learningcircuit--local-deep-research/tests/programmatic_access/test_programmatic_access.py

#!/usr/bin/env python3
"""
Test programmatic access to Local Deep Research without database dependencies.
"""

from unittest.mock import Mock
from langchain_core.retrievers import Document


def test_import_without_database():
    """Test that we can import AdvancedSearchSystem without database initialization."""
    # This should not fail with database errors
    from local_deep_research.search_system import AdvancedSearchSystem

    # Should be able to create an instance with mock components
    llm = Mock()
    search = Mock()

    # Create settings snapshot without programmatic_mode
    settings_snapshot = {
        "search.iterations": {"value": 1, "type": "int"},
        "search.questions_per_iteration": {"value": 2, "type": "int"},
        "search.strategy": {"value": "direct", "type": "str"},
        "search.max_results_per_query": {"value": 10, "type": "int"},
        "search.source_strategy.diversity_threshold": {
            "value": 0.8,
            "type": "float",
        },
        "search.source_strategy.min_relevance_score": {
            "value": 0.5,
            "type": "float",
        },
        "search.source_strategy.max_sources_per_topic": {
            "value": 5,
            "type": "int",
        },
        "search.source_strategy.enable_clustering": {
            "value": False,
            "type": "bool",
        },
        "search.cross_engine_max_results": {"value": 100, "type": "int"},
    }

    # Pass programmatic_mode as explicit parameter
    system = AdvancedSearchSystem(
        llm=llm,
        search=search,
        settings_snapshot=settings_snapshot,
        programmatic_mode=True,
    )

    assert system is not None
    assert system.model == llm
    assert system.search == search
    assert system.programmatic_mode is True


def test_analyze_topic_without_database():
    """Test analyze_topic function without database."""
    from local_deep_research.search_system import AdvancedSearchSystem

    # Create mock LLM
    llm = Mock()
    llm.invoke.return_value = Mock(
        content="This is a summary about AI research."
    )

    # Create mock search engine
    search = Mock()
    search.run.return_value = [
        {
            "title": "AI Research Paper",
            "link": "http://example.com/ai",
            "snippet": "Recent advances in AI...",
            "full_content": "Full content about AI research...",
            "rank": 1,
        }
    ]

    # Create settings snapshot without programmatic_mode
    settings_snapshot = {
        "search.iterations": {"value": 1, "type": "int"},
        "search.questions_per_iteration": {"value": 2, "type": "int"},
        "search.strategy": {"value": "direct", "type": "str"},
        "search.max_results_per_query": {"value": 10, "type": "int"},
        "search.source_strategy.diversity_threshold": {
            "value": 0.8,
            "type": "float",
        },
        "search.source_strategy.min_relevance_score": {
            "value": 0.5,
            "type": "float",
        },
        "search.source_strategy.max_sources_per_topic": {
            "value": 5,
            "type": "int",
        },
        "search.source_strategy.enable_clustering": {
            "value": False,
            "type": "bool",
        },
        "search.cross_engine_max_results": {"value": 100, "type": "int"},
    }

    # Create system with programmatic_mode as parameter
    system = AdvancedSearchSystem(
        llm=llm,
        search=search,
        settings_snapshot=settings_snapshot,
        programmatic_mode=True,
    )

    # Should be able to call analyze_topic
    result = system.analyze_topic("What is AI?")

    print(f"Result: {result}")
    print(f"Search called: {search.run.called}")
    print(f"Search call count: {search.run.call_count}")

    assert result is not None
    assert "findings" in result


def test_search_with_retriever():
    """Test using a retriever as search engine."""
    from local_deep_research.search_system import AdvancedSearchSystem
    from langchain_community.vectorstores import FAISS
    from langchain_community.embeddings import FakeEmbeddings

    # Create a simple retriever
    documents = [
        Document(
            page_content="Machine learning is a subset of artificial intelligence.",
            metadata={"source": "ml_intro.txt"},
        ),
        Document(
            page_content="Deep learning uses neural networks with multiple layers.",
            metadata={"source": "dl_intro.txt"},
        ),
    ]

    embeddings = FakeEmbeddings(size=10)
    vectorstore = FAISS.from_documents(documents, embeddings)
    retriever = vectorstore.as_retriever()

    # Create retriever wrapper
    class SimpleRetrieverWrapper:
        def __init__(self, retriever, settings_snapshot=None):
            self.retriever = retriever
            self.include_full_content = True
            self.settings_snapshot = settings_snapshot or {}

        def run(self, query, research_context=None):
            docs = self.retriever.get_relevant_documents(query)
            results = []
            for i, doc in enumerate(docs):
                results.append(
                    {
                        "title": f"Result {i + 1}",
                        "link": doc.metadata.get("source", "unknown"),
                        "snippet": doc.page_content[:200],
                        "full_content": doc.page_content
                        if self.include_full_content
                        else None,
                        "rank": i + 1,
                    }
                )
            return results

    # Create mock LLM
    llm = Mock()
    llm.invoke.return_value = Mock(content="Summary about machine learning.")

    # Create settings without programmatic_mode
    settings_snapshot = {
        "search.iterations": {"value": 1, "type": "int"},
        "search.questions_per_iteration": {"value": 2, "type": "int"},
        "search.strategy": {"value": "direct", "type": "str"},
        "search.max_results_per_query": {"value": 10, "type": "int"},
        "search.source_strategy.diversity_threshold": {
            "value": 0.8,
            "type": "float",
        },
        "search.source_strategy.min_relevance_score": {
            "value": 0.5,
            "type": "float",
        },
        "search.source_strategy.max_sources_per_topic": {
            "value": 5,
            "type": "int",
        },
        "search.source_strategy.enable_clustering": {
            "value": False,
            "type": "bool",
        },
        "search.cross_engine_max_results": {"value": 100, "type": "int"},
    }

    # Create search wrapper with settings
    search = SimpleRetrieverWrapper(retriever, settings_snapshot)

    # Create system with programmatic_mode as parameter
    system = AdvancedSearchSystem(
        llm=llm,
        search=search,
        settings_snapshot=settings_snapshot,
        programmatic_mode=True,
    )

    # Run a search
    result = system.analyze_topic("What is machine learning?")

    assert result is not None
    assert "findings" in result
    assert len(result["findings"]) > 0


def test_thread_context_without_database():
    """Test that thread context utilities work without database."""
    from local_deep_research.utilities.thread_context import (
        preserve_research_context,
    )

    # This should not fail even without database
    @preserve_research_context
    def sample_function():
        return "success"

    result = sample_function()
    assert result == "success"


if __name__ == "__main__":
    # Run tests
    test_import_without_database()
    print("✓ Import test passed")

    test_analyze_topic_without_database()
    print("✓ Analyze topic test passed")

    test_search_with_retriever()
    print("✓ Retriever search test passed")

    test_thread_context_without_database()
    print("✓ Thread context test passed")

    print("\nAll tests passed! Programmatic access works without database.")