"""Tests for initial prompt data validation."""

import json
from pathlib import Path

from kg_services.ontology import MCPPrompt


def test_initial_prompts_data_structure():
    """Test that initial_prompts.json has valid structure."""
    prompts_file = Path("data/initial_prompts.json")
    assert prompts_file.exists(), "initial_prompts.json file should exist"

    with prompts_file.open() as f:
        prompts_data = json.load(f)

    assert isinstance(prompts_data, list), "Prompts data should be a list"
    assert len(prompts_data) >= 6, "Should have at least 6 diverse prompts"

    # Check each prompt has required fields
    required_fields = [
        "prompt_id",
        "name",
        "description",
        "target_tool_id",
        "template_string",
        "tags",
        "input_variables",
        "use_case",
        "difficulty_level",
        "example_inputs",
    ]

    for prompt_data in prompts_data:
        for field in required_fields:
            assert field in prompt_data, f"Prompt missing required field: {field}"

        # Validate data types
        assert isinstance(prompt_data["prompt_id"], str)
        assert isinstance(prompt_data["name"], str)
        assert isinstance(prompt_data["description"], str)
        assert isinstance(prompt_data["target_tool_id"], str)
        assert isinstance(prompt_data["template_string"], str)
        assert isinstance(prompt_data["tags"], list)
        assert isinstance(prompt_data["input_variables"], list)
        assert isinstance(prompt_data["use_case"], str)
        assert isinstance(prompt_data["difficulty_level"], str)
        assert isinstance(prompt_data["example_inputs"], dict)


def test_initial_prompts_create_mcp_prompt_objects():
    """Test that initial prompts data can create valid MCPPrompt objects."""
    prompts_file = Path("data/initial_prompts.json")

    with prompts_file.open() as f:
        prompts_data = json.load(f)

    created_prompts = []
    for prompt_data in prompts_data:
        # Should not raise any validation errors
        prompt = MCPPrompt(**prompt_data)
        created_prompts.append(prompt)

        # Verify the object was created correctly
        assert prompt.prompt_id == prompt_data["prompt_id"]
        assert prompt.name == prompt_data["name"]
        assert prompt.description == prompt_data["description"]
        assert prompt.target_tool_id == prompt_data["target_tool_id"]
        assert prompt.template_string == prompt_data["template_string"]

    assert len(created_prompts) == len(prompts_data)


def test_initial_prompts_cover_all_tools():
    """Test that initial prompts cover all available tools."""
    # Load tools data
    tools_file = Path("data/initial_tools.json")
    with tools_file.open() as f:
        tools_data = json.load(f)

    # Load prompts data
    prompts_file = Path("data/initial_prompts.json")
    with prompts_file.open() as f:
        prompts_data = json.load(f)

    # Extract tool IDs and target tool IDs
    tool_ids = {tool["tool_id"] for tool in tools_data}
    target_tool_ids = {prompt["target_tool_id"] for prompt in prompts_data}

    # Verify all tools have at least one prompt
    for tool_id in tool_ids:
        assert (
            tool_id in target_tool_ids
        ), f"Tool {tool_id} should have at least one prompt"


def test_initial_prompts_template_variables():
    """Test that prompt templates have valid variable syntax."""
    prompts_file = Path("data/initial_prompts.json")

    with prompts_file.open() as f:
        prompts_data = json.load(f)

    for prompt_data in prompts_data:
        template = prompt_data["template_string"]
        input_variables = prompt_data["input_variables"]

        # Check that variables in template match input_variables list
        import re

        template_vars = re.findall(r"\{\{(\w+)\}\}", template)

        for var in template_vars:
            assert (
                var in input_variables
            ), f"Template variable '{var}' not in input_variables list for prompt {prompt_data['prompt_id']}"

        # Check that example_inputs cover all input_variables
        example_inputs = prompt_data["example_inputs"]
        for var in input_variables:
            assert (
                var in example_inputs
            ), f"Input variable '{var}' missing from example_inputs for prompt {prompt_data['prompt_id']}"


def test_initial_prompts_difficulty_levels():
    """Test that prompts have valid difficulty levels."""
    prompts_file = Path("data/initial_prompts.json")

    with prompts_file.open() as f:
        prompts_data = json.load(f)

    valid_levels = {"beginner", "intermediate", "advanced"}
    difficulty_distribution = {"beginner": 0, "intermediate": 0, "advanced": 0}

    for prompt_data in prompts_data:
        level = prompt_data["difficulty_level"]
        assert level in valid_levels, f"Invalid difficulty level: {level}"
        difficulty_distribution[level] += 1

    # Should have prompts at different difficulty levels
    assert difficulty_distribution["beginner"] > 0, "Should have beginner prompts"
    assert (
        difficulty_distribution["intermediate"] > 0
    ), "Should have intermediate prompts"
    assert difficulty_distribution["advanced"] > 0, "Should have advanced prompts"


def test_initial_prompts_unique_ids():
    """Test that all prompt IDs are unique."""
    prompts_file = Path("data/initial_prompts.json")

    with prompts_file.open() as f:
        prompts_data = json.load(f)

    prompt_ids = [prompt["prompt_id"] for prompt in prompts_data]
    unique_ids = set(prompt_ids)

    assert len(prompt_ids) == len(unique_ids), "All prompt IDs should be unique"