feat: harden Bluesky login with robust retry and error handling

- Ported advanced error detection (`is_auth_error`, `is_transient_error`, `is_network_error`) from the daemon script to accurately classify exceptions.
- Implemented `get_rate_limit_wait_seconds` to intelligently parse HTTP rate-limit headers (`retry-after`, `ratelimit-reset`) instead of relying on static backoffs.
- Increased default login retry attempts to 5, with jitter and bounded delays.
- Updated `login_with_backoff` to fail fast on invalid credentials while gracefully handling transient network issues and rate limits.
This commit is contained in:
Guillem Hernandez Sola
2026-05-08 08:32:40 +02:00
parent 55a327a10e
commit dd096029f1

View File

@@ -16,6 +16,7 @@ import os
import sys import sys
import time import time
import random import random
import re
from dataclasses import dataclass from dataclasses import dataclass
from atproto import Client, client_utils, models from atproto import Client, client_utils, models
@@ -27,7 +28,7 @@ from atproto import Client, client_utils, models
@dataclass(frozen=True) @dataclass(frozen=True)
class RetryConfig: class RetryConfig:
login_max_attempts: int = 5 login_max_attempts: int = 5
login_base_delay_seconds: float = 2.0 login_base_delay_seconds: float = 10.0
login_max_delay_seconds: float = 600.0 login_max_delay_seconds: float = 600.0
login_jitter_seconds: float = 1.5 login_jitter_seconds: float = 1.5
@@ -87,34 +88,125 @@ def make_rich(content: str):
# ============================================================ # ============================================================
# Error helpers # Error helpers
# ============================================================ # ============================================================
def is_rate_limited_error(e) -> bool: def is_rate_limited_error(error_obj) -> bool:
text = repr(e) text = repr(error_obj).lower()
return any(s in text for s in ["429", "RateLimitExceeded", "Too Many Requests"]) return (
"429" in text
or "ratelimitexceeded" in text
or "too many requests" in text
or "rate limit" in text
)
def is_auth_error(e) -> bool: def is_auth_error(error_obj) -> bool:
text = repr(e).lower() text = repr(error_obj).lower()
return any(s in text for s in [ return (
"401", "403", "invalid identifier or password", "401" in text
"authenticationrequired", "invalidtoken", or "403" in text
]) or "invalid identifier or password" in text
or "authenticationrequired" in text
or "invalidtoken" in text
)
def is_network_error(e) -> bool: def is_network_error(error_obj) -> bool:
text = repr(e) text = repr(error_obj)
return any(s in text for s in [ signals = [
"ConnectError", "RemoteProtocolError", "ReadTimeout", "ConnectError",
"WriteTimeout", "TimeoutException", "503", "502", "504", "RemoteProtocolError",
"ReadTimeout",
"WriteTimeout",
"TimeoutException",
"503",
"502",
"504",
"ConnectionResetError", "ConnectionResetError",
]) ]
return any(sig in text for sig in signals)
def is_timeout_error(e) -> bool: def is_transient_error(error_obj) -> bool:
text = repr(e) error_text = repr(error_obj)
return any(s in text for s in [ transient_signals = [
"InvokeTimeoutError", "ReadTimeout", "InvokeTimeoutError",
"WriteTimeout", "TimeoutException", "ReadTimeout",
]) "WriteTimeout",
"TimeoutException",
"RemoteProtocolError",
"ConnectError",
"503",
"502",
"504",
]
return any(signal in error_text for signal in transient_signals)
def get_rate_limit_wait_seconds(error_obj, default_delay: float, max_delay: float) -> float:
"""
Parse common rate-limit headers and return a bounded wait time in seconds.
Supports:
- retry-after
- x-ratelimit-after
- ratelimit-reset (unix timestamp)
"""
try:
now_ts = int(time.time())
# Direct headers on exception
headers = getattr(error_obj, "headers", None) or {}
retry_after = headers.get("retry-after") or headers.get("Retry-After")
if retry_after:
return min(max(float(retry_after), 1.0), max_delay)
x_after = headers.get("x-ratelimit-after") or headers.get("X-RateLimit-After")
if x_after:
return min(max(float(x_after), 1.0), max_delay)
reset_value = headers.get("ratelimit-reset") or headers.get("RateLimit-Reset")
if reset_value:
wait_seconds = max(float(reset_value) - now_ts + 1.0, default_delay)
return min(wait_seconds, max_delay)
except Exception:
pass
try:
# Nested response headers
response = getattr(error_obj, "response", None)
headers = getattr(response, "headers", None) or {}
now_ts = int(time.time())
retry_after = headers.get("retry-after") or headers.get("Retry-After")
if retry_after:
return min(max(float(retry_after), 1.0), max_delay)
x_after = headers.get("x-ratelimit-after") or headers.get("X-RateLimit-After")
if x_after:
return min(max(float(x_after), 1.0), max_delay)
reset_value = headers.get("ratelimit-reset") or headers.get("RateLimit-Reset")
if reset_value:
wait_seconds = max(float(reset_value) - now_ts + 1.0, default_delay)
return min(wait_seconds, max_delay)
except Exception:
pass
# repr fallback parsing
text = repr(error_obj)
m = re.search(r"'retry-after': '(\d+)'", text, re.IGNORECASE)
if m:
return min(max(float(m.group(1)), 1.0), max_delay)
m = re.search(r"'x-ratelimit-after': '(\d+)'", text, re.IGNORECASE)
if m:
return min(max(float(m.group(1)), 1.0), max_delay)
m = re.search(r"'ratelimit-reset': '(\d+)'", text, re.IGNORECASE)
if m:
now_ts = int(time.time())
wait_seconds = max(float(m.group(1)) - now_ts + 1.0, default_delay)
return min(wait_seconds, max_delay)
return default_delay
# ============================================================ # ============================================================
@@ -126,14 +218,14 @@ def login_with_backoff(
password: str, password: str,
service_url: str, service_url: str,
max_attempts: int = 5, max_attempts: int = 5,
base_delay: float = 2.0, base_delay: float = 10.0,
max_delay: float = 600.0, max_delay: float = 600.0,
jitter: float = 1.5, jitter: float = 1.5,
) -> bool: ) -> bool:
for attempt in range(1, max_attempts + 1): for attempt in range(1, max_attempts + 1):
try: try:
logging.info( logging.info(
f"🔐 Login attempt {attempt}/{max_attempts}{service_url} as {username}" f"🔑 Login attempt {attempt}/{max_attempts}{service_url} as {username}"
) )
client.login(username, password) client.login(username, password)
logging.info("✅ Login successful.") logging.info("✅ Login successful.")
@@ -142,25 +234,49 @@ def login_with_backoff(
except Exception as e: except Exception as e:
logging.exception("❌ Login exception") logging.exception("❌ Login exception")
# Fail fast on invalid credentials
if is_auth_error(e): if is_auth_error(e):
logging.error("❌ Bad credentials. Check handle/password.") logging.error("❌ Bad credentials. Check handle/password.")
return False return False
if attempt >= max_attempts: # Respect explicit rate-limit timing
logging.error("❌ Login retries exhausted.") if is_rate_limited_error(e):
if attempt < max_attempts:
wait = get_rate_limit_wait_seconds(e, default_delay=base_delay, max_delay=max_delay)
wait = wait + random.uniform(0, jitter)
logging.warning(
f"⏳ Rate-limited on login (attempt {attempt}/{max_attempts}). "
f"Retrying in {wait:.1f}s..."
)
time.sleep(wait)
continue
logging.error("❌ Exhausted login retries due to rate limiting.")
return False return False
if is_rate_limited_error(e): # Retry transient/network problems
delay = min(base_delay * (2 ** (attempt - 1)), max_delay) + random.uniform(0, jitter) if is_network_error(e) or is_transient_error(e):
logging.warning(f"⏳ Rate-limited on login. Retrying in {delay:.1f}s...") if attempt < max_attempts:
elif is_network_error(e) or is_timeout_error(e): wait = min(base_delay * attempt, max_delay) + random.uniform(0, jitter)
delay = min(base_delay * attempt, max_delay) + random.uniform(0, jitter) logging.warning(
logging.warning(f"⏳ Transient login error. Retrying in {delay:.1f}s...") f"⏳ Transient login error (attempt {attempt}/{max_attempts}). "
else: f"Retrying in {wait:.1f}s..."
delay = min(base_delay * attempt, max_delay) + random.uniform(0, jitter) )
logging.warning(f"⏳ Unknown login error. Retrying in {delay:.1f}s...") time.sleep(wait)
continue
time.sleep(delay) logging.error("❌ Exhausted login retries after transient/network errors.")
return False
# Unknown errors: bounded retry anyway
if attempt < max_attempts:
wait = min(base_delay * attempt, max_delay) + random.uniform(0, jitter)
logging.warning(
f"⏳ Unknown login error (attempt {attempt}/{max_attempts}). "
f"Retrying in {wait:.1f}s..."
)
time.sleep(wait)
continue
return False return False
@@ -283,48 +399,31 @@ def main():
parser.add_argument("text", help="Post text content") parser.add_argument("text", help="Post text content")
parser.add_argument("--username", required=True, help="Bluesky handle or email") parser.add_argument("--username", required=True, help="Bluesky handle or email")
parser.add_argument("--password", required=True, help="Bluesky app password") parser.add_argument("--password", required=True, help="Bluesky app password")
parser.add_argument("--service", default="https://eurosky.social", help="Bluesky PDS URL") parser.add_argument("--service", default="https://bsky.social", help="Bluesky PDS URL")
parser.add_argument("--lang", default="ca", help="Comma-separated language codes (e.g. ca,es)") parser.add_argument("--lang", default="ca", help="Comma-separated language codes (e.g. ca,es)")
parser.add_argument("--image", default=None, help="Path to image file to attach") parser.add_argument("--image", default=None, help="Path to image file to attach")
parser.add_argument("--video", default=None, help="Path to video file to attach") parser.add_argument("--video", default=None, help="Path to video file to attach")
parser.add_argument("--alt", default="", help="Alt text for the attached media") parser.add_argument("--alt", default="", help="Alt text for media")
args = parser.parse_args() args = parser.parse_args()
if args.image and args.video:
logging.error("❌ Cannot attach both --image and --video at the same time.")
sys.exit(1)
if args.image and not os.path.isfile(args.image):
logging.error(f"❌ Image file not found: {args.image}")
sys.exit(1)
if args.video and not os.path.isfile(args.video):
logging.error(f"❌ Video file not found: {args.video}")
sys.exit(1)
langs = [lang.strip() for lang in args.lang.split(",") if lang.strip()] or ["ca"]
logging.info(f"🌍 Language(s): {langs}")
cfg = RetryConfig()
client = Client(base_url=args.service) client = Client(base_url=args.service)
success = login_with_backoff(
logged_in = login_with_backoff( client,
client=client, args.username,
username=args.username, args.password,
password=args.password, args.service,
service_url=args.service, max_attempts=RetryConfig.login_max_attempts,
max_attempts=cfg.login_max_attempts, base_delay=RetryConfig.login_base_delay_seconds,
base_delay=cfg.login_base_delay_seconds, max_delay=RetryConfig.login_max_delay_seconds,
max_delay=cfg.login_max_delay_seconds, jitter=RetryConfig.login_jitter_seconds,
jitter=cfg.login_jitter_seconds,
) )
if not success:
if not logged_in:
logging.error("❌ Could not log in. Exiting.")
sys.exit(1) sys.exit(1)
success = post_to_bsky( langs = [l.strip() for l in args.lang.split(",") if l.strip()]
client=client, post_success = post_to_bsky(
client,
text=args.text, text=args.text,
langs=langs, langs=langs,
image_path=args.image, image_path=args.image,
@@ -332,7 +431,8 @@ def main():
alt_text=args.alt, alt_text=args.alt,
) )
sys.exit(0 if success else 1) if not post_success:
sys.exit(1)
if __name__ == "__main__": if __name__ == "__main__":