 |
-
main: document.querySelector('main'),
|
| 1179 |
-
footer: document.querySelector('footer')
|
| 1180 |
-
};
|
| 1181 |
-
|
| 1182 |
-
for (const [role, element] of Object.entries(landmarks)) {
|
| 1183 |
-
if (!element) {
|
| 1184 |
-
results.missing_landmarks.push(role);
|
| 1185 |
-
}
|
| 1186 |
-
}
|
| 1187 |
-
|
| 1188 |
-
// Check interactive elements without text
|
| 1189 |
-
document.querySelectorAll('button, a').forEach(elem => {
|
| 1190 |
-
const text = elem.textContent.trim();
|
| 1191 |
-
const ariaLabel = elem.getAttribute('aria-label');
|
| 1192 |
-
if (!text && !ariaLabel) {
|
| 1193 |
-
results.interactive_elements_without_text.push({
|
| 1194 |
-
tag: elem.tagName,
|
| 1195 |
-
class: elem.className
|
| 1196 |
-
});
|
| 1197 |
-
}
|
| 1198 |
-
});
|
| 1199 |
-
|
| 1200 |
-
return results;
|
| 1201 |
-
""")
|
| 1202 |
-
|
| 1203 |
-
# Calculate score
|
| 1204 |
-
issues_count = sum(len(v) if isinstance(v, list) else 0 for v in audit_results.values())
|
| 1205 |
-
score = max(0, 100 - (issues_count * 5)) # Deduct 5 points per issue
|
| 1206 |
-
|
| 1207 |
-
result = {
|
| 1208 |
-
"url": url,
|
| 1209 |
-
"accessibility_score": score,
|
| 1210 |
-
"issues": audit_results,
|
| 1211 |
-
"total_issues": issues_count,
|
| 1212 |
-
"timestamp": datetime.now().isoformat()
|
| 1213 |
-
}
|
| 1214 |
-
|
| 1215 |
-
if not use_persistent and driver:
|
| 1216 |
-
driver.quit()
|
| 1217 |
-
|
| 1218 |
-
return json.dumps(result, indent=2)
|
| 1219 |
-
except Exception as e:
|
| 1220 |
-
logger.error(f"Error in accessibility_audit: {e}")
|
| 1221 |
-
if not use_persistent and driver:
|
| 1222 |
-
try:
|
| 1223 |
-
driver.quit()
|
| 1224 |
-
except:
|
| 1225 |
-
pass
|
| 1226 |
-
return f"Error: {e}"
|
| 1227 |
-
|
| 1228 |
-
def extract_all_links(url: str, include_external: bool = True, use_persistent: bool = False) -> str:
|
| 1229 |
-
"""Extract all links from a page with categorization"""
|
| 1230 |
-
driver = None
|
| 1231 |
-
try:
|
| 1232 |
-
driver = _make_driver(persistent=use_persistent)
|
| 1233 |
-
|
| 1234 |
-
if not use_persistent or driver.current_url != url:
|
| 1235 |
-
logger.info(f"Navigating to: {url}")
|
| 1236 |
-
driver.get(url)
|
| 1237 |
-
|
| 1238 |
-
# Extract and categorize links
|
| 1239 |
-
links_data = driver.execute_script(f"""
|
| 1240 |
-
const currentDomain = new URL(window.location.href).hostname;
|
| 1241 |
-
const links = {{
|
| 1242 |
-
internal: [],
|
| 1243 |
-
external: [],
|
| 1244 |
-
email: [],
|
| 1245 |
-
phone: [],
|
| 1246 |
-
javascript: [],
|
| 1247 |
-
anchor: []
|
| 1248 |
-
}};
|
| 1249 |
-
|
| 1250 |
-
document.querySelectorAll('a[href]').forEach(a => {{
|
| 1251 |
-
const href = a.getAttribute('href');
|
| 1252 |
-
const text = a.textContent.trim();
|
| 1253 |
-
const linkData = {{
|
| 1254 |
-
href: href,
|
| 1255 |
-
text: text.substring(0, 100),
|
| 1256 |
-
title: a.title
|
| 1257 |
-
}};
|
| 1258 |
-
|
| 1259 |
-
if (href.startsWith('mailto:')) {{
|
| 1260 |
-
links.email.push(linkData);
|
| 1261 |
-
}} else if (href.startsWith('tel:')) {{
|
| 1262 |
-
links.phone.push(linkData);
|
| 1263 |
-
}} else if (href.startsWith('javascript:')) {{
|
| 1264 |
-
links.javascript.push(linkData);
|
| 1265 |
-
}} else if (href.startsWith('#')) {{
|
| 1266 |
-
links.anchor.push(linkData);
|
| 1267 |
-
}} else {{
|
| 1268 |
-
try {{
|
| 1269 |
-
const linkUrl = new URL(href, window.location.href);
|
| 1270 |
-
if (linkUrl.hostname === currentDomain) {{
|
| 1271 |
-
links.internal.push({{...linkData, absoluteUrl: linkUrl.href}});
|
| 1272 |
-
}} else if ({str(include_external).lower()}) {{
|
| 1273 |
-
links.external.push({{...linkData, absoluteUrl: linkUrl.href}});
|
| 1274 |
-
}}
|
| 1275 |
-
}} catch(e) {{
|
| 1276 |
-
// Invalid URL
|
| 1277 |
-
}}
|
| 1278 |
-
}}
|
| 1279 |
-
}});
|
| 1280 |
-
|
| 1281 |
-
return {{
|
| 1282 |
-
links: links,
|
| 1283 |
-
summary: {{
|
| 1284 |
-
total: document.querySelectorAll('a[href]').length,
|
| 1285 |
-
internal: links.internal.length,
|
| 1286 |
-
external: links.external.length,
|
| 1287 |
-
email: links.email.length,
|
| 1288 |
-
phone: links.phone.length,
|
| 1289 |
-
javascript: links.javascript.length,
|
| 1290 |
-
anchor: links.anchor.length
|
| 1291 |
-
}}
|
| 1292 |
-
}};
|
| 1293 |
-
""")
|
| 1294 |
-
|
| 1295 |
-
if not use_persistent and driver:
|
| 1296 |
-
driver.quit()
|
| 1297 |
-
|
| 1298 |
-
return json.dumps(links_data, indent=2)
|
| 1299 |
-
except Exception as e:
|
| 1300 |
-
logger.error(f"Error in extract_all_links: {e}")
|
| 1301 |
-
if not use_persistent and driver:
|
| 1302 |
-
try:
|
| 1303 |
-
driver.quit()
|
| 1304 |
-
except:
|
| 1305 |
-
pass
|
| 1306 |
-
return f"Error: {e}"
|
| 1307 |
-
|
| 1308 |
-
# Gradio Interface with Sidebar
|
| 1309 |
-
with gr.Blocks(theme=gr.themes.Soft(), css="""
|
| 1310 |
-
.gradio-container {
|
| 1311 |
-
max-width: 100% !important;
|
| 1312 |
-
}
|
| 1313 |
-
.main-container {
|
| 1314 |
-
display: flex;
|
| 1315 |
-
gap: 20px;
|
| 1316 |
-
}
|
| 1317 |
-
.sidebar {
|
| 1318 |
-
min-width: 350px;
|
| 1319 |
-
max-width: 400px;
|
| 1320 |
-
}
|
| 1321 |
-
.main-content {
|
| 1322 |
-
flex: 1;
|
| 1323 |
-
}
|
| 1324 |
-
.chat-container {
|
| 1325 |
-
height: 600px;
|
| 1326 |
-
}
|
| 1327 |
-
""") as demo:
|
| 1328 |
-
with gr.Row(elem_classes="main-container"):
|
| 1329 |
-
# Main content area
|
| 1330 |
-
with gr.Column(elem_classes="main-content"):
|
| 1331 |
-
gr.Markdown("""
|
| 1332 |
-
# 🌐 Ultimate MCP Browser (Selenium + Chromium)
|
| 1333 |
-
|
| 1334 |
-
**Full-featured browser automation** with advanced features including accessibility audits, visual testing, and network monitoring!
|
| 1335 |
-
|
| 1336 |
-
✅ Core Features: Auto ChromeDriver | Persistent sessions | JavaScript | Cookies | Forms | HTTP Requests
|
| 1337 |
-
|
| 1338 |
-
🚀 Advanced Features: Accessibility Audit | Visual Testing | Network Monitor | Structured Data | Link Extraction
|
| 1339 |
-
""")
|
| 1340 |
-
|
| 1341 |
-
with gr.Tab("🔍 Extract Text"):
|
| 1342 |
-
with gr.Row():
|
| 1343 |
-
url_in = gr.Textbox(label="URL", value="https://example.com")
|
| 1344 |
-
sel_in = gr.Textbox(label="CSS Selector", value="body")
|
| 1345 |
-
persist_check = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1346 |
-
out = gr.Textbox(label="Extracted Text", lines=10)
|
| 1347 |
-
gr.Button("Extract", variant="primary").click(browse_and_extract, [url_in, sel_in, persist_check], out)
|
| 1348 |
-
|
| 1349 |
-
with gr.Tab("📸 Screenshot"):
|
| 1350 |
-
with gr.Row():
|
| 1351 |
-
url_in2 = gr.Textbox(label="URL", value="https://example.com")
|
| 1352 |
-
full_page_check = gr.Checkbox(label="Full Page Screenshot", value=False)
|
| 1353 |
-
persist_check2 = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1354 |
-
out_img = gr.Image(label="Screenshot")
|
| 1355 |
-
gr.Button("Take Screenshot", variant="primary").click(screenshot, [url_in2, full_page_check, persist_check2], out_img)
|
| 1356 |
-
|
| 1357 |
-
with gr.Tab("👆 Click"):
|
| 1358 |
-
with gr.Row():
|
| 1359 |
-
url_in3 = gr.Textbox(label="URL")
|
| 1360 |
-
sel_in3 = gr.Textbox(label="CSS Selector", placeholder="button#submit")
|
| 1361 |
-
persist_check3 = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1362 |
-
out3 = gr.Textbox(label="Result", lines=4)
|
| 1363 |
-
gr.Button("Click", variant="primary").click(click, [url_in3, sel_in3, persist_check3], out3)
|
| 1364 |
-
|
| 1365 |
-
with gr.Tab("✏️ Fill"):
|
| 1366 |
-
with gr.Row():
|
| 1367 |
-
url_in4 = gr.Textbox(label="URL")
|
| 1368 |
-
sel_in4 = gr.Textbox(label="CSS Selector", placeholder="input#username")
|
| 1369 |
-
txt_in4 = gr.Textbox(label="Text to Fill")
|
| 1370 |
-
persist_check4 = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1371 |
-
out4 = gr.Textbox(label="Result", lines=2)
|
| 1372 |
-
gr.Button("Fill", variant="primary").click(fill, [url_in4, sel_in4, txt_in4, persist_check4], out4)
|
| 1373 |
-
|
| 1374 |
-
with gr.Tab("📝 Form Submit"):
|
| 1375 |
-
with gr.Row():
|
| 1376 |
-
url_in_form = gr.Textbox(label="URL")
|
| 1377 |
-
persist_check_form = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1378 |
-
form_data_in = gr.Textbox(
|
| 1379 |
-
label="Form Data",
|
| 1380 |
-
placeholder="selector1:value1;selector2:value2",
|
| 1381 |
-
lines=3,
|
| 1382 |
-
info="Format: CSS_selector:value pairs separated by semicolons"
|
| 1383 |
-
)
|
| 1384 |
-
submit_btn_in = gr.Textbox(label="Submit Button Selector (optional)", placeholder="button[type='submit']")
|
| 1385 |
-
out_form = gr.Textbox(label="Result", lines=4)
|
| 1386 |
-
gr.Button("Submit Form", variant="primary").click(
|
| 1387 |
-
submit_form,
|
| 1388 |
-
[url_in_form, form_data_in, submit_btn_in, persist_check_form],
|
| 1389 |
-
out_form
|
| 1390 |
-
)
|
| 1391 |
-
|
| 1392 |
-
with gr.Tab("💻 Execute JS"):
|
| 1393 |
-
with gr.Row():
|
| 1394 |
-
url_in5 = gr.Textbox(label="URL", value="https://example.com")
|
| 1395 |
-
persist_check5 = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1396 |
-
script_in5 = gr.Textbox(
|
| 1397 |
-
label="JavaScript Code",
|
| 1398 |
-
lines=5,
|
| 1399 |
-
value="return {title: document.title, url: document.URL, links: document.links.length};"
|
| 1400 |
-
)
|
| 1401 |
-
out5 = gr.Textbox(label="Result", lines=3)
|
| 1402 |
-
gr.Button("Execute", variant="primary").click(execute_js, [url_in5, script_in5, persist_check5], out5)
|
| 1403 |
-
|
| 1404 |
-
with gr.Tab("🍪 Cookies"):
|
| 1405 |
-
with gr.Row():
|
| 1406 |
-
with gr.Column():
|
| 1407 |
-
gr.Markdown("### Get Cookies")
|
| 1408 |
-
url_in_cookies = gr.Textbox(label="URL")
|
| 1409 |
-
persist_check_cookies = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1410 |
-
cookies_out = gr.Textbox(label="Cookies (JSON)", lines=10)
|
| 1411 |
-
gr.Button("Get Cookies").click(get_cookies, [url_in_cookies, persist_check_cookies], cookies_out)
|
| 1412 |
-
|
| 1413 |
-
with gr.Column():
|
| 1414 |
-
gr.Markdown("### Set Cookies")
|
| 1415 |
-
url_in_set_cookies = gr.Textbox(label="URL")
|
| 1416 |
-
cookies_in = gr.Textbox(label="Cookies JSON", lines=8, placeholder='[{"name": "cookie_name", "value": "cookie_value"}]')
|
| 1417 |
-
persist_check_set_cookies = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1418 |
-
set_cookies_out = gr.Textbox(label="Result")
|
| 1419 |
-
gr.Button("Set Cookies").click(set_cookies, [url_in_set_cookies, cookies_in, persist_check_set_cookies], set_cookies_out)
|
| 1420 |
-
|
| 1421 |
-
with gr.Tab("⏳ Wait for Element"):
|
| 1422 |
-
with gr.Row():
|
| 1423 |
-
url_in_wait = gr.Textbox(label="URL")
|
| 1424 |
-
sel_in_wait = gr.Textbox(label="CSS Selector", placeholder=".dynamic-content")
|
| 1425 |
-
timeout_in = gr.Slider(minimum=1, maximum=60, value=10, label="Timeout (seconds)")
|
| 1426 |
-
persist_check_wait = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1427 |
-
out_wait = gr.Textbox(label="Result", lines=3)
|
| 1428 |
-
gr.Button("Wait for Element", variant="primary").click(
|
| 1429 |
-
wait_for_element,
|
| 1430 |
-
[url_in_wait, sel_in_wait, timeout_in, persist_check_wait],
|
| 1431 |
-
out_wait
|
| 1432 |
-
)
|
| 1433 |
-
|
| 1434 |
-
with gr.Tab("📜 Scroll"):
|
| 1435 |
-
with gr.Row():
|
| 1436 |
-
url_in_scroll = gr.Textbox(label="URL")
|
| 1437 |
-
direction_in = gr.Radio(["bottom", "top", "down", "up"], value="bottom", label="Direction")
|
| 1438 |
-
pixels_in = gr.Number(label="Pixels (for up/down)", value=500)
|
| 1439 |
-
persist_check_scroll = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1440 |
-
out_scroll = gr.Textbox(label="Result", lines=2)
|
| 1441 |
-
gr.Button("Scroll", variant="primary").click(
|
| 1442 |
-
scroll_page,
|
| 1443 |
-
[url_in_scroll, direction_in, pixels_in, persist_check_scroll],
|
| 1444 |
-
out_scroll
|
| 1445 |
-
)
|
| 1446 |
-
|
| 1447 |
-
with gr.Tab("🎯 Hover"):
|
| 1448 |
-
with gr.Row():
|
| 1449 |
-
url_in_hover = gr.Textbox(label="URL")
|
| 1450 |
-
sel_in_hover = gr.Textbox(label="CSS Selector", placeholder=".menu-item")
|
| 1451 |
-
persist_check_hover = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1452 |
-
out_hover = gr.Textbox(label="Result", lines=2)
|
| 1453 |
-
gr.Button("Hover", variant="primary").click(hover_element, [url_in_hover, sel_in_hover, persist_check_hover], out_hover)
|
| 1454 |
-
|
| 1455 |
-
with gr.Tab("⌨️ Press Key"):
|
| 1456 |
-
with gr.Row():
|
| 1457 |
-
url_in_key = gr.Textbox(label="URL")
|
| 1458 |
-
key_in = gr.Textbox(label="Key", placeholder="enter, escape, tab, etc.")
|
| 1459 |
-
sel_in_key = gr.Textbox(label="Target Element", value="body")
|
| 1460 |
-
persist_check_key = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1461 |
-
out_key = gr.Textbox(label="Result", lines=2)
|
| 1462 |
-
gr.Button("Press Key", variant="primary").click(
|
| 1463 |
-
press_key,
|
| 1464 |
-
[url_in_key, key_in, sel_in_key, persist_check_key],
|
| 1465 |
-
out_key
|
| 1466 |
-
)
|
| 1467 |
-
|
| 1468 |
-
with gr.Tab("ℹ️ Page Info"):
|
| 1469 |
-
with gr.Row():
|
| 1470 |
-
url_in_info = gr.Textbox(label="URL")
|
| 1471 |
-
persist_check_info = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1472 |
-
out_info = gr.Textbox(label="Page Information", lines=15)
|
| 1473 |
-
gr.Button("Get Info", variant="primary").click(get_page_info, [url_in_info, persist_check_info], out_info)
|
| 1474 |
-
|
| 1475 |
-
with gr.Tab("📄 HTML Source"):
|
| 1476 |
-
with gr.Row():
|
| 1477 |
-
url_in_html = gr.Textbox(label="URL", value="https://example.com")
|
| 1478 |
-
persist_check_html = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1479 |
-
with gr.Row():
|
| 1480 |
-
with gr.Column():
|
| 1481 |
-
out_html = gr.Code(label="HTML Source Code", language="html", lines=20)
|
| 1482 |
-
gr.Button("Get HTML", variant="primary").click(get_html_source, [url_in_html, persist_check_html], out_html)
|
| 1483 |
-
with gr.Column():
|
| 1484 |
-
filename_in = gr.Textbox(label="Filename (optional)", placeholder="auto-generated if empty")
|
| 1485 |
-
save_out = gr.File(label="Saved HTML File")
|
| 1486 |
-
gr.Button("Save HTML to File", variant="secondary").click(
|
| 1487 |
-
save_html_to_file,
|
| 1488 |
-
[url_in_html, filename_in, persist_check_html],
|
| 1489 |
-
save_out
|
| 1490 |
-
)
|
| 1491 |
-
|
| 1492 |
-
with gr.Tab("🔗 HTTP Request"):
|
| 1493 |
-
with gr.Row():
|
| 1494 |
-
url_in_http = gr.Textbox(label="URL", value="https://api.example.com/data")
|
| 1495 |
-
method_in = gr.Dropdown(
|
| 1496 |
-
["GET", "POST", "PUT", "PATCH", "DELETE", "HEAD", "OPTIONS"],
|
| 1497 |
-
value="GET",
|
| 1498 |
-
label="Method"
|
| 1499 |
-
)
|
| 1500 |
-
persist_check_http = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1501 |
-
|
| 1502 |
-
headers_in = gr.Code(
|
| 1503 |
-
label="Headers (JSON or key:value format)",
|
| 1504 |
-
language="json",
|
| 1505 |
-
lines=5,
|
| 1506 |
-
value='{\n "Content-Type": "application/json",\n "User-Agent": "MCP-Browser/1.0"\n}'
|
| 1507 |
-
)
|
| 1508 |
-
|
| 1509 |
-
data_in = gr.Code(
|
| 1510 |
-
label="Body Data (for POST/PUT/PATCH)",
|
| 1511 |
-
language="json",
|
| 1512 |
-
lines=5,
|
| 1513 |
-
value='{\n "key": "value"\n}'
|
| 1514 |
-
)
|
| 1515 |
-
|
| 1516 |
-
with gr.Row():
|
| 1517 |
-
out_http = gr.Code(label="Response", language="json", lines=15)
|
| 1518 |
-
curl_out = gr.Textbox(label="cURL Command", lines=5)
|
| 1519 |
-
|
| 1520 |
-
with gr.Row():
|
| 1521 |
-
gr.Button("Send Request", variant="primary").click(
|
| 1522 |
-
make_http_request,
|
| 1523 |
-
[url_in_http, method_in, headers_in, data_in, persist_check_http],
|
| 1524 |
-
out_http
|
| 1525 |
-
)
|
| 1526 |
-
gr.Button("Export as cURL", variant="secondary").click(
|
| 1527 |
-
export_as_curl,
|
| 1528 |
-
[method_in, url_in_http, headers_in, data_in],
|
| 1529 |
-
curl_out
|
| 1530 |
-
)
|
| 1531 |
-
|
| 1532 |
-
with gr.Row():
|
| 1533 |
-
history_out = gr.Code(label="Request History", language="json", lines=10)
|
| 1534 |
-
gr.Button("Show History", variant="secondary").click(get_request_history, [], history_out)
|
| 1535 |
-
|
| 1536 |
-
with gr.Tab("🌐 Network Monitor"):
|
| 1537 |
-
with gr.Row():
|
| 1538 |
-
url_in_network = gr.Textbox(label="URL", value="https://example.com")
|
| 1539 |
-
duration_in = gr.Slider(minimum=1, maximum=30, value=5, label="Monitor Duration (seconds)")
|
| 1540 |
-
persist_check_network = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1541 |
-
out_network = gr.Code(label="Network Requests", language="json", lines=15)
|
| 1542 |
-
gr.Button("Monitor Network", variant="primary").click(
|
| 1543 |
-
monitor_network_requests,
|
| 1544 |
-
[url_in_network, duration_in, persist_check_network],
|
| 1545 |
-
out_network
|
| 1546 |
-
)
|
| 1547 |
-
|
| 1548 |
-
with gr.Tab("📊 Structured Data"):
|
| 1549 |
-
with gr.Row():
|
| 1550 |
-
url_in_structured = gr.Textbox(label="URL", value="https://example.com")
|
| 1551 |
-
persist_check_structured = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1552 |
-
out_structured = gr.Code(label="Structured Data (JSON-LD, Meta, OpenGraph)", language="json", lines=15)
|
| 1553 |
-
gr.Button("Extract Structured Data", variant="primary").click(
|
| 1554 |
-
extract_structured_data,
|
| 1555 |
-
[url_in_structured, persist_check_structured],
|
| 1556 |
-
out_structured
|
| 1557 |
-
)
|
| 1558 |
-
|
| 1559 |
-
with gr.Tab("🎨 Visual Testing"):
|
| 1560 |
-
with gr.Row():
|
| 1561 |
-
url1_in = gr.Textbox(label="URL 1", value="https://example.com")
|
| 1562 |
-
url2_in = gr.Textbox(label="URL 2", value="https://example.com/new")
|
| 1563 |
-
threshold_in = gr.Slider(minimum=0, maximum=1, value=0.98, label="Similarity Threshold")
|
| 1564 |
-
out_visual = gr.Code(label="Visual Comparison Result", language="json", lines=10)
|
| 1565 |
-
gr.Button("Compare Pages", variant="primary").click(
|
| 1566 |
-
visual_regression_test,
|
| 1567 |
-
[url1_in, url2_in, threshold_in],
|
| 1568 |
-
out_visual
|
| 1569 |
-
)
|
| 1570 |
-
|
| 1571 |
-
with gr.Tab("♿ Accessibility"):
|
| 1572 |
-
with gr.Row():
|
| 1573 |
-
url_in_a11y = gr.Textbox(label="URL", value="https://example.com")
|
| 1574 |
-
persist_check_a11y = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1575 |
-
out_a11y = gr.Code(label="Accessibility Audit Results", language="json", lines=20)
|
| 1576 |
-
gr.Button("Run Audit", variant="primary").click(
|
| 1577 |
-
accessibility_audit,
|
| 1578 |
-
[url_in_a11y, persist_check_a11y],
|
| 1579 |
-
out_a11y
|
| 1580 |
-
)
|
| 1581 |
-
|
| 1582 |
-
with gr.Tab("🔗 Link Extractor"):
|
| 1583 |
-
with gr.Row():
|
| 1584 |
-
url_in_links = gr.Textbox(label="URL", value="https://example.com")
|
| 1585 |
-
include_external_check = gr.Checkbox(label="Include External Links", value=True)
|
| 1586 |
-
persist_check_links = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1587 |
-
out_links = gr.Code(label="All Links (Categorized)", language="json", lines=20)
|
| 1588 |
-
gr.Button("Extract Links", variant="primary").click(
|
| 1589 |
-
extract_all_links,
|
| 1590 |
-
[url_in_links, include_external_check, persist_check_links],
|
| 1591 |
-
out_links
|
| 1592 |
-
)
|
| 1593 |
-
|
| 1594 |
-
with gr.Tab("📸 Screenshot"):
|
| 1595 |
-
with gr.Row():
|
| 1596 |
-
url_in2 = gr.Textbox(label="URL", value="https://example.com")
|
| 1597 |
-
full_page_check = gr.Checkbox(label="Full Page Screenshot", value=False)
|
| 1598 |
-
persist_check2 = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1599 |
-
out_img = gr.Image(label="Screenshot")
|
| 1600 |
-
gr.Button("Take Screenshot", variant="primary").click(screenshot, [url_in2, full_page_check, persist_check2], out_img)
|
| 1601 |
-
|
| 1602 |
-
with gr.Tab("👆 Click"):
|
| 1603 |
-
with gr.Row():
|
| 1604 |
-
url_in3 = gr.Textbox(label="URL")
|
| 1605 |
-
sel_in3 = gr.Textbox(label="CSS Selector", placeholder="button#submit")
|
| 1606 |
-
persist_check3 = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1607 |
-
out3 = gr.Textbox(label="Result", lines=4)
|
| 1608 |
-
gr.Button("Click", variant="primary").click(click, [url_in3, sel_in3, persist_check3], out3)
|
| 1609 |
-
|
| 1610 |
-
with gr.Tab("✏️ Fill"):
|
| 1611 |
-
with gr.Row():
|
| 1612 |
-
url_in4 = gr.Textbox(label="URL")
|
| 1613 |
-
sel_in4 = gr.Textbox(label="CSS Selector", placeholder="input#username")
|
| 1614 |
-
txt_in4 = gr.Textbox(label="Text to Fill")
|
| 1615 |
-
persist_check4 = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1616 |
-
out4 = gr.Textbox(label="Result", lines=2)
|
| 1617 |
-
gr.Button("Fill", variant="primary").click(fill, [url_in4, sel_in4, txt_in4, persist_check4], out4)
|
| 1618 |
-
|
| 1619 |
-
with gr.Tab("📝 Form Submit"):
|
| 1620 |
-
with gr.Row():
|
| 1621 |
-
url_in_form = gr.Textbox(label="URL")
|
| 1622 |
-
persist_check_form = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1623 |
-
form_data_in = gr.Textbox(
|
| 1624 |
-
label="Form Data",
|
| 1625 |
-
placeholder="selector1:value1;selector2:value2",
|
| 1626 |
-
lines=3,
|
| 1627 |
-
info="Format: CSS_selector:value pairs separated by semicolons"
|
| 1628 |
-
)
|
| 1629 |
-
submit_btn_in = gr.Textbox(label="Submit Button Selector (optional)", placeholder="button[type='submit']")
|
| 1630 |
-
out_form = gr.Textbox(label="Result", lines=4)
|
| 1631 |
-
gr.Button("Submit Form", variant="primary").click(
|
| 1632 |
-
submit_form,
|
| 1633 |
-
[url_in_form, form_data_in, submit_btn_in, persist_check_form],
|
| 1634 |
-
out_form
|
| 1635 |
-
)
|
| 1636 |
-
|
| 1637 |
-
with gr.Tab("💻 Execute JS"):
|
| 1638 |
-
with gr.Row():
|
| 1639 |
-
url_in5 = gr.Textbox(label="URL", value="https://example.com")
|
| 1640 |
-
persist_check5 = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1641 |
-
script_in5 = gr.Textbox(
|
| 1642 |
-
label="JavaScript Code",
|
| 1643 |
-
lines=5,
|
| 1644 |
-
value="return {title: document.title, url: document.URL, links: document.links.length};"
|
| 1645 |
-
)
|
| 1646 |
-
out5 = gr.Textbox(label="Result", lines=3)
|
| 1647 |
-
gr.Button("Execute", variant="primary").click(execute_js, [url_in5, script_in5, persist_check5], out5)
|
| 1648 |
-
|
| 1649 |
-
with gr.Tab("🍪 Cookies"):
|
| 1650 |
-
with gr.Row():
|
| 1651 |
-
with gr.Column():
|
| 1652 |
-
gr.Markdown("### Get Cookies")
|
| 1653 |
-
url_in_cookies = gr.Textbox(label="URL")
|
| 1654 |
-
persist_check_cookies = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1655 |
-
cookies_out = gr.Textbox(label="Cookies (JSON)", lines=10)
|
| 1656 |
-
gr.Button("Get Cookies").click(get_cookies, [url_in_cookies, persist_check_cookies], cookies_out)
|
| 1657 |
-
|
| 1658 |
-
with gr.Column():
|
| 1659 |
-
gr.Markdown("### Set Cookies")
|
| 1660 |
-
url_in_set_cookies = gr.Textbox(label="URL")
|
| 1661 |
-
cookies_in = gr.Textbox(label="Cookies JSON", lines=8, placeholder='[{"name": "cookie_name", "value": "cookie_value"}]')
|
| 1662 |
-
persist_check_set_cookies = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1663 |
-
set_cookies_out = gr.Textbox(label="Result")
|
| 1664 |
-
gr.Button("Set Cookies").click(set_cookies, [url_in_set_cookies, cookies_in, persist_check_set_cookies], set_cookies_out)
|
| 1665 |
-
|
| 1666 |
-
with gr.Tab("⏳ Wait for Element"):
|
| 1667 |
-
with gr.Row():
|
| 1668 |
-
url_in_wait = gr.Textbox(label="URL")
|
| 1669 |
-
sel_in_wait = gr.Textbox(label="CSS Selector", placeholder=".dynamic-content")
|
| 1670 |
-
timeout_in = gr.Slider(minimum=1, maximum=60, value=10, label="Timeout (seconds)")
|
| 1671 |
-
persist_check_wait = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1672 |
-
out_wait = gr.Textbox(label="Result", lines=3)
|
| 1673 |
-
gr.Button("Wait for Element", variant="primary").click(
|
| 1674 |
-
wait_for_element,
|
| 1675 |
-
[url_in_wait, sel_in_wait, timeout_in, persist_check_wait],
|
| 1676 |
-
out_wait
|
| 1677 |
-
)
|
| 1678 |
-
|
| 1679 |
-
with gr.Tab("📜 Scroll"):
|
| 1680 |
-
with gr.Row():
|
| 1681 |
-
url_in_scroll = gr.Textbox(label="URL")
|
| 1682 |
-
direction_in = gr.Radio(["bottom", "top", "down", "up"], value="bottom", label="Direction")
|
| 1683 |
-
pixels_in = gr.Number(label="Pixels (for up/down)", value=500)
|
| 1684 |
-
persist_check_scroll = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1685 |
-
out_scroll = gr.Textbox(label="Result", lines=2)
|
| 1686 |
-
gr.Button("Scroll", variant="primary").click(
|
| 1687 |
-
scroll_page,
|
| 1688 |
-
[url_in_scroll, direction_in, pixels_in, persist_check_scroll],
|
| 1689 |
-
out_scroll
|
| 1690 |
-
)
|
| 1691 |
-
|
| 1692 |
-
with gr.Tab("🎯 Hover"):
|
| 1693 |
-
with gr.Row():
|
| 1694 |
-
url_in_hover = gr.Textbox(label="URL")
|
| 1695 |
-
sel_in_hover = gr.Textbox(label="CSS Selector", placeholder=".menu-item")
|
| 1696 |
-
persist_check_hover = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1697 |
-
out_hover = gr.Textbox(label="Result", lines=2)
|
| 1698 |
-
gr.Button("Hover", variant="primary").click(hover_element, [url_in_hover, sel_in_hover, persist_check_hover], out_hover)
|
| 1699 |
-
|
| 1700 |
-
with gr.Tab("⌨️ Press Key"):
|
| 1701 |
-
with gr.Row():
|
| 1702 |
-
url_in_key = gr.Textbox(label="URL")
|
| 1703 |
-
key_in = gr.Textbox(label="Key", placeholder="enter, escape, tab, etc.")
|
| 1704 |
-
sel_in_key = gr.Textbox(label="Target Element", value="body")
|
| 1705 |
-
persist_check_key = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1706 |
-
out_key = gr.Textbox(label="Result", lines=2)
|
| 1707 |
-
gr.Button("Press Key", variant="primary").click(
|
| 1708 |
-
press_key,
|
| 1709 |
-
[url_in_key, key_in, sel_in_key, persist_check_key],
|
| 1710 |
-
out_key
|
| 1711 |
-
)
|
| 1712 |
-
|
| 1713 |
-
with gr.Tab("ℹ️ Page Info"):
|
| 1714 |
-
with gr.Row():
|
| 1715 |
-
url_in_info = gr.Textbox(label="URL")
|
| 1716 |
-
persist_check_info = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1717 |
-
out_info = gr.Textbox(label="Page Information", lines=15)
|
| 1718 |
-
gr.Button("Get Info", variant="primary").click(get_page_info, [url_in_info, persist_check_info], out_info)
|
| 1719 |
-
|
| 1720 |
-
with gr.Tab("📄 HTML Source"):
|
| 1721 |
-
with gr.Row():
|
| 1722 |
-
url_in_html = gr.Textbox(label="URL", value="https://example.com")
|
| 1723 |
-
persist_check_html = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1724 |
-
with gr.Row():
|
| 1725 |
-
with gr.Column():
|
| 1726 |
-
out_html = gr.Code(label="HTML Source Code", language="html", lines=20)
|
| 1727 |
-
gr.Button("Get HTML", variant="primary").click(get_html_source, [url_in_html, persist_check_html], out_html)
|
| 1728 |
-
with gr.Column():
|
| 1729 |
-
filename_in = gr.Textbox(label="Filename (optional)", placeholder="auto-generated if empty")
|
| 1730 |
-
save_out = gr.File(label="Saved HTML File")
|
| 1731 |
-
gr.Button("Save HTML to File", variant="secondary").click(
|
| 1732 |
-
save_html_to_file,
|
| 1733 |
-
[url_in_html, filename_in, persist_check_html],
|
| 1734 |
-
save_out
|
| 1735 |
-
)
|
| 1736 |
-
|
| 1737 |
-
with gr.Tab("🔗 HTTP Request"):
|
| 1738 |
-
with gr.Row():
|
| 1739 |
-
url_in_http = gr.Textbox(label="URL", value="https://api.example.com/data")
|
| 1740 |
-
method_in = gr.Dropdown(
|
| 1741 |
-
["GET", "POST", "PUT", "PATCH", "DELETE", "HEAD", "OPTIONS"],
|
| 1742 |
-
value="GET",
|
| 1743 |
-
label="Method"
|
| 1744 |
-
)
|
| 1745 |
-
persist_check_http = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1746 |
-
|
| 1747 |
-
headers_in = gr.Code(
|
| 1748 |
-
label="Headers (JSON or key:value format)",
|
| 1749 |
-
language="json",
|
| 1750 |
-
lines=5,
|
| 1751 |
-
value='{\n "Content-Type": "application/json",\n "User-Agent": "MCP-Browser/1.0"\n}'
|
| 1752 |
-
)
|
| 1753 |
-
|
| 1754 |
-
data_in = gr.Code(
|
| 1755 |
-
label="Body Data (for POST/PUT/PATCH)",
|
| 1756 |
-
language="json",
|
| 1757 |
-
lines=5,
|
| 1758 |
-
value='{\n "key": "value"\n}'
|
| 1759 |
-
)
|
| 1760 |
-
|
| 1761 |
-
with gr.Row():
|
| 1762 |
-
out_http = gr.Code(label="Response", language="json", lines=15)
|
| 1763 |
-
curl_out = gr.Textbox(label="cURL Command", lines=5)
|
| 1764 |
-
|
| 1765 |
-
with gr.Row():
|
| 1766 |
-
gr.Button("Send Request", variant="primary").click(
|
| 1767 |
-
make_http_request,
|
| 1768 |
-
[url_in_http, method_in, headers_in, data_in, persist_check_http],
|
| 1769 |
-
out_http
|
| 1770 |
-
)
|
| 1771 |
-
gr.Button("Export as cURL", variant="secondary").click(
|
| 1772 |
-
export_as_curl,
|
| 1773 |
-
[method_in, url_in_http, headers_in, data_in],
|
| 1774 |
-
curl_out
|
| 1775 |
-
)
|
| 1776 |
-
|
| 1777 |
-
with gr.Row():
|
| 1778 |
-
history_out = gr.Code(label="Request History", language="json", lines=10)
|
| 1779 |
-
gr.Button("Show History", variant="secondary").click(get_request_history, [], history_out)
|
| 1780 |
-
|
| 1781 |
-
with gr.Tab("🌐 Network Monitor"):
|
| 1782 |
-
with gr.Row():
|
| 1783 |
-
url_in_network = gr.Textbox(label="URL", value="https://example.com")
|
| 1784 |
-
duration_in = gr.Slider(minimum=1, maximum=30, value=5, label="Monitor Duration (seconds)")
|
| 1785 |
-
persist_check_network = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1786 |
-
out_network = gr.Code(label="Network Requests", language="json", lines=15)
|
| 1787 |
-
gr.Button("Monitor Network", variant="primary").click(
|
| 1788 |
-
monitor_network_requests,
|
| 1789 |
-
[url_in_network, duration_in, persist_check_network],
|
| 1790 |
-
out_network
|
| 1791 |
-
)
|
| 1792 |
-
|
| 1793 |
-
with gr.Tab("📊 Structured Data"):
|
| 1794 |
-
with gr.Row():
|
| 1795 |
-
url_in_structured = gr.Textbox(label="URL", value="https://example.com")
|
| 1796 |
-
persist_check_structured = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1797 |
-
out_structured = gr.Code(label="Structured Data (JSON-LD, Meta, OpenGraph)", language="json", lines=15)
|
| 1798 |
-
gr.Button("Extract Structured Data", variant="primary").click(
|
| 1799 |
-
extract_structured_data,
|
| 1800 |
-
[url_in_structured, persist_check_structured],
|
| 1801 |
-
out_structured
|
| 1802 |
-
)
|
| 1803 |
-
|
| 1804 |
-
with gr.Tab("🎨 Visual Testing"):
|
| 1805 |
-
with gr.Row():
|
| 1806 |
-
url1_in = gr.Textbox(label="URL 1", value="https://example.com")
|
| 1807 |
-
url2_in = gr.Textbox(label="URL 2", value="https://example.com/new")
|
| 1808 |
-
threshold_in = gr.Slider(minimum=0, maximum=1, value=0.98, label="Similarity Threshold")
|
| 1809 |
-
out_visual = gr.Code(label="Visual Comparison Result", language="json", lines=10)
|
| 1810 |
-
gr.Button("Compare Pages", variant="primary").click(
|
| 1811 |
-
visual_regression_test,
|
| 1812 |
-
[url1_in, url2_in, threshold_in],
|
| 1813 |
-
out_visual
|
| 1814 |
-
)
|
| 1815 |
-
|
| 1816 |
-
with gr.Tab("♿ Accessibility"):
|
| 1817 |
-
with gr.Row():
|
| 1818 |
-
url_in_a11y = gr.Textbox(label="URL", value="https://example.com")
|
| 1819 |
-
persist_check_a11y = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1820 |
-
out_a11y = gr.Code(label="Accessibility Audit Results", language="json", lines=20)
|
| 1821 |
-
gr.Button("Run Audit", variant="primary").click(
|
| 1822 |
-
accessibility_audit,
|
| 1823 |
-
[url_in_a11y, persist_check_a11y],
|
| 1824 |
-
out_a11y
|
| 1825 |
-
)
|
| 1826 |
-
|
| 1827 |
-
with gr.Tab("🔗 Link Extractor"):
|
| 1828 |
-
with gr.Row():
|
| 1829 |
-
url_in_links = gr.Textbox(label="URL", value="https://example.com")
|
| 1830 |
-
include_external_check = gr.Checkbox(label="Include External Links", value=True)
|
| 1831 |
-
persist_check_links = gr.Checkbox(label="Use Persistent Browser", value=False)
|
| 1832 |
-
out_links = gr.Code(label="All Links (Categorized)", language="json", lines=20)
|
| 1833 |
-
gr.Button("Extract Links", variant="primary").click(
|
| 1834 |
-
extract_all_links,
|
| 1835 |
-
[url_in_links, include_external_check, persist_check_links],
|
| 1836 |
-
out_links
|
| 1837 |
-
)
|
| 1838 |
-
|
| 1839 |
-
with gr.Tab("🔧 Browser Control"):
|
| 1840 |
-
close_out = gr.Textbox(label="Status")
|
| 1841 |
-
gr.Button("Close Persistent Browser", variant="stop").click(close_persistent_driver, [], close_out)
|
| 1842 |
-
|
| 1843 |
-
gr.Markdown("""
|
| 1844 |
-
### 📖 Feature Guide:
|
| 1845 |
-
|
| 1846 |
-
**🚀 New Advanced Features:**
|
| 1847 |
-
|
| 1848 |
-
1. **📊 Structured Data Extraction**: Extract JSON-LD, OpenGraph, Twitter Cards, and microdata
|
| 1849 |
-
2. **🌐 Network Monitor**: Track all network requests made by a page (XHR, fetch, resources)
|
| 1850 |
-
3. **♿ Accessibility Audit**: Check for common accessibility issues (alt text, labels, landmarks, etc.)
|
| 1851 |
-
4. **🎨 Visual Regression Testing**: Compare two pages visually to detect UI changes
|
| 1852 |
-
5. **🔗 Smart Link Extraction**: Categorize all links (internal, external, email, phone, etc.)
|
| 1853 |
-
6. **💾 HTML Save**: Save full rendered HTML to file
|
| 1854 |
-
7. **📜 cURL Export**: Convert HTTP requests to cURL commands
|
| 1855 |
-
8. **📝 Request History**: Track all HTTP requests made during session
|
| 1856 |
-
|
| 1857 |
-
**💡 Pro Tips:**
|
| 1858 |
-
- Use persistent browser for faster multi-step workflows
|
| 1859 |
-
- Network monitor captures AJAX/fetch requests invisible to normal inspection
|
| 1860 |
-
- Accessibility audit helps ensure WCAG compliance
|
| 1861 |
-
- Visual testing is great for detecting unintended UI changes
|
| 1862 |
-
- Structured data extraction helps with SEO analysis
|
| 1863 |
-
""")
|
| 1864 |
-
|
| 1865 |
-
# Sidebar with Gemini Chat
|
| 1866 |
-
with gr.Column(elem_classes="sidebar"):
|
| 1867 |
-
gr.Markdown("""
|
| 1868 |
-
## 🤖 Gemini AI Assistant
|
| 1869 |
-
|
| 1870 |
-
Ask me anything about web scraping, automation, or get help with using this tool!
|
| 1871 |
-
""")
|
| 1872 |
-
|
| 1873 |
-
if not GEMINI_API_KEY:
|
| 1874 |
-
gr.Markdown("""
|
| 1875 |
-
⚠️ **Setup Required**: Add your Gemini API key to Hugging Face secrets:
|
| 1876 |
-
1. Go to your Space Settings
|
| 1877 |
-
2. Add a new secret named `GEMINI_API_KEY`
|
| 1878 |
-
3. Get your API key from [Google AI Studio](https://makersuite.google.com/app/apikey)
|
| 1879 |
-
""")
|
| 1880 |
-
|
| 1881 |
-
chatbot = gr.Chatbot(
|
| 1882 |
-
value=[],
|
| 1883 |
-
elem_classes="chat-container",
|
| 1884 |
-
bubble_full_width=False,
|
| 1885 |
-
avatar_images=(None, "🤖")
|
| 1886 |
-
)
|
| 1887 |
-
|
| 1888 |
-
msg = gr.Textbox(
|
| 1889 |
-
label="Message",
|
| 1890 |
-
placeholder="Ask me about web scraping, CSS selectors, or any feature...",
|
| 1891 |
-
lines=2
|
| 1892 |
-
)
|
| 1893 |
-
|
| 1894 |
-
with gr.Row():
|
| 1895 |
-
submit = gr.Button("Send", variant="primary", scale=1)
|
| 1896 |
-
clear = gr.Button("Clear", variant="secondary", scale=1)
|
| 1897 |
-
|
| 1898 |
-
# Example prompts
|
| 1899 |
-
gr.Examples(
|
| 1900 |
-
examples=[
|
| 1901 |
-
"How do I extract all links from a webpage?",
|
| 1902 |
-
"What's the CSS selector for the first paragraph?",
|
| 1903 |
-
"How can I handle dynamic content that loads with JavaScript?",
|
| 1904 |
-
"Explain the difference between persistent and non-persistent browser",
|
| 1905 |
-
"How do I bypass CORS when making API requests?",
|
| 1906 |
-
"What's the best way to extract data from a table?",
|
| 1907 |
-
"How can I automate form filling and submission?",
|
| 1908 |
-
"Help me understand the accessibility audit results"
|
| 1909 |
-
],
|
| 1910 |
-
inputs=msg,
|
| 1911 |
-
label="Example Questions"
|
| 1912 |
-
)
|
| 1913 |
-
|
| 1914 |
-
# Chat event handlers
|
| 1915 |
-
msg.submit(chat_with_gemini, [msg, chatbot], [msg, chatbot])
|
| 1916 |
-
msg.submit(lambda: "", None, msg) # Clear input
|
| 1917 |
-
submit.click(chat_with_gemini, [msg, chatbot], [msg, chatbot])
|
| 1918 |
-
submit.click(lambda: "", None, msg) # Clear input
|
| 1919 |
-
clear.click(clear_chat, None, [msg, chatbot])
|
| 1920 |
|
| 1921 |
if __name__ == "__main__":
|
|
|
|
| 1922 |
demo.launch(mcp_server=True)
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Main application entry point for MCP Browser
|
| 3 |
+
"""
|
| 4 |
import gradio as gr
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
import logging
|
| 6 |
+
from config.settings import UI_CSS, GEMINI_API_KEY
|
| 7 |
+
from ui.layout import create_main_layout
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
|
| 9 |
# Set up logging
|
| 10 |
logging.basicConfig(level=logging.INFO)
|
| 11 |
logger = logging.getLogger(__name__)
|
| 12 |
|
| 13 |
+
# Create the main Gradio app
|
| 14 |
+
demo = create_main_layout()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
if __name__ == "__main__":
|
| 17 |
+
logger.info("Starting MCP Browser application...")
|
| 18 |
demo.launch(mcp_server=True)
|