Spaces:
Sleeping
Sleeping
Lcmind
commited on
Commit
Β·
a98a453
1
Parent(s):
ca03dab
fix: S-tier prompt - hex to color name, blur UI text, remove watermarks
Browse files- app/services/screenshot.py +44 -9
app/services/screenshot.py
CHANGED
|
@@ -41,6 +41,14 @@ async def capture_screenshot(url: str) -> tuple[str, str]:
|
|
| 41 |
if not url.startswith(('http://', 'https://')):
|
| 42 |
url = f'https://{url}'
|
| 43 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 44 |
# Set environment variables for Pyppeteer
|
| 45 |
os.environ['PYPPETEER_CHROMIUM_REVISION'] = '1181205'
|
| 46 |
os.environ['PYPPETEER_EXECUTABLE_PATH'] = settings.puppeteer_executable_path
|
|
@@ -75,19 +83,46 @@ async def capture_screenshot(url: str) -> tuple[str, str]:
|
|
| 75 |
|
| 76 |
# === Phase 2: μΉ¨ν¬ (Navigation) ===
|
| 77 |
navigation_success = False
|
|
|
|
|
|
|
|
|
|
| 78 |
try:
|
| 79 |
-
# domcontentloaded: HTML λΌλλ§ λ‘λ©λλ©΄ μ§ν (10μ΄ νμμμ)
|
| 80 |
await page.goto(url, {'waitUntil': 'domcontentloaded', 'timeout': 10000})
|
| 81 |
navigation_success = True
|
| 82 |
except Exception as e:
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 91 |
|
| 92 |
# === Phase 3: μκ° λ°μ΄ν° μμ§ (Screenshot) ===
|
| 93 |
screenshot_path = '/tmp/screenshot.png'
|
|
|
|
| 41 |
if not url.startswith(('http://', 'https://')):
|
| 42 |
url = f'https://{url}'
|
| 43 |
|
| 44 |
+
# URL μ κ·ν: www μλ μΆκ° μλ (DNS ν΄μ μ€ν¨ λ°©μ§)
|
| 45 |
+
original_url = url
|
| 46 |
+
if 'www.' not in url:
|
| 47 |
+
# instagram.com -> www.instagram.com
|
| 48 |
+
url_with_www = url.replace('://', '://www.', 1)
|
| 49 |
+
else:
|
| 50 |
+
url_with_www = None
|
| 51 |
+
|
| 52 |
# Set environment variables for Pyppeteer
|
| 53 |
os.environ['PYPPETEER_CHROMIUM_REVISION'] = '1181205'
|
| 54 |
os.environ['PYPPETEER_EXECUTABLE_PATH'] = settings.puppeteer_executable_path
|
|
|
|
| 83 |
|
| 84 |
# === Phase 2: μΉ¨ν¬ (Navigation) ===
|
| 85 |
navigation_success = False
|
| 86 |
+
last_error = None
|
| 87 |
+
|
| 88 |
+
# 첫 μλ: μλ³Έ URL
|
| 89 |
try:
|
|
|
|
| 90 |
await page.goto(url, {'waitUntil': 'domcontentloaded', 'timeout': 10000})
|
| 91 |
navigation_success = True
|
| 92 |
except Exception as e:
|
| 93 |
+
last_error = e
|
| 94 |
+
logger.warning(f"Navigation failed for {url}: {e}")
|
| 95 |
+
|
| 96 |
+
# μ¬μλ 1: www μΆκ° λ²μ (DNS ν΄μ μ€ν¨ λμ)
|
| 97 |
+
if url_with_www and url_with_www != url:
|
| 98 |
+
try:
|
| 99 |
+
logger.info(f"Retrying with www: {url_with_www}")
|
| 100 |
+
await page.goto(url_with_www, {'waitUntil': 'domcontentloaded', 'timeout': 10000})
|
| 101 |
+
navigation_success = True
|
| 102 |
+
url = url_with_www # μ±κ³΅ν URLλ‘ μ
λ°μ΄νΈ
|
| 103 |
+
except Exception as e2:
|
| 104 |
+
last_error = e2
|
| 105 |
+
logger.warning(f"www retry also failed: {e2}")
|
| 106 |
+
|
| 107 |
+
# μ¬μλ 2: λ κΈ΄ νμμμ (λλ¦° μ¬μ΄νΈ λμ)
|
| 108 |
+
if not navigation_success:
|
| 109 |
+
try:
|
| 110 |
+
logger.info(f"Final retry with longer timeout: {url}")
|
| 111 |
+
await page.goto(url, {'waitUntil': 'load', 'timeout': 15000})
|
| 112 |
+
navigation_success = True
|
| 113 |
+
except Exception as e3:
|
| 114 |
+
last_error = e3
|
| 115 |
+
|
| 116 |
+
# λͺ¨λ μλ μ€ν¨ μ λͺ
νν μλ¬ λ©μμ§
|
| 117 |
+
if not navigation_success:
|
| 118 |
+
error_msg = str(last_error)
|
| 119 |
+
if 'ERR_NAME_NOT_RESOLVED' in error_msg:
|
| 120 |
+
raise Exception(
|
| 121 |
+
f"DNS ν΄μ μ€ν¨: '{original_url}' λλ©μΈμ μ°Ύμ μ μμ΅λλ€. "
|
| 122 |
+
f"μλ² νκ²½μμ μ΄ μ¬μ΄νΈ μ μμ΄ μ°¨λ¨λμκ±°λ, μλͺ»λ URLμ
λλ€."
|
| 123 |
+
)
|
| 124 |
+
else:
|
| 125 |
+
raise Exception(f"νμ΄μ§ λ‘λ© μ€ν¨: {url}. Error: {error_msg}")
|
| 126 |
|
| 127 |
# === Phase 3: μκ° λ°μ΄ν° μμ§ (Screenshot) ===
|
| 128 |
screenshot_path = '/tmp/screenshot.png'
|