Spaces:
Paused
Paused
Update scraper.py
Browse files- scraper.py +7 -15
scraper.py
CHANGED
|
@@ -10,20 +10,14 @@ USER_AGENTS = [
|
|
| 10 |
]
|
| 11 |
|
| 12 |
async def check_fragment(word: str) -> str:
|
| 13 |
-
"""
|
| 14 |
-
Checks Fragment.com using exponential backoff to avoid 429 Rate Limits.
|
| 15 |
-
Returns: AVAILABLE, TAKEN, FOR_SALE, ON_AUCTION, SOLD, or UNAVAILABLE
|
| 16 |
-
"""
|
| 17 |
url = f"https://fragment.com/username/{word}"
|
| 18 |
|
| 19 |
-
# Create the session inside the function so bot.py can call it easily
|
| 20 |
async with aiohttp.ClientSession() as session:
|
| 21 |
for _ in range(6):
|
| 22 |
for attempt in range(3):
|
| 23 |
headers = {"User-Agent": random.choice(USER_AGENTS)}
|
| 24 |
try:
|
| 25 |
async with session.get(url, headers=headers, allow_redirects=True, timeout=15) as resp:
|
| 26 |
-
# Handle Telegram Rate Limiting (Exponential Backoff)
|
| 27 |
if resp.status == 429:
|
| 28 |
wait = (2 ** attempt * 3) + random.uniform(1, 3)
|
| 29 |
await asyncio.sleep(wait)
|
|
@@ -31,9 +25,8 @@ async def check_fragment(word: str) -> str:
|
|
| 31 |
|
| 32 |
if resp.status == 200:
|
| 33 |
html = await resp.text()
|
| 34 |
-
await asyncio.sleep(random.uniform(0.15, 0.6))
|
| 35 |
|
| 36 |
-
# Integrated Parser (Replaces external parser_html)
|
| 37 |
if "tm-status-avail" in html:
|
| 38 |
return "AVAILABLE"
|
| 39 |
|
|
@@ -42,24 +35,23 @@ async def check_fragment(word: str) -> str:
|
|
| 42 |
return "ON_AUCTION"
|
| 43 |
if "For sale" in html:
|
| 44 |
return "FOR_SALE"
|
|
|
|
|
|
|
|
|
|
| 45 |
return "TAKEN"
|
| 46 |
|
| 47 |
if "tm-status-unavail" in html:
|
| 48 |
return "UNAVAILABLE"
|
| 49 |
-
|
| 50 |
-
if "Sold" in html:
|
| 51 |
-
return "SOLD"
|
| 52 |
|
| 53 |
-
#
|
| 54 |
if "Available" in html: return "AVAILABLE"
|
| 55 |
if "Taken" in html: return "TAKEN"
|
| 56 |
|
| 57 |
return "UNAVAILABLE"
|
| 58 |
|
| 59 |
except Exception:
|
| 60 |
-
pass
|
| 61 |
|
| 62 |
-
# Deep sleep before trying the outer loop again
|
| 63 |
await asyncio.sleep(random.uniform(0.8, 2.0))
|
| 64 |
|
| 65 |
-
return "UNAVAILABLE"
|
|
|
|
| 10 |
]
|
| 11 |
|
| 12 |
async def check_fragment(word: str) -> str:
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
url = f"https://fragment.com/username/{word}"
|
| 14 |
|
|
|
|
| 15 |
async with aiohttp.ClientSession() as session:
|
| 16 |
for _ in range(6):
|
| 17 |
for attempt in range(3):
|
| 18 |
headers = {"User-Agent": random.choice(USER_AGENTS)}
|
| 19 |
try:
|
| 20 |
async with session.get(url, headers=headers, allow_redirects=True, timeout=15) as resp:
|
|
|
|
| 21 |
if resp.status == 429:
|
| 22 |
wait = (2 ** attempt * 3) + random.uniform(1, 3)
|
| 23 |
await asyncio.sleep(wait)
|
|
|
|
| 25 |
|
| 26 |
if resp.status == 200:
|
| 27 |
html = await resp.text()
|
| 28 |
+
await asyncio.sleep(random.uniform(0.15, 0.6))
|
| 29 |
|
|
|
|
| 30 |
if "tm-status-avail" in html:
|
| 31 |
return "AVAILABLE"
|
| 32 |
|
|
|
|
| 35 |
return "ON_AUCTION"
|
| 36 |
if "For sale" in html:
|
| 37 |
return "FOR_SALE"
|
| 38 |
+
# CRITICAL FIX: Looks for the specific sold text, avoiding the nav bar
|
| 39 |
+
if ">Sold for" in html or ">Sold at" in html or 'class="tm-status-text">Sold' in html:
|
| 40 |
+
return "SOLD"
|
| 41 |
return "TAKEN"
|
| 42 |
|
| 43 |
if "tm-status-unavail" in html:
|
| 44 |
return "UNAVAILABLE"
|
|
|
|
|
|
|
|
|
|
| 45 |
|
| 46 |
+
# Fallbacks
|
| 47 |
if "Available" in html: return "AVAILABLE"
|
| 48 |
if "Taken" in html: return "TAKEN"
|
| 49 |
|
| 50 |
return "UNAVAILABLE"
|
| 51 |
|
| 52 |
except Exception:
|
| 53 |
+
pass
|
| 54 |
|
|
|
|
| 55 |
await asyncio.sleep(random.uniform(0.8, 2.0))
|
| 56 |
|
| 57 |
+
return "UNAVAILABLE"
|