/tests/benchmarks/test_xbench_deepsearch_pure_logic.py

"""
Tests for XBenchDeepSearchDataset pure logic methods.

Existing tests: none — this file is the first test coverage.

Covers:
- xor_decrypt: symmetric encryption, empty data, key cycling, single byte
- get_dataset_info: expected keys and values
- get_default_dataset_path: exact string
- process_example: dict enrichment, shallow copy, evaluation criteria weights
"""

import base64


def _cls():
    """Import and return the XBenchDeepSearchDataset class."""
    from local_deep_research.benchmarks.datasets.xbench_deepsearch import (
        XBenchDeepSearchDataset,
    )

    return XBenchDeepSearchDataset


class TestXorDecrypt:
    """Tests for xor_decrypt static method."""

    def test_symmetry(self):
        """XOR decrypt is symmetric: decrypt(encrypt(x)) == x."""
        cls = _cls()
        plaintext = b"Hello, World!"
        key = "secret"
        encrypted = cls.xor_decrypt(plaintext, key)
        decrypted = cls.xor_decrypt(encrypted, key)
        assert decrypted == plaintext

    def test_known_values(self):
        """Known plaintext/key produces expected ciphertext."""
        cls = _cls()
        # XOR of b'A' (0x41) with b'k' (0x6B) = 0x2A = b'*'
        result = cls.xor_decrypt(b"A", "k")
        assert result == bytes([0x41 ^ 0x6B])

    def test_empty_data(self):
        """Empty data returns empty bytes."""
        cls = _cls()
        result = cls.xor_decrypt(b"", "anykey")
        assert result == b""

    def test_key_cycling(self):
        """Key cycles through for data longer than key."""
        cls = _cls()
        data = b"ABCD"
        key = "ab"
        # A^a, B^b, C^a, D^b
        expected = bytes(
            [
                ord("A") ^ ord("a"),
                ord("B") ^ ord("b"),
                ord("C") ^ ord("a"),
                ord("D") ^ ord("b"),
            ]
        )
        result = cls.xor_decrypt(data, key)
        assert result == expected

    def test_single_byte_key(self):
        """Single byte key applied to every byte."""
        cls = _cls()
        data = b"HELLO"
        key = "X"
        expected = bytes([b ^ ord("X") for b in data])
        result = cls.xor_decrypt(data, key)
        assert result == expected

    def test_all_zeros(self):
        """XOR with all-zero data returns key bytes repeated."""
        cls = _cls()
        data = bytes(5)  # 5 zero bytes
        key = "AB"
        # 0^A, 0^B, 0^A, 0^B, 0^A
        expected = bytes([ord("A"), ord("B"), ord("A"), ord("B"), ord("A")])
        result = cls.xor_decrypt(data, key)
        assert result == expected

    def test_unicode_key(self):
        """Unicode key is encoded as UTF-8."""
        cls = _cls()
        data = b"\x00\x00\x00"
        key = "\u00e9"  # é = 0xC3 0xA9 in UTF-8 (2 bytes)
        key_bytes = key.encode("utf-8")
        expected = bytes(
            [
                0x00 ^ key_bytes[0],
                0x00 ^ key_bytes[1],
                0x00 ^ key_bytes[0],
            ]
        )
        result = cls.xor_decrypt(data, key)
        assert result == expected

    def test_roundtrip_with_base64(self):
        """Full roundtrip: plaintext -> encrypt -> base64 -> decode -> decrypt."""
        cls = _cls()
        plaintext = b"This is a test question for benchmarking"
        key = "canary_key_123"
        encrypted = cls.xor_decrypt(plaintext, key)
        encoded = base64.b64encode(encrypted)
        decoded = base64.b64decode(encoded)
        decrypted = cls.xor_decrypt(decoded, key)
        assert decrypted == plaintext


class TestGetDatasetInfo:
    """Tests for get_dataset_info class method."""

    def test_returns_dict(self):
        """Returns a dictionary."""
        cls = _cls()
        info = cls.get_dataset_info()
        assert isinstance(info, dict)

    def test_has_required_keys(self):
        """Contains id, name, description, and url."""
        cls = _cls()
        info = cls.get_dataset_info()
        assert "id" in info
        assert "name" in info
        assert "description" in info
        assert "url" in info

    def test_id_value(self):
        """Dataset ID is 'xbench_deepsearch'."""
        cls = _cls()
        info = cls.get_dataset_info()
        assert info["id"] == "xbench_deepsearch"

    def test_name_value(self):
        """Dataset name is 'xbench-DeepSearch'."""
        cls = _cls()
        info = cls.get_dataset_info()
        assert info["name"] == "xbench-DeepSearch"

    def test_url_is_huggingface(self):
        """URL points to Hugging Face."""
        cls = _cls()
        info = cls.get_dataset_info()
        assert "huggingface.co" in info["url"]

    def test_callable_on_class_and_instance(self):
        """Can be called on both class and instance."""
        cls = _cls()
        class_result = cls.get_dataset_info()
        instance_result = cls().get_dataset_info()
        assert class_result == instance_result


class TestGetDefaultDatasetPath:
    """Tests for get_default_dataset_path class method."""

    def test_returns_string(self):
        """Returns a string."""
        cls = _cls()
        path = cls.get_default_dataset_path()
        assert isinstance(path, str)

    def test_exact_value(self):
        """Returns the expected HuggingFace path."""
        cls = _cls()
        path = cls.get_default_dataset_path()
        assert path == "xbench/DeepSearch"


class TestProcessExample:
    """Tests for process_example instance method."""

    def test_adds_requires_deep_search(self):
        """Adds requires_deep_search=True."""
        cls = _cls()
        instance = cls()
        result = instance.process_example({"id": "q1", "problem": "test"})
        assert result["requires_deep_search"] is True

    def test_adds_expected_iterations(self):
        """Adds expected_iterations=4."""
        cls = _cls()
        instance = cls()
        result = instance.process_example({"id": "q1"})
        assert result["expected_iterations"] == 4

    def test_adds_evaluation_criteria(self):
        """Adds evaluation_criteria dict."""
        cls = _cls()
        instance = cls()
        result = instance.process_example({"id": "q1"})
        criteria = result["evaluation_criteria"]
        assert "accuracy" in criteria
        assert "completeness" in criteria
        assert "reasoning" in criteria
        assert "sources" in criteria

    def test_evaluation_criteria_weights_sum_to_one(self):
        """Evaluation criteria weights sum to 1.0."""
        cls = _cls()
        instance = cls()
        result = instance.process_example({"id": "q1"})
        total = sum(result["evaluation_criteria"].values())
        assert abs(total - 1.0) < 1e-9

    def test_preserves_original_fields(self):
        """Original example fields are preserved."""
        cls = _cls()
        instance = cls()
        original = {"id": "q42", "problem": "What is X?", "answer": "Y"}
        result = instance.process_example(original)
        assert result["id"] == "q42"
        assert result["problem"] == "What is X?"
        assert result["answer"] == "Y"

    def test_does_not_modify_original(self):
        """Original dict is not modified (shallow copy)."""
        cls = _cls()
        instance = cls()
        original = {"id": "q1", "problem": "test"}
        instance.process_example(original)
        assert "requires_deep_search" not in original
        assert "evaluation_criteria" not in original

    def test_empty_input(self):
        """Empty dict gets all enrichment fields."""
        cls = _cls()
        instance = cls()
        result = instance.process_example({})
        assert result["requires_deep_search"] is True
        assert result["expected_iterations"] == 4
        assert "evaluation_criteria" in result

    def test_existing_field_overwritten(self):
        """If input has 'requires_deep_search', it gets overwritten."""
        cls = _cls()
        instance = cls()
        result = instance.process_example({"requires_deep_search": False})
        assert result["requires_deep_search"] is True


class TestXBenchInheritance:
    """Tests for XBenchDeepSearchDataset class structure."""

    def test_inherits_from_benchmark_dataset(self):
        """Inherits from BenchmarkDataset base class."""
        cls = _cls()
        from local_deep_research.benchmarks.datasets.base import (
            BenchmarkDataset,
        )

        assert issubclass(cls, BenchmarkDataset)

    def test_instantiation(self):
        """Can be instantiated without arguments."""
        cls = _cls()
        instance = cls()
        assert instance is not None
        assert instance.dataset_path == "xbench/DeepSearch"