import instructor
from pydantic import BaseModel
from instructor.core.exceptions import IncompleteOutputException, InstructorRetryException


class Report(BaseModel):
    content: str


client = instructor.from_provider("openai/gpt-4.1-mini", mode=instructor.Mode.JSON)

try:
    response = client.create(
        response_model=Report,
        messages=[{"role": "user", "content": "Write a long report..."}],
        max_tokens=50,
        max_retries=0,
    )
except (IncompleteOutputException, InstructorRetryException) as e:
    print(f"Output truncated: {e}")
    print(f"Last completion: {e.last_completion}")

 ---------------------------------------------------------------------------
IncompleteOutputException                 Traceback (most recent call last)
/usr/local/lib/python3.12/dist-packages/instructor/core/retry.py in retry_sync(func, response_model, args, kwargs, context, max_retries, strict, mode, hooks)
    308                         )
--> 309                     raise e
    310     except RetryError as e:

10 frames/usr/local/lib/python3.12/dist-packages/instructor/core/retry.py in retry_sync(func, response_model, args, kwargs, context, max_retries, strict, mode, hooks)
    203 
--> 204                     return process_response(  # type: ignore
    205                         response=response,

/usr/local/lib/python3.12/dist-packages/instructor/processing/response.py in process_response(response, response_model, stream, validation_context, strict, mode)
    369 
--> 370     model = response_model.from_response(  # type: ignore
    371         response,

/usr/local/lib/python3.12/dist-packages/instructor/processing/function_calls.py in from_response(cls, completion, validation_context, strict, mode)
    225         if completion.choices[0].finish_reason == "length":
--> 226             raise IncompleteOutputException(last_completion=completion)
    227 

IncompleteOutputException: The output is incomplete due to a max_tokens length limit.

The above exception was the direct cause of the following exception:

RetryError                                Traceback (most recent call last)
/usr/local/lib/python3.12/dist-packages/instructor/core/retry.py in retry_sync(func, response_model, args, kwargs, context, max_retries, strict, mode, hooks)
    192         response = None
--> 193         for attempt in max_retries:
    194             with attempt:

/usr/local/lib/python3.12/dist-packages/tenacity/__init__.py in __iter__(self)
    437         while True:
--> 438             do = self.iter(retry_state=retry_state)
    439             if isinstance(do, DoAttempt):

/usr/local/lib/python3.12/dist-packages/tenacity/__init__.py in iter(self, retry_state)
    370         for action in self.iter_state.actions:
--> 371             result = action(retry_state)
    372         return result

/usr/local/lib/python3.12/dist-packages/tenacity/__init__.py in exc_check(rs)
    413                     raise retry_exc.reraise()
--> 414                 raise retry_exc from fut.exception()
    415 

RetryError: RetryError[<Future at 0x7ddd7c303050 state=finished raised IncompleteOutputException>]

The above exception was the direct cause of the following exception:

InstructorRetryException                  Traceback (most recent call last)
/tmp/ipykernel_4190/812995971.py in <cell line: 0>()
     13 
     14 try:
---> 15     response = client_realapi.create(
     16         response_model=Report,
     17         messages=[{"role": "user", "content": "Write a long report..."}],

/usr/local/lib/python3.12/dist-packages/instructor/core/client.py in create(self, response_model, messages, max_retries, validation_context, context, strict, hooks, **kwargs)
    374             combined_hooks = self.hooks + hooks
    375 
--> 376         return self.create_fn(
    377             response_model=response_model,
    378             messages=messages,

/usr/local/lib/python3.12/dist-packages/instructor/core/patch.py in new_create_sync(response_model, validation_context, context, max_retries, strict, hooks, *args, **kwargs)
    256                 return obj  # type: ignore[return-value]
    257 
--> 258         response = retry_sync(
    259             func=func,  # type: ignore
    260             response_model=response_model,

/usr/local/lib/python3.12/dist-packages/instructor/core/retry.py in retry_sync(func, response_model, args, kwargs, context, max_retries, strict, mode, hooks)
    310     except RetryError as e:
    311         logger.debug(f"Retry error: {e}")
--> 312         raise InstructorRetryException(
    313             e.last_attempt._exception,
    314             last_completion=response,

InstructorRetryException: <failed_attempts>

<generation number="1">
<exception>
    The output is incomplete due to a max_tokens length limit.
</exception>
<completion>
    ChatCompletion(id='chatcmpl-DUz7hihgylbiAnmGyYCIn5w3tyrHr', choices=[Choice(finish_reason='length', index=0, logprobs=None, message=ChatCompletionMessage(content='{\n  "content": "This report provides a comprehensive overview of the subject matter at hand. It delves into various aspects, analyzing data and presenting insights to aid in understanding and decision-making. The report begins with an introduction that outlines the purpose and', refusal=None, role='assistant', annotations=[], audio=None, function_call=None, tool_calls=None))], created=1776277169, model='gpt-4.1-mini-2025-04-14', object='chat.completion', service_tier='default', system_fingerprint='fp_283a574ac4', usage=CompletionUsage(completion_tokens=50, prompt_tokens=98, total_tokens=148, completion_tokens_details=CompletionTokensDetails(accepted_prediction_tokens=None, audio_tokens=0, reasoning_tokens=0, rejected_prediction_tokens=None), prompt_tokens_details=PromptTokensDetails(audio_tokens=0, cached_tokens=0)))
</completion>
</generation>

</failed_attempts>

<last_exception>
    The output is incomplete due to a max_tokens length limit.
</last_exception>

except InstructorRetryException as e:
    # The actual exception (IncompleteOutputException) is wrapped inside InstructorRetryException
    inner_error = e.args[0] if e.args else None
    
    if isinstance(inner_error, IncompleteOutputException):
        print(f"Caught expected IncompleteOutputException via InstructorRetryException!")
        print(f"Output truncated: {inner_error}")
        print(f"Last completion: {inner_error.last_completion}")
    else:
        print(f"Caught InstructorRetryException with different inner error: {type(inner_error)}")
        print(e)
except IncompleteOutputException as e:
    # This block might still be useful if the library behavior changes or in different modes
    print(f"Caught IncompleteOutputException directly: {e}")
    print(f"Last completion: {e.last_completion}")

Caught expected IncompleteOutputException via InstructorRetryException!
Output truncated: The output is incomplete due to a max_tokens length limit.
Last completion: ChatCompletion(id='chatcmpl-DUynRyJRGqwHszFrHydyojcbuZag6', choices=[Choice(finish_reason='length', index=0, logprobs=None, message=ChatCompletionMessage(content='{\n  "content":', refusal=None, role='assistant', annotations=[], audio=None, function_call=None, tool_calls=None))], created=1776275913, model='gpt-4.1-mini-2025-04-14', object='chat.completion', service_tier='default', system_fingerprint='fp_283a574ac4', usage=CompletionUsage(completion_tokens=5, prompt_tokens=98, total_tokens=103, completion_tokens_details=CompletionTokensDetails(accepted_prediction_tokens=None, audio_tokens=0, reasoning_tokens=0, rejected_prediction_tokens=None), prompt_tokens_details=PromptTokensDetails(audio_tokens=0, cached_tokens=0)))