Spaces:
Build error
Build error
Commit Β·
d033e7d
1
Parent(s): cc95037
test
Browse files- visa_availability_scraper_playwright.py +420 -107
visa_availability_scraper_playwright.py
CHANGED
|
@@ -2,6 +2,8 @@ import asyncio
|
|
| 2 |
import json
|
| 3 |
from typing import Dict, Optional, List
|
| 4 |
from playwright.async_api import async_playwright
|
|
|
|
|
|
|
| 5 |
|
| 6 |
class PassportIndexVisaScraper:
|
| 7 |
def __init__(self, debug: bool = True):
|
|
@@ -17,57 +19,102 @@ class PassportIndexVisaScraper:
|
|
| 17 |
self.browser = None
|
| 18 |
self.context = None
|
| 19 |
self.page = None
|
|
|
|
|
|
|
| 20 |
|
| 21 |
async def __aenter__(self):
|
| 22 |
"""Initialize browser with stealth mode"""
|
| 23 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
|
| 25 |
-
# Launch browser with stealth settings
|
| 26 |
self.browser = await self.playwright.chromium.launch(
|
| 27 |
-
headless=True, # Using headless mode
|
| 28 |
args=[
|
| 29 |
'--disable-blink-features=AutomationControlled',
|
| 30 |
'--disable-dev-shm-usage',
|
| 31 |
'--no-sandbox',
|
| 32 |
'--disable-setuid-sandbox',
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
'--disable-web-security',
|
| 34 |
-
'--disable-
|
|
|
|
|
|
|
|
|
|
| 35 |
]
|
| 36 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
|
| 38 |
# Create context with realistic settings
|
| 39 |
self.context = await self.browser.new_context(
|
| 40 |
-
viewport={'width':
|
| 41 |
-
user_agent=
|
| 42 |
locale='en-US',
|
| 43 |
-
timezone_id='America/New_York'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
)
|
| 45 |
|
| 46 |
self.page = await self.context.new_page()
|
| 47 |
|
| 48 |
-
# Add stealth
|
| 49 |
await self.page.add_init_script("""
|
| 50 |
-
//
|
| 51 |
-
Object.
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
parameters.name === 'notifications' ?
|
| 64 |
-
Promise.resolve({ state: Notification.permission }) :
|
| 65 |
-
originalQuery(parameters)
|
| 66 |
-
);
|
| 67 |
""")
|
| 68 |
|
| 69 |
if self.debug:
|
| 70 |
-
print("π Browser initialized with stealth mode")
|
| 71 |
|
| 72 |
return self
|
| 73 |
|
|
@@ -79,8 +126,8 @@ class PassportIndexVisaScraper:
|
|
| 79 |
await self.context.close()
|
| 80 |
if self.browser:
|
| 81 |
await self.browser.close()
|
| 82 |
-
if self.
|
| 83 |
-
await self.
|
| 84 |
|
| 85 |
if self.debug:
|
| 86 |
print("π Browser closed")
|
|
@@ -89,44 +136,147 @@ class PassportIndexVisaScraper:
|
|
| 89 |
"""
|
| 90 |
Navigate to the website and wait for it to load properly
|
| 91 |
"""
|
| 92 |
-
|
| 93 |
-
|
| 94 |
-
print("π± Initializing session...")
|
| 95 |
-
|
| 96 |
-
# Navigate to the page
|
| 97 |
try:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
response = await self.page.goto(
|
| 99 |
self.base_url,
|
| 100 |
-
wait_until='
|
| 101 |
timeout=30000
|
| 102 |
)
|
| 103 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
|
| 105 |
-
#
|
|
|
|
|
|
|
|
|
|
| 106 |
cl_value = await self.page.evaluate("""
|
| 107 |
() => {
|
| 108 |
const clInput = document.querySelector('#cl');
|
| 109 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 110 |
}
|
| 111 |
""")
|
| 112 |
|
| 113 |
if self.debug:
|
| 114 |
-
|
| 115 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
return True
|
| 117 |
-
|
| 118 |
except Exception as e:
|
| 119 |
-
if
|
| 120 |
-
print(f"
|
| 121 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 122 |
|
| 123 |
except Exception as e:
|
| 124 |
-
|
| 125 |
-
|
| 126 |
|
| 127 |
-
async def
|
| 128 |
"""
|
| 129 |
-
Check visa requirements using
|
| 130 |
|
| 131 |
Args:
|
| 132 |
passport_country: Two-letter country code for passport
|
|
@@ -137,9 +287,12 @@ class PassportIndexVisaScraper:
|
|
| 137 |
"""
|
| 138 |
try:
|
| 139 |
if self.debug:
|
| 140 |
-
print(f"π Checking {passport_country.upper()} β {destination_country.upper()}")
|
|
|
|
|
|
|
|
|
|
| 141 |
|
| 142 |
-
# Get the
|
| 143 |
cl_value = await self.page.evaluate("""
|
| 144 |
() => {
|
| 145 |
const clInput = document.querySelector('#cl');
|
|
@@ -147,13 +300,15 @@ class PassportIndexVisaScraper:
|
|
| 147 |
}
|
| 148 |
""")
|
| 149 |
|
| 150 |
-
# Make the API request
|
| 151 |
result = await self.page.evaluate("""
|
| 152 |
async (args) => {
|
| 153 |
const [passport, destination, sessionId] = args;
|
|
|
|
|
|
|
| 154 |
const formData = new URLSearchParams();
|
| 155 |
-
formData.append('d', destination);
|
| 156 |
-
formData.append('s', passport);
|
| 157 |
formData.append('cl', sessionId);
|
| 158 |
|
| 159 |
try {
|
|
@@ -162,91 +317,254 @@ class PassportIndexVisaScraper:
|
|
| 162 |
headers: {
|
| 163 |
'Content-Type': 'application/x-www-form-urlencoded; charset=UTF-8',
|
| 164 |
'X-Requested-With': 'XMLHttpRequest',
|
| 165 |
-
'Accept': 'application/json, text/javascript, */*; q=0.01'
|
|
|
|
|
|
|
| 166 |
},
|
| 167 |
body: formData.toString(),
|
| 168 |
-
credentials: 'include'
|
|
|
|
| 169 |
});
|
| 170 |
|
| 171 |
-
|
| 172 |
-
throw new Error(`HTTP ${response.status}`);
|
| 173 |
-
}
|
| 174 |
|
| 175 |
-
|
| 176 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 177 |
} catch (error) {
|
| 178 |
return { error: error.message };
|
| 179 |
}
|
| 180 |
}
|
| 181 |
-
""", [passport_country
|
| 182 |
|
| 183 |
-
if result and 'error' not in result:
|
| 184 |
if self.debug:
|
| 185 |
-
print(f"β
Got result: {result}")
|
| 186 |
return result
|
| 187 |
elif result and 'error' in result:
|
| 188 |
-
|
|
|
|
| 189 |
return None
|
| 190 |
else:
|
| 191 |
return None
|
| 192 |
|
| 193 |
except Exception as e:
|
| 194 |
-
|
|
|
|
| 195 |
return None
|
| 196 |
|
| 197 |
-
async def
|
| 198 |
"""
|
| 199 |
-
Alternative
|
| 200 |
"""
|
| 201 |
try:
|
| 202 |
if self.debug:
|
| 203 |
-
print(f"π±οΈ Using interactive method for {passport_country.upper()} β {destination_country.upper()}")
|
| 204 |
|
| 205 |
-
#
|
| 206 |
-
await self.
|
| 207 |
-
await self.page.wait_for_timeout(500)
|
| 208 |
|
| 209 |
-
#
|
| 210 |
-
|
| 211 |
-
|
| 212 |
-
|
| 213 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 214 |
|
| 215 |
-
# Click on the destination selector
|
| 216 |
-
await self.page.click('.vch-select-des')
|
| 217 |
await self.page.wait_for_timeout(500)
|
| 218 |
|
| 219 |
-
#
|
| 220 |
-
|
| 221 |
-
|
| 222 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 223 |
await self.page.wait_for_timeout(1000)
|
| 224 |
|
| 225 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 226 |
result = await self.page.evaluate("""
|
| 227 |
() => {
|
| 228 |
const resultElement = document.querySelector('.vch-result');
|
| 229 |
if (resultElement) {
|
| 230 |
const text = resultElement.querySelector('.text');
|
| 231 |
const days = resultElement.querySelector('.days');
|
|
|
|
|
|
|
|
|
|
| 232 |
return {
|
| 233 |
-
text: text ? text.textContent : '',
|
| 234 |
-
days: days ? days.textContent : '',
|
| 235 |
-
pass:
|
| 236 |
-
dest:
|
| 237 |
};
|
| 238 |
}
|
| 239 |
return null;
|
| 240 |
}
|
| 241 |
""")
|
| 242 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 243 |
return result
|
| 244 |
|
| 245 |
except Exception as e:
|
| 246 |
if self.debug:
|
| 247 |
-
print(f"β
|
| 248 |
return None
|
| 249 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 250 |
async def check_multiple_destinations(self, passport_country: str, destinations: List[str], delay: float = 2.0) -> Dict:
|
| 251 |
"""
|
| 252 |
Check visa requirements for multiple destinations
|
|
@@ -264,12 +582,7 @@ class PassportIndexVisaScraper:
|
|
| 264 |
for i, dest in enumerate(destinations, 1):
|
| 265 |
print(f"\n[{i}/{len(destinations)}] Checking {passport_country.upper()} β {dest.upper()}...")
|
| 266 |
|
| 267 |
-
|
| 268 |
-
result = await self.check_visa_requirement_browser(passport_country, dest)
|
| 269 |
-
|
| 270 |
-
# If API fails, try interactive method
|
| 271 |
-
if not result:
|
| 272 |
-
result = await self.check_visa_interactive(passport_country, dest)
|
| 273 |
|
| 274 |
if result:
|
| 275 |
results[dest] = result
|
|
@@ -291,17 +604,20 @@ class PassportIndexVisaScraper:
|
|
| 291 |
if not result:
|
| 292 |
return "No information available"
|
| 293 |
|
| 294 |
-
|
| 295 |
-
|
| 296 |
-
|
| 297 |
-
|
| 298 |
-
|
|
|
|
|
|
|
|
|
|
| 299 |
|
| 300 |
|
| 301 |
async def main():
|
| 302 |
"""Main function to demonstrate usage"""
|
| 303 |
print("="*60)
|
| 304 |
-
print(" Passport Index Visa Checker (Playwright)")
|
| 305 |
print("="*60)
|
| 306 |
|
| 307 |
async with PassportIndexVisaScraper(debug=True) as scraper:
|
|
@@ -317,14 +633,11 @@ async def main():
|
|
| 317 |
# Test single visa requirement
|
| 318 |
print("\nπ Single visa check: US β GB")
|
| 319 |
print("-" * 40)
|
| 320 |
-
result = await scraper.
|
| 321 |
if result:
|
| 322 |
-
print(f"Result: {
|
| 323 |
else:
|
| 324 |
-
print("
|
| 325 |
-
result = await scraper.check_visa_interactive('us', 'gb')
|
| 326 |
-
if result:
|
| 327 |
-
print(f"Result: {scraper.format_result(result)}")
|
| 328 |
|
| 329 |
# Test multiple destinations
|
| 330 |
print("\nπ Multiple destinations for US passport:")
|
|
@@ -335,9 +648,9 @@ async def main():
|
|
| 335 |
print("\nπ Summary:")
|
| 336 |
for dest, result in results.items():
|
| 337 |
if result:
|
| 338 |
-
print(f" β
{
|
| 339 |
else:
|
| 340 |
print(f" β US β {dest.upper()}: Failed")
|
| 341 |
|
| 342 |
if __name__ == "__main__":
|
| 343 |
-
asyncio.run(main())
|
|
|
|
| 2 |
import json
|
| 3 |
from typing import Dict, Optional, List
|
| 4 |
from playwright.async_api import async_playwright
|
| 5 |
+
from playwright_stealth import Stealth
|
| 6 |
+
import random
|
| 7 |
|
| 8 |
class PassportIndexVisaScraper:
|
| 9 |
def __init__(self, debug: bool = True):
|
|
|
|
| 19 |
self.browser = None
|
| 20 |
self.context = None
|
| 21 |
self.page = None
|
| 22 |
+
self.playwright = None
|
| 23 |
+
self.stealth_manager = None
|
| 24 |
|
| 25 |
async def __aenter__(self):
|
| 26 |
"""Initialize browser with stealth mode"""
|
| 27 |
+
# Initialize Playwright with Stealth using context manager pattern
|
| 28 |
+
self.stealth_manager = Stealth().use_async(async_playwright())
|
| 29 |
+
self.playwright = await self.stealth_manager.__aenter__()
|
| 30 |
+
|
| 31 |
+
# Random viewport for more natural appearance
|
| 32 |
+
viewport_width = random.randint(1366, 1920)
|
| 33 |
+
viewport_height = random.randint(768, 1080)
|
| 34 |
|
| 35 |
+
# Launch browser in headless mode with stealth settings
|
| 36 |
self.browser = await self.playwright.chromium.launch(
|
| 37 |
+
headless=True, # Using headless mode as requested
|
| 38 |
args=[
|
| 39 |
'--disable-blink-features=AutomationControlled',
|
| 40 |
'--disable-dev-shm-usage',
|
| 41 |
'--no-sandbox',
|
| 42 |
'--disable-setuid-sandbox',
|
| 43 |
+
f'--window-size={viewport_width},{viewport_height}',
|
| 44 |
+
'--disable-features=IsolateOrigins,site-per-process',
|
| 45 |
+
'--enable-features=NetworkService,NetworkServiceInProcess',
|
| 46 |
+
'--disable-infobars',
|
| 47 |
+
'--hide-scrollbars',
|
| 48 |
+
'--mute-audio',
|
| 49 |
+
'--disable-background-timer-throttling',
|
| 50 |
+
'--disable-renderer-backgrounding',
|
| 51 |
+
'--disable-features=TranslateUI',
|
| 52 |
+
'--disable-ipc-flooding-protection',
|
| 53 |
+
'--force-color-profile=srgb',
|
| 54 |
+
'--metrics-recording-only',
|
| 55 |
+
'--disable-hang-monitor',
|
| 56 |
+
'--disable-popup-blocking',
|
| 57 |
+
'--disable-prompt-on-repost',
|
| 58 |
+
'--disable-background-networking',
|
| 59 |
+
'--disable-breakpad',
|
| 60 |
+
'--disable-client-side-phishing-detection',
|
| 61 |
+
'--disable-component-extensions-with-background-pages',
|
| 62 |
+
'--disable-default-apps',
|
| 63 |
+
'--disable-extensions',
|
| 64 |
+
'--disable-features=TranslateUI,BlinkGenPropertyTrees',
|
| 65 |
+
'--disable-sync',
|
| 66 |
'--disable-web-security',
|
| 67 |
+
'--disable-site-isolation-trials',
|
| 68 |
+
'--no-first-run',
|
| 69 |
+
'--no-default-browser-check',
|
| 70 |
+
'--no-pings'
|
| 71 |
]
|
| 72 |
)
|
| 73 |
+
|
| 74 |
+
# Rotate user agents
|
| 75 |
+
user_agents = [
|
| 76 |
+
'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36',
|
| 77 |
+
'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36',
|
| 78 |
+
'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36',
|
| 79 |
+
'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36'
|
| 80 |
+
]
|
| 81 |
|
| 82 |
# Create context with realistic settings
|
| 83 |
self.context = await self.browser.new_context(
|
| 84 |
+
viewport={'width': viewport_width, 'height': viewport_height},
|
| 85 |
+
user_agent=random.choice(user_agents),
|
| 86 |
locale='en-US',
|
| 87 |
+
timezone_id='America/New_York',
|
| 88 |
+
accept_downloads=False,
|
| 89 |
+
has_touch=False,
|
| 90 |
+
is_mobile=False,
|
| 91 |
+
device_scale_factor=1,
|
| 92 |
+
bypass_csp=True,
|
| 93 |
+
ignore_https_errors=True,
|
| 94 |
+
java_script_enabled=True
|
| 95 |
)
|
| 96 |
|
| 97 |
self.page = await self.context.new_page()
|
| 98 |
|
| 99 |
+
# Add additional stealth measures
|
| 100 |
await self.page.add_init_script("""
|
| 101 |
+
// Additional evasion beyond what stealth plugin provides
|
| 102 |
+
delete Object.getPrototypeOf(navigator).webdriver;
|
| 103 |
+
|
| 104 |
+
// Add some randomness to make fingerprint unique
|
| 105 |
+
const battery = navigator.getBattery;
|
| 106 |
+
if (battery) {
|
| 107 |
+
navigator.getBattery = async () => ({
|
| 108 |
+
charging: Math.random() > 0.5,
|
| 109 |
+
chargingTime: Math.floor(Math.random() * 10000),
|
| 110 |
+
dischargingTime: Math.floor(Math.random() * 10000),
|
| 111 |
+
level: Math.random()
|
| 112 |
+
});
|
| 113 |
+
}
|
|
|
|
|
|
|
|
|
|
|
|
|
| 114 |
""")
|
| 115 |
|
| 116 |
if self.debug:
|
| 117 |
+
print("π Browser initialized with stealth mode (headless)")
|
| 118 |
|
| 119 |
return self
|
| 120 |
|
|
|
|
| 126 |
await self.context.close()
|
| 127 |
if self.browser:
|
| 128 |
await self.browser.close()
|
| 129 |
+
if hasattr(self, 'stealth_manager') and self.stealth_manager:
|
| 130 |
+
await self.stealth_manager.__aexit__(exc_type, exc_val, exc_tb)
|
| 131 |
|
| 132 |
if self.debug:
|
| 133 |
print("π Browser closed")
|
|
|
|
| 136 |
"""
|
| 137 |
Navigate to the website and wait for it to load properly
|
| 138 |
"""
|
| 139 |
+
max_retries = 3
|
| 140 |
+
for attempt in range(max_retries):
|
|
|
|
|
|
|
|
|
|
| 141 |
try:
|
| 142 |
+
if self.debug:
|
| 143 |
+
print(f"π± Initializing session... (Attempt {attempt + 1}/{max_retries})")
|
| 144 |
+
|
| 145 |
+
# Add random delay to appear more human
|
| 146 |
+
if attempt > 0:
|
| 147 |
+
await self.page.wait_for_timeout(random.randint(2000, 5000))
|
| 148 |
+
|
| 149 |
+
# Navigate to the page with proper wait
|
| 150 |
response = await self.page.goto(
|
| 151 |
self.base_url,
|
| 152 |
+
wait_until='networkidle',
|
| 153 |
timeout=30000
|
| 154 |
)
|
| 155 |
+
|
| 156 |
+
# Check for Cloudflare challenge
|
| 157 |
+
page_content = await self.page.content()
|
| 158 |
+
if "Just a moment" in page_content or "cf-challenge" in page_content:
|
| 159 |
+
if self.debug:
|
| 160 |
+
print("β³ Cloudflare challenge detected, waiting...")
|
| 161 |
+
|
| 162 |
+
# Wait for Cloudflare to complete
|
| 163 |
+
await self.page.wait_for_timeout(15000)
|
| 164 |
+
|
| 165 |
+
# Try to wait for actual content
|
| 166 |
+
try:
|
| 167 |
+
await self.page.wait_for_selector('.vch-table', state='visible', timeout=20000)
|
| 168 |
+
except:
|
| 169 |
+
if attempt < max_retries - 1:
|
| 170 |
+
continue
|
| 171 |
+
else:
|
| 172 |
+
if self.debug:
|
| 173 |
+
print("β οΈ Cloudflare challenge persists")
|
| 174 |
+
# Try to reload the page
|
| 175 |
+
await self.page.reload(wait_until='networkidle', timeout=30000)
|
| 176 |
+
await self.page.wait_for_timeout(5000)
|
| 177 |
+
|
| 178 |
+
|
| 179 |
+
# Wait for page to be ready - try multiple selectors
|
| 180 |
+
try:
|
| 181 |
+
# Try to wait for any of these elements that indicate page is loaded
|
| 182 |
+
await self.page.wait_for_selector('.vch-table', state='visible', timeout=15000)
|
| 183 |
+
except:
|
| 184 |
+
# If wait fails, just continue after a delay
|
| 185 |
+
if self.debug:
|
| 186 |
+
print("β οΈ Could not find expected elements, continuing anyway...")
|
| 187 |
+
await self.page.wait_for_timeout(3000)
|
| 188 |
|
| 189 |
+
# Close/hide the virtual keyboard if it exists
|
| 190 |
+
await self.close_virtual_keyboard()
|
| 191 |
+
|
| 192 |
+
# Get the session value from the page
|
| 193 |
cl_value = await self.page.evaluate("""
|
| 194 |
() => {
|
| 195 |
const clInput = document.querySelector('#cl');
|
| 196 |
+
// Also check if the page has the visa checker elements
|
| 197 |
+
const hasVchTable = document.querySelector('.vch-table') !== null;
|
| 198 |
+
const hasPassSelect = document.querySelector('.vch-select-pass') !== null;
|
| 199 |
+
|
| 200 |
+
return {
|
| 201 |
+
cl: clInput ? clInput.value : null,
|
| 202 |
+
hasTable: hasVchTable,
|
| 203 |
+
hasSelectors: hasPassSelect,
|
| 204 |
+
url: window.location.href
|
| 205 |
+
};
|
| 206 |
}
|
| 207 |
""")
|
| 208 |
|
| 209 |
if self.debug:
|
| 210 |
+
if cl_value and cl_value.get('cl'):
|
| 211 |
+
print(f"β
Page loaded, session ID: {cl_value['cl']}")
|
| 212 |
+
print(f" Table present: {cl_value['hasTable']}, Selectors present: {cl_value['hasSelectors']}")
|
| 213 |
+
print(f" URL: {cl_value['url']}")
|
| 214 |
+
else:
|
| 215 |
+
print(f"β οΈ Page loaded but missing elements: {cl_value}")
|
| 216 |
+
# If elements are missing, try one more time with a longer wait
|
| 217 |
+
if not cl_value.get('hasTable'):
|
| 218 |
+
await self.page.wait_for_timeout(10000)
|
| 219 |
+
cl_value = await self.page.evaluate("""
|
| 220 |
+
() => {
|
| 221 |
+
const clInput = document.querySelector('#cl');
|
| 222 |
+
const hasVchTable = document.querySelector('.vch-table') !== null;
|
| 223 |
+
const hasPassSelect = document.querySelector('.vch-select-pass') !== null;
|
| 224 |
+
return {
|
| 225 |
+
cl: clInput ? clInput.value : null,
|
| 226 |
+
hasTable: hasVchTable,
|
| 227 |
+
hasSelectors: hasPassSelect,
|
| 228 |
+
url: window.location.href
|
| 229 |
+
};
|
| 230 |
+
}
|
| 231 |
+
""")
|
| 232 |
+
if cl_value.get('hasTable'):
|
| 233 |
+
print(f"β
Elements loaded after additional wait")
|
| 234 |
+
else:
|
| 235 |
+
# Continue anyway, might still work
|
| 236 |
+
pass
|
| 237 |
+
|
| 238 |
return True
|
| 239 |
+
|
| 240 |
except Exception as e:
|
| 241 |
+
if attempt == max_retries - 1:
|
| 242 |
+
print(f"β Error initializing session after {max_retries} attempts: {e}")
|
| 243 |
+
return False
|
| 244 |
+
else:
|
| 245 |
+
if self.debug:
|
| 246 |
+
print(f"β οΈ Attempt {attempt + 1} failed: {e}")
|
| 247 |
+
continue
|
| 248 |
+
return False
|
| 249 |
+
|
| 250 |
+
async def close_virtual_keyboard(self):
|
| 251 |
+
"""Close the virtual keyboard if it's open"""
|
| 252 |
+
try:
|
| 253 |
+
# Check if virtual keyboard exists and is visible
|
| 254 |
+
keyboard_visible = await self.page.evaluate("""
|
| 255 |
+
() => {
|
| 256 |
+
const keyboard = document.querySelector('#PI-VirtualKeyboard');
|
| 257 |
+
if (keyboard) {
|
| 258 |
+
// Hide the keyboard
|
| 259 |
+
keyboard.style.display = 'none';
|
| 260 |
+
keyboard.classList.remove('kioskboard-opened');
|
| 261 |
+
// Also hide any overlay elements
|
| 262 |
+
const overlays = document.querySelectorAll('.kioskboard-wrapper, .kioskboard-overlay');
|
| 263 |
+
overlays.forEach(el => el.style.display = 'none');
|
| 264 |
+
return true;
|
| 265 |
+
}
|
| 266 |
+
return false;
|
| 267 |
+
}
|
| 268 |
+
""")
|
| 269 |
+
|
| 270 |
+
if keyboard_visible and self.debug:
|
| 271 |
+
print("β¨οΈ Virtual keyboard hidden")
|
| 272 |
|
| 273 |
except Exception as e:
|
| 274 |
+
if self.debug:
|
| 275 |
+
print(f"β οΈ Could not hide virtual keyboard: {e}")
|
| 276 |
|
| 277 |
+
async def check_visa_requirement_api(self, passport_country: str, destination_country: str) -> Optional[Dict]:
|
| 278 |
"""
|
| 279 |
+
Check visa requirements using direct API call with cookies
|
| 280 |
|
| 281 |
Args:
|
| 282 |
passport_country: Two-letter country code for passport
|
|
|
|
| 287 |
"""
|
| 288 |
try:
|
| 289 |
if self.debug:
|
| 290 |
+
print(f"π Checking {passport_country.upper()} β {destination_country.upper()} via API")
|
| 291 |
+
|
| 292 |
+
# Get cookies from the page
|
| 293 |
+
cookies = await self.context.cookies()
|
| 294 |
|
| 295 |
+
# Get the session ID
|
| 296 |
cl_value = await self.page.evaluate("""
|
| 297 |
() => {
|
| 298 |
const clInput = document.querySelector('#cl');
|
|
|
|
| 300 |
}
|
| 301 |
""")
|
| 302 |
|
| 303 |
+
# Make the API request with proper cookies and headers
|
| 304 |
result = await self.page.evaluate("""
|
| 305 |
async (args) => {
|
| 306 |
const [passport, destination, sessionId] = args;
|
| 307 |
+
|
| 308 |
+
// Create form data
|
| 309 |
const formData = new URLSearchParams();
|
| 310 |
+
formData.append('d', destination.toLowerCase());
|
| 311 |
+
formData.append('s', passport.toLowerCase());
|
| 312 |
formData.append('cl', sessionId);
|
| 313 |
|
| 314 |
try {
|
|
|
|
| 317 |
headers: {
|
| 318 |
'Content-Type': 'application/x-www-form-urlencoded; charset=UTF-8',
|
| 319 |
'X-Requested-With': 'XMLHttpRequest',
|
| 320 |
+
'Accept': 'application/json, text/javascript, */*; q=0.01',
|
| 321 |
+
'Origin': 'https://www.passportindex.org',
|
| 322 |
+
'Referer': 'https://www.passportindex.org/travel-visa-checker/'
|
| 323 |
},
|
| 324 |
body: formData.toString(),
|
| 325 |
+
credentials: 'include',
|
| 326 |
+
mode: 'cors'
|
| 327 |
});
|
| 328 |
|
| 329 |
+
const responseText = await response.text();
|
|
|
|
|
|
|
| 330 |
|
| 331 |
+
// Try to parse as JSON
|
| 332 |
+
try {
|
| 333 |
+
return JSON.parse(responseText);
|
| 334 |
+
} catch {
|
| 335 |
+
// Check if it's a Cloudflare challenge
|
| 336 |
+
if (responseText.includes("Just a moment") ||
|
| 337 |
+
responseText.includes("cf-challenge") ||
|
| 338 |
+
response.status === 403) {
|
| 339 |
+
return { error: 'Cloudflare block', status: response.status };
|
| 340 |
+
}
|
| 341 |
+
// Otherwise it might be a different error
|
| 342 |
+
return { error: 'Invalid response', status: response.status };
|
| 343 |
+
}
|
| 344 |
} catch (error) {
|
| 345 |
return { error: error.message };
|
| 346 |
}
|
| 347 |
}
|
| 348 |
+
""", [passport_country, destination_country, cl_value])
|
| 349 |
|
| 350 |
+
if result and 'error' not in result and 'text' in result:
|
| 351 |
if self.debug:
|
| 352 |
+
print(f"β
Got API result: {json.dumps(result)}")
|
| 353 |
return result
|
| 354 |
elif result and 'error' in result:
|
| 355 |
+
if self.debug:
|
| 356 |
+
print(f"β οΈ API Error: {result['error']}")
|
| 357 |
return None
|
| 358 |
else:
|
| 359 |
return None
|
| 360 |
|
| 361 |
except Exception as e:
|
| 362 |
+
if self.debug:
|
| 363 |
+
print(f"β Error checking visa requirement via API: {e}")
|
| 364 |
return None
|
| 365 |
|
| 366 |
+
async def check_visa_interactive_v2(self, passport_country: str, destination_country: str) -> Optional[Dict]:
|
| 367 |
"""
|
| 368 |
+
Alternative interactive method with better element handling
|
| 369 |
"""
|
| 370 |
try:
|
| 371 |
if self.debug:
|
| 372 |
+
print(f"π±οΈ Using interactive method v2 for {passport_country.upper()} β {destination_country.upper()}")
|
| 373 |
|
| 374 |
+
# First, ensure virtual keyboard is hidden
|
| 375 |
+
await self.close_virtual_keyboard()
|
|
|
|
| 376 |
|
| 377 |
+
# Clear any previous selections via JavaScript
|
| 378 |
+
await self.page.evaluate("""
|
| 379 |
+
() => {
|
| 380 |
+
// Reset the form
|
| 381 |
+
const passSelect = document.querySelector('.vch-select-pass');
|
| 382 |
+
const destSelect = document.querySelector('.vch-select-des');
|
| 383 |
+
if (passSelect) passSelect.textContent = 'Select Passport';
|
| 384 |
+
if (destSelect) destSelect.textContent = 'Select Destination';
|
| 385 |
+
|
| 386 |
+
// Hide any open dropdowns
|
| 387 |
+
const passports = document.querySelector('.vch-passports');
|
| 388 |
+
const destinations = document.querySelector('.vch-destinations');
|
| 389 |
+
if (passports) passports.classList.remove('opener');
|
| 390 |
+
if (destinations) destinations.classList.remove('opener');
|
| 391 |
+
}
|
| 392 |
+
""")
|
| 393 |
|
|
|
|
|
|
|
| 394 |
await self.page.wait_for_timeout(500)
|
| 395 |
|
| 396 |
+
# Select passport using JavaScript
|
| 397 |
+
passport_selected = await self.page.evaluate("""
|
| 398 |
+
(countryCode) => {
|
| 399 |
+
// Click the passport selector
|
| 400 |
+
const passSelect = document.querySelector('.vch-select-pass');
|
| 401 |
+
if (!passSelect) return false;
|
| 402 |
+
passSelect.click();
|
| 403 |
+
|
| 404 |
+
// Wait a bit for dropdown to open
|
| 405 |
+
setTimeout(() => {
|
| 406 |
+
// Find and click the country
|
| 407 |
+
const countryElement = document.querySelector(`.vch-passports .s-div[data-ccode="${countryCode.toLowerCase()}"]`);
|
| 408 |
+
if (countryElement) {
|
| 409 |
+
countryElement.click();
|
| 410 |
+
return true;
|
| 411 |
+
}
|
| 412 |
+
}, 300);
|
| 413 |
+
|
| 414 |
+
return false;
|
| 415 |
+
}
|
| 416 |
+
""", passport_country)
|
| 417 |
+
|
| 418 |
await self.page.wait_for_timeout(1000)
|
| 419 |
|
| 420 |
+
# Select destination using JavaScript
|
| 421 |
+
destination_selected = await self.page.evaluate("""
|
| 422 |
+
(countryCode) => {
|
| 423 |
+
// Click the destination selector
|
| 424 |
+
const destSelect = document.querySelector('.vch-select-des');
|
| 425 |
+
if (!destSelect) return false;
|
| 426 |
+
destSelect.click();
|
| 427 |
+
|
| 428 |
+
// Wait a bit for dropdown to open
|
| 429 |
+
setTimeout(() => {
|
| 430 |
+
// Find and click the country
|
| 431 |
+
const countryElement = document.querySelector(`.vch-destinations .s-div[data-ccode="${countryCode.toLowerCase()}"]`);
|
| 432 |
+
if (countryElement) {
|
| 433 |
+
countryElement.click();
|
| 434 |
+
return true;
|
| 435 |
+
}
|
| 436 |
+
}, 300);
|
| 437 |
+
|
| 438 |
+
return false;
|
| 439 |
+
}
|
| 440 |
+
""", destination_country)
|
| 441 |
+
|
| 442 |
+
await self.page.wait_for_timeout(2000)
|
| 443 |
+
|
| 444 |
+
# Get the result
|
| 445 |
result = await self.page.evaluate("""
|
| 446 |
() => {
|
| 447 |
const resultElement = document.querySelector('.vch-result');
|
| 448 |
if (resultElement) {
|
| 449 |
const text = resultElement.querySelector('.text');
|
| 450 |
const days = resultElement.querySelector('.days');
|
| 451 |
+
const passSelect = document.querySelector('.vch-select-pass');
|
| 452 |
+
const destSelect = document.querySelector('.vch-select-des');
|
| 453 |
+
|
| 454 |
return {
|
| 455 |
+
text: text ? text.textContent.trim() : '',
|
| 456 |
+
days: days ? days.textContent.trim() : '',
|
| 457 |
+
pass: passSelect ? passSelect.textContent.trim() : '',
|
| 458 |
+
dest: destSelect ? destSelect.textContent.trim() : ''
|
| 459 |
};
|
| 460 |
}
|
| 461 |
return null;
|
| 462 |
}
|
| 463 |
""")
|
| 464 |
|
| 465 |
+
if result and result['text']:
|
| 466 |
+
return result
|
| 467 |
+
|
| 468 |
+
return None
|
| 469 |
+
|
| 470 |
+
except Exception as e:
|
| 471 |
+
if self.debug:
|
| 472 |
+
print(f"β Interactive method v2 failed: {e}")
|
| 473 |
+
return None
|
| 474 |
+
|
| 475 |
+
async def check_visa_direct_manipulation(self, passport_country: str, destination_country: str) -> Optional[Dict]:
|
| 476 |
+
"""
|
| 477 |
+
Direct manipulation method - directly trigger the visa check
|
| 478 |
+
"""
|
| 479 |
+
try:
|
| 480 |
+
if self.debug:
|
| 481 |
+
print(f"π§ Using direct manipulation for {passport_country.upper()} β {destination_country.upper()}")
|
| 482 |
+
|
| 483 |
+
# Directly call the visa checker function if it exists
|
| 484 |
+
result = await self.page.evaluate("""
|
| 485 |
+
async (args) => {
|
| 486 |
+
const [passport, destination] = args;
|
| 487 |
+
// Try to find and call the visa checker function directly
|
| 488 |
+
if (typeof visaChecker !== 'undefined' && typeof visaChecker.check === 'function') {
|
| 489 |
+
return await visaChecker.check(passport, destination);
|
| 490 |
+
}
|
| 491 |
+
|
| 492 |
+
// Alternative: manually trigger the check
|
| 493 |
+
const cl = document.querySelector('#cl')?.value || 'bc2140a2d83928ce1112d01e610bad89';
|
| 494 |
+
|
| 495 |
+
// Set the selections visually
|
| 496 |
+
const passSelect = document.querySelector('.vch-select-pass');
|
| 497 |
+
const destSelect = document.querySelector('.vch-select-des');
|
| 498 |
+
|
| 499 |
+
if (passSelect && destSelect) {
|
| 500 |
+
// Find country names
|
| 501 |
+
const passportElement = document.querySelector(`.vch-passports .s-div[data-ccode="${passport}"] .cname`);
|
| 502 |
+
const destElement = document.querySelector(`.vch-destinations .s-div[data-ccode="${destination}"] .cname`);
|
| 503 |
+
|
| 504 |
+
if (passportElement) passSelect.textContent = passportElement.textContent;
|
| 505 |
+
if (destElement) destSelect.textContent = destElement.textContent;
|
| 506 |
+
|
| 507 |
+
// Make the AJAX call
|
| 508 |
+
return new Promise((resolve) => {
|
| 509 |
+
$.ajax({
|
| 510 |
+
url: 'https://www.passportindex.org/core/visachecker.php',
|
| 511 |
+
type: 'POST',
|
| 512 |
+
data: {
|
| 513 |
+
d: destination,
|
| 514 |
+
s: passport,
|
| 515 |
+
cl: cl
|
| 516 |
+
},
|
| 517 |
+
success: function(response) {
|
| 518 |
+
// Update the UI
|
| 519 |
+
if (response.text) {
|
| 520 |
+
const resultElement = document.querySelector('.vch-result');
|
| 521 |
+
if (resultElement) {
|
| 522 |
+
resultElement.className = 'vch-result tv-grey ' + (response.vr ? 'tvvr' : 'tvvf');
|
| 523 |
+
const textElement = resultElement.querySelector('.text');
|
| 524 |
+
if (textElement) textElement.textContent = response.text;
|
| 525 |
+
}
|
| 526 |
+
}
|
| 527 |
+
resolve(response);
|
| 528 |
+
},
|
| 529 |
+
error: function(xhr) {
|
| 530 |
+
resolve({ error: 'Request failed', status: xhr.status });
|
| 531 |
+
}
|
| 532 |
+
});
|
| 533 |
+
});
|
| 534 |
+
}
|
| 535 |
+
|
| 536 |
+
return null;
|
| 537 |
+
}
|
| 538 |
+
""", [passport_country.lower(), destination_country.lower()])
|
| 539 |
+
|
| 540 |
return result
|
| 541 |
|
| 542 |
except Exception as e:
|
| 543 |
if self.debug:
|
| 544 |
+
print(f"β Direct manipulation failed: {e}")
|
| 545 |
return None
|
| 546 |
|
| 547 |
+
async def check_visa_requirement(self, passport_country: str, destination_country: str) -> Optional[Dict]:
|
| 548 |
+
"""
|
| 549 |
+
Main method to check visa requirements - tries multiple approaches
|
| 550 |
+
"""
|
| 551 |
+
# Try API method first
|
| 552 |
+
result = await self.check_visa_requirement_api(passport_country, destination_country)
|
| 553 |
+
if result:
|
| 554 |
+
return result
|
| 555 |
+
|
| 556 |
+
# Try direct manipulation
|
| 557 |
+
result = await self.check_visa_direct_manipulation(passport_country, destination_country)
|
| 558 |
+
if result:
|
| 559 |
+
return result
|
| 560 |
+
|
| 561 |
+
# Try interactive method v2
|
| 562 |
+
result = await self.check_visa_interactive_v2(passport_country, destination_country)
|
| 563 |
+
if result:
|
| 564 |
+
return result
|
| 565 |
+
|
| 566 |
+
return None
|
| 567 |
+
|
| 568 |
async def check_multiple_destinations(self, passport_country: str, destinations: List[str], delay: float = 2.0) -> Dict:
|
| 569 |
"""
|
| 570 |
Check visa requirements for multiple destinations
|
|
|
|
| 582 |
for i, dest in enumerate(destinations, 1):
|
| 583 |
print(f"\n[{i}/{len(destinations)}] Checking {passport_country.upper()} β {dest.upper()}...")
|
| 584 |
|
| 585 |
+
result = await self.check_visa_requirement(passport_country, dest)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 586 |
|
| 587 |
if result:
|
| 588 |
results[dest] = result
|
|
|
|
| 604 |
if not result:
|
| 605 |
return "No information available"
|
| 606 |
|
| 607 |
+
# Return clean JSON format
|
| 608 |
+
return json.dumps({
|
| 609 |
+
'text': result.get('text', 'N/A'),
|
| 610 |
+
'col': result.get('col', ''),
|
| 611 |
+
'link': result.get('link', 0),
|
| 612 |
+
'dest': result.get('dest', '').upper(),
|
| 613 |
+
'pass': result.get('pass', '').upper()
|
| 614 |
+
})
|
| 615 |
|
| 616 |
|
| 617 |
async def main():
|
| 618 |
"""Main function to demonstrate usage"""
|
| 619 |
print("="*60)
|
| 620 |
+
print(" Passport Index Visa Checker (Playwright - Fixed)")
|
| 621 |
print("="*60)
|
| 622 |
|
| 623 |
async with PassportIndexVisaScraper(debug=True) as scraper:
|
|
|
|
| 633 |
# Test single visa requirement
|
| 634 |
print("\nπ Single visa check: US β GB")
|
| 635 |
print("-" * 40)
|
| 636 |
+
result = await scraper.check_visa_requirement('us', 'gb')
|
| 637 |
if result:
|
| 638 |
+
print(f"Result: {json.dumps(result)}")
|
| 639 |
else:
|
| 640 |
+
print("β Failed to get result")
|
|
|
|
|
|
|
|
|
|
| 641 |
|
| 642 |
# Test multiple destinations
|
| 643 |
print("\nπ Multiple destinations for US passport:")
|
|
|
|
| 648 |
print("\nπ Summary:")
|
| 649 |
for dest, result in results.items():
|
| 650 |
if result:
|
| 651 |
+
print(f" β
{dest.upper()}: {json.dumps(result)}")
|
| 652 |
else:
|
| 653 |
print(f" β US β {dest.upper()}: Failed")
|
| 654 |
|
| 655 |
if __name__ == "__main__":
|
| 656 |
+
asyncio.run(main())
|