Skip to content
Draft
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 11 additions & 3 deletions replicate/client.py
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,8 @@
import os
import random
import time
from datetime import datetime
from datetime import datetime, timezone
from email.utils import parsedate_to_datetime
from typing import (
TYPE_CHECKING,
Any,
Expand Down Expand Up @@ -277,11 +278,18 @@ def _calculate_sleep(
return float(retry_after_header)

try:
parsed_date = datetime.fromisoformat(retry_after_header).astimezone()
try:
parsed_date = datetime.fromisoformat(
retry_after_header
).astimezone()
except ValueError:
parsed_date = parsedate_to_datetime(retry_after_header)
if parsed_date.tzinfo is None:
parsed_date = parsed_date.replace(tzinfo=timezone.utc)
diff = (parsed_date - datetime.now().astimezone()).total_seconds()
if diff > 0:
return min(diff, self.max_backoff_wait)
except ValueError:
except (TypeError, ValueError, OverflowError):
pass

backoff = self.backoff_factor * (2 ** (attempts_made - 1))
Expand Down
90 changes: 90 additions & 0 deletions tests/test_retry_after.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,90 @@
from datetime import datetime, timezone

import httpx
import pytest

from replicate import client as client_module
from replicate.client import RetryTransport


class FixedDateTime(datetime):
@classmethod
def now(cls, tz=None) -> datetime:
return datetime(2026, 10, 18, 12, 0, 0, tzinfo=timezone.utc).astimezone(tz)


CASES = [
("Sun, 18 Oct 2026 12:00:30 GMT", 30),
("Sunday, 18-Oct-26 12:00:30 GMT", 30),
("Sun Oct 18 12:00:30 2026", 30),
("Sun, 18 Oct 2026 12:01:30 GMT", 60),
("2026-10-18T12:00:30+00:00", 30),
("Sun, 18 Oct 2026 11:59:30 GMT", 0.1),
("invalid", 0.1),
("", 0.1),
("0", 0),
("3", 3),
("120", 120),
]


@pytest.mark.parametrize("header,expected", CASES)
@pytest.mark.parametrize("status", [429, 503])
def test_retry_after_sync(monkeypatch, header, expected, status):
monkeypatch.setattr(client_module, "datetime", FixedDateTime)
sleeps = []
monkeypatch.setattr(client_module.time, "sleep", sleeps.append)
responses = [
httpx.Response(status, headers={"Retry-After": header}),
httpx.Response(200, json={"ok": True}),
]
attempts = []

def respond(request):
attempts.append(request)
return responses[len(attempts) - 1]

with httpx.Client(
transport=RetryTransport(
httpx.MockTransport(respond), max_attempts=2, jitter_ratio=0
)
) as client:
response = client.get("https://example.test/predictions")

assert response.json() == {"ok": True}
assert sleeps == [expected]
assert len(attempts) == 2
assert responses[0].is_closed


@pytest.mark.parametrize("header,expected", CASES)
@pytest.mark.parametrize("status", [429, 503])
async def test_retry_after_async(monkeypatch, header, expected, status):
monkeypatch.setattr(client_module, "datetime", FixedDateTime)
sleeps = []

async def sleep(delay):
sleeps.append(delay)

monkeypatch.setattr(client_module.asyncio, "sleep", sleep)
responses = [
httpx.Response(status, headers={"Retry-After": header}),
httpx.Response(200, json={"ok": True}),
]
attempts = []

def respond(request):
attempts.append(request)
return responses[len(attempts) - 1]

async with httpx.AsyncClient(
transport=RetryTransport(
httpx.MockTransport(respond), max_attempts=2, jitter_ratio=0
)
) as client:
response = await client.get("https://example.test/predictions")

assert response.json() == {"ok": True}
assert sleeps == [expected]
assert len(attempts) == 2
assert responses[0].is_closed