learningcircuit--local-deep-research/tests/rate_limiting/test_llm_detection_extended.py

"""
Extended tests for rate_limiting/llm/detection.py — provider-specific detection.

The existing test_llm_rate_limiting.py covers basic patterns but misses:
- OpenAI-specific module check (line 52)
- Anthropic-specific module check (line 58)
- extract_retry_after edge cases (date headers, decimals, multiple patterns)

Tests cover:
- OpenAI provider-specific detection (module == "openai")
- Anthropic provider-specific detection ("anthropic" in module)
- extract_retry_after edge cases
"""

from unittest.mock import Mock


def _make_exception(class_name, module, message):
    """Create an exception instance with a custom __class__.__module__ and __name__.

    Python builtins don't support __class__ reassignment, so we create a new
    exception type with the desired module and instantiate directly.
    """
    cls = type(class_name, (Exception,), {"__module__": module})
    return cls(message)


# ---------------------------------------------------------------------------
# OpenAI-specific detection
# ---------------------------------------------------------------------------


class TestOpenAISpecificDetection:
    """Tests for OpenAI provider-specific rate limit detection."""

    def test_openai_module_ratelimiterror_429_detected(self):
        """OpenAI module + ratelimiterror type + '429' -> True."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            is_llm_rate_limit_error,
        )

        error = _make_exception("RateLimitError", "openai", "Error code: 429")
        assert is_llm_rate_limit_error(error) is True

    def test_openai_module_non_429_not_detected(self):
        """OpenAI module but non-rate-limit error -> not detected."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            is_llm_rate_limit_error,
        )

        error = _make_exception(
            "AuthenticationError", "openai", "Invalid API key"
        )
        assert is_llm_rate_limit_error(error) is False

    def test_openai_apierror_with_429_detected(self):
        """OpenAI ApiError with '429' in message -> True."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            is_llm_rate_limit_error,
        )

        error = _make_exception(
            "ApiError", "openai", "Request failed with status 429"
        )
        assert is_llm_rate_limit_error(error) is True


# ---------------------------------------------------------------------------
# Anthropic-specific detection
# ---------------------------------------------------------------------------


class TestAnthropicSpecificDetection:
    """Tests for Anthropic provider-specific rate limit detection."""

    def test_anthropic_module_rate_limit_detected(self):
        """'anthropic' in module + 'rate_limit' in message -> True."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            is_llm_rate_limit_error,
        )

        error = _make_exception(
            "RateLimitError",
            "anthropic._exceptions",
            "rate_limit_error: too many requests",
        )
        assert is_llm_rate_limit_error(error) is True

    def test_anthropic_module_429_detected(self):
        """'anthropic' in module + '429' in message -> True."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            is_llm_rate_limit_error,
        )

        error = _make_exception(
            "ApiStatusError",
            "anthropic._exceptions",
            "HTTP 429 from Anthropic API",
        )
        assert is_llm_rate_limit_error(error) is True

    def test_anthropic_module_unrelated_error_not_detected(self):
        """'anthropic' in module + unrelated message -> False."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            is_llm_rate_limit_error,
        )

        error = _make_exception(
            "AuthenticationError",
            "anthropic._exceptions",
            "Invalid API key provided",
        )
        assert is_llm_rate_limit_error(error) is False


# ---------------------------------------------------------------------------
# extract_retry_after edge cases
# ---------------------------------------------------------------------------


class TestExtractRetryAfterEdgeCases:
    """Edge cases for extract_retry_after."""

    def test_non_numeric_retry_after_header(self):
        """Non-numeric Retry-After header (date string) is handled gracefully."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            extract_retry_after,
        )

        error = Mock()
        error.response = Mock()
        error.response.headers = {
            "Retry-After": "Wed, 21 Oct 2015 07:28:00 GMT"
        }
        error.__str__ = lambda self: "Rate limited"

        # Date string can't be parsed as float, falls through to message parsing
        result = extract_retry_after(error)
        assert isinstance(result, (int, float))

    def test_decimal_seconds_in_message(self):
        """'retry after 30.5 seconds' with decimal -> 30.5."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            extract_retry_after,
        )

        error = Exception("Please retry after 30.5 seconds")

        result = extract_retry_after(error)
        assert result == 30.5

    def test_multiple_patterns_returns_first_match(self):
        """Multiple patterns in same message -> first match returned."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            extract_retry_after,
        )

        # "try again in X seconds" pattern comes before "wait X seconds"
        error = Exception(
            "Rate limited. Try again in 15 seconds. Or wait 30 seconds."
        )

        result = extract_retry_after(error)
        assert result == 15.0

    def test_no_patterns_returns_zero(self):
        """No matching patterns -> returns 0."""
        from local_deep_research.web_search_engines.rate_limiting.llm.detection import (
            extract_retry_after,
        )

        error = Exception("Something went wrong with no timing info")

        result = extract_retry_after(error)
        assert result == 0