/tests/search_engines/test_search_engine_zenodo.py

"""
Tests for the Zenodo search engine.
Tests initialization, search functionality, and error handling.
"""

import pytest
from unittest.mock import Mock


class TestZenodoSearchEngineInit:
    """Tests for Zenodo search engine initialization."""

    def test_init_default_parameters(self):
        """Test initialization with default parameters."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()

        assert engine.max_results == 10
        assert engine.resource_type is None
        assert engine.access_right is None
        assert engine.communities is None
        assert engine.sort == "bestmatch"
        assert engine.is_public is True
        assert engine.is_scientific is True

    def test_init_custom_parameters(self):
        """Test initialization with custom parameters."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine(
            max_results=50,
            resource_type="dataset",
            access_right="open",
            communities="zenodo",
            sort="mostrecent",
        )

        assert engine.max_results == 50
        assert engine.resource_type == "dataset"
        assert engine.access_right == "open"
        assert engine.communities == "zenodo"
        assert engine.sort == "mostrecent"

    def test_base_url_set(self):
        """Test that API base URL is correctly set."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()
        assert engine.base_url == "https://zenodo.org"
        assert engine.search_url == "https://zenodo.org/api/records"

    def test_user_agent_header_set(self):
        """Test that User-Agent header is set."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()
        assert "User-Agent" in engine.headers
        assert "Local-Deep-Research" in engine.headers["User-Agent"]


class TestZenodoQueryBuilding:
    """Tests for Zenodo query parameter building."""

    def test_build_query_params_basic(self):
        """Test basic query params building."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine(max_results=20)
        params = engine._build_query_params("machine learning")

        assert params["q"] == "machine learning"
        assert params["size"] == 20
        assert params["sort"] == "bestmatch"

    def test_build_query_params_with_type(self):
        """Test query params with resource type filter."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine(resource_type="dataset")
        params = engine._build_query_params("climate data")

        assert params["type"] == "dataset"

    def test_build_query_params_with_access(self):
        """Test query params with access right filter."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine(access_right="open")
        params = engine._build_query_params("test")

        assert params["access_right"] == "open"

    def test_build_query_params_with_communities(self):
        """Test query params with communities filter."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine(communities="astronomy")
        params = engine._build_query_params("test")

        assert params["communities"] == "astronomy"


class TestZenodoCreatorParsing:
    """Tests for Zenodo creator parsing."""

    def test_parse_creators_basic(self):
        """Test parsing creators."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()
        creators = [
            {"name": "John Smith"},
            {"name": "Jane Doe"},
        ]

        result = engine._parse_creators(creators)
        assert result == ["John Smith", "Jane Doe"]

    def test_parse_creators_limits_to_five(self):
        """Test that creator parsing limits to 5."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()
        creators = [{"name": f"Author{i}"} for i in range(10)]

        result = engine._parse_creators(creators)
        assert len(result) == 5

    def test_parse_creators_empty(self):
        """Test parsing empty creators list."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()
        result = engine._parse_creators([])
        assert result == []


class TestZenodoResourceType:
    """Tests for Zenodo resource type handling."""

    def test_get_resource_type_label_with_title(self):
        """Test resource type label extraction."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()
        resource_type = {"title": "Dataset", "type": "dataset"}

        result = engine._get_resource_type_label(resource_type)
        assert result == "Dataset"

    def test_get_resource_type_label_fallback(self):
        """Test resource type label fallback to type."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()
        resource_type = {"type": "software"}

        result = engine._get_resource_type_label(resource_type)
        assert result == "software"

    def test_get_resource_type_label_empty(self):
        """Test resource type label with empty dict."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()
        result = engine._get_resource_type_label({})
        assert result == "Unknown"


class TestZenodoSearchExecution:
    """Tests for Zenodo search execution."""

    @pytest.fixture
    def engine(self):
        """Create a Zenodo engine."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        return ZenodoSearchEngine(max_results=10)

    def test_get_previews_success(self, engine, monkeypatch):
        """Test successful preview retrieval."""
        mock_response = Mock()
        mock_response.status_code = 200
        mock_response.json = Mock(
            return_value={
                "hits": {
                    "total": 100,
                    "hits": [
                        {
                            "id": 12345,
                            "metadata": {
                                "title": "Test Dataset",
                                "creators": [{"name": "John Smith"}],
                                "description": "<p>A test dataset</p>",
                                "doi": "10.5281/zenodo.12345",
                                "publication_date": "2024-01-01",
                                "resource_type": {
                                    "title": "Dataset",
                                    "type": "dataset",
                                },
                                "access_right": "open",
                                "keywords": ["test", "data"],
                                "license": {"id": "cc-by-4.0"},
                            },
                            "links": {
                                "self_html": "https://zenodo.org/records/12345",
                                "doi": "https://doi.org/10.5281/zenodo.12345",
                            },
                        }
                    ],
                }
            }
        )
        mock_response.raise_for_status = Mock()

        monkeypatch.setattr(
            "local_deep_research.web_search_engines.engines.search_engine_zenodo.safe_get",
            Mock(return_value=mock_response),
        )

        previews = engine._get_previews("test query")

        assert len(previews) == 1
        assert previews[0]["title"] == "Test Dataset"
        assert "John Smith" in previews[0]["authors"]
        assert previews[0]["doi"] == "10.5281/zenodo.12345"
        assert previews[0]["source"] == "Zenodo"
        assert previews[0]["resource_type"] == "Dataset"

    def test_get_previews_rate_limit_error(self, engine, monkeypatch):
        """Test that 429 errors raise RateLimitError."""
        from local_deep_research.web_search_engines.rate_limiting import (
            RateLimitError,
        )

        mock_response = Mock()
        mock_response.status_code = 429

        monkeypatch.setattr(
            "local_deep_research.web_search_engines.engines.search_engine_zenodo.safe_get",
            Mock(return_value=mock_response),
        )

        with pytest.raises(RateLimitError):
            engine._get_previews("test")

    def test_get_previews_handles_exception(self, engine, monkeypatch):
        """Test that exceptions are handled gracefully."""
        import requests

        monkeypatch.setattr(
            "local_deep_research.web_search_engines.engines.search_engine_zenodo.safe_get",
            Mock(side_effect=requests.RequestException("Network error")),
        )

        previews = engine._get_previews("test")
        assert previews == []


class TestZenodoEdgeCases:
    """Tests for Zenodo edge cases and error handling."""

    @pytest.fixture
    def engine(self):
        """Create a Zenodo engine."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        return ZenodoSearchEngine(max_results=10)

    def test_get_previews_empty_results(self, engine, monkeypatch):
        """Test handling of empty search results."""
        mock_response = Mock()
        mock_response.status_code = 200
        mock_response.json = Mock(
            return_value={
                "hits": {
                    "total": 0,
                    "hits": [],
                }
            }
        )
        mock_response.raise_for_status = Mock()

        monkeypatch.setattr(
            "local_deep_research.web_search_engines.engines.search_engine_zenodo.safe_get",
            Mock(return_value=mock_response),
        )

        previews = engine._get_previews("xyznonexistentquery12345")
        assert previews == []

    def test_get_previews_html_in_description(self, engine, monkeypatch):
        """Test that HTML is stripped from description."""
        mock_response = Mock()
        mock_response.status_code = 200
        mock_response.json = Mock(
            return_value={
                "hits": {
                    "total": 1,
                    "hits": [
                        {
                            "id": 12345,
                            "metadata": {
                                "title": "Test Record",
                                "description": "<p>This is <strong>bold</strong> and <em>italic</em> text.</p>",
                                "creators": [{"name": "Test Author"}],
                            },
                            "links": {},
                        }
                    ],
                }
            }
        )
        mock_response.raise_for_status = Mock()

        monkeypatch.setattr(
            "local_deep_research.web_search_engines.engines.search_engine_zenodo.safe_get",
            Mock(return_value=mock_response),
        )

        previews = engine._get_previews("test")
        assert len(previews) == 1
        # HTML should be stripped
        assert "<p>" not in previews[0]["description"]
        assert "<strong>" not in previews[0]["description"]
        assert "bold" in previews[0]["description"]

    def test_get_previews_html_entities_decoded(self, engine, monkeypatch):
        """Test that HTML entities are decoded."""
        mock_response = Mock()
        mock_response.status_code = 200
        mock_response.json = Mock(
            return_value={
                "hits": {
                    "total": 1,
                    "hits": [
                        {
                            "id": 12345,
                            "metadata": {
                                "title": "Test Record",
                                "description": "Test &amp; verify &lt;data&gt;",
                                "creators": [],
                            },
                            "links": {},
                        }
                    ],
                }
            }
        )
        mock_response.raise_for_status = Mock()

        monkeypatch.setattr(
            "local_deep_research.web_search_engines.engines.search_engine_zenodo.safe_get",
            Mock(return_value=mock_response),
        )

        previews = engine._get_previews("test")
        assert len(previews) == 1
        # HTML entities should be decoded
        assert "&amp;" not in previews[0]["description"]
        assert "&" in previews[0]["description"]

    def test_get_previews_missing_optional_fields(self, engine, monkeypatch):
        """Test handling of missing optional fields."""
        mock_response = Mock()
        mock_response.status_code = 200
        mock_response.json = Mock(
            return_value={
                "hits": {
                    "total": 1,
                    "hits": [
                        {
                            "id": 12345,
                            "metadata": {
                                "title": "Minimal Record",
                                # Missing: creators, description, doi, etc.
                            },
                            "links": {},
                        }
                    ],
                }
            }
        )
        mock_response.raise_for_status = Mock()

        monkeypatch.setattr(
            "local_deep_research.web_search_engines.engines.search_engine_zenodo.safe_get",
            Mock(return_value=mock_response),
        )

        previews = engine._get_previews("test")
        assert len(previews) == 1
        assert previews[0]["title"] == "Minimal Record"
        assert previews[0]["authors"] == []
        assert previews[0]["doi"] == ""

    def test_get_previews_unicode_in_title(self, engine, monkeypatch):
        """Test handling of Unicode characters in title."""
        mock_response = Mock()
        mock_response.status_code = 200
        mock_response.json = Mock(
            return_value={
                "hits": {
                    "total": 1,
                    "hits": [
                        {
                            "id": 12345,
                            "metadata": {
                                "title": "Données géographiques français",
                                "creators": [{"name": "François Müller"}],
                            },
                            "links": {},
                        }
                    ],
                }
            }
        )
        mock_response.raise_for_status = Mock()

        monkeypatch.setattr(
            "local_deep_research.web_search_engines.engines.search_engine_zenodo.safe_get",
            Mock(return_value=mock_response),
        )

        previews = engine._get_previews("geographic data")
        assert len(previews) == 1
        assert "géographiques" in previews[0]["title"]
        assert "François Müller" in previews[0]["authors"]

    def test_get_full_content_without_raw(self, engine):
        """Test full content handling when _raw is missing."""
        items = [
            {
                "title": "Test Record",
                "authors": ["Test Author"],
            }
        ]

        results = engine._get_full_content(items)
        assert len(results) == 1
        # Should not crash

    def test_all_sort_options(self):
        """Test that various sort options can be used."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        sort_options = ["bestmatch", "mostrecent", "-mostrecent"]
        for sort in sort_options:
            engine = ZenodoSearchEngine(sort=sort)
            assert engine.sort == sort

    def test_all_access_rights(self):
        """Test that various access rights can be used."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        access_rights = [None, "open", "closed", "embargoed", "restricted"]
        for access in access_rights:
            engine = ZenodoSearchEngine(access_right=access)
            assert engine.access_right == access

    def test_all_resource_types(self):
        """Test that various resource types can be used."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        types = [
            None,
            "dataset",
            "software",
            "publication",
            "poster",
            "presentation",
        ]
        for resource_type in types:
            engine = ZenodoSearchEngine(resource_type=resource_type)
            assert engine.resource_type == resource_type


class TestZenodoFullContent:
    """Tests for Zenodo full content retrieval."""

    def test_get_full_content_builds_content(self):
        """Test that full content builds proper content string."""
        from local_deep_research.web_search_engines.engines.search_engine_zenodo import (
            ZenodoSearchEngine,
        )

        engine = ZenodoSearchEngine()

        items = [
            {
                "title": "Test Dataset",
                "authors": ["Author One", "Author Two"],
                "doi": "10.5281/zenodo.12345",
                "publication_date": "2024-01-01",
                "resource_type": "Dataset",
                "keywords": ["test", "data", "science"],
                "license": "cc-by-4.0",
                "description": "A test dataset for testing.",
                "_raw": {
                    "metadata": {
                        "description": "Full description of the dataset.",
                        "keywords": ["test", "data", "science", "research"],
                        "related_identifiers": [],
                        "references": [],
                    },
                    "files": [
                        {
                            "key": "data.csv",
                            "size": 1024,
                            # Zenodo API returns checksums - using sha256 for test data
                            "checksum": "sha256:abc123def456",
                        }
                    ],
                },
            }
        ]

        results = engine._get_full_content(items)

        assert len(results) == 1
        assert "Authors: Author One, Author Two" in results[0]["content"]
        assert "DOI: 10.5281/zenodo.12345" in results[0]["content"]
        assert "_raw" not in results[0]
        assert len(results[0]["files"]) == 1