Update app.py
Browse files
app.py
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
-
from fastapi import FastAPI, HTTPException, status, BackgroundTasks
|
| 2 |
-
from fastapi.responses import StreamingResponse
|
| 3 |
import instaloader
|
| 4 |
import requests
|
| 5 |
import os
|
|
@@ -7,12 +7,8 @@ import time
|
|
| 7 |
import logging
|
| 8 |
import random
|
| 9 |
from functools import wraps
|
| 10 |
-
import json
|
| 11 |
from datetime import datetime, timedelta
|
| 12 |
-
|
| 13 |
-
import hashlib
|
| 14 |
-
import asyncio
|
| 15 |
-
from starlette.concurrency import run_in_threadpool
|
| 16 |
|
| 17 |
# Configure logging
|
| 18 |
logging.basicConfig(
|
|
@@ -36,68 +32,83 @@ CACHE_DIR = "/tmp/instagram_cache"
|
|
| 36 |
os.makedirs(SESSION_DIR, exist_ok=True)
|
| 37 |
os.makedirs(CACHE_DIR, exist_ok=True)
|
| 38 |
|
| 39 |
-
#
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
# Cache state
|
| 45 |
-
STORY_CACHE: Dict[str, Dict] = {}
|
| 46 |
-
CACHE_EXPIRY: Dict[str, datetime] = {}
|
| 47 |
-
CACHE_LOCK = asyncio.Lock()
|
| 48 |
|
| 49 |
USER_AGENTS = [
|
| 50 |
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
|
| 51 |
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.6 Safari/605.1.15",
|
| 52 |
-
"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36"
|
| 53 |
-
"Mozilla/5.0 (iPhone; CPU iPhone OS 17_1 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.1 Mobile/15E148 Safari/604.1",
|
| 54 |
-
"Mozilla/5.0 (iPad; CPU OS 17_1 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.1 Mobile/15E148 Safari/604.1"
|
| 55 |
]
|
| 56 |
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
return
|
| 60 |
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
logger.info(f"Cache expired for {username}")
|
| 73 |
-
STORY_CACHE.pop(cache_key, None)
|
| 74 |
-
CACHE_EXPIRY.pop(cache_key, None)
|
| 75 |
-
|
| 76 |
return None
|
| 77 |
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
|
|
|
|
|
|
| 81 |
|
| 82 |
-
|
| 83 |
-
STORY_CACHE[cache_key] = stories_data
|
| 84 |
-
CACHE_EXPIRY[cache_key] = datetime.now() + timedelta(minutes=ttl_minutes)
|
| 85 |
-
|
| 86 |
-
logger.info(f"Cached stories for {username} for {ttl_minutes} minutes")
|
| 87 |
-
|
| 88 |
-
# Save to disk for persistence
|
| 89 |
-
cache_file = os.path.join(CACHE_DIR, f"{cache_key}.json")
|
| 90 |
try:
|
|
|
|
| 91 |
with open(cache_file, 'w') as f:
|
| 92 |
json.dump({
|
| 93 |
-
"data":
|
| 94 |
-
"expires": CACHE_EXPIRY[
|
| 95 |
}, f)
|
|
|
|
| 96 |
except Exception as e:
|
| 97 |
-
logger.warning(f"Failed to save cache
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
|
| 99 |
-
def
|
| 100 |
-
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 101 |
L = instaloader.Instaloader(
|
| 102 |
sleep=True,
|
| 103 |
request_timeout=30,
|
|
@@ -120,382 +131,211 @@ def get_instaloader(force_login=False) -> instaloader.Instaloader:
|
|
| 120 |
)
|
| 121 |
|
| 122 |
try:
|
| 123 |
-
if
|
| 124 |
-
logger.info(f"
|
| 125 |
try:
|
| 126 |
L.load_session_from_file(INSTAGRAM_USERNAME, SESSION_FILE)
|
| 127 |
logger.info("Session loaded successfully")
|
| 128 |
-
# Test session with a lightweight call if possible
|
| 129 |
return L
|
| 130 |
except Exception as e:
|
| 131 |
-
logger.warning(f"Session load failed: {str(e)}
|
| 132 |
-
|
| 133 |
-
# Add delay before login to mimic human behavior
|
| 134 |
-
time.sleep(random.uniform(1.0, 3.0))
|
| 135 |
|
| 136 |
-
logger.info("Starting fresh login
|
|
|
|
| 137 |
L.login(INSTAGRAM_USERNAME, INSTAGRAM_PASSWORD)
|
| 138 |
|
| 139 |
-
# Set realistic headers
|
| 140 |
-
L.context._session.headers.update({
|
| 141 |
-
'Accept-Language': 'en-US,en;q=0.9',
|
| 142 |
-
'Accept-Encoding': 'gzip, deflate, br',
|
| 143 |
-
'Referer': 'https://www.instagram.com/',
|
| 144 |
-
'Origin': 'https://www.instagram.com',
|
| 145 |
-
'DNT': '1',
|
| 146 |
-
'Connection': 'keep-alive',
|
| 147 |
-
'Sec-Fetch-Dest': 'empty',
|
| 148 |
-
'Sec-Fetch-Mode': 'cors',
|
| 149 |
-
'Sec-Fetch-Site': 'same-origin',
|
| 150 |
-
'Pragma': 'no-cache',
|
| 151 |
-
'Cache-Control': 'no-cache',
|
| 152 |
-
})
|
| 153 |
-
|
| 154 |
if SESSION_FILE:
|
| 155 |
L.save_session_to_file(SESSION_FILE)
|
| 156 |
logger.info(f"Saved new session to {SESSION_FILE}")
|
| 157 |
|
| 158 |
-
# Add delay after login
|
| 159 |
-
time.sleep(random.uniform(2.0, 4.0))
|
| 160 |
-
|
| 161 |
return L
|
| 162 |
|
| 163 |
-
except instaloader.exceptions.BadCredentialsException as e:
|
| 164 |
-
logger.error("Invalid credentials: %s", str(e))
|
| 165 |
-
raise HTTPException(
|
| 166 |
-
status_code=status.HTTP_401_UNAUTHORIZED,
|
| 167 |
-
detail="Invalid Instagram credentials"
|
| 168 |
-
)
|
| 169 |
-
except instaloader.exceptions.TwoFactorAuthRequiredException as e:
|
| 170 |
-
logger.error("2FA required: %s", str(e))
|
| 171 |
-
raise HTTPException(
|
| 172 |
-
status_code=status.HTTP_401_UNAUTHORIZED,
|
| 173 |
-
detail="Two-factor authentication required"
|
| 174 |
-
)
|
| 175 |
except Exception as e:
|
| 176 |
-
logger.error("Login failed:
|
| 177 |
raise HTTPException(
|
| 178 |
status_code=status.HTTP_503_SERVICE_UNAVAILABLE,
|
| 179 |
-
detail="Instagram login
|
| 180 |
)
|
| 181 |
|
| 182 |
-
async def is_rate_limited(username: str) -> bool:
|
| 183 |
-
"""Check if a given username is currently rate limited"""
|
| 184 |
-
async with RATE_LIMIT_LOCK:
|
| 185 |
-
# If in cooldown period, check if it's expired
|
| 186 |
-
if username in COOLDOWN_PERIODS:
|
| 187 |
-
if datetime.now() < COOLDOWN_PERIODS[username]:
|
| 188 |
-
remaining = (COOLDOWN_PERIODS[username] - datetime.now()).seconds
|
| 189 |
-
logger.warning(f"User {username} in cooldown for {remaining} more seconds")
|
| 190 |
-
return True
|
| 191 |
-
else:
|
| 192 |
-
# Cooldown expired
|
| 193 |
-
del COOLDOWN_PERIODS[username]
|
| 194 |
-
|
| 195 |
-
# Check time since last request
|
| 196 |
-
if username in LAST_REQUEST_TIME:
|
| 197 |
-
elapsed = (datetime.now() - LAST_REQUEST_TIME[username]).seconds
|
| 198 |
-
min_interval = 60 # Minimum 60 seconds between requests
|
| 199 |
-
if elapsed < min_interval:
|
| 200 |
-
logger.warning(f"Rate limiting {username}: {elapsed}s elapsed, need {min_interval}s")
|
| 201 |
-
return True
|
| 202 |
-
|
| 203 |
-
# Update last request time
|
| 204 |
-
LAST_REQUEST_TIME[username] = datetime.now()
|
| 205 |
-
return False
|
| 206 |
-
|
| 207 |
-
async def handle_rate_limit_error(username: str, retry_after: Optional[int] = None):
|
| 208 |
-
"""Handle rate limit by setting cooldown period"""
|
| 209 |
-
async with RATE_LIMIT_LOCK:
|
| 210 |
-
# Set cooldown period based on retry_after or default logic
|
| 211 |
-
if retry_after:
|
| 212 |
-
# Use server-provided retry duration if available
|
| 213 |
-
cooldown_minutes = max(retry_after / 60, 1) # At least 1 minute
|
| 214 |
-
else:
|
| 215 |
-
# Default exponential backoff
|
| 216 |
-
cooldown_minutes = 5
|
| 217 |
-
if username in COOLDOWN_PERIODS:
|
| 218 |
-
cooldown_minutes = min(cooldown_minutes * 2, 30) # Up to 30 minutes
|
| 219 |
-
|
| 220 |
-
COOLDOWN_PERIODS[username] = datetime.now() + timedelta(minutes=cooldown_minutes)
|
| 221 |
-
logger.warning(f"Setting cooldown for {username} for {cooldown_minutes} minutes")
|
| 222 |
-
|
| 223 |
def handle_instagram_errors(func):
|
| 224 |
@wraps(func)
|
| 225 |
async def wrapper(*args, **kwargs):
|
| 226 |
-
# Extract username
|
| 227 |
-
|
| 228 |
if len(args) > 1 and hasattr(args[1], '__str__'):
|
| 229 |
-
|
| 230 |
|
| 231 |
try:
|
| 232 |
# Check for rate limiting
|
| 233 |
-
if
|
| 234 |
-
|
| 235 |
-
|
| 236 |
-
|
| 237 |
-
|
| 238 |
-
return {**cached_response, "from_cache": True}
|
| 239 |
else:
|
| 240 |
-
# No cache available
|
| 241 |
raise HTTPException(
|
| 242 |
status_code=status.HTTP_429_TOO_MANY_REQUESTS,
|
| 243 |
detail="Rate limit exceeded. Please try again later."
|
| 244 |
)
|
| 245 |
-
|
|
|
|
| 246 |
response = await func(*args, **kwargs)
|
| 247 |
|
| 248 |
-
# Cache successful responses
|
| 249 |
-
if
|
| 250 |
-
|
| 251 |
|
| 252 |
return response
|
| 253 |
|
| 254 |
except instaloader.exceptions.ConnectionException as e:
|
| 255 |
error_message = str(e)
|
| 256 |
-
logger.error("Connection error:
|
| 257 |
|
| 258 |
-
#
|
| 259 |
-
retry_after = None
|
| 260 |
if "429 Too Many Requests" in error_message:
|
| 261 |
-
|
| 262 |
try:
|
| 263 |
-
|
| 264 |
-
|
| 265 |
-
|
| 266 |
-
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
await handle_rate_limit_error(request_username, retry_after)
|
| 270 |
-
|
| 271 |
-
# Try to serve from cache if available
|
| 272 |
-
if request_username:
|
| 273 |
-
cached_response = await get_cached_stories(request_username)
|
| 274 |
-
if cached_response:
|
| 275 |
-
logger.info(f"Returning cached response for rate-limited request")
|
| 276 |
-
return {**cached_response, "from_cache": True}
|
| 277 |
|
| 278 |
-
|
| 279 |
-
|
| 280 |
-
detail=f"Instagram rate limit exceeded. Please try again in {retry_after//60} minutes."
|
| 281 |
-
)
|
| 282 |
-
elif "401 Unauthorized" in error_message and "Please wait a few minutes" in error_message:
|
| 283 |
-
if request_username:
|
| 284 |
-
await handle_rate_limit_error(request_username)
|
| 285 |
|
| 286 |
-
# Try to serve from cache
|
| 287 |
-
if
|
| 288 |
-
|
| 289 |
-
if
|
| 290 |
-
|
| 291 |
-
|
| 292 |
-
|
| 293 |
raise HTTPException(
|
| 294 |
status_code=status.HTTP_429_TOO_MANY_REQUESTS,
|
| 295 |
-
detail="Instagram rate limit exceeded. Please try again
|
| 296 |
)
|
|
|
|
| 297 |
elif "404 Not Found" in error_message:
|
| 298 |
-
raise HTTPException(
|
| 299 |
-
|
| 300 |
-
detail="Profile not found or no stories available"
|
| 301 |
-
)
|
| 302 |
else:
|
| 303 |
-
raise HTTPException(
|
| 304 |
-
|
| 305 |
-
detail="Instagram service unavailable"
|
| 306 |
-
)
|
| 307 |
-
|
| 308 |
-
except instaloader.exceptions.QueryReturnedBadRequestException as e:
|
| 309 |
-
logger.error("API error 400: %s", str(e))
|
| 310 |
-
if request_username:
|
| 311 |
-
await handle_rate_limit_error(request_username)
|
| 312 |
|
| 313 |
-
# Try to serve from cache if available
|
| 314 |
-
if request_username:
|
| 315 |
-
cached_response = await get_cached_stories(request_username)
|
| 316 |
-
if cached_response:
|
| 317 |
-
logger.info(f"Returning cached response for rate-limited request")
|
| 318 |
-
return {**cached_response, "from_cache": True}
|
| 319 |
-
|
| 320 |
-
raise HTTPException(
|
| 321 |
-
status_code=status.HTTP_429_TOO_MANY_REQUESTS,
|
| 322 |
-
detail="Instagram rate limit exceeded"
|
| 323 |
-
)
|
| 324 |
-
|
| 325 |
-
except HTTPException:
|
| 326 |
-
# Re-raise HTTP exceptions without modification
|
| 327 |
-
raise
|
| 328 |
-
|
| 329 |
except Exception as e:
|
| 330 |
-
logger.error("
|
| 331 |
-
raise HTTPException(
|
| 332 |
-
|
| 333 |
-
detail="Internal server error"
|
| 334 |
-
)
|
| 335 |
return wrapper
|
| 336 |
|
| 337 |
@app.get("/stories/{username}")
|
| 338 |
@handle_instagram_errors
|
| 339 |
async def get_stories(username: str, background_tasks: BackgroundTasks, cached: bool = False):
|
| 340 |
-
"""Retrieve stories
|
| 341 |
-
logger.info(f"
|
| 342 |
|
| 343 |
-
#
|
| 344 |
if cached:
|
| 345 |
-
|
| 346 |
-
if
|
| 347 |
-
return {**
|
| 348 |
|
| 349 |
try:
|
| 350 |
-
#
|
| 351 |
-
|
| 352 |
-
|
| 353 |
-
|
| 354 |
-
|
| 355 |
-
|
| 356 |
-
|
| 357 |
-
|
| 358 |
-
|
| 359 |
-
|
| 360 |
-
|
| 361 |
-
|
| 362 |
-
|
| 363 |
-
|
| 364 |
-
try:
|
| 365 |
-
logger.info(f"Fetching profile (attempt {attempt+1}/3)")
|
| 366 |
-
profile = instaloader.Profile.from_username(L.context, username)
|
| 367 |
-
break
|
| 368 |
-
except instaloader.exceptions.QueryReturnedBadRequestException:
|
| 369 |
-
wait_time = (attempt + 1) * random.uniform(3.0, 5.0)
|
| 370 |
-
logger.warning(f"Rate limited, waiting {wait_time:.2f}s")
|
| 371 |
-
time.sleep(wait_time)
|
| 372 |
-
except instaloader.exceptions.ConnectionException as e:
|
| 373 |
-
if "401 Unauthorized" in str(e) and "Please wait a few minutes" in str(e):
|
| 374 |
-
# Try with a fresh login if session might be expired
|
| 375 |
-
if attempt < 2: # Only try this once
|
| 376 |
-
logger.warning("Session may be expired, trying with fresh login")
|
| 377 |
-
L = get_instaloader(force_login=True)
|
| 378 |
-
time.sleep(random.uniform(4.0, 6.0))
|
| 379 |
-
continue
|
| 380 |
-
raise
|
| 381 |
|
| 382 |
-
|
| 383 |
-
|
| 384 |
-
|
| 385 |
-
|
| 386 |
-
|
| 387 |
-
|
| 388 |
-
|
| 389 |
-
if not profile.has_viewable_story:
|
| 390 |
-
logger.warning("No viewable story")
|
| 391 |
-
raise HTTPException(
|
| 392 |
-
status.HTTP_404_NOT_FOUND,
|
| 393 |
-
"No accessible stories for this profile"
|
| 394 |
-
)
|
| 395 |
-
|
| 396 |
-
# Additional delay before story fetch
|
| 397 |
-
time.sleep(random.uniform(1.0, 2.5))
|
| 398 |
|
| 399 |
-
|
| 400 |
-
|
| 401 |
-
|
| 402 |
-
|
| 403 |
-
|
| 404 |
-
|
| 405 |
-
|
| 406 |
-
|
| 407 |
-
|
| 408 |
-
|
| 409 |
-
|
| 410 |
-
|
| 411 |
-
|
| 412 |
-
|
| 413 |
-
|
| 414 |
-
|
| 415 |
-
|
| 416 |
-
|
| 417 |
-
if item.is_video:
|
| 418 |
-
try:
|
| 419 |
-
if hasattr(item, 'view_count'):
|
| 420 |
-
story_data["views"] = item.view_count
|
| 421 |
-
except AttributeError:
|
| 422 |
-
pass
|
| 423 |
-
|
| 424 |
-
stories.append(story_data)
|
| 425 |
|
| 426 |
-
|
| 427 |
-
|
| 428 |
-
|
| 429 |
-
|
| 430 |
-
|
| 431 |
-
|
| 432 |
-
|
| 433 |
-
|
| 434 |
-
logger.info("No active stories found")
|
| 435 |
-
raise HTTPException(
|
| 436 |
-
status.HTTP_404_NOT_FOUND,
|
| 437 |
-
"No active stories available"
|
| 438 |
-
)
|
| 439 |
-
|
| 440 |
-
# Queue session save in background
|
| 441 |
-
background_tasks.add_task(lambda: L.save_session_to_file(SESSION_FILE) if SESSION_FILE else None)
|
| 442 |
-
|
| 443 |
-
# Final randomized delay
|
| 444 |
-
time.sleep(random.uniform(0.3, 0.7))
|
| 445 |
|
| 446 |
-
|
| 447 |
-
|
| 448 |
-
|
| 449 |
-
|
| 450 |
-
|
| 451 |
-
|
| 452 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 453 |
|
| 454 |
-
# Execute in thread pool to not block the event loop
|
| 455 |
-
return await run_in_threadpool(fetch_stories)
|
| 456 |
-
|
| 457 |
except Exception as e:
|
| 458 |
-
logger.error(f"
|
| 459 |
raise
|
| 460 |
|
| 461 |
@app.get("/download/{url:path}")
|
| 462 |
@handle_instagram_errors
|
| 463 |
async def download_media(url: str):
|
| 464 |
"""Download and proxy media content"""
|
| 465 |
-
logger.info(f"Download request for
|
| 466 |
|
| 467 |
try:
|
| 468 |
-
#
|
| 469 |
if not url.startswith(("https://instagram", "https://scontent")):
|
| 470 |
raise HTTPException(
|
| 471 |
status_code=status.HTTP_400_BAD_REQUEST,
|
| 472 |
detail="Invalid URL format"
|
| 473 |
)
|
| 474 |
|
| 475 |
-
#
|
| 476 |
-
time.sleep(random.uniform(0.5, 1.5))
|
| 477 |
-
|
| 478 |
-
# Configure headers to mimic a browser
|
| 479 |
headers = {
|
| 480 |
"User-Agent": random.choice(USER_AGENTS),
|
| 481 |
-
"Accept": "*/*",
|
| 482 |
-
"Accept-Language": "en-US,en;q=0.9",
|
| 483 |
-
"Accept-Encoding": "gzip, deflate, br",
|
| 484 |
"Referer": "https://www.instagram.com/",
|
| 485 |
-
"
|
| 486 |
-
"Sec-Fetch-Dest": "empty",
|
| 487 |
-
"Sec-Fetch-Mode": "cors",
|
| 488 |
-
"Sec-Fetch-Site": "cross-site",
|
| 489 |
-
"Pragma": "no-cache",
|
| 490 |
-
"Cache-Control": "no-cache",
|
| 491 |
}
|
| 492 |
|
| 493 |
-
#
|
| 494 |
-
|
| 495 |
-
response = session.get(url, headers=headers, stream=True, timeout=10)
|
| 496 |
response.raise_for_status()
|
| 497 |
|
| 498 |
-
# Determine content type
|
| 499 |
if "Content-Type" in response.headers:
|
| 500 |
content_type = response.headers["Content-Type"]
|
| 501 |
elif url.endswith((".jpg", ".jpeg")):
|
|
@@ -505,67 +345,49 @@ async def download_media(url: str):
|
|
| 505 |
else:
|
| 506 |
content_type = "application/octet-stream"
|
| 507 |
|
| 508 |
-
logger.info(f"Media downloaded
|
| 509 |
|
| 510 |
-
#
|
| 511 |
return StreamingResponse(
|
| 512 |
response.iter_content(chunk_size=8192),
|
| 513 |
media_type=content_type,
|
| 514 |
-
headers={
|
| 515 |
-
"Content-Disposition": f"attachment; filename={url.split('/')[-1]}",
|
| 516 |
-
"Cache-Control": "no-cache, no-store, must-revalidate",
|
| 517 |
-
"Pragma": "no-cache",
|
| 518 |
-
"Expires": "0",
|
| 519 |
-
}
|
| 520 |
)
|
| 521 |
|
| 522 |
-
except
|
| 523 |
logger.error(f"Download failed: {str(e)}")
|
| 524 |
raise HTTPException(
|
| 525 |
status_code=status.HTTP_502_BAD_GATEWAY,
|
| 526 |
detail="Failed to download media"
|
| 527 |
)
|
| 528 |
|
| 529 |
-
#
|
| 530 |
-
@app.get("/health")
|
| 531 |
-
async def health_check():
|
| 532 |
-
return {"status": "ok", "timestamp": datetime.now().isoformat()}
|
| 533 |
-
|
| 534 |
-
# Add middleware to handle rate limit headers
|
| 535 |
-
@app.middleware("http")
|
| 536 |
-
async def add_rate_limit_headers(request: Request, call_next):
|
| 537 |
-
response = await call_next(request)
|
| 538 |
-
|
| 539 |
-
# Add custom headers for rate limiting info
|
| 540 |
-
username = request.path_params.get("username", None)
|
| 541 |
-
if username and username in COOLDOWN_PERIODS:
|
| 542 |
-
remaining = max(0, int((COOLDOWN_PERIODS[username] - datetime.now()).total_seconds()))
|
| 543 |
-
response.headers["X-RateLimit-Reset"] = str(remaining)
|
| 544 |
-
response.headers["X-RateLimit-Remaining"] = "0" if remaining > 0 else "1"
|
| 545 |
-
|
| 546 |
-
return response
|
| 547 |
-
|
| 548 |
-
# Startup event to load cache from disk
|
| 549 |
@app.on_event("startup")
|
| 550 |
-
|
| 551 |
try:
|
|
|
|
| 552 |
for filename in os.listdir(CACHE_DIR):
|
| 553 |
if filename.endswith('.json'):
|
| 554 |
-
file_path = os.path.join(CACHE_DIR, filename)
|
| 555 |
try:
|
| 556 |
-
with open(
|
| 557 |
cache_data = json.load(f)
|
| 558 |
-
|
| 559 |
if "data" in cache_data and "expires" in cache_data:
|
| 560 |
-
|
| 561 |
-
|
| 562 |
-
|
| 563 |
-
|
| 564 |
-
|
| 565 |
-
|
|
|
|
| 566 |
except Exception as e:
|
| 567 |
-
logger.warning(f"
|
| 568 |
|
| 569 |
-
logger.info(f"Loaded {
|
| 570 |
except Exception as e:
|
| 571 |
-
logger.error(f"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from fastapi import FastAPI, HTTPException, status, BackgroundTasks
|
| 2 |
+
from fastapi.responses import StreamingResponse
|
| 3 |
import instaloader
|
| 4 |
import requests
|
| 5 |
import os
|
|
|
|
| 7 |
import logging
|
| 8 |
import random
|
| 9 |
from functools import wraps
|
|
|
|
| 10 |
from datetime import datetime, timedelta
|
| 11 |
+
import json
|
|
|
|
|
|
|
|
|
|
| 12 |
|
| 13 |
# Configure logging
|
| 14 |
logging.basicConfig(
|
|
|
|
| 32 |
os.makedirs(SESSION_DIR, exist_ok=True)
|
| 33 |
os.makedirs(CACHE_DIR, exist_ok=True)
|
| 34 |
|
| 35 |
+
# Simple memory cache
|
| 36 |
+
STORY_CACHE = {}
|
| 37 |
+
CACHE_EXPIRY = {}
|
| 38 |
+
LAST_REQUEST = {}
|
| 39 |
+
COOLDOWN_UNTIL = {}
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
USER_AGENTS = [
|
| 42 |
"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36",
|
| 43 |
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.6 Safari/605.1.15",
|
| 44 |
+
"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/119.0.0.0 Safari/537.36"
|
|
|
|
|
|
|
| 45 |
]
|
| 46 |
|
| 47 |
+
def get_cache_key(username):
|
| 48 |
+
"""Simple cache key generator"""
|
| 49 |
+
return username.lower()
|
| 50 |
|
| 51 |
+
def get_cached_stories(username):
|
| 52 |
+
"""Get stories from cache if available"""
|
| 53 |
+
key = get_cache_key(username)
|
| 54 |
+
if key in STORY_CACHE and key in CACHE_EXPIRY:
|
| 55 |
+
if datetime.now() < CACHE_EXPIRY[key]:
|
| 56 |
+
logger.info(f"Cache hit for {username}")
|
| 57 |
+
return STORY_CACHE[key]
|
| 58 |
+
else:
|
| 59 |
+
# Clean expired cache
|
| 60 |
+
STORY_CACHE.pop(key, None)
|
| 61 |
+
CACHE_EXPIRY.pop(key, None)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
return None
|
| 63 |
|
| 64 |
+
def save_to_cache(username, data, minutes=30):
|
| 65 |
+
"""Save stories to cache"""
|
| 66 |
+
key = get_cache_key(username)
|
| 67 |
+
STORY_CACHE[key] = data
|
| 68 |
+
CACHE_EXPIRY[key] = datetime.now() + timedelta(minutes=minutes)
|
| 69 |
|
| 70 |
+
# Save to disk
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
try:
|
| 72 |
+
cache_file = os.path.join(CACHE_DIR, f"{key}.json")
|
| 73 |
with open(cache_file, 'w') as f:
|
| 74 |
json.dump({
|
| 75 |
+
"data": data,
|
| 76 |
+
"expires": CACHE_EXPIRY[key].isoformat()
|
| 77 |
}, f)
|
| 78 |
+
logger.info(f"Cached {username} stories for {minutes} minutes")
|
| 79 |
except Exception as e:
|
| 80 |
+
logger.warning(f"Failed to save cache: {str(e)}")
|
| 81 |
+
|
| 82 |
+
def is_rate_limited(username):
|
| 83 |
+
"""Check if we should rate limit this request"""
|
| 84 |
+
# Check if in cooldown
|
| 85 |
+
if username in COOLDOWN_UNTIL:
|
| 86 |
+
if datetime.now() < COOLDOWN_UNTIL[username]:
|
| 87 |
+
seconds_left = (COOLDOWN_UNTIL[username] - datetime.now()).total_seconds()
|
| 88 |
+
logger.warning(f"{username} in cooldown for {int(seconds_left)}s")
|
| 89 |
+
return True
|
| 90 |
+
else:
|
| 91 |
+
# Cooldown expired
|
| 92 |
+
COOLDOWN_UNTIL.pop(username)
|
| 93 |
+
|
| 94 |
+
# Check request frequency
|
| 95 |
+
if username in LAST_REQUEST:
|
| 96 |
+
seconds_since = (datetime.now() - LAST_REQUEST[username]).total_seconds()
|
| 97 |
+
if seconds_since < 60: # 1 minute minimum between requests
|
| 98 |
+
logger.warning(f"Rate limiting {username}: only {int(seconds_since)}s since last request")
|
| 99 |
+
return True
|
| 100 |
+
|
| 101 |
+
# Update last request time
|
| 102 |
+
LAST_REQUEST[username] = datetime.now()
|
| 103 |
+
return False
|
| 104 |
|
| 105 |
+
def set_cooldown(username, minutes=30):
|
| 106 |
+
"""Set a cooldown period for a username"""
|
| 107 |
+
COOLDOWN_UNTIL[username] = datetime.now() + timedelta(minutes=minutes)
|
| 108 |
+
logger.warning(f"Setting {minutes}m cooldown for {username}")
|
| 109 |
+
|
| 110 |
+
def get_instaloader():
|
| 111 |
+
"""Create and configure Instaloader instance"""
|
| 112 |
L = instaloader.Instaloader(
|
| 113 |
sleep=True,
|
| 114 |
request_timeout=30,
|
|
|
|
| 131 |
)
|
| 132 |
|
| 133 |
try:
|
| 134 |
+
if SESSION_FILE and os.path.exists(SESSION_FILE):
|
| 135 |
+
logger.info(f"Loading session from {SESSION_FILE}")
|
| 136 |
try:
|
| 137 |
L.load_session_from_file(INSTAGRAM_USERNAME, SESSION_FILE)
|
| 138 |
logger.info("Session loaded successfully")
|
|
|
|
| 139 |
return L
|
| 140 |
except Exception as e:
|
| 141 |
+
logger.warning(f"Session load failed: {str(e)}")
|
|
|
|
|
|
|
|
|
|
| 142 |
|
| 143 |
+
logger.info("Starting fresh login")
|
| 144 |
+
time.sleep(random.uniform(1.0, 2.0)) # Small delay before login
|
| 145 |
L.login(INSTAGRAM_USERNAME, INSTAGRAM_PASSWORD)
|
| 146 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 147 |
if SESSION_FILE:
|
| 148 |
L.save_session_to_file(SESSION_FILE)
|
| 149 |
logger.info(f"Saved new session to {SESSION_FILE}")
|
| 150 |
|
|
|
|
|
|
|
|
|
|
| 151 |
return L
|
| 152 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 153 |
except Exception as e:
|
| 154 |
+
logger.error(f"Login failed: {str(e)}")
|
| 155 |
raise HTTPException(
|
| 156 |
status_code=status.HTTP_503_SERVICE_UNAVAILABLE,
|
| 157 |
+
detail="Instagram login failed"
|
| 158 |
)
|
| 159 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 160 |
def handle_instagram_errors(func):
|
| 161 |
@wraps(func)
|
| 162 |
async def wrapper(*args, **kwargs):
|
| 163 |
+
# Extract username if present
|
| 164 |
+
username = ""
|
| 165 |
if len(args) > 1 and hasattr(args[1], '__str__'):
|
| 166 |
+
username = str(args[1])
|
| 167 |
|
| 168 |
try:
|
| 169 |
# Check for rate limiting
|
| 170 |
+
if username and is_rate_limited(username):
|
| 171 |
+
# Try to serve from cache
|
| 172 |
+
cached = get_cached_stories(username)
|
| 173 |
+
if cached:
|
| 174 |
+
return {**cached, "from_cache": True}
|
|
|
|
| 175 |
else:
|
|
|
|
| 176 |
raise HTTPException(
|
| 177 |
status_code=status.HTTP_429_TOO_MANY_REQUESTS,
|
| 178 |
detail="Rate limit exceeded. Please try again later."
|
| 179 |
)
|
| 180 |
+
|
| 181 |
+
# Execute the handler
|
| 182 |
response = await func(*args, **kwargs)
|
| 183 |
|
| 184 |
+
# Cache successful responses
|
| 185 |
+
if username and isinstance(response, dict) and "data" in response:
|
| 186 |
+
save_to_cache(username, response)
|
| 187 |
|
| 188 |
return response
|
| 189 |
|
| 190 |
except instaloader.exceptions.ConnectionException as e:
|
| 191 |
error_message = str(e)
|
| 192 |
+
logger.error(f"Connection error: {error_message}")
|
| 193 |
|
| 194 |
+
# Handle rate limiting errors
|
|
|
|
| 195 |
if "429 Too Many Requests" in error_message:
|
| 196 |
+
retry_mins = 30 # Default
|
| 197 |
try:
|
| 198 |
+
# Try to extract minutes from error message
|
| 199 |
+
if "retried in" in error_message and "minutes" in error_message:
|
| 200 |
+
retry_text = error_message.split("retried in")[1].split("minutes")[0].strip()
|
| 201 |
+
retry_mins = int(retry_text)
|
| 202 |
+
except:
|
| 203 |
+
pass
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 204 |
|
| 205 |
+
if username:
|
| 206 |
+
set_cooldown(username, retry_mins)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 207 |
|
| 208 |
+
# Try to serve from cache
|
| 209 |
+
if username:
|
| 210 |
+
cached = get_cached_stories(username)
|
| 211 |
+
if cached:
|
| 212 |
+
return {**cached, "from_cache": True}
|
| 213 |
+
|
|
|
|
| 214 |
raise HTTPException(
|
| 215 |
status_code=status.HTTP_429_TOO_MANY_REQUESTS,
|
| 216 |
+
detail=f"Instagram rate limit exceeded. Please try again in {retry_mins} minutes."
|
| 217 |
)
|
| 218 |
+
|
| 219 |
elif "404 Not Found" in error_message:
|
| 220 |
+
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND,
|
| 221 |
+
detail="Profile or stories not found")
|
|
|
|
|
|
|
| 222 |
else:
|
| 223 |
+
raise HTTPException(status_code=status.HTTP_503_SERVICE_UNAVAILABLE,
|
| 224 |
+
detail="Instagram service unavailable")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 225 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 226 |
except Exception as e:
|
| 227 |
+
logger.error(f"Error: {str(e)}")
|
| 228 |
+
raise HTTPException(status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
|
| 229 |
+
detail="Internal server error")
|
|
|
|
|
|
|
| 230 |
return wrapper
|
| 231 |
|
| 232 |
@app.get("/stories/{username}")
|
| 233 |
@handle_instagram_errors
|
| 234 |
async def get_stories(username: str, background_tasks: BackgroundTasks, cached: bool = False):
|
| 235 |
+
"""Retrieve Instagram stories for a user"""
|
| 236 |
+
logger.info(f"Request for @{username} stories")
|
| 237 |
|
| 238 |
+
# Return from cache if requested
|
| 239 |
if cached:
|
| 240 |
+
cached_result = get_cached_stories(username)
|
| 241 |
+
if cached_result:
|
| 242 |
+
return {**cached_result, "from_cache": True}
|
| 243 |
|
| 244 |
try:
|
| 245 |
+
# Get Instagram loader
|
| 246 |
+
L = get_instaloader()
|
| 247 |
+
logger.info("Instagram loader ready")
|
| 248 |
+
|
| 249 |
+
# Small delay
|
| 250 |
+
time.sleep(random.uniform(1.0, 2.0))
|
| 251 |
+
|
| 252 |
+
# Get profile
|
| 253 |
+
logger.info(f"Fetching profile for {username}")
|
| 254 |
+
try:
|
| 255 |
+
profile = instaloader.Profile.from_username(L.context, username)
|
| 256 |
+
except Exception as e:
|
| 257 |
+
logger.error(f"Profile fetch error: {str(e)}")
|
| 258 |
+
raise
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 259 |
|
| 260 |
+
# Check story access
|
| 261 |
+
if not profile.has_viewable_story:
|
| 262 |
+
logger.warning("No viewable story")
|
| 263 |
+
raise HTTPException(
|
| 264 |
+
status_code=status.HTTP_404_NOT_FOUND,
|
| 265 |
+
detail="No accessible stories for this user"
|
| 266 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 267 |
|
| 268 |
+
# Get stories
|
| 269 |
+
logger.info("Fetching stories")
|
| 270 |
+
stories = []
|
| 271 |
+
for story in L.get_stories(userids=[profile.userid]):
|
| 272 |
+
for item in story.get_items():
|
| 273 |
+
story_data = {
|
| 274 |
+
"id": str(item.mediaid),
|
| 275 |
+
"url": item.url,
|
| 276 |
+
"type": "video" if item.is_video else "image",
|
| 277 |
+
"timestamp": item.date_utc.isoformat()
|
| 278 |
+
}
|
| 279 |
+
|
| 280 |
+
# Safe attribute access
|
| 281 |
+
if item.is_video and hasattr(item, 'view_count'):
|
| 282 |
+
try:
|
| 283 |
+
story_data["views"] = item.view_count
|
| 284 |
+
except:
|
| 285 |
+
pass
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 286 |
|
| 287 |
+
stories.append(story_data)
|
| 288 |
+
|
| 289 |
+
if not stories:
|
| 290 |
+
logger.info("No stories found")
|
| 291 |
+
raise HTTPException(
|
| 292 |
+
status_code=status.HTTP_404_NOT_FOUND,
|
| 293 |
+
detail="No active stories available"
|
| 294 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 295 |
|
| 296 |
+
# Save session in background to not delay response
|
| 297 |
+
background_tasks.add_task(lambda: L.save_session_to_file(SESSION_FILE) if SESSION_FILE else None)
|
| 298 |
+
|
| 299 |
+
logger.info(f"Returning {len(stories)} stories")
|
| 300 |
+
result = {
|
| 301 |
+
"data": stories,
|
| 302 |
+
"count": len(stories),
|
| 303 |
+
"username": username,
|
| 304 |
+
"fetched_at": datetime.now().isoformat()
|
| 305 |
+
}
|
| 306 |
+
|
| 307 |
+
return result
|
| 308 |
|
|
|
|
|
|
|
|
|
|
| 309 |
except Exception as e:
|
| 310 |
+
logger.error(f"Error: {str(e)}")
|
| 311 |
raise
|
| 312 |
|
| 313 |
@app.get("/download/{url:path}")
|
| 314 |
@handle_instagram_errors
|
| 315 |
async def download_media(url: str):
|
| 316 |
"""Download and proxy media content"""
|
| 317 |
+
logger.info(f"Download request for URL")
|
| 318 |
|
| 319 |
try:
|
| 320 |
+
# Validate URL
|
| 321 |
if not url.startswith(("https://instagram", "https://scontent")):
|
| 322 |
raise HTTPException(
|
| 323 |
status_code=status.HTTP_400_BAD_REQUEST,
|
| 324 |
detail="Invalid URL format"
|
| 325 |
)
|
| 326 |
|
| 327 |
+
# Configure request
|
|
|
|
|
|
|
|
|
|
| 328 |
headers = {
|
| 329 |
"User-Agent": random.choice(USER_AGENTS),
|
|
|
|
|
|
|
|
|
|
| 330 |
"Referer": "https://www.instagram.com/",
|
| 331 |
+
"Accept": "*/*",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 332 |
}
|
| 333 |
|
| 334 |
+
# Get media
|
| 335 |
+
response = requests.get(url, headers=headers, stream=True, timeout=15)
|
|
|
|
| 336 |
response.raise_for_status()
|
| 337 |
|
| 338 |
+
# Determine content type
|
| 339 |
if "Content-Type" in response.headers:
|
| 340 |
content_type = response.headers["Content-Type"]
|
| 341 |
elif url.endswith((".jpg", ".jpeg")):
|
|
|
|
| 345 |
else:
|
| 346 |
content_type = "application/octet-stream"
|
| 347 |
|
| 348 |
+
logger.info(f"Media downloaded: {content_type}")
|
| 349 |
|
| 350 |
+
# Return the media
|
| 351 |
return StreamingResponse(
|
| 352 |
response.iter_content(chunk_size=8192),
|
| 353 |
media_type=content_type,
|
| 354 |
+
headers={"Content-Disposition": f"attachment; filename={url.split('/')[-1]}"}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 355 |
)
|
| 356 |
|
| 357 |
+
except Exception as e:
|
| 358 |
logger.error(f"Download failed: {str(e)}")
|
| 359 |
raise HTTPException(
|
| 360 |
status_code=status.HTTP_502_BAD_GATEWAY,
|
| 361 |
detail="Failed to download media"
|
| 362 |
)
|
| 363 |
|
| 364 |
+
# Load cache from disk at startup
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 365 |
@app.on_event("startup")
|
| 366 |
+
def load_cache_from_disk():
|
| 367 |
try:
|
| 368 |
+
count = 0
|
| 369 |
for filename in os.listdir(CACHE_DIR):
|
| 370 |
if filename.endswith('.json'):
|
|
|
|
| 371 |
try:
|
| 372 |
+
with open(os.path.join(CACHE_DIR, filename), 'r') as f:
|
| 373 |
cache_data = json.load(f)
|
| 374 |
+
|
| 375 |
if "data" in cache_data and "expires" in cache_data:
|
| 376 |
+
# Convert ISO string to datetime
|
| 377 |
+
expires = datetime.fromisoformat(cache_data["expires"])
|
| 378 |
+
if expires > datetime.now():
|
| 379 |
+
key = filename.replace('.json', '')
|
| 380 |
+
STORY_CACHE[key] = cache_data["data"]
|
| 381 |
+
CACHE_EXPIRY[key] = expires
|
| 382 |
+
count += 1
|
| 383 |
except Exception as e:
|
| 384 |
+
logger.warning(f"Couldn't load cache file {filename}: {e}")
|
| 385 |
|
| 386 |
+
logger.info(f"Loaded {count} items from cache")
|
| 387 |
except Exception as e:
|
| 388 |
+
logger.error(f"Cache loading error: {e}")
|
| 389 |
+
|
| 390 |
+
# Health check endpoint
|
| 391 |
+
@app.get("/health")
|
| 392 |
+
async def health_check():
|
| 393 |
+
return {"status": "ok", "timestamp": datetime.now().isoformat()}
|