DocsMCP/tests/test_db.py

"""
Tests for backend/app/db.py

These tests verify SQLite database operations including:
- Table creation (init_db)
- Library CRUD operations
- Document chunk storage and retrieval
- Full-text search functionality

All tests use a temporary test database file.
"""
import pytest
from datetime import datetime


class TestInitDatabase:
    """Tests for init_db() - table creation."""

    def test_init_db_creates_tables(self, test_database):
        """Database should have libraries and documents tables after init."""
        import sqlite3
        from backend.app.db import get_connection, get_db_path
        
        conn = get_connection()
        cursor = conn.execute("SELECT name FROM sqlite_master WHERE type='table' ORDER BY name")
        tables = [row[0] for row in cursor.fetchall()]
        
        # Should have libraries, documents, and FTS virtual table
        assert "libraries" in tables or any("libraries" in t.lower() for t in tables)
        conn.close()

    def test_init_db_returns_success(self, test_database):
        """init_db should return success indicator."""
        from backend.app.db import init_db
        
        result = init_db()
        assert result["success"] is True


class TestLibraryOperations:
    """Tests for library CRUD operations."""

    def test_upsert_library_new(self, test_database):
        """Upsert should create new library."""
        from backend.app.db import upsert_library
        
        result = upsert_library(
            library_id="/local/testlib",
            name="Test Library",
            description="A test library for unit tests"
        )
        
        assert result["success"] is True
        assert result["id"] == "/local/testlib"

    def test_upsert_library_update(self, test_database):
        """Upsert should update existing library."""
        from backend.app.db import upsert_library
        
        # Insert first library
        upsert_library(
            library_id="/local/upsertlib",
            name="Original Name",
            description="Original description"
        )
        
        # Update it
        result = upsert_library(
            library_id="/local/upsertlib",
            name="Updated Name",
            description="Updated description"
        )
        
        assert result["success"] is True

    def test_upsert_library_id_normalization(self, test_database):
        """Library ID normalization - /local/ prefix should be preserved."""
        from backend.app.db import upsert_library
        
        # Test various ID formats
        test_ids = [
            "/local/foundryvtt",
            "foundryvtt",
            "/local/mydocs",
        ]
        
        for lib_id in test_ids:
            result = upsert_library(library_id=lib_id, name="Test", description="Desc")
            assert result["success"] is True
            # Verify we can retrieve it back
            from backend.app.db import get_chunks_for_library
            # Just ensure no errors occur

    def test_list_libraries(self, test_database):
        """list_libraries should return list of libraries."""
        from backend.app.db import upsert_library, list_libraries
        
        # Create some libraries
        for i in range(3):
            upsert_library(
                library_id=f"/local/lib{i}",
                name=f"Library {i}",
                description=f"Description {i}"
            )
        
        libs = list_libraries()
        assert isinstance(libs, list)
        assert len(libs) >= 3

    def test_search_libraries(self, test_database):
        """search_libraries should find libraries by name/description."""
        from backend.app.db import upsert_library, search_libraries
        
        # Create libraries with searchable names
        upsert_library(library_id="/local/foo1", name="Foo Library", description="Bar baz")
        upsert_library(library_id="/local/foo2", name="Other Library", description="Different content")
        
        results = search_libraries("foo")
        assert isinstance(results, list)


class TestDocumentChunkOperations:
    """Tests for document chunk storage and retrieval."""

    def test_insert_document_chunk_new(self, test_database):
        """insert_document_chunk should create new chunk record."""
        from backend.app.db import insert_document_chunk
        
        result = insert_document_chunk(
            doc_id="doc-1",
            library_id="/local/testlib",
            path="docs/example.md",
            title="Example Document",
            content="# Example\n\nThis is the content.",
            chunk_index=0,
            token_estimate=100
        )
        
        assert result["success"] is True

    def test_insert_document_chunk_update(self, test_database):
        """insert_document_chunk should update existing record."""
        from backend.app.db import insert_document_chunk
        
        # Insert first
        insert_document_chunk(
            doc_id="doc-update-test",
            library_id="/local/uplib",
            path="old-path.md",
            title="Old Title",
            content="# Old\nContent here.",
            chunk_index=0,
            token_estimate=50
        )
        
        # Update it
        result = insert_document_chunk(
            doc_id="doc-update-test",
            library_id="/local/uplib",
            path="new-path.md",
            title="New Title",
            content="# New\nUpdated content.",
            chunk_index=1,
            token_estimate=75
        )
        
        assert result["success"] is True

    def test_get_document_by_id(self, test_database):
        """get_document_by_id should retrieve document by ID."""
        from backend.app.db import insert_document_chunk, get_document_by_id
        
        # Insert document
        doc_id = "unique-doc-id-12345"
        insert_document_chunk(
            doc_id=doc_id,
            library_id="/local/testlib",
            path="docs/test.md",
            title="Test Document",
            content="# Test\n\nTest content here.",
            chunk_index=None,
            token_estimate=200
        )
        
        # Retrieve it
        doc = get_document_by_id(doc_id)
        assert doc is not None
        assert doc["id"] == doc_id

    def test_get_chunks_for_library(self, test_database):
        """get_chunks_for_library should return all chunks for a library."""
        from backend.app.db import upsert_library, insert_document_chunk, get_chunks_for_library
        
        # Create library
        upsert_library(library_id="/local/chunktest", name="Chunk Test", description="Test")
        
        # Add some chunks
        for i in range(3):
            insert_document_chunk(
                doc_id=f"chunk-{i}",
                library_id="/local/chunktest",
                path=f"path{i}.md",
                title=f"Section {i}",
                content=f"Content section {i}.",
                chunk_index=i,
                token_estimate=50
            )
        
        chunks = get_chunks_for_library("/local/chunktest")
        assert isinstance(chunks, list)
        assert len(chunks) >= 3

    def test_clear_library_documents(self, test_database):
        """clear_library_documents should delete all docs for a library."""
        from backend.app.db import upsert_library, insert_document_chunk, clear_library_documents, get_chunks_for_library
        
        # Create and populate library
        upsert_library(library_id="/local/cleartest", name="Clear Test", description="Test")
        for i in range(5):
            insert_document_chunk(
                doc_id=f"clear-{i}",
                library_id="/local/cleartest",
                path=f"path{i}.md",
                content=f"Content {i}.",
                token_estimate=20
            )
        
        # Clear it
        result = clear_library_documents("/local/cleartest")
        assert result["success"] is True
        
        # Verify cleared
        remaining = get_chunks_for_library("/local/cleartest")
        assert len(remaining) == 0


class TestDatabaseEdgeCases:
    """Tests for edge cases and error handling."""

    def test_empty_library_id(self, test_database):
        """Operations with empty ID should handle gracefully."""
        from backend.app.db import upsert_library
        
        result = upsert_library(library_id="", name="Test", description="Desc")
        # Should not crash, though may not be a valid operation

    def test_special_characters_in_content(self, test_database):
        """Content with special characters should be stored."""
        from backend.app.db import insert_document_chunk
        
        content = "Hello \"world\" <tag /> & amp; 'apostrophe'"
        result = insert_document_chunk(
            doc_id="special-test",
            library_id="/local/speciallib",
            path="special.md",
            content=content,
            token_estimate=100
        )
        
        assert result["success"] is True

    def test_very_long_content(self, test_database):
        """Long content should be stored."""
        from backend.app.db import insert_document_chunk
        
        long_content = "a" * 5000
        result = insert_document_chunk(
            doc_id="long-test",
            library_id="/local/longlib",
            path="long.md",
            content=long_content,
            token_estimate=1000
        )
        
        assert result["success"] is True

    def test_none_description(self, test_database):
        """Library with None description should work."""
        from backend.app.db import upsert_library
        
        result = upsert_library(
            library_id="/local/nonedesc",
            name="No Description Lib",
            description=None
        )
        
        assert result["success"] is True


class TestDatabaseInitialization:
    """Tests for database initialization state."""

    def test_database_is_empty_after_init(self, test_database):
        """Database should be empty right after init."""
        from backend.app.db import list_libraries
        
        libs = list_libraries()
        assert isinstance(libs, list)


# =============================================================================
# FIXTURES
# =============================================================================

@pytest.fixture
def sample_doc():
    """Sample document chunk for testing."""
    return {
        "doc_id": "sample-doc-1",
        "library_id": "/local/samplelib",
        "path": "docs/guide.md",
        "title": "Getting Started Guide",
        "content": "# Getting Started\n\nWelcome to the guide. This is a sample document for testing.\n\n## Installation\n\nInstall with pip.",
        "chunk_index": 0,
        "token_estimate": 500
    }