Merge branch 'main' into async

This commit is contained in:
Paul Gauthier 2024-08-08 15:23:19 -03:00
commit 4abb2e78b6
3 changed files with 41 additions and 55 deletions

View file

@ -30,7 +30,7 @@ from aider.llm import litellm
from aider.mdstream import MarkdownStream
from aider.repo import GitRepo
from aider.repomap import RepoMap
from aider.sendchat import send_with_retries
from aider.sendchat import retry_exceptions, send_completion
from aider.utils import format_content, format_messages, is_image_file
from ..dump import dump # noqa: F401
@ -70,6 +70,7 @@ class Coder:
lint_outcome = None
test_outcome = None
multi_response_content = ""
partial_response_content = ""
partial_response_content = None
@ -895,6 +896,8 @@ class Coder:
else:
self.mdstream = None
retry_delay = 0.125
self.usage_report = None
exhausted = False
interrupted = False
@ -903,6 +906,14 @@ class Coder:
try:
yield from self.send(messages, functions=self.functions)
break
except retry_exceptions() as err:
self.io.tool_error(str(err))
retry_delay *= 2
if retry_delay > 60:
break
self.io.tool_output(f"Retrying in {retry_delay:.1f} seconds...")
time.sleep(retry_delay)
continue
except KeyboardInterrupt:
interrupted = True
break
@ -1165,7 +1176,7 @@ class Coder:
interrupted = False
try:
hash_object, completion = send_with_retries(
hash_object, completion = send_completion(
model.name,
messages,
functions,

View file

@ -14,53 +14,28 @@ CACHE = None
# CACHE = Cache(CACHE_PATH)
def retry_exceptions():
import httpx
return (
httpx.ConnectError,
httpx.RemoteProtocolError,
httpx.ReadTimeout,
litellm.exceptions.APIConnectionError,
litellm.exceptions.APIError,
litellm.exceptions.RateLimitError,
litellm.exceptions.ServiceUnavailableError,
litellm.exceptions.Timeout,
litellm.exceptions.InternalServerError,
litellm.llms.anthropic.AnthropicError,
)
def lazy_litellm_retry_decorator(func):
def wrapper(*args, **kwargs):
import httpx
def should_giveup(e):
if not hasattr(e, "status_code"):
return False
if type(e) in (
httpx.ConnectError,
httpx.RemoteProtocolError,
httpx.ReadTimeout,
litellm.exceptions.APIConnectionError,
litellm.exceptions.APIError,
litellm.exceptions.RateLimitError,
litellm.exceptions.ServiceUnavailableError,
litellm.exceptions.Timeout,
litellm.exceptions.InternalServerError,
litellm.llms.anthropic.AnthropicError,
):
return False
# These seem to return .status_code = ""
# litellm._should_retry() expects an int and throws a TypeError
#
# litellm.llms.anthropic.AnthropicError
# litellm.exceptions.APIError
if not e.status_code:
return False
return not litellm._should_retry(e.status_code)
decorated_func = backoff.on_exception(
backoff.expo,
(
httpx.ConnectError,
httpx.RemoteProtocolError,
httpx.ReadTimeout,
litellm.exceptions.APIConnectionError,
litellm.exceptions.APIError,
litellm.exceptions.RateLimitError,
litellm.exceptions.ServiceUnavailableError,
litellm.exceptions.Timeout,
litellm.exceptions.InternalServerError,
litellm.llms.anthropic.AnthropicError,
),
giveup=should_giveup,
retry_exceptions(),
max_time=60,
on_backoff=lambda details: print(
f"{details.get('exception', 'Exception')}\nRetry in {details['wait']:.1f} seconds."
@ -71,8 +46,7 @@ def lazy_litellm_retry_decorator(func):
return wrapper
@lazy_litellm_retry_decorator
def send_with_retries(
def send_completion(
model_name, messages, functions, stream, temperature=0, extra_headers=None, max_tokens=None
):
from aider.llm import litellm
@ -108,9 +82,10 @@ def send_with_retries(
return hash_object, res
@lazy_litellm_retry_decorator
def simple_send_with_retries(model_name, messages):
try:
_hash, response = send_with_retries(
_hash, response = send_completion(
model_name=model_name,
messages=messages,
functions=None,

View file

@ -4,7 +4,7 @@ from unittest.mock import MagicMock, patch
import httpx
from aider.llm import litellm
from aider.sendchat import send_with_retries
from aider.sendchat import simple_send_with_retries
class PrintCalled(Exception):
@ -14,7 +14,7 @@ class PrintCalled(Exception):
class TestSendChat(unittest.TestCase):
@patch("litellm.completion")
@patch("builtins.print")
def test_send_with_retries_rate_limit_error(self, mock_print, mock_completion):
def test_simple_send_with_retries_rate_limit_error(self, mock_print, mock_completion):
mock = MagicMock()
mock.status_code = 500
@ -29,19 +29,19 @@ class TestSendChat(unittest.TestCase):
None,
]
# Call the send_with_retries method
send_with_retries("model", ["message"], None, False)
# Call the simple_send_with_retries method
simple_send_with_retries("model", ["message"])
mock_print.assert_called_once()
@patch("litellm.completion")
@patch("builtins.print")
def test_send_with_retries_connection_error(self, mock_print, mock_completion):
def test_simple_send_with_retries_connection_error(self, mock_print, mock_completion):
# Set up the mock to raise
mock_completion.side_effect = [
httpx.ConnectError("Connection error"),
None,
]
# Call the send_with_retries method
send_with_retries("model", ["message"], None, False)
# Call the simple_send_with_retries method
simple_send_with_retries("model", ["message"])
mock_print.assert_called_once()